Joachim M. Vis commited on
Commit
9848958
·
verified ·
1 Parent(s): 81917a3

Upload 145 files

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .env_structure.txt +60 -0
  2. GAIA_ASSESSMENT_README.md +215 -0
  3. HF_README.md +22 -0
  4. IMPLEMENTATION_PLAN.md +155 -0
  5. README.md +294 -15
  6. README_original.md +294 -0
  7. SUPABASE_TABLES_README.md +122 -0
  8. SUPABASE_TEST_README.md +104 -0
  9. agent/__init__.py +22 -0
  10. agent/__pycache__/__init__.cpython-312.pyc +0 -0
  11. agent/__pycache__/__init__.cpython-313.pyc +0 -0
  12. agent/__pycache__/agent.cpython-312.pyc +0 -0
  13. agent/__pycache__/agent.cpython-313.pyc +0 -0
  14. agent/__pycache__/config.cpython-312.pyc +0 -0
  15. agent/__pycache__/config.cpython-313.pyc +0 -0
  16. agent/__pycache__/graph.cpython-312.pyc +0 -0
  17. agent/__pycache__/graph.cpython-313.pyc +0 -0
  18. agent/__pycache__/tool_registry.cpython-313.pyc +0 -0
  19. agent/agent.py +291 -0
  20. agent/config.py +186 -0
  21. agent/graph.py +452 -0
  22. agent/tool_registry.py +319 -0
  23. app.py +340 -195
  24. backup_tests/minimal_supabase_test.py +75 -0
  25. backup_tests/test_supabase_tables_final.py +167 -0
  26. backup_tests/test_supabase_tables_fix.py +276 -0
  27. backup_tests/test_supabase_tables_production.py +481 -0
  28. cleanup_results.txt +340 -0
  29. create_env_and_test.py +77 -0
  30. deploy_to_hf.py +161 -0
  31. deploy_to_huggingface.py +333 -0
  32. deploy_to_huggingface_credential.py +320 -0
  33. deploy_to_huggingface_dotenv.py +227 -0
  34. deploy_to_huggingface_env.py +203 -0
  35. deploy_with_token.py +176 -0
  36. deployment_readiness_report.md +97 -0
  37. docs/all_apis_integration_testing.md +170 -0
  38. docs/configuration.md +357 -0
  39. docs/connection_verification.md +114 -0
  40. docs/deployment.md +303 -0
  41. docs/duckduckgo_integration.md +156 -0
  42. docs/examples.md +197 -0
  43. docs/gaia_memory_integration.md +155 -0
  44. docs/index.md +63 -0
  45. docs/langchain_openai_integration.md +227 -0
  46. docs/multi_search_integration.md +318 -0
  47. docs/openai_integration_testing.md +138 -0
  48. docs/serper_integration.md +153 -0
  49. docs/supabase_memory_integration.md +175 -0
  50. docs/troubleshooting.md +281 -0
.env_structure.txt ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # API Keys and Endpoints
2
+ SUPABASE_URL=https://tjamxhvvtnypbadvrkjq.supabase.co
3
+
4
+ # API Endpoints
5
+ OPENAI_API_BASE=https://api.openai.com/v1
6
+ SERPER_API_URL=https://google.serper.dev/search
7
+ PERPLEXITY_API_URL=https://api.perplexity.ai
8
+ HUGGINGFACE_API_URL=https://api-inference.huggingface.co/models
9
+
10
+ # Model Configuration
11
+ DEFAULT_MODEL=gpt-4-turbo
12
+ EMBEDDING_MODEL=text-embedding-ada-002
13
+ FALLBACK_MODEL=gpt-3.5-turbo
14
+ VISION_MODEL=gpt-4-vision-preview
15
+ REASONING_MODEL=gpt-4-turbo
16
+ MODEL_TEMPERATURE=0.1
17
+
18
+ # Tool Parameters
19
+ WEB_SEARCH_RESULT_COUNT=5
20
+ WEB_SEARCH_TIMEOUT=10
21
+ ARXIV_MAX_RESULTS=3
22
+ DUCKDUCKGO_TIMEOUT=10
23
+ DUCKDUCKGO_MAX_RESULTS=5
24
+ PERPLEXITY_TIMEOUT=30
25
+ PERPLEXITY_MODEL=pplx-7b-online
26
+ WEB_SCRAPING_TIMEOUT=15
27
+ WEB_SCRAPING_MAX_LINKS=3
28
+ WEB_SCRAPING_MAX_CONTENT_LENGTH=10000
29
+
30
+ # Memory Settings
31
+ MEMORY_ENABLED=True
32
+ MEMORY_TABLE_NAME=gaia_memory
33
+ MEMORY_TTL=86400
34
+ MEMORY_CACHE_SIZE=100
35
+ MEMORY_SIMILARITY_THRESHOLD=0.85
36
+ MEMORY_MAX_ENTRIES=1000
37
+
38
+ # Agent Settings
39
+ MAX_ITERATIONS=10
40
+ VERBOSE=False
41
+ TIMEOUT=300
42
+ MAX_RETRIES=3
43
+ RETRY_DELAY=2
44
+ PARALLEL_TOOL_EXECUTION=False
45
+
46
+ # Logging Configuration
47
+ LOG_LEVEL=INFO
48
+ LOG_FILE=gaia_agent.log
49
+ LOG_FORMAT=%(asctime)s - %(name)s - %(levelname)s - %(message)s
50
+
51
+ # Multimodal Settings
52
+ IMAGE_MAX_SIZE=1024
53
+ IMAGE_FORMAT=JPEG
54
+ VISION_DETAIL_LEVEL=high
55
+
56
+ # Web Browsing Settings
57
+ USER_AGENT=Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36
58
+ BROWSER_TIMEOUT=30
59
+ BROWSER_HEADLESS=True
60
+ BROWSER_IMAGES_ENABLED=True
GAIA_ASSESSMENT_README.md ADDED
@@ -0,0 +1,215 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # GAIA Assessment Final Integration Test
2
+
3
+ This document provides comprehensive information about the GAIA assessment, the Supabase tables fix, and how to run the final integration test.
4
+
5
+ ## Table of Contents
6
+
7
+ 1. [Overview](#overview)
8
+ 2. [Supabase Tables Fix](#supabase-tables-fix)
9
+ 3. [Scripts Created](#scripts-created)
10
+ 4. [Running the GAIA Assessment](#running-the-gaia-assessment)
11
+ 5. [Interpreting Results](#interpreting-results)
12
+ 6. [Troubleshooting](#troubleshooting)
13
+
14
+ ## Overview
15
+
16
+ The GAIA agent is an AI assistant that uses Supabase for memory storage. The agent uses several tables to store different types of memory:
17
+
18
+ - `gaia_memory`: Main memory table for general storage
19
+ - `gaia_memory_conversations`: Conversation history table
20
+ - `gaia_memory_cache`: Result cache table for performance optimization
21
+ - `gaia_memory_working`: Working memory table for complex reasoning
22
+
23
+ This integration test verifies that all components work together correctly, especially the Supabase tables fix.
24
+
25
+ ## Supabase Tables Fix
26
+
27
+ ### Issue Description
28
+
29
+ The original implementation had issues with the Supabase tables:
30
+
31
+ 1. Tables were not being created automatically if they didn't exist
32
+ 2. The memory system would fail if any of the required tables were missing
33
+ 3. There was no proper error handling for table operations
34
+ 4. The cache mechanism wasn't properly integrated with the other memory components
35
+
36
+ ### Fix Implementation
37
+
38
+ The fix was implemented in the following files:
39
+
40
+ 1. `memory/fix_supabase_tables.py`: Contains the main fix implementation
41
+ - `ensure_tables_exist()`: Creates all required tables if they don't exist
42
+ - `apply_monkey_patch()`: Patches the memory system to use the fixed implementation
43
+
44
+ 2. `memory/supabase_memory_patch.py`: Contains the patched methods for the memory system
45
+ - Improved error handling
46
+ - Automatic table creation
47
+ - Better integration between memory components
48
+
49
+ 3. `memory/simplified_memory_fix.py`: A simplified version of the fix for testing
50
+
51
+ The fix ensures that:
52
+
53
+ 1. All required tables are created automatically if they don't exist
54
+ 2. Operations gracefully handle missing tables by creating them
55
+ 3. Proper error messages are provided when operations fail
56
+ 4. The cache mechanism is properly integrated with the other memory components
57
+
58
+ ## Scripts Created
59
+
60
+ The following scripts were created to test and verify the GAIA agent:
61
+
62
+ 1. `verify_gaia_assessment.py`: The main integration test script
63
+ - Tests all components together
64
+ - Verifies that the Supabase tables fix is working correctly
65
+ - Tests with complex questions that exercise all memory types
66
+ - Verifies multi-turn conversations
67
+ - Checks caching for repeated questions
68
+ - Verifies working memory for complex reasoning
69
+
70
+ 2. `run_gaia_with_fix.py`: A script to run the GAIA agent with the Supabase tables fix applied
71
+
72
+ 3. `tests/test_supabase_tables.py`: A test script for the Supabase tables
73
+ - Tests table existence
74
+ - Tests basic memory operations
75
+ - Tests conversation memory
76
+ - Tests result cache
77
+ - Tests working memory
78
+
79
+ 4. `tests/test_supabase_tables_comprehensive.py`: A comprehensive test for the Supabase tables
80
+ - More detailed tests for all memory components
81
+ - Performance testing for the cache mechanism
82
+ - Multi-turn conversation testing
83
+
84
+ 5. `tests/test_all_supabase_tables.py`: A test script that verifies all Supabase tables are working correctly
85
+
86
+ 6. `tests/test_gaia_with_tables.py`: A test script that verifies the GAIA agent works correctly with the Supabase tables
87
+
88
+ ## Running the GAIA Assessment
89
+
90
+ ### Prerequisites
91
+
92
+ Before running the GAIA assessment, make sure you have:
93
+
94
+ 1. Python 3.8 or higher installed
95
+ 2. All dependencies installed: `pip install -r requirements.txt`
96
+ 3. Environment variables set:
97
+ - `OPENAI_API_KEY`: Your OpenAI API key
98
+ - `SUPABASE_URL`: Your Supabase URL
99
+ - `SUPABASE_KEY`: Your Supabase key
100
+
101
+ ### Running the Final Verification
102
+
103
+ To run the final verification, use the `verify_gaia_assessment.py` script:
104
+
105
+ ```bash
106
+ python verify_gaia_assessment.py [options]
107
+ ```
108
+
109
+ Options:
110
+ - `--verbose`: Enable verbose logging
111
+ - `--show-all`: Show all table entries
112
+ - `--metrics`: Show detailed performance metrics
113
+
114
+ Example:
115
+ ```bash
116
+ python verify_gaia_assessment.py --verbose
117
+ ```
118
+
119
+ ### What to Expect
120
+
121
+ The script will run a series of tests:
122
+
123
+ 1. Simple Question Test: Tests a basic question to verify the agent works
124
+ 2. Cached Question Test: Tests the same question again to verify caching
125
+ 3. Follow-up Question Test: Tests a follow-up question to verify context retention
126
+ 4. Complex Question Test: Tests a complex question that exercises all memory types
127
+ 5. Multi-turn Conversation Test: Tests a multi-turn conversation to verify conversation history
128
+
129
+ After each test, the script will show:
130
+ - The question and answer
131
+ - Execution time
132
+ - Table state comparison (before and after)
133
+ - Detailed table content (if `--show-all` is specified)
134
+
135
+ The script will also verify:
136
+ - Working memory usage
137
+ - Conversation memory usage
138
+ - Cache usage
139
+
140
+ Finally, the script will save the results to a JSON file: `gaia_assessment_results_YYYYMMDD_HHMMSS.json`
141
+
142
+ ## Interpreting Results
143
+
144
+ The results JSON file contains:
145
+
146
+ 1. Timestamp: When the test was run
147
+ 2. Tests: Results for each test
148
+ - Question and answer
149
+ - Execution time
150
+ - Initial and final table state
151
+ - Error (if any)
152
+
153
+ For the multi-turn conversation test, it also includes:
154
+ - All questions and answers
155
+ - Conversation history from memory
156
+
157
+ For the verification tests, it includes:
158
+ - Whether the memory component has entries
159
+ - The number of entries
160
+ - Example entries
161
+
162
+ A successful test should show:
163
+ - All questions answered correctly
164
+ - Faster execution time for cached questions
165
+ - Proper context retention for follow-up questions
166
+ - Working memory entries for complex questions
167
+ - Conversation history for multi-turn conversations
168
+ - Cache entries for repeated questions
169
+
170
+ ## Troubleshooting
171
+
172
+ ### Common Issues
173
+
174
+ 1. **Missing Environment Variables**
175
+ - Error: "Missing required environment variables"
176
+ - Solution: Set the required environment variables (OPENAI_API_KEY, SUPABASE_URL, SUPABASE_KEY)
177
+
178
+ 2. **Supabase Connection Issues**
179
+ - Error: "Error creating Supabase client"
180
+ - Solution: Check your Supabase URL and key, and make sure your Supabase instance is running
181
+
182
+ 3. **OpenAI API Issues**
183
+ - Error: "Error initializing GAIA agent"
184
+ - Solution: Check your OpenAI API key and make sure you have sufficient credits
185
+
186
+ 4. **Table Creation Issues**
187
+ - Error: "Error applying Supabase tables fix"
188
+ - Solution: Make sure your Supabase user has permission to create tables
189
+
190
+ 5. **Memory System Issues**
191
+ - Error: "Error initializing memory system"
192
+ - Solution: Check the logs for more details, and make sure the Supabase tables fix is applied correctly
193
+
194
+ ### Debugging Tips
195
+
196
+ 1. Use the `--verbose` option to get more detailed logs
197
+ 2. Check the logs for error messages and stack traces
198
+ 3. Use the `--show-all` option to see all table entries
199
+ 4. Check the Supabase dashboard to see if the tables were created correctly
200
+ 5. Try running the individual test scripts to isolate issues
201
+
202
+ ### Getting Help
203
+
204
+ If you encounter issues that you can't resolve, please:
205
+
206
+ 1. Check the logs for error messages
207
+ 2. Check the Supabase dashboard for table status
208
+ 3. Try running the individual test scripts
209
+ 4. Contact the GAIA team with the error message and logs
210
+
211
+ ## Conclusion
212
+
213
+ The GAIA assessment with the Supabase tables fix should now be working correctly. The final integration test verifies that all components work together, and the fix ensures that the Supabase tables are created and used correctly.
214
+
215
+ If you have any questions or issues, please contact the GAIA team.
HF_README.md ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # GAIA: Generative AI Assistant
2
+
3
+ A versatile AI assistant with memory, search capabilities, and multi-API integration.
4
+
5
+ ## Features
6
+
7
+ - OpenAI integration with GPT-4 and fallback models
8
+ - Memory persistence with Supabase
9
+ - Web search capabilities via Serper and DuckDuckGo
10
+ - Agent-based architecture with tool registry
11
+ - Multi-turn conversation support
12
+
13
+ ## Usage
14
+
15
+ 1. Enter your query in the chat interface
16
+ 2. GAIA will process your request, using tools as needed
17
+ 3. Responses are retained in memory for context in future interactions
18
+
19
+ ## Credits
20
+
21
+ Developed as part of the Hugging Face Agents Course
22
+
IMPLEMENTATION_PLAN.md ADDED
@@ -0,0 +1,155 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # GAIA Agent Implementation Plan
2
+
3
+ This document outlines the detailed implementation plan for creating our GAIA agent using LangChain/LangGraph and the required tools. Each section represents a distinct implementation phase with specific tasks and deliverables.
4
+
5
+ ## Phase 1: Project Setup and Dependencies
6
+
7
+ ### Task 1.1: Environment Setup
8
+ - Create project structure with appropriate folders
9
+ - Set up virtual environment
10
+ - Update requirements.txt with all necessary dependencies:
11
+ ```
12
+ gradio>=5.25.2
13
+ requests
14
+ langchain
15
+ langgraph
16
+ supabase
17
+ duckduckgo-search
18
+ beautifulsoup4
19
+ pillow
20
+ arxiv
21
+ perplexity-python
22
+ python-dotenv
23
+ ```
24
+
25
+ ### Task 1.2: Configuration Management
26
+ - Create `agent/config.py` for centralized configuration
27
+ - Implement environment variable loading with dotenv
28
+ - Define configuration for:
29
+ - API endpoints
30
+ - Model selection
31
+ - Tool parameters
32
+ - Memory settings
33
+
34
+ ## Phase 2: Core Agent Architecture
35
+
36
+ ### Task 2.1: Basic Agent Implementation
37
+ - Create `agent/agent.py` with the `GAIAAgent` class
38
+ - Implement the core `__call__` method to process questions
39
+ - Add logging and error handling
40
+
41
+ ### Task 2.2: LangGraph Workflow
42
+ - Create `agent/graph.py` to define the agent's workflow
43
+ - Implement nodes for:
44
+ - Question analysis
45
+ - Planning
46
+ - Tool selection
47
+ - Tool execution
48
+ - Answer formulation
49
+ - Define edges between nodes and conditional routing
50
+
51
+ ### Task 2.3: Memory Integration
52
+ - Create `memory/supabase_memory.py` for persistent memory
53
+ - Implement memory interfaces for:
54
+ - Storing intermediate results
55
+ - Caching previous answers
56
+ - Maintaining conversation context
57
+
58
+ ## Phase 3: Tool Implementation
59
+
60
+ ### Task 3.1: Web Browsing Tools
61
+ - Create `tools/web_tools.py` with implementations for:
62
+ - Web search (DuckDuckGo, Serper)
63
+ - Web page content extraction
64
+ - URL navigation and scraping
65
+
66
+ ### Task 3.2: Reasoning Tools
67
+ - Create `tools/reasoning_tools.py` with implementations for:
68
+ - Step-by-step reasoning
69
+ - Mathematical calculations
70
+ - Fact verification
71
+
72
+ ### Task 3.3: Multimodal Tools
73
+ - Create `tools/multimodal_tools.py` with implementations for:
74
+ - Image analysis and description
75
+ - Chart/graph interpretation
76
+ - Document parsing
77
+
78
+ ### Task 3.4: Tool Registry
79
+ - Create `tools/__init__.py` with a tool registry
80
+ - Implement tool selection logic based on question requirements
81
+ - Add metadata to tools for better selection
82
+
83
+ ## Phase 4: Question Processing
84
+
85
+ ### Task 4.1: Question Analysis
86
+ - Create `utils/parsing.py` with functions for:
87
+ - Extracting key information from questions
88
+ - Identifying required tools
89
+ - Determining question complexity
90
+
91
+ ### Task 4.2: Answer Formatting
92
+ - Implement answer extraction and formatting
93
+ - Ensure answers match the expected format (number, few words, comma-separated list)
94
+ - Add validation to verify answer format before submission
95
+
96
+ ## Phase 5: Integration with Gradio Interface
97
+
98
+ ### Task 5.1: Agent Integration
99
+ - Update `app.py` to use our new `GAIAAgent` class
100
+ - Ensure proper error handling and logging
101
+ - Implement progress tracking for long-running operations
102
+
103
+ ### Task 5.2: UI Enhancements
104
+ - Add optional UI improvements:
105
+ - Progress indicators
106
+ - Detailed result views
107
+ - Error explanations
108
+
109
+ ## Phase 6: Testing and Evaluation
110
+
111
+ ### Task 6.1: Unit Testing
112
+ - Create test cases for individual components
113
+ - Implement automated tests for tools and utilities
114
+ - Add integration tests for the complete agent workflow
115
+
116
+ ### Task 6.2: Benchmark Testing
117
+ - Test against sample GAIA questions
118
+ - Measure performance and accuracy
119
+ - Identify and address bottlenecks
120
+
121
+ ## Phase 7: Documentation and Deployment
122
+
123
+ ### Task 7.1: Code Documentation
124
+ - Add comprehensive docstrings to all functions and classes
125
+ - Create usage examples
126
+ - Document configuration options
127
+
128
+ ### Task 7.2: Deployment Preparation
129
+ - Ensure all dependencies are properly specified
130
+ - Verify compatibility with Hugging Face Spaces
131
+ - Create deployment instructions
132
+
133
+ ## Implementation Timeline
134
+
135
+ | Phase | Estimated Duration | Dependencies |
136
+ |-------|-------------------|--------------|
137
+ | 1: Project Setup | 1 day | None |
138
+ | 2: Core Architecture | 2 days | Phase 1 |
139
+ | 3: Tool Implementation | 3 days | Phase 2 |
140
+ | 4: Question Processing | 2 days | Phase 2 |
141
+ | 5: Gradio Integration | 1 day | Phases 2-4 |
142
+ | 6: Testing | 2 days | Phases 1-5 |
143
+ | 7: Documentation | 1 day | Phases 1-6 |
144
+
145
+ ## Success Criteria
146
+
147
+ - Agent successfully answers at least 30% of GAIA benchmark questions
148
+ - Code is well-organized, documented, and maintainable
149
+ - Implementation follows best practices for LangChain/LangGraph
150
+ - All tools function correctly and efficiently
151
+ - Submission process works end-to-end
152
+
153
+ ## Next Steps
154
+
155
+ After reviewing this implementation plan, we will proceed with Phase 1 by setting up the project structure and installing dependencies. Each subsequent phase will be implemented as a separate task, with regular testing and validation throughout the process.
README.md CHANGED
@@ -1,15 +1,294 @@
1
- ---
2
- title: Template Final Assignment
3
- emoji: 🕵🏻‍♂️
4
- colorFrom: indigo
5
- colorTo: indigo
6
- sdk: gradio
7
- sdk_version: 5.25.2
8
- app_file: app.py
9
- pinned: false
10
- hf_oauth: true
11
- # optional, default duration is 8 hours/480 minutes. Max duration is 30 days/43200 minutes.
12
- hf_oauth_expiration_minutes: 480
13
- ---
14
-
15
- Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # GAIA Agent - Hugging Face Agents Course Final Project
2
+
3
+ ## Project Overview
4
+
5
+ This project implements an intelligent agent capable of answering questions from the GAIA (Grounded AI Alignment) benchmark as part of the Hugging Face Agents Course final assignment. The agent leverages LangChain/LangGraph frameworks and various tools to solve complex, multi-step reasoning tasks.
6
+
7
+ ## What is GAIA?
8
+
9
+ GAIA is a benchmark designed to evaluate AI assistants on real-world tasks that require a combination of core capabilities:
10
+ - Reasoning across multiple steps
11
+ - Multimodal understanding (text, images, etc.)
12
+ - Web browsing and information retrieval
13
+ - Proficient tool use
14
+
15
+ The benchmark consists of 466 carefully curated questions that are conceptually simple for humans (~92% success rate) but challenging for current AI systems (GPT-4 with plugins: ~15% success rate).
16
+
17
+ ## Challenge Requirements
18
+
19
+ - **Task**: Create an agent that answers questions from a subset of GAIA (20 questions from level 1)
20
+ - **Success Criteria**: Score at least 30% on the benchmark
21
+ - **Evaluation Method**: Exact match comparison with ground truth answers
22
+ - **Response Format**: Answers should be concise (number, few words, or comma-separated list)
23
+
24
+ ## Technical Architecture
25
+
26
+ Our implementation uses:
27
+
28
+ 1. **LangChain/LangGraph**: For creating a flexible, modular agent architecture
29
+ 2. **Supabase**: For memory management and persistence
30
+ 3. **Web Tools**: Integration with search engines (DuckDuckGo, Serper), academic resources (arXiv), and AI assistants (Perplexity)
31
+ 4. **Gradio**: For the user interface and submission process
32
+
33
+ ## Project Structure
34
+
35
+ ```
36
+ ├── README.md # Project documentation
37
+ ├── app.py # Main application file with Gradio interface
38
+ ├── requirements.txt # Project dependencies
39
+ ├── agent/ # Agent implementation
40
+ │ ├── __init__.py # Package initialization
41
+ │ ├── agent.py # Main agent class
42
+ │ ├── config.py # Configuration settings
43
+ │ ├── graph.py # LangGraph workflow definition
44
+ │ └── tool_registry.py # Tool registration and management
45
+ ├── tools/ # Tool implementations
46
+ │ ├── __init__.py # Package initialization
47
+ │ ├── web_tools.py # Web browsing and search tools
48
+ │ ├── reasoning_tools.py # Reasoning and analysis tools
49
+ │ ├── multimodal_tools.py # Tools for handling images and other media
50
+ │ ├── arxiv_tool.py # Tools for searching academic papers
51
+ │ └── perplexity_tool.py # Integration with Perplexity AI
52
+ ├── memory/ # Memory management
53
+ │ ├── __init__.py # Package initialization
54
+ │ ├── supabase_memory.py # Supabase integration for memory
55
+ │ ├── supabase_memory_patch.py # Patches for Supabase memory
56
+ │ └── simplified_memory_fix.py # Simplified memory implementation
57
+ └── utils/ # Utility functions
58
+ ├── __init__.py # Package initialization
59
+ ├── parsing.py # Functions for parsing questions and answers
60
+ └── formatting.py # Formatting utilities
61
+ ```
62
+
63
+ ## Implementation Approach
64
+
65
+ Our agent follows a multi-step process to answer GAIA questions:
66
+
67
+ 1. **Question Analysis**: Parse and understand the question, identifying required tools and information sources
68
+ 2. **Planning**: Develop a step-by-step plan to answer the question
69
+ 3. **Tool Execution**: Use appropriate tools in the correct sequence
70
+ 4. **Answer Formulation**: Extract and format the precise answer from the results
71
+ 5. **Verification**: Validate the answer format and content before submission
72
+
73
+ ## Key Features
74
+
75
+ - **Modular Architecture**: Easily extensible with new tools and capabilities
76
+ - **Robust Error Handling**: Graceful recovery from tool failures or unexpected results
77
+ - **Efficient Tool Selection**: Smart selection of the most appropriate tools for each question
78
+ - **Answer Precision**: Focus on extracting exact answers that match the ground truth format
79
+ - **Memory Integration**: Persistent memory across sessions using Supabase
80
+
81
+ ## Getting Started
82
+
83
+ ### Prerequisites
84
+
85
+ - Python 3.8 or higher
86
+ - OpenAI API key
87
+ - Supabase account and API key (for memory features)
88
+ - Optional: Serper API key (for enhanced search)
89
+ - Optional: Perplexity API key (for enhanced reasoning)
90
+
91
+ ### Installation
92
+
93
+ 1. Clone this repository
94
+ 2. Install dependencies:
95
+ ```bash
96
+ pip install -r requirements.txt
97
+ ```
98
+
99
+ > **Note**: If you encounter issues with the `langchain_openai` dependency, make sure you have the latest version installed:
100
+ > ```bash
101
+ > pip install --upgrade langchain_openai
102
+ > ```
103
+
104
+ 3. Set up environment variables (create a `.env` file in the project root):
105
+ ```
106
+ OPENAI_API_KEY=your_openai_api_key
107
+ SUPABASE_KEY=your_supabase_api_key
108
+ SERPER_API_KEY=your_serper_api_key (optional)
109
+ PERPLEXITY_API_KEY=your_perplexity_api_key (optional)
110
+ ```
111
+
112
+ 4. Run the application:
113
+ ```bash
114
+ python app.py
115
+ ```
116
+
117
+ 5. Log in with your Hugging Face account
118
+ 6. Click "Run Evaluation & Submit All Answers" to test the agent
119
+
120
+ ## Testing
121
+
122
+ The project includes comprehensive unit tests for all components:
123
+
124
+ - **Agent Tests**: Test the GAIAAgent class and its workflow
125
+ - **Tools Tests**: Test the web, reasoning, and multimodal tools
126
+ - **Memory Tests**: Test the memory components and Supabase integration
127
+ - **Utils Tests**: Test utility functions for parsing and formatting
128
+ - **Integration Tests**: Test integration with external APIs like OpenAI and Perplexity
129
+
130
+ To run all tests:
131
+ ```bash
132
+ python -m unittest discover tests
133
+ ```
134
+
135
+ To run specific test files:
136
+ ```bash
137
+ python -m unittest tests.test_agent
138
+ python -m unittest tests.test_tools
139
+ python -m unittest tests.test_memory
140
+ python -m unittest tests.test_utils
141
+ ```
142
+
143
+ ### OpenAI Integration Testing
144
+
145
+ The project includes specialized tests for OpenAI integration with newer models:
146
+
147
+ ```bash
148
+ # Test OpenAI integration with newer models
149
+ python tests/test_openai_integration.py
150
+
151
+ # Test OpenAI integration with GAIA agent
152
+ python tests/test_openai_integration.py --test-agent
153
+
154
+ # Test GAIA agent with multiple AI providers (OpenAI and Perplexity)
155
+ python tests/test_gaia_multi_provider.py
156
+ ```
157
+
158
+ For convenience, you can run all OpenAI integration tests at once:
159
+
160
+ ```bash
161
+ # Run all OpenAI integration tests
162
+ python tests/run_openai_tests.py
163
+
164
+ # Run all tests with verbose logging
165
+ python tests/run_openai_tests.py --verbose
166
+
167
+ # Run all tests and save results to a custom directory
168
+ python tests/run_openai_tests.py --output my_test_results
169
+
170
+ # Stop running tests after the first failure
171
+ python tests/run_openai_tests.py --stop-on-failure
172
+ ```
173
+
174
+ These tests verify that:
175
+ - The OpenAI API connection works correctly
176
+ - Newer models like GPT-4o can be used successfully
177
+ - The GAIA agent can use both OpenAI and Perplexity tools together
178
+
179
+ ### API Integration Testing
180
+
181
+ To test all API integrations together:
182
+
183
+ ```bash
184
+ # Run all API integration tests
185
+ python tests/run_all_apis_integration_test.py
186
+
187
+ # Run with verbose logging
188
+ python tests/run_all_apis_integration_test.py --verbose
189
+
190
+ # Specify a custom output file
191
+ python tests/run_all_apis_integration_test.py --report-file my_report.json
192
+ ```
193
+
194
+ ### Supabase Integration Testing
195
+
196
+ To test the Supabase memory integration:
197
+
198
+ ```bash
199
+ # Run all Supabase integration tests
200
+ python tests/run_supabase_integration_tests.py
201
+
202
+ # Run with verbose logging
203
+ python tests/run_supabase_integration_tests.py --verbose
204
+
205
+ # Specify a custom output directory
206
+ python tests/run_supabase_integration_tests.py --output-dir custom_docs
207
+ ```
208
+
209
+ ## Evaluation
210
+
211
+ The agent is evaluated on its ability to provide exact matches to the ground truth answers for the GAIA benchmark questions. The submission process includes:
212
+
213
+ - Fetching questions from the course API
214
+ - Running the agent on each question
215
+ - Submitting answers for scoring
216
+ - Viewing results on the leaderboard
217
+
218
+ ## Benchmark Testing
219
+
220
+ The project includes a benchmark testing framework to evaluate the agent's performance:
221
+
222
+ ```bash
223
+ # Run benchmark tests with default settings
224
+ python -m tests.benchmark --output benchmark_results.json
225
+
226
+ # Run benchmark tests with custom questions
227
+ python -m tests.benchmark --questions custom_questions.json --output benchmark_results.json
228
+
229
+ # Run benchmark tests with a specific number of questions
230
+ python -m tests.benchmark --max-questions 5 --output benchmark_results.json
231
+ ```
232
+
233
+ The benchmark testing framework:
234
+ - Tests the agent against sample questions
235
+ - Measures performance metrics (accuracy, latency, resource usage)
236
+ - Generates detailed reports of benchmark results
237
+ - Allows comparison of different agent configurations
238
+
239
+ ## Troubleshooting
240
+
241
+ ### Common Issues
242
+
243
+ #### 1. langchain_openai Dependency Issues
244
+
245
+ If you encounter errors related to the `langchain_openai` package, such as:
246
+ - `ImportError: No module named 'langchain_openai'`
247
+ - `AttributeError: module 'langchain_openai' has no attribute 'ChatOpenAI'`
248
+
249
+ Try the following solutions:
250
+ - Ensure you have the latest version installed: `pip install --upgrade langchain_openai`
251
+ - Make sure you're using the correct import: `from langchain_openai import ChatOpenAI`
252
+ - Check for conflicts with older langchain versions: `pip uninstall langchain-openai` (note the hyphen)
253
+
254
+ #### 2. API Connection Issues
255
+
256
+ If you're having trouble connecting to external APIs:
257
+ - Verify that your API keys are correctly set in the environment variables
258
+ - Check that the API service is not experiencing downtime
259
+ - Ensure you have a stable internet connection
260
+ - Check for rate limiting or quota issues
261
+
262
+ #### 3. Supabase Connection Issues
263
+
264
+ If you're having trouble with Supabase:
265
+ - Verify that your Supabase URL and API key are correctly set
266
+ - Ensure that the required tables exist in your Supabase project
267
+ - Check that your API key has the necessary permissions
268
+ - Verify that your Supabase project is active
269
+
270
+ #### 4. Memory Issues
271
+
272
+ If you're experiencing issues with the agent's memory:
273
+ - Check that memory is enabled in the configuration
274
+ - Verify that the table names are correctly set
275
+ - Ensure that the TTL and cache size settings are appropriate
276
+ - Check for errors in the Supabase logs
277
+
278
+ ## Documentation
279
+
280
+ Comprehensive documentation is available in the `docs/` directory:
281
+
282
+ - [Usage Examples](docs/examples.md): Examples of how to use the GAIA agent in different scenarios
283
+ - [Configuration Guide](docs/configuration.md): Detailed information about configuration options
284
+ - [Deployment Guide](docs/deployment.md): Instructions for deploying the agent in various environments
285
+ - [OpenAI Integration Testing](docs/openai_integration_testing.md): Guide for testing OpenAI integration with newer models
286
+ - [Connection Verification](docs/connection_verification.md): Guide for verifying connections to external services
287
+ - [Supabase Memory Integration](docs/supabase_memory_integration.md): Guide for setting up and using Supabase memory
288
+ - [All APIs Integration Testing](docs/all_apis_integration_testing.md): Guide for testing all API integrations together
289
+ - [Multi-Search Integration](docs/multi_search_integration.md): Guide for using multiple search providers
290
+ - [GAIA Memory Integration](docs/gaia_memory_integration.md): Guide for using memory in the GAIA agent
291
+
292
+ ## License
293
+
294
+ This project is part of the Hugging Face Agents Course and follows its licensing terms.
README_original.md ADDED
@@ -0,0 +1,294 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # GAIA Agent - Hugging Face Agents Course Final Project
2
+
3
+ ## Project Overview
4
+
5
+ This project implements an intelligent agent capable of answering questions from the GAIA (Grounded AI Alignment) benchmark as part of the Hugging Face Agents Course final assignment. The agent leverages LangChain/LangGraph frameworks and various tools to solve complex, multi-step reasoning tasks.
6
+
7
+ ## What is GAIA?
8
+
9
+ GAIA is a benchmark designed to evaluate AI assistants on real-world tasks that require a combination of core capabilities:
10
+ - Reasoning across multiple steps
11
+ - Multimodal understanding (text, images, etc.)
12
+ - Web browsing and information retrieval
13
+ - Proficient tool use
14
+
15
+ The benchmark consists of 466 carefully curated questions that are conceptually simple for humans (~92% success rate) but challenging for current AI systems (GPT-4 with plugins: ~15% success rate).
16
+
17
+ ## Challenge Requirements
18
+
19
+ - **Task**: Create an agent that answers questions from a subset of GAIA (20 questions from level 1)
20
+ - **Success Criteria**: Score at least 30% on the benchmark
21
+ - **Evaluation Method**: Exact match comparison with ground truth answers
22
+ - **Response Format**: Answers should be concise (number, few words, or comma-separated list)
23
+
24
+ ## Technical Architecture
25
+
26
+ Our implementation uses:
27
+
28
+ 1. **LangChain/LangGraph**: For creating a flexible, modular agent architecture
29
+ 2. **Supabase**: For memory management and persistence
30
+ 3. **Web Tools**: Integration with search engines (DuckDuckGo, Serper), academic resources (arXiv), and AI assistants (Perplexity)
31
+ 4. **Gradio**: For the user interface and submission process
32
+
33
+ ## Project Structure
34
+
35
+ ```
36
+ ├── README.md # Project documentation
37
+ ├── app.py # Main application file with Gradio interface
38
+ ├── requirements.txt # Project dependencies
39
+ ├── agent/ # Agent implementation
40
+ │ ├── __init__.py # Package initialization
41
+ │ ├── agent.py # Main agent class
42
+ │ ├── config.py # Configuration settings
43
+ │ ├── graph.py # LangGraph workflow definition
44
+ │ └── tool_registry.py # Tool registration and management
45
+ ├── tools/ # Tool implementations
46
+ │ ├── __init__.py # Package initialization
47
+ │ ├── web_tools.py # Web browsing and search tools
48
+ │ ├── reasoning_tools.py # Reasoning and analysis tools
49
+ │ ├── multimodal_tools.py # Tools for handling images and other media
50
+ │ ├── arxiv_tool.py # Tools for searching academic papers
51
+ │ └── perplexity_tool.py # Integration with Perplexity AI
52
+ ├── memory/ # Memory management
53
+ │ ├── __init__.py # Package initialization
54
+ │ ├── supabase_memory.py # Supabase integration for memory
55
+ │ ├── supabase_memory_patch.py # Patches for Supabase memory
56
+ │ └── simplified_memory_fix.py # Simplified memory implementation
57
+ └── utils/ # Utility functions
58
+ ├── __init__.py # Package initialization
59
+ ├── parsing.py # Functions for parsing questions and answers
60
+ └── formatting.py # Formatting utilities
61
+ ```
62
+
63
+ ## Implementation Approach
64
+
65
+ Our agent follows a multi-step process to answer GAIA questions:
66
+
67
+ 1. **Question Analysis**: Parse and understand the question, identifying required tools and information sources
68
+ 2. **Planning**: Develop a step-by-step plan to answer the question
69
+ 3. **Tool Execution**: Use appropriate tools in the correct sequence
70
+ 4. **Answer Formulation**: Extract and format the precise answer from the results
71
+ 5. **Verification**: Validate the answer format and content before submission
72
+
73
+ ## Key Features
74
+
75
+ - **Modular Architecture**: Easily extensible with new tools and capabilities
76
+ - **Robust Error Handling**: Graceful recovery from tool failures or unexpected results
77
+ - **Efficient Tool Selection**: Smart selection of the most appropriate tools for each question
78
+ - **Answer Precision**: Focus on extracting exact answers that match the ground truth format
79
+ - **Memory Integration**: Persistent memory across sessions using Supabase
80
+
81
+ ## Getting Started
82
+
83
+ ### Prerequisites
84
+
85
+ - Python 3.8 or higher
86
+ - OpenAI API key
87
+ - Supabase account and API key (for memory features)
88
+ - Optional: Serper API key (for enhanced search)
89
+ - Optional: Perplexity API key (for enhanced reasoning)
90
+
91
+ ### Installation
92
+
93
+ 1. Clone this repository
94
+ 2. Install dependencies:
95
+ ```bash
96
+ pip install -r requirements.txt
97
+ ```
98
+
99
+ > **Note**: If you encounter issues with the `langchain_openai` dependency, make sure you have the latest version installed:
100
+ > ```bash
101
+ > pip install --upgrade langchain_openai
102
+ > ```
103
+
104
+ 3. Set up environment variables (create a `.env` file in the project root):
105
+ ```
106
+ OPENAI_API_KEY=your_openai_api_key
107
+ SUPABASE_KEY=your_supabase_api_key
108
+ SERPER_API_KEY=your_serper_api_key (optional)
109
+ PERPLEXITY_API_KEY=your_perplexity_api_key (optional)
110
+ ```
111
+
112
+ 4. Run the application:
113
+ ```bash
114
+ python app.py
115
+ ```
116
+
117
+ 5. Log in with your Hugging Face account
118
+ 6. Click "Run Evaluation & Submit All Answers" to test the agent
119
+
120
+ ## Testing
121
+
122
+ The project includes comprehensive unit tests for all components:
123
+
124
+ - **Agent Tests**: Test the GAIAAgent class and its workflow
125
+ - **Tools Tests**: Test the web, reasoning, and multimodal tools
126
+ - **Memory Tests**: Test the memory components and Supabase integration
127
+ - **Utils Tests**: Test utility functions for parsing and formatting
128
+ - **Integration Tests**: Test integration with external APIs like OpenAI and Perplexity
129
+
130
+ To run all tests:
131
+ ```bash
132
+ python -m unittest discover tests
133
+ ```
134
+
135
+ To run specific test files:
136
+ ```bash
137
+ python -m unittest tests.test_agent
138
+ python -m unittest tests.test_tools
139
+ python -m unittest tests.test_memory
140
+ python -m unittest tests.test_utils
141
+ ```
142
+
143
+ ### OpenAI Integration Testing
144
+
145
+ The project includes specialized tests for OpenAI integration with newer models:
146
+
147
+ ```bash
148
+ # Test OpenAI integration with newer models
149
+ python tests/test_openai_integration.py
150
+
151
+ # Test OpenAI integration with GAIA agent
152
+ python tests/test_openai_integration.py --test-agent
153
+
154
+ # Test GAIA agent with multiple AI providers (OpenAI and Perplexity)
155
+ python tests/test_gaia_multi_provider.py
156
+ ```
157
+
158
+ For convenience, you can run all OpenAI integration tests at once:
159
+
160
+ ```bash
161
+ # Run all OpenAI integration tests
162
+ python tests/run_openai_tests.py
163
+
164
+ # Run all tests with verbose logging
165
+ python tests/run_openai_tests.py --verbose
166
+
167
+ # Run all tests and save results to a custom directory
168
+ python tests/run_openai_tests.py --output my_test_results
169
+
170
+ # Stop running tests after the first failure
171
+ python tests/run_openai_tests.py --stop-on-failure
172
+ ```
173
+
174
+ These tests verify that:
175
+ - The OpenAI API connection works correctly
176
+ - Newer models like GPT-4o can be used successfully
177
+ - The GAIA agent can use both OpenAI and Perplexity tools together
178
+
179
+ ### API Integration Testing
180
+
181
+ To test all API integrations together:
182
+
183
+ ```bash
184
+ # Run all API integration tests
185
+ python tests/run_all_apis_integration_test.py
186
+
187
+ # Run with verbose logging
188
+ python tests/run_all_apis_integration_test.py --verbose
189
+
190
+ # Specify a custom output file
191
+ python tests/run_all_apis_integration_test.py --report-file my_report.json
192
+ ```
193
+
194
+ ### Supabase Integration Testing
195
+
196
+ To test the Supabase memory integration:
197
+
198
+ ```bash
199
+ # Run all Supabase integration tests
200
+ python tests/run_supabase_integration_tests.py
201
+
202
+ # Run with verbose logging
203
+ python tests/run_supabase_integration_tests.py --verbose
204
+
205
+ # Specify a custom output directory
206
+ python tests/run_supabase_integration_tests.py --output-dir custom_docs
207
+ ```
208
+
209
+ ## Evaluation
210
+
211
+ The agent is evaluated on its ability to provide exact matches to the ground truth answers for the GAIA benchmark questions. The submission process includes:
212
+
213
+ - Fetching questions from the course API
214
+ - Running the agent on each question
215
+ - Submitting answers for scoring
216
+ - Viewing results on the leaderboard
217
+
218
+ ## Benchmark Testing
219
+
220
+ The project includes a benchmark testing framework to evaluate the agent's performance:
221
+
222
+ ```bash
223
+ # Run benchmark tests with default settings
224
+ python -m tests.benchmark --output benchmark_results.json
225
+
226
+ # Run benchmark tests with custom questions
227
+ python -m tests.benchmark --questions custom_questions.json --output benchmark_results.json
228
+
229
+ # Run benchmark tests with a specific number of questions
230
+ python -m tests.benchmark --max-questions 5 --output benchmark_results.json
231
+ ```
232
+
233
+ The benchmark testing framework:
234
+ - Tests the agent against sample questions
235
+ - Measures performance metrics (accuracy, latency, resource usage)
236
+ - Generates detailed reports of benchmark results
237
+ - Allows comparison of different agent configurations
238
+
239
+ ## Troubleshooting
240
+
241
+ ### Common Issues
242
+
243
+ #### 1. langchain_openai Dependency Issues
244
+
245
+ If you encounter errors related to the `langchain_openai` package, such as:
246
+ - `ImportError: No module named 'langchain_openai'`
247
+ - `AttributeError: module 'langchain_openai' has no attribute 'ChatOpenAI'`
248
+
249
+ Try the following solutions:
250
+ - Ensure you have the latest version installed: `pip install --upgrade langchain_openai`
251
+ - Make sure you're using the correct import: `from langchain_openai import ChatOpenAI`
252
+ - Check for conflicts with older langchain versions: `pip uninstall langchain-openai` (note the hyphen)
253
+
254
+ #### 2. API Connection Issues
255
+
256
+ If you're having trouble connecting to external APIs:
257
+ - Verify that your API keys are correctly set in the environment variables
258
+ - Check that the API service is not experiencing downtime
259
+ - Ensure you have a stable internet connection
260
+ - Check for rate limiting or quota issues
261
+
262
+ #### 3. Supabase Connection Issues
263
+
264
+ If you're having trouble with Supabase:
265
+ - Verify that your Supabase URL and API key are correctly set
266
+ - Ensure that the required tables exist in your Supabase project
267
+ - Check that your API key has the necessary permissions
268
+ - Verify that your Supabase project is active
269
+
270
+ #### 4. Memory Issues
271
+
272
+ If you're experiencing issues with the agent's memory:
273
+ - Check that memory is enabled in the configuration
274
+ - Verify that the table names are correctly set
275
+ - Ensure that the TTL and cache size settings are appropriate
276
+ - Check for errors in the Supabase logs
277
+
278
+ ## Documentation
279
+
280
+ Comprehensive documentation is available in the `docs/` directory:
281
+
282
+ - [Usage Examples](docs/examples.md): Examples of how to use the GAIA agent in different scenarios
283
+ - [Configuration Guide](docs/configuration.md): Detailed information about configuration options
284
+ - [Deployment Guide](docs/deployment.md): Instructions for deploying the agent in various environments
285
+ - [OpenAI Integration Testing](docs/openai_integration_testing.md): Guide for testing OpenAI integration with newer models
286
+ - [Connection Verification](docs/connection_verification.md): Guide for verifying connections to external services
287
+ - [Supabase Memory Integration](docs/supabase_memory_integration.md): Guide for setting up and using Supabase memory
288
+ - [All APIs Integration Testing](docs/all_apis_integration_testing.md): Guide for testing all API integrations together
289
+ - [Multi-Search Integration](docs/multi_search_integration.md): Guide for using multiple search providers
290
+ - [GAIA Memory Integration](docs/gaia_memory_integration.md): Guide for using memory in the GAIA agent
291
+
292
+ ## License
293
+
294
+ This project is part of the Hugging Face Agents Course and follows its licensing terms.
SUPABASE_TABLES_README.md ADDED
@@ -0,0 +1,122 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Supabase Tables for GAIA Agent
2
+
3
+ This document explains the issue with Supabase tables in the GAIA agent and how to fix it.
4
+
5
+ ## Issue
6
+
7
+ The GAIA agent uses multiple Supabase tables for different types of memory:
8
+
9
+ 1. `gaia_memory`: Main memory table for storing key-value pairs
10
+ 2. `gaia_memory_working`: Working memory table for storing intermediate results during reasoning
11
+ 3. `gaia_memory_conversations`: Conversation history table for storing message history
12
+ 4. `gaia_memory_cache`: Cache table for storing frequently accessed results
13
+
14
+ However, only the main memory table (`gaia_memory`) and the test connection table are being populated, while the other tables remain empty. This is because the specialized memory classes (`WorkingMemory`, `ConversationMemory`, `ResultCache`) set their own table names but when calling methods on the base memory object, they were always using the base memory's table.
15
+
16
+ ## Fix
17
+
18
+ We've created two fix scripts:
19
+
20
+ 1. `memory/fix_supabase_tables.py`: A comprehensive fix that modifies all memory classes to use their respective tables
21
+ 2. `memory/simplified_memory_fix.py`: A simplified fix that focuses only on the working memory table
22
+
23
+ The fix modifies the `SupabaseMemory` class to accept a `table_name` parameter in its methods, and updates the specialized memory classes to pass their table name when calling the base memory methods.
24
+
25
+ ## How to Use the Fix
26
+
27
+ ### Option 1: Run the GAIA agent with the fix applied
28
+
29
+ ```bash
30
+ python run_gaia_with_fix.py --question "Your question here"
31
+ ```
32
+
33
+ This script applies the fix, runs the GAIA agent with your question, and shows the content of all tables after the question is answered.
34
+
35
+ ### Option 2: Apply the fix manually
36
+
37
+ ```bash
38
+ # For all tables
39
+ python -m memory.fix_supabase_tables
40
+
41
+ # For working memory only
42
+ python -m memory.simplified_memory_fix
43
+ ```
44
+
45
+ ### Option 3: Run tests
46
+
47
+ ```bash
48
+ # Simple test for working memory
49
+ python -m tests.test_supabase_tables_simple
50
+
51
+ # Comprehensive test for all tables
52
+ python -m tests.test_gaia_with_all_tables
53
+ ```
54
+
55
+ ## Table Descriptions
56
+
57
+ ### gaia_memory
58
+
59
+ The main memory table for storing key-value pairs. This is the default table used by the `SupabaseMemory` class.
60
+
61
+ ### gaia_memory_working
62
+
63
+ The working memory table for storing intermediate results during reasoning. This table is used by the `WorkingMemory` class to store temporary data during the reasoning process, such as:
64
+
65
+ - Parsed question
66
+ - Identified entities
67
+ - Intermediate reasoning steps
68
+ - Search results
69
+ - Final answer
70
+
71
+ ### gaia_memory_conversations
72
+
73
+ The conversation history table for storing message history. This table is used by the `ConversationMemory` class to store the conversation history, including:
74
+
75
+ - User messages
76
+ - Assistant messages
77
+ - Timestamps
78
+ - Metadata
79
+
80
+ ### gaia_memory_cache
81
+
82
+ The cache table for storing frequently accessed results. This table is used by the `ResultCache` class to store cached results, which can improve performance by avoiding redundant computations.
83
+
84
+ ## Performance Benefits
85
+
86
+ Using the specialized tables provides several benefits:
87
+
88
+ 1. **Improved organization**: Each type of memory has its own table, making it easier to manage and query.
89
+ 2. **Better performance**: Smaller, specialized tables are faster to query than a single large table.
90
+ 3. **Enhanced debugging**: It's easier to debug issues when each memory type has its own table.
91
+ 4. **Reduced data conflicts**: Separate tables reduce the risk of key conflicts between different memory types.
92
+
93
+ ## Implementation Details
94
+
95
+ The fix works by monkey-patching the following methods:
96
+
97
+ ### SupabaseMemory
98
+
99
+ - `store`: Modified to accept a `table_name` parameter
100
+ - `retrieve`: Modified to accept a `table_name` parameter
101
+ - `delete`: Modified to accept a `table_name` parameter
102
+ - `list_keys`: Modified to accept a `table_name` parameter
103
+
104
+ ### WorkingMemory
105
+
106
+ - `store_intermediate_result`: Modified to use the working memory table
107
+ - `get_intermediate_result`: Modified to use the working memory table
108
+ - `list_intermediate_results`: Modified to use the working memory table
109
+ - `clear`: Modified to use the working memory table
110
+
111
+ ### ConversationMemory
112
+
113
+ - `add_message`: Modified to use the conversations table
114
+ - `get_messages`: Modified to use the conversations table
115
+ - `clear`: Modified to use the conversations table
116
+
117
+ ### ResultCache
118
+
119
+ - `cache_result`: Modified to use the cache table
120
+ - `get_result`: Modified to use the cache table
121
+ - `invalidate`: Modified to use the cache table
122
+ - `clear`: Modified to use the cache table
SUPABASE_TEST_README.md ADDED
@@ -0,0 +1,104 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Standalone Supabase Connection Test
2
+
3
+ This directory contains scripts for testing the Supabase connection without relying on any project modules.
4
+
5
+ ## Main Script: `supabase_direct_test.py`
6
+
7
+ This script tests the connection to Supabase, verifies access to all required tables, and performs basic CRUD operations on the test_connection table.
8
+
9
+ ### Prerequisites
10
+
11
+ - Python 3.6+
12
+ - Required packages:
13
+ - supabase
14
+ - python-dotenv
15
+
16
+ You can install the required packages with:
17
+
18
+ ```bash
19
+ pip install supabase python-dotenv
20
+ ```
21
+
22
+ ### Usage
23
+
24
+ You can run the script in several ways:
25
+
26
+ 1. Using environment variables:
27
+
28
+ ```bash
29
+ # Set environment variables
30
+ export SUPABASE_URL=https://your-project-id.supabase.co
31
+ export SUPABASE_KEY=your-supabase-api-key
32
+
33
+ # Run the test
34
+ python supabase_direct_test.py
35
+ ```
36
+
37
+ 2. Using command-line arguments:
38
+
39
+ ```bash
40
+ python supabase_direct_test.py --key your-supabase-api-key
41
+ ```
42
+
43
+ 3. Using a .env file:
44
+
45
+ Create a `.env` file in the same directory with the following content:
46
+
47
+ ```
48
+ SUPABASE_URL=https://your-project-id.supabase.co
49
+ SUPABASE_KEY=your-supabase-api-key
50
+ ```
51
+
52
+ Then run the script:
53
+
54
+ ```bash
55
+ python supabase_direct_test.py
56
+ ```
57
+
58
+ ### Options
59
+
60
+ - `--verbose`: Enable verbose logging
61
+ - `--url`: Specify the Supabase URL (overrides environment variable)
62
+ - `--key`: Specify the Supabase API key (overrides environment variable)
63
+
64
+ Example with verbose logging:
65
+
66
+ ```bash
67
+ python supabase_direct_test.py --verbose --key your-supabase-api-key
68
+ ```
69
+
70
+ ## Helper Scripts
71
+
72
+ The repository also includes several helper scripts for troubleshooting:
73
+
74
+ - `check_supabase_env.py`: Checks if the required Supabase environment variables are set
75
+ - `supabase_import_test.py`: Tests importing the Supabase library
76
+ - `supabase_init_test.py`: Tests initializing the Supabase client
77
+ - `supabase_minimal_test.py`: Minimal test for Supabase connection
78
+ - `create_env_and_test.py`: Creates a .env file and tests the connection
79
+
80
+ ## Tables Tested
81
+
82
+ The script verifies connection to the following tables:
83
+
84
+ 1. `gaia_memory`: Main memory table
85
+ 2. `gaia_memory_conversations`: Conversation history table
86
+ 3. `gaia_memory_cache`: Cache table
87
+ 4. `gaia_memory_working`: Working memory table
88
+ 5. `test_connection`: Test table for connection verification
89
+
90
+ If any of these tables don't exist, you can create them by running the SQL statements in `supabase_tables.sql`.
91
+
92
+ ## Troubleshooting
93
+
94
+ If you encounter issues:
95
+
96
+ 1. Check that your Supabase API key is correct
97
+ 2. Verify that the Supabase URL is correct
98
+ 3. Ensure that the required tables exist in your Supabase project
99
+ 4. Check network connectivity to Supabase
100
+ 5. Run with the `--verbose` flag for more detailed error information
101
+
102
+ ## Security Note
103
+
104
+ Never hardcode your Supabase API key in your scripts. Always use environment variables, .env files, or command-line arguments to provide the key securely.
agent/__init__.py ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ Agent module for GAIA implementation.
3
+ Contains the core agent architecture and configuration.
4
+ """
5
+
6
+ from agent.agent import GAIAAgent
7
+ from agent.config import (
8
+ get_logging_config,
9
+ get_model_config,
10
+ get_tool_config,
11
+ get_memory_config,
12
+ get_agent_config
13
+ )
14
+
15
+ __all__ = [
16
+ 'GAIAAgent',
17
+ 'get_logging_config',
18
+ 'get_model_config',
19
+ 'get_tool_config',
20
+ 'get_memory_config',
21
+ 'get_agent_config'
22
+ ]
agent/__pycache__/__init__.cpython-312.pyc ADDED
Binary file (635 Bytes). View file
 
agent/__pycache__/__init__.cpython-313.pyc ADDED
Binary file (635 Bytes). View file
 
agent/__pycache__/agent.cpython-312.pyc ADDED
Binary file (13.2 kB). View file
 
agent/__pycache__/agent.cpython-313.pyc ADDED
Binary file (12.8 kB). View file
 
agent/__pycache__/config.cpython-312.pyc ADDED
Binary file (6.18 kB). View file
 
agent/__pycache__/config.cpython-313.pyc ADDED
Binary file (6.09 kB). View file
 
agent/__pycache__/graph.cpython-312.pyc ADDED
Binary file (16.8 kB). View file
 
agent/__pycache__/graph.cpython-313.pyc ADDED
Binary file (16.8 kB). View file
 
agent/__pycache__/tool_registry.cpython-313.pyc ADDED
Binary file (14.3 kB). View file
 
agent/agent.py ADDED
@@ -0,0 +1,291 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ Main agent implementation for the GAIA benchmark.
3
+
4
+ This module contains the GAIAAgent class which is responsible for:
5
+ - Processing questions from the GAIA benchmark
6
+ - Selecting and executing appropriate tools
7
+ - Formulating precise answers
8
+ - Handling errors and logging
9
+
10
+ The agent uses LangGraph for workflow management and Supabase for memory,
11
+ with configuration from the config module. It can be extended with
12
+ additional capabilities as needed.
13
+ """
14
+
15
+ import logging
16
+ import time
17
+ import traceback
18
+ import uuid
19
+ from typing import Dict, Any, Optional, List, Union, Tuple
20
+
21
+ from agent.config import (
22
+ get_logging_config,
23
+ get_model_config,
24
+ get_tool_config,
25
+ get_memory_config,
26
+ get_agent_config,
27
+ VERBOSE
28
+ )
29
+
30
+ # Import LangGraph workflow
31
+ from agent.graph import run_agent_graph
32
+
33
+ from memory import SupabaseMemory, ConversationMemory, ResultCache, WorkingMemory
34
+
35
+ logging_config = get_logging_config()
36
+ logging.basicConfig(
37
+ level=logging_config["level"],
38
+ format=logging_config["format"],
39
+ filename=logging_config["filename"]
40
+ )
41
+ logger = logging.getLogger("gaia_agent")
42
+
43
+ class GAIAAgent:
44
+ """
45
+ Agent for answering questions from the GAIA benchmark.
46
+
47
+ This agent processes questions, selects appropriate tools,
48
+ executes a reasoning process, and formulates precise answers.
49
+ """
50
+
51
+ def __init__(self, config: Optional[Dict[str, Any]] = None):
52
+ """
53
+ Initialize the GAIA agent with configuration.
54
+
55
+ Args:
56
+ config (Optional[Dict[str, Any]]): Custom configuration dictionary.
57
+ If None, the default configuration from config.py is used.
58
+ """
59
+ logger.info("Initializing GAIAAgent")
60
+
61
+ self.config = config or {
62
+ "model": get_model_config(),
63
+ "tools": get_tool_config(),
64
+ "memory": get_memory_config(),
65
+ "agent": get_agent_config()
66
+ }
67
+
68
+ if VERBOSE:
69
+ logger.setLevel(logging.DEBUG)
70
+ if not any(isinstance(h, logging.StreamHandler) for h in logger.handlers):
71
+ console_handler = logging.StreamHandler()
72
+ console_handler.setLevel(logging.DEBUG)
73
+ console_handler.setFormatter(logging.Formatter(logging_config["format"]))
74
+ logger.addHandler(console_handler)
75
+
76
+ self._initialize_memory()
77
+
78
+ self.reset_state()
79
+
80
+ logger.info("GAIAAgent initialized successfully")
81
+
82
+ def _initialize_memory(self) -> None:
83
+ """Initialize memory components."""
84
+
85
+ try:
86
+ self.memory = SupabaseMemory(self.config["memory"])
87
+
88
+ self.session_id = str(uuid.uuid4())
89
+
90
+ self.conversation_memory = ConversationMemory(self.memory, self.session_id)
91
+ self.result_cache = ResultCache(self.memory)
92
+ self.working_memory = WorkingMemory(self.memory, self.session_id)
93
+
94
+ except Exception as e:
95
+ logger.error(f"Error initializing memory components: {str(e)}")
96
+ logger.warning("Continuing without persistent memory")
97
+
98
+ self.memory = None
99
+ self.conversation_memory = None
100
+ self.result_cache = None
101
+ self.working_memory = None
102
+
103
+ def reset_state(self) -> None:
104
+ """Reset the agent's internal state."""
105
+ self.state = {
106
+ "current_question": None,
107
+ "tools_used": [],
108
+ "intermediate_results": [],
109
+ "start_time": None,
110
+ "end_time": None,
111
+ "error": None
112
+ }
113
+
114
+ def __call__(self, question: str) -> str:
115
+ """
116
+ Process a question and return an answer.
117
+
118
+ This is the main entry point for the agent. It takes a question,
119
+ processes it, and returns a concise answer.
120
+
121
+ Args:
122
+ question (str): The question to answer
123
+
124
+ Returns:
125
+ str: The answer to the question
126
+
127
+ Raises:
128
+ Exception: If an error occurs during processing
129
+ """
130
+ self.reset_state()
131
+ self.state["current_question"] = question
132
+ self.state["start_time"] = time.time()
133
+
134
+ logger.info(f"Processing question: {question[:100]}...")
135
+
136
+ try:
137
+ max_retries = self.config["agent"]["max_retries"]
138
+ retry_delay = self.config["agent"]["retry_delay"]
139
+
140
+ for attempt in range(max_retries):
141
+ try:
142
+ answer = self._process_question(question)
143
+
144
+ self.state["end_time"] = time.time()
145
+ processing_time = self.state["end_time"] - self.state["start_time"]
146
+ logger.info(f"Question processed in {processing_time:.2f} seconds. Answer: {answer[:100]}...")
147
+
148
+ return answer
149
+
150
+ except Exception as e:
151
+ if attempt < max_retries - 1:
152
+ logger.warning(f"Attempt {attempt + 1}/{max_retries} failed: {str(e)}. Retrying in {retry_delay} seconds...")
153
+ time.sleep(retry_delay)
154
+ else:
155
+ raise
156
+
157
+ except Exception as e:
158
+ self.state["error"] = str(e)
159
+ self.state["end_time"] = time.time()
160
+ processing_time = self.state["end_time"] - self.state["start_time"]
161
+
162
+ logger.error(f"Error processing question after {processing_time:.2f} seconds: {str(e)}")
163
+ logger.error(traceback.format_exc())
164
+
165
+ if self.config["agent"].get("return_error_message", False):
166
+ return f"Error: {str(e)}"
167
+ else:
168
+ raise
169
+
170
+ def _process_question(self, question: str) -> str:
171
+ """
172
+ Internal method to process a question and generate an answer.
173
+
174
+ This method implements the core logic of the agent using LangGraph:
175
+ 1. Analyze the question
176
+ 2. Plan the approach
177
+ 3. Execute tools
178
+ 4. Formulate the answer
179
+
180
+ Args:
181
+ question (str): The question to process
182
+
183
+ Returns:
184
+ str: The answer to the question
185
+ """
186
+
187
+ if self.result_cache:
188
+ import hashlib
189
+ cache_key = f"answer:{hashlib.md5(question.encode()).hexdigest()}"
190
+
191
+ cached_answer = self.result_cache.get_result(cache_key)
192
+ if cached_answer:
193
+ logger.info(f"Using cached answer for question: {question[:30]}...")
194
+ return cached_answer
195
+
196
+ graph_config = {
197
+ "recursion_limit": self.config["agent"].get("max_iterations", 10),
198
+ "timeout": self.config["agent"].get("timeout", 300)
199
+ }
200
+
201
+ try:
202
+ result = run_agent_graph(question, graph_config)
203
+
204
+ if result.get("answer"):
205
+ answer = result["answer"]
206
+ elif result.get("error"):
207
+ raise Exception(result["error"])
208
+ else:
209
+ answer = f"Unable to answer: {question[:30]}..."
210
+
211
+ if self.result_cache:
212
+ self.result_cache.cache_result(cache_key, answer)
213
+
214
+ if self.working_memory:
215
+ if result.get("analysis"):
216
+ self.working_memory.store_intermediate_result(
217
+ "analysis", result["analysis"]
218
+ )
219
+
220
+ if result.get("plan"):
221
+ self.working_memory.store_intermediate_result(
222
+ "plan", result["plan"]
223
+ )
224
+
225
+ if result.get("tool_results"):
226
+ self.working_memory.store_intermediate_result(
227
+ "tool_results", result["tool_results"]
228
+ )
229
+
230
+ if self.conversation_memory:
231
+ self.conversation_memory.add_message("user", question)
232
+ self.conversation_memory.add_message("assistant", answer)
233
+
234
+ return answer
235
+ except Exception as e:
236
+ logger.error(f"Error running LangGraph workflow: {str(e)}")
237
+ raise
238
+
239
+ def get_memory_snapshot(self) -> Dict[str, Any]:
240
+ """
241
+ Get a snapshot of the agent's memory.
242
+
243
+ This method returns a dictionary containing the current state of the
244
+ agent's memory, including conversation history, cached results, and
245
+ intermediate results.
246
+
247
+ Returns:
248
+ Dict[str, Any]: Memory snapshot
249
+ """
250
+ snapshot = {
251
+ "conversation": [],
252
+ "cached_results": {},
253
+ "intermediate_results": {}
254
+ }
255
+
256
+ if self.conversation_memory:
257
+ snapshot["conversation"] = self.conversation_memory.get_messages()
258
+
259
+ if self.result_cache and self.memory:
260
+ cache_keys = self.memory.list_keys("cache:")
261
+ for key in cache_keys:
262
+ original_key = key[6:] # Remove "cache:" prefix
263
+
264
+ result = self.memory.retrieve(key)
265
+
266
+ snapshot["cached_results"][original_key] = result
267
+
268
+ if self.working_memory:
269
+ snapshot["intermediate_results"] = self.working_memory.list_intermediate_results()
270
+
271
+ return snapshot
272
+
273
+ def clear_memory(self) -> None:
274
+ """
275
+ Clear the agent's memory.
276
+
277
+ This method clears all memory components, including conversation history,
278
+ cached results, and intermediate results.
279
+ """
280
+ logger.info("Clearing agent memory")
281
+
282
+ if self.conversation_memory:
283
+ self.conversation_memory.clear()
284
+
285
+ if self.result_cache:
286
+ self.result_cache.clear()
287
+
288
+ if self.working_memory:
289
+ self.working_memory.clear()
290
+
291
+ logger.info("Agent memory cleared")
agent/config.py ADDED
@@ -0,0 +1,186 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ Configuration module for the GAIA agent.
3
+
4
+ This module centralizes all configuration settings for the agent, including:
5
+ - API endpoints
6
+ - Model selection
7
+ - Tool parameters
8
+ - Memory settings
9
+ - Logging configuration
10
+ - Web browsing settings
11
+ - Multimodal processing parameters
12
+
13
+ It uses python-dotenv to load environment variables for sensitive credentials.
14
+ """
15
+
16
+ import os
17
+ import logging
18
+ from typing import Dict, Any, Optional, List, Union, Tuple
19
+
20
+ try:
21
+ from dotenv import load_dotenv
22
+ load_dotenv(verbose=False)
23
+ except ImportError:
24
+ def load_dotenv(*args, **kwargs):
25
+
26
+ OPENAI_API_KEY = os.getenv("OPENAI_API_KEY", "")
27
+ SERPER_API_KEY = os.getenv("SERPER_API_KEY", "")
28
+ PERPLEXITY_API_KEY = os.getenv("PERPLEXITY_API_KEY", "")
29
+ SUPABASE_KEY = os.getenv("SUPABASE_KEY", "")
30
+ HF_TOKEN = os.getenv("HF_TOKEN", "")
31
+
32
+ # Supabase Configuration (partially hardcoded)
33
+ SUPABASE_URL = "https://tjamxhvvtnypbadvrkjq.supabase.co"
34
+ SUPABASE_PROJECT_ID = "tjamxhvvtnypbadvrkjq"
35
+ SUPABASE_PROJECT_NAME = "HF_GAIA_Assessment_AGT"
36
+
37
+ OPENAI_API_BASE = "https://api.openai.com/v1"
38
+ SERPER_API_URL = "https://google.serper.dev/search"
39
+ PERPLEXITY_API_URL = "https://api.perplexity.ai"
40
+ HUGGINGFACE_API_URL = "https://api-inference.huggingface.co/models"
41
+
42
+ # Model Configuration (hardcoded with updated models)
43
+ DEFAULT_MODEL = "gpt-4o" # Updated from gpt-4-turbo
44
+ EMBEDDING_MODEL = "text-embedding-3-large" # Updated from text-embedding-ada-002
45
+ FALLBACK_MODEL = "gpt-3.5-turbo"
46
+ VISION_MODEL = "gpt-4o" # Updated from gpt-4-vision-preview
47
+ REASONING_MODEL = "gpt-4o" # Updated from gpt-4-turbo
48
+ MODEL_TEMPERATURE = 0.1
49
+ MODEL_MAX_TOKENS = 4096
50
+
51
+ # Tool Parameters (hardcoded)
52
+ WEB_SEARCH_RESULT_COUNT = 5
53
+ WEB_SEARCH_TIMEOUT = 10
54
+ ARXIV_MAX_RESULTS = 3
55
+ DUCKDUCKGO_TIMEOUT = 10
56
+ DUCKDUCKGO_MAX_RESULTS = 5
57
+ PERPLEXITY_TIMEOUT = 30
58
+ PERPLEXITY_MODEL = "sonar-reasoning"
59
+ WEB_SCRAPING_TIMEOUT = 15
60
+ WEB_SCRAPING_MAX_LINKS = 3
61
+ WEB_SCRAPING_MAX_CONTENT_LENGTH = 10000
62
+
63
+ # Memory Settings (hardcoded)
64
+ MEMORY_ENABLED = True
65
+ MEMORY_TABLE_NAME = "gaia_memory"
66
+ MEMORY_TTL = 86400 # 24 hours in seconds
67
+ MEMORY_CACHE_SIZE = 100
68
+ MEMORY_SIMILARITY_THRESHOLD = 0.85
69
+ MEMORY_MAX_ENTRIES = 1000
70
+
71
+ # Agent Settings (hardcoded)
72
+ MAX_ITERATIONS = 10
73
+ VERBOSE = False
74
+ TIMEOUT = 300 # 5 minutes in seconds
75
+ MAX_RETRIES = 3
76
+ RETRY_DELAY = 2 # seconds
77
+ PARALLEL_TOOL_EXECUTION = False
78
+
79
+ # Logging Configuration (hardcoded)
80
+ LOG_LEVEL = "INFO"
81
+ LOG_FILE = "gaia_agent.log"
82
+ LOG_FORMAT = "%(asctime)s - %(name)s - %(levelname)s - %(message)s"
83
+
84
+ # Multimodal Settings (hardcoded)
85
+ IMAGE_MAX_SIZE = 1024 # pixels
86
+ IMAGE_FORMAT = "JPEG"
87
+ VISION_DETAIL_LEVEL = "high" # low, medium, high
88
+
89
+ # Web Browsing Settings (hardcoded)
90
+ USER_AGENT = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36"
91
+ BROWSER_TIMEOUT = 30
92
+ BROWSER_HEADLESS = True
93
+ BROWSER_IMAGES_ENABLED = True
94
+
95
+ def get_logging_config() -> Dict[str, Any]:
96
+ """
97
+ Returns a dictionary with logging configuration.
98
+
99
+ Returns:
100
+ Dict[str, Any]: Logging configuration dictionary
101
+ """
102
+ return {
103
+ "level": getattr(logging, LOG_LEVEL.upper(), logging.INFO),
104
+ "format": LOG_FORMAT,
105
+ "filename": LOG_FILE if LOG_FILE else None,
106
+ }
107
+
108
+ def get_model_config() -> Dict[str, Any]:
109
+ """
110
+ Returns a dictionary with model configuration.
111
+
112
+ Returns:
113
+ Dict[str, Any]: Model configuration dictionary
114
+ """
115
+ return {
116
+ "default_model": DEFAULT_MODEL,
117
+ "embedding_model": EMBEDDING_MODEL,
118
+ "fallback_model": FALLBACK_MODEL,
119
+ "vision_model": VISION_MODEL,
120
+ "reasoning_model": REASONING_MODEL,
121
+ "temperature": MODEL_TEMPERATURE,
122
+ "max_tokens": MODEL_MAX_TOKENS,
123
+ }
124
+
125
+ def get_tool_config() -> Dict[str, Dict[str, Any]]:
126
+ """
127
+ Returns a dictionary with tool configurations.
128
+
129
+ Returns:
130
+ Dict[str, Dict[str, Any]]: Tool configuration dictionary
131
+ """
132
+ return {
133
+ "web_search": {
134
+ "result_count": WEB_SEARCH_RESULT_COUNT,
135
+ "timeout": WEB_SEARCH_TIMEOUT,
136
+ },
137
+ "arxiv": {
138
+ "max_results": ARXIV_MAX_RESULTS,
139
+ },
140
+ "duckduckgo": {
141
+ "timeout": DUCKDUCKGO_TIMEOUT,
142
+ "max_results": DUCKDUCKGO_MAX_RESULTS,
143
+ },
144
+ "perplexity": {
145
+ "timeout": PERPLEXITY_TIMEOUT,
146
+ "model": PERPLEXITY_MODEL,
147
+ "api_url": PERPLEXITY_API_URL,
148
+ },
149
+ "web_scraping": {
150
+ "timeout": WEB_SCRAPING_TIMEOUT,
151
+ "max_links": WEB_SCRAPING_MAX_LINKS,
152
+ "max_content_length": WEB_SCRAPING_MAX_CONTENT_LENGTH,
153
+ },
154
+ }
155
+
156
+ def get_memory_config() -> Dict[str, Any]:
157
+ """
158
+ Returns a dictionary with memory configuration.
159
+
160
+ Returns:
161
+ Dict[str, Any]: Memory configuration dictionary
162
+ """
163
+ return {
164
+ "enabled": MEMORY_ENABLED,
165
+ "table_name": MEMORY_TABLE_NAME,
166
+ "ttl": MEMORY_TTL,
167
+ "cache_size": MEMORY_CACHE_SIZE,
168
+ "similarity_threshold": MEMORY_SIMILARITY_THRESHOLD,
169
+ "max_entries": MEMORY_MAX_ENTRIES,
170
+ }
171
+
172
+ def get_agent_config() -> Dict[str, Any]:
173
+ """
174
+ Returns a dictionary with agent configuration.
175
+
176
+ Returns:
177
+ Dict[str, Any]: Agent configuration dictionary
178
+ """
179
+ return {
180
+ "max_iterations": MAX_ITERATIONS,
181
+ "verbose": VERBOSE,
182
+ "timeout": TIMEOUT,
183
+ "max_retries": MAX_RETRIES,
184
+ "retry_delay": RETRY_DELAY,
185
+ "parallel_tool_execution": PARALLEL_TOOL_EXECUTION,
186
+ }
agent/graph.py ADDED
@@ -0,0 +1,452 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ LangGraph workflow definition for the GAIA agent.
3
+
4
+ This module defines the workflow graph for the GAIA agent using LangGraph.
5
+ It implements nodes for question analysis, planning, tool selection,
6
+ tool execution, and answer formulation, as well as the edges between them.
7
+
8
+ The graph is designed to be integrated with the GAIAAgent class and can be
9
+ extended with additional nodes and edges as needed.
10
+ """
11
+
12
+ import logging
13
+ from typing import Dict, Any, List, Tuple, Optional, Annotated, TypedDict, Literal, Union, cast
14
+
15
+ from langchain.prompts import PromptTemplate
16
+ from langchain_core.messages import HumanMessage, AIMessage
17
+ from langchain_core.runnables import RunnableConfig
18
+ from langchain_core.output_parsers import StrOutputParser
19
+ from langchain_openai import ChatOpenAI
20
+
21
+ from langgraph.graph import StateGraph, END
22
+ from langgraph.prebuilt import ToolNode
23
+
24
+ from agent.config import get_model_config, get_agent_config, VERBOSE
25
+
26
+ logger = logging.getLogger("gaia_agent.graph")
27
+
28
+ class QuestionAnalysis(TypedDict):
29
+ """Type for question analysis results."""
30
+ question_type: str
31
+ required_tools: List[str]
32
+ information_sources: List[str]
33
+ complexity: str
34
+
35
+ class PlanStep(TypedDict):
36
+ """Type for a single step in the plan."""
37
+ step_number: int
38
+ description: str
39
+ tool: Optional[str]
40
+ tool_input: Optional[Dict[str, Any]]
41
+
42
+ class ToolResult(TypedDict):
43
+ """Type for tool execution results."""
44
+ tool_name: str
45
+ success: bool
46
+ result: Any
47
+ error: Optional[str]
48
+
49
+ class AgentState(TypedDict):
50
+ """Type for the agent state."""
51
+ question: str
52
+ analysis: Optional[QuestionAnalysis]
53
+ plan: Optional[List[PlanStep]]
54
+ current_step: int
55
+ tools_to_use: Optional[List[str]]
56
+ tool_results: List[ToolResult]
57
+ intermediate_thoughts: List[str]
58
+ answer: Optional[str]
59
+ error: Optional[str]
60
+
61
+ def analyze_question(state: AgentState) -> AgentState:
62
+ """
63
+ Analyze the question to determine its requirements.
64
+
65
+ Args:
66
+ state: The current agent state
67
+
68
+ Returns:
69
+ Updated agent state with question analysis
70
+ """
71
+
72
+ model_config = get_model_config()
73
+
74
+ analysis_prompt = PromptTemplate.from_template(
75
+ """You are an expert question analyzer for the GAIA benchmark.
76
+
77
+ Your task is to analyze the following question and determine:
78
+ 1. The question type (factual, mathematical, reasoning, etc.)
79
+ 2. The tools required to answer it (web_search, calculator, reasoning, etc.)
80
+ 3. The information sources needed (web, knowledge base, computation, etc.)
81
+ 4. The complexity level (simple, medium, complex)
82
+
83
+ Question: {question}
84
+
85
+ Provide your analysis in the following JSON format:
86
+ {{
87
+ "question_type": "type of question",
88
+ "required_tools": ["tool1", "tool2", ...],
89
+ "information_sources": ["source1", "source2", ...],
90
+ "complexity": "complexity level"
91
+ }}
92
+
93
+ Analysis:"""
94
+ )
95
+
96
+ model = ChatOpenAI(
97
+ model=model_config["reasoning_model"],
98
+ temperature=model_config["temperature"],
99
+ max_tokens=model_config["max_tokens"]
100
+ )
101
+
102
+ chain = analysis_prompt | model | StrOutputParser()
103
+
104
+ try:
105
+ analysis_str = chain.invoke({"question": state["question"]})
106
+
107
+ import json
108
+ analysis = json.loads(analysis_str)
109
+
110
+ new_state = state.copy()
111
+ new_state["analysis"] = analysis
112
+
113
+
114
+ return new_state
115
+ except Exception as e:
116
+ logger.error(f"Error analyzing question: {str(e)}")
117
+
118
+ new_state = state.copy()
119
+ new_state["error"] = f"Error analyzing question: {str(e)}"
120
+
121
+ return new_state
122
+
123
+ def create_plan(state: AgentState) -> AgentState:
124
+ """
125
+ Create a plan to answer the question based on the analysis.
126
+
127
+ Args:
128
+ state: The current agent state
129
+
130
+ Returns:
131
+ Updated agent state with plan
132
+ """
133
+
134
+ if state.get("error"):
135
+ return state
136
+
137
+ model_config = get_model_config()
138
+
139
+ planning_prompt = PromptTemplate.from_template(
140
+ """You are an expert planner for the GAIA benchmark.
141
+
142
+ Your task is to create a step-by-step plan to answer the following question:
143
+
144
+ Question: {question}
145
+
146
+ Question Analysis:
147
+ {analysis}
148
+
149
+ Create a detailed plan with the following steps:
150
+ 1. Break down the question into logical steps
151
+ 2. For each step, specify which tool to use (if any)
152
+ 3. Specify the inputs for each tool
153
+ 4. Describe how to combine the results to formulate the final answer
154
+
155
+ Provide your plan in the following JSON format:
156
+ [
157
+ {{
158
+ "step_number": 1,
159
+ "description": "description of step",
160
+ "tool": "tool_name or null if no tool needed",
161
+ "tool_input": {{ "param1": "value1", ... }} or null if no tool needed
162
+ }},
163
+ ...
164
+ ]
165
+
166
+ Plan:"""
167
+ )
168
+
169
+ model = ChatOpenAI(
170
+ model=model_config["reasoning_model"],
171
+ temperature=model_config["temperature"],
172
+ max_tokens=model_config["max_tokens"]
173
+ )
174
+
175
+ chain = planning_prompt | model | StrOutputParser()
176
+
177
+ try:
178
+ plan_str = chain.invoke({
179
+ "question": state["question"],
180
+ "analysis": state["analysis"]
181
+ })
182
+
183
+ import json
184
+ plan = json.loads(plan_str)
185
+
186
+ new_state = state.copy()
187
+ new_state["plan"] = plan
188
+ new_state["current_step"] = 0
189
+
190
+ tools_to_use = []
191
+ for step in plan:
192
+ if step.get("tool"):
193
+ tools_to_use.append(step["tool"])
194
+
195
+ new_state["tools_to_use"] = list(set(tools_to_use)) # Remove duplicates
196
+
197
+
198
+ return new_state
199
+ except Exception as e:
200
+ logger.error(f"Error creating plan: {str(e)}")
201
+
202
+ new_state = state.copy()
203
+ new_state["error"] = f"Error creating plan: {str(e)}"
204
+
205
+ return new_state
206
+
207
+ def select_tool(state: AgentState) -> Dict[str, Any]:
208
+ """
209
+ Select the next tool to execute based on the plan.
210
+
211
+ Args:
212
+ state: The current agent state
213
+
214
+ Returns:
215
+ Dictionary with the selected tool and its input
216
+ """
217
+
218
+ if state.get("error"):
219
+ return {"next": END}
220
+
221
+ if state["current_step"] >= len(state["plan"]):
222
+ return {"next": "formulate_answer"}
223
+
224
+ current_step = state["plan"][state["current_step"]]
225
+
226
+ if not current_step.get("tool"):
227
+ new_state = state.copy()
228
+ new_state["current_step"] += 1
229
+
230
+ thought = f"Step {current_step['step_number']}: {current_step['description']} - No tool needed, moving to next step."
231
+ new_state["intermediate_thoughts"] = state.get("intermediate_thoughts", []) + [thought]
232
+
233
+ return {"next": "select_tool", "state": new_state}
234
+
235
+ return {"next": "execute_tool"}
236
+
237
+ def execute_tool(state: AgentState) -> AgentState:
238
+ """
239
+ Execute the selected tool based on the current plan step.
240
+
241
+ Args:
242
+ state: The current agent state
243
+
244
+ Returns:
245
+ Updated agent state with tool execution results
246
+ """
247
+ if state.get("error"):
248
+ return state
249
+
250
+ current_step = state["plan"][state["current_step"]]
251
+ tool_name = current_step.get("tool")
252
+ tool_input = current_step.get("tool_input", {})
253
+
254
+
255
+ try:
256
+ result = f"Simulated result for {tool_name} with input {tool_input}"
257
+
258
+ tool_result = {
259
+ "tool_name": tool_name,
260
+ "success": True,
261
+ "result": result,
262
+ "error": None
263
+ }
264
+
265
+ new_state = state.copy()
266
+ new_state["tool_results"] = state.get("tool_results", []) + [tool_result]
267
+ new_state["current_step"] += 1
268
+
269
+ thought = f"Step {current_step['step_number']}: {current_step['description']} - Executed {tool_name} and got result: {result}"
270
+ new_state["intermediate_thoughts"] = state.get("intermediate_thoughts", []) + [thought]
271
+
272
+
273
+ return new_state
274
+ except Exception as e:
275
+ logger.error(f"Error executing tool {tool_name}: {str(e)}")
276
+
277
+ tool_result = {
278
+ "tool_name": tool_name,
279
+ "success": False,
280
+ "result": None,
281
+ "error": str(e)
282
+ }
283
+
284
+ new_state = state.copy()
285
+ new_state["tool_results"] = state.get("tool_results", []) + [tool_result]
286
+ new_state["current_step"] += 1
287
+
288
+ thought = f"Step {current_step['step_number']}: {current_step['description']} - Failed to execute {tool_name}: {str(e)}"
289
+ new_state["intermediate_thoughts"] = state.get("intermediate_thoughts", []) + [thought]
290
+
291
+ return new_state
292
+
293
+ def formulate_answer(state: AgentState) -> AgentState:
294
+ """
295
+ Formulate the final answer based on tool results.
296
+
297
+ Args:
298
+ state: The current agent state
299
+
300
+ Returns:
301
+ Updated agent state with answer
302
+ """
303
+
304
+ if state.get("error"):
305
+ return state
306
+
307
+ model_config = get_model_config()
308
+
309
+ answer_prompt = PromptTemplate.from_template(
310
+ """You are an expert answer formulator for the GAIA benchmark.
311
+
312
+ Your task is to formulate a precise, concise answer to the following question based on the tool results:
313
+
314
+ Question: {question}
315
+
316
+ Question Analysis:
317
+ {analysis}
318
+
319
+ Plan:
320
+ {plan}
321
+
322
+ Tool Results:
323
+ {tool_results}
324
+
325
+ Intermediate Thoughts:
326
+ {thoughts}
327
+
328
+ Remember, GAIA benchmark answers should be very concise - either a number, a few words, or a comma-separated list.
329
+ Do not provide explanations or additional context in your answer.
330
+
331
+ Concise Answer:"""
332
+ )
333
+
334
+ model = ChatOpenAI(
335
+ model=model_config["reasoning_model"],
336
+ temperature=model_config["temperature"],
337
+ max_tokens=model_config["max_tokens"]
338
+ )
339
+
340
+ chain = answer_prompt | model | StrOutputParser()
341
+
342
+ try:
343
+ answer = chain.invoke({
344
+ "question": state["question"],
345
+ "analysis": state["analysis"],
346
+ "plan": state["plan"],
347
+ "tool_results": state["tool_results"],
348
+ "thoughts": state.get("intermediate_thoughts", [])
349
+ })
350
+
351
+ new_state = state.copy()
352
+ new_state["answer"] = answer.strip()
353
+
354
+
355
+ return new_state
356
+ except Exception as e:
357
+ logger.error(f"Error formulating answer: {str(e)}")
358
+
359
+ new_state = state.copy()
360
+ new_state["error"] = f"Error formulating answer: {str(e)}"
361
+
362
+ return new_state
363
+
364
+ def should_retry(state: AgentState) -> Dict[str, Any]:
365
+ """
366
+ Determine if the agent should retry the workflow based on errors or incomplete results.
367
+
368
+ Args:
369
+ state: The current agent state
370
+
371
+ Returns:
372
+ Dictionary indicating the next node
373
+ """
374
+ if state.get("error"):
375
+ agent_config = get_agent_config()
376
+
377
+ logger.error(f"Workflow ended with error: {state['error']}")
378
+ return {"next": END}
379
+
380
+ if state.get("answer"):
381
+ return {"next": END}
382
+
383
+ return {"next": "analyze_question"}
384
+
385
+ def create_agent_graph() -> Tuple[StateGraph, Dict[str, Any]]:
386
+ """
387
+ Create the agent workflow graph.
388
+
389
+ Returns:
390
+ Tuple containing the StateGraph and the initial state
391
+ """
392
+ workflow = StateGraph(AgentState)
393
+
394
+ workflow.add_node("analyze_question", analyze_question)
395
+ workflow.add_node("create_plan", create_plan)
396
+ workflow.add_node("select_tool", select_tool)
397
+ workflow.add_node("execute_tool", execute_tool)
398
+ workflow.add_node("formulate_answer", formulate_answer)
399
+
400
+ workflow.add_edge("analyze_question", "create_plan")
401
+ workflow.add_edge("create_plan", "select_tool")
402
+ workflow.add_conditional_edges(
403
+ "select_tool",
404
+ should_retry,
405
+ {
406
+ "execute_tool": "execute_tool",
407
+ "formulate_answer": "formulate_answer",
408
+ END: END
409
+ }
410
+ )
411
+ workflow.add_edge("execute_tool", "select_tool")
412
+ workflow.add_edge("formulate_answer", END)
413
+
414
+ workflow.set_entry_point("analyze_question")
415
+
416
+ initial_state = {
417
+ "question": "",
418
+ "analysis": None,
419
+ "plan": None,
420
+ "current_step": 0,
421
+ "tools_to_use": None,
422
+ "tool_results": [],
423
+ "intermediate_thoughts": [],
424
+ "answer": None,
425
+ "error": None
426
+ }
427
+
428
+ return workflow, initial_state
429
+
430
+ def run_agent_graph(question: str, config: Optional[Dict[str, Any]] = None) -> Dict[str, Any]:
431
+ """
432
+ Run the agent workflow graph with a question.
433
+
434
+ Args:
435
+ question: The question to answer
436
+ config: Optional configuration for the graph
437
+
438
+ Returns:
439
+ The final state of the graph
440
+ """
441
+ workflow, initial_state = create_agent_graph()
442
+
443
+ app = workflow.compile()
444
+
445
+ state = initial_state.copy()
446
+ state["question"] = question
447
+
448
+ runnable_config = RunnableConfig(config or {})
449
+
450
+ result = app.invoke(state, config=runnable_config)
451
+
452
+ return result
agent/tool_registry.py ADDED
@@ -0,0 +1,319 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ Tool registry for the GAIA agent.
3
+
4
+ This module provides a registry for tools that can be used by the GAIA agent.
5
+ It includes factory functions for creating tool instances and a registry class
6
+ for managing tool instances.
7
+
8
+ The registry is designed to be used with the LangGraph workflow in agent/graph.py.
9
+ """
10
+
11
+ import logging
12
+ import os
13
+ from typing import Dict, Any, Optional, List, Callable
14
+
15
+ from tools.web_tools import (
16
+ DuckDuckGoSearchTool,
17
+ SerperSearchTool,
18
+ create_duckduckgo_search,
19
+ create_serper_search
20
+ )
21
+ from tools.perplexity_tool import PerplexityTool, create_perplexity_tool
22
+ from tools.arxiv_tool import ArxivSearchTool, create_arxiv_search
23
+
24
+ logger = logging.getLogger("gaia_agent.tool_registry")
25
+
26
+ class ToolRegistry:
27
+ """Registry for tools used by the GAIA agent."""
28
+
29
+ def __init__(self):
30
+ """Initialize an empty tool registry."""
31
+ self.tools = {}
32
+
33
+ def register_tool(self, name: str, tool: Any) -> None:
34
+ """
35
+ Register a tool in the registry.
36
+
37
+ Args:
38
+ name: The name of the tool
39
+ tool: The tool instance
40
+ """
41
+ self.tools[name] = tool
42
+
43
+ def get_tool(self, name: str) -> Optional[Any]:
44
+ """
45
+ Get a tool from the registry.
46
+
47
+ Args:
48
+ name: The name of the tool
49
+
50
+ Returns:
51
+ The tool instance, or None if not found
52
+ """
53
+ tool = self.tools.get(name)
54
+ if not tool:
55
+ logger.warning(f"Tool not found in registry: {name}")
56
+ return tool
57
+
58
+ def list_tools(self) -> List[str]:
59
+ """
60
+ List all tools in the registry.
61
+
62
+ Returns:
63
+ List of tool names
64
+ """
65
+ return list(self.tools.keys())
66
+
67
+ def execute_tool(self, name: str, **kwargs) -> Any:
68
+ """
69
+ Execute a tool from the registry.
70
+
71
+ Args:
72
+ name: The name of the tool
73
+ **kwargs: Arguments to pass to the tool
74
+
75
+ Returns:
76
+ The result of the tool execution
77
+
78
+ Raises:
79
+ Exception: If the tool is not found or execution fails
80
+ """
81
+ tool = self.get_tool(name)
82
+ if not tool:
83
+ raise Exception(f"Tool not found in registry: {name}")
84
+
85
+
86
+ try:
87
+ if name in ["duckduckgo_search", "serper_search"]:
88
+ query = kwargs.get("query")
89
+ if not query:
90
+ raise ValueError("Query is required for search tools")
91
+ return tool.search(query)
92
+
93
+ elif name == "perplexity_search":
94
+ query = kwargs.get("query")
95
+ if not query:
96
+ raise ValueError("Query is required for Perplexity search")
97
+ return tool.search(query)
98
+
99
+ elif name == "arxiv_search":
100
+ query = kwargs.get("query")
101
+ max_results = kwargs.get("max_results")
102
+ if not query:
103
+ raise ValueError("Query is required for arXiv search")
104
+ return tool.search(query, max_results)
105
+
106
+ elif name == "arxiv_get_paper":
107
+ paper_id = kwargs.get("paper_id")
108
+ if not paper_id:
109
+ raise ValueError("Paper ID is required for arXiv paper retrieval")
110
+ return tool.get_paper_by_id(paper_id)
111
+
112
+ elif name == "arxiv_search_category":
113
+ category = kwargs.get("category")
114
+ max_results = kwargs.get("max_results")
115
+ if not category:
116
+ raise ValueError("Category is required for arXiv category search")
117
+ return tool.search_by_category(category, max_results)
118
+
119
+ else:
120
+ return tool.run(**kwargs)
121
+
122
+ except Exception as e:
123
+ logger.error(f"Error executing tool {name}: {str(e)}")
124
+ raise
125
+
126
+ def create_default_registry() -> ToolRegistry:
127
+ """
128
+ Create a default tool registry with all available tools.
129
+
130
+ Returns:
131
+ ToolRegistry: A registry with all available tools
132
+ """
133
+ registry = ToolRegistry()
134
+
135
+ # Register DuckDuckGo search tool (no API key required)
136
+ try:
137
+ duckduckgo_tool = create_duckduckgo_search()
138
+ registry.register_tool("duckduckgo_search", duckduckgo_tool)
139
+ except Exception as e:
140
+ logger.warning(f"Failed to create DuckDuckGo search tool: {str(e)}")
141
+
142
+ # Register Serper search tool if API key is available
143
+ serper_api_key = os.environ.get("SERPER_API_KEY")
144
+ if serper_api_key:
145
+ try:
146
+ serper_tool = create_serper_search()
147
+ registry.register_tool("serper_search", serper_tool)
148
+ except Exception as e:
149
+ logger.warning(f"Failed to create Serper search tool: {str(e)}")
150
+ else:
151
+ logger.warning("Serper API key not available, skipping Serper search tool")
152
+
153
+ # Register Perplexity tool if API key is available
154
+ perplexity_api_key = os.environ.get("PERPLEXITY_API_KEY")
155
+ if perplexity_api_key:
156
+ try:
157
+ perplexity_tool = create_perplexity_tool()
158
+ registry.register_tool("perplexity_search", perplexity_tool)
159
+ except Exception as e:
160
+ logger.warning(f"Failed to create Perplexity tool: {str(e)}")
161
+ else:
162
+ logger.warning("Perplexity API key not available, skipping Perplexity tool")
163
+
164
+ try:
165
+ arxiv_tool = create_arxiv_search()
166
+ registry.register_tool("arxiv_search", arxiv_tool)
167
+ registry.register_tool("arxiv_get_paper", arxiv_tool)
168
+ registry.register_tool("arxiv_search_category", arxiv_tool)
169
+ except Exception as e:
170
+ logger.warning(f"Failed to create arXiv search tool: {str(e)}")
171
+
172
+ logger.info(f"Created default tool registry with {len(registry.list_tools())} tools")
173
+ return registry
174
+
175
+ def robust_search(registry: ToolRegistry, query: str) -> Dict[str, Any]:
176
+ """
177
+ Perform a robust search using multiple providers.
178
+
179
+ This function tries multiple search providers in order, falling back to
180
+ the next provider if one fails.
181
+
182
+ Args:
183
+ registry: The tool registry
184
+ query: The search query
185
+
186
+ Returns:
187
+ Dict with provider name and search results
188
+
189
+ Raises:
190
+ Exception: If all search providers fail
191
+ """
192
+ # Try DuckDuckGo first (no API key required)
193
+ if registry.get_tool("duckduckgo_search"):
194
+ try:
195
+ results = registry.execute_tool("duckduckgo_search", query=query)
196
+ if results:
197
+ return {"provider": "duckduckgo", "results": results}
198
+ except Exception as e:
199
+ logger.warning(f"DuckDuckGo search failed: {str(e)}")
200
+
201
+ # Try Serper as fallback
202
+ if registry.get_tool("serper_search"):
203
+ try:
204
+ results = registry.execute_tool("serper_search", query=query)
205
+ if results:
206
+ return {"provider": "serper", "results": results}
207
+ except Exception as e:
208
+ logger.warning(f"Serper search failed: {str(e)}")
209
+
210
+ # Try Perplexity as last resort
211
+ if registry.get_tool("perplexity_search"):
212
+ try:
213
+ results = registry.execute_tool("perplexity_search", query=query)
214
+ if results:
215
+ return {"provider": "perplexity", "results": results}
216
+ except Exception as e:
217
+ logger.warning(f"Perplexity search failed: {str(e)}")
218
+
219
+ if registry.get_tool("arxiv_search"):
220
+ try:
221
+ results = registry.execute_tool("arxiv_search", query=query)
222
+ if results:
223
+ return {"provider": "arxiv", "results": results}
224
+ except Exception as e:
225
+ logger.warning(f"arXiv search failed: {str(e)}")
226
+
227
+ raise Exception("All search providers failed")
228
+
229
+ def merged_search(registry: ToolRegistry, query: str) -> Dict[str, Any]:
230
+ """
231
+ Perform a merged search using multiple providers.
232
+
233
+ This function collects results from all available search providers and
234
+ merges them into a single result set, removing duplicates.
235
+
236
+ Args:
237
+ registry: The tool registry
238
+ query: The search query
239
+
240
+ Returns:
241
+ Dict with merged results and Perplexity context
242
+ """
243
+ results = {
244
+ "duckduckgo": [],
245
+ "serper": [],
246
+ "perplexity": None,
247
+ "arxiv": []
248
+ }
249
+
250
+ if registry.get_tool("duckduckgo_search"):
251
+ try:
252
+ results["duckduckgo"] = registry.execute_tool("duckduckgo_search", query=query)
253
+ except Exception as e:
254
+ logger.warning(f"DuckDuckGo search failed: {str(e)}")
255
+
256
+ if registry.get_tool("serper_search"):
257
+ try:
258
+ results["serper"] = registry.execute_tool("serper_search", query=query)
259
+ except Exception as e:
260
+ logger.warning(f"Serper search failed: {str(e)}")
261
+
262
+ if registry.get_tool("perplexity_search"):
263
+ try:
264
+ results["perplexity"] = registry.execute_tool("perplexity_search", query=query)
265
+ except Exception as e:
266
+ logger.warning(f"Perplexity search failed: {str(e)}")
267
+ merged = []
268
+ seen_urls = set()
269
+
270
+ # Add DuckDuckGo results
271
+ for result in results["duckduckgo"]:
272
+ if result["link"] not in seen_urls:
273
+ merged.append(result)
274
+ seen_urls.add(result["link"])
275
+
276
+ # Add Serper results
277
+ for result in results["serper"]:
278
+ if result["link"] not in seen_urls:
279
+ merged.append(result)
280
+ seen_urls.add(result["link"])
281
+
282
+ arxiv_results = []
283
+ for result in results["arxiv"]:
284
+ if "url" in result and result["url"] not in seen_urls:
285
+ formatted_result = {
286
+ "title": result.get("title", ""),
287
+ "link": result.get("url", ""),
288
+ "snippet": result.get("summary", "")[:200] + "..." if result.get("summary") else ""
289
+ }
290
+ merged.append(formatted_result)
291
+ seen_urls.add(result["url"])
292
+ arxiv_results.append(result) # Keep original format for context
293
+
294
+ # Add Perplexity content as context
295
+ perplexity_content = results["perplexity"].get("content") if results["perplexity"] else None
296
+
297
+ return {
298
+ "merged_results": merged,
299
+ "perplexity_context": perplexity_content,
300
+ "arxiv_context": arxiv_results
301
+ }
302
+
303
+ if __name__ == "__main__":
304
+ logging.basicConfig(level=logging.INFO)
305
+
306
+ registry = create_default_registry()
307
+
308
+
309
+ try:
310
+ query = "latest Python version"
311
+ result = robust_search(registry, query)
312
+ except Exception as e:
313
+
314
+ try:
315
+ query = "latest Python version"
316
+ result = merged_search(registry, query)
317
+ if result["perplexity_context"]:
318
+ if result.get("arxiv_context"):
319
+ except Exception as e:
app.py CHANGED
@@ -1,196 +1,341 @@
1
- import os
2
- import gradio as gr
3
- import requests
4
- import inspect
5
- import pandas as pd
6
-
7
- # (Keep Constants as is)
8
- # --- Constants ---
9
- DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
10
-
11
- # --- Basic Agent Definition ---
12
- # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
13
- class BasicAgent:
14
- def __init__(self):
15
- print("BasicAgent initialized.")
16
- def __call__(self, question: str) -> str:
17
- print(f"Agent received question (first 50 chars): {question[:50]}...")
18
- fixed_answer = "This is a default answer."
19
- print(f"Agent returning fixed answer: {fixed_answer}")
20
- return fixed_answer
21
-
22
- def run_and_submit_all( profile: gr.OAuthProfile | None):
23
- """
24
- Fetches all questions, runs the BasicAgent on them, submits all answers,
25
- and displays the results.
26
- """
27
- # --- Determine HF Space Runtime URL and Repo URL ---
28
- space_id = os.getenv("SPACE_ID") # Get the SPACE_ID for sending link to the code
29
-
30
- if profile:
31
- username= f"{profile.username}"
32
- print(f"User logged in: {username}")
33
- else:
34
- print("User not logged in.")
35
- return "Please Login to Hugging Face with the button.", None
36
-
37
- api_url = DEFAULT_API_URL
38
- questions_url = f"{api_url}/questions"
39
- submit_url = f"{api_url}/submit"
40
-
41
- # 1. Instantiate Agent ( modify this part to create your agent)
42
- try:
43
- agent = BasicAgent()
44
- except Exception as e:
45
- print(f"Error instantiating agent: {e}")
46
- return f"Error initializing agent: {e}", None
47
- # In the case of an app running as a hugging Face space, this link points toward your codebase ( usefull for others so please keep it public)
48
- agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
49
- print(agent_code)
50
-
51
- # 2. Fetch Questions
52
- print(f"Fetching questions from: {questions_url}")
53
- try:
54
- response = requests.get(questions_url, timeout=15)
55
- response.raise_for_status()
56
- questions_data = response.json()
57
- if not questions_data:
58
- print("Fetched questions list is empty.")
59
- return "Fetched questions list is empty or invalid format.", None
60
- print(f"Fetched {len(questions_data)} questions.")
61
- except requests.exceptions.RequestException as e:
62
- print(f"Error fetching questions: {e}")
63
- return f"Error fetching questions: {e}", None
64
- except requests.exceptions.JSONDecodeError as e:
65
- print(f"Error decoding JSON response from questions endpoint: {e}")
66
- print(f"Response text: {response.text[:500]}")
67
- return f"Error decoding server response for questions: {e}", None
68
- except Exception as e:
69
- print(f"An unexpected error occurred fetching questions: {e}")
70
- return f"An unexpected error occurred fetching questions: {e}", None
71
-
72
- # 3. Run your Agent
73
- results_log = []
74
- answers_payload = []
75
- print(f"Running agent on {len(questions_data)} questions...")
76
- for item in questions_data:
77
- task_id = item.get("task_id")
78
- question_text = item.get("question")
79
- if not task_id or question_text is None:
80
- print(f"Skipping item with missing task_id or question: {item}")
81
- continue
82
- try:
83
- submitted_answer = agent(question_text)
84
- answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
85
- results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
86
- except Exception as e:
87
- print(f"Error running agent on task {task_id}: {e}")
88
- results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"AGENT ERROR: {e}"})
89
-
90
- if not answers_payload:
91
- print("Agent did not produce any answers to submit.")
92
- return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
93
-
94
- # 4. Prepare Submission
95
- submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}
96
- status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..."
97
- print(status_update)
98
-
99
- # 5. Submit
100
- print(f"Submitting {len(answers_payload)} answers to: {submit_url}")
101
- try:
102
- response = requests.post(submit_url, json=submission_data, timeout=60)
103
- response.raise_for_status()
104
- result_data = response.json()
105
- final_status = (
106
- f"Submission Successful!\n"
107
- f"User: {result_data.get('username')}\n"
108
- f"Overall Score: {result_data.get('score', 'N/A')}% "
109
- f"({result_data.get('correct_count', '?')}/{result_data.get('total_attempted', '?')} correct)\n"
110
- f"Message: {result_data.get('message', 'No message received.')}"
111
- )
112
- print("Submission successful.")
113
- results_df = pd.DataFrame(results_log)
114
- return final_status, results_df
115
- except requests.exceptions.HTTPError as e:
116
- error_detail = f"Server responded with status {e.response.status_code}."
117
- try:
118
- error_json = e.response.json()
119
- error_detail += f" Detail: {error_json.get('detail', e.response.text)}"
120
- except requests.exceptions.JSONDecodeError:
121
- error_detail += f" Response: {e.response.text[:500]}"
122
- status_message = f"Submission Failed: {error_detail}"
123
- print(status_message)
124
- results_df = pd.DataFrame(results_log)
125
- return status_message, results_df
126
- except requests.exceptions.Timeout:
127
- status_message = "Submission Failed: The request timed out."
128
- print(status_message)
129
- results_df = pd.DataFrame(results_log)
130
- return status_message, results_df
131
- except requests.exceptions.RequestException as e:
132
- status_message = f"Submission Failed: Network error - {e}"
133
- print(status_message)
134
- results_df = pd.DataFrame(results_log)
135
- return status_message, results_df
136
- except Exception as e:
137
- status_message = f"An unexpected error occurred during submission: {e}"
138
- print(status_message)
139
- results_df = pd.DataFrame(results_log)
140
- return status_message, results_df
141
-
142
-
143
- # --- Build Gradio Interface using Blocks ---
144
- with gr.Blocks() as demo:
145
- gr.Markdown("# Basic Agent Evaluation Runner")
146
- gr.Markdown(
147
- """
148
- **Instructions:**
149
-
150
- 1. Please clone this space, then modify the code to define your agent's logic, the tools, the necessary packages, etc ...
151
- 2. Log in to your Hugging Face account using the button below. This uses your HF username for submission.
152
- 3. Click 'Run Evaluation & Submit All Answers' to fetch questions, run your agent, submit answers, and see the score.
153
-
154
- ---
155
- **Disclaimers:**
156
- Once clicking on the "submit button, it can take quite some time ( this is the time for the agent to go through all the questions).
157
- This space provides a basic setup and is intentionally sub-optimal to encourage you to develop your own, more robust solution. For instance for the delay process of the submit button, a solution could be to cache the answers and submit in a seperate action or even to answer the questions in async.
158
- """
159
- )
160
-
161
- gr.LoginButton()
162
-
163
- run_button = gr.Button("Run Evaluation & Submit All Answers")
164
-
165
- status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
166
- # Removed max_rows=10 from DataFrame constructor
167
- results_table = gr.DataFrame(label="Questions and Agent Answers", wrap=True)
168
-
169
- run_button.click(
170
- fn=run_and_submit_all,
171
- outputs=[status_output, results_table]
172
- )
173
-
174
- if __name__ == "__main__":
175
- print("\n" + "-"*30 + " App Starting " + "-"*30)
176
- # Check for SPACE_HOST and SPACE_ID at startup for information
177
- space_host_startup = os.getenv("SPACE_HOST")
178
- space_id_startup = os.getenv("SPACE_ID") # Get SPACE_ID at startup
179
-
180
- if space_host_startup:
181
- print(f"✅ SPACE_HOST found: {space_host_startup}")
182
- print(f" Runtime URL should be: https://{space_host_startup}.hf.space")
183
- else:
184
- print("ℹ️ SPACE_HOST environment variable not found (running locally?).")
185
-
186
- if space_id_startup: # Print repo URLs if SPACE_ID is found
187
- print(f"✅ SPACE_ID found: {space_id_startup}")
188
- print(f" Repo URL: https://huggingface.co/spaces/{space_id_startup}")
189
- print(f" Repo Tree URL: https://huggingface.co/spaces/{space_id_startup}/tree/main")
190
- else:
191
- print("ℹ️ SPACE_ID environment variable not found (running locally?). Repo URL cannot be determined.")
192
-
193
- print("-"*(60 + len(" App Starting ")) + "\n")
194
-
195
- print("Launching Gradio Interface for Basic Agent Evaluation...")
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
196
  demo.launch(debug=True, share=False)
 
1
+ import os
2
+ import gradio as gr
3
+ import requests
4
+ import inspect
5
+ import pandas as pd
6
+ import logging
7
+ import time
8
+ from typing import Dict, Any, List, Optional, Tuple, Union
9
+
10
+ DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
11
+
12
+ from agent import GAIAAgent
13
+ from agent.config import VERBOSE
14
+
15
+ logger = logging.getLogger("gaia_app")
16
+ if VERBOSE:
17
+ logging.basicConfig(level=logging.DEBUG, format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')
18
+ else:
19
+ logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')
20
+
21
+ def run_and_submit_all(profile: Optional[gr.OAuthProfile] = None, progress=gr.Progress()) -> Tuple[str, Optional[pd.DataFrame]]:
22
+ """
23
+ Fetches all questions, runs the GAIA Agent on them, submits all answers,
24
+ and displays the results with progress tracking.
25
+
26
+ Args:
27
+ profile: Hugging Face user profile for authentication
28
+ progress: Gradio progress tracker
29
+
30
+ Returns:
31
+ Tuple containing status message and results dataframe
32
+ """
33
+ space_id = os.getenv("SPACE_ID") # Get the SPACE_ID for sending link to the code
34
+
35
+ if profile:
36
+ username = f"{profile.username}"
37
+ logger.info(f"User logged in: {username}")
38
+ else:
39
+ logger.warning("User not logged in.")
40
+ return "Please Login to Hugging Face with the button.", None
41
+
42
+ api_url = DEFAULT_API_URL
43
+ questions_url = f"{api_url}/questions"
44
+ submit_url = f"{api_url}/submit"
45
+
46
+ progress(0, desc="Initializing GAIA Agent...")
47
+ try:
48
+ start_time = time.time()
49
+ agent = GAIAAgent()
50
+ init_time = time.time() - start_time
51
+ logger.info(f"GAIA Agent initialized successfully in {init_time:.2f} seconds.")
52
+ progress(0.1, desc=f"Agent initialized in {init_time:.2f}s")
53
+ except Exception as e:
54
+ error_msg = f"Error initializing GAIA agent: {e}"
55
+ logger.error(error_msg)
56
+ return error_msg, None
57
+
58
+ agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
59
+ logger.info(f"Agent code URL: {agent_code}")
60
+
61
+ progress(0.15, desc="Fetching questions...")
62
+ logger.info(f"Fetching questions from: {questions_url}")
63
+ try:
64
+ response = requests.get(questions_url, timeout=15)
65
+ response.raise_for_status()
66
+ questions_data = response.json()
67
+ if not questions_data:
68
+ error_msg = "Fetched questions list is empty or invalid format."
69
+ logger.warning(error_msg)
70
+ return error_msg, None
71
+
72
+ question_count = len(questions_data)
73
+ logger.info(f"Fetched {question_count} questions.")
74
+ progress(0.2, desc=f"Fetched {question_count} questions")
75
+ except requests.exceptions.RequestException as e:
76
+ error_msg = f"Error fetching questions: {e}"
77
+ logger.error(error_msg)
78
+ return error_msg, None
79
+ except requests.exceptions.JSONDecodeError as e:
80
+ error_msg = f"Error decoding JSON response from questions endpoint: {e}"
81
+ logger.error(f"{error_msg}\nResponse text: {response.text[:500]}")
82
+ return error_msg, None
83
+ except Exception as e:
84
+ error_msg = f"An unexpected error occurred fetching questions: {e}"
85
+ logger.error(error_msg)
86
+ return error_msg, None
87
+
88
+ results_log = []
89
+ answers_payload = []
90
+ total_questions = len(questions_data)
91
+
92
+ logger.info(f"Running agent on {total_questions} questions...")
93
+
94
+ question_progress_increment = 0.6 / total_questions # 60% of progress bar for processing questions
95
+
96
+ for i, item in enumerate(questions_data):
97
+ task_id = item.get("task_id")
98
+ question_text = item.get("question")
99
+
100
+ if not task_id or question_text is None:
101
+ logger.warning(f"Skipping item with missing task_id or question: {item}")
102
+ continue
103
+
104
+ progress_value = 0.2 + (i * question_progress_increment)
105
+ progress(progress_value, desc=f"Processing question {i+1}/{total_questions}...")
106
+
107
+ logger.info(f"Processing question {i+1}/{total_questions}: {question_text[:50]}...")
108
+ question_start_time = time.time()
109
+
110
+ try:
111
+ submitted_answer = agent(question_text)
112
+ processing_time = time.time() - question_start_time
113
+
114
+ answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
115
+ results_log.append({
116
+ "Task ID": task_id,
117
+ "Question": question_text,
118
+ "Submitted Answer": submitted_answer,
119
+ "Processing Time": f"{processing_time:.2f}s",
120
+ "Status": "Success"
121
+ })
122
+
123
+ logger.info(f"Question {i+1} processed in {processing_time:.2f}s. Answer: {submitted_answer[:50]}...")
124
+
125
+ except Exception as e:
126
+ error_msg = f"Error running agent on task {task_id}: {e}"
127
+ logger.error(error_msg)
128
+
129
+ results_log.append({
130
+ "Task ID": task_id,
131
+ "Question": question_text,
132
+ "Submitted Answer": f"AGENT ERROR: {e}",
133
+ "Processing Time": f"{time.time() - question_start_time:.2f}s",
134
+ "Status": "Error"
135
+ })
136
+
137
+ if not answers_payload:
138
+ error_msg = "Agent did not produce any answers to submit."
139
+ logger.warning(error_msg)
140
+ return error_msg, pd.DataFrame(results_log)
141
+
142
+ progress(0.8, desc="Preparing submission...")
143
+ submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}
144
+ status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..."
145
+ logger.info(status_update)
146
+
147
+ progress(0.9, desc=f"Submitting {len(answers_payload)} answers...")
148
+ logger.info(f"Submitting {len(answers_payload)} answers to: {submit_url}")
149
+ try:
150
+ response = requests.post(submit_url, json=submission_data, timeout=60)
151
+ response.raise_for_status()
152
+ result_data = response.json()
153
+
154
+ correct_count = result_data.get('correct_count', 0)
155
+ total_attempted = result_data.get('total_attempted', 0)
156
+ score = result_data.get('score', 'N/A')
157
+
158
+ final_status = (
159
+ f"✅ Submission Successful!\n"
160
+ f"User: {result_data.get('username')}\n"
161
+ f"Overall Score: {score}% "
162
+ f"({correct_count}/{total_attempted} correct)\n"
163
+ f"Message: {result_data.get('message', 'No message received.')}"
164
+ )
165
+ logger.info("Submission successful.")
166
+
167
+ results_df = pd.DataFrame(results_log)
168
+
169
+ if not results_df.empty:
170
+ success_count = sum(1 for status in results_df.get("Status", []) if status == "Success")
171
+ error_count = sum(1 for status in results_df.get("Status", []) if status == "Error")
172
+
173
+ try:
174
+ avg_time = sum(float(time.strip('s')) for time in results_df["Processing Time"]) / len(results_df)
175
+ avg_time_str = f"{avg_time:.2f}s"
176
+ except:
177
+ avg_time_str = "N/A"
178
+
179
+ logger.info(f"Processed {len(results_df)} questions: {success_count} successes, {error_count} errors, avg time: {avg_time_str}")
180
+
181
+ progress(1.0, desc="Evaluation complete!")
182
+ return final_status, results_df
183
+ except requests.exceptions.HTTPError as e:
184
+ progress(1.0, desc="Submission failed!")
185
+ error_detail = f"Server responded with status {e.response.status_code}."
186
+ try:
187
+ error_json = e.response.json()
188
+ error_detail += f" Detail: {error_json.get('detail', e.response.text)}"
189
+ except requests.exceptions.JSONDecodeError:
190
+ error_detail += f" Response: {e.response.text[:500]}"
191
+ status_message = f"❌ Submission Failed: {error_detail}"
192
+ logger.error(status_message)
193
+ results_df = pd.DataFrame(results_log)
194
+ return status_message, results_df
195
+ except requests.exceptions.Timeout:
196
+ progress(1.0, desc="Submission timed out!")
197
+ status_message = "❌ Submission Failed: The request timed out."
198
+ logger.error(status_message)
199
+ results_df = pd.DataFrame(results_log)
200
+ return status_message, results_df
201
+ except requests.exceptions.RequestException as e:
202
+ progress(1.0, desc="Network error!")
203
+ status_message = f"❌ Submission Failed: Network error - {e}"
204
+ logger.error(status_message)
205
+ results_df = pd.DataFrame(results_log)
206
+ return status_message, results_df
207
+ except Exception as e:
208
+ progress(1.0, desc="Error during submission!")
209
+ status_message = f"❌ An unexpected error occurred during submission: {e}"
210
+ logger.error(status_message)
211
+ results_df = pd.DataFrame(results_log)
212
+ return status_message, results_df
213
+
214
+
215
+ with gr.Blocks(theme=gr.themes.Soft()) as demo:
216
+ gr.Markdown("# GAIA Agent Evaluation Runner")
217
+
218
+ with gr.Tabs():
219
+ with gr.TabItem("Run Evaluation"):
220
+ gr.Markdown(
221
+ """
222
+ **Instructions:**
223
+
224
+ 1. Log in to your Hugging Face account using the button below. This uses your HF username for submission.
225
+ 2. Click 'Run Evaluation & Submit All Answers' to fetch questions, run the GAIA agent, submit answers, and see the score.
226
+ 3. The progress bar will show you the current status of the evaluation.
227
+ 4. Results will appear in the table below once processing is complete.
228
+
229
+ **Note:** Running the evaluation may take some time as the agent processes all questions. Please be patient.
230
+ """
231
+ )
232
+
233
+ with gr.Row():
234
+ with gr.Column(scale=1):
235
+ gr.LoginButton()
236
+ with gr.Column(scale=2):
237
+ run_button = gr.Button("Run Evaluation & Submit All Answers", variant="primary", size="lg")
238
+
239
+ status_output = gr.Textbox(
240
+ label="Run Status / Submission Result",
241
+ lines=6,
242
+ interactive=False,
243
+ elem_id="status_output"
244
+ )
245
+
246
+ with gr.Accordion("Results", open=False, elem_id="results_accordion"):
247
+ results_table = gr.DataFrame(
248
+ label="Questions and Agent Answers",
249
+ wrap=True,
250
+ elem_id="results_table"
251
+ )
252
+
253
+ with gr.Row():
254
+ with gr.Column():
255
+ gr.Markdown("### Result Statistics")
256
+ stats_output = gr.JSON(label="Statistics", visible=True)
257
+
258
+ with gr.TabItem("About GAIA Agent"):
259
+ gr.Markdown(
260
+ """
261
+ ## About GAIA Agent
262
+
263
+ This agent is designed to answer questions from the GAIA (Grounded AI Alignment) benchmark.
264
+
265
+ ### Key Features:
266
+
267
+ - **Architecture**: Built with LangChain/LangGraph for workflow management
268
+ - **Tools**: Integrates multiple tools for web search, reasoning, and multimodal understanding
269
+ - **Memory**: Implements memory components for context retention
270
+ - **Structured Approach**: Uses a systematic process for question analysis and answer formulation
271
+
272
+ ### Components:
273
+
274
+ 1. **Web Tools**: Search engines, content extraction, and web navigation
275
+ 2. **Reasoning Tools**: Step-by-step reasoning, mathematical calculations, fact verification
276
+ 3. **Multimodal Tools**: Image analysis, chart interpretation, document parsing
277
+ 4. **Memory Components**: Conversation history, working memory, result caching
278
+
279
+ ### Performance Metrics:
280
+
281
+ The agent's performance is evaluated based on:
282
+ - Accuracy of answers
283
+ - Reasoning quality
284
+ - Ability to use appropriate tools
285
+ - Handling of edge cases
286
+ """
287
+ )
288
+
289
+ with gr.TabItem("Help & Troubleshooting"):
290
+ gr.Markdown(
291
+ """
292
+ ## Help & Troubleshooting
293
+
294
+ ### Common Issues:
295
+
296
+ 1. **Login Problems**: Make sure you're logged in to Hugging Face before running the evaluation.
297
+ 2. **Timeout Errors**: If the agent takes too long to process a question, try running the evaluation again.
298
+ 3. **API Errors**: If you encounter API errors, check your internet connection and try again.
299
+
300
+ ### Error Messages:
301
+
302
+ - **"Please Login to Hugging Face"**: You need to log in using the login button.
303
+ - **"Error initializing GAIA agent"**: There was a problem setting up the agent. Check the logs for details.
304
+ - **"Error fetching questions"**: The system couldn't retrieve questions from the API.
305
+ - **"Submission Failed"**: There was a problem submitting your answers. Check the error details.
306
+
307
+ ### Contact:
308
+
309
+ If you continue to experience issues, please open an issue on the GitHub repository or contact the course instructors.
310
+ """
311
+ )
312
+
313
+ run_button.click(
314
+ fn=run_and_submit_all,
315
+ outputs=[status_output, results_table]
316
+ ).then(
317
+ fn=lambda df: {
318
+ "Total Questions": len(df) if df is not None else 0,
319
+ "Successful": sum(1 for status in df.get("Status", []) if status == "Success") if df is not None else 0,
320
+ "Errors": sum(1 for status in df.get("Status", []) if status == "Error") if df is not None else 0,
321
+ "Average Processing Time": f"{sum(float(time.strip('s')) for time in df['Processing Time']) / len(df):.2f}s" if df is not None and len(df) > 0 else "N/A"
322
+ } if df is not None else {},
323
+ inputs=[results_table],
324
+ outputs=[stats_output]
325
+ ).then(
326
+ fn=lambda: gr.update(open=True),
327
+ outputs=[gr.Accordion.update(visible=True)]
328
+ )
329
+
330
+ if __name__ == "__main__":
331
+ space_host_startup = os.getenv("SPACE_HOST")
332
+ space_id_startup = os.getenv("SPACE_ID") # Get SPACE_ID at startup
333
+
334
+ if space_host_startup:
335
+ else:
336
+
337
+ if space_id_startup: # Print repo URLs if SPACE_ID is found
338
+ else:
339
+
340
+
341
  demo.launch(debug=True, share=False)
backup_tests/minimal_supabase_test.py ADDED
@@ -0,0 +1,75 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ Minimal Supabase Connection Test
3
+
4
+ This script performs a minimal test of the Supabase connection.
5
+ """
6
+
7
+ import os
8
+ import sys
9
+ import logging
10
+ from dotenv import load_dotenv
11
+
12
+ logging.basicConfig(level=logging.DEBUG)
13
+ logger = logging.getLogger("minimal_supabase_test")
14
+
15
+ dotenv_path = os.path.join(os.path.dirname(os.path.dirname(os.path.abspath(__file__))), '.env')
16
+ logger.info(f"Loading environment variables from {dotenv_path}")
17
+ load_dotenv(dotenv_path)
18
+
19
+ supabase_url = os.environ.get("SUPABASE_URL")
20
+ supabase_key = os.environ.get("SUPABASE_KEY")
21
+
22
+ logger.info(f"SUPABASE_URL: {supabase_url}")
23
+ logger.info(f"SUPABASE_KEY: {'***' if supabase_key else None}")
24
+
25
+ if not supabase_url or not supabase_key:
26
+ logger.error("Supabase URL or key not provided")
27
+ logger.error("Please set SUPABASE_URL and SUPABASE_KEY environment variables")
28
+ sys.exit(1)
29
+
30
+ try:
31
+ from supabase import create_client, Client
32
+ logger.info("Supabase package imported successfully")
33
+ except ImportError as e:
34
+ logger.error(f"Error importing supabase: {str(e)}")
35
+ logger.error("Please install the supabase package with: pip install supabase")
36
+ sys.exit(1)
37
+
38
+ try:
39
+ logger.info(f"Connecting to Supabase at {supabase_url}")
40
+ client = create_client(supabase_url, supabase_key)
41
+ logger.info("Supabase client created successfully")
42
+
43
+ import time
44
+ from datetime import datetime
45
+
46
+ test_id = int(time.time())
47
+ test_name = f"Test Record {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}"
48
+
49
+ logger.info(f"Inserting test record with ID {test_id} and name '{test_name}'...")
50
+ insert_result = client.table("test_connection").insert({
51
+ "id": test_id,
52
+ "name": test_name
53
+ }).execute()
54
+
55
+ if insert_result.data:
56
+ logger.info("Test record inserted successfully")
57
+ else:
58
+ logger.warning("Failed to insert test record")
59
+
60
+ logger.info("Querying test_connection table...")
61
+ result = client.table("test_connection").select("*").limit(5).execute()
62
+
63
+ if result.data is not None:
64
+ logger.info(f"Query successful, found {len(result.data)} records")
65
+ for record in result.data:
66
+ else:
67
+ logger.warning("Query successful but no data returned")
68
+
69
+ logger.info("Connection to Supabase successful")
70
+ except Exception as e:
71
+ logger.error(f"Error connecting to Supabase: {str(e)}")
72
+ import traceback
73
+ logger.error(f"Traceback: {traceback.format_exc()}")
74
+ sys.exit(1)
75
+
backup_tests/test_supabase_tables_final.py ADDED
@@ -0,0 +1,167 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ Final test script for Supabase tables in GAIA Agent.
3
+
4
+ This script applies the fix to ensure all Supabase tables are properly utilized,
5
+ then runs a simple test to verify the fix works correctly.
6
+
7
+ Usage:
8
+ python -m tests.test_supabase_tables_final
9
+ """
10
+
11
+ import os
12
+ import sys
13
+ import json
14
+ import logging
15
+ import time
16
+ import uuid
17
+ import datetime
18
+ from typing import Dict, Any, List, Optional
19
+
20
+ logging.basicConfig(
21
+ level=logging.INFO,
22
+ format="%(asctime)s - %(name)s - %(levelname)s - %(message)s"
23
+ )
24
+ logger = logging.getLogger("test_supabase_tables_final")
25
+
26
+ sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
27
+
28
+ try:
29
+ from memory.supabase_memory import (
30
+ SupabaseMemory,
31
+ WorkingMemory
32
+ )
33
+ from agent.config import get_memory_config
34
+
35
+ from memory.fix_supabase_tables import apply_monkey_patch, ensure_tables_exist
36
+ except ImportError as e:
37
+ logger.error(f"Error importing required components: {str(e)}")
38
+ sys.exit(1)
39
+
40
+ def apply_fix_and_verify():
41
+ """Apply the fix and verify it works correctly."""
42
+ logger.info("Applying fix for Supabase tables...")
43
+
44
+ if not ensure_tables_exist():
45
+ logger.error("Failed to ensure tables exist")
46
+ return False
47
+
48
+ if not apply_monkey_patch():
49
+ logger.error("Failed to apply monkey patch")
50
+ return False
51
+
52
+ logger.info("Fix applied successfully")
53
+ return True
54
+
55
+ def test_working_memory():
56
+ """Test if working memory is using the correct table."""
57
+ logger.info("Testing working memory...")
58
+
59
+ memory_config = get_memory_config()
60
+ memory_config["enabled"] = True
61
+
62
+ base_memory = SupabaseMemory(config=memory_config)
63
+ session_id = f"test-session-{uuid.uuid4()}"
64
+ working_memory = WorkingMemory(base_memory, session_id)
65
+
66
+ logger.info(f"Base memory table: {base_memory.table_name}")
67
+ logger.info(f"Working memory table: {working_memory.table_name}")
68
+
69
+ test_key = f"test_working_{int(time.time())}"
70
+ test_value = {
71
+ "test_data": "This is test data for working memory",
72
+ "timestamp": time.time(),
73
+ "random_id": str(uuid.uuid4())
74
+ }
75
+
76
+ logger.info(f"Storing data in working memory: {test_key}")
77
+ store_result = working_memory.store_intermediate_result(test_key, test_value)
78
+
79
+ if not store_result:
80
+ logger.error("Failed to store test value")
81
+ return False
82
+
83
+ logger.info(f"Retrieving data from working memory: {test_key}")
84
+ retrieved_value = working_memory.get_intermediate_result(test_key)
85
+
86
+ if retrieved_value != test_value:
87
+ logger.error(f"Retrieved value does not match test value")
88
+ logger.error(f"Expected: {test_value}")
89
+ logger.error(f"Got: {retrieved_value}")
90
+ return False
91
+
92
+ logger.info("✅ Store and retrieve operations successful")
93
+
94
+ logger.info("Verifying data was stored in the working memory table...")
95
+
96
+ if base_memory.client and base_memory.initialized:
97
+ try:
98
+ working_key = f"working:{session_id}:{test_key}"
99
+ response = base_memory.client.table(working_memory.table_name).select("*").eq("key", working_key).execute()
100
+
101
+ if not response.data or len(response.data) == 0:
102
+ logger.error("No data found in working memory table")
103
+
104
+ response = base_memory.client.table(base_memory.table_name).select("*").eq("key", working_key).execute()
105
+
106
+ if response.data and len(response.data) > 0:
107
+ logger.error("❌ Data was stored in base memory table instead of working memory table")
108
+ return False
109
+ else:
110
+ logger.error("❌ Data was not stored in any table")
111
+ return False
112
+
113
+ logger.info(f"✅ Data found in working memory table: {len(response.data)} entries")
114
+
115
+ entry_data = response.data[0]
116
+ stored_value = entry_data.get("value")
117
+
118
+ if isinstance(stored_value, str):
119
+ try:
120
+ stored_value = json.loads(stored_value)
121
+ except json.JSONDecodeError:
122
+ pass
123
+
124
+ if stored_value == test_value or stored_value == json.dumps(test_value):
125
+ logger.info("✅ Stored data matches expected value")
126
+ else:
127
+ logger.error("❌ Stored data does not match expected value")
128
+ logger.error(f"Expected: {test_value}")
129
+ logger.error(f"Got: {stored_value}")
130
+ return False
131
+
132
+ logger.info("Cleaning up test data...")
133
+ working_memory.clear()
134
+
135
+ return True
136
+ except Exception as e:
137
+ logger.error(f"Error verifying data storage: {str(e)}")
138
+ return False
139
+
140
+ return False
141
+
142
+ def main():
143
+ """Main entry point for the test script."""
144
+
145
+ if not apply_fix_and_verify():
146
+ sys.exit(1)
147
+
148
+ if test_working_memory():
149
+
150
+ timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
151
+ filename = f"gaia_memory_test_results_{timestamp}.json"
152
+
153
+ results = {
154
+ "timestamp": timestamp,
155
+ "success": True,
156
+ "message": "Working memory table is now being used correctly"
157
+ }
158
+
159
+ with open(filename, "w") as f:
160
+ json.dump(results, f, indent=2)
161
+
162
+ sys.exit(0)
163
+ else:
164
+ sys.exit(1)
165
+
166
+ if __name__ == "__main__":
167
+ main()
backup_tests/test_supabase_tables_fix.py ADDED
@@ -0,0 +1,276 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ Test script for verifying the Supabase tables fix.
3
+
4
+ This script tests if the working memory table (gaia_memory_working) is properly
5
+ utilized by the GAIA agent. It verifies that data is correctly stored in and
6
+ retrieved from the working memory table.
7
+
8
+ Usage:
9
+ python -m tests.test_supabase_tables_fix
10
+ """
11
+
12
+ import os
13
+ import sys
14
+ import json
15
+ import logging
16
+ import time
17
+ import uuid
18
+ from typing import Dict, Any, List, Optional
19
+
20
+ logging.basicConfig(
21
+ level=logging.INFO,
22
+ format="%(asctime)s - %(name)s - %(levelname)s - %(message)s"
23
+ )
24
+ logger = logging.getLogger("test_supabase_tables")
25
+
26
+ sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
27
+
28
+ try:
29
+ from memory.supabase_memory import (
30
+ SupabaseMemory,
31
+ WorkingMemory
32
+ )
33
+ from agent.config import get_memory_config
34
+
35
+ from memory.simplified_memory_fix import apply_fix, create_working_memory_table
36
+ except ImportError as e:
37
+ logger.error(f"Error importing required components: {str(e)}")
38
+ sys.exit(1)
39
+
40
+ class SupabaseTablesTest:
41
+ """Test class for verifying Supabase tables."""
42
+
43
+ def __init__(self):
44
+ """Initialize the test class."""
45
+ self.memory_config = get_memory_config()
46
+ self.memory_config["enabled"] = True
47
+
48
+ self.base_memory = None
49
+ self.working_memory = None
50
+ self.session_id = str(uuid.uuid4())
51
+
52
+ self.table_names = {
53
+ "base": "gaia_memory",
54
+ "working": "gaia_memory_working"
55
+ }
56
+
57
+ self.results = {
58
+ "initialization": False,
59
+ "table_existence": {},
60
+ "working_memory": {}
61
+ }
62
+
63
+ def initialize_memory(self) -> bool:
64
+ """Initialize memory components."""
65
+ logger.info("Initializing memory components...")
66
+
67
+ try:
68
+ self.base_memory = SupabaseMemory(self.memory_config)
69
+
70
+ self.working_memory = WorkingMemory(self.base_memory, self.session_id)
71
+
72
+ if self.base_memory and self.base_memory.initialized:
73
+ logger.info("✅ Memory components initialized successfully")
74
+ self.results["initialization"] = True
75
+ return True
76
+ else:
77
+ logger.error("❌ Memory components initialization failed")
78
+ self.results["initialization"] = False
79
+ return False
80
+
81
+ except Exception as e:
82
+ logger.error(f"Error initializing memory components: {str(e)}")
83
+ self.results["initialization"] = False
84
+ return False
85
+
86
+ def check_table_existence(self) -> Dict[str, bool]:
87
+ """Check if the required tables exist."""
88
+ logger.info("Checking table existence...")
89
+
90
+ results = {}
91
+
92
+ if not self.base_memory or not self.base_memory.initialized:
93
+ logger.error("Memory components not initialized")
94
+ return {"error": "Memory components not initialized"}
95
+
96
+ for table_type, table_name in self.table_names.items():
97
+ try:
98
+ response = self.base_memory.client.table(table_name).select("*").limit(1).execute()
99
+
100
+ if hasattr(response, "data"):
101
+ logger.info(f"✅ Table {table_name} exists")
102
+ results[table_name] = True
103
+ else:
104
+ logger.warning(f"❌ Table {table_name} query failed")
105
+ results[table_name] = False
106
+
107
+ except Exception as e:
108
+ logger.error(f"Error checking table {table_name}: {str(e)}")
109
+ results[table_name] = False
110
+
111
+ self.results["table_existence"] = results
112
+ return results
113
+
114
+ def test_working_memory(self) -> Dict[str, bool]:
115
+ """Test working memory operations."""
116
+ logger.info("Testing working memory operations...")
117
+
118
+ results = {}
119
+
120
+ if not self.working_memory:
121
+ logger.error("Working memory not initialized")
122
+ return {"error": "Working memory not initialized"}
123
+
124
+ try:
125
+ test_key = f"test_working_{int(time.time())}"
126
+ test_value = {
127
+ "test_data": "This is test data for working memory",
128
+ "timestamp": time.time(),
129
+ "random_id": str(uuid.uuid4())
130
+ }
131
+
132
+ logger.info(f"Storing data in working memory: {test_key}")
133
+ store_result = self.working_memory.store_intermediate_result(test_key, test_value)
134
+
135
+ if store_result:
136
+ logger.info("✅ Store operation successful")
137
+ results["store"] = True
138
+ else:
139
+ logger.warning("❌ Store operation failed")
140
+ results["store"] = False
141
+
142
+ logger.info(f"Retrieving data from working memory: {test_key}")
143
+ retrieved_value = self.working_memory.get_intermediate_result(test_key)
144
+
145
+ if retrieved_value and retrieved_value == test_value:
146
+ logger.info("✅ Retrieve operation successful")
147
+ results["retrieve"] = True
148
+ else:
149
+ logger.warning(f"❌ Retrieve operation failed or data mismatch")
150
+ logger.warning(f"Expected: {test_value}")
151
+ logger.warning(f"Got: {retrieved_value}")
152
+ results["retrieve"] = False
153
+
154
+ logger.info("Verifying data was stored in the correct table...")
155
+
156
+ if self.base_memory.client and self.base_memory.initialized:
157
+ try:
158
+ working_key = f"working:{self.session_id}:{test_key}"
159
+ response = self.base_memory.client.table(self.table_names["working"]).select("*").eq("key", working_key).execute()
160
+
161
+ if response.data and len(response.data) > 0:
162
+ logger.info(f"✅ Data found in working memory table: {len(response.data)} entries")
163
+ results["correct_table"] = True
164
+
165
+ entry_data = response.data[0]
166
+ stored_value = entry_data.get("value")
167
+
168
+ if isinstance(stored_value, str):
169
+ try:
170
+ stored_value = json.loads(stored_value)
171
+ except json.JSONDecodeError:
172
+ pass
173
+
174
+ if stored_value == test_value or stored_value == json.dumps(test_value):
175
+ logger.info("✅ Stored data matches expected value")
176
+ results["data_match"] = True
177
+ else:
178
+ logger.warning("❌ Stored data does not match expected value")
179
+ logger.warning(f"Expected: {test_value}")
180
+ logger.warning(f"Got: {stored_value}")
181
+ results["data_match"] = False
182
+ else:
183
+ logger.warning("❌ No data found in working memory table")
184
+ results["correct_table"] = False
185
+
186
+ response = self.base_memory.client.table(self.table_names["base"]).select("*").eq("key", working_key).execute()
187
+
188
+ if response.data and len(response.data) > 0:
189
+ logger.warning("⚠️ Data was stored in base memory table instead of working memory table")
190
+ results["stored_in_base"] = True
191
+ else:
192
+ logger.warning("❌ Data was not stored in any table")
193
+ results["stored_in_base"] = False
194
+
195
+ except Exception as e:
196
+ logger.error(f"Error verifying data storage: {str(e)}")
197
+ results["correct_table"] = False
198
+
199
+ logger.info("Cleaning up test data...")
200
+ self.working_memory.clear()
201
+
202
+ except Exception as e:
203
+ logger.error(f"Error testing working memory: {str(e)}")
204
+ results["error"] = str(e)
205
+
206
+ self.results["working_memory"] = results
207
+ return results
208
+
209
+ def run_all_tests(self) -> Dict[str, Any]:
210
+ """Run all tests."""
211
+ logger.info("Running all tests...")
212
+
213
+ if not self.initialize_memory():
214
+ logger.error("Memory initialization failed, aborting tests")
215
+ return self.results
216
+
217
+ self.check_table_existence()
218
+
219
+ if self.results["table_existence"].get(self.table_names["working"], False):
220
+ logger.info("Working memory table exists, applying fix...")
221
+ apply_fix()
222
+ else:
223
+ logger.info("Working memory table does not exist, creating it...")
224
+ create_working_memory_table()
225
+
226
+ self.check_table_existence()
227
+
228
+ self.test_working_memory()
229
+
230
+ return self.results
231
+
232
+ def print_test_summary(self) -> None:
233
+ """Print a summary of the test results."""
234
+
235
+ if self.results["initialization"]:
236
+ else:
237
+
238
+ for table_name, exists in self.results["table_existence"].items():
239
+ if exists:
240
+ else:
241
+
242
+ working_results = self.results["working_memory"]
243
+
244
+ if working_results.get("store", False):
245
+ else:
246
+
247
+ if working_results.get("retrieve", False):
248
+ else:
249
+
250
+ if working_results.get("correct_table", False):
251
+ else:
252
+
253
+ if working_results.get("stored_in_base", False):
254
+
255
+ if working_results.get("data_match", False):
256
+ else:
257
+
258
+ if (self.results["initialization"] and
259
+ all(self.results["table_existence"].values()) and
260
+ working_results.get("store", False) and
261
+ working_results.get("retrieve", False) and
262
+ working_results.get("correct_table", False) and
263
+ working_results.get("data_match", False)):
264
+ else:
265
+
266
+
267
+ def main():
268
+ """Main entry point for the test script."""
269
+
270
+ test = SupabaseTablesTest()
271
+ test.run_all_tests()
272
+
273
+ test.print_test_summary()
274
+
275
+ if __name__ == "__main__":
276
+ main()
backup_tests/test_supabase_tables_production.py ADDED
@@ -0,0 +1,481 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ Comprehensive Unit Tests for Supabase Tables
3
+
4
+ This script tests all Supabase tables used by the GAIA agent:
5
+ - gaia_memory: Main memory table
6
+ - gaia_memory_working: Working memory table
7
+ - gaia_memory_conversations: Conversation history table
8
+ - gaia_memory_cache: Result cache table
9
+
10
+ The tests verify:
11
+ 1. Each table is created correctly
12
+ 2. Each memory class uses its dedicated table
13
+ 3. Data is properly isolated between tables
14
+ 4. Performance metrics are captured
15
+ 5. All test data is cleaned up after tests
16
+
17
+ Usage:
18
+ python tests/test_supabase_tables_production.py
19
+ """
20
+
21
+ import os
22
+ import sys
23
+ import json
24
+ import time
25
+ import uuid
26
+ import logging
27
+ import unittest
28
+ from datetime import datetime, timedelta
29
+ from typing import Dict, Any, List, Optional, Union, Tuple
30
+
31
+ logging.basicConfig(
32
+ level=logging.INFO,
33
+ format="%(asctime)s - %(name)s - %(levelname)s - %(message)s"
34
+ )
35
+ logger = logging.getLogger("test_supabase_tables_production")
36
+
37
+ parent_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
38
+ sys.path.insert(0, parent_dir)
39
+
40
+ try:
41
+ from supabase import create_client, Client
42
+ except ImportError as e:
43
+ logger.error(f"Error importing supabase: {str(e)}")
44
+ sys.exit(1)
45
+
46
+ try:
47
+ from agent.config import (
48
+ SUPABASE_URL,
49
+ SUPABASE_KEY,
50
+ SUPABASE_PROJECT_ID,
51
+ SUPABASE_PROJECT_NAME
52
+ )
53
+ except ImportError as e:
54
+ logger.error(f"Error importing configuration: {str(e)}")
55
+ sys.exit(1)
56
+
57
+ def get_table_schema(client: Client, table_name: str) -> List[str]:
58
+ """Get the schema of a table."""
59
+ try:
60
+ result = client.table(table_name).select("*").limit(1).execute()
61
+
62
+ if result.data and len(result.data) > 0:
63
+ return list(result.data[0].keys())
64
+ else:
65
+ schema_result = client.from_("information_schema.columns").select("column_name").eq("table_name", table_name).execute()
66
+ if schema_result.data:
67
+ return [col["column_name"] for col in schema_result.data]
68
+ else:
69
+ return []
70
+ except Exception as e:
71
+ logger.error(f"Error getting schema for table {table_name}: {str(e)}")
72
+ return []
73
+
74
+ def ensure_table_exists(client: Client, table_name: str) -> bool:
75
+ """Ensure a table exists in Supabase."""
76
+ try:
77
+ client.table(table_name).select("*").limit(1).execute()
78
+ logger.info(f"Table {table_name} exists")
79
+ return True
80
+ except Exception as e:
81
+ logger.warning(f"Error checking table {table_name}: {str(e)}")
82
+ logger.info(f"Creating table {table_name}")
83
+
84
+ try:
85
+ sql = f"""
86
+ CREATE TABLE IF NOT EXISTS {table_name} (
87
+ id UUID PRIMARY KEY DEFAULT uuid_generate_v4(),
88
+ key TEXT NOT NULL UNIQUE,
89
+ value TEXT NOT NULL,
90
+ metadata JSONB,
91
+ created_at TIMESTAMPTZ NOT NULL DEFAULT NOW(),
92
+ expires_at TIMESTAMPTZ
93
+ );
94
+
95
+ CREATE INDEX IF NOT EXISTS idx_{table_name}_key ON {table_name}(key);
96
+ CREATE INDEX IF NOT EXISTS idx_{table_name}_expires_at ON {table_name}(expires_at);
97
+ """
98
+
99
+ client.query(sql).execute()
100
+
101
+ client.table(table_name).select("*").limit(1).execute()
102
+
103
+ logger.info(f"Table {table_name} created successfully")
104
+ return True
105
+ except Exception as e:
106
+ logger.error(f"Error creating table {table_name}: {str(e)}")
107
+ return False
108
+
109
+ def store_value(client: Client, table_name: str, key: str, value: Any, metadata: Optional[Dict[str, Any]] = None, ttl: Optional[int] = None, session_id: Optional[str] = None) -> bool:
110
+ """Store a value in a table."""
111
+ try:
112
+ entry = {
113
+ "key": key,
114
+ "value": json.dumps(value) if not isinstance(value, str) else value,
115
+ "metadata": metadata or {}
116
+ }
117
+
118
+ if ttl:
119
+ entry["expires_at"] = (datetime.now() + timedelta(seconds=ttl)).isoformat()
120
+
121
+ if session_id and table_name.endswith("_working"):
122
+ entry["session_id"] = session_id
123
+
124
+ client.table(table_name).upsert(entry).execute()
125
+
126
+ logger.info(f"Stored value in table {table_name}: {key}")
127
+ return True
128
+ except Exception as e:
129
+ logger.error(f"Error storing value in table {table_name}: {key} - {str(e)}")
130
+ return False
131
+
132
+ def retrieve_value(client: Client, table_name: str, key: str, default: Any = None) -> Any:
133
+ """Retrieve a value from a table."""
134
+ try:
135
+ result = client.table(table_name).select("*").eq("key", key).execute()
136
+
137
+ if result.data and len(result.data) > 0:
138
+ entry = result.data[0]
139
+
140
+ value = entry.get("value")
141
+ if isinstance(value, str):
142
+ try:
143
+ value = json.loads(value)
144
+ except json.JSONDecodeError:
145
+ pass
146
+
147
+ expires_at = entry.get("expires_at")
148
+ if expires_at:
149
+ try:
150
+ expiry = datetime.fromisoformat(expires_at)
151
+ if expiry < datetime.now():
152
+ delete_value(client, table_name, key)
153
+ return default
154
+ except (ValueError, TypeError):
155
+ pass
156
+
157
+ logger.info(f"Retrieved value from table {table_name}: {key}")
158
+ return value
159
+ except Exception as e:
160
+ logger.error(f"Error retrieving value from table {table_name}: {key} - {str(e)}")
161
+
162
+ return default
163
+
164
+ def delete_value(client: Client, table_name: str, key: str) -> bool:
165
+ """Delete a value from a table."""
166
+ try:
167
+ client.table(table_name).delete().eq("key", key).execute()
168
+
169
+ logger.info(f"Deleted value from table {table_name}: {key}")
170
+ return True
171
+ except Exception as e:
172
+ logger.error(f"Error deleting value from table {table_name}: {key} - {str(e)}")
173
+ return False
174
+
175
+ def list_keys(client: Client, table_name: str, prefix: Optional[str] = None) -> List[str]:
176
+ """List keys in a table."""
177
+ keys = []
178
+
179
+ try:
180
+ query = client.table(table_name).select("key")
181
+
182
+ if prefix:
183
+ query = query.like("key", f"{prefix}%")
184
+
185
+ result = query.execute()
186
+
187
+ if result.data:
188
+ for entry in result.data:
189
+ key = entry.get("key")
190
+ if key:
191
+ keys.append(key)
192
+
193
+ logger.info(f"Listed keys from table {table_name}: {len(keys)} keys")
194
+ except Exception as e:
195
+ logger.error(f"Error listing keys from table {table_name}: {str(e)}")
196
+
197
+ return keys
198
+
199
+ def clear_table(client: Client, table_name: str, prefix: Optional[str] = None) -> bool:
200
+ """Clear a table."""
201
+ try:
202
+ if prefix:
203
+ keys = list_keys(client, table_name, prefix)
204
+
205
+ for key in keys:
206
+ delete_value(client, table_name, key)
207
+ else:
208
+ client.table(table_name).delete().execute()
209
+
210
+ logger.info(f"Cleared table {table_name}")
211
+ return True
212
+ except Exception as e:
213
+ logger.error(f"Error clearing table {table_name}: {str(e)}")
214
+ return False
215
+
216
+ def run_tests():
217
+ """Run the tests."""
218
+ logger.info("Running Supabase tables tests...")
219
+
220
+ if not SUPABASE_URL or not SUPABASE_KEY:
221
+ logger.error("Supabase URL or key not provided")
222
+ logger.error("Please set SUPABASE_URL and SUPABASE_KEY environment variables")
223
+ return False
224
+
225
+ # Initialize Supabase client
226
+ client = create_client(SUPABASE_URL, SUPABASE_KEY)
227
+
228
+ test_run_id = str(uuid.uuid4())
229
+ session_id = f"test-session-{test_run_id}"
230
+ conversation_id = f"test-conversation-{test_run_id}"
231
+
232
+ base_table = "gaia_memory"
233
+ tables = {
234
+ "base": base_table,
235
+ "working": f"{base_table}_working",
236
+ "conversations": f"{base_table}_conversations",
237
+ "cache": f"{base_table}_cache"
238
+ }
239
+
240
+ for table_type, table_name in tables.items():
241
+ if not ensure_table_exists(client, table_name):
242
+ logger.error(f"Failed to ensure table {table_name} exists")
243
+ return False
244
+
245
+ schema = get_table_schema(client, table_name)
246
+ logger.info(f"Table {table_name} schema: {schema}")
247
+
248
+ if table_type == "conversations":
249
+ if "conversation_id" in schema and "message_id" in schema and "content" in schema:
250
+ logger.info(f"Table {table_name} has a different schema for conversations")
251
+
252
+ logger.info("Testing base table operations...")
253
+
254
+ test_key = f"test-{test_run_id}-base"
255
+ test_value = {
256
+ "test_data": f"This is test data for {test_key}",
257
+ "timestamp": datetime.now().isoformat(),
258
+ "random_id": str(uuid.uuid4())
259
+ }
260
+
261
+ if not store_value(client, tables["base"], test_key, test_value):
262
+ logger.error("Failed to store test value in base table")
263
+ return False
264
+
265
+ retrieved_value = retrieve_value(client, tables["base"], test_key)
266
+ if retrieved_value != test_value:
267
+ logger.error("Retrieved value does not match test value")
268
+ logger.error(f"Expected: {test_value}")
269
+ logger.error(f"Got: {retrieved_value}")
270
+ return False
271
+
272
+ logger.info("[PASS] Base table operations successful")
273
+
274
+ logger.info("Testing working memory table operations...")
275
+
276
+ working_key = f"working:{session_id}:test-{test_run_id}-working"
277
+ working_value = {
278
+ "test_data": f"This is test data for {working_key}",
279
+ "timestamp": datetime.now().isoformat(),
280
+ "random_id": str(uuid.uuid4())
281
+ }
282
+
283
+ if not store_value(client, tables["working"], working_key, working_value, session_id=session_id):
284
+ logger.error("Failed to store test value in working table")
285
+ return False
286
+
287
+ retrieved_value = retrieve_value(client, tables["working"], working_key)
288
+ if retrieved_value != working_value:
289
+ logger.error("Retrieved value does not match test value")
290
+ logger.error(f"Expected: {working_value}")
291
+ logger.error(f"Got: {retrieved_value}")
292
+ return False
293
+
294
+ base_value = retrieve_value(client, tables["base"], working_key)
295
+ if base_value is not None:
296
+ logger.error("Working memory data was incorrectly stored in base table")
297
+ return False
298
+
299
+ logger.info("[PASS] Working memory table operations successful")
300
+
301
+ logger.info("Testing conversation memory table operations...")
302
+
303
+ schema = get_table_schema(client, tables["conversations"])
304
+
305
+ if "key" in schema and "value" in schema:
306
+ conversation_key = f"conversation:{conversation_id}:message:{datetime.now().isoformat()}"
307
+ conversation_value = {
308
+ "role": "user",
309
+ "content": "Hello, how are you?",
310
+ "timestamp": datetime.now().isoformat()
311
+ }
312
+
313
+ if not store_value(client, tables["conversations"], conversation_key, conversation_value):
314
+ logger.error("Failed to store test value in conversations table")
315
+ return False
316
+
317
+ retrieved_value = retrieve_value(client, tables["conversations"], conversation_key)
318
+ if retrieved_value != conversation_value:
319
+ logger.error("Retrieved value does not match test value")
320
+ logger.error(f"Expected: {conversation_value}")
321
+ logger.error(f"Got: {retrieved_value}")
322
+ return False
323
+
324
+ base_value = retrieve_value(client, tables["base"], conversation_key)
325
+ if base_value is not None:
326
+ logger.error("Conversation memory data was incorrectly stored in base table")
327
+ return False
328
+ else:
329
+ logger.info("Using different schema for conversations table")
330
+
331
+ try:
332
+ message_id = f"message-{datetime.now().isoformat()}"
333
+ entry = {
334
+ "conversation_id": conversation_id,
335
+ "message_id": message_id,
336
+ "role": "user",
337
+ "content": "Hello, how are you?",
338
+ "created_at": datetime.now().isoformat()
339
+ }
340
+
341
+ client.table(tables["conversations"]).insert(entry).execute()
342
+ logger.info(f"Inserted message in conversations table: {message_id}")
343
+
344
+ result = client.table(tables["conversations"]).select("*").eq("conversation_id", conversation_id).execute()
345
+
346
+ if not result.data or len(result.data) == 0:
347
+ logger.error("No messages found in conversations table")
348
+ return False
349
+
350
+ logger.info(f"Found {len(result.data)} messages in conversations table")
351
+ except Exception as e:
352
+ logger.error(f"Error testing conversations table: {str(e)}")
353
+ return False
354
+
355
+ logger.info("[PASS] Conversation memory table operations successful")
356
+
357
+ logger.info("Testing result cache table operations...")
358
+
359
+ cache_key = f"cache:test-{test_run_id}-cache"
360
+ cache_value = {
361
+ "test_data": f"This is test data for {cache_key}",
362
+ "timestamp": datetime.now().isoformat(),
363
+ "random_id": str(uuid.uuid4())
364
+ }
365
+
366
+ if not store_value(client, tables["cache"], cache_key, cache_value, ttl=3600):
367
+ logger.error("Failed to store test value in cache table")
368
+ return False
369
+
370
+ retrieved_value = retrieve_value(client, tables["cache"], cache_key)
371
+ if retrieved_value != cache_value:
372
+ logger.error("Retrieved value does not match test value")
373
+ logger.error(f"Expected: {cache_value}")
374
+ logger.error(f"Got: {retrieved_value}")
375
+ return False
376
+
377
+ base_value = retrieve_value(client, tables["base"], cache_key)
378
+ if base_value is not None:
379
+ logger.error("Result cache data was incorrectly stored in base table")
380
+ return False
381
+
382
+ logger.info("[PASS] Result cache table operations successful")
383
+
384
+ logger.info("Testing data isolation between tables...")
385
+
386
+ base_keys = list_keys(client, tables["base"])
387
+ working_keys = list_keys(client, tables["working"])
388
+ cache_keys = list_keys(client, tables["cache"])
389
+
390
+ schema = get_table_schema(client, tables["conversations"])
391
+ if "key" in schema:
392
+ conversation_keys = list_keys(client, tables["conversations"])
393
+
394
+ if conversation_key not in conversation_keys:
395
+ logger.error(f"Conversation key {conversation_key} not found in conversations table")
396
+ return False
397
+ else:
398
+ try:
399
+ result = client.table(tables["conversations"]).select("*").eq("conversation_id", conversation_id).execute()
400
+ if not result.data or len(result.data) == 0:
401
+ logger.error(f"No messages found for conversation ID {conversation_id}")
402
+ return False
403
+ logger.info(f"Found {len(result.data)} messages for conversation ID {conversation_id}")
404
+ except Exception as e:
405
+ logger.error(f"Error checking conversations table: {str(e)}")
406
+ return False
407
+
408
+ if test_key not in base_keys:
409
+ logger.error(f"Test key {test_key} not found in base table")
410
+ return False
411
+
412
+ if working_key not in working_keys:
413
+ logger.error(f"Working key {working_key} not found in working table")
414
+ return False
415
+
416
+ if cache_key not in cache_keys:
417
+ logger.error(f"Cache key {cache_key} not found in cache table")
418
+ return False
419
+
420
+ logger.info("[PASS] Data isolation between tables verified")
421
+
422
+ logger.info("Testing performance metrics...")
423
+
424
+ num_operations = 10
425
+
426
+ base_store_times = []
427
+ base_retrieve_times = []
428
+
429
+ for i in range(num_operations):
430
+ key = f"test-{test_run_id}-perf-base-{i}"
431
+ value = {"index": i, "data": f"Performance test data {i}"}
432
+
433
+ start_time = time.time()
434
+ store_value(client, tables["base"], key, value)
435
+ base_store_times.append(time.time() - start_time)
436
+
437
+ start_time = time.time()
438
+ retrieve_value(client, tables["base"], key)
439
+ base_retrieve_times.append(time.time() - start_time)
440
+
441
+ working_store_times = []
442
+ working_retrieve_times = []
443
+
444
+ for i in range(num_operations):
445
+ key = f"working:{session_id}:test-{test_run_id}-perf-working-{i}"
446
+ value = {"index": i, "data": f"Performance test data {i}"}
447
+
448
+ start_time = time.time()
449
+ store_value(client, tables["working"], key, value, session_id=session_id)
450
+ working_store_times.append(time.time() - start_time)
451
+
452
+ start_time = time.time()
453
+ retrieve_value(client, tables["working"], key)
454
+ working_retrieve_times.append(time.time() - start_time)
455
+
456
+ logger.info(f"Base table: avg store {sum(base_store_times)/num_operations:.4f}s, avg retrieve {sum(base_retrieve_times)/num_operations:.4f}s")
457
+ logger.info(f"Working table: avg store {sum(working_store_times)/num_operations:.4f}s, avg retrieve {sum(working_retrieve_times)/num_operations:.4f}s")
458
+
459
+ logger.info("[PASS] Performance metrics collected")
460
+
461
+ logger.info("Cleaning up test data...")
462
+
463
+ clear_table(client, tables["base"], f"test-{test_run_id}")
464
+ clear_table(client, tables["working"], f"working:{session_id}")
465
+ clear_table(client, tables["conversations"], f"conversation:{conversation_id}")
466
+ clear_table(client, tables["cache"], f"cache:test-{test_run_id}")
467
+
468
+ logger.info("[PASS] Test data cleaned up")
469
+
470
+ return True
471
+
472
+ def main():
473
+ """Main entry point for the test script."""
474
+
475
+ if run_tests():
476
+ sys.exit(0)
477
+ else:
478
+ sys.exit(1)
479
+
480
+ if __name__ == "__main__":
481
+ main()
cleanup_results.txt ADDED
@@ -0,0 +1,340 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2025-04-30 22:18:05,446 - INFO - Finding temporary files to remove...
2
+ 2025-04-30 22:18:05,563 - INFO - Found 39 temporary files to remove
3
+ 2025-04-30 22:18:05,603 - INFO - Removed: gaia_assessment_results_20250430_205848.json
4
+ 2025-04-30 22:18:05,615 - INFO - Removed: gaia_assessment_results_20250430_210047.json
5
+ 2025-04-30 22:18:05,616 - INFO - Removed: gaia_assessment_results_20250430_210252.json
6
+ 2025-04-30 22:18:05,619 - INFO - Removed: gaia_assessment_results_20250430_210521.json
7
+ 2025-04-30 22:18:05,621 - INFO - Removed: gaia_assessment_results_20250430_211034.json
8
+ 2025-04-30 22:18:05,622 - INFO - Removed: gaia_assessment_results_20250430_211508.json
9
+ 2025-04-30 22:18:05,624 - INFO - Removed: gaia_assessment_results_20250430_213743.json
10
+ 2025-04-30 22:18:05,627 - INFO - Removed: gaia_assessment_results_20250430_214034.json
11
+ 2025-04-30 22:18:05,629 - INFO - Removed: gaia_assessment_results_20250430_214159.json
12
+ 2025-04-30 22:18:05,631 - INFO - Removed: gaia_assessment_results_20250430_214711.json
13
+ 2025-04-30 22:18:05,634 - INFO - Removed: gaia_assessment_results_20250430_220729.json
14
+ 2025-04-30 22:18:05,636 - INFO - Removed: gaia_memory_test_results_20250430_185256.json
15
+ 2025-04-30 22:18:05,639 - INFO - Removed: gaia_memory_test_results_20250430_185420.json
16
+ 2025-04-30 22:18:05,641 - INFO - Removed: gaia_memory_test_results_20250430_185453.json
17
+ 2025-04-30 22:18:05,644 - INFO - Removed: gaia_memory_test_results_20250430_185550.json
18
+ 2025-04-30 22:18:05,648 - INFO - Removed: gaia_memory_test_results_20250430_185641.json
19
+ 2025-04-30 22:18:05,671 - INFO - Removed: cleanup_script.bat
20
+ 2025-04-30 22:18:05,672 - INFO - Removed: check_env_vars.py
21
+ 2025-04-30 22:18:05,675 - INFO - Removed: check_imports.py
22
+ 2025-04-30 22:18:05,678 - INFO - Removed: check_serper_config.py
23
+ 2025-04-30 22:18:05,679 - INFO - Removed: check_supabase_env.py
24
+ 2025-04-30 22:18:05,681 - INFO - Removed: direct_config_check.py
25
+ 2025-04-30 22:18:05,683 - INFO - Removed: direct_serper_test.py
26
+ 2025-04-30 22:18:05,685 - INFO - Removed: minimal_serper_test.py
27
+ 2025-04-30 22:18:05,686 - INFO - Removed: simple_serper_test.py
28
+ 2025-04-30 22:18:05,689 - INFO - Removed: simple_serper_test_final.py
29
+ 2025-04-30 22:18:05,693 - INFO - Removed: supabase_direct_test.py
30
+ 2025-04-30 22:18:05,695 - INFO - Removed: supabase_dotenv_test.py
31
+ 2025-04-30 22:18:05,697 - INFO - Removed: supabase_import_test.py
32
+ 2025-04-30 22:18:05,698 - INFO - Removed: supabase_init_test.py
33
+ 2025-04-30 22:18:05,700 - INFO - Removed: supabase_minimal_test.py
34
+ 2025-04-30 22:18:05,702 - INFO - Removed: test_serper_direct.py
35
+ 2025-04-30 22:18:05,704 - INFO - Removed: verify_python.py
36
+ 2025-04-30 22:18:05,706 - INFO - Removed: run_gaia_assessment.py
37
+ 2025-04-30 22:18:05,707 - INFO - Removed: run_gaia_assessment_simple.py
38
+ 2025-04-30 22:18:05,709 - INFO - Removed: run_gaia_fixed.py
39
+ 2025-04-30 22:18:05,713 - INFO - Removed: run_gaia_minimal.py
40
+ 2025-04-30 22:18:05,714 - INFO - Removed: run_gaia_with_fix.py
41
+ 2025-04-30 22:18:05,717 - INFO - Removed: setup_serper_env.py
42
+ 2025-04-30 22:18:05,717 - INFO - Finding duplicate test files...
43
+ 2025-04-30 22:18:05,720 - INFO - Found 7 duplicate test files
44
+ 2025-04-30 22:18:05,720 - INFO - Duplicate tests for test_config.py:
45
+ 2025-04-30 22:18:05,720 - INFO - - tests\test_config.py
46
+ 2025-04-30 22:18:05,720 - INFO - - tests\test_config_simple.py
47
+ 2025-04-30 22:18:05,721 - INFO - Keeping: tests\test_config.py
48
+ 2025-04-30 22:18:05,721 - INFO - Removing: tests\test_config_simple.py
49
+ 2025-04-30 22:18:05,722 - INFO - Removed: tests\test_config_simple.py
50
+ 2025-04-30 22:18:05,722 - INFO - Duplicate tests for test_serper_api.py:
51
+ 2025-04-30 22:18:05,723 - INFO - - tests\test_serper_api.py
52
+ 2025-04-30 22:18:05,723 - INFO - - tests\test_serper_api_final.py
53
+ 2025-04-30 22:18:05,723 - INFO - Keeping: tests\test_serper_api_final.py
54
+ 2025-04-30 22:18:05,723 - INFO - Removing: tests\test_serper_api.py
55
+ 2025-04-30 22:18:05,724 - INFO - Removed: tests\test_serper_api.py
56
+ 2025-04-30 22:18:05,726 - INFO - Duplicate tests for test_supabase_connection.py:
57
+ 2025-04-30 22:18:05,726 - INFO - - tests\test_supabase_connection.py
58
+ 2025-04-30 22:18:05,726 - INFO - - tests\test_supabase_connection_simple.py
59
+ 2025-04-30 22:18:05,727 - INFO - Keeping: tests\test_supabase_connection.py
60
+ 2025-04-30 22:18:05,727 - INFO - Removing: tests\test_supabase_connection_simple.py
61
+ 2025-04-30 22:18:05,728 - INFO - Removed: tests\test_supabase_connection_simple.py
62
+ 2025-04-30 22:18:05,729 - INFO - Duplicate tests for test_supabase_tables.py:
63
+ 2025-04-30 22:18:05,729 - INFO - - tests\test_supabase_tables.py
64
+ 2025-04-30 22:18:05,729 - INFO - - tests\test_supabase_tables_comprehensive.py
65
+ 2025-04-30 22:18:05,729 - INFO - - tests\test_supabase_tables_final.py
66
+ 2025-04-30 22:18:05,729 - INFO - - tests\test_supabase_tables_production.py
67
+ 2025-04-30 22:18:05,729 - INFO - - tests\test_supabase_tables_simple.py
68
+ 2025-04-30 22:18:05,730 - INFO - Keeping: tests\test_supabase_tables_comprehensive.py
69
+ 2025-04-30 22:18:05,730 - INFO - Removing: tests\test_supabase_tables.py, tests\test_supabase_tables_simple.py
70
+ 2025-04-30 22:18:05,731 - INFO - Removed: tests\test_supabase_tables.py
71
+ 2025-04-30 22:18:05,733 - INFO - Removed: tests\test_supabase_tables_simple.py
72
+ 2025-04-30 22:18:05,733 - INFO - Checking for debug statements and commented-out code...
73
+ 2025-04-30 22:18:05,803 - INFO - Found 248 instances of debug statements or commented-out code
74
+ 2025-04-30 22:18:05,803 - INFO - Issues in agent\agent.py:
75
+ 2025-04-30 22:18:05,803 - INFO - Line 72 (Commented code): # Set up verbose logging if enabled
76
+ 2025-04-30 22:18:05,803 - INFO - Line 75 (Commented code): # Add console handler if not already present
77
+ 2025-04-30 22:18:05,803 - INFO - Line 92 (Debug statement): logger.debug("Initializing memory components")
78
+ 2025-04-30 22:18:05,803 - INFO - Line 98 (Commented code): # Create a unique session ID for this agent instance
79
+ 2025-04-30 22:18:05,803 - INFO - Line 106 (Debug statement): logger.debug("Memory components initialized successfully")
80
+ 2025-04-30 22:18:05,804 - INFO - Line 161 (Commented code): # Record end time and return the answer
81
+ 2025-04-30 22:18:05,804 - INFO - Line 207 (Debug statement): logger.debug(f"Processing question: {question}")
82
+ 2025-04-30 22:18:05,804 - INFO - Line 209 (Commented code): # Check if we have a cached answer
83
+ 2025-04-30 22:18:05,804 - INFO - Line 215 (Commented code): # Check if we have a cached answer
84
+ 2025-04-30 22:18:05,804 - INFO - Line 221 (Commented code): # Create a config for the graph
85
+ 2025-04-30 22:18:05,804 - INFO - Line 237 (Commented code): # Fallback answer if no answer or error is provided
86
+ 2025-04-30 22:18:05,804 - INFO - Line 240 (Commented code): # Store the answer in the result cache if available
87
+ 2025-04-30 22:18:05,804 - INFO - Line 244 (Commented code): # Store the intermediate results in working memory if available
88
+ 2025-04-30 22:18:05,804 - INFO - Line 264 (Commented code): # Add the question and answer to conversation memory if available
89
+ 2025-04-30 22:18:05,804 - INFO - Line 291 (Commented code): # Get conversation history if available
90
+ 2025-04-30 22:18:05,804 - INFO - Line 295 (Commented code): # Get cached results if available
91
+ 2025-04-30 22:18:05,805 - INFO - Line 308 (Commented code): # Get intermediate results if available
92
+ 2025-04-30 22:18:05,805 - INFO - Line 323 (Commented code): # Clear conversation memory if available
93
+ 2025-04-30 22:18:05,805 - INFO - Line 327 (Commented code): # Clear result cache if available
94
+ 2025-04-30 22:18:05,805 - INFO - Line 331 (Commented code): # Clear working memory if available
95
+ 2025-04-30 22:18:05,805 - INFO - Issues in agent\config.py:
96
+ 2025-04-30 22:18:05,805 - INFO - Line 20 (Commented code): # Try to import dotenv, but don't fail if it's not available
97
+ 2025-04-30 22:18:05,805 - INFO - Line 23 (Commented code): # Load environment variables from .env file if it exists
98
+ 2025-04-30 22:18:05,805 - INFO - Line 28 (Debug statement): print("Warning: python-dotenv not installed. Environment variables will not be loaded from .env file.")
99
+ 2025-04-30 22:18:05,806 - INFO - Line 30 (Commented code): # API Keys and Credentials (kept as environment variables)
100
+ 2025-04-30 22:18:05,806 - INFO - Line 37 (Commented code): # Supabase Configuration (partially hardcoded)
101
+ 2025-04-30 22:18:05,806 - INFO - Line 42 (Commented code): # API Endpoints (hardcoded)
102
+ 2025-04-30 22:18:05,806 - INFO - Line 48 (Commented code): # Model Configuration (hardcoded with updated models)
103
+ 2025-04-30 22:18:05,806 - INFO - Line 57 (Commented code): # Tool Parameters (hardcoded)
104
+ 2025-04-30 22:18:05,806 - INFO - Line 69 (Commented code): # Memory Settings (hardcoded)
105
+ 2025-04-30 22:18:05,806 - INFO - Line 77 (Commented code): # Agent Settings (hardcoded)
106
+ 2025-04-30 22:18:05,806 - INFO - Line 85 (Commented code): # Logging Configuration (hardcoded)
107
+ 2025-04-30 22:18:05,807 - INFO - Line 90 (Commented code): # Multimodal Settings (hardcoded)
108
+ 2025-04-30 22:18:05,807 - INFO - Line 95 (Commented code): # Web Browsing Settings (hardcoded)
109
+ 2025-04-30 22:18:05,807 - INFO - Line 101 (Commented code): # Helper functions for configuration
110
+ 2025-04-30 22:18:05,807 - INFO - Issues in agent\graph.py:
111
+ 2025-04-30 22:18:05,807 - INFO - Line 30 (Commented code): # Type definitions for state
112
+ 2025-04-30 22:18:05,807 - INFO - Line 75 (Debug statement): logger.debug(f"Analyzing question: {state['question']}")
113
+ 2025-04-30 22:18:05,807 - INFO - Line 117 (Commented code): # Parse the analysis (in a real implementation, use a proper JSON parser)
114
+ 2025-04-30 22:18:05,807 - INFO - Line 118 (Commented code): # This is a simplified version for demonstration
115
+ 2025-04-30 22:18:05,807 - INFO - Line 126 (Debug statement): logger.debug(f"Question analysis completed: {analysis}")
116
+ 2025-04-30 22:18:05,807 - INFO - Line 148 (Debug statement): logger.debug("Creating plan based on question analysis")
117
+ 2025-04-30 22:18:05,808 - INFO - Line 150 (Commented code): # If there's an error, return the state unchanged
118
+ 2025-04-30 22:18:05,808 - INFO - Line 205 (Commented code): # Parse the plan (in a real implementation, use a proper JSON parser)
119
+ 2025-04-30 22:18:05,808 - INFO - Line 222 (Debug statement): logger.debug(f"Plan created with {len(plan)} steps")
120
+ 2025-04-30 22:18:05,809 - INFO - Line 244 (Debug statement): logger.debug(f"Selecting tool for step {state['current_step'] + 1}")
121
+ 2025-04-30 22:18:05,809 - INFO - Line 246 (Commented code): # If there's an error, return END
122
+ 2025-04-30 22:18:05,809 - INFO - Line 281 (Commented code): # If there's an error, return the state unchanged
123
+ 2025-04-30 22:18:05,809 - INFO - Line 290 (Debug statement): logger.debug(f"Executing tool {tool_name} for step {current_step['step_number']}")
124
+ 2025-04-30 22:18:05,809 - INFO - Line 315 (Debug statement): logger.debug(f"Tool {tool_name} executed successfully")
125
+ 2025-04-30 22:18:05,810 - INFO - Line 350 (Debug statement): logger.debug("Formulating answer based on tool results")
126
+ 2025-04-30 22:18:05,810 - INFO - Line 352 (Commented code): # If there's an error, return the state unchanged
127
+ 2025-04-30 22:18:05,810 - INFO - Line 409 (Debug statement): logger.debug(f"Answer formulated: {answer.strip()}")
128
+ 2025-04-30 22:18:05,810 - INFO - Line 431 (Commented code): # If there's an error, check if we should retry
129
+ 2025-04-30 22:18:05,810 - INFO - Issues in agent\tool_registry.py:
130
+ 2025-04-30 22:18:05,810 - INFO - Line 34 (Debug statement): logger.debug("Initialized empty tool registry")
131
+ 2025-04-30 22:18:05,811 - INFO - Line 45 (Debug statement): logger.debug(f"Registered tool: {name}")
132
+ 2025-04-30 22:18:05,811 - INFO - Line 89 (Debug statement): logger.debug(f"Executing tool: {name} with args: {kwargs}")
133
+ 2025-04-30 22:18:05,811 - INFO - Line 147 (Commented code): # Register DuckDuckGo search tool (no API key required)
134
+ 2025-04-30 22:18:05,811 - INFO - Line 154 (Commented code): # Register Serper search tool if API key is available
135
+ 2025-04-30 22:18:05,811 - INFO - Line 165 (Commented code): # Register Perplexity tool if API key is available
136
+ 2025-04-30 22:18:05,811 - INFO - Line 176 (Commented code): # Register arXiv tool (no API key required)
137
+ 2025-04-30 22:18:05,812 - INFO - Line 180 (Commented code): # Register the same tool instance for different operations
138
+ 2025-04-30 22:18:05,812 - INFO - Line 207 (Commented code): # Try DuckDuckGo first (no API key required)
139
+ 2025-04-30 22:18:05,812 - INFO - Line 334 (Debug statement): print(f"Available tools: {registry.list_tools()}")
140
+ 2025-04-30 22:18:05,812 - INFO - Line 340 (Debug statement): print(f"Robust search using {result['provider']} found {len(result['results'])} results")
141
+ 2025-04-30 22:18:05,812 - INFO - Line 342 (Debug statement): print(f"Robust search failed: {str(e)}")
142
+ 2025-04-30 22:18:05,812 - INFO - Line 348 (Debug statement): print(f"Merged search found {len(result['merged_results'])} results")
143
+ 2025-04-30 22:18:05,813 - INFO - Line 350 (Debug statement): print(f"Perplexity context: {result['perplexity_context'][:100]}...")
144
+ 2025-04-30 22:18:05,813 - INFO - Line 352 (Debug statement): print(f"arXiv papers: {len(result['arxiv_context'])}")
145
+ 2025-04-30 22:18:05,813 - INFO - Line 354 (Debug statement): print(f"Merged search failed: {str(e)}")
146
+ 2025-04-30 22:18:05,813 - INFO - Issues in memory\fix_supabase_tables.py:
147
+ 2025-04-30 22:18:05,813 - INFO - Line 55 (Commented code): # Check if the client is initialized
148
+ 2025-04-30 22:18:05,813 - INFO - Line 71 (Commented code): # Check if the table exists
149
+ 2025-04-30 22:18:05,813 - INFO - Line 79 (Commented code): # SQL for creating the table
150
+ 2025-04-30 22:18:05,814 - INFO - Line 130 (Commented code): # Trim cache if it's too large
151
+ 2025-04-30 22:18:05,814 - INFO - Line 141 (Commented code): # Store in Supabase if initialized
152
+ 2025-04-30 22:18:05,814 - INFO - Line 150 (Debug statement): logger.debug(f"Stored entry in table {current_table}: {key}")
153
+ 2025-04-30 22:18:05,814 - INFO - Line 170 (Commented code): # Check if the entry has expired
154
+ 2025-04-30 22:18:05,814 - INFO - Line 187 (Commented code): # Check if any entries were found
155
+ 2025-04-30 22:18:05,814 - INFO - Line 196 (Commented code): # Check if the entry has expired
156
+ 2025-04-30 22:18:05,814 - INFO - Line 208 (Debug statement): logger.debug(f"Retrieved entry from table {current_table}: {key}")
157
+ 2025-04-30 22:18:05,814 - INFO - Line 227 (Commented code): # Remove from Supabase if initialized
158
+ 2025-04-30 22:18:05,814 - INFO - Line 233 (Debug statement): logger.debug(f"Deleted entry from table {current_table}: {key}")
159
+ 2025-04-30 22:18:05,814 - INFO - Line 256 (Commented code): # Get keys from Supabase if initialized
160
+ 2025-04-30 22:18:05,815 - INFO - Line 262 (Commented code): # Filter by prefix if provided
161
+ 2025-04-30 22:18:05,815 - INFO - Line 278 (Debug statement): logger.debug(f"Listed keys from table {current_table}: {len(keys)} keys")
162
+ 2025-04-30 22:18:05,815 - INFO - Line 350 (Commented code): # Create a key for the message
163
+ 2025-04-30 22:18:05,815 - INFO - Line 360 (Commented code): # Add metadata if provided
164
+ 2025-04-30 22:18:05,816 - INFO - Line 384 (Commented code): # Filter by before/after if provided
165
+ 2025-04-30 22:18:05,816 - INFO - Line 391 (Commented code): # Limit the number of messages if provided
166
+ 2025-04-30 22:18:05,816 - INFO - Line 526 (Commented code): # Check if any entries were found
167
+ 2025-04-30 22:18:05,816 - INFO - Line 546 (Debug statement): print("\nFixing Supabase Tables for GAIA Agent")
168
+ 2025-04-30 22:18:05,816 - INFO - Line 547 (Debug statement): print("====================================\n")
169
+ 2025-04-30 22:18:05,837 - INFO - Line 562 (Debug statement): print("All Supabase tables are now being used correctly.")
170
+ 2025-04-30 22:18:05,839 - INFO - Line 566 (Debug statement): print("Please check the logs for more information")
171
+ 2025-04-30 22:18:05,839 - INFO - Issues in memory\simplified_memory_fix.py:
172
+ 2025-04-30 22:18:05,839 - INFO - Line 50 (Commented code): # Check if the client is initialized
173
+ 2025-04-30 22:18:05,840 - INFO - Line 59 (Commented code): # Check if the table exists
174
+ 2025-04-30 22:18:05,840 - INFO - Line 68 (Commented code): # SQL for creating the table
175
+ 2025-04-30 22:18:05,840 - INFO - Line 118 (Commented code): # Trim cache if it's too large
176
+ 2025-04-30 22:18:05,840 - INFO - Line 129 (Commented code): # Store in Supabase if initialized
177
+ 2025-04-30 22:18:05,840 - INFO - Line 138 (Debug statement): logger.debug(f"Stored entry in table {current_table}: {key}")
178
+ 2025-04-30 22:18:05,840 - INFO - Line 158 (Commented code): # Check if the entry has expired
179
+ 2025-04-30 22:18:05,840 - INFO - Line 175 (Commented code): # Check if any entries were found
180
+ 2025-04-30 22:18:05,840 - INFO - Line 184 (Commented code): # Check if the entry has expired
181
+ 2025-04-30 22:18:05,841 - INFO - Line 196 (Debug statement): logger.debug(f"Retrieved entry from table {current_table}: {key}")
182
+ 2025-04-30 22:18:05,841 - INFO - Line 215 (Commented code): # Remove from Supabase if initialized
183
+ 2025-04-30 22:18:05,841 - INFO - Line 221 (Debug statement): logger.debug(f"Deleted entry from table {current_table}: {key}")
184
+ 2025-04-30 22:18:05,841 - INFO - Line 244 (Commented code): # Get keys from Supabase if initialized
185
+ 2025-04-30 22:18:05,841 - INFO - Line 250 (Commented code): # Filter by prefix if provided
186
+ 2025-04-30 22:18:05,841 - INFO - Line 266 (Debug statement): logger.debug(f"Listed keys from table {current_table}: {len(keys)} keys")
187
+ 2025-04-30 22:18:05,842 - INFO - Line 388 (Commented code): # Check if any entries were found
188
+ 2025-04-30 22:18:05,842 - INFO - Line 408 (Debug statement): print("\nSimplified Fix for Supabase Tables in GAIA Agent")
189
+ 2025-04-30 22:18:05,842 - INFO - Line 409 (Debug statement): print("==============================================\n")
190
+ 2025-04-30 22:18:05,843 - INFO - Line 411 (Commented code): # Create the working memory table if it doesn't exist
191
+ 2025-04-30 22:18:05,845 - INFO - Line 424 (Debug statement): print("The working memory table is now being used correctly.")
192
+ 2025-04-30 22:18:05,846 - INFO - Line 425 (Debug statement): print("This should be sufficient for the GAIA benchmark test.")
193
+ 2025-04-30 22:18:05,848 - INFO - Line 429 (Debug statement): print("Please check the logs for more information")
194
+ 2025-04-30 22:18:05,848 - INFO - Issues in memory\supabase_memory.py:
195
+ 2025-04-30 22:18:05,849 - INFO - Line 15 (Commented code): # Try to import supabase, but don't fail if it's not available
196
+ 2025-04-30 22:18:05,849 - INFO - Line 19 (Commented code): # Define dummy types for type checking
197
+ 2025-04-30 22:18:05,849 - INFO - Line 60 (Commented code): # Set expiration time if TTL is provided
198
+ 2025-04-30 22:18:05,849 - INFO - Line 147 (Commented code): # Check if the main table exists
199
+ 2025-04-30 22:18:05,849 - INFO - Line 181 (Commented code): # Check if the table exists by querying it
200
+ 2025-04-30 22:18:05,850 - INFO - Line 183 (Debug statement): logger.debug(f"Table {actual_table} exists")
201
+ 2025-04-30 22:18:05,850 - INFO - Line 249 (Commented code): # Trim cache if it's too large
202
+ 2025-04-30 22:18:05,850 - INFO - Line 260 (Commented code): # Store in Supabase if initialized
203
+ 2025-04-30 22:18:05,850 - INFO - Line 300 (Commented code): # Check if the entry has expired
204
+ 2025-04-30 22:18:05,850 - INFO - Line 316 (Commented code): # Check if any entries were found
205
+ 2025-04-30 22:18:05,850 - INFO - Line 324 (Commented code): # Check if the entry has expired
206
+ 2025-04-30 22:18:05,851 - INFO - Line 359 (Commented code): # Remove from Supabase if initialized
207
+ 2025-04-30 22:18:05,851 - INFO - Line 388 (Commented code): # Clear Supabase if initialized
208
+ 2025-04-30 22:18:05,851 - INFO - Line 422 (Commented code): # Get keys from Supabase if initialized
209
+ 2025-04-30 22:18:05,851 - INFO - Line 428 (Commented code): # Filter by prefix if provided
210
+ 2025-04-30 22:18:05,851 - INFO - Line 484 (Commented code): # Create a key for the message
211
+ 2025-04-30 22:18:05,851 - INFO - Line 494 (Commented code): # Add metadata if provided
212
+ 2025-04-30 22:18:05,851 - INFO - Line 531 (Commented code): # Filter by before/after if provided
213
+ 2025-04-30 22:18:05,851 - INFO - Line 538 (Commented code): # Limit the number of messages if provided
214
+ 2025-04-30 22:18:05,852 - INFO - Issues in memory\supabase_memory_patch.py:
215
+ 2025-04-30 22:18:05,852 - INFO - Line 60 (Commented code): # Check if the client is initialized
216
+ 2025-04-30 22:18:05,852 - INFO - Line 73 (Commented code): # Check if the tables exist
217
+ 2025-04-30 22:18:05,852 - INFO - Line 90 (Commented code): # SQL for creating the table
218
+ 2025-04-30 22:18:05,853 - INFO - Line 151 (Debug statement): print("All tables are now properly utilized:")
219
+ 2025-04-30 22:18:05,854 - INFO - Line 152 (Debug statement): print("- gaia_memory: Main memory table")
220
+ 2025-04-30 22:18:05,854 - INFO - Line 153 (Debug statement): print("- gaia_memory_conversations: Conversation history table")
221
+ 2025-04-30 22:18:05,854 - INFO - Line 154 (Debug statement): print("- gaia_memory_cache: Result cache table")
222
+ 2025-04-30 22:18:05,854 - INFO - Line 155 (Debug statement): print("- gaia_memory_working: Working memory table")
223
+ 2025-04-30 22:18:05,856 - INFO - Line 159 (Debug statement): print("Please check the logs for more information")
224
+ 2025-04-30 22:18:05,856 - INFO - Issues in tools\arxiv_tool.py:
225
+ 2025-04-30 22:18:05,856 - INFO - Line 60 (Debug statement): logger.debug(f"Searching arXiv for: {query}")
226
+ 2025-04-30 22:18:05,856 - INFO - Line 82 (Debug statement): logger.debug(f"Found {len(results)} results from arXiv")
227
+ 2025-04-30 22:18:05,856 - INFO - Line 134 (Debug statement): logger.debug(f"Getting arXiv paper with ID: {paper_id}")
228
+ 2025-04-30 22:18:05,856 - INFO - Line 143 (Commented code): # Create a search for the specific paper
229
+ 2025-04-30 22:18:05,856 - INFO - Line 201 (Debug statement): logger.debug(f"Searching arXiv for papers in category: {category}")
230
+ 2025-04-30 22:18:05,856 - INFO - Line 203 (Commented code): # Create a query for the category
231
+ 2025-04-30 22:18:05,856 - INFO - Line 222 (Commented code): # Search for papers
232
+ 2025-04-30 22:18:05,857 - INFO - Line 225 (Debug statement): print(f"Found {len(results)} papers")
233
+ 2025-04-30 22:18:05,857 - INFO - Line 227 (Debug statement): print(f"\nPaper {i+1}:")
234
+ 2025-04-30 22:18:05,857 - INFO - Line 228 (Debug statement): print(f"Title: {paper['title']}")
235
+ 2025-04-30 22:18:05,857 - INFO - Line 229 (Debug statement): print(f"Authors: {paper['authors']}")
236
+ 2025-04-30 22:18:05,857 - INFO - Line 230 (Debug statement): print(f"Published: {paper['published']}")
237
+ 2025-04-30 22:18:05,857 - INFO - Line 231 (Debug statement): print(f"URL: {paper['url']}")
238
+ 2025-04-30 22:18:05,857 - INFO - Line 232 (Debug statement): print(f"Summary: {paper['summary'][:200]}...")
239
+ 2025-04-30 22:18:05,857 - INFO - Line 234 (Debug statement): print(f"Error: {str(e)}")
240
+ 2025-04-30 22:18:05,857 - INFO - Issues in tools\multimodal_tools.py:
241
+ 2025-04-30 22:18:05,858 - INFO - Line 81 (Debug statement): logger.debug(f"Analyzing image: {image_path}")
242
+ 2025-04-30 22:18:05,858 - INFO - Line 126 (Debug statement): logger.debug("Image analysis completed successfully")
243
+ 2025-04-30 22:18:05,858 - INFO - Line 129 (Commented code): # If the result is not valid JSON, return it as plain text
244
+ 2025-04-30 22:18:05,858 - INFO - Line 157 (Debug statement): logger.debug(f"Detecting objects in image: {image_path}")
245
+ 2025-04-30 22:18:05,858 - INFO - Line 206 (Debug statement): logger.debug(f"Object detection completed, found {parsed_result.get('object_count', 0)} objects")
246
+ 2025-04-30 22:18:05,859 - INFO - Line 255 (Debug statement): logger.debug(f"Interpreting chart: {chart_path}")
247
+ 2025-04-30 22:18:05,859 - INFO - Line 306 (Debug statement): logger.debug(f"Chart interpretation completed: {parsed_result.get('chart_type', 'unknown')} chart")
248
+ 2025-04-30 22:18:05,859 - INFO - Line 338 (Debug statement): logger.debug(f"Extracting data from chart: {chart_path}")
249
+ 2025-04-30 22:18:05,859 - INFO - Line 386 (Debug statement): logger.debug(f"Chart data extraction completed with {len(parsed_result.get('data', []))} data points")
250
+ 2025-04-30 22:18:05,859 - INFO - Line 422 (Commented code): # Check for required dependencies
251
+ 2025-04-30 22:18:05,860 - INFO - Line 443 (Debug statement): logger.debug(f"Parsing document: {document_path}")
252
+ 2025-04-30 22:18:05,860 - INFO - Line 468 (Commented code): # Truncate if too long
253
+ 2025-04-30 22:18:05,860 - INFO - Line 504 (Debug statement): logger.debug(f"Extracting structured data from document: {document_path}")
254
+ 2025-04-30 22:18:05,860 - INFO - Line 536 (Debug statement): logger.debug("Structured data extraction completed successfully")
255
+ 2025-04-30 22:18:05,860 - INFO - Issues in tools\perplexity_tool.py:
256
+ 2025-04-30 22:18:05,860 - INFO - Line 20 (Commented code): # Set up basic logging if not already configured
257
+ 2025-04-30 22:18:05,860 - INFO - Line 40 (Commented code): # Get API key from config or environment (only configuration from environment)
258
+ 2025-04-30 22:18:05,860 - INFO - Line 45 (Commented code): # Hardcoded values for API URL and model
259
+ 2025-04-30 22:18:05,860 - INFO - Line 50 (Commented code): # Endpoint for chat completions
260
+ 2025-04-30 22:18:05,861 - INFO - Line 53 (Debug statement): logger.debug(f"Initialized Perplexity tool with model {self.model}")
261
+ 2025-04-30 22:18:05,861 - INFO - Line 73 (Debug statement): logger.debug(f"Querying Perplexity API: {question[:50]}...")
262
+ 2025-04-30 22:18:05,861 - INFO - Line 82 (Commented code): # Add system prompt if provided
263
+ 2025-04-30 22:18:05,861 - INFO - Line 134 (Debug statement): logger.debug(f"Perplexity API query successful, received {len(content) if content else 0} characters")
264
+ 2025-04-30 22:18:05,862 - INFO - Line 169 (Debug statement): logger.debug(f"Searching with Perplexity API: {query[:50]}...")
265
+ 2025-04-30 22:18:05,862 - INFO - Issues in tools\reasoning_tools.py:
266
+ 2025-04-30 22:18:05,862 - INFO - Line 65 (Debug statement): logger.debug(f"Performing step-by-step reasoning on: {question}")
267
+ 2025-04-30 22:18:05,862 - INFO - Line 93 (Commented code): # Add context if provided
268
+ 2025-04-30 22:18:05,862 - INFO - Line 111 (Debug statement): logger.debug(f"Reasoning completed with {len(parsed_result.get('steps', []))} steps")
269
+ 2025-04-30 22:18:05,862 - INFO - Line 134 (Debug statement): logger.debug(f"Performing chain-of-thought reasoning on: {question}")
270
+ 2025-04-30 22:18:05,862 - INFO - Line 148 (Commented code): # Add context if provided
271
+ 2025-04-30 22:18:05,862 - INFO - Line 163 (Debug statement): logger.debug("Chain-of-thought reasoning completed")
272
+ 2025-04-30 22:18:05,862 - INFO - Line 203 (Debug statement): logger.debug(f"Calculating expression: {expression}")
273
+ 2025-04-30 22:18:05,863 - INFO - Line 209 (Commented code): # Evaluate using sympy for symbolic math
274
+ 2025-04-30 22:18:05,863 - INFO - Line 241 (Debug statement): logger.debug(f"Solving equation: {equation} for {variable}")
275
+ 2025-04-30 22:18:05,863 - INFO - Line 262 (Commented code): # Convert to float or int if possible
276
+ 2025-04-30 22:18:05,863 - INFO - Line 317 (Commented code): # Only replace if it's a whole word (not part of another word)
277
+ 2025-04-30 22:18:05,863 - INFO - Line 354 (Debug statement): logger.debug(f"Verifying claim: {claim}")
278
+ 2025-04-30 22:18:05,863 - INFO - Line 395 (Debug statement): logger.debug(f"Verification completed: {parsed_result.get('verification_result', 'unknown')}")
279
+ 2025-04-30 22:18:05,863 - INFO - Line 418 (Debug statement): logger.debug(f"Comparing claims: '{claim1}' and '{claim2}'")
280
+ 2025-04-30 22:18:05,863 - INFO - Line 455 (Debug statement): logger.debug(f"Comparison completed: {parsed_result.get('relationship', 'unknown')}")
281
+ 2025-04-30 22:18:05,863 - INFO - Issues in tools\web_tools.py:
282
+ 2025-04-30 22:18:05,864 - INFO - Line 124 (Debug statement): logger.debug(f"Searching DuckDuckGo for: {query}")
283
+ 2025-04-30 22:18:05,864 - INFO - Line 137 (Debug statement): logger.debug(f"Found {len(results)} results from DuckDuckGo")
284
+ 2025-04-30 22:18:05,864 - INFO - Line 187 (Debug statement): logger.debug(f"Searching Serper for: {query}")
285
+ 2025-04-30 22:18:05,864 - INFO - Line 217 (Debug statement): logger.debug(f"Found {len(organic_results)} results from Serper")
286
+ 2025-04-30 22:18:05,864 - INFO - Line 270 (Debug statement): logger.debug(f"Extracting content from: {url}")
287
+ 2025-04-30 22:18:05,864 - INFO - Line 306 (Commented code): # Truncate if too long
288
+ 2025-04-30 22:18:05,864 - INFO - Line 392 (Debug statement): logger.debug(f"Following links from: {url}")
289
+ 2025-04-30 22:18:05,864 - INFO - Line 401 (Commented code): # Filter links by pattern if provided
290
+ 2025-04-30 22:18:05,864 - INFO - Issues in tools\__init__.py:
291
+ 2025-04-30 22:18:05,865 - INFO - Line 129 (Debug statement): logger.debug(f"Registered tool: {metadata.name} in category {metadata.category}")
292
+ 2025-04-30 22:18:05,865 - INFO - Line 189 (Commented code): # Check for direct tool name match
293
+ 2025-04-30 22:18:05,865 - INFO - Line 194 (Commented code): # Check for keyword matches
294
+ 2025-04-30 22:18:05,865 - INFO - Line 659 (Commented code): # Convenience functions for accessing tools
295
+ 2025-04-30 22:18:05,865 - INFO - Issues in utils\formatting.py:
296
+ 2025-04-30 22:18:05,865 - INFO - Line 38 (Debug statement): logger.debug(f"Extracting answer from response (expected format: {expected_format})")
297
+ 2025-04-30 22:18:05,865 - INFO - Line 40 (Commented code): # If the response is already concise, return it directly
298
+ 2025-04-30 22:18:05,865 - INFO - Line 44 (Commented code): # Look for explicit answer indicators
299
+ 2025-04-30 22:18:05,865 - INFO - Line 75 (Commented code): # Default: return the last sentence as the answer
300
+ 2025-04-30 22:18:05,866 - INFO - Line 92 (Commented code): # Look for numbers with optional decimal points and commas
301
+ 2025-04-30 22:18:05,866 - INFO - Line 108 (Commented code): # If no answer-indicating phrase, find all numbers and return the last one
302
+ 2025-04-30 22:18:05,866 - INFO - Line 147 (Commented code): # If no answer-indicating phrase, find all dates and return the first one
303
+ 2025-04-30 22:18:05,866 - INFO - Line 167 (Commented code): # Check for yes/no indicators at the beginning of the text or after answer indicators
304
+ 2025-04-30 22:18:05,866 - INFO - Line 198 (Commented code): # If no clear yes/no, look for sentiment
305
+ 2025-04-30 22:18:05,866 - INFO - Line 213 (Commented code): # Default to "Unknown" if can't determine
306
+ 2025-04-30 22:18:05,866 - INFO - Line 226 (Commented code): # Look for explicit list indicators
307
+ 2025-04-30 22:18:05,866 - INFO - Line 244 (Commented code): # Look for bullet points or numbered lists
308
+ 2025-04-30 22:18:05,866 - INFO - Line 250 (Commented code): # Look for any comma-separated sequence
309
+ 2025-04-30 22:18:05,866 - INFO - Line 268 (Commented code): # Look for entity indicators
310
+ 2025-04-30 22:18:05,866 - INFO - Line 281 (Commented code): # Look for capitalized proper nouns
311
+ 2025-04-30 22:18:05,867 - INFO - Line 306 (Commented code): # Look for JSON blocks
312
+ 2025-04-30 22:18:05,867 - INFO - Line 318 (Commented code): # Look for JSON without code blocks
313
+ 2025-04-30 22:18:05,867 - INFO - Line 341 (Debug statement): logger.debug(f"Formatting answer to {format_type} format")
314
+ 2025-04-30 22:18:05,867 - INFO - Line 401 (Commented code): # Default: return as is
315
+ 2025-04-30 22:18:05,867 - INFO - Line 415 (Debug statement): logger.debug(f"Validating answer format (expected: {expected_format})")
316
+ 2025-04-30 22:18:05,867 - INFO - Line 435 (Commented code): # Validate date format (simple check)
317
+ 2025-04-30 22:18:05,867 - INFO - Line 469 (Commented code): # Validate entity format (simple check for proper noun)
318
+ 2025-04-30 22:18:05,867 - INFO - Line 502 (Debug statement): logger.debug(f"Processing answer (expected format: {expected_format})")
319
+ 2025-04-30 22:18:05,867 - INFO - Line 504 (Commented code): # Default to text format if none specified
320
+ 2025-04-30 22:18:05,867 - INFO - Issues in utils\parsing.py:
321
+ 2025-04-30 22:18:05,868 - INFO - Line 34 (Debug statement): logger.debug(f"Extracting key information from question: {question}")
322
+ 2025-04-30 22:18:05,868 - INFO - Line 47 (Commented code): # Extract entities (proper nouns, potentially)
323
+ 2025-04-30 22:18:05,868 - INFO - Line 52 (Commented code): # Extract keywords (important nouns, verbs, adjectives)
324
+ 2025-04-30 22:18:05,868 - INFO - Line 143 (Debug statement): logger.debug(f"Identifying required tools for question: {question}")
325
+ 2025-04-30 22:18:05,868 - INFO - Line 197 (Debug statement): logger.debug(f"Determining complexity for question: {question}")
326
+ 2025-04-30 22:18:05,868 - INFO - Line 293 (Debug statement): logger.debug(f"Parsing question: {question}")
327
+ 2025-04-30 22:18:05,868 - INFO - Issues in utils\__init__.py:
328
+ 2025-04-30 22:18:05,868 - INFO - Line 96 (Commented code): # Check for number format
329
+ 2025-04-30 22:18:05,868 - INFO - Line 100 (Commented code): # Check for date format
330
+ 2025-04-30 22:18:05,868 - INFO - Line 104 (Commented code): # Check for boolean format
331
+ 2025-04-30 22:18:05,869 - INFO - Line 109 (Commented code): # Check for list format
332
+ 2025-04-30 22:18:05,869 - INFO - Line 113 (Commented code): # Check for entity format
333
+ 2025-04-30 22:18:05,869 - INFO - Checking for potential hardcoded credentials...
334
+ 2025-04-30 22:18:05,990 - INFO - No potential hardcoded credentials found
335
+ 2025-04-30 22:18:05,990 - INFO -
336
+ Cleanup Summary:
337
+ 2025-04-30 22:18:05,991 - INFO - - Temporary files to remove: 39
338
+ 2025-04-30 22:18:05,991 - INFO - - Duplicate test files to remove: 7
339
+ 2025-04-30 22:18:05,991 - INFO - - Files with debug statements or commented code: 17
340
+ 2025-04-30 22:18:05,992 - INFO - - Files with potential hardcoded credentials: 0
create_env_and_test.py ADDED
@@ -0,0 +1,77 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ Create .env File and Test Supabase Connection
3
+
4
+ This script creates a .env file with hardcoded values and then tests the connection to Supabase.
5
+ """
6
+
7
+ import os
8
+ import sys
9
+ import time
10
+
11
+ def create_env_file(api_key):
12
+ """
13
+ Create a .env file with the provided API key.
14
+
15
+ Args:
16
+ api_key: The Supabase API key
17
+ """
18
+
19
+ env_content = f"""
20
+ # Supabase Configuration
21
+ SUPABASE_URL=https://tjamxhvvtnypbadvrkjq.supabase.co
22
+ SUPABASE_KEY={api_key}
23
+ """
24
+
25
+ with open(".env", "w") as f:
26
+ f.write(env_content)
27
+
28
+
29
+ def test_supabase():
30
+ """Test the connection to Supabase."""
31
+
32
+ try:
33
+ from supabase import create_client
34
+ from dotenv import load_dotenv
35
+
36
+ load_dotenv(override=True)
37
+
38
+ # Get Supabase URL and key
39
+ supabase_url = os.getenv("SUPABASE_URL")
40
+ supabase_key = os.getenv("SUPABASE_KEY")
41
+
42
+ if supabase_key:
43
+ masked_key = supabase_key[:4] + "*" * (len(supabase_key) - 8) + supabase_key[-4:] if len(supabase_key) > 8 else "****"
44
+ else:
45
+ return False
46
+
47
+ client = create_client(supabase_url, supabase_key)
48
+
49
+ result = client.table("test_connection").select("*").limit(1).execute()
50
+
51
+ return True
52
+ except Exception as e:
53
+ import traceback
54
+ traceback.print_exc()
55
+ return False
56
+
57
+ def main():
58
+ """Main function."""
59
+
60
+ api_key = input("> ").strip()
61
+
62
+ if not api_key:
63
+ return
64
+
65
+ create_env_file(api_key)
66
+
67
+ time.sleep(1)
68
+
69
+ # Test Supabase connection
70
+ success = test_supabase()
71
+
72
+ if success:
73
+ else:
74
+
75
+
76
+ if __name__ == "__main__":
77
+ main()
deploy_to_hf.py ADDED
@@ -0,0 +1,161 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ Simplified Git Deployment Script for Hugging Face Spaces
3
+
4
+ This script handles the git operations required to deploy to a Hugging Face Space.
5
+ It assumes you have Git installed and have already configured git authentication.
6
+
7
+ Usage:
8
+ python deploy_to_hf.py
9
+
10
+ """
11
+ import os
12
+ import subprocess
13
+ import sys
14
+
15
+ # Hugging Face Space details
16
+ SPACE_ID = "JoachimVC/Final_Assignment_Template"
17
+ REPO_URL = f"https://huggingface.co/spaces/{SPACE_ID}"
18
+
19
+ def run_command(cmd, silent=False):
20
+ """Run a shell command and return output"""
21
+ if not silent:
22
+ print(f"Running: {cmd}")
23
+ try:
24
+ process = subprocess.Popen(
25
+ cmd,
26
+ shell=True,
27
+ stdout=subprocess.PIPE,
28
+ stderr=subprocess.PIPE,
29
+ text=True
30
+ )
31
+ stdout, stderr = process.communicate()
32
+
33
+ if process.returncode != 0:
34
+ print(f"Command failed with error code {process.returncode}")
35
+ print(f"Error: {stderr.strip()}")
36
+ return False
37
+
38
+ if not silent and stdout.strip():
39
+ print(stdout.strip())
40
+
41
+ return True
42
+ except Exception as e:
43
+ print(f"Failed to execute command: {str(e)}")
44
+ return False
45
+
46
+ def check_git_installed():
47
+ """Check if git is installed"""
48
+ print("Checking if git is installed...")
49
+ return run_command("git --version", silent=True)
50
+
51
+ def setup_git():
52
+ """Initialize git if needed and configure remote"""
53
+ # Check if git is already initialized
54
+ if not os.path.exists(".git"):
55
+ print("Initializing git repository...")
56
+ if not run_command("git init"):
57
+ return False
58
+
59
+ # Configure git user if not set
60
+ result = subprocess.run("git config user.name", shell=True, capture_output=True, text=True)
61
+ if not result.stdout.strip():
62
+ print("Setting up git user name...")
63
+ run_command('git config user.name "GAIA Deployment"')
64
+
65
+ result = subprocess.run("git config user.email", shell=True, capture_output=True, text=True)
66
+ if not result.stdout.strip():
67
+ print("Setting up git user email...")
68
+ run_command('git config user.email "gaia-deployment@example.com"')
69
+
70
+ # Set up the Hugging Face remote
71
+ print("Setting up git remote for Hugging Face...")
72
+
73
+ # Check if remote exists
74
+ result = subprocess.run("git remote -v", shell=True, capture_output=True, text=True)
75
+ remote_output = result.stdout
76
+
77
+ if "huggingface" in remote_output:
78
+ run_command("git remote remove huggingface")
79
+
80
+ # Add the HF remote
81
+ return run_command(f"git remote add huggingface https://huggingface.co/spaces/{SPACE_ID}")
82
+
83
+ def create_gitignore():
84
+ """Create .gitignore file with common patterns"""
85
+ patterns = [
86
+ ".env",
87
+ "__pycache__/",
88
+ "*.py[cod]",
89
+ "*$py.class",
90
+ ".Python",
91
+ "env/",
92
+ "venv/",
93
+ ".venv",
94
+ ".coverage",
95
+ "*.log"
96
+ ]
97
+
98
+ print("Creating .gitignore file...")
99
+ with open(".gitignore", "w") as f:
100
+ for pattern in patterns:
101
+ f.write(f"{pattern}\n")
102
+
103
+ return True
104
+
105
+ def commit_changes():
106
+ """Stage and commit all changes"""
107
+ print("Staging files for commit...")
108
+ if not run_command("git add -A"):
109
+ return False
110
+
111
+ print("Committing changes...")
112
+ return run_command('git commit -m "Deploy GAIA to Hugging Face Space"')
113
+
114
+ def push_to_huggingface():
115
+ """Push to Hugging Face Space"""
116
+ print("Pushing to Hugging Face Space...")
117
+
118
+ # Try regular push first
119
+ if run_command("git push huggingface main"):
120
+ return True
121
+
122
+ # If regular push fails, try force push
123
+ print("Regular push failed, attempting force push...")
124
+ if run_command("git push huggingface main --force"):
125
+ return True
126
+
127
+ print("Force push failed. You may need to manually push:")
128
+ print("1. Ensure you have the correct permissions")
129
+ print("2. Try: git push huggingface main --force")
130
+ return False
131
+
132
+ def main():
133
+ """Run deployment process"""
134
+ print("\n===== DEPLOYING TO HUGGING FACE SPACE =====")
135
+ print(f"Target Space: {REPO_URL}")
136
+
137
+ # Ensure Git is installed
138
+ if not check_git_installed():
139
+ print("Git is not installed. Please install Git and try again.")
140
+ sys.exit(1)
141
+
142
+ # Setup steps
143
+ steps = [
144
+ ("Setting up git", setup_git),
145
+ ("Creating .gitignore", create_gitignore),
146
+ ("Committing changes", commit_changes),
147
+ ("Pushing to Hugging Face", push_to_huggingface)
148
+ ]
149
+
150
+ for name, func in steps:
151
+ print(f"\n>> {name}...")
152
+ if not func():
153
+ print(f"Step '{name}' failed. Deployment process halted.")
154
+ sys.exit(1)
155
+
156
+ print("\n===== DEPLOYMENT COMPLETED SUCCESSFULLY =====")
157
+ print(f"Your project has been deployed to: {REPO_URL}")
158
+ print("Note: It may take a few minutes for the changes to be visible.")
159
+
160
+ if __name__ == "__main__":
161
+ main()
deploy_to_huggingface.py ADDED
@@ -0,0 +1,333 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python
2
+ """
3
+ Hugging Face Space Deployment Script
4
+
5
+ This script deploys the GAIA project to Hugging Face Spaces using git.
6
+ It handles authentication, remote setup, and deployment.
7
+
8
+ Usage:
9
+ python deploy_to_huggingface.py [--space-id SPACE_ID]
10
+
11
+ Author: DevOps Team
12
+ Date: April 30, 2025
13
+ """
14
+
15
+ import os
16
+ import sys
17
+ import argparse
18
+ import subprocess
19
+ import logging
20
+ import json
21
+ import time
22
+ from pathlib import Path
23
+
24
+ # Configure logging
25
+ logging.basicConfig(
26
+ level=logging.INFO,
27
+ format="%(asctime)s - %(levelname)s - %(message)s"
28
+ )
29
+ logger = logging.getLogger("hf_deploy")
30
+
31
+ def parse_arguments():
32
+ """Parse command line arguments"""
33
+ parser = argparse.ArgumentParser(description="Deploy GAIA to Hugging Face Space")
34
+ parser.add_argument("--space-id",
35
+ help="Hugging Face Space ID (username/space-name)",
36
+ default="JoachimVC/GAIA-Assistant")
37
+ parser.add_argument("--check-only", action="store_true",
38
+ help="Only check prerequisites without deploying")
39
+ return parser.parse_args()
40
+
41
+ def run_command(command, shell=False):
42
+ """Run a shell command and return the result"""
43
+ try:
44
+ logger.info(f"Running command: {command}")
45
+ result = subprocess.run(
46
+ command,
47
+ shell=shell,
48
+ check=True,
49
+ text=True,
50
+ capture_output=True
51
+ )
52
+ return result.stdout.strip()
53
+ except subprocess.CalledProcessError as e:
54
+ logger.error(f"Command failed with error: {e}")
55
+ logger.error(f"Error output: {e.stderr}")
56
+ return None
57
+
58
+ def check_prerequisites():
59
+ """Check if all prerequisites are met"""
60
+ # Check if git is installed
61
+ git_version = run_command(["git", "--version"])
62
+ if not git_version:
63
+ logger.error("Git is not installed or not in PATH")
64
+ return False
65
+ logger.info(f"Git version: {git_version}")
66
+
67
+ # Check if huggingface_hub is installed
68
+ try:
69
+ import huggingface_hub
70
+ logger.info(f"huggingface_hub version: {huggingface_hub.__version__}")
71
+ except ImportError:
72
+ logger.error("huggingface_hub is not installed")
73
+ logger.info("Installing huggingface_hub...")
74
+ if run_command(["pip", "install", "huggingface_hub"]) is None:
75
+ return False
76
+
77
+ # Check if user is logged in to Hugging Face
78
+ hf_token_exists = os.path.exists(os.path.expanduser("~/.huggingface/token"))
79
+ if not hf_token_exists:
80
+ logger.error("Not logged in to Hugging Face")
81
+ logger.info("Please run 'huggingface-cli login' to log in")
82
+ return False
83
+
84
+ # Check if the project has necessary files
85
+ required_files = ["app.py", "requirements.txt", "space_config.json", "HF_README.md"]
86
+ missing_files = [f for f in required_files if not os.path.exists(f)]
87
+ if missing_files:
88
+ logger.error(f"Missing required files: {', '.join(missing_files)}")
89
+ return False
90
+
91
+ logger.info("All prerequisites met")
92
+ return True
93
+
94
+ def setup_git_repo(space_id):
95
+ """Set up git repository for deployment"""
96
+ # Check if .git exists
97
+ if not os.path.exists(".git"):
98
+ logger.info("Initializing git repository...")
99
+ if run_command(["git", "init"]) is None:
100
+ return False
101
+
102
+ # Check if the Hugging Face remote exists
103
+ remotes = run_command(["git", "remote"])
104
+ if remotes is None:
105
+ return False
106
+
107
+ if "huggingface" in remotes.split():
108
+ logger.info("Updating Hugging Face remote...")
109
+ if run_command(["git", "remote", "set-url", "huggingface", f"https://huggingface.co/spaces/{space_id}"]) is None:
110
+ return False
111
+ else:
112
+ logger.info("Adding Hugging Face remote...")
113
+ if run_command(["git", "remote", "add", "huggingface", f"https://huggingface.co/spaces/{space_id}"]) is None:
114
+ return False
115
+
116
+ logger.info("Git repository set up successfully")
117
+ return True
118
+
119
+ def validate_space_id(space_id):
120
+ """Validate the Hugging Face Space ID format"""
121
+ if "/" not in space_id:
122
+ logger.error(f"Invalid Space ID format: {space_id}")
123
+ logger.error("Space ID should be in format 'username/space-name'")
124
+ return False
125
+
126
+ username, space_name = space_id.split("/", 1)
127
+ if not username or not space_name:
128
+ logger.error(f"Invalid Space ID components: {space_id}")
129
+ return False
130
+
131
+ return True
132
+
133
+ def rename_files_for_hf():
134
+ """Rename files for Hugging Face Space"""
135
+ if os.path.exists("HF_README.md"):
136
+ if os.path.exists("README.md"):
137
+ if not os.path.exists("README_original.md"):
138
+ logger.info("Backing up README.md to README_original.md...")
139
+ os.rename("README.md", "README_original.md")
140
+ else:
141
+ logger.info("README_original.md already exists, keeping it")
142
+ # Remove README.md to avoid confusion
143
+ if os.path.exists("README.md"):
144
+ os.remove("README.md")
145
+
146
+ logger.info("Renaming HF_README.md to README.md...")
147
+ os.rename("HF_README.md", "README.md")
148
+
149
+ return True
150
+
151
+ def update_git_ignore():
152
+ """Update .gitignore file for Hugging Face Space"""
153
+ gitignore_entries = [
154
+ ".env",
155
+ "__pycache__/",
156
+ "*.py[cod]",
157
+ "*$py.class",
158
+ "*.so",
159
+ ".Python",
160
+ "env/",
161
+ "build/",
162
+ "develop-eggs/",
163
+ "dist/",
164
+ "downloads/",
165
+ "eggs/",
166
+ ".eggs/",
167
+ "lib/",
168
+ "lib64/",
169
+ "parts/",
170
+ "sdist/",
171
+ "var/",
172
+ "*.egg-info/",
173
+ ".installed.cfg",
174
+ "*.egg",
175
+ ".env",
176
+ ".venv",
177
+ "venv/",
178
+ "ENV/",
179
+ ".idea/",
180
+ ".vscode/",
181
+ "*.log",
182
+ "deploy_to_huggingface.py",
183
+ "test_components.py",
184
+ "verify_production.py",
185
+ "production_verification_report.md",
186
+ "deployment_readiness_report.md",
187
+ "final_verification_summary.md",
188
+ "prepare_for_hf_deployment.py"
189
+ ]
190
+
191
+ # Read existing .gitignore if it exists
192
+ existing_entries = []
193
+ if os.path.exists(".gitignore"):
194
+ with open(".gitignore", "r") as f:
195
+ existing_entries = [line.strip() for line in f.readlines()]
196
+
197
+ # Combine entries, removing duplicates
198
+ combined_entries = list(set(existing_entries + gitignore_entries))
199
+ combined_entries.sort()
200
+
201
+ # Write updated .gitignore
202
+ with open(".gitignore", "w") as f:
203
+ f.write("\n".join(combined_entries) + "\n")
204
+
205
+ logger.info("Updated .gitignore file")
206
+ return True
207
+
208
+ def commit_and_push(space_id):
209
+ """Commit changes and push to Hugging Face Space"""
210
+ # Stage all files
211
+ if run_command(["git", "add", "."]) is None:
212
+ return False
213
+
214
+ # Commit changes
215
+ commit_message = f"Deploy GAIA to Hugging Face Space {space_id}"
216
+ if run_command(["git", "commit", "-m", commit_message]) is None:
217
+ logger.warning("No changes to commit or commit failed")
218
+ # Continue anyway, as there might be no changes
219
+
220
+ # Push to Hugging Face
221
+ logger.info(f"Pushing to Hugging Face Space {space_id}...")
222
+ if run_command(["git", "push", "huggingface", "main"]) is None:
223
+ # Try pushing to master branch if main fails
224
+ logger.info("Pushing to master branch instead...")
225
+ if run_command(["git", "push", "huggingface", "master"]) is None:
226
+ # Try force pushing to main as last resort
227
+ logger.info("Force pushing to main branch...")
228
+ if run_command(["git", "push", "-f", "huggingface", "main"]) is None:
229
+ return False
230
+
231
+ logger.info(f"Successfully pushed to Hugging Face Space {space_id}")
232
+ return True
233
+
234
+ def create_space_if_needed(space_id):
235
+ """Create Hugging Face Space if it doesn't exist"""
236
+ try:
237
+ from huggingface_hub import HfApi, SpaceHardware, SpaceSdk
238
+
239
+ api = HfApi()
240
+ username, space_name = space_id.split("/", 1)
241
+
242
+ # Check if space exists
243
+ try:
244
+ api.get_space_info(space_id)
245
+ logger.info(f"Space {space_id} already exists")
246
+ return True
247
+ except Exception:
248
+ logger.info(f"Space {space_id} doesn't exist, creating it...")
249
+
250
+ # Create space
251
+ api.create_space(
252
+ space_id=space_name,
253
+ token=None, # Use token from ~/.huggingface/token
254
+ sdk=SpaceSdk.GRADIO,
255
+ hardware=SpaceHardware.CPU_BASIC,
256
+ private=False
257
+ )
258
+
259
+ logger.info(f"Created Hugging Face Space {space_id}")
260
+ return True
261
+
262
+ except ImportError:
263
+ logger.warning("huggingface_hub not installed properly, skipping space creation check")
264
+ return True
265
+ except Exception as e:
266
+ logger.error(f"Error creating space: {e}")
267
+ return False
268
+
269
+ def wait_for_deployment(space_id):
270
+ """Wait for deployment to complete and provide status link"""
271
+ logger.info(f"Deployment initiated to {space_id}")
272
+ logger.info(f"You can check the deployment status at: https://huggingface.co/spaces/{space_id}")
273
+ logger.info(f"Once deployment is complete, your app will be available at: https://huggingface.co/spaces/{space_id}")
274
+
275
+ logger.info("Waiting for 60 seconds to allow deployment to start...")
276
+ for i in range(12):
277
+ time.sleep(5)
278
+ logger.info(f"Elapsed time: {(i+1)*5} seconds")
279
+
280
+ return True
281
+
282
+ def main():
283
+ """Main function to deploy to Hugging Face Space"""
284
+ args = parse_arguments()
285
+ space_id = args.space_id
286
+
287
+ logger.info(f"Starting deployment to Hugging Face Space: {space_id}")
288
+
289
+ # Validate Space ID
290
+ if not validate_space_id(space_id):
291
+ return False
292
+
293
+ # Check prerequisites
294
+ if not check_prerequisites():
295
+ return False
296
+
297
+ if args.check_only:
298
+ logger.info("Check-only mode: prerequisites satisfied, stopping before deployment")
299
+ return True
300
+
301
+ # Rename files for Hugging Face
302
+ if not rename_files_for_hf():
303
+ return False
304
+
305
+ # Update .gitignore
306
+ if not update_git_ignore():
307
+ return False
308
+
309
+ # Create Space if needed
310
+ if not create_space_if_needed(space_id):
311
+ logger.warning("Failed to create space, but continuing with deployment")
312
+
313
+ # Set up git repository
314
+ if not setup_git_repo(space_id):
315
+ return False
316
+
317
+ # Commit and push changes
318
+ if not commit_and_push(space_id):
319
+ return False
320
+
321
+ # Wait for deployment and show status
322
+ if not wait_for_deployment(space_id):
323
+ return False
324
+
325
+ logger.info("Deployment process completed successfully!")
326
+ logger.info(f"Your GAIA application will be available at: https://huggingface.co/spaces/{space_id}")
327
+ logger.info("Note: It may take a few minutes for the deployment to complete on Hugging Face")
328
+
329
+ return True
330
+
331
+ if __name__ == "__main__":
332
+ success = main()
333
+ sys.exit(0 if success else 1)
deploy_to_huggingface_credential.py ADDED
@@ -0,0 +1,320 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ Git Credential-based Deployment Script for Hugging Face Spaces
3
+
4
+ This script handles the git operations required to deploy to a Hugging Face Space
5
+ using credential helper for token-based authentication.
6
+
7
+ Usage:
8
+ python deploy_to_huggingface_credential.py [--space SPACE_ID]
9
+
10
+ Environment variables (from .env file):
11
+ HF_TOKEN - Your Hugging Face API token
12
+ """
13
+ import os
14
+ import subprocess
15
+ import sys
16
+ import argparse
17
+ from pathlib import Path
18
+
19
+ def parse_args():
20
+ parser = argparse.ArgumentParser(description="Deploy to Hugging Face Space using git credential helper")
21
+ parser.add_argument("--space", help="Space ID (username/space-name)",
22
+ default="JoachimVC/Final_Assignment_Template")
23
+ return parser.parse_args()
24
+
25
+ def load_dotenv():
26
+ """Load environment variables from .env file"""
27
+ env_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), '.env')
28
+ if not os.path.exists(env_path):
29
+ print(f"ERROR: .env file not found at {env_path}")
30
+ return False
31
+
32
+ print(f"Loading environment variables from {env_path}")
33
+ with open(env_path, 'r') as file:
34
+ for line in file:
35
+ line = line.strip()
36
+ if not line or line.startswith('#'):
37
+ continue
38
+
39
+ key, value = line.split('=', 1)
40
+ os.environ[key] = value
41
+
42
+ return True
43
+
44
+ def run_command(cmd, silent=False):
45
+ """Run a shell command and return output"""
46
+ if not silent:
47
+ print(f"Running: {cmd}")
48
+ try:
49
+ process = subprocess.Popen(
50
+ cmd,
51
+ shell=True,
52
+ stdout=subprocess.PIPE,
53
+ stderr=subprocess.PIPE,
54
+ text=True
55
+ )
56
+ stdout, stderr = process.communicate()
57
+
58
+ if process.returncode != 0:
59
+ print(f"Command failed with error code {process.returncode}")
60
+ print(f"Error: {stderr.strip()}")
61
+ return False
62
+
63
+ if not silent and stdout.strip():
64
+ print(stdout.strip())
65
+
66
+ return True
67
+ except Exception as e:
68
+ print(f"Failed to execute command: {str(e)}")
69
+ return False
70
+
71
+ def check_git_installed():
72
+ """Check if git is installed"""
73
+ print("Checking if git is installed...")
74
+ return run_command("git --version", silent=True)
75
+
76
+ def get_token_from_env():
77
+ """Get Hugging Face token from environment variables"""
78
+ # Try various environment variable names that might contain the token
79
+ token_vars = ["HF_TOKEN", "HUGGINGFACE_TOKEN", "HUGGING_FACE_TOKEN",
80
+ "HF_API_TOKEN", "HUGGINGFACE_API_TOKEN"]
81
+
82
+ for var_name in token_vars:
83
+ token = os.environ.get(var_name)
84
+ if token:
85
+ print(f"Found token in environment variable: {var_name}")
86
+ return token
87
+
88
+ return None
89
+
90
+ def create_deploy_token_file(token):
91
+ """Create a temporary token file for deployment"""
92
+ token_dir = Path.home() / ".huggingface"
93
+ token_dir.mkdir(exist_ok=True)
94
+
95
+ token_file = token_dir / "token"
96
+ with open(token_file, "w") as f:
97
+ f.write(token)
98
+
99
+ print(f"Created HF token file at {token_file}")
100
+ return token_file
101
+
102
+ def setup_git_credential_helper(token):
103
+ """Set up git credential helper to use the token"""
104
+ # Store the token in the git credential helper
105
+ credential_input = f"protocol=https\nhost=huggingface.co\nusername=hf_token\npassword={token}\n"
106
+
107
+ # Use git credential-store to save the token
108
+ print("Setting up git credential helper...")
109
+ process = subprocess.Popen(
110
+ ["git", "credential-store", "store"],
111
+ stdin=subprocess.PIPE,
112
+ stdout=subprocess.PIPE,
113
+ stderr=subprocess.PIPE,
114
+ text=True
115
+ )
116
+
117
+ process.communicate(input=credential_input)
118
+
119
+ # Configure git to use the credential helper
120
+ return run_command("git config --global credential.helper store")
121
+
122
+ def setup_git(space_id):
123
+ """Initialize git if needed and configure remote"""
124
+ # Check if git is already initialized
125
+ if not os.path.exists(".git"):
126
+ print("Initializing git repository...")
127
+ if not run_command("git init"):
128
+ return False
129
+
130
+ # Configure git user if not set
131
+ result = subprocess.run("git config user.name", shell=True, capture_output=True, text=True)
132
+ if not result.stdout.strip():
133
+ print("Setting up git user name...")
134
+ run_command('git config user.name "GAIA Deployment"')
135
+
136
+ result = subprocess.run("git config user.email", shell=True, capture_output=True, text=True)
137
+ if not result.stdout.strip():
138
+ print("Setting up git user email...")
139
+ run_command('git config user.email "gaia-deployment@example.com"')
140
+
141
+ # Set up the Hugging Face remote without token in URL
142
+ print("Setting up git remote for Hugging Face...")
143
+
144
+ # Check if remote exists
145
+ result = subprocess.run("git remote -v", shell=True, capture_output=True, text=True)
146
+ remote_output = result.stdout
147
+
148
+ if "huggingface" in remote_output:
149
+ run_command("git remote remove huggingface")
150
+
151
+ # Add the HF remote without token (will use credential helper)
152
+ repo_url = f"https://huggingface.co/spaces/{space_id}"
153
+ return run_command(f"git remote add huggingface {repo_url}")
154
+
155
+ def create_gitignore():
156
+ """Create .gitignore file with common patterns"""
157
+ patterns = [
158
+ ".env",
159
+ "__pycache__/",
160
+ "*.py[cod]",
161
+ "*$py.class",
162
+ ".Python",
163
+ "env/",
164
+ "venv/",
165
+ ".venv",
166
+ ".coverage",
167
+ "*.log",
168
+ "deploy_with_token.py",
169
+ "deploy_to_huggingface_env.py",
170
+ "deploy_to_huggingface_dotenv.py",
171
+ "deploy_to_huggingface_credential.py" # Don't include this file with token handling
172
+ ]
173
+
174
+ print("Creating .gitignore file...")
175
+ with open(".gitignore", "w") as f:
176
+ for pattern in patterns:
177
+ f.write(f"{pattern}\n")
178
+
179
+ return True
180
+
181
+ def commit_changes():
182
+ """Stage and commit all changes"""
183
+ print("Staging files for commit...")
184
+ if not run_command("git add -A"):
185
+ return False
186
+
187
+ print("Committing changes...")
188
+ return run_command('git commit -m "Deploy GAIA to Hugging Face Space"')
189
+
190
+ def push_to_huggingface():
191
+ """Push to Hugging Face Space"""
192
+ print("Pushing to Hugging Face Space...")
193
+
194
+ # Try direct push with git credential store
195
+ cmd = "git push huggingface main"
196
+
197
+ # Execute the push command
198
+ if run_command(cmd):
199
+ return True
200
+
201
+ # If regular push fails, try force push
202
+ print("Regular push failed, attempting force push...")
203
+ if run_command("git push huggingface main --force"):
204
+ return True
205
+
206
+ print("Force push failed. You may need to manually push:")
207
+ print("1. Ensure your token has write permissions")
208
+ print("2. Try: git push huggingface main --force")
209
+ return False
210
+
211
+ def setup_cli_auth_file(token):
212
+ """Set up CLI authentication file for Hugging Face"""
213
+ # Create .huggingface directory in home if it doesn't exist
214
+ hf_dir = os.path.expanduser("~/.huggingface")
215
+ os.makedirs(hf_dir, exist_ok=True)
216
+
217
+ # Create or update token file
218
+ token_file = os.path.join(hf_dir, "token")
219
+ with open(token_file, "w") as f:
220
+ f.write(token)
221
+
222
+ print(f"Created HF token file at {token_file}")
223
+ return True
224
+
225
+ def setup_git_with_netrc(token):
226
+ """Set up .netrc file for git authentication"""
227
+ netrc_path = os.path.expanduser("~/.netrc")
228
+
229
+ # Check if file exists and has huggingface.co entry
230
+ netrc_content = ""
231
+ if os.path.exists(netrc_path):
232
+ with open(netrc_path, "r") as f:
233
+ netrc_content = f.read()
234
+
235
+ # Add or update huggingface.co entry
236
+ if "machine huggingface.co" not in netrc_content:
237
+ with open(netrc_path, "a") as f:
238
+ f.write(f"\nmachine huggingface.co\nlogin hf_token\npassword {token}\n")
239
+ else:
240
+ # Replace existing entry
241
+ import re
242
+ pattern = r"machine huggingface\.co\s+login.*\s+password.*\s+"
243
+ replacement = f"machine huggingface.co\nlogin hf_token\npassword {token}\n"
244
+ netrc_content = re.sub(pattern, replacement, netrc_content)
245
+ with open(netrc_path, "w") as f:
246
+ f.write(netrc_content)
247
+
248
+ # Set permissions for .netrc file (Unix only)
249
+ if os.name != "nt": # not Windows
250
+ os.chmod(netrc_path, 0o600)
251
+
252
+ print(f"Updated .netrc file at {netrc_path}")
253
+ return True
254
+
255
+ def main():
256
+ """Run deployment process"""
257
+ args = parse_args()
258
+ space_id = args.space
259
+
260
+ # Load .env file
261
+ if not load_dotenv():
262
+ print("WARNING: Failed to load .env file, continuing with existing environment variables")
263
+
264
+ # Get token from environment
265
+ token = get_token_from_env()
266
+ if not token:
267
+ print("ERROR: No Hugging Face token found in environment variables.")
268
+ print("Please ensure your .env file contains one of the following variables:")
269
+ print(" - HF_TOKEN")
270
+ print(" - HUGGINGFACE_TOKEN")
271
+ print(" - HUGGING_FACE_TOKEN")
272
+ print(" - HF_API_TOKEN")
273
+ print(" - HUGGINGFACE_API_TOKEN")
274
+ sys.exit(1)
275
+
276
+ repo_url = f"https://huggingface.co/spaces/{space_id}"
277
+
278
+ print("\n===== DEPLOYING TO HUGGING FACE SPACE =====")
279
+ print(f"Target Space: {repo_url}")
280
+
281
+ # Ensure Git is installed
282
+ if not check_git_installed():
283
+ print("Git is not installed. Please install Git and try again.")
284
+ sys.exit(1)
285
+
286
+ # Set up token file in ~/.huggingface/token
287
+ if not setup_cli_auth_file(token):
288
+ print("Failed to set up CLI authentication file")
289
+ sys.exit(1)
290
+
291
+ # Set up .netrc file with token
292
+ if not setup_git_with_netrc(token):
293
+ print("Failed to set up .netrc file")
294
+ sys.exit(1)
295
+
296
+ # Set up git credential helper
297
+ if not setup_git_credential_helper(token):
298
+ print("Failed to set up git credential helper")
299
+ sys.exit(1)
300
+
301
+ # Setup steps
302
+ steps = [
303
+ ("Setting up git", lambda: setup_git(space_id)),
304
+ ("Creating .gitignore", create_gitignore),
305
+ ("Committing changes", commit_changes),
306
+ ("Pushing to Hugging Face", push_to_huggingface)
307
+ ]
308
+
309
+ for name, func in steps:
310
+ print(f"\n>> {name}...")
311
+ if not func():
312
+ print(f"Step '{name}' failed. Deployment process halted.")
313
+ sys.exit(1)
314
+
315
+ print("\n===== DEPLOYMENT COMPLETED SUCCESSFULLY =====")
316
+ print(f"Your project has been deployed to: {repo_url}")
317
+ print("Note: It may take a few minutes for the changes to be visible.")
318
+
319
+ if __name__ == "__main__":
320
+ main()
deploy_to_huggingface_dotenv.py ADDED
@@ -0,0 +1,227 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ Dotenv-based Git Deployment Script for Hugging Face Spaces
3
+
4
+ This script handles the git operations required to deploy to a Hugging Face Space
5
+ using token-based authentication, with the token loaded from a .env file.
6
+
7
+ Usage:
8
+ python deploy_to_huggingface_dotenv.py [--space SPACE_ID]
9
+
10
+ Environment variables (from .env file):
11
+ HF_TOKEN - Your Hugging Face API token
12
+ """
13
+ import os
14
+ import subprocess
15
+ import sys
16
+ import argparse
17
+
18
+ def parse_args():
19
+ parser = argparse.ArgumentParser(description="Deploy to Hugging Face Space using .env file")
20
+ parser.add_argument("--space", help="Space ID (username/space-name)",
21
+ default="JoachimVC/Final_Assignment_Template")
22
+ return parser.parse_args()
23
+
24
+ def load_dotenv():
25
+ """Load environment variables from .env file"""
26
+ env_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), '.env')
27
+ if not os.path.exists(env_path):
28
+ print(f"ERROR: .env file not found at {env_path}")
29
+ return False
30
+
31
+ print(f"Loading environment variables from {env_path}")
32
+ with open(env_path, 'r') as file:
33
+ for line in file:
34
+ line = line.strip()
35
+ if not line or line.startswith('#'):
36
+ continue
37
+
38
+ key, value = line.split('=', 1)
39
+ os.environ[key] = value
40
+
41
+ return True
42
+
43
+ def run_command(cmd, silent=False):
44
+ """Run a shell command and return output"""
45
+ if not silent:
46
+ print(f"Running: {cmd}")
47
+ try:
48
+ process = subprocess.Popen(
49
+ cmd,
50
+ shell=True,
51
+ stdout=subprocess.PIPE,
52
+ stderr=subprocess.PIPE,
53
+ text=True
54
+ )
55
+ stdout, stderr = process.communicate()
56
+
57
+ if process.returncode != 0:
58
+ print(f"Command failed with error code {process.returncode}")
59
+ print(f"Error: {stderr.strip()}")
60
+ return False
61
+
62
+ if not silent and stdout.strip():
63
+ print(stdout.strip())
64
+
65
+ return True
66
+ except Exception as e:
67
+ print(f"Failed to execute command: {str(e)}")
68
+ return False
69
+
70
+ def check_git_installed():
71
+ """Check if git is installed"""
72
+ print("Checking if git is installed...")
73
+ return run_command("git --version", silent=True)
74
+
75
+ def get_token_from_env():
76
+ """Get Hugging Face token from environment variables"""
77
+ # Try various environment variable names that might contain the token
78
+ token_vars = ["HF_TOKEN", "HUGGINGFACE_TOKEN", "HUGGING_FACE_TOKEN",
79
+ "HF_API_TOKEN", "HUGGINGFACE_API_TOKEN"]
80
+
81
+ for var_name in token_vars:
82
+ token = os.environ.get(var_name)
83
+ if token:
84
+ print(f"Found token in environment variable: {var_name}")
85
+ return token
86
+
87
+ return None
88
+
89
+ def setup_git(space_id, token):
90
+ """Initialize git if needed and configure remote with token"""
91
+ # Check if git is already initialized
92
+ if not os.path.exists(".git"):
93
+ print("Initializing git repository...")
94
+ if not run_command("git init"):
95
+ return False
96
+
97
+ # Configure git user if not set
98
+ result = subprocess.run("git config user.name", shell=True, capture_output=True, text=True)
99
+ if not result.stdout.strip():
100
+ print("Setting up git user name...")
101
+ run_command('git config user.name "GAIA Deployment"')
102
+
103
+ result = subprocess.run("git config user.email", shell=True, capture_output=True, text=True)
104
+ if not result.stdout.strip():
105
+ print("Setting up git user email...")
106
+ run_command('git config user.email "gaia-deployment@example.com"')
107
+
108
+ # Set up the Hugging Face remote with token
109
+ print("Setting up git remote for Hugging Face...")
110
+
111
+ # Check if remote exists
112
+ result = subprocess.run("git remote -v", shell=True, capture_output=True, text=True)
113
+ remote_output = result.stdout
114
+
115
+ if "huggingface" in remote_output:
116
+ run_command("git remote remove huggingface")
117
+
118
+ # Add the HF remote with token authentication
119
+ token_url = f"https://{token}@huggingface.co/spaces/{space_id}"
120
+
121
+ # Don't print the token in the console output
122
+ print(f"Adding remote: huggingface -> https://****@huggingface.co/spaces/{space_id}")
123
+ return run_command(f"git remote add huggingface {token_url}")
124
+
125
+ def create_gitignore():
126
+ """Create .gitignore file with common patterns"""
127
+ patterns = [
128
+ ".env",
129
+ "__pycache__/",
130
+ "*.py[cod]",
131
+ "*$py.class",
132
+ ".Python",
133
+ "env/",
134
+ "venv/",
135
+ ".venv",
136
+ ".coverage",
137
+ "*.log",
138
+ "deploy_with_token.py",
139
+ "deploy_to_huggingface_env.py",
140
+ "deploy_to_huggingface_dotenv.py" # Don't include this file with token handling
141
+ ]
142
+
143
+ print("Creating .gitignore file...")
144
+ with open(".gitignore", "w") as f:
145
+ for pattern in patterns:
146
+ f.write(f"{pattern}\n")
147
+
148
+ return True
149
+
150
+ def commit_changes():
151
+ """Stage and commit all changes"""
152
+ print("Staging files for commit...")
153
+ if not run_command("git add -A"):
154
+ return False
155
+
156
+ print("Committing changes...")
157
+ return run_command('git commit -m "Deploy GAIA to Hugging Face Space"')
158
+
159
+ def push_to_huggingface():
160
+ """Push to Hugging Face Space"""
161
+ print("Pushing to Hugging Face Space...")
162
+
163
+ # Try regular push first
164
+ if run_command("git push huggingface main"):
165
+ return True
166
+
167
+ # If regular push fails, try force push
168
+ print("Regular push failed, attempting force push...")
169
+ if run_command("git push huggingface main --force"):
170
+ return True
171
+
172
+ print("Force push failed. You may need to manually push:")
173
+ print("1. Ensure your token has write permissions")
174
+ print("2. Try: git push huggingface main --force")
175
+ return False
176
+
177
+ def main():
178
+ """Run deployment process"""
179
+ args = parse_args()
180
+ space_id = args.space
181
+
182
+ # Load .env file
183
+ if not load_dotenv():
184
+ print("WARNING: Failed to load .env file, continuing with existing environment variables")
185
+
186
+ # Get token from environment
187
+ token = get_token_from_env()
188
+ if not token:
189
+ print("ERROR: No Hugging Face token found in environment variables.")
190
+ print("Please ensure your .env file contains one of the following variables:")
191
+ print(" - HF_TOKEN")
192
+ print(" - HUGGINGFACE_TOKEN")
193
+ print(" - HUGGING_FACE_TOKEN")
194
+ print(" - HF_API_TOKEN")
195
+ print(" - HUGGINGFACE_API_TOKEN")
196
+ sys.exit(1)
197
+
198
+ repo_url = f"https://huggingface.co/spaces/{space_id}"
199
+
200
+ print("\n===== DEPLOYING TO HUGGING FACE SPACE =====")
201
+ print(f"Target Space: {repo_url}")
202
+
203
+ # Ensure Git is installed
204
+ if not check_git_installed():
205
+ print("Git is not installed. Please install Git and try again.")
206
+ sys.exit(1)
207
+
208
+ # Setup steps
209
+ steps = [
210
+ ("Setting up git", lambda: setup_git(space_id, token)),
211
+ ("Creating .gitignore", create_gitignore),
212
+ ("Committing changes", commit_changes),
213
+ ("Pushing to Hugging Face", push_to_huggingface)
214
+ ]
215
+
216
+ for name, func in steps:
217
+ print(f"\n>> {name}...")
218
+ if not func():
219
+ print(f"Step '{name}' failed. Deployment process halted.")
220
+ sys.exit(1)
221
+
222
+ print("\n===== DEPLOYMENT COMPLETED SUCCESSFULLY =====")
223
+ print(f"Your project has been deployed to: {repo_url}")
224
+ print("Note: It may take a few minutes for the changes to be visible.")
225
+
226
+ if __name__ == "__main__":
227
+ main()
deploy_to_huggingface_env.py ADDED
@@ -0,0 +1,203 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ Environment-based Git Deployment Script for Hugging Face Spaces
3
+
4
+ This script handles the git operations required to deploy to a Hugging Face Space
5
+ using token-based authentication, with the token taken from environment variables.
6
+
7
+ Usage:
8
+ python deploy_to_huggingface_env.py [--space SPACE_ID]
9
+
10
+ Environment variables:
11
+ HUGGINGFACE_TOKEN - Your Hugging Face API token
12
+ """
13
+ import os
14
+ import subprocess
15
+ import sys
16
+ import argparse
17
+
18
+ def parse_args():
19
+ parser = argparse.ArgumentParser(description="Deploy to Hugging Face Space using environment token")
20
+ parser.add_argument("--space", help="Space ID (username/space-name)",
21
+ default="JoachimVC/Final_Assignment_Template")
22
+ return parser.parse_args()
23
+
24
+ def run_command(cmd, silent=False):
25
+ """Run a shell command and return output"""
26
+ if not silent:
27
+ print(f"Running: {cmd}")
28
+ try:
29
+ process = subprocess.Popen(
30
+ cmd,
31
+ shell=True,
32
+ stdout=subprocess.PIPE,
33
+ stderr=subprocess.PIPE,
34
+ text=True
35
+ )
36
+ stdout, stderr = process.communicate()
37
+
38
+ if process.returncode != 0:
39
+ print(f"Command failed with error code {process.returncode}")
40
+ print(f"Error: {stderr.strip()}")
41
+ return False
42
+
43
+ if not silent and stdout.strip():
44
+ print(stdout.strip())
45
+
46
+ return True
47
+ except Exception as e:
48
+ print(f"Failed to execute command: {str(e)}")
49
+ return False
50
+
51
+ def check_git_installed():
52
+ """Check if git is installed"""
53
+ print("Checking if git is installed...")
54
+ return run_command("git --version", silent=True)
55
+
56
+ def get_token_from_env():
57
+ """Get Hugging Face token from environment variables"""
58
+ # Try various environment variable names that might contain the token
59
+ token_vars = ["HUGGINGFACE_TOKEN", "HF_TOKEN", "HUGGING_FACE_TOKEN",
60
+ "HF_API_TOKEN", "HUGGINGFACE_API_TOKEN"]
61
+
62
+ for var_name in token_vars:
63
+ token = os.environ.get(var_name)
64
+ if token:
65
+ print(f"Found token in environment variable: {var_name}")
66
+ return token
67
+
68
+ return None
69
+
70
+ def setup_git(space_id, token):
71
+ """Initialize git if needed and configure remote with token"""
72
+ # Check if git is already initialized
73
+ if not os.path.exists(".git"):
74
+ print("Initializing git repository...")
75
+ if not run_command("git init"):
76
+ return False
77
+
78
+ # Configure git user if not set
79
+ result = subprocess.run("git config user.name", shell=True, capture_output=True, text=True)
80
+ if not result.stdout.strip():
81
+ print("Setting up git user name...")
82
+ run_command('git config user.name "GAIA Deployment"')
83
+
84
+ result = subprocess.run("git config user.email", shell=True, capture_output=True, text=True)
85
+ if not result.stdout.strip():
86
+ print("Setting up git user email...")
87
+ run_command('git config user.email "gaia-deployment@example.com"')
88
+
89
+ # Set up the Hugging Face remote with token
90
+ print("Setting up git remote for Hugging Face...")
91
+
92
+ # Check if remote exists
93
+ result = subprocess.run("git remote -v", shell=True, capture_output=True, text=True)
94
+ remote_output = result.stdout
95
+
96
+ if "huggingface" in remote_output:
97
+ run_command("git remote remove huggingface")
98
+
99
+ # Add the HF remote with token authentication
100
+ token_url = f"https://{token}@huggingface.co/spaces/{space_id}"
101
+
102
+ # Don't print the token in the console output
103
+ print(f"Adding remote: huggingface -> https://****@huggingface.co/spaces/{space_id}")
104
+ return run_command(f"git remote add huggingface {token_url}")
105
+
106
+ def create_gitignore():
107
+ """Create .gitignore file with common patterns"""
108
+ patterns = [
109
+ ".env",
110
+ "__pycache__/",
111
+ "*.py[cod]",
112
+ "*$py.class",
113
+ ".Python",
114
+ "env/",
115
+ "venv/",
116
+ ".venv",
117
+ ".coverage",
118
+ "*.log",
119
+ "deploy_with_token.py",
120
+ "deploy_to_huggingface_env.py" # Don't include this file with token handling
121
+ ]
122
+
123
+ print("Creating .gitignore file...")
124
+ with open(".gitignore", "w") as f:
125
+ for pattern in patterns:
126
+ f.write(f"{pattern}\n")
127
+
128
+ return True
129
+
130
+ def commit_changes():
131
+ """Stage and commit all changes"""
132
+ print("Staging files for commit...")
133
+ if not run_command("git add -A"):
134
+ return False
135
+
136
+ print("Committing changes...")
137
+ return run_command('git commit -m "Deploy GAIA to Hugging Face Space"')
138
+
139
+ def push_to_huggingface():
140
+ """Push to Hugging Face Space"""
141
+ print("Pushing to Hugging Face Space...")
142
+
143
+ # Try regular push first
144
+ if run_command("git push huggingface main"):
145
+ return True
146
+
147
+ # If regular push fails, try force push
148
+ print("Regular push failed, attempting force push...")
149
+ if run_command("git push huggingface main --force"):
150
+ return True
151
+
152
+ print("Force push failed. You may need to manually push:")
153
+ print("1. Ensure your token has write permissions")
154
+ print("2. Try: git push huggingface main --force")
155
+ return False
156
+
157
+ def main():
158
+ """Run deployment process"""
159
+ args = parse_args()
160
+ space_id = args.space
161
+
162
+ # Get token from environment
163
+ token = get_token_from_env()
164
+ if not token:
165
+ print("ERROR: No Hugging Face token found in environment variables.")
166
+ print("Please set one of the following environment variables:")
167
+ print(" - HUGGINGFACE_TOKEN")
168
+ print(" - HF_TOKEN")
169
+ print(" - HUGGING_FACE_TOKEN")
170
+ print(" - HF_API_TOKEN")
171
+ print(" - HUGGINGFACE_API_TOKEN")
172
+ sys.exit(1)
173
+
174
+ repo_url = f"https://huggingface.co/spaces/{space_id}"
175
+
176
+ print("\n===== DEPLOYING TO HUGGING FACE SPACE =====")
177
+ print(f"Target Space: {repo_url}")
178
+
179
+ # Ensure Git is installed
180
+ if not check_git_installed():
181
+ print("Git is not installed. Please install Git and try again.")
182
+ sys.exit(1)
183
+
184
+ # Setup steps
185
+ steps = [
186
+ ("Setting up git", lambda: setup_git(space_id, token)),
187
+ ("Creating .gitignore", create_gitignore),
188
+ ("Committing changes", commit_changes),
189
+ ("Pushing to Hugging Face", push_to_huggingface)
190
+ ]
191
+
192
+ for name, func in steps:
193
+ print(f"\n>> {name}...")
194
+ if not func():
195
+ print(f"Step '{name}' failed. Deployment process halted.")
196
+ sys.exit(1)
197
+
198
+ print("\n===== DEPLOYMENT COMPLETED SUCCESSFULLY =====")
199
+ print(f"Your project has been deployed to: {repo_url}")
200
+ print("Note: It may take a few minutes for the changes to be visible.")
201
+
202
+ if __name__ == "__main__":
203
+ main()
deploy_with_token.py ADDED
@@ -0,0 +1,176 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """
2
+ Token-Based Git Deployment Script for Hugging Face Spaces
3
+
4
+ This script handles the git operations required to deploy to a Hugging Face Space
5
+ using token-based authentication.
6
+
7
+ Usage:
8
+ python deploy_with_token.py --token YOUR_HF_TOKEN
9
+
10
+ """
11
+ import os
12
+ import subprocess
13
+ import sys
14
+ import argparse
15
+
16
+ def parse_args():
17
+ parser = argparse.ArgumentParser(description="Deploy to Hugging Face Space with token")
18
+ parser.add_argument("--token", help="Hugging Face API token", required=True)
19
+ parser.add_argument("--space", help="Space ID (username/space-name)",
20
+ default="JoachimVC/Final_Assignment_Template")
21
+ return parser.parse_args()
22
+
23
+ def run_command(cmd, silent=False):
24
+ """Run a shell command and return output"""
25
+ if not silent:
26
+ print(f"Running: {cmd}")
27
+ try:
28
+ process = subprocess.Popen(
29
+ cmd,
30
+ shell=True,
31
+ stdout=subprocess.PIPE,
32
+ stderr=subprocess.PIPE,
33
+ text=True
34
+ )
35
+ stdout, stderr = process.communicate()
36
+
37
+ if process.returncode != 0:
38
+ print(f"Command failed with error code {process.returncode}")
39
+ print(f"Error: {stderr.strip()}")
40
+ return False
41
+
42
+ if not silent and stdout.strip():
43
+ print(stdout.strip())
44
+
45
+ return True
46
+ except Exception as e:
47
+ print(f"Failed to execute command: {str(e)}")
48
+ return False
49
+
50
+ def check_git_installed():
51
+ """Check if git is installed"""
52
+ print("Checking if git is installed...")
53
+ return run_command("git --version", silent=True)
54
+
55
+ def setup_git(space_id, token):
56
+ """Initialize git if needed and configure remote with token"""
57
+ # Check if git is already initialized
58
+ if not os.path.exists(".git"):
59
+ print("Initializing git repository...")
60
+ if not run_command("git init"):
61
+ return False
62
+
63
+ # Configure git user if not set
64
+ result = subprocess.run("git config user.name", shell=True, capture_output=True, text=True)
65
+ if not result.stdout.strip():
66
+ print("Setting up git user name...")
67
+ run_command('git config user.name "GAIA Deployment"')
68
+
69
+ result = subprocess.run("git config user.email", shell=True, capture_output=True, text=True)
70
+ if not result.stdout.strip():
71
+ print("Setting up git user email...")
72
+ run_command('git config user.email "gaia-deployment@example.com"')
73
+
74
+ # Set up the Hugging Face remote with token
75
+ print("Setting up git remote for Hugging Face...")
76
+
77
+ # Check if remote exists
78
+ result = subprocess.run("git remote -v", shell=True, capture_output=True, text=True)
79
+ remote_output = result.stdout
80
+
81
+ if "huggingface" in remote_output:
82
+ run_command("git remote remove huggingface")
83
+
84
+ # Add the HF remote with token authentication
85
+ token_url = f"https://{token}@huggingface.co/spaces/{space_id}"
86
+
87
+ # Don't print the token in the console output
88
+ print(f"Adding remote: huggingface -> https://****@huggingface.co/spaces/{space_id}")
89
+ return run_command(f"git remote add huggingface {token_url}")
90
+
91
+ def create_gitignore():
92
+ """Create .gitignore file with common patterns"""
93
+ patterns = [
94
+ ".env",
95
+ "__pycache__/",
96
+ "*.py[cod]",
97
+ "*$py.class",
98
+ ".Python",
99
+ "env/",
100
+ "venv/",
101
+ ".venv",
102
+ ".coverage",
103
+ "*.log",
104
+ "deploy_with_token.py" # Don't include this file with token handling
105
+ ]
106
+
107
+ print("Creating .gitignore file...")
108
+ with open(".gitignore", "w") as f:
109
+ for pattern in patterns:
110
+ f.write(f"{pattern}\n")
111
+
112
+ return True
113
+
114
+ def commit_changes():
115
+ """Stage and commit all changes"""
116
+ print("Staging files for commit...")
117
+ if not run_command("git add -A"):
118
+ return False
119
+
120
+ print("Committing changes...")
121
+ return run_command('git commit -m "Deploy GAIA to Hugging Face Space"')
122
+
123
+ def push_to_huggingface():
124
+ """Push to Hugging Face Space"""
125
+ print("Pushing to Hugging Face Space...")
126
+
127
+ # Try regular push first
128
+ if run_command("git push huggingface main"):
129
+ return True
130
+
131
+ # If regular push fails, try force push
132
+ print("Regular push failed, attempting force push...")
133
+ if run_command("git push huggingface main --force"):
134
+ return True
135
+
136
+ print("Force push failed. You may need to manually push:")
137
+ print("1. Ensure your token has write permissions")
138
+ print("2. Try: git push huggingface main --force")
139
+ return False
140
+
141
+ def main():
142
+ """Run deployment process"""
143
+ args = parse_args()
144
+ token = args.token
145
+ space_id = args.space
146
+
147
+ repo_url = f"https://huggingface.co/spaces/{space_id}"
148
+
149
+ print("\n===== DEPLOYING TO HUGGING FACE SPACE =====")
150
+ print(f"Target Space: {repo_url}")
151
+
152
+ # Ensure Git is installed
153
+ if not check_git_installed():
154
+ print("Git is not installed. Please install Git and try again.")
155
+ sys.exit(1)
156
+
157
+ # Setup steps
158
+ steps = [
159
+ ("Setting up git", lambda: setup_git(space_id, token)),
160
+ ("Creating .gitignore", create_gitignore),
161
+ ("Committing changes", commit_changes),
162
+ ("Pushing to Hugging Face", push_to_huggingface)
163
+ ]
164
+
165
+ for name, func in steps:
166
+ print(f"\n>> {name}...")
167
+ if not func():
168
+ print(f"Step '{name}' failed. Deployment process halted.")
169
+ sys.exit(1)
170
+
171
+ print("\n===== DEPLOYMENT COMPLETED SUCCESSFULLY =====")
172
+ print(f"Your project has been deployed to: {repo_url}")
173
+ print("Note: It may take a few minutes for the changes to be visible.")
174
+
175
+ if __name__ == "__main__":
176
+ main()
deployment_readiness_report.md ADDED
@@ -0,0 +1,97 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # GAIA Deployment Readiness Report
2
+
3
+ ## 1. Deployment Status: ✅ SUCCESSFUL
4
+
5
+ The GAIA agent system has been successfully deployed to Hugging Face Space at:
6
+ `https://huggingface.co/spaces/JoachimVC/Final_Assignment_Template`
7
+
8
+ **Deployment Timestamp:** April 30, 2025
9
+
10
+ ## 2. Environment Configuration
11
+
12
+ | Category | Status | Details |
13
+ |----------|--------|---------|
14
+ | Code Files | ✅ Complete | All necessary code files have been uploaded |
15
+ | Environment Variables | ✅ Configured | All required API keys and configuration variables set |
16
+ | Secret Management | ✅ Secure | Using Hugging Face Secrets for sensitive credentials |
17
+
18
+ ### Environment Variables Configured:
19
+ - `OPENAI_API_KEY` ✅
20
+ - `SERPER_API_KEY` ✅
21
+ - `PERPLEXITY_API_KEY` ✅
22
+ - `SUPABASE_URL` ✅
23
+ - `SUPABASE_KEY` ✅
24
+ - `HF_TOKEN` ✅
25
+
26
+ ## 3. Component Integration Status
27
+
28
+ | Component | Status | Verification |
29
+ |-----------|--------|-------------|
30
+ | OpenAI API | ✅ Working | Successfully connected to API and retrieved model list |
31
+ | Serper Search | ✅ Working | Successfully performed search query with expected results |
32
+ | Supabase Memory | ⚠️ Needs Verification | Environment variables set, but actual integration needs testing |
33
+ | DuckDuckGo Search | ⚠️ Needs Verification | Functionality available but needs end-to-end testing |
34
+ | Perplexity | ⚠️ Needs Verification | API key configured, but integration needs testing |
35
+ | Agent Graph Workflow | ⚠️ Needs Verification | Core functionality needs verification in the deployed environment |
36
+
37
+ ## 4. Cleanup Recommendations
38
+
39
+ Several files and directories in the deployed space are not required for runtime operation and can be safely removed to optimize the environment. Refer to `huggingface_space_cleanup.md` for a detailed list of files that can be removed, including:
40
+
41
+ - Test directories (`tests/`, `backup_tests/`)
42
+ - Development configuration (`.roo/`, `.roomodes/`, etc.)
43
+ - Deployment and cleaning scripts
44
+ - Non-essential documentation
45
+
46
+ ## 5. Verification Steps
47
+
48
+ The following steps are recommended to complete verification of the deployed system:
49
+
50
+ 1. **Run Production Verification Script**
51
+ - Execute `verify_production.py` on the Hugging Face Space to test all components
52
+ - This script will generate a detailed report on component status
53
+
54
+ 2. **Test Conversation Memory**
55
+ - Initiate a multi-turn conversation that relies on memory
56
+ - Verify that the agent can recall information from previous interactions
57
+
58
+ 3. **Test Search Tool Integration**
59
+ - Ask a question requiring web search
60
+ - Verify that search results are properly integrated into responses
61
+
62
+ 4. **Performance Monitoring**
63
+ - Monitor response times for different query types
64
+ - Check for any memory leaks or increasing latency over time
65
+
66
+ ## 6. Recommendations
67
+
68
+ 1. **Memory Persistence**
69
+ - Verify Supabase table structure matches the required schema
70
+ - Check that memory persistence works across session restarts
71
+
72
+ 2. **Error Handling**
73
+ - Implement robust error handling for API failures
74
+ - Add automatic fallback to alternative models if primary model is unavailable
75
+
76
+ 3. **Monitoring**
77
+ - Set up periodic health checks to verify all integrations remain functional
78
+ - Implement usage tracking to monitor API consumption and costs
79
+
80
+ 4. **Documentation**
81
+ - Update user-facing documentation with examples of supported capabilities
82
+ - Provide clear guidance on limitations and expected behaviors
83
+
84
+ ## 7. Rollback Plan
85
+
86
+ In case issues are discovered in the deployed system:
87
+
88
+ 1. Identify the specific component causing the issue
89
+ 2. If possible, fix directly in the Hugging Face interface
90
+ 3. For more complex issues, fix locally and redeploy
91
+ 4. If critical failure occurs, consider reverting to previous stable version
92
+
93
+ ## Conclusion
94
+
95
+ The GAIA system has been successfully deployed to Hugging Face Space with all required environment variables configured. Initial component testing shows successful integration with OpenAI and Serper APIs. Comprehensive testing of the Supabase memory integration and other components is recommended to ensure full functionality in the production environment.
96
+
97
+ The deployment is currently in a good state for final verification and production use, with minor cleanup recommended to optimize the space.
docs/all_apis_integration_testing.md ADDED
@@ -0,0 +1,170 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # All APIs Integration Testing
2
+
3
+ This document provides information about the comprehensive integration test for all APIs in the GAIA agent.
4
+
5
+ ## Overview
6
+
7
+ The integration test verifies that all APIs work together seamlessly in the GAIA agent. It tests:
8
+
9
+ 1. **OpenAI** - For language model capabilities
10
+ 2. **Perplexity** - For enhanced search and reasoning
11
+ 3. **Serper** - For Google search results
12
+ 4. **DuckDuckGo** - For web search without API key requirements
13
+ 5. **arXiv** - For academic and scientific paper search
14
+
15
+ The test ensures that the agent can switch between different tools based on the task and maintain memory across different tool executions.
16
+
17
+ ## Prerequisites
18
+
19
+ Before running the integration test, ensure you have the following environment variables set:
20
+
21
+ - `OPENAI_API_KEY` (required) - Your OpenAI API key
22
+ - `SERPER_API_KEY` (optional) - Your Serper API key for Google search
23
+ - `PERPLEXITY_API_KEY` (optional) - Your Perplexity API key
24
+
25
+ If optional API keys are not provided, the corresponding tests will be skipped.
26
+
27
+ ## Running the Test
28
+
29
+ You can run the integration test using the provided runner script:
30
+
31
+ ```bash
32
+ python tests/run_all_apis_integration_test.py
33
+ ```
34
+
35
+ ### Command-line Options
36
+
37
+ - `--verbose` or `-v`: Enable verbose logging for detailed output
38
+ - `--report-file` or `-r`: Specify a custom output file for the test report (default: `gaia_all_apis_integration_report.json`)
39
+
40
+ Example with options:
41
+
42
+ ```bash
43
+ python tests/run_all_apis_integration_test.py --verbose --report-file my_report.json
44
+ ```
45
+
46
+ ## Test Components
47
+
48
+ The integration test consists of several components:
49
+
50
+ ### 1. API Connection Tests
51
+
52
+ Verifies that the agent can connect to each API service. For each API, it:
53
+ - Checks if the required API key is available
54
+ - Attempts to establish a connection
55
+ - Verifies that the connection is successful
56
+
57
+ ### 2. Individual Tool Execution Tests
58
+
59
+ Tests each tool individually with simple queries to ensure they work correctly:
60
+ - DuckDuckGo search
61
+ - Serper search
62
+ - Perplexity search
63
+ - arXiv search
64
+ - Robust search (fallback mechanism)
65
+ - Merged search (combining results from multiple providers)
66
+
67
+ ### 3. Tool Switching Tests
68
+
69
+ Verifies that the agent can switch between different tools based on the task:
70
+ - Scientific questions (should use arXiv and Perplexity)
71
+ - Factual questions (should use DuckDuckGo and Serper)
72
+ - Explanatory questions (should use Perplexity)
73
+
74
+ ### 4. Memory Integration Tests
75
+
76
+ Tests the agent's ability to maintain memory across different tool executions:
77
+ - Processes a sequence of related questions
78
+ - Verifies that the conversation history is maintained
79
+ - Checks for cached results and intermediate results
80
+
81
+ ## Test Report
82
+
83
+ After running the test, a comprehensive report is generated in JSON format. The report includes:
84
+
85
+ - API connection results
86
+ - Tool execution results
87
+ - Complex query results
88
+ - Memory integration results
89
+ - Summary statistics
90
+
91
+ The report is displayed in the console and saved to the specified file.
92
+
93
+ ## Interpreting Results
94
+
95
+ The test results are displayed with the following indicators:
96
+
97
+ - ✅ **PASS**: The test was successful
98
+ - ❌ **FAIL**: The test failed
99
+ - ⚠️ **SKIP**: The test was skipped (usually due to missing API keys)
100
+
101
+ ### Example Output
102
+
103
+ ```
104
+ ================================================================================
105
+ GAIA ALL APIs INTEGRATION TEST REPORT
106
+ ================================================================================
107
+
108
+ --------------------------------------------------------------------------------
109
+ SUMMARY
110
+ --------------------------------------------------------------------------------
111
+ Timestamp: 2025-04-30 18:00:00
112
+ Total APIs Tested: 5
113
+ Successful Connections: 4
114
+ Successful Tool Executions: 6
115
+ Successful Complex Queries: 3
116
+ Memory Integration Success: True
117
+
118
+ --------------------------------------------------------------------------------
119
+ API CONNECTIONS
120
+ --------------------------------------------------------------------------------
121
+ ✅ PASS | openai: Successfully connected to OpenAI API
122
+ ✅ PASS | perplexity: Successfully connected to Perplexity API
123
+ ⚠️ SKIP | serper: Serper API key not available
124
+ ✅ PASS | duckduckgo: Successfully connected to DuckDuckGo
125
+ ✅ PASS | arxiv: Successfully connected to arXiv
126
+
127
+ ...
128
+ ```
129
+
130
+ ## Troubleshooting
131
+
132
+ ### Missing API Keys
133
+
134
+ If you see warnings about missing API keys, ensure that the environment variables are properly set:
135
+
136
+ ```bash
137
+ export OPENAI_API_KEY=your_openai_api_key
138
+ export SERPER_API_KEY=your_serper_api_key
139
+ export PERPLEXITY_API_KEY=your_perplexity_api_key
140
+ ```
141
+
142
+ ### Connection Issues
143
+
144
+ If API connection tests fail:
145
+ 1. Check your internet connection
146
+ 2. Verify that your API keys are valid and not expired
147
+ 3. Check if the API service is experiencing downtime
148
+
149
+ ### Test Failures
150
+
151
+ If specific tests fail:
152
+ 1. Check the error message in the report
153
+ 2. Look for detailed error logs if running with `--verbose`
154
+ 3. Verify that the required dependencies are installed
155
+ 4. Check if there are any rate limits or quotas on the APIs
156
+
157
+ ## Extending the Test
158
+
159
+ To add more test cases or test additional APIs:
160
+
161
+ 1. Add new test methods to the `TestGAIAAllAPIsIntegration` class in `tests/test_gaia_all_apis_integration.py`
162
+ 2. Update the report generation to include the new test results
163
+ 3. Update this documentation to reflect the changes
164
+
165
+ ## Related Documentation
166
+
167
+ - [OpenAI Integration Testing](openai_integration_testing.md)
168
+ - [Serper Integration](serper_integration.md)
169
+ - [DuckDuckGo Integration](duckduckgo_integration.md)
170
+ - [Multi-Search Integration](multi_search_integration.md)
docs/configuration.md ADDED
@@ -0,0 +1,357 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # GAIA Agent Configuration Guide
2
+
3
+ This document provides detailed information about the configuration options available for the GAIA agent. Understanding these options will help you customize the agent's behavior to suit your specific needs.
4
+
5
+ ## Configuration Overview
6
+
7
+ The GAIA agent uses a hierarchical configuration structure with the following main sections:
8
+
9
+ 1. **Model Configuration**: Settings related to the language models used by the agent
10
+ 2. **Tools Configuration**: Settings for the various tools used by the agent
11
+ 3. **Memory Configuration**: Settings for the agent's memory components
12
+ 4. **Agent Configuration**: General settings for the agent's behavior
13
+
14
+ ## Configuration Methods
15
+
16
+ There are several ways to configure the GAIA agent:
17
+
18
+ 1. **Environment Variables**: Set configuration values using environment variables
19
+ 2. **Configuration Files**: Use a `.env` file or JSON configuration files
20
+ 3. **Direct Configuration**: Pass a configuration dictionary when initializing the agent
21
+
22
+ ## Model Configuration
23
+
24
+ The model configuration controls which language models are used by the agent and how they behave.
25
+
26
+ ```python
27
+ model_config = {
28
+ "reasoning_model": "gpt-4o", # Model used for reasoning tasks (updated from gpt-4-turbo)
29
+ "tool_model": "gpt-3.5-turbo", # Model used for tool execution
30
+ "vision_model": "gpt-4o", # Model used for image analysis (updated from gpt-4-vision-preview)
31
+ "embedding_model": "text-embedding-3-large", # Model used for embeddings (updated from text-embedding-ada-002)
32
+ "temperature": 0.1, # Controls randomness (0.0 to 1.0)
33
+ "max_tokens": 4096, # Maximum number of tokens to generate
34
+ "top_p": 0.95, # Controls diversity via nucleus sampling
35
+ "frequency_penalty": 0.0, # Penalizes repeated tokens
36
+ "presence_penalty": 0.0 # Penalizes repeated topics
37
+ }
38
+ ```
39
+
40
+ ### Environment Variables
41
+
42
+ ```
43
+ OPENAI_API_KEY=your-api-key
44
+ REASONING_MODEL=gpt-4o
45
+ TOOL_MODEL=gpt-3.5-turbo
46
+ VISION_MODEL=gpt-4o
47
+ EMBEDDING_MODEL=text-embedding-3-large
48
+ MODEL_TEMPERATURE=0.1
49
+ MODEL_MAX_TOKENS=4096
50
+ ```
51
+
52
+ ## Tools Configuration
53
+
54
+ The tools configuration controls which tools are available to the agent and how they behave.
55
+
56
+ ```python
57
+ tools_config = {
58
+ "web_search": {
59
+ "enabled": True,
60
+ "providers": ["duckduckgo", "serper"],
61
+ "result_count": 5,
62
+ "timeout": 10
63
+ },
64
+ "web_content": {
65
+ "enabled": True,
66
+ "max_content_length": 10000,
67
+ "timeout": 15
68
+ },
69
+ "reasoning": {
70
+ "enabled": True,
71
+ "step_by_step": True,
72
+ "verification": True
73
+ },
74
+ "multimodal": {
75
+ "enabled": True,
76
+ "image_analysis": True,
77
+ "chart_interpretation": True,
78
+ "document_parsing": True
79
+ },
80
+ "arxiv": {
81
+ "enabled": True,
82
+ "max_results": 5,
83
+ "timeout": 10
84
+ },
85
+ "perplexity": {
86
+ "enabled": True,
87
+ "model": "sonar-medium-online",
88
+ "timeout": 15
89
+ }
90
+ }
91
+ ```
92
+
93
+ ### Environment Variables
94
+
95
+ ```
96
+ ENABLE_WEB_SEARCH=true
97
+ SEARCH_PROVIDERS=duckduckgo,serper
98
+ SEARCH_RESULT_COUNT=5
99
+ SEARCH_TIMEOUT=10
100
+ ENABLE_WEB_CONTENT=true
101
+ MAX_CONTENT_LENGTH=10000
102
+ ENABLE_REASONING=true
103
+ ENABLE_MULTIMODAL=true
104
+ ENABLE_ARXIV=true
105
+ ARXIV_MAX_RESULTS=5
106
+ ENABLE_PERPLEXITY=true
107
+ PERPLEXITY_MODEL=sonar-medium-online
108
+ ```
109
+
110
+ ## Memory Configuration
111
+
112
+ The memory configuration controls how the agent stores and retrieves information.
113
+
114
+ ```python
115
+ memory_config = {
116
+ "supabase": {
117
+ "url": "https://tjamxhvvtnypbadvrkjq.supabase.co",
118
+ "key": "your-api-key",
119
+ "project_id": "tjamxhvvtnypbadvrkjq",
120
+ "project_name": "HF_GAIA_Assessment_AGT",
121
+ "table": "gaia_memory"
122
+ },
123
+ "conversation_memory": {
124
+ "enabled": True,
125
+ "max_messages": 100,
126
+ "table": "gaia_memory_conversations"
127
+ },
128
+ "result_cache": {
129
+ "enabled": True,
130
+ "ttl": 3600, # Time-to-live in seconds
131
+ "table": "gaia_memory_cache"
132
+ },
133
+ "working_memory": {
134
+ "enabled": True,
135
+ "max_items": 50,
136
+ "table": "gaia_memory_working"
137
+ }
138
+ }
139
+ ```
140
+
141
+ ### Environment Variables
142
+
143
+ ```
144
+ # Supabase Configuration
145
+ SUPABASE_URL=https://tjamxhvvtnypbadvrkjq.supabase.co
146
+ SUPABASE_KEY=your-api-key
147
+ SUPABASE_PROJECT_ID=tjamxhvvtnypbadvrkjq
148
+ SUPABASE_PROJECT_NAME=HF_GAIA_Assessment_AGT
149
+ MEMORY_TABLE_NAME=gaia_memory
150
+
151
+ # Memory Settings
152
+ MEMORY_ENABLED=true
153
+ MEMORY_TTL=86400
154
+ MEMORY_CACHE_SIZE=100
155
+ MEMORY_SIMILARITY_THRESHOLD=0.85
156
+ MEMORY_MAX_ENTRIES=1000
157
+ ```
158
+
159
+ ## Agent Configuration
160
+
161
+ The agent configuration controls the general behavior of the agent.
162
+
163
+ ```python
164
+ agent_config = {
165
+ "max_retries": 3, # Maximum number of retries for failed operations
166
+ "retry_delay": 1.0, # Delay between retries in seconds
167
+ "timeout": 300, # Overall timeout for processing a question
168
+ "max_iterations": 10, # Maximum number of iterations in the workflow
169
+ "return_error_message": False, # Whether to return error messages or raise exceptions
170
+ "verbose": False # Whether to enable verbose logging
171
+ }
172
+ ```
173
+
174
+ ### Environment Variables
175
+
176
+ ```
177
+ AGENT_MAX_RETRIES=3
178
+ AGENT_RETRY_DELAY=1.0
179
+ AGENT_TIMEOUT=300
180
+ AGENT_MAX_ITERATIONS=10
181
+ RETURN_ERROR_MESSAGE=false
182
+ VERBOSE=false
183
+ ```
184
+
185
+ ## Complete Configuration Example
186
+
187
+ Here's a complete example of a configuration dictionary that you can pass when initializing the agent:
188
+
189
+ ```python
190
+ config = {
191
+ "model": {
192
+ "reasoning_model": "gpt-4o",
193
+ "tool_model": "gpt-3.5-turbo",
194
+ "vision_model": "gpt-4o",
195
+ "embedding_model": "text-embedding-3-large",
196
+ "temperature": 0.1,
197
+ "max_tokens": 4096
198
+ },
199
+ "tools": {
200
+ "web_search": {
201
+ "enabled": True,
202
+ "providers": ["duckduckgo", "serper"],
203
+ "result_count": 5,
204
+ "timeout": 10
205
+ },
206
+ "web_content": {
207
+ "enabled": True,
208
+ "max_content_length": 10000,
209
+ "timeout": 15
210
+ },
211
+ "reasoning": {
212
+ "enabled": True,
213
+ "step_by_step": True,
214
+ "verification": True
215
+ },
216
+ "multimodal": {
217
+ "enabled": True,
218
+ "image_analysis": True,
219
+ "chart_interpretation": True,
220
+ "document_parsing": True
221
+ },
222
+ "arxiv": {
223
+ "enabled": True,
224
+ "max_results": 5,
225
+ "timeout": 10
226
+ },
227
+ "perplexity": {
228
+ "enabled": True,
229
+ "model": "sonar-medium-online",
230
+ "timeout": 15
231
+ }
232
+ },
233
+ "memory": {
234
+ "supabase": {
235
+ "url": "https://tjamxhvvtnypbadvrkjq.supabase.co",
236
+ "key": "your-api-key",
237
+ "project_id": "tjamxhvvtnypbadvrkjq",
238
+ "project_name": "HF_GAIA_Assessment_AGT",
239
+ "table": "gaia_memory"
240
+ },
241
+ "conversation_memory": {
242
+ "enabled": True,
243
+ "max_messages": 100,
244
+ "table": "gaia_memory_conversations"
245
+ },
246
+ "result_cache": {
247
+ "enabled": True,
248
+ "ttl": 3600,
249
+ "table": "gaia_memory_cache"
250
+ },
251
+ "working_memory": {
252
+ "enabled": True,
253
+ "max_items": 50,
254
+ "table": "gaia_memory_working"
255
+ }
256
+ },
257
+ "agent": {
258
+ "max_retries": 3,
259
+ "retry_delay": 1.0,
260
+ "timeout": 300,
261
+ "max_iterations": 10,
262
+ "return_error_message": False,
263
+ "verbose": False
264
+ }
265
+ }
266
+
267
+ from agent import GAIAAgent
268
+ agent = GAIAAgent(config)
269
+ ```
270
+
271
+ ## Environment Variables File (.env)
272
+
273
+ You can also use a `.env` file to configure the agent. Here's an example:
274
+
275
+ ```
276
+ # API Keys and Endpoints
277
+ OPENAI_API_KEY=your-openai-api-key
278
+ SERPER_API_KEY=your-serper-api-key
279
+ PERPLEXITY_API_KEY=your-perplexity-api-key
280
+
281
+ # Supabase Configuration
282
+ SUPABASE_URL=https://tjamxhvvtnypbadvrkjq.supabase.co
283
+ SUPABASE_KEY=your-supabase-api-key
284
+ SUPABASE_PROJECT_ID=tjamxhvvtnypbadvrkjq
285
+ SUPABASE_PROJECT_NAME=HF_GAIA_Assessment_AGT
286
+
287
+ # Hugging Face Configuration
288
+ HF_TOKEN=your-huggingface-token
289
+
290
+ # Model Configuration
291
+ REASONING_MODEL=gpt-4o
292
+ TOOL_MODEL=gpt-3.5-turbo
293
+ VISION_MODEL=gpt-4o
294
+ EMBEDDING_MODEL=text-embedding-3-large
295
+ MODEL_TEMPERATURE=0.1
296
+ MODEL_MAX_TOKENS=4096
297
+
298
+ # Tools Configuration
299
+ ENABLE_WEB_SEARCH=true
300
+ SEARCH_PROVIDERS=duckduckgo,serper
301
+ SEARCH_RESULT_COUNT=5
302
+ ENABLE_WEB_CONTENT=true
303
+ ENABLE_REASONING=true
304
+ ENABLE_MULTIMODAL=true
305
+ ENABLE_ARXIV=true
306
+ ENABLE_PERPLEXITY=true
307
+
308
+ # Memory Configuration
309
+ ENABLE_CONVERSATION_MEMORY=true
310
+ ENABLE_RESULT_CACHE=true
311
+ ENABLE_WORKING_MEMORY=true
312
+
313
+ # Agent Configuration
314
+ AGENT_MAX_RETRIES=3
315
+ AGENT_TIMEOUT=300
316
+ VERBOSE=false
317
+ ```
318
+
319
+ ## Dependencies
320
+
321
+ The GAIA agent relies on several key dependencies:
322
+
323
+ 1. **langchain**: Core framework for building language model applications
324
+ 2. **langchain_openai**: OpenAI integration for LangChain (separate package)
325
+ 3. **langgraph**: Framework for building complex agent workflows
326
+ 4. **supabase**: Python client for Supabase
327
+ 5. **duckduckgo-search**: Client for DuckDuckGo search
328
+ 6. **perplexity-python**: Client for Perplexity AI
329
+ 7. **arxiv**: Client for arXiv search
330
+
331
+ Make sure these dependencies are installed and up-to-date:
332
+
333
+ ```bash
334
+ pip install --upgrade langchain langchain_openai langgraph supabase duckduckgo-search perplexity-python arxiv
335
+ ```
336
+
337
+ > **Note**: The `langchain_openai` package is now separate from the core `langchain` package. If you encounter import errors, make sure you have the latest version installed.
338
+
339
+ ## Configuration Precedence
340
+
341
+ The agent uses the following precedence order when determining configuration values:
342
+
343
+ 1. Direct configuration passed to the constructor
344
+ 2. Environment variables
345
+ 3. Default values defined in the code
346
+
347
+ This means that values passed directly to the constructor will override environment variables, which in turn override the default values.
348
+
349
+ ## Troubleshooting
350
+
351
+ If you encounter configuration issues, check the [Troubleshooting Guide](troubleshooting.md) for common problems and solutions.
352
+
353
+ ## Next Steps
354
+
355
+ For usage examples, see [Usage Examples](examples.md).
356
+
357
+ For deployment instructions, see [Deployment Guide](deployment.md).
docs/connection_verification.md ADDED
@@ -0,0 +1,114 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Connection Verification Guide
2
+
3
+ This guide explains how to verify connections to external services used by the GAIA agent.
4
+
5
+ ## Overview
6
+
7
+ The GAIA agent relies on several external services:
8
+
9
+ 1. **OpenAI API** - For language model access
10
+ 2. **Perplexity API** - For web search and reasoning
11
+ 3. **Supabase** - For memory storage
12
+ 4. **Serper API** - For web search
13
+
14
+ Proper connection to these services is essential for the agent to function correctly.
15
+
16
+ ## Verification Tools
17
+
18
+ The project includes several tools to verify connections:
19
+
20
+ 1. `tests/verify_connections.py` - A comprehensive test script that verifies all connections
21
+ 2. `direct_config_check.py` - A simple script to verify configuration values
22
+
23
+ ## Verifying All Connections
24
+
25
+ To verify all connections at once, run:
26
+
27
+ ```bash
28
+ python -m tests.verify_connections
29
+ ```
30
+
31
+ This script will:
32
+ - Check if all required API keys are set
33
+ - Test connections to each service
34
+ - Report success or failure for each connection
35
+
36
+ ## Verifying Individual Connections
37
+
38
+ ### OpenAI API
39
+
40
+ To verify the OpenAI API connection:
41
+
42
+ ```bash
43
+ python -m tests.verify_connections --service openai
44
+ ```
45
+
46
+ ### Perplexity API
47
+
48
+ To verify the Perplexity API connection:
49
+
50
+ ```bash
51
+ python -m tests.verify_connections --service perplexity
52
+ ```
53
+
54
+ You can also use the dedicated Perplexity test:
55
+
56
+ ```bash
57
+ python -m tests.test_perplexity_sonar
58
+ ```
59
+
60
+ ### Supabase
61
+
62
+ To verify the Supabase connection:
63
+
64
+ ```bash
65
+ python -m tests.verify_connections --service supabase
66
+ ```
67
+
68
+ ### Serper API
69
+
70
+ To verify the Serper API connection:
71
+
72
+ ```bash
73
+ python -m tests.verify_connections --service serper
74
+ ```
75
+
76
+ ## Troubleshooting Connection Issues
77
+
78
+ ### API Key Issues
79
+
80
+ If you encounter API key issues:
81
+
82
+ 1. Check that the key is correctly set in your `.env` file
83
+ 2. Verify the key format (no extra spaces or characters)
84
+ 3. Ensure the key has not expired
85
+ 4. Check if the service has usage limits that you've exceeded
86
+
87
+ ### Network Issues
88
+
89
+ If you encounter network issues:
90
+
91
+ 1. Check your internet connection
92
+ 2. Verify that your firewall or network settings allow connections to the service
93
+ 3. Check if the service is experiencing downtime
94
+
95
+ ### Configuration Issues
96
+
97
+ If you encounter configuration issues:
98
+
99
+ 1. Run `python direct_config_check.py` to verify configuration values
100
+ 2. Check that the configuration in `agent/config.py` is correct
101
+ 3. Ensure environment variables are properly loaded
102
+
103
+ ## Setting Up API Keys
104
+
105
+ All API keys should be stored in a `.env` file in the project root:
106
+
107
+ ```
108
+ OPENAI_API_KEY=your_openai_key
109
+ PERPLEXITY_API_KEY=your_perplexity_key
110
+ SERPER_API_KEY=your_serper_key
111
+ SUPABASE_KEY=your_supabase_key
112
+ ```
113
+
114
+ Never commit this file to version control.
docs/deployment.md ADDED
@@ -0,0 +1,303 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # GAIA Agent Deployment Guide
2
+
3
+ This document provides instructions for deploying the GAIA agent in various environments, from local development to production settings.
4
+
5
+ ## Prerequisites
6
+
7
+ Before deploying the GAIA agent, ensure you have:
8
+
9
+ 1. Python 3.8 or higher installed
10
+ 2. All required dependencies installed
11
+ 3. API keys for any external services (OpenAI, Serper, Supabase, etc.)
12
+ 4. Sufficient resources for your expected workload
13
+
14
+ ## Local Deployment
15
+
16
+ ### Installation
17
+
18
+ 1. Clone the repository:
19
+ ```bash
20
+ git clone https://github.com/yourusername/gaia-agent.git
21
+ cd gaia-agent
22
+ ```
23
+
24
+ 2. Create a virtual environment:
25
+ ```bash
26
+ python -m venv venv
27
+ source venv/bin/activate # On Windows: venv\Scripts\activate
28
+ ```
29
+
30
+ 3. Install dependencies:
31
+ ```bash
32
+ pip install -r requirements.txt
33
+ ```
34
+
35
+ 4. Create a `.env` file with your configuration:
36
+ ```
37
+ OPENAI_API_KEY=your-openai-api-key
38
+ SERPER_API_KEY=your-serper-api-key
39
+ SUPABASE_URL=https://your-project-url.supabase.co
40
+ SUPABASE_KEY=your-supabase-api-key
41
+ ```
42
+
43
+ 5. Run the application:
44
+ ```bash
45
+ python app.py
46
+ ```
47
+
48
+ ### Testing
49
+
50
+ Before deploying to production, run the test suite to ensure everything is working correctly:
51
+
52
+ ```bash
53
+ # Run all tests
54
+ python -m unittest discover tests
55
+
56
+ # Run benchmark tests
57
+ python -m tests.benchmark --output benchmark_results.json
58
+ ```
59
+
60
+ ## Deployment Options
61
+
62
+ ### Hugging Face Spaces
63
+
64
+ The GAIA agent is designed to be easily deployed to Hugging Face Spaces:
65
+
66
+ 1. Create a new Space on Hugging Face:
67
+ - Go to https://huggingface.co/spaces
68
+ - Click "Create new Space"
69
+ - Select "Gradio" as the SDK
70
+ - Choose a name for your Space
71
+
72
+ 2. Configure your Space:
73
+ - Add your API keys as secrets in the Space settings
74
+ - Set the Python version to 3.8 or higher
75
+
76
+ 3. Push your code to the Space:
77
+ ```bash
78
+ git remote add space https://huggingface.co/spaces/yourusername/your-space-name
79
+ git push space main
80
+ ```
81
+
82
+ 4. Your agent will be automatically deployed and available at `https://huggingface.co/spaces/yourusername/your-space-name`
83
+
84
+ ### Docker Deployment
85
+
86
+ For containerized deployment, you can use Docker:
87
+
88
+ 1. Create a Dockerfile:
89
+ ```dockerfile
90
+ FROM python:3.9-slim
91
+
92
+ WORKDIR /app
93
+
94
+ COPY requirements.txt .
95
+ RUN pip install --no-cache-dir -r requirements.txt
96
+
97
+ COPY . .
98
+
99
+ CMD ["python", "app.py"]
100
+ ```
101
+
102
+ 2. Build the Docker image:
103
+ ```bash
104
+ docker build -t gaia-agent .
105
+ ```
106
+
107
+ 3. Run the container:
108
+ ```bash
109
+ docker run -p 7860:7860 --env-file .env gaia-agent
110
+ ```
111
+
112
+ ### Cloud Deployment
113
+
114
+ #### AWS Elastic Beanstalk
115
+
116
+ 1. Install the EB CLI:
117
+ ```bash
118
+ pip install awsebcli
119
+ ```
120
+
121
+ 2. Initialize your EB application:
122
+ ```bash
123
+ eb init -p python-3.8 gaia-agent
124
+ ```
125
+
126
+ 3. Create an environment:
127
+ ```bash
128
+ eb create gaia-agent-env
129
+ ```
130
+
131
+ 4. Configure environment variables:
132
+ ```bash
133
+ eb setenv OPENAI_API_KEY=your-openai-api-key SERPER_API_KEY=your-serper-api-key
134
+ ```
135
+
136
+ 5. Deploy your application:
137
+ ```bash
138
+ eb deploy
139
+ ```
140
+
141
+ #### Google Cloud Run
142
+
143
+ 1. Build and push your Docker image to Google Container Registry:
144
+ ```bash
145
+ gcloud builds submit --tag gcr.io/your-project-id/gaia-agent
146
+ ```
147
+
148
+ 2. Deploy to Cloud Run:
149
+ ```bash
150
+ gcloud run deploy gaia-agent \
151
+ --image gcr.io/your-project-id/gaia-agent \
152
+ --platform managed \
153
+ --set-env-vars OPENAI_API_KEY=your-openai-api-key,SERPER_API_KEY=your-serper-api-key
154
+ ```
155
+
156
+ ## Scaling Considerations
157
+
158
+ ### Memory Usage
159
+
160
+ The GAIA agent can be memory-intensive, especially when processing complex questions or using large language models. Consider the following:
161
+
162
+ - Monitor memory usage during benchmark testing
163
+ - Adjust `max_tokens` and other model parameters to control memory usage
164
+ - Consider using a memory cache for frequently asked questions
165
+
166
+ ### API Rate Limits
167
+
168
+ Be aware of rate limits for external APIs:
169
+
170
+ - OpenAI has rate limits based on your subscription tier
171
+ - Serper and other search APIs typically have daily or monthly quotas
172
+ - Implement retry logic with exponential backoff for rate limit errors
173
+
174
+ ### Cost Management
175
+
176
+ Using external APIs can incur costs:
177
+
178
+ - Monitor your API usage and costs
179
+ - Consider caching results to reduce API calls
180
+ - Use smaller or less expensive models for simpler tasks
181
+
182
+ ## Production Best Practices
183
+
184
+ ### Logging
185
+
186
+ Implement comprehensive logging:
187
+
188
+ ```python
189
+ import logging
190
+
191
+ logging.basicConfig(
192
+ level=logging.INFO,
193
+ format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
194
+ filename='gaia_agent.log'
195
+ )
196
+ ```
197
+
198
+ ### Monitoring
199
+
200
+ Set up monitoring for:
201
+
202
+ - API response times
203
+ - Error rates
204
+ - Memory and CPU usage
205
+ - Cost metrics
206
+
207
+ ### Security
208
+
209
+ Protect sensitive information:
210
+
211
+ - Never hardcode API keys in your code
212
+ - Use environment variables or a secure secrets manager
213
+ - Implement proper authentication for your API endpoints
214
+ - Validate and sanitize all user inputs
215
+
216
+ ### Backup and Recovery
217
+
218
+ Implement backup and recovery procedures:
219
+
220
+ - Regularly backup your Supabase database
221
+ - Create snapshots of your deployment environment
222
+ - Document recovery procedures
223
+
224
+ ## Continuous Integration/Continuous Deployment (CI/CD)
225
+
226
+ Set up CI/CD pipelines to automate testing and deployment:
227
+
228
+ ### GitHub Actions Example
229
+
230
+ Create a `.github/workflows/main.yml` file:
231
+
232
+ ```yaml
233
+ name: CI/CD Pipeline
234
+
235
+ on:
236
+ push:
237
+ branches: [ main ]
238
+ pull_request:
239
+ branches: [ main ]
240
+
241
+ jobs:
242
+ test:
243
+ runs-on: ubuntu-latest
244
+ steps:
245
+ - uses: actions/checkout@v2
246
+ - name: Set up Python
247
+ uses: actions/setup-python@v2
248
+ with:
249
+ python-version: '3.9'
250
+ - name: Install dependencies
251
+ run: |
252
+ python -m pip install --upgrade pip
253
+ pip install -r requirements.txt
254
+ - name: Run tests
255
+ run: |
256
+ python -m unittest discover tests
257
+
258
+ deploy:
259
+ needs: test
260
+ if: github.event_name == 'push' && github.ref == 'refs/heads/main'
261
+ runs-on: ubuntu-latest
262
+ steps:
263
+ - uses: actions/checkout@v2
264
+ - name: Deploy to Hugging Face Spaces
265
+ uses: huggingface/huggingface-spaces-deploy-action@main
266
+ with:
267
+ space_id: ${{ secrets.SPACE_ID }}
268
+ token: ${{ secrets.HF_TOKEN }}
269
+ ```
270
+
271
+ ## Troubleshooting
272
+
273
+ ### Common Issues
274
+
275
+ 1. **API Key Errors**:
276
+ - Ensure all API keys are correctly set in your environment variables
277
+ - Check for typos or trailing spaces in your API keys
278
+
279
+ 2. **Memory Issues**:
280
+ - Reduce the `max_tokens` parameter in your model configuration
281
+ - Implement memory cleanup after processing each question
282
+
283
+ 3. **Timeout Errors**:
284
+ - Increase the timeout values in your configuration
285
+ - Optimize your agent's workflow to reduce processing time
286
+
287
+ 4. **Rate Limit Errors**:
288
+ - Implement retry logic with exponential backoff
289
+ - Consider upgrading your API subscription tier
290
+
291
+ ### Getting Help
292
+
293
+ If you encounter issues not covered in this guide:
294
+
295
+ 1. Check the project's GitHub issues for similar problems
296
+ 2. Consult the documentation for the specific components (LangChain, LangGraph, etc.)
297
+ 3. Reach out to the community for assistance
298
+
299
+ ## Next Steps
300
+
301
+ For usage examples, see [Usage Examples](examples.md).
302
+
303
+ For configuration options, see [Configuration Guide](configuration.md).
docs/duckduckgo_integration.md ADDED
@@ -0,0 +1,156 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # DuckDuckGo Integration for GAIA Agent
2
+
3
+ This document describes how to set up and use the DuckDuckGo search integration with the GAIA agent.
4
+
5
+ ## Overview
6
+
7
+ The GAIA agent now supports DuckDuckGo as an alternative search engine for web searches. This integration allows the agent to perform web searches without requiring an API key, making it a good fallback option when other search providers are unavailable or rate-limited.
8
+
9
+ ## Features
10
+
11
+ - **No API Key Required**: DuckDuckGo search works without requiring an API key
12
+ - **Configurable Result Count**: Control how many search results are returned
13
+ - **Error Handling**: Robust error handling for various search scenarios
14
+ - **Standardized Results**: Results are formatted consistently with other search providers
15
+
16
+ ## Installation
17
+
18
+ The DuckDuckGo integration requires the `duckduckgo-search` package:
19
+
20
+ ```bash
21
+ pip install duckduckgo-search
22
+ ```
23
+
24
+ ## Configuration
25
+
26
+ The DuckDuckGo search tool can be configured in the GAIA agent's configuration:
27
+
28
+ ```python
29
+ # In agent/config.py
30
+ class GaiaConfig:
31
+ def __init__(self):
32
+ # ...
33
+ self.duckduckgo_enabled = True
34
+ self.duckduckgo_result_count = 5
35
+ # ...
36
+ ```
37
+
38
+ ### Configuration Options
39
+
40
+ | Option | Description | Default |
41
+ |--------|-------------|---------|
42
+ | `duckduckgo_enabled` | Enable/disable DuckDuckGo search | `True` |
43
+ | `duckduckgo_result_count` | Number of results to return | `5` |
44
+ | `duckduckgo_timeout` | Timeout in seconds for search requests | `30` |
45
+
46
+ ## Usage
47
+
48
+ ### Basic Usage
49
+
50
+ ```python
51
+ from tools.web_tools import create_duckduckgo_search
52
+
53
+ # Create a DuckDuckGo search tool
54
+ search_tool = create_duckduckgo_search()
55
+
56
+ # Perform a search
57
+ results = search_tool.search("python programming")
58
+
59
+ # Process the results
60
+ for result in results:
61
+ print(f"Title: {result['title']}")
62
+ print(f"URL: {result['link']}")
63
+ print(f"Snippet: {result['snippet']}")
64
+ print("---")
65
+ ```
66
+
67
+ ### Integration with GAIA Agent
68
+
69
+ The DuckDuckGo search tool is automatically integrated with the GAIA agent if enabled in the configuration:
70
+
71
+ ```python
72
+ from agent.agent import GaiaAgent
73
+ from agent.config import GaiaConfig
74
+
75
+ # Create a configuration with DuckDuckGo enabled
76
+ config = GaiaConfig()
77
+ config.duckduckgo_enabled = True
78
+
79
+ # Create a GAIA agent with the configuration
80
+ agent = GaiaAgent(config=config)
81
+
82
+ # The agent will now use DuckDuckGo for web searches
83
+ response = agent.process("What is the latest version of Python?")
84
+ ```
85
+
86
+ ## Result Format
87
+
88
+ The DuckDuckGo search tool returns results in the following format:
89
+
90
+ ```json
91
+ [
92
+ {
93
+ "title": "Result title",
94
+ "link": "https://example.com/result",
95
+ "snippet": "A brief excerpt from the search result..."
96
+ },
97
+ // More results...
98
+ ]
99
+ ```
100
+
101
+ ## Fallback Mechanism
102
+
103
+ The GAIA agent can be configured to use DuckDuckGo as a fallback when other search providers fail:
104
+
105
+ ```python
106
+ # In agent/config.py
107
+ class GaiaConfig:
108
+ def __init__(self):
109
+ # ...
110
+ self.search_fallback_order = ["serper", "duckduckgo", "perplexity"]
111
+ # ...
112
+ ```
113
+
114
+ With this configuration, the agent will try Serper first, then DuckDuckGo if Serper fails, and finally Perplexity if both fail.
115
+
116
+ ## Testing
117
+
118
+ You can test the DuckDuckGo integration using the provided test scripts:
119
+
120
+ ```bash
121
+ # Test just the DuckDuckGo search
122
+ python tests/test_duckduckgo_search.py
123
+
124
+ # Test all search providers together
125
+ python tests/test_gaia_multi_search_providers.py
126
+ ```
127
+
128
+ ## Troubleshooting
129
+
130
+ ### Common Issues
131
+
132
+ 1. **No Results Returned**: DuckDuckGo may return no results for certain queries. Try reformulating the query.
133
+
134
+ 2. **Connection Errors**: If you're experiencing connection errors, check your internet connection and ensure that DuckDuckGo is accessible from your network.
135
+
136
+ 3. **Rate Limiting**: While DuckDuckGo doesn't require an API key, it may still rate-limit excessive requests. Consider adding delays between searches if you're making many requests.
137
+
138
+ ### Debugging
139
+
140
+ To enable debug logging for the DuckDuckGo search tool:
141
+
142
+ ```python
143
+ import logging
144
+ logging.basicConfig(level=logging.DEBUG)
145
+ logger = logging.getLogger("tools.web_tools")
146
+ ```
147
+
148
+ ## Limitations
149
+
150
+ - DuckDuckGo search may not provide as many results as commercial search APIs
151
+ - Some advanced search features (like filtering by date) may not be available
152
+ - Results may be less comprehensive than those from paid search providers
153
+
154
+ ## Conclusion
155
+
156
+ The DuckDuckGo integration provides a reliable, no-cost alternative for web searches in the GAIA agent. It's particularly useful as a fallback option or for development and testing purposes.
docs/examples.md ADDED
@@ -0,0 +1,197 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # GAIA Agent Usage Examples
2
+
3
+ This document provides examples of how to use the GAIA agent in different scenarios. These examples demonstrate the agent's capabilities and how to integrate it into your applications.
4
+
5
+ ## Basic Usage
6
+
7
+ ### Initializing the Agent
8
+
9
+ ```python
10
+ from agent import GAIAAgent
11
+
12
+ # Initialize with default configuration
13
+ agent = GAIAAgent()
14
+
15
+ # Or with custom configuration
16
+ custom_config = {
17
+ "model": {
18
+ "reasoning_model": "gpt-4",
19
+ "temperature": 0.2,
20
+ "max_tokens": 1000
21
+ },
22
+ "tools": {
23
+ "web_search": {
24
+ "enabled": True,
25
+ "providers": ["duckduckgo", "serper"]
26
+ }
27
+ }
28
+ }
29
+ agent = GAIAAgent(custom_config)
30
+ ```
31
+
32
+ ### Asking Questions
33
+
34
+ ```python
35
+ # Ask a simple factual question
36
+ answer = agent("What is the capital of France?")
37
+ print(f"Answer: {answer}")
38
+
39
+ # Ask a more complex question requiring reasoning
40
+ answer = agent("If a train travels at 120 km/h for 2.5 hours, how far does it travel?")
41
+ print(f"Answer: {answer}")
42
+
43
+ # Ask a question requiring web search
44
+ answer = agent("What were the key announcements at the most recent Apple WWDC?")
45
+ print(f"Answer: {answer}")
46
+ ```
47
+
48
+ ## Advanced Usage
49
+
50
+ ### Using Memory Components
51
+
52
+ ```python
53
+ # Get a snapshot of the agent's memory
54
+ memory_snapshot = agent.get_memory_snapshot()
55
+ print(f"Conversation history: {memory_snapshot['conversation']}")
56
+ print(f"Cached results: {memory_snapshot['cached_results']}")
57
+
58
+ # Clear the agent's memory
59
+ agent.clear_memory()
60
+ ```
61
+
62
+ ### Error Handling
63
+
64
+ ```python
65
+ try:
66
+ answer = agent("What is the meaning of life?")
67
+ print(f"Answer: {answer}")
68
+ except Exception as e:
69
+ print(f"Error: {e}")
70
+
71
+ # You can also configure the agent to return error messages instead of raising exceptions
72
+ agent_with_error_messages = GAIAAgent({
73
+ "agent": {
74
+ "return_error_message": True
75
+ }
76
+ })
77
+
78
+ answer = agent_with_error_messages("What is the meaning of life?")
79
+ if answer.startswith("Error:"):
80
+ print(f"Agent encountered an error: {answer}")
81
+ else:
82
+ print(f"Answer: {answer}")
83
+ ```
84
+
85
+ ## Integration Examples
86
+
87
+ ### Integration with Gradio
88
+
89
+ ```python
90
+ import gradio as gr
91
+ from agent import GAIAAgent
92
+
93
+ # Initialize the agent
94
+ agent = GAIAAgent()
95
+
96
+ # Define the Gradio interface
97
+ def process_question(question):
98
+ try:
99
+ answer = agent(question)
100
+ return answer
101
+ except Exception as e:
102
+ return f"Error: {e}"
103
+
104
+ # Create the interface
105
+ demo = gr.Interface(
106
+ fn=process_question,
107
+ inputs=gr.Textbox(lines=2, placeholder="Enter your question here..."),
108
+ outputs=gr.Textbox(label="Answer"),
109
+ title="GAIA Agent Demo",
110
+ description="Ask a question and get an answer from the GAIA agent."
111
+ )
112
+
113
+ # Launch the interface
114
+ demo.launch()
115
+ ```
116
+
117
+ ### Integration with FastAPI
118
+
119
+ ```python
120
+ from fastapi import FastAPI, HTTPException
121
+ from pydantic import BaseModel
122
+ from agent import GAIAAgent
123
+
124
+ # Initialize the agent
125
+ agent = GAIAAgent()
126
+
127
+ # Create the FastAPI app
128
+ app = FastAPI(title="GAIA Agent API")
129
+
130
+ # Define the request model
131
+ class QuestionRequest(BaseModel):
132
+ question: str
133
+
134
+ # Define the response model
135
+ class AnswerResponse(BaseModel):
136
+ answer: str
137
+ processing_time: float
138
+
139
+ # Define the endpoint
140
+ @app.post("/ask", response_model=AnswerResponse)
141
+ async def ask_question(request: QuestionRequest):
142
+ try:
143
+ import time
144
+ start_time = time.time()
145
+ answer = agent(request.question)
146
+ processing_time = time.time() - start_time
147
+ return AnswerResponse(answer=answer, processing_time=processing_time)
148
+ except Exception as e:
149
+ raise HTTPException(status_code=500, detail=str(e))
150
+
151
+ # Run with: uvicorn api:app --reload
152
+ ```
153
+
154
+ ## Benchmark Testing
155
+
156
+ ```python
157
+ from tests.benchmark import run_benchmark, SAMPLE_QUESTIONS
158
+
159
+ # Run the benchmark with default settings
160
+ result = run_benchmark(SAMPLE_QUESTIONS)
161
+
162
+ # Print the summary
163
+ result.print_summary()
164
+
165
+ # Save the results to a file
166
+ result.save_to_file("benchmark_results.json")
167
+
168
+ # Run with custom configuration
169
+ custom_config = {
170
+ "model": {
171
+ "reasoning_model": "gpt-4",
172
+ "temperature": 0.1
173
+ }
174
+ }
175
+ result = run_benchmark(SAMPLE_QUESTIONS, agent_config=custom_config)
176
+ ```
177
+
178
+ ## Command-Line Usage
179
+
180
+ You can also run the agent from the command line using the provided scripts:
181
+
182
+ ```bash
183
+ # Run the Gradio interface
184
+ python app.py
185
+
186
+ # Run benchmark tests
187
+ python -m tests.benchmark --output benchmark_results.json
188
+
189
+ # Run benchmark tests with custom questions
190
+ python -m tests.benchmark --questions custom_questions.json --output benchmark_results.json
191
+ ```
192
+
193
+ ## Next Steps
194
+
195
+ For more detailed information on configuration options, see [Configuration Guide](configuration.md).
196
+
197
+ For deployment instructions, see [Deployment Guide](deployment.md).
docs/gaia_memory_integration.md ADDED
@@ -0,0 +1,155 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # GAIA Memory Integration with Supabase
2
+
3
+ This document describes the integration of Supabase as a persistent memory store for the GAIA agent. The integration enables the agent to maintain conversation history, cache results, and store intermediate data across multiple sessions.
4
+
5
+ ## Overview
6
+
7
+ The GAIA agent uses Supabase as a backend database to store various types of memory:
8
+
9
+ 1. **Conversation Memory**: Stores the history of interactions between users and the agent
10
+ 2. **Result Cache**: Caches tool results and answers to avoid redundant computations
11
+ 3. **Working Memory**: Stores intermediate results during agent execution
12
+
13
+ This memory architecture allows the agent to:
14
+ - Maintain context across multiple turns in a conversation
15
+ - Remember previous interactions even after the agent is restarted
16
+ - Build upon previous knowledge in follow-up questions
17
+ - Optimize performance by caching expensive operations
18
+
19
+ ## Architecture
20
+
21
+ The memory system consists of several components:
22
+
23
+ ### SupabaseMemory
24
+
25
+ The core memory implementation that handles the connection to Supabase and provides basic CRUD operations:
26
+
27
+ - `store(key, value, metadata, ttl)`: Store a value in memory
28
+ - `retrieve(key, default)`: Retrieve a value from memory
29
+ - `delete(key)`: Delete a value from memory
30
+ - `clear()`: Clear all values from memory
31
+ - `list_keys(prefix)`: List all keys in memory
32
+
33
+ ### ConversationMemory
34
+
35
+ A specialized memory interface for storing conversation history:
36
+
37
+ - `add_message(role, content, metadata)`: Add a message to the conversation history
38
+ - `get_messages(limit, before, after)`: Get messages from the conversation history
39
+ - `clear()`: Clear the conversation history
40
+
41
+ ### ResultCache
42
+
43
+ A specialized memory interface for caching tool results and answers:
44
+
45
+ - `cache_result(tool_name, input_hash, result)`: Cache a tool result
46
+ - `get_result(tool_name, input_hash)`: Get a cached result
47
+ - `invalidate(key)`: Invalidate a cached result
48
+ - `clear()`: Clear all cached results
49
+
50
+ ### WorkingMemory
51
+
52
+ A specialized memory interface for storing intermediate results during agent execution:
53
+
54
+ - `store_intermediate_result(key, value, metadata)`: Store an intermediate result
55
+ - `get_intermediate_result(key, default)`: Get an intermediate result
56
+ - `list_intermediate_results()`: List all intermediate results
57
+ - `clear()`: Clear all intermediate results
58
+
59
+ ## Database Schema
60
+
61
+ The Supabase database uses the following tables:
62
+
63
+ ### gaia_memory
64
+
65
+ The main memory table with the following columns:
66
+
67
+ - `id`: UUID primary key
68
+ - `key`: Text, unique key for the memory entry
69
+ - `value`: Text, JSON-serialized value
70
+ - `metadata`: JSONB, additional metadata
71
+ - `created_at`: Timestamp with timezone, creation time
72
+ - `expires_at`: Timestamp with timezone, expiration time (optional)
73
+
74
+ ## Configuration
75
+
76
+ The memory system is configured through the `agent/config.py` file:
77
+
78
+ ```python
79
+ # Memory Settings
80
+ MEMORY_ENABLED = True
81
+ MEMORY_TABLE_NAME = "gaia_memory"
82
+ MEMORY_TTL = 86400 # 24 hours in seconds
83
+ MEMORY_CACHE_SIZE = 100
84
+ MEMORY_SIMILARITY_THRESHOLD = 0.85
85
+ MEMORY_MAX_ENTRIES = 1000
86
+ ```
87
+
88
+ ## Usage
89
+
90
+ ### Initializing the Agent with Memory
91
+
92
+ ```python
93
+ from agent.agent import GAIAAgent
94
+
95
+ # Create an agent with memory enabled
96
+ agent = GAIAAgent({
97
+ "memory": {
98
+ "enabled": True,
99
+ "table_name": "gaia_memory",
100
+ "conversation_id": "unique-conversation-id"
101
+ }
102
+ })
103
+ ```
104
+
105
+ ### Using Memory Across Sessions
106
+
107
+ The memory system allows the agent to maintain context across multiple sessions:
108
+
109
+ ```python
110
+ # Session 1
111
+ agent1 = GAIAAgent({"memory": {"conversation_id": "conversation-123"}})
112
+ answer1 = agent1("What is the capital of France?")
113
+
114
+ # Session 2 (could be a different process or even a different day)
115
+ agent2 = GAIAAgent({"memory": {"conversation_id": "conversation-123"}})
116
+ answer2 = agent2("What is the population of that city?") # Agent remembers "that city" refers to Paris
117
+ ```
118
+
119
+ ## Testing
120
+
121
+ The integration can be tested using the `tests/test_gaia_with_memory.py` script:
122
+
123
+ ```bash
124
+ # Test memory persistence
125
+ python tests/test_gaia_with_memory.py --test memory
126
+
127
+ # Test API integration with memory
128
+ python tests/test_gaia_with_memory.py --test api
129
+
130
+ # Run all tests
131
+ python tests/test_gaia_with_memory.py --test all
132
+ ```
133
+
134
+ ## Error Handling
135
+
136
+ The memory system is designed to be resilient to failures:
137
+
138
+ 1. If Supabase is unavailable, the agent falls back to in-memory storage
139
+ 2. If a memory entry expires, it is automatically removed from the system
140
+ 3. If a memory operation fails, the error is logged and the agent continues operation
141
+
142
+ ## Security Considerations
143
+
144
+ 1. Supabase credentials are stored as environment variables
145
+ 2. The memory system uses Supabase's built-in security features
146
+ 3. Sensitive data should be encrypted before storage
147
+ 4. Memory entries can have a time-to-live (TTL) to automatically expire
148
+
149
+ ## Future Improvements
150
+
151
+ 1. Implement vector embeddings for semantic search
152
+ 2. Add memory compression to reduce storage requirements
153
+ 3. Implement memory prioritization to focus on relevant information
154
+ 4. Add support for multi-user memory isolation
155
+ 5. Implement memory summarization for long conversations
docs/index.md ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # GAIA Agent Documentation
2
+
3
+ Welcome to the GAIA Agent documentation. This comprehensive guide provides detailed information about the GAIA agent, its components, configuration options, and usage examples.
4
+
5
+ ## Table of Contents
6
+
7
+ ### Getting Started
8
+ - [Project Overview](../README.md): Introduction to the GAIA agent project
9
+ - [Configuration Guide](configuration.md): Detailed information about configuration options
10
+ - [Deployment Guide](deployment.md): Instructions for deploying the agent in various environments
11
+
12
+ ### Core Components
13
+ - [Agent Architecture](agent_architecture.md): Overview of the agent's architecture and components
14
+ - [Memory System](gaia_memory_integration.md): Information about the agent's memory system
15
+ - [Tool Integration](tool_integration.md): Guide to the agent's tool integration system
16
+
17
+ ### API Integrations
18
+ - [OpenAI Integration](openai_integration_testing.md): Guide for testing OpenAI integration with newer models
19
+ - [LangChain OpenAI Integration](langchain_openai_integration.md): Guide for the langchain_openai package integration
20
+ - [Serper Integration](serper_integration.md): Guide for using the Serper API for Google search
21
+ - [DuckDuckGo Integration](duckduckgo_integration.md): Guide for using DuckDuckGo search
22
+ - [Multi-Search Integration](multi_search_integration.md): Guide for using multiple search providers
23
+ - [Perplexity Integration](perplexity_integration.md): Guide for using Perplexity AI
24
+
25
+ ### Memory Integrations
26
+ - [Supabase Memory Integration](supabase_memory_integration.md): Guide for setting up and using Supabase memory
27
+
28
+ ### Testing and Verification
29
+ - [Connection Verification](connection_verification.md): Guide for verifying connections to external services
30
+ - [All APIs Integration Testing](all_apis_integration_testing.md): Guide for testing all API integrations together
31
+
32
+ ### Examples and Tutorials
33
+ - [Usage Examples](examples.md): Examples of how to use the GAIA agent in different scenarios
34
+
35
+ ### Troubleshooting
36
+ - [Troubleshooting Guide](troubleshooting.md): Solutions for common issues and problems
37
+
38
+ ## Quick Links
39
+
40
+ - [GitHub Repository](https://github.com/your-repo): Source code repository
41
+ - [Issue Tracker](https://github.com/your-repo/issues): Report issues and request features
42
+ - [Hugging Face Agents Course](https://huggingface.co/learn/agents): Learn more about the Hugging Face Agents Course
43
+
44
+ ## Recent Updates
45
+
46
+ - **April 30, 2025**: Added troubleshooting guide and langchain_openai integration documentation
47
+ - **April 25, 2025**: Updated model configurations to use GPT-4o
48
+ - **April 20, 2025**: Added multi-search integration documentation
49
+ - **April 15, 2025**: Added Perplexity integration documentation
50
+ - **April 10, 2025**: Initial documentation release
51
+
52
+ ## Contributing
53
+
54
+ If you'd like to contribute to the documentation, please follow these steps:
55
+
56
+ 1. Fork the repository
57
+ 2. Create a new branch for your changes
58
+ 3. Make your changes
59
+ 4. Submit a pull request
60
+
61
+ ## License
62
+
63
+ This documentation is part of the Hugging Face Agents Course and follows its licensing terms.
docs/langchain_openai_integration.md ADDED
@@ -0,0 +1,227 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # LangChain OpenAI Integration Guide
2
+
3
+ This document provides detailed information about the integration between LangChain and OpenAI in the GAIA agent, focusing on the `langchain_openai` package.
4
+
5
+ ## Overview
6
+
7
+ The GAIA agent uses the `langchain_openai` package to interact with OpenAI's language models. This package provides a clean interface for using OpenAI models within the LangChain framework, enabling features like:
8
+
9
+ - Chat completions with various OpenAI models
10
+ - Embeddings generation
11
+ - Function calling
12
+ - Streaming responses
13
+ - Vision capabilities
14
+
15
+ ## Recent Changes
16
+
17
+ As of recent updates to the LangChain ecosystem, the OpenAI integration has been moved from the core `langchain` package to a separate `langchain_openai` package. This change was made to:
18
+
19
+ 1. Modularize the LangChain ecosystem
20
+ 2. Allow for more frequent updates to specific integrations
21
+ 3. Reduce the size of the core package
22
+ 4. Enable more flexible dependency management
23
+
24
+ ## Installation
25
+
26
+ To use the `langchain_openai` package, you need to install it separately:
27
+
28
+ ```bash
29
+ pip install langchain_openai
30
+ ```
31
+
32
+ For the GAIA agent, this dependency is included in the `requirements.txt` file, so it should be installed automatically when you run:
33
+
34
+ ```bash
35
+ pip install -r requirements.txt
36
+ ```
37
+
38
+ If you encounter issues, try upgrading to the latest version:
39
+
40
+ ```bash
41
+ pip install --upgrade langchain_openai
42
+ ```
43
+
44
+ ## Usage in GAIA Agent
45
+
46
+ The GAIA agent uses the `langchain_openai` package in several components:
47
+
48
+ ### 1. Agent Graph (agent/graph.py)
49
+
50
+ ```python
51
+ from langchain_openai import ChatOpenAI
52
+
53
+ # Initialize the model
54
+ model = ChatOpenAI(
55
+ model=get_model_config().get("reasoning_model", "gpt-4o"),
56
+ temperature=get_model_config().get("temperature", 0.1),
57
+ max_tokens=get_model_config().get("max_tokens", 4096)
58
+ )
59
+ ```
60
+
61
+ ### 2. Reasoning Tools (tools/reasoning_tools.py)
62
+
63
+ ```python
64
+ from langchain_openai import ChatOpenAI
65
+
66
+ # Initialize the model for reasoning
67
+ reasoning_model = ChatOpenAI(
68
+ model=get_model_config().get("reasoning_model", "gpt-4o"),
69
+ temperature=get_model_config().get("temperature", 0.1)
70
+ )
71
+ ```
72
+
73
+ ### 3. Multimodal Tools (tools/multimodal_tools.py)
74
+
75
+ ```python
76
+ from langchain_openai import ChatOpenAI
77
+
78
+ # Initialize the model for vision tasks
79
+ vision_model = ChatOpenAI(
80
+ model=get_model_config().get("vision_model", "gpt-4o"),
81
+ temperature=get_model_config().get("temperature", 0.1)
82
+ )
83
+ ```
84
+
85
+ ## Common Issues and Solutions
86
+
87
+ ### 1. Import Errors
88
+
89
+ If you encounter import errors like:
90
+
91
+ ```
92
+ ImportError: No module named 'langchain_openai'
93
+ ```
94
+
95
+ or
96
+
97
+ ```
98
+ ImportError: cannot import name 'ChatOpenAI' from 'langchain_openai'
99
+ ```
100
+
101
+ **Solutions:**
102
+
103
+ 1. Ensure the package is installed:
104
+ ```bash
105
+ pip install langchain_openai
106
+ ```
107
+
108
+ 2. Check for conflicts with older versions:
109
+ ```bash
110
+ pip uninstall langchain-openai # Note the hyphen
111
+ pip install langchain_openai # Note the underscore
112
+ ```
113
+
114
+ 3. Update your import statements:
115
+ ```python
116
+ # Old import (no longer works)
117
+ # from langchain.chat_models import ChatOpenAI
118
+
119
+ # New import (correct)
120
+ from langchain_openai import ChatOpenAI
121
+ ```
122
+
123
+ ### 2. Version Compatibility
124
+
125
+ If you encounter compatibility issues between `langchain`, `langchain_openai`, and other LangChain packages:
126
+
127
+ **Solution:**
128
+ ```bash
129
+ pip install --upgrade langchain langchain_openai langgraph
130
+ ```
131
+
132
+ This ensures all LangChain-related packages are updated to compatible versions.
133
+
134
+ ### 3. API Key Issues
135
+
136
+ If you encounter authentication errors:
137
+
138
+ **Solution:**
139
+ Ensure your OpenAI API key is correctly set:
140
+ ```python
141
+ import os
142
+ from dotenv import load_dotenv
143
+
144
+ load_dotenv()
145
+
146
+ # Check if the API key is set
147
+ api_key = os.getenv("OPENAI_API_KEY")
148
+ if not api_key:
149
+ raise ValueError("OPENAI_API_KEY environment variable is not set")
150
+ ```
151
+
152
+ ## Model Configuration
153
+
154
+ The GAIA agent uses the following OpenAI models by default:
155
+
156
+ | Purpose | Model | Environment Variable |
157
+ |---------|-------|----------------------|
158
+ | Reasoning | gpt-4o | REASONING_MODEL |
159
+ | Tool Execution | gpt-3.5-turbo | TOOL_MODEL |
160
+ | Vision | gpt-4o | VISION_MODEL |
161
+ | Embeddings | text-embedding-3-large | EMBEDDING_MODEL |
162
+
163
+ You can customize these models by setting the appropriate environment variables or by passing a configuration dictionary when initializing the agent.
164
+
165
+ ## Best Practices
166
+
167
+ 1. **Always specify the model version**: This ensures consistent behavior even when OpenAI releases new model versions.
168
+ ```python
169
+ ChatOpenAI(model="gpt-4o") # Specific model version
170
+ ```
171
+
172
+ 2. **Set appropriate temperature**: Lower values (0.0-0.3) for factual tasks, higher values (0.5-0.8) for creative tasks.
173
+ ```python
174
+ ChatOpenAI(model="gpt-4o", temperature=0.1) # Low temperature for factual responses
175
+ ```
176
+
177
+ 3. **Handle rate limits gracefully**: Implement exponential backoff and retry logic.
178
+ ```python
179
+ from tenacity import retry, stop_after_attempt, wait_exponential
180
+
181
+ @retry(stop=stop_after_attempt(3), wait=wait_exponential(multiplier=1, min=2, max=10))
182
+ def query_model(prompt):
183
+ model = ChatOpenAI(model="gpt-4o")
184
+ return model.invoke(prompt)
185
+ ```
186
+
187
+ 4. **Use streaming for long responses**: This provides a better user experience for long-running queries.
188
+ ```python
189
+ model = ChatOpenAI(model="gpt-4o", streaming=True)
190
+ for chunk in model.stream("Tell me about AI"):
191
+ print(chunk.content, end="", flush=True)
192
+ ```
193
+
194
+ 5. **Set max_tokens appropriately**: This prevents unnecessarily long responses and reduces token usage.
195
+ ```python
196
+ ChatOpenAI(model="gpt-4o", max_tokens=1000)
197
+ ```
198
+
199
+ ## Testing
200
+
201
+ The GAIA agent includes tests for OpenAI integration in `tests/test_openai_integration.py`. These tests verify:
202
+
203
+ 1. Connection to the OpenAI API
204
+ 2. Compatibility with newer models like GPT-4o
205
+ 3. Integration with the GAIA agent
206
+
207
+ To run these tests:
208
+
209
+ ```bash
210
+ python tests/test_openai_integration.py
211
+ ```
212
+
213
+ For more comprehensive testing:
214
+
215
+ ```bash
216
+ python tests/run_openai_tests.py
217
+ ```
218
+
219
+ ## Troubleshooting
220
+
221
+ If you encounter issues with the `langchain_openai` integration, check the [Troubleshooting Guide](troubleshooting.md) for common problems and solutions.
222
+
223
+ ## Related Documentation
224
+
225
+ - [OpenAI API Documentation](https://platform.openai.com/docs/api-reference)
226
+ - [LangChain Documentation](https://python.langchain.com/docs/get_started/introduction)
227
+ - [LangChain OpenAI Integration Documentation](https://python.langchain.com/docs/integrations/llms/openai)
docs/multi_search_integration.md ADDED
@@ -0,0 +1,318 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Multi-Search Provider Integration for GAIA Agent
2
+
3
+ This document explains how to integrate and use multiple search providers with the GAIA agent, including:
4
+
5
+ 1. DuckDuckGo
6
+ 2. Serper (Google Search API)
7
+ 3. Perplexity
8
+ 4. arXiv
9
+
10
+ ## Overview
11
+
12
+ The GAIA agent can leverage multiple search providers to get comprehensive and diverse search results. Each provider has its strengths:
13
+
14
+ - **DuckDuckGo**: Privacy-focused search with no API key required
15
+ - **Serper**: Google search results with structured data
16
+ - **Perplexity**: AI-powered search with reasoning capabilities
17
+ - **arXiv**: Academic paper search for scientific research
18
+
19
+ ## Configuration
20
+
21
+ ### Environment Variables
22
+
23
+ Set up the necessary API keys in your environment:
24
+
25
+ ```bash
26
+ # Required for Serper
27
+ export SERPER_API_KEY="your_serper_api_key"
28
+
29
+ # Required for Perplexity
30
+ export PERPLEXITY_API_KEY="your_perplexity_api_key"
31
+
32
+ # DuckDuckGo and arXiv don't require API keys
33
+ ```
34
+
35
+ ### Tool Configuration
36
+
37
+ The search providers are configured in `agent/config.py`:
38
+
39
+ ```python
40
+ # Tool Parameters (hardcoded)
41
+ WEB_SEARCH_RESULT_COUNT = 5
42
+ WEB_SEARCH_TIMEOUT = 10
43
+ DUCKDUCKGO_TIMEOUT = 10
44
+ DUCKDUCKGO_MAX_RESULTS = 5
45
+ PERPLEXITY_TIMEOUT = 30
46
+ PERPLEXITY_MODEL = "sonar-reasoning"
47
+ ARXIV_MAX_RESULTS = 3
48
+ ```
49
+
50
+ ## Using Search Providers
51
+
52
+ ### Creating Search Tool Instances
53
+
54
+ ```python
55
+ # Import the necessary tools
56
+ from tools.web_tools import create_duckduckgo_search, create_serper_search
57
+ from tools.perplexity_tool import PerplexityTool
58
+ from tools.arxiv_tool import create_arxiv_search
59
+
60
+ # Create DuckDuckGo search tool (no API key required)
61
+ duckduckgo_tool = create_duckduckgo_search()
62
+
63
+ # Create Serper search tool (requires API key in environment)
64
+ serper_tool = create_serper_search()
65
+
66
+ # Create Perplexity tool (requires API key in environment)
67
+ perplexity_tool = PerplexityTool()
68
+
69
+ # Create arXiv search tool (no API key required)
70
+ arxiv_tool = create_arxiv_search()
71
+ ```
72
+
73
+ ### Performing Searches
74
+
75
+ Each search provider has a similar interface:
76
+
77
+ ```python
78
+ # DuckDuckGo search
79
+ duckduckgo_results = duckduckgo_tool.search("your search query")
80
+ # Returns a list of dictionaries with 'title', 'link', and 'snippet' keys
81
+
82
+ # Serper search
83
+ serper_results = serper_tool.search("your search query")
84
+ # Returns a list of dictionaries with 'title', 'link', and 'snippet' keys
85
+
86
+ # Perplexity search
87
+ perplexity_results = perplexity_tool.search("your search query")
88
+ # Returns a dictionary with 'content', 'citations', and other metadata
89
+
90
+ # arXiv search
91
+ arxiv_results = arxiv_tool.search("your search query")
92
+ # Returns a list of dictionaries with paper information
93
+
94
+ # arXiv get paper by ID
95
+ paper = arxiv_tool.get_paper_by_id("2303.08774")
96
+ # Returns a dictionary with detailed paper information
97
+
98
+ # arXiv search by category
99
+ category_results = arxiv_tool.search_by_category("cs.AI")
100
+ # Returns a list of dictionaries with papers in the specified category
101
+ ```
102
+
103
+ ## Integrating with LangGraph
104
+
105
+ To integrate these search providers with the LangGraph-based GAIA agent, you need to modify the `execute_tool` function in `agent/graph.py`:
106
+
107
+ ```python
108
+ # Create a tool registry
109
+ tool_registry = {
110
+ "duckduckgo_search": create_duckduckgo_search(),
111
+ "serper_search": create_serper_search(),
112
+ "perplexity_search": PerplexityTool(),
113
+ "arxiv_search": create_arxiv_search(),
114
+ "arxiv_get_paper": create_arxiv_search(),
115
+ "arxiv_search_category": create_arxiv_search()
116
+ }
117
+
118
+ def execute_tool(state: AgentState) -> AgentState:
119
+ # Get the current step from the plan
120
+ current_step = state["plan"][state["current_step"]]
121
+ tool_name = current_step.get("tool")
122
+ tool_input = current_step.get("tool_input", {})
123
+
124
+ logger.debug(f"Executing tool {tool_name} for step {current_step['step_number']}")
125
+
126
+ try:
127
+ # Get the tool from the registry
128
+ tool = tool_registry.get(tool_name)
129
+ if not tool:
130
+ raise Exception(f"Tool {tool_name} not found in registry")
131
+
132
+ # Execute the tool based on its type
133
+ if tool_name in ["duckduckgo_search", "serper_search"]:
134
+ result = tool.search(tool_input.get("query", ""))
135
+ elif tool_name == "perplexity_search":
136
+ result = tool.search(tool_input.get("query", ""))
137
+ elif tool_name == "arxiv_search":
138
+ result = tool.search(tool_input.get("query", ""), tool_input.get("max_results"))
139
+ elif tool_name == "arxiv_get_paper":
140
+ result = tool.get_paper_by_id(tool_input.get("paper_id", ""))
141
+ elif tool_name == "arxiv_search_category":
142
+ result = tool.search_by_category(tool_input.get("category", ""), tool_input.get("max_results"))
143
+ else:
144
+ # Generic tool execution
145
+ result = tool.run(tool_input)
146
+
147
+ # Create a tool result
148
+ tool_result = {
149
+ "tool_name": tool_name,
150
+ "success": True,
151
+ "result": result,
152
+ "error": None
153
+ }
154
+
155
+ # Update the state
156
+ new_state = state.copy()
157
+ new_state["tool_results"] = state.get("tool_results", []) + [tool_result]
158
+ new_state["current_step"] += 1
159
+
160
+ return new_state
161
+ except Exception as e:
162
+ # Handle error...
163
+ # (rest of the function remains the same)
164
+ }
165
+ ```
166
+
167
+ ## Comparison of Search Providers
168
+
169
+ | Provider | API Key Required | Result Format | Strengths | Limitations |
170
+ |-------------|------------------|------------------------------|--------------------------------------------|--------------------------------------------|
171
+ | DuckDuckGo | No | List of result dictionaries | Privacy-focused, No API key | Sometimes less comprehensive |
172
+ | Serper | Yes | List of result dictionaries | Google results, Structured data | Requires API key, Usage limits |
173
+ | Perplexity | Yes | Content with citations | AI-powered, Reasoning capabilities | Requires API key, Slower response time |
174
+ | arXiv | No | List of paper dictionaries | Academic papers, Scientific research | Limited to academic content |
175
+
176
+ ## Best Practices
177
+
178
+ 1. **Fallback Mechanism**: Implement a fallback mechanism to try alternative search providers if one fails.
179
+
180
+ ```python
181
+ def robust_search(query):
182
+ # Try DuckDuckGo first (no API key required)
183
+ try:
184
+ results = duckduckgo_tool.search(query)
185
+ if results:
186
+ return {"provider": "duckduckgo", "results": results}
187
+ except Exception as e:
188
+ logger.warning(f"DuckDuckGo search failed: {str(e)}")
189
+
190
+ # Try Serper as fallback
191
+ try:
192
+ results = serper_tool.search(query)
193
+ if results:
194
+ return {"provider": "serper", "results": results}
195
+ except Exception as e:
196
+ logger.warning(f"Serper search failed: {str(e)}")
197
+
198
+ # Try Perplexity as third option
199
+ try:
200
+ results = perplexity_tool.search(query)
201
+ if results:
202
+ return {"provider": "perplexity", "results": results}
203
+ except Exception as e:
204
+ logger.warning(f"Perplexity search failed: {str(e)}")
205
+
206
+ # Try arXiv as last resort
207
+ try:
208
+ results = arxiv_tool.search(query)
209
+ if results:
210
+ return {"provider": "arxiv", "results": results}
211
+ except Exception as e:
212
+ logger.warning(f"arXiv search failed: {str(e)}")
213
+
214
+ # All providers failed
215
+ raise Exception("All search providers failed")
216
+ ```
217
+
218
+ 2. **Query Optimization**: Different search providers may respond better to different query formats.
219
+
220
+ 3. **Result Merging**: For comprehensive results, consider merging results from multiple providers.
221
+
222
+ ```python
223
+ def merged_search(query):
224
+ results = {
225
+ "duckduckgo": [],
226
+ "serper": [],
227
+ "perplexity": None,
228
+ "arxiv": []
229
+ }
230
+
231
+ # Collect results from all providers
232
+ try:
233
+ results["duckduckgo"] = duckduckgo_tool.search(query)
234
+ except Exception:
235
+ pass
236
+
237
+ try:
238
+ results["serper"] = serper_tool.search(query)
239
+ except Exception:
240
+ pass
241
+
242
+ try:
243
+ results["perplexity"] = perplexity_tool.search(query)
244
+ except Exception:
245
+ pass
246
+
247
+ # Try arXiv
248
+ try:
249
+ results["arxiv"] = arxiv_tool.search(query)
250
+ except Exception:
251
+ pass
252
+
253
+ # Merge and deduplicate results
254
+ merged = []
255
+ seen_urls = set()
256
+
257
+ # Add DuckDuckGo results
258
+ for result in results["duckduckgo"]:
259
+ if result["link"] not in seen_urls:
260
+ merged.append(result)
261
+ seen_urls.add(result["link"])
262
+
263
+ # Add Serper results
264
+ for result in results["serper"]:
265
+ if result["link"] not in seen_urls:
266
+ merged.append(result)
267
+ seen_urls.add(result["link"])
268
+
269
+ # Add arXiv results
270
+ arxiv_results = []
271
+ for result in results["arxiv"]:
272
+ if "url" in result and result["url"] not in seen_urls:
273
+ # Convert arXiv result to standard format
274
+ formatted_result = {
275
+ "title": result.get("title", ""),
276
+ "link": result.get("url", ""),
277
+ "snippet": result.get("summary", "")[:200] + "..." if result.get("summary") else ""
278
+ }
279
+ merged.append(formatted_result)
280
+ seen_urls.add(result["url"])
281
+ arxiv_results.append(result) # Keep original format for context
282
+
283
+ # Add Perplexity content as context
284
+ perplexity_content = results["perplexity"].get("content") if results["perplexity"] else None
285
+
286
+ return {
287
+ "merged_results": merged,
288
+ "perplexity_context": perplexity_content,
289
+ "arxiv_context": arxiv_results
290
+ }
291
+ ```
292
+
293
+ ## Testing
294
+
295
+ To test the integration of all search providers, run:
296
+
297
+ ```bash
298
+ python tests/test_duckduckgo_integration.py
299
+ python tests/test_arxiv_integration.py
300
+ ```
301
+
302
+ These tests verify:
303
+ 1. Direct connection to each search provider
304
+ 2. Proper functioning of each search tool
305
+ 3. Error handling
306
+ 4. Using multiple search providers together
307
+ 5. Integration with the GAIA agent
308
+
309
+ ## Conclusion
310
+
311
+ By integrating multiple search providers, the GAIA agent can provide more robust and comprehensive search capabilities. The agent can fall back to alternative providers if one fails, and can leverage the strengths of each provider for different types of queries:
312
+
313
+ - Use DuckDuckGo for general web searches without API key requirements
314
+ - Use Serper for more comprehensive Google search results
315
+ - Use Perplexity for AI-powered reasoning and analysis
316
+ - Use arXiv for academic and scientific research papers
317
+
318
+ This multi-provider approach ensures the GAIA agent can handle a wide range of search queries and provide the most relevant information to users.
docs/openai_integration_testing.md ADDED
@@ -0,0 +1,138 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Testing OpenAI Integration with Newer Models
2
+
3
+ This document provides instructions for testing the OpenAI integration with newer models like GPT-4o in the GAIA agent.
4
+
5
+ ## Overview
6
+
7
+ The GAIA agent has been updated to support newer OpenAI models, including:
8
+
9
+ - **GPT-4o**: The latest multimodal model that replaces both GPT-4-turbo and GPT-4-vision-preview
10
+ - **text-embedding-3-large**: The latest embedding model that replaces text-embedding-ada-002
11
+
12
+ These test scripts verify that the configuration changes are working properly and that the GAIA agent can use these newer models effectively.
13
+
14
+ ## Prerequisites
15
+
16
+ Before running the tests, make sure you have:
17
+
18
+ 1. Set up your environment variables with the necessary API keys:
19
+ - `OPENAI_API_KEY`: Your OpenAI API key
20
+ - `PERPLEXITY_API_KEY`: Your Perplexity API key (for multi-provider tests)
21
+
22
+ 2. Installed all required dependencies:
23
+ ```bash
24
+ pip install -r requirements.txt
25
+ ```
26
+
27
+ ## Test Scripts
28
+
29
+ ### 1. OpenAI Integration Test
30
+
31
+ This script tests the OpenAI integration with newer models.
32
+
33
+ **File**: `tests/test_openai_integration.py`
34
+
35
+ **Features**:
36
+ - Verifies connection to the OpenAI API
37
+ - Tests sending requests to newer models (GPT-4o, etc.)
38
+ - Tests vision capabilities with GPT-4o
39
+ - Optionally tests the GAIA agent with OpenAI models
40
+
41
+ **Usage**:
42
+ ```bash
43
+ python tests/test_openai_integration.py [--verbose] [--output FILENAME] [--test-agent]
44
+ ```
45
+
46
+ **Options**:
47
+ - `--verbose`, `-v`: Enable verbose logging
48
+ - `--output`, `-o`: Specify output file for test results (default: openai_test_results.json)
49
+ - `--test-agent`, `-a`: Test the GAIA agent with OpenAI models
50
+
51
+ ### 2. GAIA Multi-Provider Test
52
+
53
+ This script tests the GAIA agent's ability to use both OpenAI and Perplexity tools together.
54
+
55
+ **File**: `tests/test_gaia_multi_provider.py`
56
+
57
+ **Features**:
58
+ - Tests OpenAI-only questions
59
+ - Tests Perplexity-only questions
60
+ - Tests questions that require both providers
61
+ - Tests memory components with multi-provider data
62
+
63
+ **Usage**:
64
+ ```bash
65
+ python tests/test_gaia_multi_provider.py [--verbose] [--output FILENAME]
66
+ ```
67
+
68
+ **Options**:
69
+ - `--verbose`, `-v`: Enable verbose logging
70
+ - `--output`, `-o`: Specify output file for test results (default: gaia_multi_provider_test_report.json)
71
+
72
+ ## Test Results
73
+
74
+ Both test scripts generate detailed reports in JSON format, which include:
75
+
76
+ - Connection status
77
+ - Model test results
78
+ - Vision test results (for OpenAI)
79
+ - Agent test results
80
+ - Memory component test results
81
+ - Summary statistics
82
+
83
+ The reports are saved to the specified output file and also printed to the console.
84
+
85
+ ## Interpreting Results
86
+
87
+ ### OpenAI Integration Test
88
+
89
+ A successful test should show:
90
+ - ✅ Connection to OpenAI API established
91
+ - ✅ All model tests passed
92
+ - ✅ Vision test passed (if applicable)
93
+ - ✅ Agent test passed (if enabled)
94
+
95
+ ### GAIA Multi-Provider Test
96
+
97
+ A successful test should show:
98
+ - ✅ Both API keys available
99
+ - ✅ OpenAI-only questions answered correctly
100
+ - ✅ Perplexity-only questions answered correctly
101
+ - ✅ Combined questions answered correctly
102
+ - ✅ All memory components working
103
+
104
+ ## Troubleshooting
105
+
106
+ ### Common Issues
107
+
108
+ 1. **API Key Issues**:
109
+ - Make sure your API keys are set correctly in your environment variables
110
+ - Check that your API keys have not expired
111
+ - Verify that your API keys have access to the models being tested
112
+
113
+ 2. **Model Availability**:
114
+ - Some newer models may not be available to all accounts
115
+ - Check the OpenAI dashboard to see which models you have access to
116
+
117
+ 3. **Rate Limiting**:
118
+ - If you encounter rate limiting errors, try running the tests with fewer questions
119
+ - Add delays between requests by modifying the test scripts
120
+
121
+ 4. **Memory Issues**:
122
+ - If memory tests fail, check your Supabase configuration
123
+ - Verify that the memory tables exist and have the correct schema
124
+
125
+ ## Next Steps
126
+
127
+ After verifying that the OpenAI integration works with newer models, you can:
128
+
129
+ 1. Update your application to use the newer models
130
+ 2. Optimize your prompts for the newer models
131
+ 3. Implement additional features that leverage the capabilities of the newer models
132
+ 4. Monitor performance and costs to ensure efficient usage
133
+
134
+ ## Additional Resources
135
+
136
+ - [OpenAI API Documentation](https://platform.openai.com/docs/api-reference)
137
+ - [GPT-4o Documentation](https://platform.openai.com/docs/models/gpt-4o)
138
+ - [Perplexity API Documentation](https://docs.perplexity.ai/)
docs/serper_integration.md ADDED
@@ -0,0 +1,153 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Serper API Integration
2
+
3
+ This document provides information about the Serper API integration in the GAIA agent, including configuration, usage, and testing procedures.
4
+
5
+ ## Overview
6
+
7
+ Serper is a Google Search API that our GAIA agent uses for web searches. It provides a reliable way to access Google search results programmatically, which enhances the agent's ability to retrieve up-to-date information from the web.
8
+
9
+ ## Configuration
10
+
11
+ The Serper API integration requires the following configuration:
12
+
13
+ 1. **API Key**: Set the `SERPER_API_KEY` environment variable with your Serper API key.
14
+ 2. **API URL**: The API endpoint is configured in `agent/config.py` as `SERPER_API_URL`.
15
+
16
+ Example configuration in `.env` file:
17
+
18
+ ```
19
+ SERPER_API_KEY=your_serper_api_key_here
20
+ ```
21
+
22
+ ## Testing Results
23
+
24
+ We have successfully tested the Serper API integration and confirmed that it works correctly. The key findings from our testing are:
25
+
26
+ 1. **API Connection**: The Serper API connection is working correctly. We can successfully connect to the API and receive a 200 status code.
27
+ 2. **Search Functionality**: The search functionality works as expected. We can send search queries and receive relevant results.
28
+ 3. **Response Format**: The API returns results in the expected format, with organic search results containing title, link, and snippet information.
29
+
30
+ ### Sample Response
31
+
32
+ Here's an example of the response format from the Serper API:
33
+
34
+ ```json
35
+ {
36
+ "organic": [
37
+ {
38
+ "title": "Artificial Intelligence News - ScienceDaily",
39
+ "link": "https://www.sciencedaily.com/news/computers_math/artificial_intelligence/",
40
+ "snippet": "Latest Headlines • AI Tool Outperforms Most Doctors On Exams • Brain-Inspired AI Breakthrough • Robot See, Robot Do • Explainable AI for Ship Navigation Raises ..."
41
+ },
42
+ {
43
+ "title": "...",
44
+ "link": "...",
45
+ "snippet": "..."
46
+ }
47
+ ]
48
+ }
49
+ ```
50
+
51
+ ### Environment Configuration
52
+
53
+ The Serper API key must be properly configured in the `.env` file. Our testing confirmed that the key is correctly loaded from the `.env` file and used in API requests.
54
+
55
+ ### Integration with GAIA Agent
56
+
57
+ The Serper API is integrated with the GAIA agent through the `SerperSearchTool` class in `tools/web_tools.py`. This integration allows the agent to use the Serper API for web searches as part of its capabilities.
58
+
59
+ ## Implementation
60
+
61
+ The Serper API is integrated through the `SerperSearchTool` class in `tools/web_tools.py`. This class:
62
+
63
+ 1. Initializes with the Serper API key and URL
64
+ 2. Provides a `search` method that accepts a query string
65
+ 3. Formats the request with appropriate headers and payload
66
+ 4. Processes the response to extract relevant search results
67
+ 5. Returns the results in a standardized format
68
+
69
+ ## Usage
70
+
71
+ ### Basic Usage
72
+
73
+ ```python
74
+ from tools.web_tools import create_serper_search
75
+
76
+ # Create a Serper search tool instance
77
+ serper_tool = create_serper_search()
78
+
79
+ # Perform a search
80
+ results = serper_tool.search("artificial intelligence latest developments")
81
+
82
+ # Process the results
83
+ for result in results:
84
+ print(f"Title: {result['title']}")
85
+ print(f"Link: {result['link']}")
86
+ print(f"Snippet: {result['snippet']}")
87
+ print()
88
+ ```
89
+
90
+ ### Integration with GAIA Agent
91
+
92
+ ```python
93
+ from agent.agent import GaiaAgent
94
+ from tools.web_tools import create_serper_search
95
+
96
+ # Create a GAIA agent instance
97
+ agent = GaiaAgent()
98
+
99
+ # Create and register the Serper search tool
100
+ serper_tool = create_serper_search()
101
+ agent.register_tool("serper_search", serper_tool.search)
102
+
103
+ # Run the agent with a query that would use the Serper tool
104
+ response = agent.run(
105
+ "What are the latest developments in artificial intelligence?",
106
+ tools=["serper_search"]
107
+ )
108
+
109
+ print(response)
110
+ ```
111
+
112
+ ## Testing
113
+
114
+ We have implemented several tests to verify the Serper API integration:
115
+
116
+ 1. **Standalone API Test**: `tests/test_serper_api.py` - Tests the direct connection to the Serper API.
117
+ 2. **Unit Tests**: `tests/test_tools.py` - Contains unit tests for the `SerperSearchTool` class.
118
+ 3. **Integration Tests**: `tests/test_gaia_serper_integration.py` - Tests the integration of Serper with the GAIA agent.
119
+ 4. **Connection Verification**: `tests/verify_connections.py` - Includes a function to verify the Serper API connection.
120
+
121
+ ### Running the Tests
122
+
123
+ To test the Serper API connection:
124
+
125
+ ```bash
126
+ python tests/test_serper_api.py
127
+ ```
128
+
129
+ To run the integration tests:
130
+
131
+ ```bash
132
+ python tests/test_gaia_serper_integration.py
133
+ ```
134
+
135
+ To verify all connections including Serper:
136
+
137
+ ```bash
138
+ python tests/verify_connections.py
139
+ ```
140
+
141
+ ## Troubleshooting
142
+
143
+ Common issues and their solutions:
144
+
145
+ 1. **API Key Not Set**: Ensure the `SERPER_API_KEY` environment variable is set correctly.
146
+ 2. **Connection Timeout**: Check your internet connection and try again later.
147
+ 3. **Rate Limiting**: Serper API has rate limits. Check your usage and consider upgrading your plan if necessary.
148
+ 4. **Invalid Response Format**: If the API response format changes, update the `SerperSearchTool.search` method to handle the new format.
149
+
150
+ ## References
151
+
152
+ - [Serper API Documentation](https://serper.dev/docs)
153
+ - [GAIA Agent Documentation](./configuration.md)
docs/supabase_memory_integration.md ADDED
@@ -0,0 +1,175 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Supabase Memory Integration for GAIA Agent
2
+
3
+ This document provides an overview of the Supabase memory integration for the GAIA agent, including setup instructions, table structure, and testing procedures.
4
+
5
+ ## Overview
6
+
7
+ The GAIA agent uses Supabase as a persistent memory storage solution. This allows the agent to maintain memory across different sessions and provides a reliable way to store conversation history, cache results, and track intermediate results during agent execution.
8
+
9
+ The memory implementation consists of the following components:
10
+
11
+ 1. **SupabaseMemory**: Base memory implementation using Supabase for storage
12
+ 2. **ConversationMemory**: Memory implementation for storing conversation history
13
+ 3. **ResultCache**: Memory implementation for caching tool results and answers
14
+ 4. **WorkingMemory**: Memory implementation for storing intermediate results during agent execution
15
+
16
+ ## Supabase Setup
17
+
18
+ ### Prerequisites
19
+
20
+ - A Supabase account
21
+ - A Supabase project
22
+ - Supabase API key with appropriate permissions
23
+
24
+ ### Environment Variables
25
+
26
+ The following environment variables need to be set:
27
+
28
+ ```
29
+ SUPABASE_KEY=your_supabase_api_key
30
+ ```
31
+
32
+ The Supabase URL and project details are configured in `agent/config.py`:
33
+
34
+ ```python
35
+ # Supabase Configuration
36
+ SUPABASE_URL = "https://tjamxhvvtnypbadvrkjq.supabase.co"
37
+ SUPABASE_PROJECT_ID = "tjamxhvvtnypbadvrkjq"
38
+ SUPABASE_PROJECT_NAME = "HF_GAIA_Assessment_AGT"
39
+ ```
40
+
41
+ ### Table Structure
42
+
43
+ The Supabase integration uses the following tables:
44
+
45
+ 1. **gaia_memory**: Main memory table for storing key-value pairs
46
+ 2. **gaia_memory_conversations**: Conversation history table
47
+ 3. **gaia_memory_cache**: Cache table for storing tool results and answers
48
+ 4. **gaia_memory_working**: Working memory table for storing intermediate results
49
+ 5. **test_connection**: Test table for verifying connection
50
+
51
+ You can create these tables by running the SQL statements in `supabase_tables.sql` in the Supabase SQL Editor.
52
+
53
+ ## Memory Implementation
54
+
55
+ ### SupabaseMemory
56
+
57
+ The `SupabaseMemory` class provides the base memory implementation using Supabase for storage. It supports the following operations:
58
+
59
+ - `store(key, value, metadata, ttl)`: Store a value in memory
60
+ - `retrieve(key, default)`: Retrieve a value from memory
61
+ - `delete(key)`: Delete a value from memory
62
+ - `clear()`: Clear all values from memory
63
+ - `list_keys(prefix)`: List all keys in memory
64
+
65
+ ### ConversationMemory
66
+
67
+ The `ConversationMemory` class provides memory implementation for storing conversation history. It supports the following operations:
68
+
69
+ - `add_message(role, content, metadata)`: Add a message to the conversation history
70
+ - `get_messages(limit, before, after)`: Get messages from the conversation history
71
+ - `clear()`: Clear the conversation history
72
+
73
+ ### ResultCache
74
+
75
+ The `ResultCache` class provides memory implementation for caching tool results and answers. It supports the following operations:
76
+
77
+ - `cache_result(query, result, metadata, ttl)`: Cache a result
78
+ - `get_result(query, default)`: Get a cached result
79
+ - `invalidate(key)`: Invalidate a cached result
80
+ - `clear()`: Clear all cached results
81
+
82
+ ### WorkingMemory
83
+
84
+ The `WorkingMemory` class provides memory implementation for storing intermediate results during agent execution. It supports the following operations:
85
+
86
+ - `store_intermediate_result(key, value, metadata, ttl)`: Store an intermediate result
87
+ - `get_intermediate_result(key, default)`: Get an intermediate result
88
+ - `list_intermediate_results()`: List all intermediate results
89
+ - `clear()`: Clear all intermediate results
90
+
91
+ ## Testing
92
+
93
+ ### Test Scripts
94
+
95
+ The following test scripts are available for testing the Supabase memory integration:
96
+
97
+ 1. **test_supabase_connection.py**: Tests the connection to Supabase and verifies that the required tables exist
98
+ 2. **test_supabase_memory.py**: Tests the Supabase memory module implementation
99
+ 3. **test_gaia_supabase_integration.py**: Tests the integration of the GAIA agent with Supabase memory
100
+ 4. **run_supabase_integration_tests.py**: Runs all the Supabase integration tests and documents the results
101
+
102
+ ### Running the Tests
103
+
104
+ To run all the Supabase integration tests, use the following command:
105
+
106
+ ```bash
107
+ python tests/run_supabase_integration_tests.py
108
+ ```
109
+
110
+ For verbose output, add the `--verbose` flag:
111
+
112
+ ```bash
113
+ python tests/run_supabase_integration_tests.py --verbose
114
+ ```
115
+
116
+ To specify a custom output directory for the test results, use the `--output-dir` option:
117
+
118
+ ```bash
119
+ python tests/run_supabase_integration_tests.py --output-dir custom_docs
120
+ ```
121
+
122
+ ### Test Results
123
+
124
+ The test results are documented in a markdown file in the specified output directory (default: `docs`). The file includes:
125
+
126
+ - Test summary with date and Supabase project details
127
+ - Results for each test with status (passed/failed)
128
+ - Detailed test output in collapsible sections
129
+ - Overall assessment and recommendations
130
+
131
+ ## Troubleshooting
132
+
133
+ If the tests fail, check the following:
134
+
135
+ 1. **Supabase Connection**:
136
+ - Verify that the Supabase URL and API key are correctly set
137
+ - Check that the Supabase project is active
138
+ - Ensure that the API key has the necessary permissions
139
+
140
+ 2. **Table Structure**:
141
+ - Verify that all required tables exist in the Supabase database
142
+ - Check that the tables have the correct structure
143
+ - Ensure that the indexes are properly created
144
+
145
+ 3. **Memory Configuration**:
146
+ - Check that memory is enabled in the configuration
147
+ - Verify that the table names are correctly set
148
+ - Ensure that the TTL and cache size settings are appropriate
149
+
150
+ 4. **Agent Integration**:
151
+ - Check that the agent is correctly using the memory components
152
+ - Verify that the conversation ID and session ID are properly set
153
+ - Ensure that the agent is correctly processing questions
154
+
155
+ ## Best Practices
156
+
157
+ 1. **Memory Management**:
158
+ - Regularly clear unused memory to prevent excessive storage usage
159
+ - Use appropriate TTL values for different types of memory entries
160
+ - Monitor memory usage and adjust cache sizes as needed
161
+
162
+ 2. **Error Handling**:
163
+ - Implement proper error handling for memory operations
164
+ - Gracefully handle connection issues and retry operations when appropriate
165
+ - Log memory-related errors for debugging purposes
166
+
167
+ 3. **Security**:
168
+ - Use a secure API key with appropriate permissions
169
+ - Do not store sensitive information in memory
170
+ - Regularly rotate API keys for security
171
+
172
+ 4. **Performance**:
173
+ - Use efficient queries to minimize latency
174
+ - Implement caching to reduce database load
175
+ - Monitor query performance and optimize as needed
docs/troubleshooting.md ADDED
@@ -0,0 +1,281 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Troubleshooting Guide for GAIA Agent
2
+
3
+ This guide addresses common issues you might encounter when setting up, configuring, or running the GAIA agent. It provides solutions and workarounds for these issues.
4
+
5
+ ## Table of Contents
6
+
7
+ - [Dependency Issues](#dependency-issues)
8
+ - [langchain_openai Issues](#langchain_openai-issues)
9
+ - [Other Dependency Issues](#other-dependency-issues)
10
+ - [API Connection Issues](#api-connection-issues)
11
+ - [OpenAI API Issues](#openai-api-issues)
12
+ - [Serper API Issues](#serper-api-issues)
13
+ - [Perplexity API Issues](#perplexity-api-issues)
14
+ - [Supabase Issues](#supabase-issues)
15
+ - [Connection Issues](#connection-issues)
16
+ - [Table Structure Issues](#table-structure-issues)
17
+ - [Memory Issues](#memory-issues)
18
+ - [Agent Execution Issues](#agent-execution-issues)
19
+ - [Testing Issues](#testing-issues)
20
+
21
+ ## Dependency Issues
22
+
23
+ ### langchain_openai Issues
24
+
25
+ The `langchain_openai` package is a critical dependency for the GAIA agent, as it provides the interface to OpenAI's language models. Here are common issues and their solutions:
26
+
27
+ #### ImportError: No module named 'langchain_openai'
28
+
29
+ This error occurs when the `langchain_openai` package is not installed or not in your Python path.
30
+
31
+ **Solution:**
32
+ ```bash
33
+ pip install langchain_openai
34
+ ```
35
+
36
+ If you already have it installed, try upgrading to the latest version:
37
+ ```bash
38
+ pip install --upgrade langchain_openai
39
+ ```
40
+
41
+ #### AttributeError: module 'langchain_openai' has no attribute 'ChatOpenAI'
42
+
43
+ This error can occur if you have an outdated version of `langchain_openai` or if there's a conflict with other packages.
44
+
45
+ **Solutions:**
46
+
47
+ 1. Upgrade to the latest version:
48
+ ```bash
49
+ pip install --upgrade langchain_openai
50
+ ```
51
+
52
+ 2. Check for conflicts with older versions of langchain:
53
+ ```bash
54
+ pip uninstall langchain-openai # Note the hyphen instead of underscore
55
+ pip install langchain_openai # Install the correct package
56
+ ```
57
+
58
+ 3. Verify your import statement:
59
+ ```python
60
+ # Correct import
61
+ from langchain_openai import ChatOpenAI
62
+
63
+ # Incorrect import (old style)
64
+ # from langchain.chat_models import ChatOpenAI
65
+ ```
66
+
67
+ #### Version Compatibility Issues
68
+
69
+ If you're experiencing compatibility issues between `langchain`, `langchain_openai`, and other LangChain packages:
70
+
71
+ **Solution:**
72
+ ```bash
73
+ pip install --upgrade langchain langchain_openai langgraph
74
+ ```
75
+
76
+ This ensures all LangChain-related packages are updated to compatible versions.
77
+
78
+ ### Other Dependency Issues
79
+
80
+ #### Missing Dependencies
81
+
82
+ If you encounter errors about missing dependencies:
83
+
84
+ **Solution:**
85
+ ```bash
86
+ pip install -r requirements.txt
87
+ ```
88
+
89
+ #### Conflicting Dependencies
90
+
91
+ If you have conflicting dependencies:
92
+
93
+ **Solution:**
94
+ ```bash
95
+ # Create a fresh virtual environment
96
+ python -m venv fresh_env
97
+ source fresh_env/bin/activate # On Windows: fresh_env\Scripts\activate
98
+ pip install -r requirements.txt
99
+ ```
100
+
101
+ ## API Connection Issues
102
+
103
+ ### OpenAI API Issues
104
+
105
+ #### Authentication Error
106
+
107
+ If you see an error like `AuthenticationError: Incorrect API key provided`:
108
+
109
+ **Solutions:**
110
+ 1. Verify your API key is correct and active in your OpenAI account
111
+ 2. Check that the environment variable is set correctly:
112
+ ```bash
113
+ # In your .env file
114
+ OPENAI_API_KEY=your_actual_api_key
115
+ ```
116
+ 3. Try setting the API key directly in your code (for testing only):
117
+ ```python
118
+ import os
119
+ os.environ["OPENAI_API_KEY"] = "your_actual_api_key"
120
+ ```
121
+
122
+ #### Rate Limit Exceeded
123
+
124
+ If you encounter rate limit errors:
125
+
126
+ **Solutions:**
127
+ 1. Implement exponential backoff and retry logic
128
+ 2. Reduce the frequency of API calls
129
+ 3. Consider upgrading your OpenAI plan for higher rate limits
130
+
131
+ ### Serper API Issues
132
+
133
+ #### Authentication Error
134
+
135
+ If you encounter authentication errors with Serper:
136
+
137
+ **Solution:**
138
+ Verify your Serper API key is correctly set in the environment variables:
139
+ ```bash
140
+ # In your .env file
141
+ SERPER_API_KEY=your_serper_api_key
142
+ ```
143
+
144
+ #### No Results Returned
145
+
146
+ If Serper searches return no results:
147
+
148
+ **Solutions:**
149
+ 1. Check your query formatting
150
+ 2. Verify that Serper is available and operational
151
+ 3. Try using an alternative search provider like DuckDuckGo
152
+
153
+ ### Perplexity API Issues
154
+
155
+ #### Authentication Error
156
+
157
+ If you encounter authentication errors with Perplexity:
158
+
159
+ **Solution:**
160
+ Verify your Perplexity API key is correctly set in the environment variables:
161
+ ```bash
162
+ # In your .env file
163
+ PERPLEXITY_API_KEY=your_perplexity_api_key
164
+ ```
165
+
166
+ #### Model Not Available
167
+
168
+ If you encounter errors about a model not being available:
169
+
170
+ **Solution:**
171
+ Check the available models in your Perplexity account and update the configuration accordingly.
172
+
173
+ ## Supabase Issues
174
+
175
+ ### Connection Issues
176
+
177
+ #### Authentication Error
178
+
179
+ If you encounter authentication errors with Supabase:
180
+
181
+ **Solution:**
182
+ Verify your Supabase URL and API key are correctly set:
183
+ ```bash
184
+ # In your .env file
185
+ SUPABASE_URL=https://your-project-id.supabase.co
186
+ SUPABASE_KEY=your_supabase_api_key
187
+ ```
188
+
189
+ #### Connection Timeout
190
+
191
+ If you encounter connection timeouts:
192
+
193
+ **Solutions:**
194
+ 1. Check your internet connection
195
+ 2. Verify that your Supabase project is active
196
+ 3. Implement retry logic with exponential backoff
197
+
198
+ ### Table Structure Issues
199
+
200
+ #### Missing Tables
201
+
202
+ If you encounter errors about missing tables:
203
+
204
+ **Solution:**
205
+ Run the SQL statements in `supabase_tables.sql` in the Supabase SQL Editor to create the required tables.
206
+
207
+ #### Incorrect Table Structure
208
+
209
+ If you encounter errors about incorrect table structure:
210
+
211
+ **Solution:**
212
+ Compare your table structure with the expected structure in `supabase_tables.sql` and make any necessary adjustments.
213
+
214
+ ## Memory Issues
215
+
216
+ #### Memory Not Persisting
217
+
218
+ If the agent's memory is not persisting between sessions:
219
+
220
+ **Solutions:**
221
+ 1. Verify that memory is enabled in the configuration
222
+ 2. Check that the Supabase connection is working correctly
223
+ 3. Verify that the memory tables exist and have the correct structure
224
+
225
+ #### Memory Retrieval Issues
226
+
227
+ If the agent is unable to retrieve memory:
228
+
229
+ **Solutions:**
230
+ 1. Check the memory key format
231
+ 2. Verify that the TTL (time-to-live) settings are appropriate
232
+ 3. Check for errors in the Supabase logs
233
+
234
+ ## Agent Execution Issues
235
+
236
+ #### Agent Not Responding
237
+
238
+ If the agent is not responding or taking too long:
239
+
240
+ **Solutions:**
241
+ 1. Check the timeout settings in the configuration
242
+ 2. Verify that all required APIs are available
243
+ 3. Check for errors in the logs
244
+
245
+ #### Incorrect Answers
246
+
247
+ If the agent is providing incorrect answers:
248
+
249
+ **Solutions:**
250
+ 1. Check the question format
251
+ 2. Verify that the agent is using the correct tools
252
+ 3. Check the answer extraction logic
253
+
254
+ ## Testing Issues
255
+
256
+ #### Tests Failing
257
+
258
+ If tests are failing:
259
+
260
+ **Solutions:**
261
+ 1. Check that all required environment variables are set
262
+ 2. Verify that all required APIs are available
263
+ 3. Check for errors in the test logs
264
+
265
+ #### Test Timeouts
266
+
267
+ If tests are timing out:
268
+
269
+ **Solutions:**
270
+ 1. Increase the timeout settings
271
+ 2. Run tests individually to identify the problematic test
272
+ 3. Check for resource-intensive operations that might be causing the timeout
273
+
274
+ ## Additional Resources
275
+
276
+ If you're still experiencing issues after trying the solutions in this guide, consider the following resources:
277
+
278
+ 1. Check the [GitHub repository](https://github.com/your-repo) for open issues or create a new one
279
+ 2. Consult the [LangChain documentation](https://python.langchain.com/docs/get_started/introduction)
280
+ 3. Refer to the [OpenAI API documentation](https://platform.openai.com/docs/api-reference)
281
+ 4. Check the [Supabase documentation](https://supabase.com/docs)