Spaces:
Sleeping
Sleeping
Update core_agent.py
Browse files- core_agent.py +175 -0
core_agent.py
CHANGED
|
@@ -128,6 +128,181 @@ def analyze_csv_file(file_path: str, query: str) -> str:
|
|
| 128 |
except Exception as e:
|
| 129 |
return f"Error analyzing CSV file: {str(e)}"
|
| 130 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 131 |
|
| 132 |
|
| 133 |
|
|
|
|
| 128 |
except Exception as e:
|
| 129 |
return f"Error analyzing CSV file: {str(e)}"
|
| 130 |
|
| 131 |
+
|
| 132 |
+
@tool
|
| 133 |
+
def analyze_excel_file(file_path: str, query: str) -> str:
|
| 134 |
+
"""
|
| 135 |
+
Analyze an Excel file using pandas and answer a question about it.
|
| 136 |
+
|
| 137 |
+
Args:
|
| 138 |
+
file_path: Path to the Excel file
|
| 139 |
+
query: Question about the data
|
| 140 |
+
|
| 141 |
+
Returns:
|
| 142 |
+
Analysis result or error message
|
| 143 |
+
"""
|
| 144 |
+
try:
|
| 145 |
+
import pandas as pd
|
| 146 |
+
|
| 147 |
+
# Read the Excel file
|
| 148 |
+
df = pd.read_excel(file_path)
|
| 149 |
+
|
| 150 |
+
# Run various analyses based on the query
|
| 151 |
+
result = f"Excel file loaded with {len(df)} rows and {len(df.columns)} columns.\n"
|
| 152 |
+
result += f"Columns: {', '.join(df.columns)}\n\n"
|
| 153 |
+
|
| 154 |
+
# Add summary statistics
|
| 155 |
+
result += "Summary statistics:\n"
|
| 156 |
+
result += str(df.describe())
|
| 157 |
+
|
| 158 |
+
return result
|
| 159 |
+
except ImportError:
|
| 160 |
+
return "Error: pandas and openpyxl are not installed. Please install them with 'pip install pandas openpyxl'."
|
| 161 |
+
except Exception as e:
|
| 162 |
+
return f"Error analyzing Excel file: {str(e)}"
|
| 163 |
+
|
| 164 |
+
|
| 165 |
+
class GAIAAgent:
|
| 166 |
+
def __init__(
|
| 167 |
+
self,
|
| 168 |
+
model_type: str = "HfApiModel",
|
| 169 |
+
model_id: Optional[str] = None,
|
| 170 |
+
api_key: Optional[str] = None,
|
| 171 |
+
api_base: Optional[str] = None,
|
| 172 |
+
temperature: float = 0.2,
|
| 173 |
+
executor_type: str = "local", # Changed from use_e2b to executor_type
|
| 174 |
+
additional_imports: List[str] = None,
|
| 175 |
+
additional_tools: List[Any] = None,
|
| 176 |
+
system_prompt: Optional[str] = None, # We'll still accept this parameter but not use it directly
|
| 177 |
+
verbose: bool = False,
|
| 178 |
+
provider: Optional[str] = None, # Add provider for InferenceClientModel
|
| 179 |
+
timeout: Optional[int] = None
|
| 180 |
+
):
|
| 181 |
+
"""
|
| 182 |
+
Initialize a GAIAAgent with specified configuration
|
| 183 |
+
|
| 184 |
+
Args:
|
| 185 |
+
model_type: Type of model to use (HfApiModel, LiteLLMModel, OpenAIServerModel, InferenceClientModel)
|
| 186 |
+
model_id: ID of the model to use
|
| 187 |
+
api_key: API key for the model provider
|
| 188 |
+
api_base: Base URL for API calls
|
| 189 |
+
temperature: Temperature for text generation
|
| 190 |
+
executor_type: Type of executor for code execution ('local' or 'e2b')
|
| 191 |
+
additional_imports: Additional Python modules to allow importing
|
| 192 |
+
additional_tools: Additional tools to provide to the agent
|
| 193 |
+
system_prompt: Custom system prompt to use (not directly used, kept for backward compatibility)
|
| 194 |
+
verbose: Enable verbose logging
|
| 195 |
+
provider: Provider for InferenceClientModel (e.g., "hf-inference")
|
| 196 |
+
timeout: Timeout in seconds for API calls
|
| 197 |
+
"""
|
| 198 |
+
self.verbose = verbose
|
| 199 |
+
self.system_prompt = system_prompt # Store for potential future use
|
| 200 |
+
|
| 201 |
+
if model_type == "HfApiModel":
|
| 202 |
+
if api_key is None:
|
| 203 |
+
api_key = os.getenv("HUGGINGFACEHUB_API_TOKEN")
|
| 204 |
+
if not api_key:
|
| 205 |
+
raise ValueError("No Hugging Face token provided. Please set HUGGINGFACEHUB_API_TOKEN environment variable or pass api_key parameter.")
|
| 206 |
+
|
| 207 |
+
if self.verbose:
|
| 208 |
+
print(f"Using Hugging Face token: {api_key[:5]}...")
|
| 209 |
+
|
| 210 |
+
self.model = HfApiModel(
|
| 211 |
+
model_id=model_id or "meta-llama/Llama-3-70B-Instruct",
|
| 212 |
+
token=api_key,
|
| 213 |
+
temperature=temperature
|
| 214 |
+
)
|
| 215 |
+
else:
|
| 216 |
+
raise ValueError(f"Unknown model type: {model_type}")
|
| 217 |
+
|
| 218 |
+
if self.verbose:
|
| 219 |
+
print(f"Initialized model: {model_type} - {model_id}")
|
| 220 |
+
|
| 221 |
+
# Initialize default tools
|
| 222 |
+
self.tools = [
|
| 223 |
+
DuckDuckGoSearchTool(),
|
| 224 |
+
PythonInterpreterTool(),
|
| 225 |
+
save_file,
|
| 226 |
+
download_file_from_url,
|
| 227 |
+
analyze_csv_file,
|
| 228 |
+
analyze_excel_file
|
| 229 |
+
]
|
| 230 |
+
|
| 231 |
+
try:
|
| 232 |
+
from langchain.document_loaders import UnstructuredImageLoader
|
| 233 |
+
self.tools.append(extract_text_from_image)
|
| 234 |
+
if self.verbose:
|
| 235 |
+
print("Added image processing tool")
|
| 236 |
+
except ImportError:
|
| 237 |
+
if self.verbose:
|
| 238 |
+
print("Image processing libraries not available")
|
| 239 |
+
|
| 240 |
+
if additional_tools:
|
| 241 |
+
self.tools.extend(additional_tools)
|
| 242 |
+
|
| 243 |
+
if self.verbose:
|
| 244 |
+
print(f"Initialized with {len(self.tools)} tools")
|
| 245 |
+
|
| 246 |
+
self.imports = ["pandas", "numpy", "datetime", "json", "re", "math", "os", "requests", "csv", "urllib"]
|
| 247 |
+
if additional_imports:
|
| 248 |
+
self.imports.extend(additional_imports)
|
| 249 |
+
|
| 250 |
+
# Initialize the CodeAgent
|
| 251 |
+
executor_kwargs = {}
|
| 252 |
+
if executor_type == "e2b":
|
| 253 |
+
try:
|
| 254 |
+
# Try to import e2b dependencies to check if they're available
|
| 255 |
+
from e2b_code_interpreter import Sandbox
|
| 256 |
+
if self.verbose:
|
| 257 |
+
print("Using e2b executor")
|
| 258 |
+
except ImportError:
|
| 259 |
+
if self.verbose:
|
| 260 |
+
print("e2b dependencies not found, falling back to local executor")
|
| 261 |
+
executor_type = "local" # Fallback to local if e2b is not available
|
| 262 |
+
|
| 263 |
+
self.agent = CodeAgent(
|
| 264 |
+
tools=self.tools,
|
| 265 |
+
model=self.model,
|
| 266 |
+
additional_authorized_imports=self.imports,
|
| 267 |
+
executor_type=executor_type,
|
| 268 |
+
executor_kwargs=executor_kwargs,
|
| 269 |
+
verbosity_level=2 if self.verbose else 0
|
| 270 |
+
)
|
| 271 |
+
|
| 272 |
+
if self.verbose:
|
| 273 |
+
print("Agent initialized and ready")
|
| 274 |
+
|
| 275 |
+
|
| 276 |
+
def answer_question(self,question:str,task_file_path: Optional[str]=None):
|
| 277 |
+
"""
|
| 278 |
+
Process a GAIA benchmark question and return the answer
|
| 279 |
+
|
| 280 |
+
Args:
|
| 281 |
+
question: The question to answer
|
| 282 |
+
task_file_path: Optional path to a file associated with the question
|
| 283 |
+
|
| 284 |
+
Returns:
|
| 285 |
+
The answer to the question
|
| 286 |
+
"""
|
| 287 |
+
try:
|
| 288 |
+
if self.verbose:
|
| 289 |
+
print(f"Processing Question {question}")
|
| 290 |
+
if task_file_path:
|
| 291 |
+
print(f"with associated task file {task_file_path}")
|
| 292 |
+
|
| 293 |
+
context=question
|
| 294 |
+
fileContent=None
|
| 295 |
+
|
| 296 |
+
if task_file_path:
|
| 297 |
+
try:
|
| 298 |
+
with open(task_file_path,'r') as f:
|
| 299 |
+
fileContent=f.read()
|
| 300 |
+
|
| 301 |
+
import os
|
| 302 |
+
file_ext=os.path.splitext(task_file_path)[1].lower()
|
| 303 |
+
context=f"""Question: {question} This question has an associated file. Here is the file content:```{file_ext} {file_content}``` Analyze the file content above to answer the question."""
|
| 304 |
+
|
| 305 |
+
|
| 306 |
|
| 307 |
|
| 308 |
|