Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -6,13 +6,13 @@ import logging
|
|
| 6 |
import io
|
| 7 |
import json
|
| 8 |
import requests
|
|
|
|
| 9 |
from flask import Flask, request, jsonify, Response
|
| 10 |
from flask_cors import CORS
|
| 11 |
from datetime import datetime
|
| 12 |
from dotenv import load_dotenv
|
| 13 |
from llama_cpp import Llama
|
| 14 |
from huggingface_hub import hf_hub_download
|
| 15 |
-
from duckduckgo_search import DDGS
|
| 16 |
import pdfplumber
|
| 17 |
import docx
|
| 18 |
from PIL import Image
|
|
@@ -31,11 +31,11 @@ app = Flask(__name__)
|
|
| 31 |
CORS(app)
|
| 32 |
|
| 33 |
# --- Configuration & Model Selection ---
|
| 34 |
-
OLLAMA_URL = "http://localhost:11434"
|
| 35 |
HF_FALLBACK_REPO = "TheBloke/TinyLlama-1.1B-Chat-v1.0-GGUF"
|
| 36 |
HF_FALLBACK_FILE = "tinyllama-1.1b-chat-v1.0.Q4_K_M.gguf"
|
| 37 |
|
| 38 |
-
# Load Fallback Local Model
|
| 39 |
llm_fallback = None
|
| 40 |
try:
|
| 41 |
logger.info("Loading Hugging Face fallback model...")
|
|
@@ -78,151 +78,251 @@ Maintain a helpful, modern, and production-ready chat style (like ChatGPT).
|
|
| 78 |
Current Date: {date}
|
| 79 |
"""
|
| 80 |
|
| 81 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 82 |
def detect_tools(message):
|
| 83 |
-
"""
|
| 84 |
-
|
| 85 |
-
"""
|
| 86 |
-
message = message.lower()
|
| 87 |
tools = []
|
| 88 |
-
|
| 89 |
-
# Search Intent
|
| 90 |
search_keywords = [
|
| 91 |
"search", "latest", "news", "find out", "who is", "what is the current",
|
| 92 |
"research", "browse", "today", "current", "facts", "data", "tell me about",
|
| 93 |
-
"what happened", "weather", "stock", "price", "how many", "who won"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 94 |
]
|
|
|
|
|
|
|
| 95 |
if any(kw in message for kw in search_keywords):
|
| 96 |
tools.append("search")
|
| 97 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 98 |
# Image Intent
|
| 99 |
-
|
| 100 |
-
|
| 101 |
-
("make" in message and "image" in message):
|
| 102 |
tools.append("image")
|
| 103 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 104 |
return tools
|
| 105 |
|
| 106 |
-
#
|
| 107 |
-
def web_search(query: str, max_results: int = 5) -> list:
|
| 108 |
-
"""Search the web using DuckDuckGo v7.5.0 syntax. No API key required."""
|
| 109 |
-
try:
|
| 110 |
-
from duckduckgo_search import DDGS
|
| 111 |
-
with DDGS() as ddgs:
|
| 112 |
-
results = list(ddgs.text(query, max_results=max_results))
|
| 113 |
-
return results
|
| 114 |
-
except Exception as e:
|
| 115 |
-
logger.error(f"Search failed: {str(e)}")
|
| 116 |
-
return [{"error": f"Search failed: {str(e)}"}]
|
| 117 |
|
| 118 |
-
def
|
| 119 |
"""Format search results with markdown citations."""
|
| 120 |
-
if not data
|
| 121 |
-
|
| 122 |
-
|
| 123 |
-
|
| 124 |
-
|
| 125 |
-
|
| 126 |
-
|
| 127 |
-
|
|
|
|
|
|
|
| 128 |
for i, r in enumerate(results, 1):
|
| 129 |
title = r.get("title", "No title")
|
| 130 |
url = r.get("href", "#")
|
| 131 |
-
body = r.get("body", "No description")
|
| 132 |
-
|
| 133 |
-
|
| 134 |
-
|
| 135 |
-
|
| 136 |
-
|
| 137 |
-
|
| 138 |
-
|
| 139 |
-
|
| 140 |
-
|
| 141 |
-
|
| 142 |
-
|
| 143 |
-
|
| 144 |
-
|
| 145 |
-
|
| 146 |
-
|
| 147 |
-
|
| 148 |
-
|
| 149 |
-
|
| 150 |
-
|
| 151 |
-
|
| 152 |
-
|
| 153 |
-
|
| 154 |
-
|
| 155 |
-
# --- Image Generation Feature ---
|
| 156 |
-
@app.route('/api/image', methods=['POST'])
|
| 157 |
-
def generate_image():
|
| 158 |
-
data = request.json or {}
|
| 159 |
-
prompt = data.get("prompt", "").strip()
|
| 160 |
-
if not prompt:
|
| 161 |
-
return jsonify({"error": "No prompt provided"}), 400
|
| 162 |
-
|
| 163 |
-
try:
|
| 164 |
-
seed = int(time.time())
|
| 165 |
-
width = data.get("width", 1024)
|
| 166 |
-
height = data.get("height", 1024)
|
| 167 |
-
image_url = f"https://image.pollinations.ai/prompt/{requests.utils.quote(prompt)}?width={width}&height={height}&seed={seed}&nologo=true"
|
| 168 |
-
|
| 169 |
-
return jsonify({
|
| 170 |
-
"status": "success",
|
| 171 |
-
"image_url": image_url,
|
| 172 |
-
"prompt": prompt
|
| 173 |
-
})
|
| 174 |
-
except Exception as e:
|
| 175 |
-
logger.error(f"Image generation failed: {e}")
|
| 176 |
-
return jsonify({"error": "Image generation failed"}), 500
|
| 177 |
-
|
| 178 |
-
# --- File Analysis Feature ---
|
| 179 |
-
@app.route('/api/analyze-file', methods=['POST'])
|
| 180 |
-
def analyze_file():
|
| 181 |
-
if 'file' not in request.files:
|
| 182 |
-
return jsonify({"error": "No file uploaded"}), 400
|
| 183 |
-
|
| 184 |
-
file = request.files['file']
|
| 185 |
-
filename = file.filename
|
| 186 |
-
ext = filename.split('.')[-1].lower()
|
| 187 |
-
|
| 188 |
-
try:
|
| 189 |
-
text_content = ""
|
| 190 |
-
if ext == 'pdf':
|
| 191 |
-
with pdfplumber.open(file) as pdf:
|
| 192 |
-
text_content = "\n".join([page.extract_text() for page in pdf.pages if page.extract_text()])
|
| 193 |
-
elif ext == 'docx':
|
| 194 |
-
doc = docx.Document(file)
|
| 195 |
-
text_content = "\n".join([para.text for para in doc.paragraphs])
|
| 196 |
-
elif ext == 'txt':
|
| 197 |
-
text_content = file.read().decode('utf-8', errors='ignore')
|
| 198 |
-
elif ext in ['png', 'jpg', 'jpeg', 'webp']:
|
| 199 |
-
img = Image.open(file)
|
| 200 |
-
text_content = pytesseract.image_to_string(img)
|
| 201 |
-
else:
|
| 202 |
-
return jsonify({"error": f"Unsupported file type: {ext}"}), 400
|
| 203 |
-
|
| 204 |
-
if not text_content.strip():
|
| 205 |
-
return jsonify({"error": "Could not extract text from the file."}), 400
|
| 206 |
-
|
| 207 |
-
truncated_text = text_content[:4000]
|
| 208 |
-
return jsonify({
|
| 209 |
-
"filename": filename,
|
| 210 |
-
"content": truncated_text,
|
| 211 |
-
"full_length": len(text_content),
|
| 212 |
-
"status": "success"
|
| 213 |
-
})
|
| 214 |
-
except Exception as e:
|
| 215 |
-
logger.error(f"File analysis failed: {e}")
|
| 216 |
-
return jsonify({"error": f"Failed to process file: {str(e)}"}), 500
|
| 217 |
|
| 218 |
-
#
|
| 219 |
-
def get_time_response(message):
|
| 220 |
-
message = message.lower()
|
| 221 |
-
time_keywords = ["time", "date", "today", "now", "clock", "day"]
|
| 222 |
-
if any(re.search(rf"\b{word}\b", message) for word in time_keywords):
|
| 223 |
-
now = datetime.now()
|
| 224 |
-
return f"๐ The current time is **{now.strftime('%H:%M:%S')}**.\n๐
Today is **{now.strftime('%A, %Y-%m-%d')}**."
|
| 225 |
-
return None
|
| 226 |
|
| 227 |
@app.route('/')
|
| 228 |
def home():
|
|
@@ -243,6 +343,37 @@ def login():
|
|
| 243 |
data = request.json or {}
|
| 244 |
return jsonify({"user": {"username": data.get("username", "Guest"), "messages_used": 0}}), 200
|
| 245 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 246 |
@app.route('/api/chat', methods=['POST'])
|
| 247 |
def chat():
|
| 248 |
data = request.json or {}
|
|
@@ -251,7 +382,7 @@ def chat():
|
|
| 251 |
memory = data.get("memory", {})
|
| 252 |
mode = data.get("mode", "instant")
|
| 253 |
is_deep_research = data.get("deep_research", False)
|
| 254 |
-
|
| 255 |
if not message:
|
| 256 |
return jsonify({"error": "No message"}), 400
|
| 257 |
|
|
@@ -263,15 +394,20 @@ def chat():
|
|
| 263 |
# 2. Tool Detection
|
| 264 |
tools_needed = detect_tools(message)
|
| 265 |
tool_context = ""
|
| 266 |
-
|
| 267 |
if "search" in tools_needed:
|
| 268 |
-
|
| 269 |
-
|
| 270 |
-
|
| 271 |
-
|
| 272 |
-
tool_context += f"
|
|
|
|
|
|
|
|
|
|
| 273 |
else:
|
| 274 |
-
tool_context += "
|
|
|
|
|
|
|
| 275 |
|
| 276 |
# 3. Dynamic Prompting
|
| 277 |
current_date = datetime.now().strftime("%Y-%m-%d")
|
|
@@ -282,16 +418,19 @@ def chat():
|
|
| 282 |
"research": "Act as a professional researcher. Provide deep, structured answers with logical sections.",
|
| 283 |
"creative": "Act as a creative writer. Use evocative language and focus on ideation."
|
| 284 |
}
|
| 285 |
-
|
| 286 |
selected_mode_instruction = mode_instructions.get(mode, mode_instructions["instant"])
|
| 287 |
dynamic_system_prompt = SYSTEM_PROMPT.format(date=current_date)
|
| 288 |
-
dynamic_system_prompt += f"
|
| 289 |
-
|
|
|
|
| 290 |
if is_deep_research:
|
| 291 |
-
dynamic_system_prompt += "
|
|
|
|
| 292 |
|
| 293 |
if memory:
|
| 294 |
-
dynamic_system_prompt += f"
|
|
|
|
| 295 |
|
| 296 |
full_prompt = f"{message}{tool_context}"
|
| 297 |
|
|
@@ -305,7 +444,7 @@ def chat():
|
|
| 305 |
ollama_role = "assistant" if role == "ai" else "user"
|
| 306 |
messages.append({"role": ollama_role, "content": h.get('content', '')})
|
| 307 |
messages.append({"role": "user", "content": full_prompt})
|
| 308 |
-
|
| 309 |
response = requests.post(
|
| 310 |
f"{OLLAMA_URL}/api/chat",
|
| 311 |
json={
|
|
@@ -317,7 +456,7 @@ def chat():
|
|
| 317 |
timeout=5,
|
| 318 |
stream=True
|
| 319 |
)
|
| 320 |
-
|
| 321 |
if response.status_code == 200:
|
| 322 |
for line in response.iter_lines():
|
| 323 |
if line:
|
|
@@ -328,7 +467,8 @@ def chat():
|
|
| 328 |
"source": "colab",
|
| 329 |
"provider": "Ollama (Phi-3)",
|
| 330 |
"tools_used": tools_needed
|
| 331 |
-
}) + "
|
|
|
|
| 332 |
return
|
| 333 |
|
| 334 |
except Exception as colab_err:
|
|
@@ -337,19 +477,26 @@ def chat():
|
|
| 337 |
# Step B: Fallback to Local HF Model
|
| 338 |
if llm_fallback:
|
| 339 |
try:
|
| 340 |
-
llama_prompt = f"<|system|>
|
|
|
|
|
|
|
| 341 |
for h in history:
|
| 342 |
role = h.get('role', 'user')
|
| 343 |
-
llama_prompt += f"<|{role}|>
|
| 344 |
-
|
| 345 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 346 |
output = llm_fallback(
|
| 347 |
llama_prompt,
|
| 348 |
max_tokens=512,
|
| 349 |
stop=["</s>", "<|user|>"],
|
| 350 |
stream=True
|
| 351 |
)
|
| 352 |
-
|
| 353 |
for chunk in output:
|
| 354 |
token = chunk["choices"][0]["text"]
|
| 355 |
if token:
|
|
@@ -358,48 +505,101 @@ def chat():
|
|
| 358 |
"source": "hf",
|
| 359 |
"provider": "Nexa Local (HF Fallback)",
|
| 360 |
"tools_used": tools_needed
|
| 361 |
-
}) + "
|
|
|
|
| 362 |
except Exception as hf_err:
|
| 363 |
logger.error(f"HF Fallback failed: {hf_err}")
|
| 364 |
-
yield json.dumps({"error": "All backends failed"}) + "
|
|
|
|
| 365 |
else:
|
| 366 |
-
yield json.dumps({"error": "No backend available"}) + "
|
|
|
|
| 367 |
|
| 368 |
return Response(generate(), mimetype='application/x-ndjson')
|
| 369 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 370 |
@app.route('/api/status', methods=['GET'])
|
| 371 |
def status_check():
|
| 372 |
return jsonify({
|
| 373 |
"status": "online",
|
| 374 |
"backend": "Nexa AI Unified",
|
| 375 |
"timestamp": time.ctime(),
|
| 376 |
-
"fallback_loaded": llm_fallback is not None
|
|
|
|
| 377 |
})
|
| 378 |
|
| 379 |
-
@app.route('/api/search', methods=['POST'])
|
| 380 |
-
def search_api():
|
| 381 |
-
try:
|
| 382 |
-
data = request.json or {}
|
| 383 |
-
query = data.get("query", "").strip()
|
| 384 |
-
if not query:
|
| 385 |
-
return jsonify({"error": "No query provided"}), 400
|
| 386 |
-
|
| 387 |
-
search_data, error = perform_search(query)
|
| 388 |
-
if error and not search_data:
|
| 389 |
-
return jsonify({"response": error, "results": []})
|
| 390 |
-
if not search_data:
|
| 391 |
-
return jsonify({"response": "No results found.", "results": []})
|
| 392 |
-
|
| 393 |
-
return jsonify({
|
| 394 |
-
"response": format_results(search_data),
|
| 395 |
-
"results": search_data["results"],
|
| 396 |
-
"engine": search_data["engine"],
|
| 397 |
-
"status": "success"
|
| 398 |
-
})
|
| 399 |
-
except Exception as e:
|
| 400 |
-
logger.error(f"Search API crashed: {e}")
|
| 401 |
-
return jsonify({"error": "Search API internal error", "response": f"Internal error: {str(e)}"}), 500
|
| 402 |
-
|
| 403 |
@app.errorhandler(500)
|
| 404 |
def internal_error(error):
|
| 405 |
return jsonify({"error": "Internal Server Error", "message": "An unexpected error occurred on the server."}), 500
|
|
|
|
| 6 |
import io
|
| 7 |
import json
|
| 8 |
import requests
|
| 9 |
+
import urllib.parse
|
| 10 |
from flask import Flask, request, jsonify, Response
|
| 11 |
from flask_cors import CORS
|
| 12 |
from datetime import datetime
|
| 13 |
from dotenv import load_dotenv
|
| 14 |
from llama_cpp import Llama
|
| 15 |
from huggingface_hub import hf_hub_download
|
|
|
|
| 16 |
import pdfplumber
|
| 17 |
import docx
|
| 18 |
from PIL import Image
|
|
|
|
| 31 |
CORS(app)
|
| 32 |
|
| 33 |
# --- Configuration & Model Selection ---
|
| 34 |
+
OLLAMA_URL = "http://localhost:11434"
|
| 35 |
HF_FALLBACK_REPO = "TheBloke/TinyLlama-1.1B-Chat-v1.0-GGUF"
|
| 36 |
HF_FALLBACK_FILE = "tinyllama-1.1b-chat-v1.0.Q4_K_M.gguf"
|
| 37 |
|
| 38 |
+
# Load Fallback Local Model
|
| 39 |
llm_fallback = None
|
| 40 |
try:
|
| 41 |
logger.info("Loading Hugging Face fallback model...")
|
|
|
|
| 78 |
Current Date: {date}
|
| 79 |
"""
|
| 80 |
|
| 81 |
+
# ============ BULLETPROOF SEARCH SYSTEM ============
|
| 82 |
+
|
| 83 |
+
class SearchManager:
|
| 84 |
+
"""Multi-engine search with automatic fallback. Never fails completely."""
|
| 85 |
+
|
| 86 |
+
def __init__(self):
|
| 87 |
+
self.last_search_time = 0
|
| 88 |
+
self.min_delay = 2 # seconds between searches
|
| 89 |
+
self.user_agents = [
|
| 90 |
+
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.0",
|
| 91 |
+
"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.0",
|
| 92 |
+
"Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.0",
|
| 93 |
+
"Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:120.0) Gecko/20100101 Firefox/120.0"
|
| 94 |
+
]
|
| 95 |
+
|
| 96 |
+
def _get_headers(self):
|
| 97 |
+
return {"User-Agent": random.choice(self.user_agents)}
|
| 98 |
+
|
| 99 |
+
def _rate_limit(self):
|
| 100 |
+
"""Ensure minimum delay between searches to avoid blocking."""
|
| 101 |
+
elapsed = time.time() - self.last_search_time
|
| 102 |
+
if elapsed < self.min_delay:
|
| 103 |
+
time.sleep(self.min_delay - elapsed)
|
| 104 |
+
self.last_search_time = time.time()
|
| 105 |
+
|
| 106 |
+
def search_duckduckgo(self, query: str, max_results: int = 5) -> list:
|
| 107 |
+
"""Primary search: DuckDuckGo (fastest, most reliable)."""
|
| 108 |
+
try:
|
| 109 |
+
self._rate_limit()
|
| 110 |
+
from duckduckgo_search import DDGS
|
| 111 |
+
|
| 112 |
+
# Try with custom headers and timeout
|
| 113 |
+
with DDGS(headers=self._get_headers(), timeout=15) as ddgs:
|
| 114 |
+
results = list(ddgs.text(query, max_results=max_results))
|
| 115 |
+
if results:
|
| 116 |
+
logger.info(f"โ
DuckDuckGo: {len(results)} results")
|
| 117 |
+
return [{"title": r["title"], "href": r["href"], "body": r["body"]}
|
| 118 |
+
for r in results]
|
| 119 |
+
except Exception as e:
|
| 120 |
+
logger.warning(f"DuckDuckGo failed: {e}")
|
| 121 |
+
return []
|
| 122 |
+
|
| 123 |
+
def search_bing(self, query: str, max_results: int = 5) -> list:
|
| 124 |
+
"""Fallback 1: Bing HTML scraping."""
|
| 125 |
+
try:
|
| 126 |
+
self._rate_limit()
|
| 127 |
+
encoded_query = urllib.parse.quote(query)
|
| 128 |
+
url = f"https://www.bing.com/search?q={encoded_query}"
|
| 129 |
+
|
| 130 |
+
response = requests.get(url, headers=self._get_headers(), timeout=10)
|
| 131 |
+
response.raise_for_status()
|
| 132 |
+
|
| 133 |
+
# Extract results using regex patterns
|
| 134 |
+
html = response.text
|
| 135 |
+
results = []
|
| 136 |
+
|
| 137 |
+
# Pattern 1: Standard Bing results
|
| 138 |
+
import re
|
| 139 |
+
|
| 140 |
+
# Find all result blocks
|
| 141 |
+
result_blocks = re.findall(
|
| 142 |
+
r'<li class="b_algo"[^>]*>(.*?)</li>',
|
| 143 |
+
html,
|
| 144 |
+
re.DOTALL
|
| 145 |
+
)
|
| 146 |
+
|
| 147 |
+
for block in result_blocks[:max_results]:
|
| 148 |
+
# Extract title and URL
|
| 149 |
+
title_match = re.search(r'<a[^>]*href="([^"]*)"[^>]*>(.*?)</a>', block, re.DOTALL)
|
| 150 |
+
if title_match:
|
| 151 |
+
href = title_match.group(1)
|
| 152 |
+
title = re.sub(r'<[^>]+>', '', title_match.group(2)).strip()
|
| 153 |
+
|
| 154 |
+
# Extract snippet
|
| 155 |
+
snippet_match = re.search(r'<p[^>]*>(.*?)</p>', block, re.DOTALL)
|
| 156 |
+
body = re.sub(r'<[^>]+>', '', snippet_match.group(1)).strip() if snippet_match else ""
|
| 157 |
+
|
| 158 |
+
# Clean up URL
|
| 159 |
+
if href.startswith('/'):
|
| 160 |
+
href = 'https://www.bing.com' + href
|
| 161 |
+
|
| 162 |
+
results.append({"title": title, "href": href, "body": body})
|
| 163 |
+
|
| 164 |
+
if results:
|
| 165 |
+
logger.info(f"โ
Bing: {len(results)} results")
|
| 166 |
+
return results
|
| 167 |
+
|
| 168 |
+
except Exception as e:
|
| 169 |
+
logger.warning(f"Bing failed: {e}")
|
| 170 |
+
return []
|
| 171 |
+
|
| 172 |
+
def search_wikipedia(self, query: str, max_results: int = 3) -> list:
|
| 173 |
+
"""Fallback 2: Wikipedia API (never blocks, always free)."""
|
| 174 |
+
try:
|
| 175 |
+
self._rate_limit()
|
| 176 |
+
url = "https://en.wikipedia.org/w/api.php"
|
| 177 |
+
params = {
|
| 178 |
+
"action": "query",
|
| 179 |
+
"list": "search",
|
| 180 |
+
"srsearch": query,
|
| 181 |
+
"format": "json",
|
| 182 |
+
"srlimit": max_results
|
| 183 |
+
}
|
| 184 |
+
response = requests.get(url, params=params, headers=self._get_headers(), timeout=10)
|
| 185 |
+
data = response.json()
|
| 186 |
+
|
| 187 |
+
results = []
|
| 188 |
+
for item in data.get("query", {}).get("search", []):
|
| 189 |
+
title = item["title"]
|
| 190 |
+
page_url = f"https://en.wikipedia.org/wiki/{title.replace(' ', '_')}"
|
| 191 |
+
snippet = re.sub(r'<span class="searchmatch">|</span>', '', item["snippet"])
|
| 192 |
+
results.append({"title": title, "href": page_url, "body": snippet})
|
| 193 |
+
|
| 194 |
+
if results:
|
| 195 |
+
logger.info(f"โ
Wikipedia: {len(results)} results")
|
| 196 |
+
return results
|
| 197 |
+
|
| 198 |
+
except Exception as e:
|
| 199 |
+
logger.warning(f"Wikipedia failed: {e}")
|
| 200 |
+
return []
|
| 201 |
+
|
| 202 |
+
def search(self, query: str, max_results: int = 5) -> dict:
|
| 203 |
+
"""
|
| 204 |
+
Master search function. Tries engines in order until one succeeds.
|
| 205 |
+
Returns: {"engine": str, "results": list, "success": bool}
|
| 206 |
+
"""
|
| 207 |
+
if not query or not query.strip():
|
| 208 |
+
return {"engine": "None", "results": [], "success": False, "error": "Empty query"}
|
| 209 |
+
|
| 210 |
+
logger.info(f"๐ Searching: {query}")
|
| 211 |
+
|
| 212 |
+
# Engine 1: DuckDuckGo
|
| 213 |
+
results = self.search_duckduckgo(query, max_results)
|
| 214 |
+
if results:
|
| 215 |
+
return {"engine": "DuckDuckGo", "results": results, "success": True, "error": None}
|
| 216 |
+
|
| 217 |
+
# Engine 2: Bing
|
| 218 |
+
results = self.search_bing(query, max_results)
|
| 219 |
+
if results:
|
| 220 |
+
return {"engine": "Bing", "results": results, "success": True, "error": None}
|
| 221 |
+
|
| 222 |
+
# Engine 3: Wikipedia
|
| 223 |
+
results = self.search_wikipedia(query, max_results)
|
| 224 |
+
if results:
|
| 225 |
+
return {"engine": "Wikipedia", "results": results, "success": True, "error": None}
|
| 226 |
+
|
| 227 |
+
# All failed
|
| 228 |
+
logger.error("โ All search engines failed")
|
| 229 |
+
return {
|
| 230 |
+
"engine": "None",
|
| 231 |
+
"results": [],
|
| 232 |
+
"success": False,
|
| 233 |
+
"error": "All search engines are temporarily unavailable. Please try again in a few minutes."
|
| 234 |
+
}
|
| 235 |
+
|
| 236 |
+
# Initialize search manager
|
| 237 |
+
search_manager = SearchManager()
|
| 238 |
+
|
| 239 |
+
# ============ TOOL DETECTION ============
|
| 240 |
+
|
| 241 |
def detect_tools(message):
|
| 242 |
+
"""Improved intent detection for tool routing."""
|
| 243 |
+
message = message.lower().strip()
|
|
|
|
|
|
|
| 244 |
tools = []
|
| 245 |
+
|
| 246 |
+
# Search Intent - comprehensive keyword list
|
| 247 |
search_keywords = [
|
| 248 |
"search", "latest", "news", "find out", "who is", "what is the current",
|
| 249 |
"research", "browse", "today", "current", "facts", "data", "tell me about",
|
| 250 |
+
"what happened", "weather", "stock", "price", "how many", "who won",
|
| 251 |
+
"temperature", "forecast", "update", "recent", "now", "happening",
|
| 252 |
+
"look up", "check", "get info", "information about", "details on",
|
| 253 |
+
"what's", "whats", "how's", "hows", "where is", "when did", "why did",
|
| 254 |
+
"did ", "does ", "is ", "are ", "was ", "were ", "will ", "can ", "could ",
|
| 255 |
+
"2024", "2025", "2026", "yesterday", "tomorrow", "this week", "this month"
|
| 256 |
]
|
| 257 |
+
|
| 258 |
+
# Check for search keywords
|
| 259 |
if any(kw in message for kw in search_keywords):
|
| 260 |
tools.append("search")
|
| 261 |
+
|
| 262 |
+
# Question patterns that likely need search (who, what, when, where, why, how)
|
| 263 |
+
question_starters = ["who ", "what ", "when ", "where ", "why ", "how "]
|
| 264 |
+
if any(message.startswith(kw) for kw in question_starters):
|
| 265 |
+
# Only add if not already added
|
| 266 |
+
if "search" not in tools:
|
| 267 |
+
# Check if it's a factual question (not personal/opinion)
|
| 268 |
+
personal_words = ["you", "your", "we", "our", "i ", "my ", "me "]
|
| 269 |
+
if not any(pw in message for pw in personal_words):
|
| 270 |
+
tools.append("search")
|
| 271 |
+
|
| 272 |
# Image Intent
|
| 273 |
+
image_keywords = ["image", "draw", "visualize", "photo", "paint", "picture", "generate image"]
|
| 274 |
+
if any(kw in message for kw in image_keywords) or ("create" in message and "image" in message) or ("make" in message and "image" in message):
|
|
|
|
| 275 |
tools.append("image")
|
| 276 |
+
|
| 277 |
+
# File analysis intent
|
| 278 |
+
file_keywords = ["pdf", "document", "file", "analyze file", "read file", "upload"]
|
| 279 |
+
if any(kw in message for kw in file_keywords):
|
| 280 |
+
tools.append("file")
|
| 281 |
+
|
| 282 |
return tools
|
| 283 |
|
| 284 |
+
# ============ FORMATTING ============
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 285 |
|
| 286 |
+
def format_search_results(data: dict) -> str:
|
| 287 |
"""Format search results with markdown citations."""
|
| 288 |
+
if not data.get("success"):
|
| 289 |
+
error = data.get("error", "Search failed")
|
| 290 |
+
return f"โ ๏ธ {error}"
|
| 291 |
+
|
| 292 |
+
results = data.get("results", [])
|
| 293 |
+
if not results:
|
| 294 |
+
return "No results found for this query."
|
| 295 |
+
|
| 296 |
+
formatted = [f"๐ก Source: {data['engine']}
|
| 297 |
+
"]
|
| 298 |
for i, r in enumerate(results, 1):
|
| 299 |
title = r.get("title", "No title")
|
| 300 |
url = r.get("href", "#")
|
| 301 |
+
body = r.get("body", "No description")
|
| 302 |
+
# Clean up body text
|
| 303 |
+
body = re.sub(r'\s+', ' ', body).strip()[:300]
|
| 304 |
+
formatted.append(f"{i}. [{title}]({url})
|
| 305 |
+
{body}")
|
| 306 |
+
|
| 307 |
+
return "
|
| 308 |
+
|
| 309 |
+
".join(formatted)
|
| 310 |
+
|
| 311 |
+
def clean_search_query(query: str) -> str:
|
| 312 |
+
"""Clean and optimize query for better search results."""
|
| 313 |
+
# Remove common prefixes
|
| 314 |
+
prefixes = [
|
| 315 |
+
"search for", "find out about", "look up", "check",
|
| 316 |
+
"tell me about", "what is", "who is", "where is",
|
| 317 |
+
"/search", "search:", "find:"
|
| 318 |
+
]
|
| 319 |
+
query = query.lower().strip()
|
| 320 |
+
for prefix in prefixes:
|
| 321 |
+
if query.startswith(prefix):
|
| 322 |
+
query = query[len(prefix):].strip()
|
| 323 |
+
return query
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 324 |
|
| 325 |
+
# ============ API ROUTES ============
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 326 |
|
| 327 |
@app.route('/')
|
| 328 |
def home():
|
|
|
|
| 343 |
data = request.json or {}
|
| 344 |
return jsonify({"user": {"username": data.get("username", "Guest"), "messages_used": 0}}), 200
|
| 345 |
|
| 346 |
+
@app.route('/api/search', methods=['POST'])
|
| 347 |
+
def search_api():
|
| 348 |
+
"""Dedicated search endpoint with bulletproof fallback."""
|
| 349 |
+
try:
|
| 350 |
+
data = request.json or {}
|
| 351 |
+
query = data.get("query", "").strip()
|
| 352 |
+
max_results = data.get("max_results", 5)
|
| 353 |
+
|
| 354 |
+
if not query:
|
| 355 |
+
return jsonify({"error": "No query provided", "success": False}), 400
|
| 356 |
+
|
| 357 |
+
# Perform search with fallback
|
| 358 |
+
search_data = search_manager.search(query, max_results)
|
| 359 |
+
|
| 360 |
+
return jsonify({
|
| 361 |
+
"success": search_data["success"],
|
| 362 |
+
"query": query,
|
| 363 |
+
"engine": search_data["engine"],
|
| 364 |
+
"results": search_data["results"],
|
| 365 |
+
"formatted": format_search_results(search_data),
|
| 366 |
+
"result_count": len(search_data["results"])
|
| 367 |
+
})
|
| 368 |
+
|
| 369 |
+
except Exception as e:
|
| 370 |
+
logger.error(f"Search API error: {e}")
|
| 371 |
+
return jsonify({
|
| 372 |
+
"error": "Search service temporarily unavailable",
|
| 373 |
+
"success": False,
|
| 374 |
+
"query": data.get("query", "") if 'data' in dir() else ""
|
| 375 |
+
}), 500
|
| 376 |
+
|
| 377 |
@app.route('/api/chat', methods=['POST'])
|
| 378 |
def chat():
|
| 379 |
data = request.json or {}
|
|
|
|
| 382 |
memory = data.get("memory", {})
|
| 383 |
mode = data.get("mode", "instant")
|
| 384 |
is_deep_research = data.get("deep_research", False)
|
| 385 |
+
|
| 386 |
if not message:
|
| 387 |
return jsonify({"error": "No message"}), 400
|
| 388 |
|
|
|
|
| 394 |
# 2. Tool Detection
|
| 395 |
tools_needed = detect_tools(message)
|
| 396 |
tool_context = ""
|
| 397 |
+
|
| 398 |
if "search" in tools_needed:
|
| 399 |
+
search_query = clean_search_query(message)
|
| 400 |
+
search_data = search_manager.search(search_query, max_results=5)
|
| 401 |
+
|
| 402 |
+
if search_data["success"]:
|
| 403 |
+
tool_context += f"
|
| 404 |
+
|
| 405 |
+
[Web Search Results from {search_data['engine']}]:
|
| 406 |
+
{format_search_results(search_data)}"
|
| 407 |
else:
|
| 408 |
+
tool_context += f"
|
| 409 |
+
|
| 410 |
+
[Search Status]: {search_data.get('error', 'Search temporarily unavailable')}"
|
| 411 |
|
| 412 |
# 3. Dynamic Prompting
|
| 413 |
current_date = datetime.now().strftime("%Y-%m-%d")
|
|
|
|
| 418 |
"research": "Act as a professional researcher. Provide deep, structured answers with logical sections.",
|
| 419 |
"creative": "Act as a creative writer. Use evocative language and focus on ideation."
|
| 420 |
}
|
| 421 |
+
|
| 422 |
selected_mode_instruction = mode_instructions.get(mode, mode_instructions["instant"])
|
| 423 |
dynamic_system_prompt = SYSTEM_PROMPT.format(date=current_date)
|
| 424 |
+
dynamic_system_prompt += f"
|
| 425 |
+
[Active Mode]: {mode.upper()} - {selected_mode_instruction}"
|
| 426 |
+
|
| 427 |
if is_deep_research:
|
| 428 |
+
dynamic_system_prompt += "
|
| 429 |
+
[DEEP RESEARCH ACTIVE]: Use a multi-step reasoning approach. Analyze thoroughly."
|
| 430 |
|
| 431 |
if memory:
|
| 432 |
+
dynamic_system_prompt += f"
|
| 433 |
+
[User Memory]: {json.dumps(memory)}"
|
| 434 |
|
| 435 |
full_prompt = f"{message}{tool_context}"
|
| 436 |
|
|
|
|
| 444 |
ollama_role = "assistant" if role == "ai" else "user"
|
| 445 |
messages.append({"role": ollama_role, "content": h.get('content', '')})
|
| 446 |
messages.append({"role": "user", "content": full_prompt})
|
| 447 |
+
|
| 448 |
response = requests.post(
|
| 449 |
f"{OLLAMA_URL}/api/chat",
|
| 450 |
json={
|
|
|
|
| 456 |
timeout=5,
|
| 457 |
stream=True
|
| 458 |
)
|
| 459 |
+
|
| 460 |
if response.status_code == 200:
|
| 461 |
for line in response.iter_lines():
|
| 462 |
if line:
|
|
|
|
| 467 |
"source": "colab",
|
| 468 |
"provider": "Ollama (Phi-3)",
|
| 469 |
"tools_used": tools_needed
|
| 470 |
+
}) + "
|
| 471 |
+
"
|
| 472 |
return
|
| 473 |
|
| 474 |
except Exception as colab_err:
|
|
|
|
| 477 |
# Step B: Fallback to Local HF Model
|
| 478 |
if llm_fallback:
|
| 479 |
try:
|
| 480 |
+
llama_prompt = f"<|system|>
|
| 481 |
+
{dynamic_system_prompt}</s>
|
| 482 |
+
"
|
| 483 |
for h in history:
|
| 484 |
role = h.get('role', 'user')
|
| 485 |
+
llama_prompt += f"<|{role}|>
|
| 486 |
+
{h.get('content', '')}</s>
|
| 487 |
+
"
|
| 488 |
+
llama_prompt += f"<|user|>
|
| 489 |
+
{full_prompt}</s>
|
| 490 |
+
<|assistant|>
|
| 491 |
+
"
|
| 492 |
+
|
| 493 |
output = llm_fallback(
|
| 494 |
llama_prompt,
|
| 495 |
max_tokens=512,
|
| 496 |
stop=["</s>", "<|user|>"],
|
| 497 |
stream=True
|
| 498 |
)
|
| 499 |
+
|
| 500 |
for chunk in output:
|
| 501 |
token = chunk["choices"][0]["text"]
|
| 502 |
if token:
|
|
|
|
| 505 |
"source": "hf",
|
| 506 |
"provider": "Nexa Local (HF Fallback)",
|
| 507 |
"tools_used": tools_needed
|
| 508 |
+
}) + "
|
| 509 |
+
"
|
| 510 |
except Exception as hf_err:
|
| 511 |
logger.error(f"HF Fallback failed: {hf_err}")
|
| 512 |
+
yield json.dumps({"error": "All backends failed"}) + "
|
| 513 |
+
"
|
| 514 |
else:
|
| 515 |
+
yield json.dumps({"error": "No backend available"}) + "
|
| 516 |
+
"
|
| 517 |
|
| 518 |
return Response(generate(), mimetype='application/x-ndjson')
|
| 519 |
|
| 520 |
+
# --- Time Bot Logic ---
|
| 521 |
+
def get_time_response(message):
|
| 522 |
+
message = message.lower()
|
| 523 |
+
time_keywords = ["time", "date", "today", "now", "clock", "day"]
|
| 524 |
+
if any(re.search(rf"\b{word}\b", message) for word in time_keywords):
|
| 525 |
+
now = datetime.now()
|
| 526 |
+
return f"๐ The current time is **{now.strftime('%H:%M:%S')}**.
|
| 527 |
+
๐
Today is **{now.strftime('%A, %Y-%m-%d')}**."
|
| 528 |
+
return None
|
| 529 |
+
|
| 530 |
+
# --- Image Generation Feature ---
|
| 531 |
+
@app.route('/api/image', methods=['POST'])
|
| 532 |
+
def generate_image():
|
| 533 |
+
data = request.json or {}
|
| 534 |
+
prompt = data.get("prompt", "").strip()
|
| 535 |
+
if not prompt:
|
| 536 |
+
return jsonify({"error": "No prompt provided"}), 400
|
| 537 |
+
|
| 538 |
+
try:
|
| 539 |
+
seed = int(time.time())
|
| 540 |
+
width = data.get("width", 1024)
|
| 541 |
+
height = data.get("height", 1024)
|
| 542 |
+
image_url = f"https://image.pollinations.ai/prompt/{requests.utils.quote(prompt)}?width={width}&height={height}&seed={seed}&nologo=true"
|
| 543 |
+
|
| 544 |
+
return jsonify({
|
| 545 |
+
"status": "success",
|
| 546 |
+
"image_url": image_url,
|
| 547 |
+
"prompt": prompt
|
| 548 |
+
})
|
| 549 |
+
except Exception as e:
|
| 550 |
+
logger.error(f"Image generation failed: {e}")
|
| 551 |
+
return jsonify({"error": "Image generation failed"}), 500
|
| 552 |
+
|
| 553 |
+
# --- File Analysis Feature ---
|
| 554 |
+
@app.route('/api/analyze-file', methods=['POST'])
|
| 555 |
+
def analyze_file():
|
| 556 |
+
if 'file' not in request.files:
|
| 557 |
+
return jsonify({"error": "No file uploaded"}), 400
|
| 558 |
+
|
| 559 |
+
file = request.files['file']
|
| 560 |
+
filename = file.filename
|
| 561 |
+
ext = filename.split('.')[-1].lower()
|
| 562 |
+
|
| 563 |
+
try:
|
| 564 |
+
text_content = ""
|
| 565 |
+
if ext == 'pdf':
|
| 566 |
+
with pdfplumber.open(file) as pdf:
|
| 567 |
+
text_content = "\n".join([page.extract_text() for page in pdf.pages if page.extract_text()])
|
| 568 |
+
elif ext == 'docx':
|
| 569 |
+
doc = docx.Document(file)
|
| 570 |
+
text_content = "\n".join([para.text for para in doc.paragraphs])
|
| 571 |
+
elif ext == 'txt':
|
| 572 |
+
text_content = file.read().decode('utf-8', errors='ignore')
|
| 573 |
+
elif ext in ['png', 'jpg', 'jpeg', 'webp']:
|
| 574 |
+
img = Image.open(file)
|
| 575 |
+
text_content = pytesseract.image_to_string(img)
|
| 576 |
+
else:
|
| 577 |
+
return jsonify({"error": f"Unsupported file type: {ext}"}), 400
|
| 578 |
+
|
| 579 |
+
if not text_content.strip():
|
| 580 |
+
return jsonify({"error": "Could not extract text from the file."}), 400
|
| 581 |
+
|
| 582 |
+
truncated_text = text_content[:4000]
|
| 583 |
+
return jsonify({
|
| 584 |
+
"filename": filename,
|
| 585 |
+
"content": truncated_text,
|
| 586 |
+
"full_length": len(text_content),
|
| 587 |
+
"status": "success"
|
| 588 |
+
})
|
| 589 |
+
except Exception as e:
|
| 590 |
+
logger.error(f"File analysis failed: {e}")
|
| 591 |
+
return jsonify({"error": f"Failed to process file: {str(e)}"}), 500
|
| 592 |
+
|
| 593 |
@app.route('/api/status', methods=['GET'])
|
| 594 |
def status_check():
|
| 595 |
return jsonify({
|
| 596 |
"status": "online",
|
| 597 |
"backend": "Nexa AI Unified",
|
| 598 |
"timestamp": time.ctime(),
|
| 599 |
+
"fallback_loaded": llm_fallback is not None,
|
| 600 |
+
"search_engines": ["DuckDuckGo", "Bing", "Wikipedia"]
|
| 601 |
})
|
| 602 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 603 |
@app.errorhandler(500)
|
| 604 |
def internal_error(error):
|
| 605 |
return jsonify({"error": "Internal Server Error", "message": "An unexpected error occurred on the server."}), 500
|