Spaces:
Sleeping
Sleeping
File size: 17,709 Bytes
ce673e5 db8ee02 ce673e5 db8ee02 ce673e5 db8ee02 ce673e5 8ff58c6 ce673e5 8ff58c6 ce673e5 8ff58c6 fb45fc6 8ff58c6 fb45fc6 8ff58c6 db8ee02 fb45fc6 db8ee02 fb45fc6 db8ee02 8ff58c6 ce673e5 8ff58c6 fb45fc6 8ff58c6 fb45fc6 8ff58c6 fb45fc6 8ff58c6 ce673e5 fb45fc6 ce673e5 fb45fc6 70cc489 fb45fc6 ce673e5 fb45fc6 ce673e5 fb45fc6 ce673e5 fb45fc6 ce673e5 fb45fc6 ce673e5 fb45fc6 ce673e5 fb45fc6 ce673e5 fb45fc6 ce673e5 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 | """
RAG Chatbot Service for OpenTriage.
AI assistant trained on repo docs and closed issues to answer contributor questions.
"""
import logging
from typing import List, Dict, Any, Optional
from datetime import datetime, timezone
from pydantic import BaseModel
import time
from config.settings import settings
logger = logging.getLogger(__name__)
class RAGAnswer(BaseModel):
"""Response from the RAG chatbot."""
question: str
answer: str
sources: List[Dict[str, Any]] = [] # List of source documents used
confidence: float = 0.0
related_issues: List[Dict[str, Any]] = []
repo_name: Optional[str] = None
generated_at: datetime = None
class RAGChatbotService:
"""
RAG-powered Q&A chatbot service.
Uses vector similarity search on indexed documents
(issues, PRs, docs) to provide context-aware answers.
Note: For production, integrate with ChromaDB or Pinecone.
This implementation uses in-memory search as fallback.
"""
# Cache settings
README_CACHE_TTL = 600 # 10 minutes in seconds
def __init__(self):
self.use_vector_db = False # Set True when ChromaDB is available
self._embeddings_cache = {}
self._readme_cache = {} # {repo_key: {"content": str, "timestamp": float}}
def _get_cached_readme(self, repo_name: str) -> Optional[str]:
"""
Get cached README if it exists and hasn't expired.
Args:
repo_name: Repository name (owner/repo format)
Returns:
README content if cached and valid, None otherwise
"""
if repo_name not in self._readme_cache:
return None
cache_entry = self._readme_cache[repo_name]
age = time.time() - cache_entry["timestamp"]
if age > self.README_CACHE_TTL:
# Cache expired, remove it
del self._readme_cache[repo_name]
logger.info(f"README cache expired for {repo_name} (age: {age:.1f}s)")
return None
logger.info(f"✅ Serving README from cache for {repo_name} (age: {age:.1f}s)")
return cache_entry["content"]
def _cache_readme(self, repo_name: str, content: str) -> None:
"""
Cache README content with timestamp.
Args:
repo_name: Repository name (owner/repo format)
content: README content to cache
"""
self._readme_cache[repo_name] = {
"content": content,
"timestamp": time.time()
}
logger.info(f"📝 Cached README for {repo_name} ({len(content)} chars)")
async def answer_question(
self,
question: str,
repo_name: Optional[str] = None,
top_k: int = 5,
github_access_token: Optional[str] = None
) -> RAGAnswer:
"""
Answer a question using RAG.
Args:
question: The question to answer
repo_name: Optional repo context
top_k: Number of documents to retrieve
github_access_token: Optional GitHub token for README fetching
Returns:
RAGAnswer with the response and sources
"""
from config.database import db
import httpx
# Check if we have any indexed content for this repo
has_indexed_content = False
readme_content = None
if repo_name:
# Check for existing RAG chunks
try:
existing_chunks = await db.rag_chunks.count_documents({"sourceRepo": repo_name})
has_indexed_content = existing_chunks > 0
except:
has_indexed_content = False
# If no indexed content, try to fetch README directly from GitHub
if not has_indexed_content:
logger.info(f"No indexed content for {repo_name}, checking cache and fetching README if needed...")
# Stage 1: Check cache first
cached_readme = self._get_cached_readme(repo_name)
if cached_readme:
readme_content = cached_readme
else:
# Cache miss - fetch from GitHub
try:
owner, repo = repo_name.split('/')
url = f"https://raw.githubusercontent.com/{owner}/{repo}/main/README.md"
async with httpx.AsyncClient(timeout=10) as client:
response = await client.get(url)
if response.status_code == 200:
readme_content = response.text
# Cache the fetched content
self._cache_readme(repo_name, readme_content)
logger.info(f"Successfully fetched README for {repo_name} ({len(readme_content)} chars)")
else:
# Try master branch instead
url = f"https://raw.githubusercontent.com/{owner}/{repo}/master/README.md"
response = await client.get(url)
if response.status_code == 200:
readme_content = response.text
# Cache the fetched content
self._cache_readme(repo_name, readme_content)
logger.info(f"Successfully fetched README (master) for {repo_name} ({len(readme_content)} chars)")
else:
logger.warning(f"README not found at {url}")
except Exception as e:
logger.error(f"Error fetching README for {repo_name}: {e}")
# Search for relevant documents (from indexed chunks)
relevant_docs = await self.search_documents(question, repo_name, top_k)
# Build context from documents
context = self._build_context(relevant_docs)
# If we have a fresh README but no indexed content, prepend it to context
if readme_content and not has_indexed_content:
# Truncate README if too long (keep first 5000 chars for better context)
truncated_readme = readme_content[:5000] if len(readme_content) > 5000 else readme_content
context = f"[PROJECT README]\n{truncated_readme}\n\n---\n\n{context}"
# Add README to sources
relevant_docs.insert(0, {
"id": f"{repo_name}_readme_live",
"title": "Project README (Live from GitHub)",
"type": "readme",
"relevance": 1.0
})
logger.info(f"Added README to context for {repo_name}")
# Generate answer using AI
answer, confidence = await self._generate_answer(question, context, repo_name)
# Find related issues
related_issues = await self._find_related_issues(question, repo_name)
return RAGAnswer(
question=question,
answer=answer,
sources=[{
"id": doc.get("id", ""),
"title": doc.get("title", ""),
"type": doc.get("type", ""),
"relevance": doc.get("relevance", 0)
} for doc in relevant_docs],
confidence=confidence,
related_issues=related_issues,
repo_name=repo_name,
generated_at=datetime.now(timezone.utc)
)
async def search_documents(
self,
query: str,
repo_name: Optional[str] = None,
top_k: int = 5
) -> List[Dict[str, Any]]:
"""
Search for relevant documents.
Uses vector search if available, falls back to keyword search.
"""
from config.database import db
results = []
# First, try to find RAG chunks
chunk_query = {}
if repo_name:
chunk_query["sourceRepo"] = repo_name
# First, search specifically for README content if repo_name is provided
if repo_name:
readme_query = {
"sourceRepo": repo_name,
"documentType": "readme"
}
cursor = db.rag_chunks.find(readme_query, {"_id": 0}).sort("chunkIndex", 1).limit(3)
readme_chunks = await cursor.to_list(length=3)
for chunk in readme_chunks:
results.append({
"id": chunk.get("chunkId", ""),
"title": "Project README",
"content": chunk.get("content", ""),
"type": "readme",
"relevance": 1.0 # High relevance for README context
})
# Simple text search as fallback
if query:
chunk_query["$or"] = [
{"content": {"$regex": query, "$options": "i"}},
{"metadata.title": {"$regex": query, "$options": "i"}}
]
cursor = db.rag_chunks.find(chunk_query, {"_id": 0}).limit(top_k * 2)
chunks = await cursor.to_list(length=top_k * 2)
for chunk in chunks:
results.append({
"id": chunk.get("chunkId", ""),
"title": chunk.get("metadata", {}).get("title", "Document"),
"content": chunk.get("content", ""),
"type": chunk.get("documentType", ""),
"relevance": self._calculate_relevance(query, chunk.get("content", ""))
})
# Also search closed issues for answers
issue_query = {"state": "closed"}
if repo_name:
issue_query["repoName"] = repo_name
if query:
issue_query["$or"] = [
{"title": {"$regex": query, "$options": "i"}},
{"body": {"$regex": query, "$options": "i"}}
]
cursor = db.issues.find(issue_query, {"_id": 0}).limit(top_k)
issues = await cursor.to_list(length=top_k)
for issue in issues:
results.append({
"id": issue.get("id", ""),
"title": issue.get("title", ""),
"content": issue.get("body", ""),
"type": "closed_issue",
"relevance": self._calculate_relevance(query, f"{issue.get('title', '')} {issue.get('body', '')}")
})
# Sort by relevance and return top_k
results.sort(key=lambda x: x.get("relevance", 0), reverse=True)
return results[:top_k]
def _calculate_relevance(self, query: str, content: str) -> float:
"""Calculate simple relevance score based on keyword matching."""
if not query or not content:
return 0.0
query_terms = set(query.lower().split())
content_lower = content.lower()
matches = sum(1 for term in query_terms if term in content_lower)
if len(query_terms) == 0:
return 0.0
return matches / len(query_terms)
def _build_context(self, documents: List[Dict[str, Any]]) -> str:
"""Build context string from documents."""
if not documents:
return ""
context_parts = []
for i, doc in enumerate(documents[:5]):
title = doc.get("title", "Document")
content = doc.get("content", "")[:500] # Limit content
doc_type = doc.get("type", "document")
context_parts.append(f"[{doc_type.upper()}] {title}:\n{content}\n")
return "\n---\n".join(context_parts)
async def _generate_answer(
self,
question: str,
context: str,
repo_name: Optional[str]
) -> tuple[str, float]:
"""Generate answer using AI with context and model fallbacks."""
from openai import OpenAI
# Use the same fallback models as chat service
models = [
"google/gemini-2.0-flash-001",
"cohere/rerank-4-pro",
"arcee-ai/trinity-large-preview:free",
"liquid/lfm-2.5-1.2b-thinking:free",
]
client = OpenAI(
base_url="https://openrouter.ai/api/v1",
api_key=settings.OPENROUTER_API_KEY
)
system_prompt = f"""You are a knowledgeable guide for contributors to{f' the {repo_name} project' if repo_name else ' this open source project'}, acting much like a senior developer who has worked on the codebase for years.
When answering questions, draw from the provided documentation and issue history naturally. Present information as helpful conversation rather than recitation.
If the context provides information, use it to give a solid answer. If context is limited, acknowledge what you don't know while offering general guidance.
Provide clear, practical answers. Reference specific issues or documentation sections naturally."""
user_prompt = f"""Context from project:
{context}
---
Question: {question}
Provide a helpful answer based on the context above. If context is limited, be honest about it."""
# Try each model
for model in models:
try:
logger.info(f"RAG: Attempting answer generation with {model}")
response = client.chat.completions.create(
model=model,
messages=[
{"role": "system", "content": system_prompt},
{"role": "user", "content": user_prompt}
],
max_tokens=800,
temperature=0.3
)
answer = response.choices[0].message.content.strip()
confidence = 0.9 if context and len(context) > 500 else 0.5
logger.info(f"RAG: Successfully generated answer with {model}")
return answer, confidence
except Exception as e:
logger.warning(f"RAG: Model {model} failed: {e}")
continue
# All models failed
logger.error("RAG: All models failed for answer generation")
return (
"I apologize, I'm currently unable to generate an answer. Please try again in a moment.",
0.0
)
async def _find_related_issues(
self,
question: str,
repo_name: Optional[str],
limit: int = 3
) -> List[Dict[str, Any]]:
"""Find issues related to the question."""
from config.database import db
query = {}
if repo_name:
query["repoName"] = repo_name
if question:
query["$or"] = [
{"title": {"$regex": question.split()[0] if question.split() else "", "$options": "i"}},
]
cursor = db.issues.find(query, {"_id": 0}).limit(limit)
issues = await cursor.to_list(length=limit)
return [{
"id": issue.get("id"),
"number": issue.get("number"),
"title": issue.get("title"),
"state": issue.get("state"),
"url": issue.get("htmlUrl", "")
} for issue in issues]
async def index_repository(self, repo_name: str, github_access_token: Optional[str] = None) -> Dict[str, Any]:
"""
Index a repository's content for RAG.
Uses the existing rag_data_prep service.
"""
try:
from services.rag_data_prep import rag_data_prep
result = await rag_data_prep.prepare_and_store(
doc_types=["issue", "pr", "comment", "readme"],
repo_names=[repo_name],
collection_name="rag_chunks",
github_access_token=github_access_token
)
return {
"status": "success",
"repo_name": repo_name,
"documents_indexed": result.get("documents_processed", 0),
"chunks_created": result.get("chunks_created", 0)
}
except Exception as e:
logger.error(f"Indexing failed: {e}")
return {
"status": "error",
"error": str(e)
}
async def get_suggested_questions(
self,
repo_name: Optional[str] = None
) -> List[str]:
"""Get suggested questions based on common topics."""
from config.database import db
suggestions = [
"How do I get started contributing?",
"What is the development setup?",
"How do I run the tests?",
"What coding style should I follow?",
"How do I submit a pull request?"
]
# Add repo-specific suggestions based on common issues
if repo_name:
cursor = db.issues.find(
{"repoName": repo_name, "state": "closed"},
{"_id": 0, "title": 1}
).limit(5)
issues = await cursor.to_list(length=5)
for issue in issues:
title = issue.get("title", "")
if "?" in title:
suggestions.append(title)
return suggestions[:8]
# Singleton instance
rag_chatbot_service = RAGChatbotService()
|