Spaces:
Sleeping
Sleeping
File size: 3,459 Bytes
402298d cd1c169 402298d 74b575c 402298d 08d20f8 a123e22 402298d 819f7c5 a123e22 819f7c5 a123e22 74b575c 402298d a123e22 402298d 08d20f8 92f791e 08d20f8 402298d | 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 | """Retrieval service for semantic search"""
from typing import List, Dict, Any
from app.services.embeddings import embedding_service
from app.services.vector_store import vector_store
from app.config import settings
from app.utils.logger import setup_logger
import numpy as np
logger = setup_logger(__name__)
class RetrieverService:
"""Handles semantic search over vector database"""
def __init__(self):
self.embedding_service = embedding_service
self.vector_store = vector_store
def retrieve(self, query: str, top_k: int = None) -> List[Dict[str, Any]]:
"""Retrieve relevant documents for a query"""
logger.debug(f"top_k: {top_k}")
logger.debug(f"User Query: {query}")
if top_k is None:
top_k = settings.TOP_K
# Generate query embedding
logger.info(f"[RETRIEVER] Retrieving documents for query: {query}")
query_embedding = self.embedding_service.embed_text(query,is_query=True)
#logger.debug(f"Embedded query: {query_embedding}")
#FAISS
results = self.vector_store.search(
query_vector=query_embedding,
limit=top_k,
score_threshold=settings.SCORE_THRESHOLD
)
'''
try:
logger.warning(f"FAISS index object: {self.vector_store.index}")
if self.vector_store.index is None:
logger.warning("❌ FAISS index is None")
else:
logger.warning(f"FAISS total vectors: {self.vector_store.index.ntotal}")
D, I = self.vector_store.index.search(
np.array([query_embedding]).astype("float32"), k=top_k
)
logger.warning(f"Distances: {D}, Indices: {I}")
except Exception as e:
import traceback
logger.error(f"FAISS search error: {e}\n{traceback.format_exc()}")
'''
#Qdrant
# Search vector database
# results = self.vector_store.search(
# query_vector=query_embedding,
# limit=top_k,
# score_threshold=settings.SCORE_THRESHOLD
# )
logger.info(f"[RETRIEVER] Retrieved {len(results)} documents")
if results:
logger.debug("[RETRIEVER] Raw FAISS top-5 scores: " + ", ".join(f"{r['score']:.4f}" for r in results[:5]))
return results
def format_context(self, results: List[Dict[str, Any]]) -> str:
"""Format retrieved documents into context string"""
context_parts = []
for idx, result in enumerate(results, 1):
payload = result['payload']
score = result['score']
context_parts.append(f"[Document {idx}] (Relevance: {score:.2f})")
context_parts.append(f"Ticket: {payload.get('ticket_id', 'N/A')}")
context_parts.append(f"Project: {payload.get('project', 'N/A')}")
context_parts.append(f"Status: {payload.get('status', 'N/A')}")
context_parts.append(f"Priority: {payload.get('priority', 'N/A')}")
context_parts.append(f"Summary: {payload.get('summary', 'N/A')}")
if payload.get('description'):
context_parts.append(f"Description: {payload['description'][:200]}...")
context_parts.append("")
return "\n".join(context_parts)
# Global instance
retriever = RetrieverService() |