rag-backend-api / vector_store.py
seraj-b2
runtime err fix
03c219f
Raw History Blame Contribute Delete
5.91 kB
import os
import uuid
import logging
from typing import List, Dict, Any, Optional
from qdrant_client import QdrantClient
from qdrant_client.http import models
from qdrant_client.http.exceptions import UnexpectedResponse
logger = logging.getLogger("rag_backend")
class QdrantVectorStore:
def __init__(self):
self.url = os.getenv("QDRANT_URL")
self.api_key = os.getenv("QDRANT_API_KEY")
self.collection_name = os.getenv("QDRANT_COLLECTION", "rag_documents")
self.vector_size = 384 # Default dimension for all-MiniLM-L6-v2
# Connect to Qdrant Cloud if credentials are provided, otherwise fallback to in-memory
if self.url and self.api_key:
url = self.url
# Qdrant Cloud URLs typically require port 443 for HTTPS REST APIs.
# If the user provided a cloud URL without specifying any port, default to 443.
clean_url = url.replace("https://", "").replace("http://", "")
if "qdrant.io" in url.lower() and ":" not in clean_url:
url = f"{url.rstrip('/')}:443"
logger.info(f"Detected Qdrant Cloud URL without port. Appending secure port 443: {url}")
else:
logger.info(f"Connecting to Qdrant Cloud at {url}...")
self.client = QdrantClient(
url=url,
api_key=self.api_key
)
self.is_cloud = True
else:
logger.warning("QDRANT_URL and QDRANT_API_KEY not found in environment variables. Falling back to local In-Memory Qdrant...")
self.client = QdrantClient(location=":memory:")
self.is_cloud = False
self._ensure_collection()
def _ensure_collection(self):
"""Ensures that the target collection exists in Qdrant."""
try:
# Check if collection exists by getting its info
self.client.get_collection(self.collection_name)
logger.info(f"Collection '{self.collection_name}' already exists.")
except (UnexpectedResponse, Exception):
# Collection does not exist or error, create it
logger.info(f"Collection '{self.collection_name}' not found. Creating a new one...")
self.client.create_collection(
collection_name=self.collection_name,
vectors_config=models.VectorParams(
size=self.vector_size,
distance=models.Distance.COSINE
)
)
logger.info(f"Collection '{self.collection_name}' created successfully.")
def add_documents(self, texts: List[str], embeddings: List[List[float]], metadatas: Optional[List[Dict[str, Any]]] = None) -> int:
"""
Upserts document chunks and their corresponding embeddings and payloads into Qdrant.
"""
if not texts:
return 0
if metadatas is None:
metadatas = [{} for _ in range(len(texts))]
points = []
for i, (text, embedding, metadata) in enumerate(zip(texts, embeddings, metadatas)):
point_id = str(uuid.uuid4())
payload = {
"text": text,
"metadata": metadata
}
points.append(
models.PointStruct(
id=point_id,
vector=embedding,
payload=payload
)
)
# Qdrant client upsert in batches
self.client.upsert(
collection_name=self.collection_name,
wait=True,
points=points
)
return len(points)
def search(self, query_embedding: List[float], top_k: int = 3) -> List[Dict[str, Any]]:
"""
Searches the Qdrant collection for similar embeddings.
Returns a list of dicts with text, metadata, and score.
"""
results = self.client.query_points(
collection_name=self.collection_name,
query=query_embedding,
limit=top_k
)
search_results = []
for hit in results.points:
search_results.append({
"text": hit.payload.get("text", ""),
"metadata": hit.payload.get("metadata", {}),
"score": hit.score
})
return search_results
def clear(self) -> bool:
"""
Deletes and recreates the collection to reset the database.
"""
try:
self.client.delete_collection(self.collection_name)
self.client.create_collection(
collection_name=self.collection_name,
vectors_config=models.VectorParams(
size=self.vector_size,
distance=models.Distance.COSINE
)
)
logger.info(f"Collection '{self.collection_name}' has been reset.")
return True
except Exception as e:
logger.error(f"Error resetting collection '{self.collection_name}': {e}")
return False
def get_status(self) -> Dict[str, Any]:
"""
Returns connection and collection stats.
"""
try:
collection_info = self.client.get_collection(self.collection_name)
vectors_count = collection_info.points_count
status = collection_info.status.value if hasattr(collection_info.status, "value") else str(collection_info.status)
except Exception as e:
logger.error(f"Error retrieving collection info: {e}")
vectors_count = 0
status = "unknown"
return {
"connection_type": "Cloud" if self.is_cloud else "In-Memory (Local)",
"collection_name": self.collection_name,
"vector_size": self.vector_size,
"vectors_count": vectors_count,
"status": status
}