Spaces:
Sleeping
Sleeping
File size: 10,483 Bytes
d0f35dc |
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 |
from typing import List, Optional, Dict, Any
from langchain_classic.schema import Document
from langchain_google_genai import GoogleGenerativeAIEmbeddings
from langchain_qdrant import QdrantVectorStore
from qdrant_client import QdrantClient
from qdrant_client.models import Distance, VectorParams, PointStruct
from config import Config
import uuid
class VectorStoreManager:
"""Manages Qdrant vector store operations for insurance documents"""
def __init__(self):
"""Initialize Qdrant client and embeddings"""
# Validate configuration
Config.validate_config()
# Get configuration
self.qdrant_config = Config.get_qdrant_config()
self.retrieval_config = Config.get_retrieval_config()
# Initialize Qdrant client
self.client = QdrantClient(
url=self.qdrant_config["url"],
api_key=self.qdrant_config["api_key"],
)
# Initialize embeddings
self.embeddings = GoogleGenerativeAIEmbeddings(
model=Config.EMBEDDING_MODEL,
output_dimensionality=Config.EMBEDDING_DIMENSION,
google_api_key=Config.GEMINI_API_KEY
)
self.collection_name = self.qdrant_config["collection_name"]
print("Vector store manager initialized")
def create_collection(self, recreate: bool = False) -> bool:
"""
Create a new collection in Qdrant
Args:
recreate: If True, delete existing collection and create new one
Returns:
Boolean indicating success
"""
try:
# Check if collection exists
collections = self.client.get_collections().collections
collection_exists = any(c.name == self.collection_name for c in collections)
if collection_exists:
if recreate:
print(f"⚠ Deleting existing collection: {self.collection_name}")
self.client.delete_collection(self.collection_name)
else:
print(f" Collection '{self.collection_name}' already exists")
return True
# Create new collection
self.client.create_collection(
collection_name=self.collection_name,
vectors_config=VectorParams(
size=self.qdrant_config["vector_size"],
distance=Distance.COSINE
)
)
print(f" Created collection: {self.collection_name}")
return True
except Exception as e:
print(f" Error creating collection: {str(e)}")
raise
def add_documents(self, documents: List[Document], batch_size: int = 100) -> List[str]:
"""
Add documents to Qdrant vector store
Args:
documents: List of Document objects to add
batch_size: Number of documents to process in each batch
Returns:
List of document IDs
"""
try:
print(f"Adding {len(documents)} documents to vector store...")
# Ensure collection exists
self.create_collection(recreate=False)
# Initialize vector store
vector_store = QdrantVectorStore(
client=self.client,
collection_name=self.collection_name,
embedding=self.embeddings
)
# Add documents in batches
all_ids = []
for i in range(0, len(documents), batch_size):
batch = documents[i:i + batch_size]
# Generate unique IDs for this batch
batch_ids = [str(uuid.uuid4()) for _ in batch]
# Add to vector store
vector_store.add_documents(documents=batch, ids=batch_ids)
all_ids.extend(batch_ids)
print(f" Processed batch {i//batch_size + 1}/{(len(documents)-1)//batch_size + 1}")
print(f" Successfully added {len(documents)} documents")
return all_ids
except Exception as e:
print(f" Error adding documents: {str(e)}")
raise
def similarity_search(
self,
query: str,
k: Optional[int] = None,
filter_dict: Optional[Dict[str, Any]] = None
) -> List[Document]:
"""
Search for similar documents using semantic similarity
Args:
query: Search query string
k: Number of results to return (default from config)
filter_dict: Optional metadata filters (e.g., {"section_type": "exclusions"})
Returns:
List of most similar Documents
"""
try:
if k is None:
k = self.retrieval_config["top_k"]
# Initialize vector store for querying
vector_store = QdrantVectorStore(
client=self.client,
collection_name=self.collection_name,
embedding=self.embeddings
)
if filter_dict:
# Get more results than needed
results = vector_store.similarity_search(query=query, k=k*3)
# Filter by metadata
filtered_results = []
for doc in results:
match = True
for key, value in filter_dict.items():
if doc.metadata.get(key) != value:
match = False
break
if match:
filtered_results.append(doc)
# Stop when we have enough results
if len(filtered_results) >= k:
break
return filtered_results[:k]
else:
results = vector_store.similarity_search(query=query, k=k)
return results
except Exception as e:
print(f" Error during similarity search: {str(e)}")
raise
def similarity_search_with_score(
self,
query: str,
k: Optional[int] = None,
score_threshold: Optional[float] = None
) -> List[tuple[Document, float]]:
"""
Search with similarity scores
Args:
query: Search query string
k: Number of results to return
score_threshold: Minimum similarity score (default from config)
Returns:
List of (Document, score) tuples
"""
try:
if k is None:
k = self.retrieval_config["top_k"]
if score_threshold is None:
score_threshold = self.retrieval_config["similarity_threshold"]
# Initialize vector store
vector_store = QdrantVectorStore(
client=self.client,
collection_name=self.collection_name,
embedding=self.embeddings
)
# Search with scores
results = vector_store.similarity_search_with_score(query=query, k=k)
# Filter by score threshold
filtered_results = [
(doc, score) for doc, score in results
if score >= score_threshold
]
print(f" Found {len(filtered_results)} results above threshold {score_threshold}")
return filtered_results
except Exception as e:
print(f" Error during similarity search with score: {str(e)}")
raise
def search_by_section_type(
self,
query: str,
section_type: str,
k: Optional[int] = None
) -> List[Document]:
"""
Search within a specific section type (e.g., 'exclusions', 'addons')
Args:
query: Search query string
section_type: Type of section to search in
k: Number of results to return
Returns:
List of Documents from specified section type
"""
filter_dict = {"section_type": section_type}
return self.similarity_search(query=query, k=k, filter_dict=filter_dict)
def get_collection_info(self) -> Dict:
"""
Get information about the current collection
Returns:
Dictionary with collection statistics
"""
try:
collection_info = self.client.get_collection(self.collection_name)
return {
"name": self.collection_name,
"vectors_count": collection_info.vectors_count,
"points_count": collection_info.points_count,
"status": collection_info.status,
}
except Exception as e:
print(f" Error getting collection info: {str(e)}")
return {}
def delete_collection(self) -> bool:
"""
Delete the current collection
Returns:
Boolean indicating success
"""
try:
self.client.delete_collection(self.collection_name)
print(f" Deleted collection: {self.collection_name}")
return True
except Exception as e:
print(f" Error deleting collection: {str(e)}")
return False
def get_retriever(self, **kwargs):
"""
Get a LangChain retriever object for use in chains
Args:
**kwargs: Additional arguments for retriever configuration
Returns:
VectorStoreRetriever object
"""
vector_store = QdrantVectorStore(
client=self.client,
collection_name=self.collection_name,
embedding=self.embeddings
)
# Set default search kwargs
search_kwargs = {
"k": self.retrieval_config["top_k"]
}
search_kwargs.update(kwargs)
return vector_store.as_retriever(search_kwargs=search_kwargs)
|