Spaces:
Sleeping
Sleeping
File size: 7,816 Bytes
f05e8f9 |
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 |
from __future__ import annotations
import json
import os
import hashlib
from pathlib import Path
from typing import List
import chromadb
from chromadb.config import Settings
from sentence_transformers import SentenceTransformer
from src.utils.logger import get_logger
from config.settings import settings
logger = get_logger(__name__)
class ChromaVectorDBManager:
"""Corporate-friendly ChromaDB manager - completely offline."""
def __init__(self, model_name: str = None, db_path: str = None):
self.model_name = model_name or getattr(
settings, 'EMBEDDING_MODEL', 'sentence-transformers/all-MiniLM-L6-v2'
)
self.embedding_model = SentenceTransformer(self.model_name)
self.db_path = db_path or getattr(settings, 'CHROMADB_PATH', './chroma_db')
os.makedirs(self.db_path, exist_ok=True)
self.client = chromadb.PersistentClient(
path=self.db_path,
settings=Settings(
anonymized_telemetry=False,
allow_reset=True,
is_persistent=True
)
)
self.collection_name = getattr(settings, 'COLLECTION_NAME', 'rag_chunks')
self.collection = self._get_collection()
logger.info(f"ChromaDB initialized at: {self.db_path}")
def _get_collection(self):
"""Get or create collection without embedding function."""
try:
return self.client.get_collection(name=self.collection_name)
except Exception:
try:
self.client.delete_collection(name=self.collection_name)
except Exception:
pass
return self.client.create_collection(
name=self.collection_name,
metadata={"description": "RAG chunks"}
)
def generate_embeddings(self, texts: List[str]) -> List[List[float]]:
"""Generate embeddings using local sentence-transformers."""
embeddings = self.embedding_model.encode(
texts,
batch_size=32,
show_progress_bar=len(texts) > 100,
convert_to_tensor=False
)
return embeddings.tolist()
def add_chunks_to_db(self, chunks: list, source_file: str) -> bool:
"""Add chunks (list of dicts) to ChromaDB with manual embedding generation."""
if not chunks:
return True
texts, ids, metadatas = [], [], []
seen_hashes = set()
for chunk in chunks:
text = chunk.get("text", "").strip()
if not text:
continue
text_hash = hashlib.md5(text.encode()).hexdigest()
if text_hash in seen_hashes:
continue
seen_hashes.add(text_hash)
chunk_id = f"{source_file}_{chunk.get('chunk_id', 0)}_{text_hash[:8]}"
try:
if self.collection.get(ids=[chunk_id])['ids']:
continue
except Exception:
pass
texts.append(text)
ids.append(chunk_id)
metadata = {
"source_file": source_file,
"chunk_index": chunk.get("chunk_id", 0),
"char_length": len(text),
"text_hash": text_hash
}
metadatas.append(metadata)
if not texts:
return True
embeddings = self.generate_embeddings(texts)
self.collection.add(
embeddings=embeddings,
documents=texts,
metadatas=metadatas,
ids=ids
)
logger.info(f"Added {len(texts)} chunks from {source_file} to ChromaDB")
return True
def search_for_rag(
self,
query: str,
n_results: int = 5,
use_truncated: bool = False,
filter_128_context: bool = False
) -> List[dict]:
"""Search using manual query embedding generation - completely offline."""
query_embedding = self.generate_embeddings([query])[0]
results = self.collection.query(
query_embeddings=[query_embedding],
n_results=min(n_results * 2, 20),
include=["documents", "metadatas", "distances"]
)
search_results = []
for i, (doc, metadata, distance) in enumerate(zip(
results['documents'][0], results['metadatas'][0], results['distances'][0]
)):
if len(search_results) >= n_results:
break
similarity_score = 1 / (1 + distance)
result = {
"id": results['ids'][0][i],
"score": similarity_score,
"distance": distance,
"text": doc,
"source_file": metadata["source_file"],
"chunk_index": metadata["chunk_index"]
}
search_results.append(result)
return search_results
def reset_database(self):
"""Reset/delete existing collection."""
try:
self.client.delete_collection(name=self.collection_name)
self.collection = self._get_collection()
logger.info(f"Reset collection: {self.collection_name}")
return True
except Exception as e:
logger.error(f"Failed to reset database: {e}")
return False
def get_collection_stats(self) -> dict:
"""Get collection statistics."""
count = self.collection.count()
db_size_mb = 0
try:
for file_path in Path(self.db_path).rglob("*"):
if file_path.is_file():
db_size_mb += file_path.stat().st_size
db_size_mb /= (1024 * 1024)
except Exception:
db_size_mb = 0
return {
"total_chunks": count,
"collection_name": self.collection_name,
"embedding_model": self.model_name,
"db_path": self.db_path,
"db_size_mb": db_size_mb
}
def process_all_chunks(self, chunks_dir: str = None) -> bool:
"""Process all *_extracted.json files and build ChromaDB."""
if not chunks_dir:
chunks_dir = getattr(settings, 'PROCESSED_TEXT_DIR', './data/processed_text')
chunk_files = list(Path(chunks_dir).glob("*_extracted.json"))
logger.info(f"Found {len(chunk_files)} extracted JSON files to process")
total_processed = 0
for chunk_file in chunk_files:
try:
with open(chunk_file, "r", encoding="utf-8") as f:
data = json.load(f)
# Handle the actual structure of extracted JSON files
if isinstance(data, dict) and "initial_chunks" in data:
# New format: { "source_info": {...}, "initial_chunks": [...] }
chunks = data["initial_chunks"]
elif isinstance(data, list):
# Old format: list of chunks directly
chunks = data
else:
logger.warning(f"Unexpected format in {chunk_file.name}")
continue
if chunks and self.add_chunks_to_db(chunks, source_file=chunk_file.name):
total_processed += len(chunks)
logger.info(f"Processed {chunk_file.name}: {len(chunks)} chunks")
except Exception as e:
logger.error(f"Error processing {chunk_file}: {e}")
continue
logger.info(f"Successfully processed {total_processed} total chunks")
return total_processed > 0
|