Spaces:
Sleeping
Sleeping
File size: 11,208 Bytes
729a1f7 20fd4b7 0fee802 729a1f7 20fd4b7 729a1f7 20fd4b7 729a1f7 20fd4b7 729a1f7 20fd4b7 729a1f7 20fd4b7 729a1f7 18b46d7 729a1f7 20fd4b7 18b46d7 729a1f7 20fd4b7 18b46d7 20fd4b7 18b46d7 729a1f7 c48df01 729a1f7 c48df01 729a1f7 20fd4b7 729a1f7 20fd4b7 729a1f7 |
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 |
# ────────────────────────────── utils/rag.py ──────────────────────────────
import os
import math
from typing import List, Dict, Any, Optional
from pymongo import MongoClient, ASCENDING, TEXT
from pymongo.collection import Collection
from pymongo.errors import PyMongoError
import numpy as np
import logging
from ..logger import get_logger
VECTOR_DIM = 384 # all-MiniLM-L6-v2
INDEX_NAME = os.getenv("MONGO_VECTOR_INDEX", "vector_index")
USE_ATLAS_VECTOR = os.getenv("ATLAS_VECTOR", "0") == "1"
logger = get_logger("RAG", __name__)
class RAGStore:
def __init__(self, mongo_uri: str, db_name: str = "studybuddy"):
self.client = MongoClient(mongo_uri)
self.db = self.client[db_name]
self.chunks: Collection = self.db["chunks"]
self.files: Collection = self.db["files"]
# ── Write ────────────────────────────────────────────────────────────────
def store_cards(self, cards: List[Dict[str, Any]]):
if not cards:
return
for c in cards:
# basic validation
emb = c.get("embedding")
if not emb or len(emb) != VECTOR_DIM:
raise ValueError("Invalid embedding length; expected %d" % VECTOR_DIM)
self.chunks.insert_many(cards, ordered=False)
logger.info(f"Inserted {len(cards)} cards into MongoDB")
def upsert_file_summary(self, user_id: str, project_id: str, filename: str, summary: str):
self.files.update_one(
{"user_id": user_id, "project_id": project_id, "filename": filename},
{"$set": {"summary": summary}},
upsert=True
)
logger.info(f"Upserted summary for {filename} (user {user_id}, project {project_id})")
# ── Read ────────────────────────────────────────────────────────────────
def list_cards(self, user_id: str, project_id: str, filename: Optional[str], limit: int, skip: int):
q = {"user_id": user_id, "project_id": project_id}
if filename:
q["filename"] = filename
cur = self.chunks.find(q, {"embedding": 0}).skip(skip).limit(limit).sort([("_id", ASCENDING)])
# Convert MongoDB documents to JSON-serializable format
cards = []
for card in cur:
serializable_card = {}
for key, value in card.items():
if key == '_id':
serializable_card[key] = str(value) # Convert ObjectId to string
elif hasattr(value, 'isoformat'): # Handle datetime objects
serializable_card[key] = value.isoformat()
else:
serializable_card[key] = value
cards.append(serializable_card)
return cards
def get_file_summary(self, user_id: str, project_id: str, filename: str):
doc = self.files.find_one({"user_id": user_id, "project_id": project_id, "filename": filename})
if doc:
# Convert MongoDB document to JSON-serializable format
serializable_doc = {}
for key, value in doc.items():
if key == '_id':
serializable_doc[key] = str(value) # Convert ObjectId to string
elif hasattr(value, 'isoformat'): # Handle datetime objects
serializable_doc[key] = value.isoformat()
else:
serializable_doc[key] = value
return serializable_doc
return None
def list_files(self, user_id: str, project_id: str):
"""List all files for a project with their summaries"""
files_cursor = self.files.find(
{"user_id": user_id, "project_id": project_id},
{"_id": 0, "filename": 1, "summary": 1}
).sort("filename", ASCENDING)
# Convert MongoDB documents to JSON-serializable format
files = []
for file_doc in files_cursor:
serializable_file = {}
for key, value in file_doc.items():
if hasattr(value, 'isoformat'): # Handle datetime objects
serializable_file[key] = value.isoformat()
else:
serializable_file[key] = value
files.append(serializable_file)
return files
def vector_search(self, user_id: str, project_id: str, query_vector: List[float], k: int = 6, filenames: Optional[List[str]] = None, search_type: str = "hybrid"):
"""
Enhanced vector search with multiple strategies:
- hybrid: Combines Atlas and local search
- flat: Exhaustive search for maximum accuracy
- atlas: Uses Atlas Vector Search only
- local: Uses local cosine similarity only
"""
if search_type == "flat" or (search_type == "hybrid" and not USE_ATLAS_VECTOR):
return self._flat_vector_search(user_id, project_id, query_vector, k, filenames)
elif search_type == "atlas" and USE_ATLAS_VECTOR:
return self._atlas_vector_search(user_id, project_id, query_vector, k, filenames)
elif search_type == "local":
return self._local_vector_search(user_id, project_id, query_vector, k, filenames)
else:
# Default hybrid approach
if USE_ATLAS_VECTOR:
atlas_results = self._atlas_vector_search(user_id, project_id, query_vector, k, filenames)
if atlas_results:
return atlas_results
return self._local_vector_search(user_id, project_id, query_vector, k, filenames)
def _atlas_vector_search(self, user_id: str, project_id: str, query_vector: List[float], k: int, filenames: Optional[List[str]] = None):
"""Atlas Vector Search implementation"""
match_stage = {"user_id": user_id, "project_id": project_id}
if filenames:
match_stage["filename"] = {"$in": filenames}
pipeline = [
{
"$search": {
"index": INDEX_NAME,
"knnBeta": {
"vector": query_vector,
"path": "embedding",
"k": k,
}
}
},
{"$match": match_stage},
{"$project": {"doc": "$$ROOT", "score": {"$meta": "searchScore"}}},
{"$limit": k},
]
hits = list(self.chunks.aggregate(pipeline))
return self._serialize_hits(hits)
def _local_vector_search(self, user_id: str, project_id: str, query_vector: List[float], k: int, filenames: Optional[List[str]] = None):
"""Local cosine similarity search with improved sampling"""
q = {"user_id": user_id, "project_id": project_id}
if filenames:
q["filename"] = {"$in": filenames}
# Increase sample size for better accuracy
sample_limit = max(5000, k * 50)
sample = list(self.chunks.find(q).sort([("_id", -1)]).limit(sample_limit))
if not sample:
return []
qv = np.array(query_vector, dtype="float32")
scores = []
for d in sample:
v = np.array(d.get("embedding", [0]*VECTOR_DIM), dtype="float32")
denom = (np.linalg.norm(qv) * np.linalg.norm(v)) or 1.0
sim = float(np.dot(qv, v) / denom)
scores.append((sim, d))
scores.sort(key=lambda x: x[0], reverse=True)
top = scores[:k]
logger.info(f"Local vector search: {len(sample)} docs sampled, {len(top)} results")
return self._serialize_results(top)
def _flat_vector_search(self, user_id: str, project_id: str, query_vector: List[float], k: int, filenames: Optional[List[str]] = None):
"""Flat exhaustive search for maximum accuracy"""
q = {"user_id": user_id, "project_id": project_id}
if filenames:
q["filename"] = {"$in": filenames}
# Get ALL relevant documents for exhaustive search
all_docs = list(self.chunks.find(q))
if not all_docs:
return []
qv = np.array(query_vector, dtype="float32")
scores = []
for doc in all_docs:
v = np.array(doc.get("embedding", [0]*VECTOR_DIM), dtype="float32")
denom = (np.linalg.norm(qv) * np.linalg.norm(v)) or 1.0
sim = float(np.dot(qv, v) / denom)
scores.append((sim, doc))
scores.sort(key=lambda x: x[0], reverse=True)
top = scores[:k]
logger.info(f"Flat vector search: {len(all_docs)} docs searched, {len(top)} results")
return self._serialize_results(top)
def _serialize_hits(self, hits):
"""Serialize Atlas search hits"""
serializable_hits = []
for hit in hits:
doc = hit["doc"]
serializable_doc = self._serialize_doc(doc)
serializable_hits.append({
"doc": serializable_doc,
"score": float(hit.get("score", 0.0))
})
return serializable_hits
def _serialize_results(self, results):
"""Serialize local search results"""
serializable_results = []
for score, doc in results:
serializable_doc = self._serialize_doc(doc)
serializable_results.append({
"doc": serializable_doc,
"score": float(score)
})
return serializable_results
def _serialize_doc(self, doc):
"""Convert MongoDB document to JSON-serializable format"""
serializable_doc = {}
for key, value in doc.items():
if key == '_id':
serializable_doc[key] = str(value)
elif hasattr(value, 'isoformat'):
serializable_doc[key] = value.isoformat()
else:
serializable_doc[key] = value
return serializable_doc
def ensure_indexes(store: RAGStore):
# Basic text index for fallback keyword search (optional)
try:
store.chunks.create_index([("user_id", ASCENDING), ("project_id", ASCENDING), ("filename", ASCENDING)])
store.chunks.create_index([("content", TEXT), ("topic_name", TEXT), ("summary", TEXT)], name="text_idx")
store.files.create_index([("user_id", ASCENDING), ("project_id", ASCENDING), ("filename", ASCENDING)], unique=True)
except PyMongoError as e:
logger.warning(f"Index creation warning: {e}")
# Note: For Atlas Vector, create an Atlas Search index named INDEX_NAME on field "embedding" with vector options.
# Example (in Atlas UI):
# {
# "mappings": {
# "dynamic": false,
# "fields": {
# "embedding": {
# "type": "knnVector",
# "dimensions": 384,
# "similarity": "cosine"
# }
# }
# }
# } |