File size: 6,774 Bytes
5b2f824 |
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 |
"""
Custom FAISS vectorstore to replace langchain FAISS.
"""
import os
import pickle
import tempfile
import zipfile
from pathlib import Path
from typing import List, Tuple, Optional, Dict, Any
import numpy as np
import faiss
from uuid import uuid4
from tools.document import Document
class InMemoryDocstore:
"""Simple in-memory document store."""
def __init__(self):
self._dict: Dict[str, Document] = {}
def add(self, mapping: Dict[str, Document]):
"""Add documents to the store."""
if not isinstance(self._dict, dict):
# Ensure _dict is a dictionary
if hasattr(self._dict, '_dict'):
self._dict = self._dict._dict
else:
self._dict = {}
self._dict.update(mapping)
def get(self, key: str) -> Optional[Document]:
"""Get a document by key."""
if not isinstance(self._dict, dict):
# Ensure _dict is a dictionary
if hasattr(self._dict, '_dict'):
self._dict = self._dict._dict
else:
self._dict = {}
return self._dict.get(key)
class FAISS:
"""FAISS vectorstore wrapper."""
def __init__(
self,
embedding_function,
index: Optional[faiss.Index] = None,
docstore: Optional[InMemoryDocstore] = None,
index_to_docstore_id: Optional[Dict[int, str]] = None
):
self.embedding_function = embedding_function
self.index = index
self.docstore = docstore if docstore else InMemoryDocstore()
self.index_to_docstore_id = index_to_docstore_id if index_to_docstore_id else {}
@classmethod
def from_documents(
cls,
documents: List[Document],
embedding
) -> "FAISS":
"""Create a FAISS index from documents."""
if not documents:
raise ValueError("No documents provided")
# Generate embeddings
texts = [doc.page_content for doc in documents]
embeddings = embedding.embed_documents(texts)
embeddings_np = np.array(embeddings).astype("float32")
# Create FAISS index
dimension = embeddings_np.shape[1]
index = faiss.IndexFlatIP(dimension)
index.add(embeddings_np)
# Create docstore
docstore = InMemoryDocstore()
index_to_docstore_id = {}
for i, doc in enumerate(documents):
doc_id = str(uuid4())
docstore.add({doc_id: doc})
index_to_docstore_id[i] = doc_id
return cls(
embedding_function=embedding.embed_query,
index=index,
docstore=docstore,
index_to_docstore_id=index_to_docstore_id
)
def similarity_search_with_score(
self,
query: str,
k: int = 4
) -> List[Tuple[Document, float]]:
"""Search for similar documents with scores."""
if self.index is None:
return []
# Get query embedding
query_embedding = self.embedding_function(query)
query_vector = np.array([query_embedding]).astype("float32")
# Search
scores, indices = self.index.search(query_vector, k)
results = []
for score, idx in zip(scores[0], indices[0]):
if idx < 0: # FAISS returns -1 for invalid indices
continue
doc_id = self.index_to_docstore_id.get(idx)
if doc_id:
doc = self.docstore.get(doc_id)
if doc:
results.append((doc, float(score)))
return results
def save_local(self, folder_path: str):
"""Save the FAISS index and docstore to disk."""
folder = Path(folder_path)
folder.mkdir(parents=True, exist_ok=True)
# Save FAISS index
faiss.write_index(self.index, str(folder / "index.faiss"))
# Save docstore and mapping
save_dict = {
"docstore": self.docstore._dict,
"index_to_docstore_id": self.index_to_docstore_id
}
with open(folder / "index.pkl", "wb") as f:
pickle.dump(save_dict, f)
@classmethod
def load_local(
cls,
folder_path: str,
embeddings,
allow_dangerous_deserialization: bool = False
) -> "FAISS":
"""Load a FAISS index from disk."""
if not allow_dangerous_deserialization:
raise ValueError("allow_dangerous_deserialization must be True to load pickled files")
folder = Path(folder_path)
# Load FAISS index
index = faiss.read_index(str(folder / "index.faiss"))
# Load docstore and mapping
with open(folder / "index.pkl", "rb") as f:
save_dict = pickle.load(f)
# Handle different pickle formats (dict or tuple)
if isinstance(save_dict, dict):
# Expected format: dictionary with keys
docstore_data = save_dict.get("docstore", {})
index_to_docstore_id = save_dict.get("index_to_docstore_id", {})
elif isinstance(save_dict, tuple):
# Legacy format: might be a tuple, try to unpack
# If tuple has 2 elements, assume (docstore_dict, index_to_docstore_id)
if len(save_dict) == 2:
docstore_data, index_to_docstore_id = save_dict
else:
raise ValueError(
f"Unexpected pickle format: tuple with {len(save_dict)} elements. "
f"Expected dictionary or tuple with 2 elements."
)
else:
raise TypeError(
f"Unexpected pickle format: {type(save_dict)}. "
f"Expected dictionary or tuple."
)
# Handle docstore_data - could be a dict or InMemoryDocstore object
docstore = InMemoryDocstore()
if isinstance(docstore_data, dict):
# It's a dictionary, use it directly
docstore._dict = docstore_data
elif isinstance(docstore_data, InMemoryDocstore):
# It's already an InMemoryDocstore object, copy its _dict
docstore._dict = docstore_data._dict.copy()
else:
# Try to convert to dict or raise error
raise TypeError(
f"Unexpected docstore format: {type(docstore_data)}. "
f"Expected dictionary or InMemoryDocstore object."
)
return cls(
embedding_function=embeddings.embed_query,
index=index,
docstore=docstore,
index_to_docstore_id=index_to_docstore_id
)
|