Spaces:
Sleeping
Sleeping
making classes and test for vectorstore handling
Browse files
pytest.ini
ADDED
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
1 |
+
[pytest]
|
2 |
+
pythonpath = .
|
rag_app/vector_store_handler/__init__.py
ADDED
File without changes
|
rag_app/vector_store_handler/vectorstores.py
ADDED
@@ -0,0 +1,203 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from abc import ABC, abstractmethod
|
2 |
+
from langchain.vectorstores import Chroma, FAISS
|
3 |
+
from langchain.embeddings import OpenAIEmbeddings
|
4 |
+
from langchain.text_splitter import CharacterTextSplitter
|
5 |
+
from langchain.document_loaders import TextLoader
|
6 |
+
|
7 |
+
class BaseVectorStore(ABC):
|
8 |
+
"""
|
9 |
+
Abstract base class for vector stores.
|
10 |
+
|
11 |
+
This class defines the interface for vector stores and implements
|
12 |
+
common functionality.
|
13 |
+
"""
|
14 |
+
|
15 |
+
def __init__(self, embedding_model, persist_directory=None):
|
16 |
+
"""
|
17 |
+
Initialize the BaseVectorStore.
|
18 |
+
|
19 |
+
Args:
|
20 |
+
embedding_model: The embedding model to use for vectorizing text.
|
21 |
+
persist_directory (str, optional): Directory to persist the vector store.
|
22 |
+
"""
|
23 |
+
self.persist_directory = persist_directory
|
24 |
+
self.embeddings = embedding_model
|
25 |
+
self.vectorstore = None
|
26 |
+
|
27 |
+
def load_and_process_documents(self, file_path, chunk_size=1000, chunk_overlap=0):
|
28 |
+
"""
|
29 |
+
Load and process documents from a file.
|
30 |
+
|
31 |
+
Args:
|
32 |
+
file_path (str): Path to the file to load.
|
33 |
+
chunk_size (int): Size of text chunks for processing.
|
34 |
+
chunk_overlap (int): Overlap between chunks.
|
35 |
+
|
36 |
+
Returns:
|
37 |
+
list: Processed documents.
|
38 |
+
"""
|
39 |
+
loader = TextLoader(file_path)
|
40 |
+
documents = loader.load()
|
41 |
+
text_splitter = CharacterTextSplitter(chunk_size=chunk_size, chunk_overlap=chunk_overlap)
|
42 |
+
return text_splitter.split_documents(documents)
|
43 |
+
|
44 |
+
@abstractmethod
|
45 |
+
def create_vectorstore(self, texts):
|
46 |
+
"""
|
47 |
+
Create a new vector store from the given texts.
|
48 |
+
|
49 |
+
Args:
|
50 |
+
texts (list): List of texts to vectorize and store.
|
51 |
+
"""
|
52 |
+
pass
|
53 |
+
|
54 |
+
@abstractmethod
|
55 |
+
def load_existing_vectorstore(self):
|
56 |
+
"""
|
57 |
+
Load an existing vector store from the persist directory.
|
58 |
+
"""
|
59 |
+
pass
|
60 |
+
|
61 |
+
def similarity_search(self, query):
|
62 |
+
"""
|
63 |
+
Perform a similarity search on the vector store.
|
64 |
+
|
65 |
+
Args:
|
66 |
+
query (str): The query text to search for.
|
67 |
+
|
68 |
+
Returns:
|
69 |
+
list: Search results.
|
70 |
+
|
71 |
+
Raises:
|
72 |
+
ValueError: If the vector store is not initialized.
|
73 |
+
"""
|
74 |
+
if not self.vectorstore:
|
75 |
+
raise ValueError("Vector store not initialized. Call create_vectorstore or load_existing_vectorstore first.")
|
76 |
+
return self.vectorstore.similarity_search(query)
|
77 |
+
|
78 |
+
@abstractmethod
|
79 |
+
def save(self):
|
80 |
+
"""
|
81 |
+
Save the current state of the vector store.
|
82 |
+
"""
|
83 |
+
pass
|
84 |
+
|
85 |
+
class ChromaVectorStore(BaseVectorStore):
|
86 |
+
"""
|
87 |
+
Implementation of BaseVectorStore using Chroma as the backend.
|
88 |
+
"""
|
89 |
+
|
90 |
+
def create_vectorstore(self, texts):
|
91 |
+
"""
|
92 |
+
Create a new Chroma vector store from the given texts.
|
93 |
+
|
94 |
+
Args:
|
95 |
+
texts (list): List of texts to vectorize and store.
|
96 |
+
"""
|
97 |
+
self.vectorstore = Chroma.from_documents(
|
98 |
+
texts,
|
99 |
+
self.embeddings,
|
100 |
+
persist_directory=self.persist_directory
|
101 |
+
)
|
102 |
+
|
103 |
+
def load_existing_vectorstore(self):
|
104 |
+
"""
|
105 |
+
Load an existing Chroma vector store from the persist directory.
|
106 |
+
|
107 |
+
Raises:
|
108 |
+
ValueError: If persist_directory is not set.
|
109 |
+
"""
|
110 |
+
if self.persist_directory:
|
111 |
+
self.vectorstore = Chroma(
|
112 |
+
persist_directory=self.persist_directory,
|
113 |
+
embedding_function=self.embeddings
|
114 |
+
)
|
115 |
+
else:
|
116 |
+
raise ValueError("Persist directory is required for loading Chroma.")
|
117 |
+
|
118 |
+
def save(self):
|
119 |
+
"""
|
120 |
+
Save the current state of the Chroma vector store.
|
121 |
+
|
122 |
+
Raises:
|
123 |
+
ValueError: If the vector store is not initialized.
|
124 |
+
"""
|
125 |
+
if not self.vectorstore:
|
126 |
+
raise ValueError("Vector store not initialized. Nothing to save.")
|
127 |
+
self.vectorstore.persist()
|
128 |
+
|
129 |
+
class FAISSVectorStore(BaseVectorStore):
|
130 |
+
"""
|
131 |
+
Implementation of BaseVectorStore using FAISS as the backend.
|
132 |
+
"""
|
133 |
+
|
134 |
+
def create_vectorstore(self, texts):
|
135 |
+
"""
|
136 |
+
Create a new FAISS vector store from the given texts.
|
137 |
+
|
138 |
+
Args:
|
139 |
+
texts (list): List of texts to vectorize and store.
|
140 |
+
"""
|
141 |
+
self.vectorstore = FAISS.from_documents(texts, self.embeddings)
|
142 |
+
|
143 |
+
def load_existing_vectorstore(self):
|
144 |
+
"""
|
145 |
+
Load an existing FAISS vector store from the persist directory.
|
146 |
+
|
147 |
+
Raises:
|
148 |
+
ValueError: If persist_directory is not set.
|
149 |
+
"""
|
150 |
+
if self.persist_directory:
|
151 |
+
self.vectorstore = FAISS.load_local(self.persist_directory, self.embeddings)
|
152 |
+
else:
|
153 |
+
raise ValueError("Persist directory is required for loading FAISS.")
|
154 |
+
|
155 |
+
def save(self):
|
156 |
+
"""
|
157 |
+
Save the current state of the FAISS vector store.
|
158 |
+
|
159 |
+
Raises:
|
160 |
+
ValueError: If the vector store is not initialized.
|
161 |
+
"""
|
162 |
+
if not self.vectorstore:
|
163 |
+
raise ValueError("Vector store not initialized. Nothing to save.")
|
164 |
+
self.vectorstore.save_local(self.persist_directory)
|
165 |
+
|
166 |
+
# Usage example:
|
167 |
+
def main():
|
168 |
+
"""
|
169 |
+
Example usage of the vector store classes.
|
170 |
+
"""
|
171 |
+
# Create an embedding model
|
172 |
+
embedding_model = OpenAIEmbeddings()
|
173 |
+
|
174 |
+
# Using Chroma
|
175 |
+
chroma_store = ChromaVectorStore(embedding_model, persist_directory="./chroma_store")
|
176 |
+
texts = chroma_store.load_and_process_documents("path/to/your/file.txt")
|
177 |
+
chroma_store.create_vectorstore(texts)
|
178 |
+
results = chroma_store.similarity_search("Your query here")
|
179 |
+
print("Chroma results:", results[0].page_content)
|
180 |
+
chroma_store.save()
|
181 |
+
|
182 |
+
# Load existing Chroma store
|
183 |
+
existing_chroma = ChromaVectorStore(embedding_model, persist_directory="./chroma_store")
|
184 |
+
existing_chroma.load_existing_vectorstore()
|
185 |
+
results = existing_chroma.similarity_search("Another query")
|
186 |
+
print("Existing Chroma results:", results[0].page_content)
|
187 |
+
|
188 |
+
# Using FAISS
|
189 |
+
faiss_store = FAISSVectorStore(embedding_model, persist_directory="./faiss_store")
|
190 |
+
texts = faiss_store.load_and_process_documents("path/to/your/file.txt")
|
191 |
+
faiss_store.create_vectorstore(texts)
|
192 |
+
results = faiss_store.similarity_search("Your query here")
|
193 |
+
print("FAISS results:", results[0].page_content)
|
194 |
+
faiss_store.save()
|
195 |
+
|
196 |
+
# Load existing FAISS store
|
197 |
+
existing_faiss = FAISSVectorStore(embedding_model, persist_directory="./faiss_store")
|
198 |
+
existing_faiss.load_existing_vectorstore()
|
199 |
+
results = existing_faiss.similarity_search("Another query")
|
200 |
+
print("Existing FAISS results:", results[0].page_content)
|
201 |
+
|
202 |
+
if __name__ == "__main__":
|
203 |
+
main()
|
tests/vector_store_handler/test_vectorstores.py
ADDED
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import unittest
|
2 |
+
from unittest.mock import MagicMock, patch
|
3 |
+
from langchain.embeddings import OpenAIEmbeddings
|
4 |
+
from langchain.schema import Document
|
5 |
+
|
6 |
+
# Update the import to reflect your project structure
|
7 |
+
from rag_app.vector_store_handler.vectorstores import BaseVectorStore, ChromaVectorStore, FAISSVectorStore
|
8 |
+
|
9 |
+
class TestBaseVectorStore(unittest.TestCase):
|
10 |
+
def setUp(self):
|
11 |
+
self.embedding_model = MagicMock(spec=OpenAIEmbeddings)
|
12 |
+
self.base_store = BaseVectorStore(self.embedding_model, "test_dir")
|
13 |
+
|
14 |
+
def test_init(self):
|
15 |
+
self.assertEqual(self.base_store.persist_directory, "test_dir")
|
16 |
+
self.assertEqual(self.base_store.embeddings, self.embedding_model)
|
17 |
+
self.assertIsNone(self.base_store.vectorstore)
|
18 |
+
|
19 |
+
@patch('rag_app.vector_store_handler.vectorstores.TextLoader')
|
20 |
+
@patch('rag_app.vector_store_handler.vectorstores.CharacterTextSplitter')
|
21 |
+
def test_load_and_process_documents(self, mock_splitter, mock_loader):
|
22 |
+
mock_loader.return_value.load.return_value = ["doc1", "doc2"]
|
23 |
+
mock_splitter.return_value.split_documents.return_value = ["split1", "split2"]
|
24 |
+
|
25 |
+
result = self.base_store.load_and_process_documents("test.txt")
|
26 |
+
|
27 |
+
mock_loader.assert_called_once_with("test.txt")
|
28 |
+
mock_splitter.assert_called_once_with(chunk_size=1000, chunk_overlap=0)
|
29 |
+
self.assertEqual(result, ["split1", "split2"])
|
30 |
+
|
31 |
+
def test_similarity_search_not_initialized(self):
|
32 |
+
with self.assertRaises(ValueError):
|
33 |
+
self.base_store.similarity_search("query")
|
34 |
+
|
35 |
+
class TestChromaVectorStore(unittest.TestCase):
|
36 |
+
def setUp(self):
|
37 |
+
self.embedding_model = MagicMock(spec=OpenAIEmbeddings)
|
38 |
+
self.chroma_store = ChromaVectorStore(self.embedding_model, "test_dir")
|
39 |
+
|
40 |
+
@patch('rag_app.vector_store_handler.vectorstores.Chroma')
|
41 |
+
def test_create_vectorstore(self, mock_chroma):
|
42 |
+
texts = [Document(page_content="test")]
|
43 |
+
self.chroma_store.create_vectorstore(texts)
|
44 |
+
mock_chroma.from_documents.assert_called_once_with(
|
45 |
+
texts,
|
46 |
+
self.embedding_model,
|
47 |
+
persist_directory="test_dir"
|
48 |
+
)
|
49 |
+
|
50 |
+
@patch('rag_app.vector_store_handler.vectorstores.Chroma')
|
51 |
+
def test_load_existing_vectorstore(self, mock_chroma):
|
52 |
+
self.chroma_store.load_existing_vectorstore()
|
53 |
+
mock_chroma.assert_called_once_with(
|
54 |
+
persist_directory="test_dir",
|
55 |
+
embedding_function=self.embedding_model
|
56 |
+
)
|
57 |
+
|
58 |
+
def test_save(self):
|
59 |
+
self.chroma_store.vectorstore = MagicMock()
|
60 |
+
self.chroma_store.save()
|
61 |
+
self.chroma_store.vectorstore.persist.assert_called_once()
|
62 |
+
|
63 |
+
class TestFAISSVectorStore(unittest.TestCase):
|
64 |
+
def setUp(self):
|
65 |
+
self.embedding_model = MagicMock(spec=OpenAIEmbeddings)
|
66 |
+
self.faiss_store = FAISSVectorStore(self.embedding_model, "test_dir")
|
67 |
+
|
68 |
+
@patch('rag_app.vector_store_handler.vectorstores.FAISS')
|
69 |
+
def test_create_vectorstore(self, mock_faiss):
|
70 |
+
texts = [Document(page_content="test")]
|
71 |
+
self.faiss_store.create_vectorstore(texts)
|
72 |
+
mock_faiss.from_documents.assert_called_once_with(texts, self.embedding_model)
|
73 |
+
|
74 |
+
@patch('rag_app.vector_store_handler.vectorstores.FAISS')
|
75 |
+
def test_load_existing_vectorstore(self, mock_faiss):
|
76 |
+
self.faiss_store.load_existing_vectorstore()
|
77 |
+
mock_faiss.load_local.assert_called_once_with("test_dir", self.embedding_model)
|
78 |
+
|
79 |
+
@patch('rag_app.vector_store_handler.vectorstores.FAISS')
|
80 |
+
def test_save(self, mock_faiss):
|
81 |
+
self.faiss_store.vectorstore = MagicMock()
|
82 |
+
self.faiss_store.save()
|
83 |
+
self.faiss_store.vectorstore.save_local.assert_called_once_with("test_dir")
|
84 |
+
|
85 |
+
if __name__ == '__main__':
|
86 |
+
unittest.main()
|