Update ingest.py

ingest.py

@@ -1,84 +1,63 @@
-import os
-import logging
-from langchain_community.document_loaders import PDFMinerLoader
-from langchain.text_splitter import RecursiveCharacterTextSplitter
-from langchain_community.embeddings import HuggingFaceEmbeddings
-from langchain_community.vectorstores import FAISS
-
-logging.basicConfig(level=logging.INFO)
-logger = logging.getLogger(__name__)
-
-def create_faiss_index():
-    documents = []
-    docs_dir = "docs"
-
-    if not os.path.exists(docs_dir):
-        logger.error(f"The directory '{docs_dir}' does not exist.")
-        return
-
-    for root, dirs, files in os.walk(docs_dir):
-        for file in files:
-            if file.endswith(".pdf"):
-                file_path = os.path.join(root, file)
-                logger.info(f"Loading document: {file_path}")
-                try:
-                    loader = PDFMinerLoader(file_path)
-                    loaded_docs = loader.load()
-                    if loaded_docs:
-                        logger.info(f"Loaded {len(loaded_docs)} documents from {file_path}")
-                        documents.extend(loaded_docs)
-                    else:
-                        logger.warning(f"No documents loaded from {file_path}")
-                except Exception as e:
-                    logger.error(f"Error loading {file_path}: {e}")
-
-    if not documents:
-        logger.error("No documents were loaded. Check the 'docs' directory and file paths.")
-        return
-
-    logger.info(f"Loaded {len(documents)} documents.")
-
-    text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)
-    texts = text_splitter.split_documents(documents)
-    logger.info(f"Created {len(texts)} text chunks.")
-    if not texts:
-        logger.error("No text chunks created. Check the text splitting process.")
-        return
-
-    try:
-        embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
-        logger.info("Embeddings initialized successfully.")
-    except Exception as e:
-        logger.error(f"Failed to initialize embeddings: {e}")
-        return
-
-    try:
-        db = FAISS.from_documents(texts, embeddings)
-        if db.index.ntotal > 0:
-            logger.info(f"Created FAISS index with {db.index.ntotal} vectors.")
-        else:
-            logger.error("FAISS index contains 0 vectors.")
-            return
-    except Exception as e:
-        logger.error(f"Failed to create FAISS index: {e}")
-        return
-
-    index_dir = "faiss_index"
-    if not os.path.exists(index_dir):
-        os.makedirs(index_dir)
-
-    try:
-        db.save_local(index_dir)
-        index_file_path = os.path.join(index_dir, "index.faiss")
-        file_size = os.path.getsize(index_file_path)
-        logger.info(f"FAISS index saved to {index_dir}")
-        logger.info(f"Index file size: {file_size} bytes")
-        if file_size == 0:
-            logger.error(f"Index file '{index_file_path}' is empty.")
-        else:
-            logger.info(f"Index file '{index_file_path}' created successfully.")
-    except Exception as e:
-        logger.error(f"Failed to save FAISS index: {e}")
-
-if __name__ == "__main__":
-    create_faiss_index()
+import os
+import logging
+from langchain_community.document_loaders import PDFMinerLoader
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain_community.embeddings import HuggingFaceEmbeddings
+from langchain_community.vectorstores import Chroma
+
+logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger(__name__)
+
+def create_chroma_db():
+    documents = []
+    docs_dir = "docs"
+
+    if not os.path.exists(docs_dir):
+        logger.error(f"The directory '{docs_dir}' does not exist.")
+        return
+
+    for root, dirs, files in os.walk(docs_dir):
+        for file in files:
+            if file.endswith(".pdf"):
+                file_path = os.path.join(root, file)
+                logger.info(f"Loading document: {file_path}")
+                try:
+                    loader = PDFMinerLoader(file_path)
+                    loaded_docs = loader.load()
+                    if loaded_docs:
+                        logger.info(f"Loaded {len(loaded_docs)} documents from {file_path}")
+                        documents.extend(loaded_docs)
+                    else:
+                        logger.warning(f"No documents loaded from {file_path}")
+                except Exception as e:
+                    logger.error(f"Error loading {file_path}: {e}")
+
+    if not documents:
+        logger.error("No documents were loaded. Check the 'docs' directory and file paths.")
+        return
+
+    logger.info(f"Loaded {len(documents)} documents.")
+
+    text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)
+    texts = text_splitter.split_documents(documents)
+    logger.info(f"Created {len(texts)} text chunks.")
+    if not texts:
+        logger.error("No text chunks created. Check the text splitting process.")
+        return
+
+    try:
+        embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
+        logger.info("Embeddings initialized successfully.")
+    except Exception as e:
+        logger.error(f"Failed to initialize embeddings: {e}")
+        return
+
+    try:
+        db = Chroma.from_documents(texts, embeddings, persist_directory="chroma_db")
+        logger.info(f"Created Chroma database with {len(texts)} vectors.")
+    except Exception as e:
+        logger.error(f"Failed to create Chroma database: {e}")
+        return
+
+if __name__ == "__main__":
+    create_chroma_db()