fix

app/rag.py

@@ -22,35 +22,38 @@ QDRANT_API_KEY = os.getenv('QDRANT_API_KEY')
 
 
 class ChatPDF:
-    text_chunks = []
-    doc_ids = []
-    nodes = []
     hyde_query_engine = None
+    text_parser = None
+    vector_store = None
+    embed_model = None
     logger = None
 
     def __init__(self):
         logging.basicConfig(level=logging.INFO)
         self.logger = logging.getLogger(__name__)
 
-        text_parser = SentenceSplitter(chunk_size=512, chunk_overlap=100)
+        self.text_parser = SentenceSplitter(chunk_size=512, chunk_overlap=100)
 
         self.logger.info("initializing the vector store related objects")
         client = QdrantClient(url=QDRANT_API_URL, api_key=QDRANT_API_KEY)
-        vector_store = QdrantVectorStore(client=client, collection_name="rag_documents")
+        self.vector_store = QdrantVectorStore(client=client, collection_name="rag_documents")
 
         self.logger.info("initializing the OllamaEmbedding")
-        embed_model = OllamaEmbedding(model_name='mxbai-embed-large')
+        self.embed_model = OllamaEmbedding(model_name='mxbai-embed-large')
         self.logger.info("initializing the global settings")
-        Settings.embed_model = embed_model
+        Settings.embed_model = self.embed_model
         Settings.llm = Ollama(model="qwen:1.8b", request_timeout=1000000)
-        Settings.transformations = [text_parser]
+        Settings.transformations = [self.text_parser]
 
     def ingest(self, dir_path: str):
         docs = SimpleDirectoryReader(input_dir=dir_path).load_data()
+        text_chunks = []
+        doc_ids = []
+        nodes = []
 
         self.logger.info("enumerating docs")
         for doc_idx, doc in enumerate(docs):
-            curr_text_chunks = text_parser.split_text(doc.text)
+            curr_text_chunks = self.text_parser.split_text(doc.text)
             text_chunks.extend(curr_text_chunks)
             doc_ids.extend([doc_idx] * len(curr_text_chunks))
 
@@ -63,13 +66,13 @@ class ChatPDF:
 
         self.logger.info("enumerating nodes")
         for node in nodes:
-            node_embedding = embed_model.get_text_embedding(
+            node_embedding = self.embed_model.get_text_embedding(
                 node.get_content(metadata_mode=MetadataMode.ALL)
             )
             node.embedding = node_embedding
 
         self.logger.info("initializing the storage context")
-        storage_context = StorageContext.from_defaults(vector_store=vector_store)
+        storage_context = StorageContext.from_defaults(vector_store=self.vector_store)
         self.logger.info("indexing the nodes in VectorStoreIndex")
         index = VectorStoreIndex(
             nodes=nodes,
@@ -99,7 +102,4 @@ class ChatPDF:
         return response
 
     def clear(self):
-        self.text_chunks = []
-        self.doc_ids = []
-        self.nodes = []
         self.hyde_query_engine = None