Update app.py

app.py

@@ -6,13 +6,17 @@ from torch import cuda
 from langchain_community.embeddings import HuggingFaceEmbeddings, HuggingFaceInferenceAPIEmbeddings
 from langchain_community.vectorstores import Qdrant
 from qdrant_client import QdrantClient
+from langchain.retrievers import ContextualCompressionRetriever
+from langchain.retrievers.document_compressors import CrossEncoderReranker
+from langchain_community.cross_encoders import HuggingFaceCrossEncoder
+
 device = 'cuda' if cuda.is_available() else 'cpu'
 
 
 st.set_page_config(page_title="SEARCH IATI",layout='wide')
 st.title("SEARCH IATI Database")
 var=st.text_input("enter keyword")
-title = var.replace(' ','+')
+
 
 def create_chunks(text):
     text_splitter = TokenTextSplitter(chunk_size=500, chunk_overlap=0)
@@ -75,6 +79,41 @@ def embed_chunks(chunks):
     print("vector embeddings done")
     return qdrant_collections
     
-chunks = get_chunks()
-qdrant_col = embed_chunks(chunks)
-st.write("Success")
+def get_local_qdrant(): 
+    """once the local qdrant server is created this is used to make the connection to exisitng server"""
+
+    qdrant_collections = {}
+    embeddings = HuggingFaceEmbeddings(
+        model_kwargs = {'device': device},
+        encode_kwargs = {'normalize_embeddings': True},
+        model_name='BAAI/bge-m3')
+    client = QdrantClient(path="/data/local_qdrant") 
+    print("Collections in local Qdrant:",client.get_collections())
+    qdrant_collections['all'] = Qdrant(client=client, collection_name='all', embeddings=embeddings, )
+    return qdrant_collections    
+
+def get_context(vectorstore,query):
+    # create metadata filter
+
+
+    # getting context
+    retriever = vectorstore.as_retriever(search_type="similarity_score_threshold", 
+                                         search_kwargs={"score_threshold": 0.5, 
+                                                        "k": 10), 
+                                                        })
+#    # re-ranking the retrieved results
+#    model = HuggingFaceCrossEncoder(model_name=model_config.get('ranker','MODEL'))
+#   compressor = CrossEncoderReranker(model=model, top_n=int(model_config.get('ranker','TOP_K')))
+#    compression_retriever = ContextualCompressionRetriever(
+#           base_compressor=compressor, base_retriever=retriever
+#       )
+    context_retrieved = retriever.invoke(query)
+    print(f"retrieved paragraphs:{len(context_retrieved)}")
+
+    return context_retrieved
+#chunks = get_chunks()
+vectorstores = get_local_qdrant()    
+button=st.button("search")
+results= get_context(vectorstores, f"find the relvant paragraphs for: {var}")
+
+st.write(results)