Update app.py

app.py

@@ -273,12 +273,15 @@ def check_for_new_retrievers():
     # see if retrievers/vector stores created by user's own uploaded PDF or newly scraped data is found
     new_documents_chroma = glob.glob("chromadb/new*")
     new_documents_bm25 = glob.glob("bm25/new*")
+    new_documents_chroma = [os.path.split(doc)[-1] for doc in new_documents_chroma]
+    new_documents_bm25 = [os.path.split(doc)[-1] for doc in new_documents_bm25]
     new_countries = []
-    print(new_documents_bm25)
+
     # loop through new docs in chroma retrievers created by user scraping/pdf (if any)
     try:
         for doc in new_documents_chroma:
-            if ((("bm25\\" + doc.split('\\')[1] + ".pickle") in new_documents_bm25) | (("bm25/" + doc.split('/')[1] + ".pickle") in new_documents_bm25)): # check that the doc also exists for bm25 retriever
+            #print(doc)
+            if ((doc + ".pickle") in new_documents_bm25): # check that the doc also exists for bm25 retriever
 
                 new_doc_country = doc.split('_')[1]
                 new_doc_chunk_size = doc.split('_')[3]
@@ -643,7 +646,7 @@ if page == "Scrape or Upload Own Docs":
     # how user wishes to populate documents
     options = [
                 "Upload Own PDF", 
-                "Automatically Scrape Web Data using DuckDuckGo (Will take 5 mins or more)"
+                "Automatically Scrape Web Data using DuckDuckGo (may take more than 5 mins)"
             ]
 
     option = st.radio(
@@ -657,12 +660,12 @@ if page == "Scrape or Upload Own Docs":
 
     # save new retrievers in local directory
     def save_new_retrievers(all_documents, chunk_size, chunk_overlap, country_scrape_upload):
-        with st.spinner('Setting up new bm25 retrievers with documents, can take 5 mins and above...'):
+        with st.spinner('Setting up new bm25 retrievers with documents, may take more than 5 mins...'):
             # vectorstore for this country will be stored in "bm25/new_{country}_chunk_{chunk_size}_overlap_{chunk_overlap}_"
             # can be used to override existing vectorstore for this country in sidebar document configuration
             setup_bm25_retriever(all_documents, chunk_size, chunk_overlap, country_scrape_upload)
         
-        with st.spinner('Setting up new chromadb vector stores with documents, can take 5 mins and above...'):
+        with st.spinner('Setting up new chromadb vector stores with documents, may take more than 5 mins...'):
             # vectorstore for this country will be stored in "chroma_db/new_{country}_chunk_{chunk_size}_overlap_{chunk_overlap}_"
             # can be used to override existing vectorstore for this country in sidebar document configuration
             setup_chromadb_vectorstore(hf_embeddings, all_documents, chunk_size, chunk_overlap, country_scrape_upload) 
@@ -682,7 +685,7 @@ if page == "Scrape or Upload Own Docs":
                 with open(temp_file, "wb") as file:
                     file.write(uploaded_pdf.getvalue())
                     pdf_filename = uploaded_pdf.name
-            submit_upload_pdf = st.form_submit_button(label='Upload and Create Vector Store')   
+            submit_upload_pdf = st.form_submit_button(label='Upload and Create Vector Store (Scroll down after clicking)')   
             st.markdown(":blue[NOTE:] After you are done creating the vectore store, the country will appear under :blue[Countries to Override in the 'Document Config'] section of the left sidebar. Select the country to override it.")