Update app.py

app.py

@@ -648,77 +648,71 @@ if __name__ == '__main__':
         
             uploaded_files = st.file_uploader("Upload a file :", accept_multiple_files=True, type=['pptx', 'ppt'])
             print(f'uploaded-files : {uploaded_files}')
-            for uploaded_file in uploaded_files:
-                
-                processing_time = datetime.now().strftime('%d %b %Y, %I:%M %p')
-                file_name = os.path.basename(uploaded_file.name)
-                base_name, ext = os.path.splitext(file_name)
-
-                full_path = os.path.realpath(uploaded_file.name)
-                file_type = ext.lstrip('.')
-
-                d = {
-                    'icon': icon_to_types[file_type][0],
-                    'document': base_name, 
-                    'type': icon_to_types[file_type][1],
-                    'path': full_path,
-                    'time': [datetime.strptime(processing_time, '%d %b %Y, %I:%M %p')],
-                    'toggle': True
-                }
-                
-                st.session_state.df = pd.concat(
-                    [st.session_state.df, pd.DataFrame(data=d)]
-                )
-                st.session_state.df.to_parquet(
-                    os.path.join(
-                        data_editor_path, 
-                        'data_editor.parquet.br'
-                    ), 
-                    compression='brotli', 
-                    engine='pyarrow'
-                )
-
-                elements = partition_pptx(file=uploaded_file)
-                
-                for elem in elements:
-                    elem.text = clean(elem.text, bullets=True)
-                    text_type = elem.to_dict()['type']
-                    print(f'UNSTRUCTURED TEXT: {text_type} , {elem.text}')
+            if uploaded_files[-1] not in uploaded_files.deleted:
+                for uploaded_file in uploaded_files:
+                    
+                    processing_time = datetime.now().strftime('%d %b %Y, %I:%M %p')
+                    file_name = os.path.basename(uploaded_file.name)
+                    base_name, ext = os.path.splitext(file_name)
 
-                weakDict, tables = ppt_chunk(uploaded_file, nlp)
-                documents = weakDict.all_texts()
+                    full_path = os.path.realpath(uploaded_file.name)
+                    file_type = ext.lstrip('.')
+                    
+                    d = {
+                        'icon': icon_to_types[file_type][0],
+                        'document': base_name, 
+                        'type': icon_to_types[file_type][1],
+                        'path': full_path,
+                        'time': [datetime.strptime(processing_time, '%d %b %Y, %I:%M %p')],
+                        'toggle': True
+                    }
+                    
+                    st.session_state.df = pd.concat(
+                        [st.session_state.df, pd.DataFrame(data=d)]
+                    )
+                    st.session_state.df.to_parquet(
+                        os.path.join(
+                            data_editor_path, 
+                            'data_editor.parquet.br'
+                        ), 
+                        compression='brotli', 
+                        engine='pyarrow'
+                    )
+                    
+                    weakDict, tables = ppt_chunk(uploaded_file, nlp)
+                    documents = weakDict.all_texts()
 
-                dense = dense_model.embed_documents(documents)
-                print(f'dense: {type(dense)}')
-                sparse = [SparseVector(indices=s.indices.tolist(), values=s.values.tolist()) for s in sparse_model.embed(documents, 32)]
-                print(f'dense: {type(sparse)}')
+                    dense = dense_model.embed_documents(documents)
+                    print(f'dense: {type(dense)}')
+                    sparse = [SparseVector(indices=s.indices.tolist(), values=s.values.tolist()) for s in sparse_model.embed(documents, 32)]
+                    print(f'dense: {type(sparse)}')
 
-                print(f'LEN: {len(documents)}, {len(weakDict.all_metadatas())}')
+                    print(f'LEN: {len(documents)}, {len(weakDict.all_metadatas())}')
                 
-                client.upsert(
-                    collection_name=collection_name,
-                    points=Batch(
-                        ids=weakDict.all_ids(),
-                        payloads=[{ 'text': documents[i], 'metadata': metadata } for i, metadata in enumerate(weakDict.all_metadatas())],
-                        vectors={
-                            'text-dense': dense,
-                            'text-sparse': sparse
-                        }
+                    client.upsert(
+                        collection_name=collection_name,
+                        points=Batch(
+                            ids=weakDict.all_ids(),
+                            payloads=[{ 'text': documents[i], 'metadata': metadata } for i, metadata in enumerate(weakDict.all_metadatas())],
+                            vectors={
+                                'text-dense': dense,
+                                'text-sparse': sparse
+                            }
+                        )
                     )
-                )
 
-                st.session_state.dialog_open = True
-                dialog_container = float_dialog(
-                    show=st.session_state.dialog_open, 
-                    width=7, 
-                    background="#EEE3D3", 
-                    transition=7, 
-                    transition_from="top", 
-                    transition_to="center"
-                )
-
-                with dialog_container:
-                    st.subheader("Documents Ingested !")
-                    if st.button("Ok", key="ok"):
-                        st.session_state.dialog_open = False
-                        st.rerun()
+                    st.session_state.dialog_open = True
+                    dialog_container = float_dialog(
+                        show=st.session_state.dialog_open, 
+                        width=7, 
+                        background="#EEE3D3", 
+                        transition=7, 
+                        transition_from="top", 
+                        transition_to="center"
+                    )
+                    
+                    with dialog_container:
+                        st.subheader("Documents Ingested !")
+                        if st.button("Ok", key="ok"):
+                            st.session_state.dialog_open = False
+                            st.rerun()