OuroborosM commited on
Commit
13ee1ce
·
1 Parent(s): 591e68e

pop info in chat for doc upload

Browse files
Files changed (1) hide show
  1. app.py +28 -16
app.py CHANGED
@@ -134,20 +134,29 @@ def process_documents(ignored_files: List[str] = []) -> List[Document]:
134
  print(f"Split into {len(texts)} chunks of text (max. {chunk_size} tokens each)")
135
  return texts
136
 
137
- def UpdateDb(file_path: str):
138
- global vectordb_p
139
- # pinecone.Index(index_name).delete(delete_all=True, namespace='')
140
- # collection = vectordb_p.get()
141
- # split_docs = process_documents([metadata['source'] for metadata in collection['metadatas']])
142
- # split_docs = process_documents()
143
- documents = load_single_document(file_path)
144
  if not documents:
145
  print("No new documents to load")
146
  exit(0)
147
  print(f"Loaded {len(documents)} new documents from {source_directory}")
148
  text_splitter = RecursiveCharacterTextSplitter(chunk_size=chunk_size, chunk_overlap=chunk_overlap)
149
- split_docs = text_splitter.split_documents(documents)
150
- print(f"Split into {len(split_docs)} chunks of text (max. {chunk_size} tokens each)")
 
 
 
 
 
 
 
 
 
151
  tt = len(split_docs)
152
  print(split_docs[tt-1])
153
  print(f"Creating embeddings. May take some minutes...")
@@ -316,18 +325,21 @@ def chathmi2(message, history):
316
  # chatbot = gr.Chatbot().style(color_map =("blue", "pink"))
317
  # chatbot = gr.Chatbot(color_map =("blue", "pink"))
318
 
319
- def upload_file(files):
320
- print(files)
321
- pass
 
 
322
 
323
  with gr.Blocks() as demo:
324
- gr.ChatInterface(
325
  chathmi2,
326
- title="STLA BABY - YOUR FRIENDLY GUIDE ",
327
- description= "v0.2: Powered by MECH Core Team",
 
328
  )
329
  upload_button = gr.UploadButton("Upload File", file_count="multiple")
330
- upload_button.upload(upload_file, upload_button)
331
 
332
  # demo = gr.Interface(
333
  # chathmi,
 
134
  print(f"Split into {len(texts)} chunks of text (max. {chunk_size} tokens each)")
135
  return texts
136
 
137
+ def process_documents_2(ignored_files: List[str] = []) -> List[Document]:
138
+ """
139
+ Load documents and split in chunks
140
+ """
141
+ print(f"Loading documents from {source_directory}")
142
+ print("File Path to start processing:", file_path)
143
+ documents = load_documents(file_path, ignored_files)
144
  if not documents:
145
  print("No new documents to load")
146
  exit(0)
147
  print(f"Loaded {len(documents)} new documents from {source_directory}")
148
  text_splitter = RecursiveCharacterTextSplitter(chunk_size=chunk_size, chunk_overlap=chunk_overlap)
149
+ texts = text_splitter.split_documents(documents)
150
+ print(f"Split into {len(texts)} chunks of text (max. {chunk_size} tokens each)")
151
+ return texts
152
+
153
+ def UpdateDb():
154
+ global vectordb_p
155
+ # pinecone.Index(index_name).delete(delete_all=True, namespace='')
156
+ # collection = vectordb_p.get()
157
+ # split_docs = process_documents([metadata['source'] for metadata in collection['metadatas']])
158
+ # split_docs = process_documents()
159
+ split_docs = process_documents_2()
160
  tt = len(split_docs)
161
  print(split_docs[tt-1])
162
  print(f"Creating embeddings. May take some minutes...")
 
325
  # chatbot = gr.Chatbot().style(color_map =("blue", "pink"))
326
  # chatbot = gr.Chatbot(color_map =("blue", "pink"))
327
 
328
+ def func_upload_file(files):
329
+ file_path = files
330
+ print(file_path)
331
+ # UpdateDb()
332
+ return("Test Upload")
333
 
334
  with gr.Blocks() as demo:
335
+ main = gr.ChatInterface(
336
  chathmi2,
337
+ examples=["What is PDP?"],
338
+ title="STLA BABY - YOUR FRIENDLY GUIDE",
339
+ description= "v0.3: Powered by MECH Core Team",
340
  )
341
  upload_button = gr.UploadButton("Upload File", file_count="multiple")
342
+ upload_button.upload(func_upload_file, upload_button, main)
343
 
344
  # demo = gr.Interface(
345
  # chathmi,