OuroborosM commited on
Commit
66d2fc7
·
1 Parent(s): b8d3c04

Update app.py

Browse files

update DB loader func

Files changed (1) hide show
  1. app.py +31 -7
app.py CHANGED
@@ -119,6 +119,27 @@ def load_documents(source_dir: str, ignored_files: List[str] = []) -> List[Docum
119
 
120
  return results
121
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
122
  def process_documents(ignored_files: List[str] = []) -> List[Document]:
123
  """
124
  Load documents and split in chunks
@@ -140,7 +161,7 @@ def process_documents_2(ignored_files: List[str] = []) -> List[Document]:
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)
@@ -328,20 +349,23 @@ def chathmi2(message, history):
328
  def func_upload_file(files, chat_history):
329
  file_path = files
330
  print(file_path)
331
- # UpdateDb()
332
- print(chat_history)
333
- test_msg = ["", "Test Upload Ongoing...."]
 
 
 
334
  chat_history.append(test_msg)
335
- return chat_history
336
 
337
  with gr.Blocks() as demo:
338
  main = gr.ChatInterface(
339
  chathmi2,
340
  title="STLA BABY - YOUR FRIENDLY GUIDE",
341
  description= "v0.3: Powered by MECH Core Team",
 
342
  )
343
- upload_button = gr.UploadButton("Upload File", file_count="multiple")
344
- upload_button.upload(func_upload_file, [upload_button, main.chatbot], main.chatbot)
345
 
346
  # demo = gr.Interface(
347
  # chathmi,
 
119
 
120
  return results
121
 
122
+ def load_documents_2(all_files: List[str] = [], ignored_files: List[str] = []) -> List[Document]:
123
+ """
124
+ Loads all documents from the source documents directory, ignoring specified files
125
+ """
126
+ # all_files = []
127
+ # for ext in LOADER_MAPPING:
128
+ # all_files.extend(
129
+ # glob.glob(os.path.join(source_dir, f"**/*{ext}"), recursive=True)
130
+ # )
131
+ filtered_files = [file_path for file_path in all_files if file_path not in ignored_files]
132
+
133
+ with Pool(processes=os.cpu_count()) as pool:
134
+ results = []
135
+ with tqdm(total=len(filtered_files), desc='Loading new documents', ncols=80) as pbar:
136
+ for i, docs in enumerate(pool.imap_unordered(load_single_document, filtered_files)):
137
+ results.extend(docs)
138
+ pbar.update()
139
+
140
+ return results
141
+
142
+
143
  def process_documents(ignored_files: List[str] = []) -> List[Document]:
144
  """
145
  Load documents and split in chunks
 
161
  """
162
  print(f"Loading documents from {source_directory}")
163
  print("File Path to start processing:", file_path)
164
+ documents = load_documents_2(file_path, ignored_files)
165
  if not documents:
166
  print("No new documents to load")
167
  exit(0)
 
349
  def func_upload_file(files, chat_history):
350
  file_path = files
351
  print(file_path)
352
+ # print(chat_history)
353
+ test_msg = ["Request Upload File into DB", "Operation Ongoing...."]
354
+ chat_history.append(test_msg)
355
+ yield chat_history
356
+ UpdateDb()
357
+ test_msg = ["Request Upload File into DB", "Operation Finished"]
358
  chat_history.append(test_msg)
359
+ yield chat_history
360
 
361
  with gr.Blocks() as demo:
362
  main = gr.ChatInterface(
363
  chathmi2,
364
  title="STLA BABY - YOUR FRIENDLY GUIDE",
365
  description= "v0.3: Powered by MECH Core Team",
366
+ upload_button = gr.UploadButton("Upload File", file_count="multiple")
367
  )
368
+ main.upload_button.upload(func_upload_file, [main.upload_button, main.chatbot], main.chatbot)
 
369
 
370
  # demo = gr.Interface(
371
  # chathmi,