Implementation of multi-pdf chat

I have added a simple implementation of multi-pdf chats by making it possible to upload multiple pdfs via file upload. It's not possible to use multiple URLs yet.

Changes
- Making sure chunks get added from text of all files
- Added a [file number] parameter to each chunk
- Adapted the prompts to also cite the file number

app.py

@@ -18,6 +18,7 @@ def preprocess(text):
     return text
 
 
+# converts pdf to text
 def pdf_to_text(path, start_page=1, end_page=None):
     doc = fitz.open(path)
     total_pages = doc.page_count
@@ -35,8 +36,8 @@ def pdf_to_text(path, start_page=1, end_page=None):
     doc.close()
     return text_list
 
-
-def text_to_chunks(texts, word_length=150, start_page=1):
+# one text converts a list of chunks
+def text_to_chunks(texts, word_length=150, start_page=1, file_number=1):
     text_toks = [t.split(' ') for t in texts]
     page_nums = []
     chunks = []
@@ -49,7 +50,7 @@ def text_to_chunks(texts, word_length=150, start_page=1):
                 text_toks[idx+1] = chunk + text_toks[idx+1]
                 continue
             chunk = ' '.join(chunk).strip()
-            chunk = f'[Page no. {idx+start_page}]' + ' ' + '"' + chunk + '"'
+            chunk = f'[File no. {file_number}] [Page no. {idx+start_page}]' + ' ' + '"' + chunk + '"'
             chunks.append(chunk)
     return chunks
 
@@ -91,10 +92,12 @@ class SemanticSearch:
 
 
 
-def load_recommender(path, start_page=1):
+def load_recommender(paths, start_page=1):
     global recommender
-    texts = pdf_to_text(path, start_page=start_page)
-    chunks = text_to_chunks(texts, start_page=start_page)
+    texts = []
+    chunks = []
+    for idx, path in enumerate(paths):
+        chunks += text_to_chunks(pdf_to_text(path, start_page=start_page), start_page=start_page, file_number=idx+1)
     recommender.fit(chunks)
     return 'Corpus Loaded.'
 
@@ -140,7 +143,7 @@ def generate_answer(question, openAI_key, model):
         prompt += c + '\n\n'
         
     prompt += "Instructions: Compose a comprehensive reply to the query using the search results given. "\
-              "Cite each reference using [ Page Number] notation. "\
+              "Cite each reference using [File number][ Page Number] notation. "\
               "Only answer what is asked. The answer should be short and concise. \n\nQuery: "
     
     prompt += f"{question}\nAnswer:"
@@ -148,13 +151,15 @@ def generate_answer(question, openAI_key, model):
     return answer
 
 
-def question_answer(chat_history, url, file, question, openAI_key, model):
+def question_answer(chat_history, url, files, question, openAI_key, model):    
     try:
+        if files == None: 
+            files = []
         if openAI_key.strip()=='':
             return '[ERROR]: Please enter your Open AI Key. Get your key here : https://platform.openai.com/account/api-keys'
-        if url.strip() == '' and file is None:
+        if url.strip() == '' and files == []:
             return '[ERROR]: Both URL and PDF is empty. Provide at least one.'
-        if url.strip() != '' and file is not None:
+        if url.strip() != '' and files is not []:
             return '[ERROR]: Both URL and PDF is provided. Please provide only one (either URL or PDF).'
         if model is None or model =='':
             return '[ERROR]: You have not selected any model. Please choose an LLM model.'
@@ -163,11 +168,16 @@ def question_answer(chat_history, url, file, question, openAI_key, model):
             download_pdf(glob_url, 'corpus.pdf')
             load_recommender('corpus.pdf')
         else:
-            old_file_name = file.name
-            file_name = file.name
-            file_name = file_name[:-12] + file_name[-4:]
-            os.rename(old_file_name, file_name)
-            load_recommender(file_name)
+            filenames = []
+            for file in files:
+                old_file_name = file.name
+                file_name = file.name
+                file_name = file_name[:-12] + file_name[-4:]
+                os.rename(old_file_name, file_name)
+                filenames.append(file_name)
+            load_recommender(filenames)
+			
+                 
         if question.strip() == '':
             return '[ERROR]: Question field is empty'
         if model == "text-davinci-003" or model == "gpt-4" or model == "gpt-4-32k":
@@ -203,7 +213,7 @@ def generate_answer_text_davinci_003(question,openAI_key):
         prompt += c + '\n\n'
         
     prompt += "Instructions: Compose a comprehensive reply to the query using the search results given. "\
-              "Cite each reference using [ Page Number] notation (every result has this number at the beginning). "\
+              "Cite each reference using [File number] [ Page Number] notation (every result has this number at the beginning). "\
               "Citation should be done at the end of each sentence. If the search results mention multiple subjects "\
               "with the same name, create separate answers for each. Only include information found in the results and "\
               "don't add any additional information. Make sure the answer is correct and don't output false content. "\
@@ -212,6 +222,7 @@ def generate_answer_text_davinci_003(question,openAI_key):
               "answer should be short and concise. \n\nQuery: {question}\nAnswer: "
     
     prompt += f"Query: {question}\nAnswer:"
+    # print("prompt == " + str(prompt))
     answer = generate_text_text_davinci_003(openAI_key, prompt,"text-davinci-003")
     return answer
 
@@ -242,15 +253,14 @@ with gr.Blocks(css="""#chatbot { font-size: 14px; min-height: 1200; }""") as dem
     gr.Markdown(f'<center><h3>{title}</h3></center>')
     gr.Markdown(description)
 
-    with gr.Row():
-        
+    with gr.Row():     
         with gr.Group():
             gr.Markdown(f'<p style="text-align:center">Get your Open AI API key <a href="https://platform.openai.com/account/api-keys">here</a></p>')
             with gr.Accordion("API Key"):
                 openAI_key = gr.Textbox(label='Enter your OpenAI API key here', password=True)
-                url = gr.Textbox(label='Enter PDF URL here   (Example: https://arxiv.org/pdf/1706.03762.pdf )')
+                url = gr.Textbox(label='Enter PDF URL here  (Example: https://arxiv.org/pdf/1706.03762.pdf )')
                 gr.Markdown("<center><h4>OR<h4></center>")
-                file = gr.File(label='Upload your PDF/ Research Paper / Book here', file_types=['.pdf'])
+                files = gr.File(label='Upload your PDF/ Research Paper / Book here', file_types=['.pdf'], file_count="multiple")
             question = gr.Textbox(label='Enter your question here')
             gr.Examples(
                 [[q] for q in questions],
@@ -273,15 +283,12 @@ with gr.Blocks(css="""#chatbot { font-size: 14px; min-height: 1200; }""") as dem
         with gr.Group():
             chatbot = gr.Chatbot(placeholder="Chat History", label="Chat History", lines=50, elem_id="chatbot")
 
-
-#
+ 
     # Bind the click event of the button to the question_answer function
     btn.click(
         question_answer,
-        inputs=[chatbot, url, file, question, openAI_key, model],
+        inputs=[chatbot, url, files, question, openAI_key, model],
         outputs=[chatbot],
     )
 
-demo.launch()
-
-
+demo.launch()