Pijush2023 commited on
Commit
2c0e685
·
verified ·
1 Parent(s): e6f0b98

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +12 -17
app.py CHANGED
@@ -604,8 +604,6 @@ import re
604
 
605
  # return final_response
606
 
607
- import re
608
-
609
  def clean_response(response_text):
610
  # Remove system and user tags
611
  response_text = re.sub(r'<\|system\|>.*?<\|end\|>', '', response_text, flags=re.DOTALL)
@@ -624,9 +622,9 @@ def clean_response(response_text):
624
 
625
  # Remove the entire 'Document(metadata=...' and any mention of it from the response
626
  response_text = re.sub(r'Document\(metadata=\{.*?\},page_content=', '', response_text, flags=re.DOTALL)
627
-
628
- # Remove any mention of "Document:" in the response
629
- response_text = re.sub(r'- Document:.*', '', response_text)
630
 
631
  # Remove any unwanted escape characters like \u and \u00
632
  response_text = re.sub(r'\\u[0-9A-Fa-f]{4}', '', response_text)
@@ -638,24 +636,21 @@ def clean_response(response_text):
638
  # Remove the phrase "Sure! The Responses are as follows:" from the actual content
639
  response_text = re.sub(r'Sure! The Responses are as follows:', '', response_text).strip()
640
 
641
- # Extract and limit to top 5 results
642
- result_lines = response_text.splitlines()[:5]
643
- top_5_results = "\n".join(result_lines)
644
-
645
  # Clean up the text by removing extra whitespace
646
  cleaned_response = re.sub(r'\s+', ' ', response_text).strip()
647
 
648
- # Format the final response with bullet points in Italian
649
  final_response = f"""
650
- Nome del documento: {document_name}
651
- Numero di pagina: {page_number}
 
 
 
 
 
652
 
653
- Risultati principali (Top 5):
654
- {top_5_results}
655
 
656
- Risultato effettivo:
657
- {cleaned_response}
658
- """
659
 
660
 
661
 
 
604
 
605
  # return final_response
606
 
 
 
607
  def clean_response(response_text):
608
  # Remove system and user tags
609
  response_text = re.sub(r'<\|system\|>.*?<\|end\|>', '', response_text, flags=re.DOTALL)
 
622
 
623
  # Remove the entire 'Document(metadata=...' and any mention of it from the response
624
  response_text = re.sub(r'Document\(metadata=\{.*?\},page_content=', '', response_text, flags=re.DOTALL)
625
+
626
+ # Extract top 5 fetched results (based on some identifier you have)
627
+ top_results = response_text.split('\n')[:5] # Assuming top 5 results are first 5 lines of content
628
 
629
  # Remove any unwanted escape characters like \u and \u00
630
  response_text = re.sub(r'\\u[0-9A-Fa-f]{4}', '', response_text)
 
636
  # Remove the phrase "Sure! The Responses are as follows:" from the actual content
637
  response_text = re.sub(r'Sure! The Responses are as follows:', '', response_text).strip()
638
 
 
 
 
 
639
  # Clean up the text by removing extra whitespace
640
  cleaned_response = re.sub(r'\s+', ' ', response_text).strip()
641
 
642
+ # Format the final response with bullet points
643
  final_response = f"""
644
+ Sure! Here is the response for your Query:
645
+ Document name - {document_name}
646
+ • Page No - {page_number}
647
+ • Top 5 Fetched Results:
648
+ {''.join([f'{i+1}. {result.strip()}\n' for i, result in enumerate(top_results)])}
649
+ • Actual Response: {cleaned_response}
650
+ """
651
 
652
+ return final_response
 
653
 
 
 
 
654
 
655
 
656