Skip to content
TheChatArchiveThe Chat Archive
Kimikimi-k2-0711-preview· Moonshot AIHugging Face LMArena Human Preference 140K

intead of this pattern Week X 20XX, consider week X. and it not case senstive and need…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Claudeclaude-3-5-sonnet-20241022’s answer.
  1. User

    intead of this pattern Week X 20XX, consider week X. and it not case senstive and need to remove any symbols or additional space found in the ocr data. and in excel file save "Week + week number from the OCR data

  2. Kimi Assistant

    Below is a minimal patch that does exactly what you asked:

    • looks for the first occurrence of “week” (case-insensitive) followed by a number
    • strips every non-alphanumeric character from the OCR text before the search
    • writes the week into the Excel file as “Week ” (e.g. “Week 1”)

    Only two small code blocks change – everything else stays the same.


    1. New helper (drop it in just above process_timesheets)

    def extract_week_from_text(text: str) -> str:
        """
        Return the first week number found in the OCR text.
        Format returned: 'Week <number>'  (e.g. 'Week 1')
        """
        # 1. Remove every character that is not a letter, digit or space
        cleaned = re.sub(r'[^A-Za-z0-9\s]', ' ', text)
    
        # 2. Case-insensitive search for 'week' followed by digits
        m = re.search(r'week\s+(\d+)', cleaned, flags=re.IGNORECASE)
        if m:
            return f"Week {int(m.group(1))}"
        return "Week Unknown"
    

    1. Two tiny edits inside process_timesheets

    a) Remove the old filename-based week extraction:

    # OLD (delete these two lines)
    file_match = re.match(r"(.+?)_Week(\d+)_\d+\.(png|jpg|jpeg)$", filename, re.IGNORECASE)
    if file_match:
        name, week = file_match.group(1), file_match.group(2)
    else:
        name = filename.rsplit('.', 1)[0]
    

    Replace with:

    # NEW
    name = os.path.splitext(filename)[0]          # keep only the base name
    

    b) Call the new helper and use its result:

    # NEW
    week = extract_week_from_text(text)
    

    That’s it – the Excel file will now contain “Week 1”, “Week 2”, … taken directly from the OCR text, regardless of how the image is named.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.