ChatGPT/Tool/Idea/Script/To Clean Up Output of Canvas

• script to clean up NNBSP and other garbage characters from documents produced by #ChatGPT

  • prior art

    • Humanize Your AI Text in One Click for Free
      • replaces the wrong thing
      • not a script - online tool
    • Person/unixwzrd/GitHub/UnicodeFix https://unixwzrd.ai/
      • Code
        • UnicodeFix/bin/cleanup-text.py at main · unixwzrd/UnicodeFix
      • Person/unixwzrd/Project/Unicodefix UnicodeFix | Distributed Thinking Systems

        • What UnicodeFix Does

          • Converts smart quotes, em-dashes, en-dashes, ellipses, and other punctuation into standard ASCII equivalents
          • Removes invisible or disruptive Unicode characters, including:
          • U+200B (Zero-Width Space)
          • U+200C (Zero-Width Non-Joiner)
          • U+200D (Zero-Width Joiner)
          • U+2018 (Left Single Quotation Mark ‘ )
          • U+2019 (Right Single Quotation Mark ’ )
          • U+201C (Left Double Quotation Mark “ )
          • U+201D (Right Double Quotation Mark ” )
          • U+2013 (En Dash – )
          • U+2014 (Em Dash — )
          • U+2026 (Horizontal Ellipsis … )
          • Cleans AI-generated, copied, or web-sourced content to prevent downstream parsing errors
          • Reduces risks when working across editors, file formats, and systems

            
            Bonus Trouble makers
            

          • U+FEFF (Byte Order Mark, BOM) — invisible but can totally screw up parsing in some languages.
          • U+00A0 (Non-Breaking Space) — looks like a regular space but acts differently in many contexts.

  • observations

    • perhaps having a tool like Py/Lib/black but for Markdown would be best; ChatGPT/Tool/Idea/Script/To Clean Up Output of Canvas/Report summarizes these