Language Typography & Character Mapping for AI Text Cleanup

This project provides language-specific mapping tables to standardize and normalize typographic and special characters in AI-generated or processed texts.
Mappings cover characters that are regionally uncommon, discouraged, or problematic for further text processing.

Why?

• Many LLMs and online AIs output Unicode characters that are not desirable or conventional for end users in specific countries (e.g. curly quotes, em dashes, narrow spaces, ß, etc).
• This repo defines clear rules to convert these to standard, widely-supported, or country-preferred equivalents.
• Main use case: LLM or AI text pre/post-processing in privacy-focused, in-browser, or server-side tools.

Usage

Import the JSON mapping for your target language/region and apply the character replacements to your text pipeline.

Usage Example (pseudo code):

for char, replacement in mapping.items():
    text = text.replace(char, replacement)

Supported Languages

• Swiss German (swiss-german.json)
• German (german.json)
• French (french.json)
• Italian (italian.json)
• English (International) (english-international.json)
• English (US) (english-us.json)

Contributing

Feel free to open issues or PRs for new mappings, edge cases, or country-specific improvements!