Refactor language handling and fix Tesseract OCR integration

[divanshu-go]

• Introduce macro-based Language enum with ISO and Tesseract codes

• Replaces old enum + TESSERACT_LANGUAGES array

• Compile-time mapping ensures type safety

• Supports optional Tesseract codes for unsupported languages

• Build language string in O(n) instead of nested searches

• Eliminates brittle string matching and formatting

Before

After

[divanshu-go]

Why it failed:

• Language::English (enum variant) was being compared to "english" (string)
• The PartialEq<&str> implementation used the Display trait, which returns ISO codes like "en" (not "english")
• So Language::English == "english" compared "en" == "english" → always false
• When languages were specified, the filter returned empty results (no matches found)
• Tesseract got called with -l (empty language parameter)
• Tesseract segfaulted when given an invalid language argument

Now when you specify languages, Tesseract gets proper language codes like -l eng+fra+deu instead of crashing with -l . The macro system ensures type-safe, compile-time guaranteed language mappings with no runtime string operations.

[divanshu-go]

@louis030195

[louis030195]

awesome

[divanshu-go]

@louis030195 any tip on this one.

[louis030195]

we don't use OCR, nor linux, so it's not very useful to us @divanshu-go

[divanshu-go]

no problem.