What screenplay-parser is good for, and what we'd accept as contributions

[mcqx4]

A common need in AI pre-production tools is parsing screenplay files. We open-sourced screenplay-parser (this repo) because every team rebuilding an AI storyboard tool, script-analysis tool, or shot-list generator hits the same problem and wastes a week on .fdx XML parsing.

This library is the same code we use in STORYLINER to preprocess scripts before our character-memory-conditioned image generation pipeline. Sharing because if you're building in this space, you shouldn't need to write this layer.

What we'd love to see contributed

• Celtx CSV export parser — Celtx exports differ from Fountain
• PDF→Fountain heuristic — for the common case of getting only a PDF screenplay
• Better main-character detection — currently counts dialogue cues, doesn't weight by speech length
• More robust scene-heading detection for Fountain (currently uses regex on INT/EXT prefix)

What we deliberately won't add

• Heavyweight dependencies (we want pure-stdlib parsing)
• AI-based parsing — defeats the point of this library
• DOCX/RTF parsing — those formats need their own libraries

If you're using this in production, drop a note here so we can think about API stability before tagging v1.0.