smoke_signal/tessdata/README.md

# Smoke Signal Tessdata Drop Folder

Put custom Tesseract model files here as `.traineddata`.

Naming rule:
- Use lowercase + underscores based on detected font name.
- Example: `Gill Sans Infant` -> `gill_sans_infant.traineddata`

The OCR runtime will auto-pick models from either:
- `/tmp/smoke_signal/tessdata/`
- `smoke_signal/tessdata/` (this folder)

Important:
- `.otf` / `.ttf` files are not Tesseract models.
- A file like `gill_sans_infant.traineddata.otf` is still a font file, not usable by Tesseract OCR.

If your model name is custom and does not match slug, register it once:

```python
from smoke_signal.scripts.font_library import register_font_model
register_font_model("Gill Sans Infant", model_name="my_model_name")
```