search

miyuchina / mistletoe

A fast, extensible and spec-compliant Markdown parser in pure Python.
MIT License
841 stars 119 forks source link

BaseRenderer.render_line_break does not work without override #208

Open victorvess opened 11 months ago

victorvess commented 11 months ago

Ran into an issue implementing a simple renderer for my project - someone submitted from Windows with CRLF data (similar to what is described in Issue #64 ) and caught me out during testing.

I can handle normalizing my own newlines, but I noticed that BaseRenderer.render_line_break just calls render_inner(token) which cannot succeed, because LineBreak token has no children. This feels like a bug - there should be a safer default (space?? empty string??) or something else here.

I'm happy to contribute a PR for this but I'm curious your opinion on the ideal behavior?

pbodnar commented 11 months ago

Hi @victorvess, thanks for the report.

It looks like the problem of missing children on the LineBreak token is there from the very beginnings of the project. But it went unnoticed, because every *Renderer class has got its own implementation of the render_line_break method, implementation which is appropriate for the target document language.

So, I would infer from this, that probably the best ("safest") we can do now is to change BaseRenderer.render_line_break implementation to raise NotImplementedError() (already used elsewhere in the code) or similar, so that it is clear that this method needs to be implemented specifically for a given language.

What do you think?