Bridgeconn
Improve performance of conversion
The parsing with the tree-sitter module is quite fast even for large and complex usfm files. Then we do a sequential parsing of the output syntax tree to convert them to USX , JSON etc. In doing so, the performance is greatly affected. Need to look into some alternate programming methodologies like callbacks to improve this.
Yes, just to give some real world stats:
https://github.com/schierlm/BibleMultiConverter can do USFM->USX conversion for a whole Bible in ~6 seconds.
usfm-grammar (Node) takes 3-60 seconds per book, so probably ~2000 seconds for whole Bible. I didn't run the whole thing as might have taken half an hour.
But there's different use cases, and it looks like this could be really useful for a more feature rich converter. I'll be keen to hear if there are performance improvements.
Note: In python could use https://docs.python.org/3/library/profile.html to find out where improvement is needed