Unparsing?

[jthrilly]

Sorry for the issue spam (maybe you could open the discussions feature up?) but I was wondering if you have plans for or interest in implementing unparsing (CSV to JS collection). I love the approach you have taken here, but I am looking for a parse/unparse combined solution in my project. I would be happy to try to contribute toward development.

[leeoniya]

hey fellow chicago dweller :)

yeah i think adding unparsing is fine, i kinda figured it would be requested eventually. im thinking something that uses the existing schema for options as much as possible/makes sense.

[leeoniya]

started some unparse work in https://github.com/leeoniya/uDSV/commit/bd533b0559a941c6d7ade83d25c2f72db506b1c8. currently:

• only unparsing from array of tuples or objects for now (still missing columnar and deep objects unparsing)
• no schema inference function and no header customization (just uses what's in supplied schema)
• perf looks to be about 2x-3x faster than Papaparse

not super sure about the API.

https://github.com/leeoniya/uDSV/commit/bd533b0559a941c6d7ade83d25c2f72db506b1c8#diff-4d5318447ff8d96281084320d0aa05527c243618044cede1bfa99881cba4a16d

• might need multiple schema inference functions (or opts) cause it's not possible to disambiguate deep vs shallow objects and tuples vs columnar arrays. inferArrs, inferObjs, inferCols, inferDeep.
• the current approach of allowing to pinky-swear away the need for quoting string fields doesnt feel great, and maybe is not necessary

[leeoniya]

(i've moved this WIP work to the unparse branch)