feat: uri decode function

[fmgornick]

hi! i love this tool and have been using it for years, but recently noticed it doesn't have built in uri decoding. i didn't feel like writing a bash script or installing another program for something this trivial, plus i thought it'd be a nice addition to jq, so i thought i'd try adding it!

I didn't add too many tests because i wanted to keep it consistent with the other format strings, but i did personally test a few files in the src dir, and i also tested it on this file containing every unicode point.

closes #798, closes #2261.

[wader]

Hey, i think that makes sense to add. gojq has is already and it seems like jaq used to have it 01mf02/jaq#105 but got removed as jq didn't have it

[emanuele6]

I also think it makes sense to add a @urid since there is also @base64d; it can be useful.

[itchyny]

Should + be converted to (space) or no?

[fmgornick]

Should + be converted to (space) or no?

it seems like real percent encoding uses %20 as the space encoding, and only application/x-www-form-urlencoded form data uses +. I have no opinion but since the @uri formatter uses %20 i figured it'd be best to keep it consistent

[emanuele6]

The += syntax is specify to query strings; if users want it, they can just gsub("\\+";" ") before @urid.

[itchyny]

I think @uri is commonly used for query string (@uri"q=\(.)"), so @urid can be used to decode them. They often use + for spaces (like ?q=jq+manual&sourceid=chrome&ie=UTF-8 which was copied from Chrome address bar).

[fmgornick]

I think @uri is commonly used for query string (@uri"q=\(.)"), so @urid can be used to decode them. They often use + for spaces (like ?q=jq+manual&sourceid=chrome&ie=UTF-8 which was copied from Chrome address bar).

looking at the code for @uri, it just uses percent encoding for any char not in the unreserved buffer ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~, since isn't unreserved, it gets encoded to %20.

If people want i can change the behavior of both @uri and @urid to encode to + and vice versa, but there prolly needs to be some more discussion there 🤷‍♀️

[wader]

Played around a bit with how go does it https://go.dev/play/p/Itnr8EfdOjK for the query it seem to decode both + and %20 as whitespace. Should we do that but only encode to %20? otherwise i guess we need separate formats (or arg?) for url parts and query encode/decode?

For reference this is how gojq behaves atm, uses query behavior in both cases

$ gojq -n '"+%20" | @urid'
"  "
$ gojq -n '" " | @uri'
"+"

[emanuele6]

+= is specific to queries and is completely wrong in all other cases; i don't think @uri/@urid should do that conversion; there could be dedicated @query and @queryd formats

[itchyny]

Understood. These formats are for path segments, not for query params.

[wader]

@fmgornick Thanks for working on this! 🥳

[fmgornick]

@fmgornick Thanks for working on this! 🥳

of course! thank you all for providing such constructive feedback 😊