Set field color using JQ_COLORS

[haguenau]

This should address #1739.

As far as I can tell, this works and passes the test suite (which I updated to test this specific addition). There is a chance that the code changes are not perfect; for example I'm not sure how much sense it makes to have introduced an enum member JV_KIND_FIELD for this (although it keeps the code mostly regular).

[coveralls]

Coverage remained the same at 84.457% when pulling 0d53abc on haguenau:feature/set-field-color into 4b4fefa on stedolan:master.

[nicowilliams]

Interesting. Obviously there is no use for this new "kind" other than as an index into the array of colors. I'll think about it. Thanks for your contribution!

[haguenau]

Cool. Please don't hesitate to let me know if you see changes that you think would make sense.

[ericpruitt]

@nicowilliams , would you be willing to merge an updated version of this patch or even #1977?

[ericpruitt]

I'm attaching a patch that's a updated version of this PR for anyone that builds jq themselves. Thanks, @haguenau, for the original code.

jq-src-JQ_COLORS-field-color.patch.txt

EDIT: I noticed that in cf61515, jq.1.prebuilt was not updated with the new default scheme.

[t829702]

is this abandoned?

[ericpruitt]

Depends on what you mean by abandoned. I'm still using this patch, and the last time I synchronized my copy of jq with the upstream repo (December), the patch still applied cleanly, so I haven't needed to post any updates here.

[zpangwin]

is this abandoned?

I see commits on the repo as recently as a month and a half ago... guessing the dev's are just busy / this isn't a high priority for them. I'm not familar with travis builds but I see that it has something flagged for:

continuous-integration/travis-ci/pr — The Travis CI build could not complete due to an error"

It could be something as simple as that holding things back.

Anyway, I hope folks are able to make it happen too... can't stand the hard-to-read dark blue key/property names on my terminal. Even if the default was a bright purple, that would at least have decent contrast on light and dark backgrounds. Obviously customizing would be better but I probably wouldn't have bothered to look this up / subscribe to the ticket and PR if the defaults were more reasonable :-)

Guess it's back to alias jq='/usr/bin/jq --monochrome-output' for me until this makes it in / someone posts idiot-proof build instructions that I can follow for patching lol

[haguenau]

I have not followed jq development very closely since submitting this pull request two years ago. I must admit I do not fully understand nicowilliam's reasons for having to think more about this; I assume the change is somehow unclean.

If there is interest, I would be happy to look into reworking the change so that it applies successfully to the current code.

[fdevibe]

I would, too, be very interested in hearing counter-arguments against this. The hard-coded field color is very difficult to read on a dark background, and dark backgrounds are pretty common in what I would assume jq's user base is.

Changing the field color is also something quite a few people request in https://stackoverflow.com/questions/51338701/how-do-i-customize-the-colors-used-by-jq-c

[NickCis]

Really looking into getting this merge, Can I offer any help with it?.

[itchyny]

Added to the next release milestone. Needs rebasing though.

[ericpruitt]

I'll open a PR later today with my latest version of the patch.

[haguenau]

I'll open a PR later today with my latest version of the patch.

Much appreciated. Given how old this pull request is, I expect a lot of changes would be needed to make it up to date, so I am very happy to see you offer to take care of this.

[ericpruitt]

I've opened #2638.

[itchyny]

Taking a careful look into this patch, I finally understand why this is hard to merge. This patch adds JV_KIND_FIELD to jv_kind, but this is not what jv_kind is intended. This enum type is a basic and core type of JSON representation in jq, and such a new value shouldn't be added just for colors. So I propose; do not touch jv_kind and color_kinds, but color_bufs can have 8 (==sizeof(color_kinds)/sizeof(color_kinds[0]) + 1) entries, and FIELD_COLOR can be colors[sizeof(color_kinds)/sizeof(color_kinds[0])] or simply colors[7].