Print line number and/or filename for selected entries

[drowning-cat]

Hello everyone. I’m very new to jq (a complete newbie).

Let’s take the vitejs package.json file as an example and assume it has been downloaded locally as ./package.json. The final goal is to parse the "scripts" field and print the corresponding line numbers for each entry.

Convert this:

"scripts": {
  "preinstall": "npx only-allow pnpm",
  "postinstall": "simple-git-hooks",
  "format": "prettier --write --cache .",

Into something like this:

{
  { "line": 21, "key": "preinstall", "val": "npx only-allow pnpm" },
  { "line": 22, "key": "postinstall", "val": "simple-git-hooks" },
  { "line": 23, "key": "format", "val": "prettier --write --cache ." },

For simplicity, you can choose any output format that makes the most sense.

I know that jq provides something like input_line_number and input_filename, but I’m not sure how to use them. Ideally, the solution should rely only on jq, since the command is intended to be used in a Neovim/Lua configuration.

Is there any way to do this?

Any help would be greatly appreciated. Thank you in advance!

[wader]

Hi, sadly `input_line_number' does not give you the line number of individual value, it's the (last i think) line number of the whitepace-ish separated JSON values read. ex:

$ (echo -e '{"a":\n123}'; echo '"hello"')
{"a":
123}
"hello"

$ (echo -e '{"a":\n123}'; echo '"hello"') | jq -c '{input: ., line: input_line_number}'
{"input":{"a":123},"line":2}
{"input":"hello","line":3}

It is probably possible to use --stream somehow but i think that might be quite tricky

[drowning-cat]

Thank you for the response ❤️.

Sad, but I guess it’s only useful for very narrow use cases, not for "selecting", but rather for "finding" entries for further navigation and jumping.

I’ll leave the discussion unanswered, just in case, if you don’t mind.

[wader]

Had to try using streamed input and it somewhat works except for the last entry that ends up one line too far, not really sure why but it seems line number if counted from "last" part of the value somehow, which probably also hints on that a package.json with lots of whitespace might cause weird line numbers also,

$ jq --stream -n '[inputs | select((.[0] | .[0] == "scripts" and length > 1) and .[1]) as [[$_,$key], $script] | {$key, value: {$script, line: input_line_number}}] | from_entries' package.json
{
  "preinstall": {
    "script": "npx only-allow pnpm",
    "line": 21
  },
  "postinstall": {
    "script": "simple-git-hooks",
    "line": 22
  },
  "format": {
    "script": "prettier --write --cache .",
    "line": 23
  },
  "lint": {
    "script": "eslint --cache .",
    "line": 24
  },
....

[drowning-cat]

I guess the problem is that --stream returns the last entry twice:

jq --stream -c '.' package.json

[["scripts","ci-docs"],"pnpm build && pnpm docs-build"]
[["scripts","ci-docs"]

[drowning-cat]

Anyway, thank you for your help. I hope your answer will help others like me. If you come up with a solution to this problem later, please share it 🚀

[wader]

👍 yeah i'm not that familiar with the streaming parsing code but something fishy is going on:

$ jq --stream -c '[., input_line_number]' package.json | grep ci-docs
[[["scripts","ci-docs"],"pnpm build && pnpm docs-build"],41]
[[["scripts","ci-docs"]],41]