How to map split strings from file into fields with jq?

[drankinatty]

This seems simple (and it is for those skilled with jq), but my experience is lacking. I have a simple file that contains words and definitions, one per-line, pipe ('|') separated, e.g.:

$ head -n1 words.txt
Appreciable|describes things that can be perceived or measured.

I am trying to read the file into a json document with the normal array of arrays structure but giving the words and definitions tags of word: and defn:. The structure will be:

[
  [
    "word": "Appreciable",
    "defn": "describes things that can be perceived or measured."
  ]
]

Testing with input to work out the filter, I can get the data in the form needed, but I cannot figure out how to add the tags. For example using split ("|") produces the document I want:

$ jq -R -n '[input | split("|")]' words.txt
[
  [
    "Appreciable",
    "describes things that can be perceived or measured."
  ]
]

But when I try to map() the fields, I end up with errors of Cannot index string with number or the like. I can't sort out how I need to continue the pipeline to ... | map ({word: <what here?>}, {defn: <what here?>}) to complete the document. I've gone through the 1.7 manual and I cannot find how to do what I'm trying to do. (that tells me I don't know enough to recognize what I'm looking at)

I suspect I need to take the result of split() and create an array with it (something like as [$word, $defn] and then be able to complete the map with the variables -- but the implementation escapes me.

How do I simply add the "word": and "defn": tags to the split strings?

[wader]

Hi, your close! maybe something like this?

$ echo -e 'a|b\nc|d' | jq -nR '[inputs | split("|") | {word: .[0], defn: .[1]}]'
[
  {
    "word": "a",
    "defn": "b"
  },
  {
    "word": "c",
    "defn": "d"
  }
]

[drankinatty]

Aaarrrgggghhh! Brilliant. I didn't need map() at all, it was just providing the tags!

Thank you! It's funny, I'm quite adept with sed, awk, POSIX shell, C, C++, FORTRAN, assembly, etc.., but jq is something I don't use enough to have made friends with it yet. I'd spent an hour or so going though posts on StackOverflow, Unix & LInux, and the wider web and pawing through the jq manual -- and I had it stuck in my head that map() was the key. One certain way for humans to waste time is to start off down the wrong path.... It has the momentum of an aircraft carrier at that point.

Thank you for getting me back on the proper course.

[wader]

👍 it's a bit special with its tacit syntax and generators, i can recommend having a look at https://github.com/jqlang/jq/blob/master/src/builtin.jq, that was enlightening for me :)

[pkoppstein]

@drankinatty - Nice question. I hope you'll find your jq explorations worthwhile. For future reference, please ask usage questions at stackoverflow.com, not least because that will likely help others find good questions and helpful answers.

[drankinatty]

I wondered about that. I've contributed on SO for a decade and the only reason I asked here, was my search turned up Simplest way to add content of the text file as json string field using jq which led me to ask here. Will post all additional usage questions (hopefully few after I wrap my head around the pipeline order and flow), on StackOverflow. Sorry for the noise.

[wader]

I used to contribute a lot to SO also but felt it got so competitive and hostile at times so i stopped :( so personally i would not mind if the jq discussions was used more and we built a knowledge base here.