Suppose I have this ndjson file:
{"id": 99, "labeled_values": [["one", "green"], ["two", "red"], ["three", "blue"]]}
{"id": 100, "labeled_values": [["four", "green"], ["five", "red"]]}
How do I get the output below (tab-separated)? That is: id on every line, and the pairs of (value, label) flattened.
99 one green
99 two red
99 three blue
100 four green
100 five red
Here are two failed attempts:
$ cat tmp/foo.ndjson | jq -r '.id as $id | [$id, .labeled_values.[0], .labeled_values.[1]] '
[
99,
[
"one",
"green"
],
[
"two",
"red"
]
]
[
100,
[
"four",
"green"
],
[
"five",
"red"
]
]
$ cat tmp/foo.ndjson | jq -r '.id as $id | [$id, .labeled_values[].[0], .labeled_values[].[1]] '
[
99,
"one",
"two",
"three",
"green",
"red",
"blue"
]
[
100,
"four",
"five",
"green",
"red"
]
This question is very related, but I still don’t understand jq well enough.
A related question: how do I learn enough about the processing model of jq to understand how to break down nested structures like this into flat structures?
Most of the examples I find are flat and simple. They pick out single fields, not nested things.
>Solution :
.[] will create an output for each .labeled_values.
To convert .id to a string and add a \t use a string interpolation.
Use @tsv to convert the array to tsv
"\(.id)\t\(.labeled_values[] | @tsv)"