jx is intended to be used to trivially extract fields from json data. Text table format is the default output format, but can be changed to TSV with the -t option, or to another delimiter with with the -d option.
jx will also auto-detect array input, and paginated input by looking at the first line, so you can send paginated results (with a total and items key in the first line), or first line starts with '['
use -s/--smart to disable this smartness
jx also provide (basic) flattening for extracting nested fields. nested json can be flattened with --flatten, and if so, nested keys can be references with the syntax key1 + . + key2 + index
e.g. name.first or addresses.0.zipcode
Use --join to join keys with a different delimiter, like "-"
jq is a much better alternative for advanced json processing, but for my very common use case of slicing out some fields, jx is more concise. I frequently use jq to pre-filter data, and then jx to extract fields.
jq '.[] | .firstname + " " + .lastname'
or
jq '.[] | "\(.firstname) \(.lastname)"'
vs.
jx firstname lastname
# extract the 'c' and 'a' fields from each object
$ printf '{"a": 1, "b": 2, "c":3}\n{"a":7,"b":2,"c":20}' | jx c a
c a
3 1
20 7
# columnar output is the default. No fields prints all values
$ printf '{"a": "foo", "b": 1}\n{"a":"loooooong", "b":2}' | jx
foo 1
loooooong 2
# no headers
$ echo '{"a": 1, "b": 2, "c":3}' | jx -H c a
3 1
# flatten with dot
# with -f or -F, should be easy to separate keys from parents
$ echo '{"a": {"b": 2, "c":3, "d":[5,6] }}' | jx -f a.c a.b
a.c a.b
3 2
# flatten with _
$ echo '{"a": {"b": 2, "c":3, "d":[5,6] }}' | jx -F a_c a_b
a_c a_b
3 2
# flatten with array indexing
$ echo '{"a": {"b": 2, "c":3, "d": [7,10]}}' | jx -F a_c a_d_0
a_c a_d_0
3 7
# get field names from first line
echo '{"a": 1, "b": 2, "c":3}' | jx --names
a
b
c
# alternate json level joiner. Default is '.'
$ echo '{"a": {"b": 2, "c":3, "d":[5,6] }}' | jx -F -j- a-c a-b
a-c a-b
3 2
# tab-separated output
$ echo '{"a":1, "b": 2, "c":3}' | jx -t a c
a c
1 3
# alternate output delimiter
$ echo '{"a":1, "b": 2, "c":3}' | jx -s '|' a c
a|c
1|3
# collapse whitespace in output fields, useful for columnar output
$ echo '{"first": "Jud D"}' | jx -w
first
Jud_D
# smart parsing: autodetect arrays
echo '[{"a":1},{"a":2}]' | jx a
a
1
2
# smart detecting multiline arrays (json pretty printed)
echo '[{"a":1},{"a":2}]' | jq | jx a
a
1
2
# smart parsing: autodetect paged sets (with items and total)
$ echo '{"total":2,"items":[{"a":1},{"a":2"}]}' | jx a
a
1
2
# smart detecting multiline paged sets (json pretty printed)
echo '{"items":[{"a":1},{"a":2}]}' | jq | jx a
a
2
# turn off smart parsing
$ echo '{"total":2,"items":[{"a":1},{"a":2}]}' | jx -s items
items
[{"a":1},{"a":2}]}
# extract all keys from all objects
# useful for exploring json data that may have different keynames across objects
printf '{"b":1,"a":2}\n{"b":2,"c":3}' | jx --all-keys
b
a
c
# combine with jq to extract from arbitrarily deep structures
curl https://dummyjson.com/products | jq .products | jx id price title
copy jx from the bin directory
Add a better install procedure.
Add positional field extraction. Will require validation of fields on every line to match the first line to prevent silently dropping/munging columns
Make the smart page set detection configurable. Better for environments that use different conventions to indicate paged sets. Note that you can use jq to extract only the objects you want, then extract the desired fields with jx:
jq '.permissions.users[]' | jx id name
Jud Dagnall github@dagnall.net