JSON is an inherently hierarchical structure, which makes searching it for path information difficult. The JQG script flattens the hierarchical structure so that the path for each JSON end node is represented as a single string, thereby enabling easy searching and producing contextually meaningful results. It also produces valid JSON, which can be further processed, as needed.
For searching, JQG uses the PCRE engine built into JQ, which is much more powerful than grep or egrep (and it's certainly easier to use). For added flexibility, JQG can read from STDIN instead of from a file, allowing it to be used in pipelines, too. Finally, there are many options to control what is searched and how, as well as the format and content of the output.
Alternately, JQG can unflatten JSON that has been previously flattened (or structured to look that way) and it can also extract a subset of the JSON input, including the extraction terms in the results. Both of these alternatives can be combined with searching in a "composite mode". The convenience scripts jqu and jqx are provided to invoke those composite modes more easily.
$ jq . odd-values.json
{
"one": {
"start-string": "foo",
"null-value": null,
"integer-number": 101,
"string-with-pipe": "this|that",
"key|with|pipe": true,
"string-with-parens": "(this and that)",
"key(with)parens": true,
"bare-parens()": true,
"left(paren-only": true,
"unmatched-left)-paren": false,
"dollar $ign": "both-sides-$now"
},
"two": [
{
"two-a": {
"non-integer-number": -101.75,
"number-zero": 0
},
"true-boolean": true,
"two-b": {
"false-boolean": false
}
},
{
"two-c": {
"alpha-num-1": "a1",
"alpha-num-2": "2b",
"alpha-num-3": "a12b"
}
}
],
"three": {
"empty-string": "",
"empty-object": {},
"empty-array": []
},
"four": [
"first",
null,
{},
999,
"fourth"
],
"end-string": "bar"
}
# some not-too-useful jq/grep results
$ jq . odd-values.json | grep string
"start-string": "foo",
"string-with-pipe": "this|that",
"string-with-parens": "(this and that)",
"empty-string": "",
"end-string": "bar"
$ jq . odd-values.json | grep 0
"integer-number": 101,
"non-integer-number": -101.75,
"number-zero": 0
$ jq . odd-values.json | grep 'int\|false'
"integer-number": 101,
"unmatched-left)-paren": false,
"non-integer-number": -101.75,
"false-boolean": false
# much more useful jqg results
$ jqg string odd-values.json
{
"one.start-string": "foo",
"one.string-with-pipe": "this|that",
"one.string-with-parens": "(this and that)",
"three.empty-string": "",
"end-string": "bar"
}
$ jqg -v 0 odd-values.json
{
"one.integer-number": 101,
"two.0.two-a.non-integer-number": -101.75,
"two.0.two-a.number-zero": 0
}
$ jqg 'int|false' odd-values.json
{
"one.integer-number": 101,
"one.unmatched-left)-paren": false,
"two.0.two-a.non-integer-number": -101.75,
"two.0.two-b.false-boolean": false
}
# The power of PCRE
# - search values looking for a 0 without a preceding number
$ jqg -v '(?<!\d)0' odd-values.json
{
"two.0.two-a.number-zero": 0
}
# - the same or an empty array
$ jqg -v '(?<!\d)0|\[]' odd-values.json
{
"two.0.two-a.number-zero": 0,
"three.empty-array": []
}
# can be used in pipelines, too
$ curl -s https://raw.githubusercontent.com/NorthboundTrain/jqg/main/test/odd-values.json | jqg -v '(?<!\d)0|\[]'
{
"two.0.two-a.number-zero": 0,
"three.empty-array": []
}
# flatten & search
$ jqg 'int|false' odd-values.json
{
"one.integer-number": 101,
"one.unmatched-left)-paren": false,
"two.0.two-a.non-integer-number": -101.75,
"two.0.two-b.false-boolean": false
}
# flatten & search, then unflatten
$ jqg 'int|false' odd-values.json | jqg -u
{
"one": {
"integer-number": 101,
"unmatched-left)-paren": false
},
"two": [
{
"two-a": {
"non-integer-number": -101.75
},
"two-b": {
"false-boolean": false
}
}
]
}
# regular JQ command
$ jq .two odd-values.json
[
{
"two-a": {
"non-integer-number": -101.75,
"number-zero": 0
},
"true-boolean": true,
"two-b": {
"false-boolean": false
}
},
{
"two-c": {
"alpha-num-1": "a1",
"alpha-num-2": "2b",
"alpha-num-3": "a12b"
}
}
]
# JQG's extract retains the selector given
$ jqg -x .two odd-values.json
{
"two": [
{
"two-a": {
"non-integer-number": -101.75,
"number-zero": 0
},
"true-boolean": true,
"two-b": {
"false-boolean": false
}
},
{
"two-c": {
"alpha-num-1": "a1",
"alpha-num-2": "2b",
"alpha-num-3": "a12b"
}
}
]
}
# extract deep into a structure
$ jqg -x '.two[0]."two-b"' odd-values.json
{
"two": [
{
"two-b": {
"false-boolean": false
}
}
]
}
# composite mode - extract & search
$ jqx .two 'int|false' odd-values.json
{
"two.0.two-a.non-integer-number": -101.75,
"two.0.two-b.false-boolean": false
}
Many more examples are provided in jqg-examples.md.
The JQG script is entirely self-contained except for the need to have both jq and bash on the system somewhere; bash itself needs to be on your $PATH, but jq does not -- see the documentation for more details on how to work when jq is not in your $PATH.
It's easy to get the latest stable version of the script. First decide where you want to put it, and then grab it using wget:
cd /path/to/script/dir
wget https://github.com/NorthboundTrain/jqg/raw/main/src/jqg
wget https://github.com/NorthboundTrain/jqg/raw/main/src/jqu
wget https://github.com/NorthboundTrain/jqg/raw/main/src/jqx
chmod +x jqg jqu jqx
Alternately, you can clone the whole repo:
# HTTP
cd /path/to/git/parent/dir
git clone https://github.com/NorthboundTrain/jqg.git
# SSH
cd /path/to/git/parent/dir
git clone git@github.com:NorthboundTrain/jqg.git
If you want to run the unit tests, you will also need the BATS sub-modules; you can clone them at the same time by adding in the --recurse-submodules option for git clone:
git clone --recurse-submodules https://github.com/NorthboundTrain/jqg.git
git clone --recurse-submodules git@github.com:NorthboundTrain/jqg.git
- Bash 3.0.27+
- JQ 1.6+
Execute JQG against a specific JSON file:
jqg search-string foo.json
jqg --unflatten flat.json
jqg --extract .some_elem foo.json
Execute JQG in a pipeline:
curl -s 'https://api.github.com/repos/NorthboundTrain/jqg' | jqg 'name|count'
- jqg.md - the JQG man page
- jqg-examples.md - an exhaustive look at the different invocation methods as well as each command line option
- jqg-filters.md - all JQG filters fully annotated
- runing-tests.md - information on running the provided unit tests
see CHANGELOG.md
The filters to convert the hierarchical structure of JSON into a flat structure are largely based on a blog post from Fabian Keller entitled 5 Useful jq Commands to Parse JSON on the CLI.
The core of the unflatten filter is taken from a StackOverflow answer given by user pmf.
The core of the extract filter is taken from a StackOverflow answer given again by user pmf.
This project uses code from the following projects:
- pure-getopt - a drop-in replacement for GNU
getopt, written in Bash - bats-core - Bash Automated Testing System
Bugs & feature requests are tracked as GitHub issues.
- Check the open issues or open a new issue to start a discussion around your feature idea or the bug you found
- Fork the repository and make your changes
- Make sure all unit tests pass (and add new ones, if appropriate)
- Update documentation as needed
- Open a new pull request
Apache-2.0
© 2021 Joseph Casadonte