https://github.com/pitr/fj
Flatten JSON
https://github.com/pitr/fj
Last synced: 10 months ago
JSON representation
Flatten JSON
- Host: GitHub
- URL: https://github.com/pitr/fj
- Owner: pitr
- License: mit
- Created: 2021-06-30T22:19:39.000Z (almost 5 years ago)
- Default Branch: main
- Last Pushed: 2023-12-31T16:26:09.000Z (over 2 years ago)
- Last Synced: 2025-04-14T12:05:35.639Z (about 1 year ago)
- Language: Go
- Size: 186 KB
- Stars: 8
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# fj
Flatten your JSON.
`fj` flattens JSON into a value and its full path per line, so data can be worked on using UNIX tools such as grep/awk/cut/etc.
❯ curl -s "https://api.github.com/repos/golang/go/commits?per_page=1" | fj | grep commit.author
json[0].commit.author.name "Josh Bleecher Snyder"
json[0].commit.author.email "josharian@gmail.com"
json[0].commit.author.date "2021-06-30T16:44:30Z"
fj can un-flatten too, which is useful to get a subset of JSON:
❯ curl -s "https://api.github.com/repos/golang/go/commits?per_page=1" | fj | grep commit.author | fj -u | jq
[
{
"commit": {
"author": {
"name": "Josh Bleecher Snyder",
"email": "josharian@gmail.com",
"date": "2021-06-30T16:44:30Z"
}
}
}
]
## Installation
1. Download [latest release for Linux, Mac, Windows or FreeBSD](https://github.com/pitr/fj/releases),
2. Put the binary in your `$PATH` (e.g. in `/usr/local/bin`) to make it easy to use:
Alternatively, if you have Go compiler:
```
❯ go install pitr.ca/fj@latest # OR
❯ go get -u pitr.ca/fj # legacy way
```
## Usage
Flatten JSON file or stdin:
```
❯ fj file.json
❯ cat file.json | fj
```
Or many JSON files. Use `-s` or `--stream` to treat input as a stream of JSON documents:
```
❯ fj -s file1.json file2.json file3.json
```
Use `grep` to search, `diff <(fj file1.json) <(fj file2.json)` to diff, and other tools such as `awk/sort/uniq/fzf`.
## FAQ
### Why shouldn't I just use gron?
[gron](https://github.com/tomnomnom/gron) is a very similar tool. However, `fj` is different from it in a few key ways:
- fj does not keep the whole JSON object in memory, which allows it to be 10-20 times faster than gron.
```
❯ hyperfine --warmup 5 'gron testdata/big.json' 'fj testdata/big.json'
Benchmark #1: gron testdata/big.json
Time (mean ± σ): 136.2 ms ± 1.9 ms [User: 57.1 ms, System: 97.2 ms]
Range (min … max): 132.9 ms … 140.2 ms 21 runs
Benchmark #2: fj testdata/big.json
Time (mean ± σ): 9.1 ms ± 0.9 ms [User: 4.9 ms, System: 2.4 ms]
Range (min … max): 7.8 ms … 12.2 ms 203 runs
Summary
'fj testdata/big.json' ran
14.93 ± 1.48 times faster than 'gron testdata/big.json'
```
- `fj` does not preserve array structures by padding with null.
```
❯ echo '[1,2,3,4,5]' | gron | grep 5 | gron -u
[
null,
null,
null,
null,
5
]
❯ echo '[1,2,3,4,5]' | fj | grep 5 | fj -u
[5]
```
- `fj` does not try to convert JSON into valid JavaScript that can be pasted into JS console. There are no extra semicolons and array/object initializations.
### Why shouldn't I just use jq?
[jq](https://stedolan.github.io/jq/) is a great and powerful tool with its own language. `fj` simply flattens (and un-flattens) JSON, and is expected to integrate with existing tools.