libmir / je

json to csv conversion. about 10x faster than jq.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Circle CI Docs

JE

Fast JSON to TSV/CSV/JSON/User-defined-format Extractor.

Build je

Requirements

  1. dub package manager.
  2. D compiler. Options:
  • LDC (LLVM D Compiler) >= 1.1.0-beta2
  • DMD (DigitalMars D Compiler) >= 2.072.1

Build with the dub package manager

To build project with LDC run following command from je project:

dub build --build=release-sse42 --compiler=ldmd2

or

dub build --build=release-native --compiler=ldmd2

To build project with DMD run

dub build --build=release

For more details run je --help.

Usage

After building, you can try je like this:

./je test.json --columns name:name,asdf:dependencies.asdf --input test.json
User defined output format
$ cat in.jsonl 
{"a":{"b":"\n"}, "d":2}
{"a":{"b":0}, "d":1}
{"a":{"b":2}}

# query with non-positional style 
$ ./je -c a.b,d -i in.jsonl --out=$'{"a":%s,"t":%s}\n'
{"a":"\n","t":2}
{"a":0,"t":1}
{"a":2,"t":null}

# query with positional style
./je -c a.b,d -i in.jsonl --out=$'{"a":%2$s,"t":%1$s}\n'
{"a":2,"t":"\n"}
{"a":1,"t":0}
{"a":null,"t":2}

##### [Probabilistic Linear Counting](https://github.com/tamediadigital/lincount)

```json
$ cat in.jsonl
{"a":{"b":0}, "d":1}
{"a":{"b":0}, "d":2}
{"a":{"b":"\n"}, "d":1}
{"a":{"b":"\n"}, "d":2}

$je -i in.jsonl --count=a.b --count=d --count="a.b&d"

{"counter":1,"count":2}
{"counter":2,"count":2}
{"counter":3,"count":4}
$ cat in.jsonl
{"a":{"b":10.0}, "d":null}
{"a":{"b":10.0}}
{"a":{"b":10.00}, "d":null}
{"a":{"b":10.00}}

$je -i in.jsonl --counta.b --count=d --count="a.b&d"
{"_counter_1":2,"_counter_2":1,"_counter_3":2}

// with named counters
$je -i in.jsonl --count=a.b:a.b --count=d --count="expr:a.b&d"
{"a.b":2,"_counter_2":1,"expr":2}
Unix Time-stamp partition
$ cat in.jsonl
{"a":{"ts":1485831253}, "param":0}
{"a":{"ts":1485831254}, "param":1}
{"a":{"ts":1485831255}, "param":1}
{"a":{"ts":1485831400}, "param":0}
{"a":{"ts":1485831401}, "param":1}

$ ./je -i in.jsonl --count=param --count-algo=timestamp --count-algo-params=a.ts,60
{"ts":1485831360,"_counter_1":2}
{"ts":1485831240,"_counter_1":2}

// with named counters
$ ./je -i in.jsonl --count=param:param --count-algo=timestamp --count-algo-params=a.ts,60
{"ts":1485831360,"param":2}
{"ts":1485831240,"param":2}

About

json to csv conversion. about 10x faster than jq.


Languages

Language:D 100.0%