pjy - JSON Python processor

pjy is a command-line tool to process JSON data and execute queries on it. It is a bit like jq but with a Python syntax for queries.

Install

From PyPI:

pipx install pjy

With colors:

pipx install "pjy[pygments]"

Usage

pjy [EXPR] [FILES]

pjy will read JSON data from FILES and print the evaluation result of the Python expression EXPR.

If FILES is missing or is "-", pjy will use stdin.

The simplest expression to use, which outputs the input unchanged is "d" (for data).

It's possible to use multiple input files.

Examples

In pjy, expressions are also called "filters", as in jq.

Just pretty-print

d (short for "data") is the most basic filter, it represents the whole input:

pjy 'd'
    {"foo":"bar","baz":[1,2,3]}

Prints:

{
  "foo": "bar",
  "baz": [
    1,
    2,
    3
  ]
}

Select a dict key

The filters are Python expressions, hence we can select a dict key:

pjy 'd["baz"]'
    {"foo":"bar","baz":[1,2,3]}

Alternatively, in pjy, dicts keys are also attributes:

pjy 'd.baz'
    {"foo":"bar","baz":[1,2,3]}

Both filters will print:

[
  1,
  2,
  3
]

In case a key has a reserved name, like import (keyword) or keys (dict method), simply use the bracket form.

Non-existent keys

Non-existent keys:

pjy 'd.baz'
    {"foo":"bar"}

will return None:

null

Same for out-of-bounds indices:

pjy 'd[3]'
    [1, 2]

Do a basic operation

It's possible to use everything that a Python expression can contain:

pjy '[i + 1 for i in d["baz"]]'
    {"foo":"bar","baz":[1,2,3]}

Prints:

[
  2,
  3,
  4
]

Lambda-placeholder

A special identifier, _ can be used to create lambdas. This identifier will absorb most operations done to it and return a lambda applying them. Then, the returned lambda can be applied:

pjy 'map(_ + 1, d.baz)'
    {"foo":"bar","baz":[1,2,3]}

Is equivalent to:

pjy 'map((lambda x: x + 1), d.baz)'
    {"foo":"bar","baz":[1,2,3]}

Which will print:

[
  2,
  3,
  4
]

The lambda-placeholder will absorb chained operations:

pjy 'map((_ + 1) * 2, d.baz)'
    {"foo":"bar","baz":[1,2,3]}

Will result in:

[
  4,
  6,
  8
]

And:

pjy 'map(_[1:3] * 2, d)'
    {"foo":"bar","baz":[1,2,3]}

Will return:

{
  "foo": "arar",
  "baz": [
    2,
    3,
    2,
    3
  ]
}

Pipe-like iteration

The pipe (|) can be used to iterate on a list, it accepts a function as right operand:

pjy 'd.baz | _ + 1'
    {"foo":"bar","baz":[1,2,3]}

Which prints:

[
  2,
  3,
  4
]

It also operates on a dict's values, and returns a dict:

pjy 'd | (lambda x: repr(x))'
    {"foo":"bar","baz":[1,2,3]}

The values are replaced by the right operand value, the keys are unchanged:

{
  "foo": "'bar'",
  "baz": "[1, 2, 3]"
}

Ampersand for filtering

Similar to the pipe, the ampersand (&) is used on a list and a function, but its purpose is to filter:

pjy 'd & (_ % 2 == 0)'
    [0, 1, 2, 3]

outputs:

[
  0,
  2
]

Which is equivalent to running:

pjy 'filter(_ % 2 == 0, d)'
    [0, 1, 2, 3]

Like the pipe, it works on a dict, and the filter is applied on the dict values.

Partial placeholder

It's not possible to call a function on a placeholder, for example, len(_) will not work. However, it's possible to use the partial helper to prepare the function call:

pjy 'd | partial(len, _)'
    {"foo":"bar","baz":[1,2,3]}

Prints:

{
  "foo": 3,
  "baz": 3
}

partial ressembles the functools.partial function: it returns a function wrapping the function passed as first argument. The returned function will call the original function with the fixed arguments passed. The difference is that lambda-placeholders can be passed, and they will be replaced by the wrapper's argument.

p is a short alias for the partial function which can be used in pjy expressions.

Imports

It's possible to import modules with the imp function:

pjy 'filter(p(imp("fnmatch").fnmatch, _, "f*"), d.keys())'
     {"foo":"bar","baz":[1,2,3]}

Will print:

[
  "foo"
]

The math and re modules are already imported and available directly without having to call imp.

Multiple inputs

In pjy, an inputs variable exists, which is a list containing the JSON data of each input file passed on the command line. The d variable is simply an alias to inputs[0].

For example:

pjy 'filter(_[0] != _[1], zip(inputs[0], inputs[1]))' before.json after.json

will read 2 files before.json and after.json, which consist in a list of objects, and pjy will compare each zipped-pair of objects together. Then it will print the list of differing pairs.

Options

Input options

--null-input

Don't read any input, act as if the input was only null.

--arg VAR VALUE

Inject a variable named VAR with a VALUE in the expression.

Output options

--monochrome-output

Force no colors even if output is a TTY. Can also set NO_COLOR environment variable to do the same.

--ascii-output

When outputting non-ASCII strings, use \uXXXX notation instead of directly Unicode characters by default.

--tab

Indent output with tabs instead of 2 spaces.

--indent N

Indent output with N spaces instead of 2 spaces.

--compact-output

Don't indent output and don't add extra whitespace between key/values and list elements.

Environment variables

NO_COLOR: set it to disable colors even if output is a TTY.

CLICOLOR_FORCE: set it to enable colors even if output is not a TTY.

Security

pjy by itself does not write files (except stdout/stderr) or sockets, or run external commands. However, pjy runs the given expressions passed as argument, in the Python interpreter, without a sandbox. Hence, do NOT pass dangerous or untrusted Python expressions to pjy.

Dependencies

pjy is written in Python 3. Its setup.py requires setuptools.

If pygments is installed, pjy's output will be colorized, but it's entirely optional.

Version and license

pjy is at version 0.13.0, it uses semantic versioning. It is licensed under the WTFPLv2, see COPYING.WTFPL for license text.