The jsonstrip Python module

The jsonstrip removes comments from a JSON document. It can be used as a Python module:

import jsonstrip

...
with open(filename, 'r', encoding='utf-8') as file:
    data = json.loads(jsonstrip.strip(file.read()))

or as a command line filter:

python3 -m jsonstrip < commented.json > regular.json

What makes jsonstrip unique is that it preserves line/column position of text.

Why do we need to strip comments from JSON?

The JSON specification does not allow comments. A commented JSON is invalid JSON. This limitation does not matter when computer programs exchange data, but sometimes an admin needs to write a configuration or a similar file in JSON format and wants to comment the contents. Those comments need to be removed before the document can be loaded.

Which comment types are recognized?

JavaScript (JS) comments - both single line and multi-line - and since version 22.10.20 also the widely-used (but not valid in JS) hash sign comments.

Features

• The main feature is that line/column position of text is preserved. A hand-edited file may contain mistakes and in case of an error like below, the exact position can be easily found in the original document and corrected there.

  json.decoder.JSONDecodeError: Expecting value: line 15 column 40
  

• The input document is properly parsed in order not to modify strings that happen to look like comments:

  [
  /* this is a comment and will be removed */
  "but /* this is a string and it will be left untouched */",
  "this is // not a line comment"   // the real comment is here
  ]
  

• It comes with a jsoncheck utility described below.

• The jsonstrip is short, simple, fast, open-source and free.

What it doesn't do

The author has decided to keep the jsonstrip as simple as possible.

• The jsonstrip does not minimize the JSON document. It does not strip any unneeded whitespace.

• The jsonstrip does not verify the JSON. It is assumed that its output will be loaded by a decoder and will be verified there.

• The jsonstrip does not even verify the comments. If a comment is open with /*, but not properly closed with */, the rest of the document becomes a comment and as such it will be removed without a warning.

Installation

Install with pip from the PyPI.

API

jsonstrip.strip(json_str: str) -> str

Return a copy of the JSON document json_str with comments removed.

How it works

Single line comments

A single line comment starts either with two forward slashes // or a single hash sign # and continues until the end of a line. Single line comments are removed.

Input:

[10, 20, // this is a comment and will be removed
30, 40]  # this text is also a comment

Output:

[10, 20, 
30, 40]  

Multi-line comments

A multi-line comment (also called a block comment) starts with /* and ends with */. These comments are removed when occurring at the end of line and replaced by whitespace elsewhere.

Input:

{"name": "foo", "value": 1001 /* comment */, "flag": true }

Output; note the unchanged position of the text:

{"name": "foo", "value": 1001              , "flag": true }

Note that multi-line comments cannot be nested, the */ sequence always terminates a comment:

/*
 * text
 * text  /* a comment inside a comment (WRONG!) */
 * BEWARE: no longer a comment
 */

The jsoncheck command line utility

This little program reads a commented JSON document, strips the comments and decodes the resulting JSON to check if there are any errors. All JSON objects (i.e. Python dicts) are checked for duplicate keys. (Note that the standard Python JSON decoder accepts duplicate keys and each occurrence overwrites the previous one. That makes some errors difficult to spot.)

The result of the check is printed on the standard output, one line for each input file. The contents of the file is not printed.

Usage

# print help and exit:
jsoncheck --help

# without arguments it reads from the stdin:
some-program | jsoncheck

# with arguments it checks the named files:
jsoncheck filename1 [filename2 ...]

Exit code

• 0 = all input files are OK
• 1 = invalid JSON in some input file(s)
• 2 = I/O error reading some input file(s)
• 3 = both errors 1 and 2 occurred
• 5 = incorrect usage (unrecognized option)