Pretty print JSON in Python: the json.dumps guide

The one-liner and what it produces

import json

data = {"name": "red sneaker", "sizes": [40, 41, 42], "in_stock": True, "price": 59.0, "meta": None}
print(json.dumps(data, indent=2))

Output, exactly as Python prints it:

{
  "name": "red sneaker",
  "sizes": [
    40,
    41,
    42
  ],
  "in_stock": true,
  "price": 59.0,
  "meta": null
}

Three things to notice, because they surprise people on real data. Booleans print lowercase (true), None prints as null, and floats keep their Python form. Lists expand too: a three-element list becomes four lines, whether you like it or not, because indent applies to every container, not just objects. Dropping the indent argument gives a single line with single spaces, like {"name": "red sneaker", "sizes": [40, 41, 42], "in_stock": true, "price": 59.0, "meta": null}, which is fine for logs but hopeless for reading.

To get sorted keys, which makes diffs between two dumps meaningful, add sort_keys=True:

json.dumps(data, indent=2, sort_keys=True)

Pretty print a JSON file

In code, read and dump:

import json

with open("data.json") as f:
    data = json.load(f)

print(json.dumps(data, indent=2))

To write the pretty version back to disk, use json.dump (no s: it writes to a file object) with the same indent:

with open("data-pretty.json", "w") as f:
    json.dump(data, f, indent=2)

Without writing any code, the standard library ships a formatter you can run from the terminal:

python -m json.tool data.json

It indents by 4 spaces and validates the file while it formats, so a broken file fails with a clear error instead of formatted output. It also reads stdin, which makes it the quickest way to pretty print an API response:

curl -s https://api.example.com/data | python -m json.tool

Choosing the indentation

• indent=2. The common choice, and what you see in most JSON you meet. Use it unless something below contradicts you.
• indent=4. Matches Python's own indentation style, which some teams prefer when JSON lives next to Python source in review.
• indent with a string. Any string works as the unit: indent="\t" indents with real tabs. Editors and terminals both handle it.
• indent=None. Compact single-line output. For machine-to-machine payloads this is what you want anyway; pretty formatting is for humans.

One separator nuance bites people comparing outputs: with indent set, items separate with a bare comma and the spacing comes from newlines; without it, the default separators are a comma plus space after items. The bytes differ between the two styles, so byte-level comparisons of formatted files should pin both indent and separators explicitly.

The errors you will actually hit

• "Object of type datetime is not JSON serializable". The classic. JSON has no date type, so json.dumps refuses and names the offending type for you. For quick output, json.dumps(data, default=str) stringifies anything it does not recognize: a datetime becomes something like "2026-10-11 20:00:00", microseconds and all when they exist. For a format other people will parse, serialize dates yourself (ISO 8601 via datetime.isoformat()) instead of letting str improvise.
• Non-ASCII characters look wrong. By default json.dumps escapes every non-ASCII character: a café prints as caf\u00e9. It is valid JSON either way, but for human-readable output add ensure_ascii=False and the real letters survive.
• The file you read is not valid JSON. json.load raises json.decoder.JSONDecodeError with line and column. The usual suspects: trailing commas (JSON has none), single quotes (JSON wants double), and pasted output that was never JSON to begin with.