Why search-and-replace is the wrong way to fix a trailing comma in JSON

The comma at the end of a JSON file can cause it to become unusable. For example, { name : ada , roles : [ admin , ops], } results in a Python json.decoder.JSONDecodeError or Node SyntaxError. JSON syntax does not allow trailing commas, so the parser stops at the first problem.

The one-liner fix, re.sub(r ',\s*([}\]]) , r '\1', text), works for simple cases. However, it can corrupt values inside strings, like changing the value of note from "ends with ," to "ends with } " when applied to a string containing commas.

To remove trailing commas correctly, you must track whether you are inside a string and watch for backslash escapes. A proper solution would be to use a function like remove_trailing_commas(text: str) -> str, which appends characters to a list while tracking string and escape state. It removes the comma only if it is outside a string and the next non-whitespace character closes an object or array.

In summary, repair only trailing commas outside strings, as they have a single defensible fix. Fail loudly with a line and column for other issues. This approach ensures the integrity of the JSON data.