 Deserialization

Deserialization reconstructs values from a representation produced according to a format such as JSON. The decoder's mapping determines the resulting runtime data types. In Python, json.loads maps a JSON object to a dictionary, an array to a list, false to False and null to None by default.^[ref]

For an original example, json.loads('{"enabled":false,"items":[null,0,"0"]}') produces {'enabled': False, 'items': [None, 0, '0']}. The last two items differ: the integer zero and a string containing a digit are not interchangeable. The decoder reconstructs nested values rather than returning the displayed source text.

Syntax and application meaning require separate checks. The malformed JSON {"items":[1,2,]} raises JSONDecodeError. By contrast, {"items":"many"} parses successfully but fails a program's requirement that items be a list. That second check is data validation, beyond recognizing the format's syntax.^[ref]

Nor does successful deserialization guarantee recovery of every original property. A Python tuple encoded as a JSON array normally returns as a list. A round-trip claim must specify whether it preserves values, runtime types or object identity. Character encoding and parsing contribute to the reconstruction, while the selected mapping determines which distinctions survive.