Fix double-encoded JSON without stripping backslashes

First check the parsed type

A log or API response can contain a JSON document inside a JSON string. The outer document is valid, so a syntax check alone will not catch the mismatch. You may see escaped quotes such as \" and try to remove them. Those backslashes can be necessary syntax for the outer string.

The following self-contained example deliberately introduces a second layer. It prints the runtime type after each parse:

import json

payload = {"status": "ready", "attempts": 2}
wire = json.dumps(json.dumps(payload))
first = json.loads(wire)
second = json.loads(first)
print(type(first).__name__)
print(type(second).__name__)
print(second["status"])

Expected output:

str
dict
ready

The first parse has worked: its result is a string containing another document. The second parse is appropriate here because this example's contract explicitly says the inner value is JSON. Valid top-level JSON strings also exist; receiving a string is not automatically a producer bug.

Fix the extra serialization at the boundary

Use one serialization step for each transport boundary that expects JSON. If an HTTP library accepts a native object through a JSON parameter, hand it the object rather than an already serialized string. If a lower-level interface expects bytes or text, serialize once there. Check the library's interface before changing the caller.

import json

payload = {"status": "ready", "attempts": 2}
wire = json.dumps(payload)
received = json.loads(wire)
assert isinstance(received, dict)
assert received == payload
print(received["attempts"])

Expected output:

2

Fixing the producer prevents every consumer from needing its own workaround. For a deliberately nested field, decode only that documented field rather than treating all strings as serialized JSON.

Why removing backslashes is destructive

Backslashes also represent literal backslashes, newlines, and quotes inside a value. A path, quoted message, or regular expression can change meaning if you remove them globally. This payload must survive decoding with its message intact:

{"message":"She said \"ready\"","path":"C:\\logs\\today.json"}

Keep an untouched copy of the original response and compare parsed values rather than deciding from a debugger's representation. Debuggers often add another visual layer of quotes around a string; that display is not necessarily the raw HTTP body.

What the tools show

Paste the raw document into JSON Viewer. If the root is a string, it will not behave like an expandable object. The converter formats the parsed value; it does not recursively decode strings until they resemble objects. That avoids silently changing legitimate text.

When you know an extra layer exists, decode that layer in the producing program or a controlled Python script, then inspect the result again. Do not repeatedly decode arbitrary strings as a repair strategy. For the standard parser API, see the Python json reference.

By Python JSON Formatter · Published 19 October 2026