Enhance documentation on serializing untrusted input in dump() (#5304)

This commit is contained in:
Luke Banicevic
2026-07-26 08:29:31 +00:00
committed by GitHub
parent d0d29039da
commit dfa51af692
2 changed files with 32 additions and 0 deletions
+11
View File
@@ -43,6 +43,17 @@ Strong guarantee: if an exception is thrown, there are no changes to any JSON va
Throws [`type_error.316`](../../home/exceptions.md#jsonexceptiontype_error316) if a string stored inside the JSON value
is not UTF-8 encoded and `error_handler` is set to `strict`
!!! warning "Serializing untrusted input"
When serializing values that may contain invalid or untrusted UTF-8 (e.g., bytes taken directly from network
input), `dump()` throws [`type_error.316`](../../home/exceptions.md#jsonexceptiontype_error316) in the default
`strict` mode. To serialize such data without throwing, pass
[`error_handler_t::replace`](error_handler_t.md) (substitutes U+FFFD) or
[`error_handler_t::ignore`](error_handler_t.md). Callers that serialize untrusted input on a crash-sensitive path
should either choose a non-strict error handler or wrap `dump()` in a `#!cpp try`/`#!cpp catch`.
See the [FAQ](../../home/faq.md#serializing-untrusted-or-invalid-utf-8) for details.
## Complexity
Linear.
+21
View File
@@ -194,6 +194,27 @@ The library uses `std::numeric_limits<number_float_t>::digits10` (15 for IEEE `d
See [this section](../features/types/number_handling.md#number-serialization) on the library's number handling for more information.
### Serializing untrusted or invalid UTF-8
!!! question "Questions"
- Why does `dump()` throw when I serialize data that came from the network?
- Is CVE-2024-34363 a vulnerability in this library?
Crashes reported against this library that stem from an uncaught
[`type_error.316`](exceptions.md#jsonexceptiontype_error316) while serializing unvalidated input (e.g.,
CVE-2024-34363) are a usage issue, not a library vulnerability:
[`dump()`](../api/basic_json/dump.md) throws in its default `strict` mode because
[RFC 8259](https://datatracker.ietf.org/doc/html/rfc8259#section-8.1) requires JSON text to be valid UTF-8.
The recommended pattern is to pass a non-strict [`error_handler`](../api/basic_json/error_handler_t.md) or to handle the
exception:
```cpp
// replace invalid sequences with U+FFFD instead of throwing
const auto s = j.dump(-1, ' ', false, json::error_handler_t::replace);
```
### Using JSON values with `std::format` or `fmt`
!!! question