# jiter

> Fast iterable JSON parser.

- **URL**: https://www.freshcrate.ai/projects/jiter
- **Author**: pypi
- **Category**: Frameworks
- **Latest version**: `v0.15.0` (2026-05-19)
- **License**: Unknown
- **Source**: https://github.com/pydantic/jiter/
- **Language**: Rust
- **GitHub**: 524 stars, 58 forks
- **Registry**: pypi (`jiter`)
- **Tags**: `pypi`

## Description

# jiter

[![CI](https://github.com/pydantic/jiter/workflows/CI/badge.svg?event=push)](https://github.com/pydantic/jiter/actions?query=event%3Apush+branch%3Amain+workflow%3ACI)
[![pypi](https://img.shields.io/pypi/v/jiter.svg)](https://pypi.python.org/pypi/jiter)
[![versions](https://img.shields.io/pypi/pyversions/jiter.svg)](https://github.com/pydantic/jiter)
[![license](https://img.shields.io/github/license/pydantic/jiter.svg)](https://github.com/pydantic/jiter/blob/main/LICENSE)

This is a standalone version of the JSON parser used in `pydantic-core`. The recommendation is to only use this package directly if you do not use `pydantic`.

The API is extremely minimal:

```python
def from_json(
    json_data: bytes,
    /,
    *,
    allow_inf_nan: bool = True,
    cache_mode: Literal[True, False, 'all', 'keys', 'none'] = 'all',
    partial_mode: Literal[True, False, 'off', 'on', 'trailing-strings'] = False,
    catch_duplicate_keys: bool = False,
    float_mode: Literal['float', 'decimal', 'lossless-float'] = 'float',
) -> Any:
    """
    Parse input bytes into a JSON object.

    Arguments:
        json_data: The JSON data to parse
        allow_inf_nan: Whether to allow infinity (`Infinity` an `-Infinity`) and `NaN` values to float fields.
            Defaults to True.
        cache_mode: cache Python strings to improve performance at the cost of some memory usage
            - True / 'all' - cache all strings
            - 'keys' - cache only object keys
            - False / 'none' - cache nothing
        partial_mode: How to handle incomplete strings:
            - False / 'off' - raise an exception if the input is incomplete
            - True / 'on' - allow incomplete JSON but discard the last string if it is incomplete
            - 'trailing-strings' - allow incomplete JSON, and include the last incomplete string in the output
        catch_duplicate_keys: if True, raise an exception if objects contain the same key multiple times
        float_mode: How to return floats: as a `float`, `Decimal` or `LosslessFloat`

    Returns:
        Python object built from the JSON input.
    """


def cache_clear() -> None:
    """
    Reset the string cache.
    """


def cache_usage() -> int:
    """
    get the size of the string cache.

    Returns:
        Size of the string cache in bytes.
    """

```
## Examples

The main function provided by Jiter is `from_json()`, which accepts a bytes object containing JSON and returns a Python dictionary, list or other value.

```python
import jiter

json_data = b'{"name": "John", "age": 30}'
parsed_data = jiter.from_json(json_data)
print(parsed_data)  # Output: {'name': 'John', 'age': 30}
```

### Handling Partial JSON

Incomplete JSON objects can be parsed using the `partial_mode=` parameter.

```python
import jiter

partial_json = b'{"name": "John", "age": 30, "city": "New Yor'

# Raise error on incomplete JSON
try:
    jiter.from_json(partial_json, partial_mode=False)
except ValueError as e:
    print(f'Error: {e}')

# Parse incomplete JSON, discarding incomplete last field
result = jiter.from_json(partial_json, partial_mode=True)
print(result)  # Output: {'name': 'John', 'age': 30}

# Parse incomplete JSON, including incomplete last field
result = jiter.from_json(partial_json, partial_mode='trailing-strings')
print(result)  # Output: {'name': 'John', 'age': 30, 'city': 'New Yor'}
```

### Catching Duplicate Keys

The `catch_duplicate_keys=True` option can be used to raise a `ValueError` if an object contains duplicate keys.

```python
import jiter

json_with_dupes = b'{"foo": 1, "foo": 2}'

# Default behavior (last value wins)
result = jiter.from_json(json_with_dupes)
print(result)  # Output: {'foo': 2}

# Catch duplicate keys
try:
    jiter.from_json(json_with_dupes, catch_duplicate_keys=True)
except ValueError as e:
    print(f'Error: {e}')
```

## Recent releases

| Version | Date | Urgency | Changes |
| --- | --- | --- | --- |
| `v0.15.0` | 2026-05-19 | High | ## What's Changed * expose `known_number_bytes` and parser methods for bytes -> number by @davidhewitt in https://github.com/pydantic/jiter/pull/250   **Full Changelog**: https://github.com/pydantic/jiter/compare/v0.14.0...v0.15.0 |
| `0.14.0` | 2026-04-21 | Low | Imported from PyPI (0.14.0) |
| `v0.14.0` | 2026-04-10 | Medium | Release v0.14.0 |
| `v0.14.0` | 2026-04-10 | Medium | Latest release: v0.14.0 |
| `v0.14.0` | 2026-04-10 | Medium | Latest release: v0.14.0 |
| `v0.14.0` | 2026-04-10 | Medium | Latest release: v0.14.0 |
| `v0.14.0` | 2026-04-10 | Medium | Latest release: v0.14.0 |
| `v0.14.0` | 2026-04-10 | Medium | Latest release: v0.14.0 |
| `v0.14.0` | 2026-04-10 | Medium | Latest release: v0.14.0 |
| `v0.14.0` | 2026-04-10 | Medium | Latest release: v0.14.0 |

## Citation

- HTML: https://www.freshcrate.ai/projects/jiter
- Markdown: https://www.freshcrate.ai/projects/jiter.md
- Dependencies JSON: https://www.freshcrate.ai/api/projects/jiter/deps

_Generated by freshcrate.ai. Indexes pypi releases for AI-agent ecosystem packages._