ijl/orjson
Orjson
orjson is a high-performance Python library for serializing and deserializing JSON data. It functions as both a JSON parsing library and a serialization engine, converting data between native Python objects and UTF-8 encoded bytes.
The project provides specialized support for converting complex Python data structures, including dataclasses, datetime objects, and NumPy arrays and scalars, into JSON format. It also allows for the insertion of pre-serialized JSON blobs into documents to maintain processing speed.
The library includes capabilities for memory-efficient deserialization through key caching and the conversion of large datasets with minimal latency. These features support high-throughput data interchange and scientific computing pipelines.
Features
- High-Performance Parsers - Provides a high-performance JSON parser optimized for extreme throughput and minimal latency.
- JSON Converters - Provides high-performance conversion of JSON strings and bytes into native Python objects.
- JSON Serializers - Functions as a fast serialization engine for Python objects, dataclasses, and NumPy arrays.
- JSON Parsers - Provides a memory-efficient library for parsing JSON strings into Python dictionaries and lists.
- JSON Serialization Libraries - Converts native Python data objects and complex types into formatted JSON strings.
- Rust - Provides a high-performance serialization engine implemented in Rust.
- Message Encoders and Decoders - Enables rapid encoding and decoding of JSON for high-throughput network services.
- Complex Data Serialization - Supports serialization of non-primitive Python types, such as dataclasses and datetime objects.
- Key Caching - Reduces memory allocations during object creation by caching frequently encountered JSON keys.
- Memory-Efficient Parsing - Minimizes memory overhead during JSON deserialization through specialized key caching techniques.
- Python Dataclass Serialization - Converts complex Python dataclasses and datetime objects into JSON for API responses.
- Runtime Type Dispatching - Uses specialized runtime dispatch logic to optimize serialization for NumPy arrays and Python dataclasses.
- JSON Export - Exports numeric arrays and scalars to JSON for scientific computing and data science pipelines.
- JSON Serialization - Converts NumPy arrays and scalars into JSON format without requiring the array library.
- Zero-Copy Buffers - Shares memory buffers between Rust and Python to minimize data duplication during parsing.
- Data Serialization - Fast JSON library for Python.
- Data Serialization - Fast, correct JSON library with support for dataclasses and datetimes.
