Skip to content

pyxsdata-core

Blazing-Fast Native Rust XML Deserializer for Python
Accelerating dataclasses and Pydantic v2 with PyO3 & quick-xml

What is pyxsdata-core?

pyxsdata-core is a dedicated native acceleration library written in Rust using PyO3 and quick-xml. It delivers high-throughput XML deserialization for Python by bypassing the traditional DOM allocation, intermediate node wrapping, and generic Python event pumps used by standard XML parsers.

Instead, pyxsdata-core compiles Python class introspection into an in-memory Rust schema registry. As XML tokens stream through quick-xml's zero-copy pull parser, Rust directly instantiates Python dataclass and Pydantic v2 objects via the CPython C-API at rates exceeding 300,000 objects per second.

flowchart LR
    A[XML Document / Bytes] --> B[quick-xml Tokenizer]
    B --> C[Rust Schema Registry]
    C -->|Direct PyO3 C-API Call| D[Python dataclasses / Pydantic v2]

    style C fill:#f96,stroke:#333,stroke-width:2px

Key Highlights

  • Extreme Performance:
  • ~311,000 objects/sec for Pydantic v2 models (7.7x faster than pure Python).
  • ~290,000 objects/sec for standard dataclasses (9.5x faster than pure Python, 15x faster than legacy xsdata).
  • Native Pydantic v2 & Dataclass Support: Works seamlessly with both standard library @dataclass models and pydantic.BaseModel schemas without configuration shims.
  • Zero Intermediate DOM Allocation: No xml.etree.Element, lxml.etree.Element, or intermediate event queues created during the parse pass.
  • Precompiled Multi-Platform Wheels: Ready-to-use binaries on Linux (x86_64, aarch64), macOS (Apple Silicon, Intel), and Windows (x64) for Python 3.12+.
  • Drop-in Acceleration for pyxsdata: Powers CoreXmlParser and CoreEventHandler in pyxsdata[core].

At a Glance

from pydantic import BaseModel
import pyxsdata_core

class Item(BaseModel):
    id: int
    name: str
    price: float

xml = b"<Item><id>42</id><name>Widget</name><price>19.99</price></Item>"
item = pyxsdata_core.deserialize(xml, Item)

print(item)
# Item(id=42, name='Widget', price=19.99)

from dataclasses import dataclass
import pyxsdata_core

@dataclass
class Item:
    id: int
    name: str
    price: float

xml = b"<Item><id>42</id><name>Widget</name><price>19.99</price></Item>"
item = pyxsdata_core.deserialize(xml, Item)

print(item)
# Item(id=42, name='Widget', price=19.99)

from pyxsdata.pydantic import CoreXmlParser
# or: from pyxsdata.formats.dataclass.parsers import CoreXmlParser

parser = CoreXmlParser()
item = parser.from_bytes(xml, Item)

Documentation Guide