Skip to content

README.md

Latest commit

 

History

History
28 lines (19 loc) · 1.25 KB

README.md

File metadata and controls

28 lines (19 loc) · 1.25 KB
Raw

Python JSON Benchmark Suite: PolyXML vs xsdata

Self-contained benchmark comparing PolyXML native JSON APIs (polyxml.loads_json, dumps_json, JsonParser, JsonSerializer) against the xsdata JSON data-binding runtime on the exact same Python dataclass models and JSON bytes.

Workloads

  1. Small Sensor Telemetry (SensorReading): Flat entity with 6 scalar fields (~128 bytes JSON).
  2. Nested Order (Order): Nested domain entity with 10 line items, optional fields, and attribute mappings (~793 bytes JSON).
  3. Repeated Batch (SensorBatch): Collection of 100 sensor readings (~12.8 KB JSON).

Semantic Verification

Every run verifies:

  • assert poly_parsed == expected == xs_parsed for all payloads.
  • JSON emitted by polyxml is validated by parsing it back into typed models.
  • Deserialization and serialization are timed separately.

How to Run

From the repository root with virtual environment activated:

# Smoke test (5 iterations):
./benchmarks/python-json/run.sh --smoke

# Full benchmark (50 iterations, with raw JSON & Markdown output):
./benchmarks/python-json/run.sh --iterations 50 --output-json docs/benchmarks/data/2026-09-26/python-json.json --output-md docs/benchmarks/data/2026-09-26/python-json.md