litellm/tests/rust-python-harness/strategies/e2e_parity/AGENTS.md
yujonglee ee08c36fc0
refactor(tests): restructure rust python harness around strategy definitions (#39628)
* wip

* refactor(tests): move sdk function tracing into rust python harness

* dead code

* fix: handle harness keyboard interrupts

* refactor(tests): deduplicate rust python harness helpers

* fix(harness): expose validated strategy choices

* wip

* refactor(harness): let strategies own parity reports

* docs(harness): update strategy structure

* refactor(harness): localize strategy report views

* wip

* fix(harness): satisfy mapping runner type checks

* fix(harness): clarify trace parity output

* wip

* fix(harness): clarify unit mapping report

* fix(harness): finalize trace parity contracts

* refactor(harness): structure parity contracts

* feat: derive unit test mapping from traces

* feat(harness): map rstest test families

* feat(ocr): port Azure document intelligence tests

* feat(harness): enforce complete unit mappings

* feat(ocr): add reducto core transforms

* feat(harness): classify host-only unit tests

* fix(ocr): complete Rust provider plumbing

* fix(harness): reuse OCR parity workers
2026-09-03 21:15:01 -07:00

238 B

Switches between the Rust implementation and the existing Python core, then compares their observable behavior for parity across SDK objects, exceptions, callbacks, streams, and gateway HTTP responses using generated and recorded inputs.