Description

With validate_input_modes=True, a card that declares text/plain rejects a part sent as Text/Plain or text/plain; charset=utf-8, because the check compares exact strings. The spec defines input modes as media types, RFC 2045 section 5.1 makes type and subtype case-insensitive, and a charset parameter does not change the type.

The validator now compares lower-cased type/subtype on both sides. The error still quotes what the client sent and what the card declares, and text/html; charset=utf-8 still fails against text/plain.

$ uv run pytest tests/utils/test_input_mode_validator.py   # 11f1116, fix reverted
FAILED ...::test_media_types_match_on_type_and_subtype[Text/Plain-text/plain]
FAILED ...::test_media_types_match_on_type_and_subtype[text/plain; charset=utf-8-text/plain]
5 failed, 12 passed in 0.12s

$ uv run pytest   # 11f1116 plus this branch, Python 3.10
2530 passed, 207 skipped, 4 xfailed, 1 xpassed, 200 warnings in 44.04s
$ uv run pytest --cov=src
src/a2a/utils/input_mode_validator.py   100%
TOTAL   94%
$ ./scripts/lint.sh
SUCCESS: All linting and formatting tasks complete!
# its 3 ruff diagnostics are in client_factory.py and transports/__init__.py, which this does not touch

a2aproject/a2a-js#745 still compares exact strings, so the two SDKs differ until it gets the same fix, which I can send there next.

@mykytanetipa a card that declares image/* still only matches that literal string. I kept wildcards out, since that is a policy call on top of #1269 rather than a parsing fix. Would you rather have them handled here?

• Follow the CONTRIBUTING Guide.
• Make your Pull Request title in the https://www.conventionalcommits.org/ specification.
• Ensure the tests and linter pass (Run bash scripts/format.sh from the repository root to format)
• Appropriate docs were updated (if necessary)
  • The validate_input_modes docstring now states the matching rule.

Refs #1269. No issue exists, since the tests above reproduce it.