fix(models): skip location-source documents in OpenAIResponsesModel message formatting - #4706

Description

OpenAIResponsesModel reads document["source"]["bytes"] unconditionally in _format_request_message_content, so a document whose source is a location (for example an S3 URI) raises KeyError: 'bytes' instead of being skipped with a warning. Every other provider has gated this case since #1572 with the shared _has_location_source helper; the Responses provider never got the gate.

This change adds the same check to _format_request_messages: content blocks with a location source are skipped with the warning "Location sources are not supported by OpenAI Responses | skipping content block", matching the behavior of the other eight providers.

Scope note: this PR covers the message path only, as reported in #4016. The tool-result path (_format_request_tool_message) has the same unguarded read but is tracked separately in #4018 and is not changed here. #3582 widens DocumentSource further (text/content variants), which makes this gate more relevant; it does not overlap with this change.

Thanks to @strandly-the-agent for the report.

Related Issues

Fixes #4016

Related: #1572, #4018, #3582

Documentation PR

No documentation changes needed. This brings the provider in line with the behavior the other providers already have.

Type of Change

Bug fix

Testing

• Reproduced the issue's script on current main: KeyError: 'bytes'. With the fix, the location-source document is skipped with a warning and the request formats normally; a document-only message drops out of the input instead of crashing.

• Added test_format_request_filters_location_source_document in tests/strands/models/test_openai_responses.py, mirroring the standard test the other providers carry: text content is kept, the location document is filtered, and the warning is logged. The test links the issue.

• pytest tests/strands/models/test_openai_responses.py: 161 passed.

• ruff format and ruff check are clean on the changed files. I ran these tools directly (the environment had no hatch); they are the same checks hatch run prepare wraps.

• I ran hatch run prepare (ran its equivalent directly: ruff formatter, ruff linter, pytest - see above)

Checklist

• I have read the CONTRIBUTING document
• I have reviewed and understand every line of code in this PR, including any generated by AI tools, and I can explain why it works
• My change is focused and reasonably small; I have split unrelated work into separate PRs
• I have added any necessary tests that prove my fix is effective or my feature works
• I have updated the documentation accordingly
• I have added an appropriate example to the documentation to outline the feature, or no new docs are needed
• My changes generate no new warnings
• Any dependent changes have been merged and published

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.