Support OpenAI Responses API content parts

[RitwijParmar]

Fixes #1901.

This fills in the Responses API content-part gap in the OpenAI instrumentation:

• input_text now maps to a text part/event instead of falling through as unknown
• input_image handles the Responses API's flat image_url string as well as file_id and inline file data
• input_file maps URL and file-id references to document URI parts, and inline file_data to a document blob part
• the legacy events path now records useful content for these same parts instead of gen_ai.unknown

I kept unknown/future part types on the existing generic fallback, so this should be narrowly scoped to the documented Responses API part types.

Tests run:

• uv run pytest tests/otel_integrations/test_openai.py -q -k "input_to_events or convert_responses_inputs"
• uv run ruff check logfire/_internal/integrations/llm_providers/openai.py tests/otel_integrations/test_openai.py
• uv run ruff format --check --diff logfire/_internal/integrations/llm_providers/openai.py tests/otel_integrations/test_openai.py
• uv run pyright logfire/_internal/integrations/llm_providers/openai.py tests/otel_integrations/test_openai.py

AI disclosure: I used AI assistance while editing and reviewing this patch, then ran the tests and checks above locally.

[cubic-dev-ai]

No issues found across 2 files

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

_{Re-trigger cubic}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[RitwijParmar]

Updated this.

File IDs now use the semconv file part instead of an openai://file/... URI, and known image/file Responses parts no longer fall through to gen_ai.unknown on the events path.

Focused checks pass:

• uv run pytest tests/otel_integrations/test_openai.py -k "responses_api_content_parts or multimodal_content_parts" -q
• uv run ruff check logfire/_internal/integrations/llm_providers/openai.py logfire/_internal/integrations/llm_providers/semconv.py tests/otel_integrations/test_openai.py
• uv run ruff format --check logfire/_internal/integrations/llm_providers/openai.py logfire/_internal/integrations/llm_providers/semconv.py tests/otel_integrations/test_openai.py

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: c712c3cb-91c0-403b-a1a1-d24c4fe88a60

📥 Commits

Reviewing files that changed from the base of the PR and between 0524aa2 and 508b299.

📒 Files selected for processing (3)

• logfire/_internal/integrations/llm_providers/openai.py
• logfire/_internal/integrations/llm_providers/semconv.py
• tests/otel_integrations/test_openai.py

🔗 Linked repositories identified

CodeRabbit considers these linked repositories for cross-repo context during reviews:

• pydantic/pydantic (auto-detected)
• pydantic/pydantic-ai (auto-detected)

🚧 Files skipped from review as they are similar to previous changes (3)

• logfire/_internal/integrations/llm_providers/semconv.py
• logfire/_internal/integrations/llm_providers/openai.py
• tests/otel_integrations/test_openai.py

---

📝 Walkthrough

Walkthrough

FilePart is added as a new TypedDict to the semconv type system with a type='file' discriminator, required file_id, and optional media_type and modality, and is included in the MessagePart union. _convert_content_part is rewritten to map Responses API types (input_text, input_image, input_file, input_audio) alongside existing Chat Completions types to appropriate semconv parts (TextPart, UriPart, BlobPart, FilePart), with unknown types returning a normalized dict. The input_to_events legacy path delegates per-item handling to a new _content_part_to_event helper. Two new tests cover both conversion paths.

🚥 Pre-merge checks | ✅ 4

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and concisely summarizes the main change: adding support for OpenAI Responses API content parts to the instrumentation.
Description check	✅ Passed	The description is directly related to the changeset, explaining which Responses API content types are now supported and how the implementation addresses the gap.
Linked Issues check	✅ Passed	The code changes fully address issue #1901: input_text now maps to text parts, input_image handles flat URL/file_id/file_data formats, input_file maps to document URI/blob parts, and legacy events path records useful content instead of unknown.
Out of Scope Changes check	✅ Passed	All changes are narrowly scoped to supporting the three Responses API content types identified in #1901, with helper functions and type definitions appropriately limited to this feature.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[alexmojaki]

Please fix coverage

[RitwijParmar]

Fixed coverage. CI and Codecov are green now.

[cubic-dev-ai]

No issues found across 3 files

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

_{Re-trigger cubic}

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no bugs or issues to report.