fix(integrations): pydantic-ai: properly format binary input message parts to be conformant with the gen_ai.request.messages structure

[constantinius]

Issues

Closes https://linear.app/getsentry/issue/TET-1634/redact-images-pydantic-ai

[linear]

TET-1634 Redact images: Pydantic-AI

[github-actions]

Semver Impact of This PR

🟢 Patch (bug fixes)

📋 Changelog Preview

This is how your changes will appear in the changelog.
Entries from this PR are highlighted with a left border (blockquote style).

---

New Features ✨

• feat(asyncio): Add on-demand way to enable AsyncioIntegration by sentrivana in #5288

Bug Fixes 🐛

• fix(ai): redact message parts content of type blob by constantinius in #5243
• fix(clickhouse): Guard against module shadowing by alexander-alderman-webb in #5250
• fix(gql): Revert signature change of patched gql.Client.execute by alexander-alderman-webb in #5289

• fix(integrations): pydantic-ai: properly format binary input message parts to be conformant with the gen_ai.request.messages structure by constantinius in #5251

• fix(litellm): Guard against module shadowing by alexander-alderman-webb in #5249
• fix(pure-eval): Guard against module shadowing by alexander-alderman-webb in #5252
• fix(ray): Guard against module shadowing by alexander-alderman-webb in #5254
• fix(threading): Handle channels shadowing by sentrivana in #5299
• fix(typer): Guard against module shadowing by alexander-alderman-webb in #5253

Documentation 📚

• docs: Update Python versions banner in README by sentrivana in #5287

Internal Changes 🔧

• ci(release): Switch from action-prepare-release to Craft by BYK in #5290

---

_{🤖 This preview updates automatically when you update the PR.}

[sentry]

Bug: The Pydantic AI integration unconditionally redacts binary message content, ignoring the send_default_pii setting and preventing users from seeing this data in traces.
_{Severity: HIGH}

🔍 Detailed Analysis

In the Pydantic AI integration, the function truncate_and_annotate_messages is called to process AI messages. This function unconditionally calls redact_blob_message_parts, which replaces any binary blob content with the string "[Blob substitute]". This redaction occurs regardless of the user's send_default_pii setting. As a result, users who have enabled PII (send_default_pii=True) to view binary content in their traces will find it redacted, which contradicts the intended behavior and the implementation pattern seen in the OpenAI integration, where a should_send_default_pii() check is performed.

💡 Suggested Fix

Wrap the call to truncate_and_annotate_messages within a conditional that checks should_send_default_pii(). This will align the Pydantic AI integration's behavior with the existing OpenAI integration, ensuring binary content is only redacted when PII is disabled.

🤖 Prompt for AI Agent

Review the code at the location below. A potential bug has been identified by an AI
agent.
Verify if this is a real issue. If it is, propose a fix; if not, explain why it's not
valid.

Location: sentry_sdk/integrations/pydantic_ai/spans/ai_client.py#L145-L147

Potential issue: In the Pydantic AI integration, the function
`truncate_and_annotate_messages` is called to process AI messages. This function
unconditionally calls `redact_blob_message_parts`, which replaces any binary blob
content with the string `"[Blob substitute]"`. This redaction occurs regardless of the
user's `send_default_pii` setting. As a result, users who have enabled PII
(`send_default_pii=True`) to view binary content in their traces will find it redacted,
which contradicts the intended behavior and the implementation pattern seen in the
OpenAI integration, where a `should_send_default_pii()` check is performed.

_{Did we get this right? 👍 / 👎 to inform future reviews.
Reference ID: 8534615}