Validate FreshRSS unread responses with Pydantic

[Skulldorom]

Motivation

• Ensure the FreshRSS /reader/api/0/stream/contents payload is well-formed before processing to avoid hidden data-quality errors and crashes.
• Apply a clear policy for malformed upstream items so callers see consistent all-or-nothing behavior instead of silent partial results.

Description

• Add strict Pydantic models FreshRSSOrigin, FreshRSSAlternate, FreshRSSItem, and FreshRSSResponse and a validate_freshrss_response helper that uses FreshRSSResponse.model_validate to validate the top-level object and nested containers.
• Validate published timestamps with a field validator that requires numeric, finite values and verifies datetime.fromtimestamp can represent the value.
• Implement an all-or-nothing policy that logs validation failures and returns a sanitized HTTP 502 with detail "FreshRSS returned an invalid unread response" when the upstream response or any item is malformed.
• Update the unread processing to consume validated model instances and add pydantic>=2.0 to pyproject.toml (and update uv.lock).

Testing

• Ran dependency and lock checks with uv lock --check, which completed successfully.
• Ran the test suite with uv run --frozen pytest -q, and all tests passed (27 passed).
• Confirmed uv run pytest -q during development also passed (27 passed).

---

Codex Task

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 5743a77755

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Catch overflow before checking timestamp finiteness

When published is a sufficiently large JSON integer (for example, a few hundred digits), math.isfinite(value) raises OverflowError while converting it to a float. Because this occurs before the guarded datetime.fromtimestamp call and Pydantic does not wrap OverflowError from a field validator as ValidationError, the endpoint returns an internal 500 instead of the sanitized 502 promised for malformed FreshRSS responses.

Useful? React with 👍 / 👎.

[Skulldorom]

Superseded by #25 — all four PRs merged with conflict resolution and review fixes applied.