Retry expired Instagram media URLs via instaById before giving up

[rejas]

Summary

• Instagram issues some (typically older) media noticeably shorter-lived signed CDN URLs than others fetched in the same batch, so a media_url embedded on the WordPress side can already be expired by the time a visitor loads it — even though our own cache entry for that batch is still well within its 12h TTL. resolve_media_url previously just returned a plain 400 with no way to recover. (Root cause originally diagnosed while fixing a stuck Instagram carousel on the frontpage — see the paired frontend fallback fix in vvp_divi5_extensions.)
• Thread the Instagram post id (and, for a carousel slide, its child id) and account through to the proxy URL, bound into the HMAC signature via a new generate_token_with_context() alongside the existing url-only generate_token().
• On a failed primary fetch, resolve_media_url now does one retry: re-fetch that single post from Instagram (fetch_fresh_media_url, new) to get a freshly signed URL, then fetch that.
• Backward compatible: links generated before this deploy (or from a post missing an id, which shouldn't happen but is technically optional in the API response) carry no post_id and are verified/served exactly as before — generate_token(url) is unchanged, replace_media_urls() is unchanged and still directly tested. replace_media_urls_for_posts() is the new, additive, Instagram-feed-specific counterpart that instaFeed now uses instead.
• Only adds a live Instagram API call on the request path when the first fetch fails, not on every image load, so this shouldn't meaningfully increase Graph API call volume.

Test plan

• python manage.py test proxycache — 68 tests pass (11 new, all existing ones untouched/still green)
• ruff check / ruff format --check clean on all touched files
• bandit -c pyproject.toml — no new findings (2 pre-existing low-severity findings in insta_feed.py's ACCOUNTS dict are unrelated env-var-name string matches, not part of this diff)
• Spot check on production once deployed: confirm a previously-broken Instagram carousel slide recovers without a full cache cycle

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR improves resilience of the Instagram media proxy by allowing a single, signed, context-aware retry when an embedded Instagram CDN media_url has expired before the proxy cache TTL.

Changes:

• Add context-bound HMAC tokens (generate_token_with_context) and Instagram-feed-specific URL rewriting (replace_media_urls_for_posts) so proxy URLs carry post_id/child_id/account.
• Update resolve_media_url to retry once by calling the Instagram Graph API for a fresh media_url when the initial fetch fails (context links only).
• Extend test coverage to validate context binding, fallback behavior, and backward compatibility for legacy (url-only) tokens.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
proxycache/helper.py	Adds context-bound token generation and an Instagram-feed-specific media URL rewriter that includes post/account context.
proxycache/services/resolve_media_url.py	Adds a fetch helper and a one-shot retry path that re-fetches a fresh Instagram media_url on failed primary fetch (context links only).
proxycache/services/insta_feed.py	Switches instaFeed to the new post-aware URL rewriting and adds fetch_fresh_media_url used by the proxy retry.
proxycache/tests.py	Adds tests for context binding, carousel child handling, retry behavior, and legacy behavior preservation.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.