Fix social URL host matching

[anirudh5harma]

Summary

• Parse social links as URLs before classifying Twitter/X and LinkedIn profile URLs
• Match social domains by hostname instead of substring search
• Add regression tests for spoofed LinkedIn paths and non-Twitter domains containing x.com/twitter.com text

Closes #1173

Tests

• bun test apps/web/lib/url-helpers.test.ts
• bun test apps/web/lib/*.test.ts
• bunx biome check apps/web/lib/url-helpers.ts apps/web/lib/url-helpers.test.ts
• bunx turbo run check-types --filter='@supermemory/ai-sdk' --filter='@supermemory/memory-graph'

---

Session Details

• Session: View Session
• Requested by: Mahesh Sanikommu (mahesh@supermemory.com)
• Address comments on this PR. Add (aside) to your comment to have me ignore it.

[vorflux]

Vorflux skipped this auto review because this account has reached its Auto Review daily review limit (20/20). You can change this in Auto Review Settings: https://us1.vorflux.com/supermemory/settings?section=pull-requests

[vorflux]

Same review finding from the superseded PR still applies here.

---

Review with Vorflux

[vorflux]

Verified the update: the uppercase HTTP:// / HTTPS:// regression is fixed and the new collectValidUrls regression cases pass.

Checks run:

• npx --yes bun@1.3.6 test apps/web/lib/url-helpers.test.ts passed
• npx --yes bun@1.3.6 test apps/web/lib/*.test.ts passed
• npx --yes bun@1.3.6 x biome check apps/web/lib/url-helpers.ts apps/web/lib/url-helpers.test.ts passed
• Typecheck command still fails on the pre-existing/local @supermemory/memory-graph missing vitest type declarations, unrelated to this diff.

[vorflux]

Validated this PR. I think it is needed: main currently classifies URLs by substring, so a normal webpage like https://notx.com/profile is treated as Twitter in the document modal before the webpage branch runs. This patch fixes that by parsing the URL and matching the hostname/path instead.

No blockers found. Focused tests pass, all apps/web/lib tests pass, and Biome passes on the touched files. Typecheck still fails on the existing @supermemory/memory-graph missing vitest declarations, which is unrelated to this diff.

---

Review with Vorflux

[MaheshtheDev]

LGTM

[anirudh5harma]

@MaheshtheDev this needs someone for another review

[MaheshtheDev]

LGTM. Re-reviewed the latest head after the test file removal and verified the URL helper behavior manually.

Checks run:

• npx --yes bun@1.3.6 test apps/web/lib/*.test.ts passed
• npx --yes bun@1.3.6 x biome check apps/web/lib/url-helpers.ts passed
• Manual Bun check confirms uppercase HTTP/HTTPS Twitter links are filtered, notx.com is kept, spoofed LinkedIn paths are rejected, and LinkedIn company pages are rejected
• Typecheck still fails only on the existing @supermemory/memory-graph missing vitest type declarations, unrelated to this diff