refactor(server): extract the loops routes into routers/loops.py (R3)

[byrongamatos]

Third router in the R3 train (after audio-effects #834, artist-aliases #838). Move-only.

Practice loops — saved A/B regions per song. router_scan.py ranks it at 0 setattr targets, 0 helpers to relocate; meta_db-only, own banner, contiguous.

• GET /api/loops · POST /api/loops · DELETE /api/loops/{loop_id}

Bodies verbatim; @app → @router, meta_db → appstate.meta_db. include_router at the original site. server.py: 9,337 → 9,302.

Verification

• Route table identical to origin/main (143, paths/methods/order).
• No test retargeting (test_demo_mode only names the paths in middleware regexes).
• Boot smoke: POST auto-names Loop N, GET reads back, DELETE removes, missing fields → error; demo mode 403s both writes, allows the read.
• pytest 2398 passed; packaging guard green; pyflakes clean; npm run lint 0 errors; Codex 0 findings.

Next candidates (router_scan.py): wanted/saved need _clean_str (shared by 10 callers) relocated to a lib/ home first — a small shared-helpers decision before those move.

🤖 Generated with Claude Code

Summary by CodeRabbit

• New Features

  • Added API support for listing, saving, and deleting audio loops.
  • Automatically assigns a default name when saving an unnamed loop.
  • Loop results are ordered by their start time.

• Documentation

  • Updated release notes and project size documentation to reflect router extraction progress.

[coderabbitai]

📝 Walkthrough

Walkthrough

The Loops API was extracted from server.py into lib/routers/loops.py, preserving its three endpoints and database behavior. server.py now mounts the router, and related changelog and size-exemption documentation were updated.

Changes

Loops router extraction

Layer / File(s)	Summary
Implement Loops router lib/routers/loops.py	Adds GET, POST, and DELETE loop endpoints using appstate.meta_db, including validation, default naming, persistence, ordering, and deletion.
Mount router and update progress records server.py, CHANGELOG.md, docs/size-exemptions.md	Replaces the inlined Loops endpoints with router registration and updates documented extraction progress and line counts.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Possibly related PRs

• got-feedBack/feedBack#833: Adds the appstate.meta_db setup consumed by the extracted Loops router.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately summarizes the main change: extracting the loops routes from server.py into routers/loops.py.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch refactor/r3-router-loops

---

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

[coderabbitai]

Actionable comments posted: 2

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@lib/routers/loops.py`:
- Around line 18-21: The loops list query currently accesses the shared SQLite
connection without synchronization. In the relevant list handler, wrap both the
`appstate.meta_db.conn.execute(...)` call and its `.fetchall()` in `with
appstate.meta_db._lock:` so the entire read is protected from concurrent
POST/DELETE transactions.
- Around line 34-43: Make loop name generation atomic in the unnamed-loop
handling code: acquire appstate.meta_db._lock before executing the COUNT query,
then keep the same lock held through name assignment, INSERT, and commit.
Consolidate the existing lock scopes so the count and persistence occur within
one critical section.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: cec452fc-86f7-41c5-9684-f4bde2cbbe31

📥 Commits

Reviewing files that changed from the base of the PR and between 6c98aba and b9c3565.

📒 Files selected for processing (4)

• CHANGELOG.md
• docs/size-exemptions.md
• lib/routers/loops.py
• server.py

[byrongamatos]

Both findings are real, and both are pre-existing — the COUNT(*) sat outside _lock and list_loops' SELECT was unlocked on origin/main too (verified byte-identical to 6c98aba:server.py). This PR moved the code verbatim, so they correctly weren't part of a move-only change. Fixing them here would have made the extraction non-verbatim.

Landing both in a small follow-up: move the COUNT(*) inside the with _lock block (atomic name generation) and wrap the list read in the lock. Severity is low in context — FeedBack is single-user (Principle I), so two concurrent unnamed-loop POSTs essentially can't happen — but the fix is one line each, so it's worth doing. Follow-up PR incoming; will link it here.

[byrongamatos]

Follow-up fix opened: #840 — locks the count+insert and the list read, with a barrier-based concurrency test (negative-checked 5/5).