feat(serve): add ingestion API routes + integration tests (GH-328)

[fagemx]

Summary

• Add 6 HTTP API routes to edda-serve exposing the ingestion trigger engine and suggestion queue from edda-ingestion (feat(ingestion): add trigger tables + evaluator engine #326, feat(ingestion): add suggestion queue + workflow #327):
  • POST /api/ingestion/evaluate — evaluate trigger (auto/suggest/skip)
  • POST /api/ingestion/records — manual ingestion (bypass trigger)
  • GET /api/ingestion/records — list ingestion records (with filters)
  • GET /api/ingestion/suggestions — list pending suggestions
  • POST /api/ingestion/suggestions/{id}/accept — accept and write to ledger
  • POST /api/ingestion/suggestions/{id}/reject — reject without writing
• Add 8 integration tests covering all three ingestion flows (auto/suggest/skip), manual ingestion, error cases (404/400), and suggest-then-reject
• Thin pass-through handler pattern: all business logic stays in edda-ingestion, handlers are mechanical adapters
• Pre-check pattern for accept/reject to produce proper HTTP status codes (404/409) consistent with post_draft_approve

Closes #328

Test plan

• cargo fmt --check -p edda-serve — formatting clean
• cargo clippy -p edda-serve -- -D warnings — zero warnings
• cargo test -p edda-serve — 105 tests pass (8 new ingestion tests)
• cargo test -p edda-ingestion — 53 tests pass (no regressions)

🤖 Generated with Claude Code

[fagemx]

Code Review: PR #372 (Round 1)

Summary

This PR adds 6 ingestion API routes to edda-serve exposing the trigger engine and suggestion queue from edda-ingestion. The implementation follows the established thin pass-through handler pattern, with all business logic staying in edda-ingestion. Code quality is high overall — proper error handling, workspace locking, and input validation.

Key Findings

Critical Issues (P0)

None.

High Priority (P1)

1. Inefficient trigger_type filter (crates/edda-serve/src/lib.rs:4316): get_ingestion_records uses serde_json::to_value(r.trigger_type) to serialize a Copy enum into JSON just to compare it as a string. This allocates a serde_json::Value per record. A simpler approach: match against the enum variants directly or use Display/to_string().

2. Missing test for 409 Conflict on double-accept/reject (crates/edda-serve/src/lib.rs:4351-4355, 4375-4379): The handlers have pre-check logic returning 409 Conflict when a suggestion is already accepted/rejected, but no test exercises this path. This is the only untested handler code path.

Testing Review

Coverage

• Auto-ingest via evaluate: covered (ingestion_auto_ingest_via_evaluate)
• Suggest-then-accept flow: covered (ingestion_suggest_then_accept)
• Skip flow: covered (ingestion_never_ingest_skips)
• Manual ingest: covered (ingestion_manual_record)
• 404 on nonexistent accept/reject: covered (ingestion_accept_nonexistent_404, ingestion_reject_nonexistent_404)
• Suggest-then-reject: covered (ingestion_suggest_then_reject)
• Invalid source_layer 400: covered (ingestion_evaluate_invalid_layer_400)
• Missing: 409 Conflict on already-accepted/rejected suggestion

Convention Compliance

• All tests use tempfile::tempdir() with real SQLite (no mocking) — correct
• All tests use #[tokio::test] — correct
• Test naming follows existing convention — correct

Testing Verdict: Adequate (minor gap on 409 path)

Verdict: Changes Requested

Fixing P1 issues and will re-review.

---

Round 1 of automated review-fix loop

[fagemx]

Code Review: PR #372 (Round 2) — LGTM 🎉

All P0 and P1 issues have been resolved.

Summary

This PR adds 6 ingestion API routes to edda-serve with 9 integration tests (8 original + 1 added for 409 Conflict coverage). The implementation follows the established thin pass-through handler pattern — all business logic stays in edda-ingestion, handlers are mechanical adapters with proper input validation, workspace locking, and HTTP status codes.

Fixes applied in round 1:

• Replaced serde_json::to_value trigger_type filter with direct match (avoids unnecessary allocation)
• Added ingestion_accept_already_accepted_409 test covering the 409 Conflict code path

Verdict: LGTM ✅

No critical or high-priority issues remaining. This PR is ready for merge.

---

Completed after 2 round(s) of automated review-fix loop