feat: add consistent image job cancellation

CHANGELOG.md

@@ -55,6 +55,8 @@ Follows [Keep a Changelog](https://keepachangelog.com/en/1.0.0/) · versioned wi
   original temporary result; reusing a key for different request data is rejected.
 - **Casting Idempotency.** Character-to-voice matching now applies the same bounded, expiring idempotency contract,
   preventing repeated clicks from issuing duplicate LLM casting requests.
+- **Image Job Cancellation.** Consistent profile/reference jobs now expose a cooperative cancellation endpoint and
+  progress-panel Cancel control; queued work can be cancelled without interrupting unrelated ComfyUI jobs.
 - **Preview Blob Lifecycle.** Character-library and Rehearser local voice previews now revoke replaced or completed
   object URLs, reducing retained browser memory during repeated auditioning.
 - **Character Library Smoke Coverage.** Added deterministic browser coverage for opening Characters / Cast, profile

TODO.md

@@ -44,8 +44,8 @@ the first block before investing in architecture or polish.
   2,227 segments, 1,043 dialogue segments and 585 newline-fused segments, but zero
   adjacent same-speaker dialogue pairs for `_audiobookMergeAdjacentDialogue` to merge.
   The remaining segmentation work is upstream splitting/fusion, not another merge pass.
-- [ ] **S — Finish the remaining Playwright smoke checks.** Characters/Cast, profile
+- [x] **S — Finish the remaining Playwright smoke checks.** Characters/Cast, profile
-  sorting, column visibility, voice ratings, and starting/cancelling a generation job.
+  sorting, column visibility, voice ratings, and starting/cancelling a generation job are covered.
 - [x] **M — Fix the Schwester Larissa unclosed-guillemet split.** Attribution normalization now repairs dialogue
   split before mixed German guillemet closes (`»…‹` / `«…›`), with an epigraph regression test.
 - [ ] **M — Make profile portraits and character sheets use one canonical identity.**
@@ -120,7 +120,8 @@ the first block before investing in architecture or polish.
 ## P1 — Backend stability and bounded concurrency
 
 - [ ] Introduce background jobs for voice design, transcription, casting, and image generation.
-- [ ] Give every job an ID, progress, cancellation, timeout, retry, and idempotency key.
+- [ ] Give every job an ID, progress, cancellation, timeout, retry, and idempotency key. Consistent image jobs now
+  provide all except retry; voice design and casting remain synchronous request paths.
 - [x] Prevent duplicate generation when a user clicks twice for image, voice-design, and casting submissions.
 - [ ] Centralize provider health checks and error handling.
 - [x] Limit concurrency to 1–2 image generations, 1–2 voice generations, and 1 STT request per user. Consistent
@@ -170,7 +171,7 @@ the first block before investing in architecture or polish.
 - [x] Sort the profile column.
 - [x] Hide and show columns.
 - [x] Rate a voice.
-- [ ] Start and cancel a generation job.
+- [x] Start and cancel a generation job.
 
 ## P2 — Observability
 

routes/conversation.py

@@ -60,6 +60,7 @@ _attribution_llm_lock = threading.Lock()
 # entries are pruned an hour after completion so this can't grow unbounded.
 _consistent_image_jobs: dict[str, dict] = {}
 _consistent_image_jobs_lock = threading.Lock()
+_consistent_image_cancel_events: dict[str, threading.Event] = {}
 _consistent_image_idempotency: dict[str, str] = {}
 _consistent_image_slots = threading.BoundedSemaphore(2)
 _CAST_IDEMPOTENCY: dict[str, tuple[str, dict, float]] = {}
@@ -89,9 +90,11 @@ def _cij_new(job_id: str, name: str) -> None:
             "name": name, "stage": "queued", "detail": "Queued…",
             "started": now, "updated": now, "done": False, "error": None, "result": None,
         }
+        _consistent_image_cancel_events[job_id] = threading.Event()
         cutoff = now - 3600
         for stale_id in [j for j, v in _consistent_image_jobs.items() if v.get("done") and v.get("updated", 0) < cutoff]:
             _consistent_image_jobs.pop(stale_id, None)
+            _consistent_image_cancel_events.pop(stale_id, None)
             for key, mapped_id in list(_consistent_image_idempotency.items()):
                 if mapped_id == stale_id:
                     _consistent_image_idempotency.pop(key, None)
@@ -106,6 +109,17 @@ def _cij_set(job_id: str, **fields) -> None:
         job["updated"] = time.time()
 
 
+class _ConsistentImageCancelled(Exception):
+    """Internal cooperative cancellation signal for a consistent-image worker."""
+
+
+def _cij_check_cancelled(job_id: str) -> None:
+    with _consistent_image_jobs_lock:
+        event = _consistent_image_cancel_events.get(job_id)
+    if event is not None and event.is_set():
+        raise _ConsistentImageCancelled()
+
+
 @contextlib.contextmanager
 def _watchdog_close(resp, timeout_seconds):
     """Force-close `resp`'s connection if the wrapped block hasn't finished
@@ -3191,12 +3205,27 @@ async def character_generate_consistent_images(request: Request):
             _consistent_image_idempotency[idempotency_key] = job_id
 
     def _worker() -> None:
-        acquired = _consistent_image_slots.acquire(timeout=900)
+        acquired = False
+        try:
+            for _ in range(900):
+                _cij_check_cancelled(job_id)
+                if _consistent_image_slots.acquire(timeout=1):
+                    acquired = True
+                    break
+        except _ConsistentImageCancelled:
+            _cij_set(job_id, stage="cancelled", detail="Cancelled", done=True,
+                     cancelled=True, error="Generation cancelled")
+            with _consistent_image_jobs_lock:
+                _consistent_image_cancel_events.pop(job_id, None)
+            return
         if not acquired:
             _cij_set(job_id, stage="error", detail="Image queue timed out", done=True,
                      error="No image-generation slot became available within 15 minutes")
+            with _consistent_image_jobs_lock:
+                _consistent_image_cancel_events.pop(job_id, None)
             return
         try:
+            _cij_check_cancelled(job_id)
             identity_image = identity_image_input
             if identity_image.startswith("/"):
                 # Library records store same-origin image URLs rather than
@@ -3210,6 +3239,7 @@ async def character_generate_consistent_images(request: Request):
                 except Exception as exc:
                     raise RuntimeError(f"Could not load the existing identity image: {exc}")
             if not identity_image:
+                _cij_check_cancelled(job_id)
                 _cij_set(job_id, stage="identity", detail="Generating shared identity portrait…")
                 identity_image = _comfyui_run(
                     comfy_url, _bundled_character_identity_workflow(),
@@ -3218,14 +3248,15 @@ async def character_generate_consistent_images(request: Request):
                     # stale_seconds doesn't fire a false "may be stuck"
                     # warning) — the elapsed counter itself is computed
                     # client-side from "started", not baked into this text.
-                    progress=lambda elapsed: _cij_set(job_id),
+                    progress=lambda elapsed: (_cij_check_cancelled(job_id), _cij_set(job_id)),
                 )
+            _cij_check_cancelled(job_id)
             _cij_set(job_id, stage="render", detail="Rendering profile, turnaround, scene & props…")
             result = _comfyui_run_consistent_images(
                 comfy_url, _bundled_consistent_comfyui_workflow(), identity_image,
                 profile_prompt, front_prompt, side_prompt, back_prompt, scene_prompt, prop_prompts,
                 generation_key,
-                progress=lambda elapsed: _cij_set(job_id),
+                progress=lambda elapsed: (_cij_check_cancelled(job_id), _cij_set(job_id)),
             )
             result = dict(result or {})
             result["prompt_version"] = _CHARACTER_VISUAL_PROMPT_VERSION
@@ -3234,10 +3265,15 @@ async def character_generate_consistent_images(request: Request):
             _cij_set(job_id, stage="done", detail="Done", done=True, result=result)
         except HTTPException as exc:
             _cij_set(job_id, stage="error", detail="Failed", done=True, error=str(exc.detail))
+        except _ConsistentImageCancelled:
+            _cij_set(job_id, stage="cancelled", detail="Cancelled", done=True,
+                     cancelled=True, error="Generation cancelled")
         except Exception as exc:
             _cij_set(job_id, stage="error", detail="Failed", done=True, error=str(exc))
         finally:
             _consistent_image_slots.release()
+            with _consistent_image_jobs_lock:
+                _consistent_image_cancel_events.pop(job_id, None)
 
     threading.Thread(target=_worker, daemon=True, name=f"consistent-images-{job_id}").start()
     return {"job_id": job_id}
@@ -3261,6 +3297,27 @@ async def character_generate_consistent_images_status(job_id: str):
     return payload
 
 
+@router.post("/api/character-generate-consistent-images/cancel/{job_id}")
+async def character_generate_consistent_images_cancel(job_id: str):
+    """Request cooperative cancellation of a queued or running image job.
+
+    ComfyUI does not provide a safe per-client interrupt in all supported
+    versions, so the worker stops at its next stage/poll checkpoint without
+    interrupting another user's GPU job.
+    """
+    with _consistent_image_jobs_lock:
+        job = _consistent_image_jobs.get(job_id)
+        event = _consistent_image_cancel_events.get(job_id)
+        if job is None:
+            raise HTTPException(404, "Unknown or expired job id")
+        if job.get("done"):
+            return {"ok": True, "already_done": True, "cancelled": bool(job.get("cancelled"))}
+        if event is not None:
+            event.set()
+        job.update({"stage": "cancelling", "detail": "Cancelling…", "updated": time.time()})
+    return {"ok": True, "cancelled": True}
+
+
 def _character_generate_image_sync(provider: str, model: str, prompt: str, keys: dict, _settings: dict, image_kind: str = "portrait") -> dict:
     if provider == "openai":
         api_key = (keys.get("openai_image") or "").strip()

static/js/library-characters.js

@@ -3312,6 +3312,7 @@ function _cijShowPanel(name) {
   panel.innerHTML = '<div class="cij-progress-head">'
     + '<span class="mdi mdi-face-recognition"></span>'
     + '<span class="cij-progress-title">Generating images — ' + escHtml(name || 'Character') + '</span>'
+    + '<button type="button" class="btn-secondary btn-sm cij-progress-cancel">Cancel</button>'
     + '<button type="button" class="cij-progress-close" title="Hide (keeps running in the background)"><span class="mdi mdi-close"></span></button>'
     + '</div>'
     + '<div class="cij-progress-steps">' + stepsHtml + '</div>'
@@ -3319,6 +3320,19 @@ function _cijShowPanel(name) {
     + '<div class="cij-progress-elapsed">Elapsed 0:00</div>';
   document.body.appendChild(panel);
   panel.querySelector('.cij-progress-close').addEventListener('click', function () { panel.remove(); });
+  const cancelButton = panel.querySelector('.cij-progress-cancel');
+  cancelButton.addEventListener('click', async function () {
+    cancelButton.disabled = true;
+    cancelButton.textContent = 'Cancelling…';
+    try {
+      const response = await fetch('/api/character-generate-consistent-images/cancel/' + encodeURIComponent(panel.dataset.jobId), { method: 'POST' });
+      if (!response.ok) throw new Error('Cancellation request failed');
+    } catch (error) {
+      cancelButton.disabled = false;
+      cancelButton.textContent = 'Cancel';
+      toast(error.message || 'Could not cancel generation', 'error');
+    }
+  });
   return {
     update: function (stage, detail, elapsed, warn) {
       const p = document.getElementById('cij-progress-panel');
@@ -3356,6 +3370,7 @@ function _cijShowPanel(name) {
       p.classList.add(ok ? 'cij-progress-ok' : 'cij-progress-err');
       setTimeout(function () { p.remove(); }, ok ? 2500 : 7000);
     },
+    setJobId: function (jobId) { panel.dataset.jobId = jobId; },
   };
 }
 
@@ -3366,7 +3381,8 @@ function _cijFmtElapsed(seconds) {
 }
 
 // Polls the background job until it finishes, updating the panel each tick.
-// Flagged "stuck" (not aborted — there's no server-side cancel wired up yet)
+// Flagged "stuck" while the backend reports no progress; cancellation is
+// cooperative and completes at the next provider polling checkpoint.
 // once 60s pass with no backend-side progress update, since the poll interval
 // itself is only ~1.5s so any longer gap means the worker thread stopped
 // reporting, not that ComfyUI is merely slow.
@@ -3420,6 +3436,7 @@ async function _charAutoGenerateConsistentImages(rec, visualStyle) {
   if (!r.ok) { const e = await r.json().catch(function () { return {}; }); throw new Error(e.detail || r.statusText); }
   const { job_id } = await r.json();
   const panel = _cijShowPanel(rec.name);
+  panel.setJobId(job_id);
   const d = await _cijPollJob(job_id, panel);
   if (!d.profile || !d.reference) throw new Error('ComfyUI returned an incomplete image pair');
   if (typeof clSetImage === 'function') await clSetImage(rec.id, d.profile);

tests/characters.spec.js

@@ -57,3 +57,35 @@ test('Characters / Cast supports profile sorting, column visibility, and voice r
   await aliceRating.click();
   await expect(aliceRating).toHaveClass(/is-on/);
 });
+
+test('consistent image generation can be cancelled from its progress panel', async ({ page }) => {
+  let cancelled = false;
+  let cancelRequests = 0;
+  await page.route('**/api/characters*', route => route.fulfill({ json: { characters: [] } }));
+  await page.route('**/api/voices*', route => route.fulfill({ json: [] }));
+  await page.route('**/api/settings*', route => route.fulfill({ json: {} }));
+  await page.route('**/api/character-generate-consistent-images', route => route.fulfill({ json: { job_id: 'job-cancel' } }));
+  await page.route('**/api/character-generate-consistent-images/cancel/job-cancel', async route => {
+    cancelRequests += 1;
+    cancelled = true;
+    await route.fulfill({ json: { ok: true, cancelled: true } });
+  });
+  await page.route('**/api/character-generate-consistent-images/status/job-cancel', route => route.fulfill({
+    json: cancelled
+      ? { stage: 'cancelled', detail: 'Cancelled', elapsed: 1, stale_seconds: 0, done: true, cancelled: true, error: 'Generation cancelled', result: null }
+      : { stage: 'identity', detail: 'Generating shared identity portrait…', elapsed: 0, stale_seconds: 0, done: false, error: null, result: null },
+  }));
+
+  await page.goto('/', { waitUntil: 'domcontentloaded' });
+  await page.waitForFunction(() => typeof window._charAutoGenerateConsistentImages === 'function', null, { timeout: 15_000 });
+  await page.evaluate(() => {
+    window.__cancelRun = window._charAutoGenerateConsistentImages({
+      id: 'cancel-character', name: 'Cancel Character', book: 'Smoke Book',
+      sheet: { image_prompt: 'portrait', concept_art_prompt: 'reference' },
+    }, 'cinematic').catch(() => null);
+  });
+  await expect(page.locator('#cij-progress-panel')).toBeVisible({ timeout: 10_000 });
+  await page.locator('#cij-progress-panel .cij-progress-cancel').click();
+  await expect.poll(() => cancelRequests).toBe(1);
+  await expect(page.locator('#cij-progress-panel')).toHaveClass(/cij-progress-err/, { timeout: 5_000 });
+});

tests/test_conversation.py

@@ -7,7 +7,14 @@ import unittest
 
 os.environ.setdefault("CONFIG_DIR", tempfile.mkdtemp(prefix="ttvc-conversation-test-config-"))
 
-from routes.conversation import _attribution_parse, _fallback_attribute_response
+from routes.conversation import (
+    _attribution_parse,
+    _cij_check_cancelled,
+    _cij_new,
+    _consistent_image_cancel_events,
+    _consistent_image_jobs,
+    _fallback_attribute_response,
+)
 
 
 class AttributionParserTests(unittest.TestCase):
@@ -29,6 +36,26 @@ class AttributionParserTests(unittest.TestCase):
         self.assertTrue(dialogue["text"].endswith("Land. Lieblich bisweilen, mild und voll der guten Menschen. ‹"))
         self.assertEqual(narration["text"], "Zitiert aus dem Buch.")
 
+    def test_consistent_image_job_can_be_cancelled(self):
+        from fastapi.testclient import TestClient
+        from server import app
+
+        job_id = "cancel-test"
+        _cij_new(job_id, "Test character")
+        try:
+            with TestClient(app) as client:
+                response = client.post(
+                    f"/api/character-generate-consistent-images/cancel/{job_id}",
+                    headers={"origin": "http://testserver"},
+                )
+            self.assertEqual(response.status_code, 200)
+            self.assertTrue(response.json()["cancelled"])
+            with self.assertRaises(Exception):
+                _cij_check_cancelled(job_id)
+        finally:
+            _consistent_image_jobs.pop(job_id, None)
+            _consistent_image_cancel_events.pop(job_id, None)
+
 
 if __name__ == "__main__":
     unittest.main()