Commit Graph

16 Commits

Author SHA1 Message Date
soroush.asadi 076c2e577f fix(render): resolve template id for render jobs + mock-fallback when no .aep
Build backend images / build content-svc (push) Failing after 1m30s
Build backend images / build file-svc (push) Failing after 1m23s
Build backend images / build gateway (push) Failing after 5m47s
Build backend images / build identity-svc (push) Failing after 1m23s
Build backend images / build notification-svc (push) Failing after 1m51s
Build backend images / build render-svc (push) Failing after 1m23s
Build backend images / build studio-svc (push) Failing after 1m23s
THE bug behind "AEPFilePath is required for real AE render": CreateJob inserted
original_project_id = saved_project_id (VALUES $3,$3), so the claim looked for the
render bundle at templates/{saved_project_id}/ — which never exists. The bundle
lives at templates/{TEMPLATE_id}/. Now original_project_id is resolved from
studio.saved_projects.original_project_id (the template the project was built from).
(Direct-SQL test renders masked this by setting the template id explicitly.)

Also harden the node-agent: Run() falls back to mock render when AEPFilePath is
empty even if AE is installed (previously hard-errored), so a missing/un-promoted
template degrades gracefully instead of failing the job.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-06 19:37:59 +03:30
soroush.asadi 62807f5f41 fix(node-agent): resilient output upload — 60s HTTP timeout + 4× retry on upload-URL
After a CPU-heavy AE render+transcode the orchestrator/DB can be briefly slow;
the 15s client timeout made the post-render output-upload-url call fail and the
finished MP4 was dropped (completed without export). Bumped client timeout to 60s
and retry the upload-URL call up to 4× with backoff so a finished render's output
is never lost to a transient stall.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-06 18:57:09 +03:30
soroush.asadi e59f07df4e fix(node-agent): transcode AE render to MP4 with ffmpeg (real renders deliver MP4)
aerender can't reliably write H.264 directly in modern AE — it renders the
project's output module (Lossless AVI/MOV) and ignores the .mp4 extension,
producing a multi-GB .avi the agent then failed to find/upload.

- findRenderedOutput(): locate the file aerender actually wrote (output.avi/.mov/.mp4)
- transcodeToMP4(): ffmpeg → H.264 yuv420p + AAC + faststart; drops the lossless
  intermediate. ffmpeg located via $FFMPEG_PATH, beside the agent exe, or PATH.
- Graceful fallback: if ffmpeg is missing/fails, upload the raw render so the job
  still delivers a (large but valid) file.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-06 07:50:43 +03:30
soroush.asadi 9d499a89de fix(render): real AE render — pass -comp, fix export insert, ensure exports bucket
Three bugs surfaced bringing up a real After Effects node (verified: AE 2026
claimed + ran, but produced no usable output):

1. aerender got no -comp/-rqindex → "output argument ignored", nothing rendered.
   - Claim now returns comp_name from content.projects.render_aep_comp (e.g. "frfinal")
     via new Store.GetTemplateCompName; threaded through ClaimedJob → runner.Job →
     aerender args (`-comp <name>`, or `-rqindex 1` fallback when unknown).

2. CreateExportForJob INSERT passed render_quality as a bare param into an enum
   column → 500 ("output-upload-url HTTP 500"), so completed renders had no export.
   - Cast $8::render.render_quality (+ explicit casts for file_type/create_type enums).

3. flatrender-exports bucket didn't exist → uploads would fail anyway.
   - render-svc now MakeBucket(exports, templates) idempotently at startup.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-05 22:40:20 +03:30
soroush.asadi 52be5be93f feat(node-agent): production ops kit — Windows service + WireGuard mesh
config:
- LoadEnvFile(): reads agent.env beside the exe (or $AGENT_ENV_FILE) before env,
  so the sc.exe service needs no per-service environment plumbing; real env wins

deploy/ (new):
- build-windows.ps1     cross-compile → dist\ + stage the deploy kit
- agent.env.example     fully documented config template
- install-service.ps1   register as auto-start Windows service (native sc.exe),
                        crash-restart 3×/5s, no NSSM dependency
- uninstall-service.ps1 stop + remove
- wireguard-node.conf.template + setup-wireguard.ps1  node dials out only, no
                        public IP / inbound rules; tunnel installed as boot service
- README.md             full control-plane + node walkthrough, ops table, troubleshooting

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-05 12:20:48 +03:30
soroush.asadi f0ce286527 fix(scan): force-kill stale AE processes before each launch (fresh start)
Build backend images / build content-svc (push) Failing after 54s
Build backend images / build file-svc (push) Failing after 56s
Build backend images / build gateway (push) Failing after 57s
Build backend images / build identity-svc (push) Failing after 58s
Build backend images / build notification-svc (push) Failing after 1m4s
Build backend images / build render-svc (push) Failing after 2m27s
Build backend images / build studio-svc (push) Failing after 55s
PrepareFreshAE = taskkill AfterFX/aerender/AfterFXLib/dynamiclinkmanager/QT32
+ 2s settle + clear crash markers, then launch. A hung/zombie AE from a prior
job would otherwise block or corrupt the new run. RunScan now calls it.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-04 21:27:34 +03:30
soroush.asadi 6e5efbdb2c fix(scan): also clear AE AppStates registry to stop Safe Mode 'Crash Repair' dialog
Build backend images / build content-svc (push) Failing after 2m1s
Build backend images / build file-svc (push) Failing after 1m0s
Build backend images / build gateway (push) Failing after 56s
Build backend images / build identity-svc (push) Failing after 54s
Build backend images / build notification-svc (push) Failing after 54s
Build backend images / build render-svc (push) Failing after 46s
Build backend images / build studio-svc (push) Failing after 48s
SCRPriorState.json alone didn't suppress it — AE's per-session GUID under
HKCU\Software\Adobe\After Effects\AppStates persists after a kill/crash and
trips Safe Mode. ClearAECrashState now reg-deletes AppStates too (reg.exe, no dep).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-04 21:21:04 +03:30
soroush.asadi 47dd87c60b fix(scan): launch AE with the project as arg to bypass the Home screen
Build backend images / build content-svc (push) Failing after 1m13s
Build backend images / build file-svc (push) Failing after 1m35s
Build backend images / build gateway (push) Failing after 57s
Build backend images / build identity-svc (push) Failing after 1m28s
Build backend images / build notification-svc (push) Failing after 53s
Build backend images / build render-svc (push) Failing after 1m4s
Build backend images / build studio-svc (push) Failing after 55s
afterfx -r alone leaves AE on its empty Home/Start screen, which blocks the
script from running (AE sits idle on Untitled Project until the scan times out).
Now launch 'afterfx <aep> -r scan.jsx' so the project opens directly; scan.jsx
uses the already-open project and only app.open()s as a fallback.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-04 21:13:16 +03:30
soroush.asadi 0c461ff841 fix(import): IgnoreQueryFilters so revive sees soft-deleted scenes; clear AE crash state
Build backend images / build content-svc (push) Failing after 53s
Build backend images / build file-svc (push) Failing after 58s
Build backend images / build gateway (push) Failing after 1m1s
Build backend images / build identity-svc (push) Failing after 57s
Build backend images / build notification-svc (push) Failing after 59s
Build backend images / build render-svc (push) Failing after 49s
Build backend images / build studio-svc (push) Failing after 49s
- AepImportService: the global Scene HasQueryFilter(DeletedAt==null) was hiding
  soft-deleted rows, so the revive never matched and the importer re-inserted →
  scenes_project_id_key violation. Add .IgnoreQueryFilters() to the load. (apply
  now revives + returns 200, verified.)
- node-agent: ClearAECrashState() deletes AE's SCRPriorState.json before each
  launch so the 'Crash Repair Options' dialog can't hang a headless scan/render.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-04 20:20:41 +03:30
soroush.asadi 0a7dd9b84c feat(nodes): live CPU/RAM/disk monitoring in the node list
Build backend images / build content-svc (push) Failing after 45s
Build backend images / build file-svc (push) Failing after 55s
Build backend images / build gateway (push) Failing after 53s
Build backend images / build identity-svc (push) Failing after 54s
Build backend images / build notification-svc (push) Failing after 53s
Build backend images / build render-svc (push) Failing after 47s
Build backend images / build studio-svc (push) Failing after 51s
- node-agent: internal/metrics — read CPU% (GetSystemTimes), RAM (GlobalMemoryStatusEx),
  disk used%/total (GetDiskFreeSpaceEx) via stdlib kernel32 (no external dep; windows
  build + non-windows stub). Heartbeat now reports cpu_pct/ram_available_mb/disk_used_pct/
  disk_total_gb + ae_running.
- render-svc: heartbeat persists last_disk_pct + disk_total_gb (migration 29); RenderNode
  model + node SELECT/scan carry them.
- admin: rewrite NodesTable to the real RenderNode shape (fixes a pre-existing items/V2Node
  mismatch that left the list empty) + a CPU/RAM/disk bars column + stale-heartbeat flag.
- assets-bundle ingestion: ProjectMediaBundle (jszip) auto-maps project.zip → project/scene
  image/demo/colour + music; PatchProject gains image/full_demo/shared_colors_svg.
- scan: RGBA (4-number) colours recognised + frshare single-int controls detected.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-04 20:01:18 +03:30
soroush.asadi 6661f53734 fix(scan): Fix-mode scanner + dialog suppression + cancel/timer + importer revive
Build backend images / build content-svc (push) Failing after 1m25s
Build backend images / build file-svc (push) Failing after 1m10s
Build backend images / build gateway (push) Failing after 56s
Build backend images / build identity-svc (push) Failing after 53s
Build backend images / build notification-svc (push) Failing after 57s
Build backend images / build render-svc (push) Failing after 48s
Build backend images / build studio-svc (push) Failing after 1m5s
- scan.jsx: app.beginSuppressDialogs() + clean quit (no AE hang on font/footage
  dialogs); FIX-mode branch parses frl_c(x)t/m(y) layer names → scenes by c(x);
  flexible/mockup keep comp-based walk; FR_SCAN_MODE selects.
- render-svc: scan job carries project mode; cancel endpoint + node watchdog that
  kills AE on cancel; parseObjectURL handles minio:// (bucket in host); scan with
  no template fails cleanly; status guards so late results can't un-cancel.
- content importer: revive soft-deleted scenes instead of duplicate-inserting
  (fixes scenes_project_id_key unique violation); orphan diff ignores deleted.
- admin: scan dialog gets project-type picker + elapsed timer + Cancel button.
- node-agent: AE-2026 wiring (host port 5010, host-reachable presign endpoint),
  FR_SCAN_MODE plumbing. docs/aep-template-convention.md: per-type naming + bundles.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-04 19:06:08 +03:30
soroush.asadi 1ff6e494c0 @
Build backend images / build content-svc (push) Failing after 19s
Build backend images / build file-svc (push) Failing after 1m53s
Build backend images / build gateway (push) Failing after 16s
Build backend images / build identity-svc (push) Failing after 7m1s
Build backend images / build notification-svc (push) Failing after 7m24s
Build backend images / build render-svc (push) Failing after 3m12s
Build backend images / build studio-svc (push) Failing after 43s
feat: AE template scanner + scene editor + AEP bundle pipeline

Scene editor (admin): per-project Scenes / Shared Colors / Color Presets
manager (ProjectScenes) reachable from each project.

AEP bundle pipeline: upload .aep or .zip → stored once per template at
templates/{project_id}/(bundle.zip|template.aep); render claim probes and
returns is_bundle+md5; node-agent extracts the bundle, locates the .aep
(zip-slip guarded), and caches by md5 so repeated renders extract once.

AE template scanner ("read scenes/colours/configs from the AEP"):
- content-svc importer: POST /v1/projects/{id}/scan/{preview,apply} —
  review-diff-then-merge into scenes/elements/colours (manual edits kept).
- render-svc Go quick-scan: stdlib RIFX parser extracts comp names+durations
  (no AE) → POST /v1/template-scans/{id}/quick.
- render-svc AE scan jobs + node-agent runner: queue → node runs scan.jsx
  (reverse of legacy JSXGenerator conventions: frfinal/frshare/frl_/frd_) →
  posts ScanResult back. Migration 26_render_scan_jobs.
- admin UI: "اسکن از افترافکت" with quick/full engines + diff-review modal.

Verified: importer preview/apply, Go quick-scan end-to-end (synthetic .aep →
scene imported), bundle extract unit tests, RIFX parser unit tests.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
@
2026-06-04 10:39:45 +03:30
soroush.asadi 7f2f65dd8a feat(render+node-agent+admin): install fonts on all render nodes + verify
Build backend images / build content-svc (push) Failing after 53s
Build backend images / build file-svc (push) Failing after 47s
Build backend images / build gateway (push) Failing after 52s
Build backend images / build identity-svc (push) Failing after 58s
Build backend images / build notification-svc (push) Failing after 55s
Build backend images / build render-svc (push) Failing after 59s
Build backend images / build studio-svc (push) Failing after 48s
Push a font once → every node installs it → admin sees per-node status.

- render-svc: font_requests + node_fonts tables (mig 25); admin GET/POST/DELETE
  /v1/node-fonts (with per-node status matrix); internal (HMAC) GET pending +
  POST status for node-agents
- node-agent: fontSyncLoop polls pending fonts every 60s, downloads, installs
  (Windows Fonts dir + registry / macOS / linux fc-cache), reports Installed/Failed
- gateway: /v1/node-fonts/* → render
- admin /admin/node-fonts: upload a .ttf/.otf → install on all nodes; per-node
  Installed/Pending/Failed badges + counts + delete

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-03 06:33:48 +03:30
soroush.asadi bcc69f0a2e feat: complete node-agent pipeline, TLS proxy, billing cancel, password reset
Node-agent — full render pipeline (items 1-3):
- render-svc: ClaimedJob now includes aep_download_url (presigned MinIO GET,
  2h TTL, path=templates/{original_project_id}/template.aep)
- render-svc: POST /v1/internal/render/jobs/:id/output-upload-url
  allocates Export row + returns presigned MinIO PUT URL + export_id
- render-svc: db.CreateExportForJob() inserts export row with 30-day retention
- render-svc: InternalHandler now owns minio client (templatesBucket + exportsBucket)
  MINIO_TEMPLATES_BUCKET env var (default flatrender-templates)
- node-agent: runner/download.go — DownloadFile() + UploadFile() (stdlib only)
- node-agent: client.GetOutputUploadURL() + ClaimedJob.AEPDownloadURL field
- node-agent: runJob() full flow: download AEP → render → get upload URL →
  PUT output to MinIO → Complete(export_id)
  All steps are non-fatal with fallback (AEP miss → mock, upload fail → no export)

TLS reverse proxy (item 15):
- Caddyfile: three virtual hosts (DOMAIN, API_DOMAIN, STORAGE_DOMAIN)
  auto-TLS via Let's Encrypt; security headers; 512MB upload limit on API
- docker-compose.v2.yml: caddy:2-alpine service, ports 80/443/443udp,
  caddy_data + caddy_config volumes; env vars DOMAIN/API_DOMAIN/STORAGE_DOMAIN/ACME_EMAIL
- .env.v2.example: new Caddy + MINIO_TEMPLATES_BUCKET entries

Billing portal (item 5):
- Identity: POST /v1/users/me/plan/cancel — sets cancelled_at, auto_renew=false
  (access continues to expiry); 404 when no active plan
- POST /api/billing/cancel — frontend proxy, validates auth
- GET /api/billing/portal — redirects to /dashboard/settings?tab=billing
- SettingsBilling: "Cancel plan" button with confirm dialog + optimistic UI,
  "Change plan" button; becomes "use client" component

Password reset UI (item 7):
- POST /api/auth/password-reset — proxies /v1/auth/password/reset/request
  (always 200, anti-enumeration)
- POST /api/auth/password-reset-confirm — proxies /v1/auth/password/reset/confirm
- AuthPageContent: "Forgot password?" link on sign-in tab opens 2-step reset flow
  (email → OTP+new-password) without leaving the auth page

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-06-01 16:41:13 +03:30
soroush.asadi d7743a6fbe feat: live render preview — node agent pushes PNG frames, frontend displays them in real time
render-svc:
- db.UpdateJobPreview(): writes base64 PNG to render_jobs.image_preview_b64
  (only on active jobs; Done/Failed/Cancelled rows ignored)
- POST /v1/internal/render/jobs/:job_id/preview — node agent endpoint
- Route registered under /v1/internal (nodeAuth)

node-agent:
- runner.PreviewFn callback type alongside ProgressFn
- runner.preview.go: GeneratePreviewB64(percent, quality, resolution)
  — pure stdlib (image/png + encoding/base64), no external deps
  — 320×180 dark frame with animated progress bar + colored indicator dots
- mock render: pushes a preview frame at every step (5→95%)
- real AE render: pushes a preview frame every 30s
- client.UpdatePreview(): POST /v1/internal/render/jobs/:job_id/preview
- main.go: onPreview callback wires client.UpdatePreview() into runner.Run()

frontend:
- render-jobs.ts: RenderJobRow.preview_b64 field; read from progress endpoint
- status/route.ts: previewB64 included in JSON response
- RenderModal: aspect-ratio preview pane during polling — shows spinner until
  first frame arrives, then live-updates with each poll (every 3s);
  step label overlaid as badge bottom-right

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-06-01 09:42:03 +03:30
soroush.asadi ee421ccc68 feat(render-svc+node-agent): add job-claim endpoint and build node-agent skeleton
render-svc:
- db: ClaimJob() — atomic SELECT FOR UPDATE SKIP LOCKED; transitions job to
  Preparing, marks node Busy in a single transaction
- models: ClaimJobRequest + ClaimedJob types
- handlers/internal: POST /v1/internal/render/jobs/claim — 200 with job or 204 when queue empty
- main: register the claim route under /v1/internal (nodeAuth)

services/node-agent/ (new Go module github.com/flatrender/node-agent):
- internal/config: env-var based config (NODE_ID required, sensible defaults)
- internal/client: typed orchestrator HTTP client (Online, Heartbeat, ClaimJob,
  Complete, Fail, ReportCrash) — X-Node-Signature auth
- internal/runner: AE render via aerender.exe or mock (for dev without AE)
- cmd/agent/main: register online → heartbeat loop (5s) + poll loop (3s) →
  claim job → run render → report complete/fail; health endpoint on :7777
- Dockerfile: cross-compiles to Windows amd64 static binary

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-06-01 09:28:31 +03:30