Description

Upload the downloaded Vimeo masters (TASK-246) to YouTube via the upload pipeline (TASK-249), then update _data platform records so YouTube becomes primary. Newly-on-YouTube videos then re-enter transcription backfill (TASK-250) and caption/metadata (TASK-252/TASK-248). Outward-facing: private/unlisted first, public only on approval.

Acceptance Criteria

• #1 All migration-set masters are uploaded to YouTube (private/unlisted first)
• #2 _data/video_assets.yml is updated with new youtube ids and platform set to primary
• #3 Explicit human approval is required before each video goes public
• #4 Migrated videos are enqueued for transcription backfill and captions

Implementation Plan

Re-upload migrated Vimeo masters to YouTube + update platform records

Inputs / dependencies

• Masters from TASK-246 in videos/masters/ (27, integrity-passed).
• Upload pipeline from TASK-249 (do NOT reimplement the uploader here; call it).
• Manifest _data/vimeo_migration_manifest.yml drives ordering + carries caption_readiness (19 caption-ready / 8 needs-transcription).

Upload flow (outward-facing safety)

1. Upload each master via TASK-249’s pipeline with privacyStatus = unlisted (never public on first upload). Set title/description/tags from the existing asset record (content is owner-provided; do not rewrite).
2. Capture the returned YouTube video id; write it into the manifest (youtube_id, migration_state: unlisted).
3. Per-video human-approval gate before public: publishing to public is a separate, explicit, per-asset step (e.g. approving in the manifest / a --publish <asset_id> action gated on migration_state == unlisted). No batch flip-all-public. Only on approval does the pipeline set privacy public and migration_state: public.

Update _data/video_assets.yml (structure/data only)

• For each migrated asset, APPEND a platform: youtube entry to platforms[] (asset_id, url: https://www.youtube.com/watch?v=<id>, embed_url: https://www.youtube.com/embed/<id>, carry duration/thumbnail as available) and set primary_platform: youtube.
• KEEP the existing vimeo platform entry for now — removal is TASK-256’s job, gated on migration-complete. This makes the change reversible and lets templates (which bias to youtube; see _includes/video-stage.html, schema-factory.html, video-asset-player.html) auto-repoint the embed the moment the youtube entry exists.

Transcription backfill + captions (do NOT blanket-enqueue)

• 19 caption-ready assets already have _data/transcripts/<transcript_id>.yml with structured turns → they need only a caption track (TASK-252): push the existing transcript to the new YouTube id as a caption/subtitle track. Do not re-transcribe.
• 8 needs-transcription assets have no transcript → enqueue for transcription backfill (TASK-250). Reuse bin/batch_ztranscribe.rb’s existing selection logic (transcript_id empty AND a youtube platform id present) — post-upload those exact 8 become eligible automatically, so no custom re-enqueue of the 19. Enqueue via zdots-ctx enqueue transcription '{"url":"https://www.youtube.com/watch?v=<id>","video_asset_id":"<id>"}' (fingerprint-idempotent; worker drains).

Acceptance mapping

• AC#1: all 27 masters uploaded, unlisted first.
• AC#2: _data/video_assets.yml gains youtube ids + primary_platform: youtube (vimeo entry retained until 256).
• AC#3: public transition is per-asset, explicit, human-approved — enforced by the migration_state gate.
• AC#4: 8 enqueued for transcription (reusing batch_ztranscribe selection); 19 handed to captions (TASK-252) from existing transcripts.

Boundary

No content rewriting. No public publish without explicit approval. Planning only — no uploads/API calls performed now.

Definition of Done

• #1 AC criteria is completed and the change has been verified