feat(archive): add automated batch transcription script for pending youtube assets

Author: just3ws

bin/batch_ztranscribe.rb

@@ -0,0 +1,46 @@
+#!/usr/bin/env ruby
+require 'yaml'
+require 'fileutils'
+
+interviews = YAML.load_file("_data/interviews.yml")["items"]
+assets = YAML.load_file("_data/video_assets.yml")["items"]
+
+pending_interviews = interviews.select do |i|
+  v = assets.find { |a| a["id"] == i["video_asset_id"] }
+  v && (!v["transcript_id"] || v["transcript_id"].empty?) && v["platforms"]&.any? { |p| p["platform"] == "youtube" && p["asset_id"] }
+end
+
+puts "Found #{pending_interviews.size} pending interviews to transcribe."
+
+staging_dir = "tmp/transcript-id-staging"
+FileUtils.mkdir_p(staging_dir)
+
+pending_interviews.each_with_index do |i, idx|
+  v = assets.find { |a| a["id"] == i["video_asset_id"] }
+  yt = v["platforms"].find { |p| p["platform"] == "youtube" }
+  yt_id = yt["asset_id"]
+  video_asset_id = v["id"]
+  
+  puts "[#{idx + 1}/#{pending_interviews.size}] Transcribing Interview: #{i['id']} (YouTube ID: #{yt_id})"
+  
+  # Run the transcription pipeline
+  url = "https://www.youtube.com/watch?v=#{yt_id}"
+  cmd = "/Users/mike/.config/zsh/recipes/yt-transcribe \"#{url}\""
+  
+  system(cmd)
+  
+  # Move the output to staging
+  # yt-transcribe saves to ~/Downloads/transcripts/<yt_id>/<yt_id>.txt
+  source_txt = File.expand_path("~/Downloads/transcripts/#{yt_id}/#{yt_id}.txt")
+  
+  if File.exist?(source_txt)
+    # Prefixing with video_asset_id ensures deterministic mapping during ingest
+    dest_txt = File.join(staging_dir, "#{video_asset_id}.txt")
+    FileUtils.cp(source_txt, dest_txt)
+    puts "  ✓ Staged transcript as #{dest_txt}"
+  else
+    puts "  ✕ Failed to find output transcript at #{source_txt}"
+  end
+end
+
+puts "\nTranscription batch complete. You can now run './bin/transcripts ingest --source-dir tmp/transcript-id-staging --min-confidence 0.9 --auto-commit'"