fix(archive): add script to stage completed transcripts and fix transcripts pipeline

Author: just3ws

bin/stage_completed_transcripts.rb

@@ -0,0 +1,30 @@
+#!/usr/bin/env ruby
+require 'yaml'
+require 'fileutils'
+
+assets = YAML.load_file("_data/video_assets.yml")["items"]
+staging_dir = "tmp/transcript-id-staging"
+FileUtils.mkdir_p(staging_dir)
+
+staged_count = 0
+Dir.glob(File.expand_path("~/Downloads/transcripts/*/*.txt")).each do |txt_file|
+  yt_id = File.basename(txt_file, ".txt")
+  
+  # Find the corresponding video_asset_id
+  asset = assets.find do |a|
+    a["platforms"]&.any? { |p| p["platform"] == "youtube" && p["asset_id"] == yt_id }
+  end
+  
+  if asset
+    video_asset_id = asset["id"]
+    dest_txt = File.join(staging_dir, "#{video_asset_id}.txt")
+    
+    unless File.exist?(dest_txt)
+      FileUtils.cp(txt_file, dest_txt)
+      puts "Staged: #{yt_id} -> #{video_asset_id}.txt"
+      staged_count += 1
+    end
+  end
+end
+
+puts "Staged #{staged_count} new transcripts."

bin/transcripts

@@ -144,7 +144,7 @@ case "$cmd" in
         --apply
     fi
 
-    ./bin/audit_transcripts.rb
+    bundle exec rake transcript:audit
     if [ "$SKIP_VALIDATE" -ne 1 ]; then
       run_validate
     fi
@@ -162,7 +162,7 @@ case "$cmd" in
     fi
     ;;
   audit|status)
-    ./bin/audit_transcripts.rb
+    bundle exec rake transcript:audit
     ;;
   validate)
     run_validate