Add new dalle export script

lib/tasks/media_review.rake

@@ -42,4 +42,75 @@ namespace :media_review do
       puts "No AWS key for #{ai.id}"
     end
   end
+
+  desc "output for DallE stuff"
+  task dalle_output: :environment do |t, args|
+    s3 = Aws::S3::Client.new
+
+    # Create a folder in S3 for the output
+    time = Time.now
+    bucket_name = "dalle-export-#{time.strftime("%F")}-#{time.strftime("%H-%M")}"
+
+    bucket = Aws::S3::Bucket.new(bucket_name)
+
+    begin
+      bucket.create
+    rescue Aws::Errors::ServiceError => e
+      puts "Couldn't create bucket. Here's why: #{e.message}"
+    end
+
+    # Sanity check to make sure the bucket's there
+    resp = s3.list_buckets
+    raise "Bucket wasn't created??? #{bucket_name}" unless resp.buckets.map(&:name).include? bucket_name
+
+
+    # Get all media review that's images
+    # Go through all the media review
+      # output the high res image to the folder
+      # save the line to the CSV
+
+    image_archive_items = ArchiveItem.publically_viewable.select {|ai| !ai.images.empty? }
+    progress_bar = ProgressBar.create(title: "Media Review Items", total: MediaReview.count)
+
+    # Create a CSV object
+    csv_string = CSV.generate do |csv|
+      csv << ["id", "media_review.media_url", "media_review.date_published",
+              "media_authenticity_category_humanized", 'claim_review.author["name"]',
+              "media_review.url"]
+
+      image_archive_items.map do |archive_item|
+        # Download and then reupload (yes there's probably better ways to do this. Like a direct S3 copy... but meh)
+        media_review = archive_item.media_review
+        next if media_review.nil?
+        author_name = media_review.claim_reviews.empty? ? "" : media_review.claim_reviews.first.author["name"]
+
+        csv << [archive_item.id, media_review.media_url,  media_review.date_published,
+                media_review.media_authenticity_category_humanized, author_name,
+                media_review.url]
+
+        archive_item.images.each do |image|
+          image.image.download do |tempfile|
+            object = Aws::S3::Object.new(bucket_name, archive_item.id)
+            object.upload_file(tempfile.path)
+          end
+        end
+
+        progress_bar.increment
+      end
+    end
+
+    # Save the CSV
+    object_key = "results.csv"
+    object = Aws::S3::Object.new(bucket_name, object_key)
+    object.put(body: csv_string)
+
+    # Get a link to the folder for ease of access
+    begin
+      url = bucket.object(object_key).presigned_url(:get)
+      puts "Created presigned URL: #{url}"
+      # URI(url)
+    rescue Aws::Errors::ServiceError => e
+      puts "Couldn't create presigned URL for #{bucket.name}:#{object_key}. Here's why: #{e.message}"
+    end
+  end
 end