added a job to clean up orphan uploads

app/jobs/scheduled/clean_up_uploads.rb

@@ -0,0 +1,24 @@
+module Jobs
+
+  class CleanUpUploads < Jobs::Scheduled
+    recurrence { hourly }
+
+    def execute(args)
+
+      uploads_used_in_posts = PostUpload.uniq.pluck(:upload_id)
+      uploads_used_as_avatars = User.uniq.where('uploaded_avatar_id IS NOT NULL').pluck(:uploaded_avatar_id)
+
+      grace_period = [SiteSetting.uploads_grace_period_in_hours, 1].max
+
+      Upload.where("created_at < ?", grace_period.hour.ago)
+            .where("id NOT IN (?)", uploads_used_in_posts + uploads_used_as_avatars)
+            .find_each do |upload|
+        # disable this for now.
+        #upload.destroy
+      end
+
+    end
+
+  end
+
+end

app/models/site_setting.rb

@@ -184,6 +184,7 @@ class SiteSetting < ActiveRecord::Base
   setting(:enforce_global_nicknames, true)
   setting(:discourse_org_access_key, '')
 
+  setting(:uploads_grace_period_in_hours, 1)
   setting(:enable_s3_uploads, false)
   setting(:s3_access_key_id, '')
   setting(:s3_secret_access_key, '')

config/locales/server.en.yml

@@ -619,6 +619,7 @@ en:
 
     suggested_topics: "Number of suggested topics shown at the bottom of a topic"
 
+    uploads_grace_period_in_hours: "Grace period (in hours) before an orphan upload is removed."
     enable_s3_uploads: "Place uploads on Amazon S3"
     s3_upload_bucket: "The Amazon S3 bucket name that files will be uploaded into. WARNING: must be lowercase (cf. http://docs.aws.amazon.com/AmazonS3/latest/dev/BucketRestrictions.html)"
     s3_access_key_id: "The Amazon S3 access key id that will be used to upload images"

db/migrate/20131014203951_backfill_post_upload_reverse_index.rb

@@ -0,0 +1,42 @@
+class BackfillPostUploadReverseIndex < ActiveRecord::Migration
+
+  def up
+    # clean the reverse index
+    execute "TRUNCATE TABLE post_uploads"
+
+    # fill the reverse index up
+    Post.select([:id, :cooked]).find_each do |post|
+      doc = Nokogiri::HTML::fragment(post.cooked)
+      # images
+      doc.search("img").each { |img| add_to_reverse_index(img['src'], post.id) }
+      # thumbnails and/or attachments
+      doc.search("a").each { |a| add_to_reverse_index(a['href'], post.id) }
+    end
+  end
+
+  def add_to_reverse_index(url, post_id)
+    # make sure we have a url to insert
+    return unless url.present?
+    # local uploads are relative
+    if index = url.index(local_base_url)
+      url = url[index..-1]
+    end
+    # filter out non-uploads
+    return unless url.starts_with?(local_base_url) || url.starts_with?(s3_base_url)
+    # update the reverse index
+    execute "INSERT INTO post_uploads (upload_id, post_id)
+             SELECT u.id, #{post_id}
+             FROM uploads u
+             WHERE u.url = '#{url}'
+             AND NOT EXISTS (SELECT 1 FROM post_uploads WHERE upload_id = u.id AND post_id = #{post_id})"
+  end
+
+  def local_base_url
+    @local_base_url ||= "/uploads/#{RailsMultisite::ConnectionManagement.current_db}"
+  end
+
+  def s3_base_url
+    @s3_base_url ||= "//#{SiteSetting.s3_upload_bucket.downcase}.s3.amazonaws.com"
+  end
+
+end

lib/cooked_post_processor.rb

@@ -16,11 +16,16 @@ class CookedPostProcessor
   end
 
   def post_process
+    clean_up_reverse_index
     post_process_attachments
     post_process_images
     post_process_oneboxes
   end
 
+  def clean_up_reverse_index
+    PostUpload.delete_all(post_id: @post.id)
+  end
+
   def post_process_attachments
     attachments.each do |attachment|
       href = attachment['href']

lib/tasks/images.rake

@@ -10,25 +10,6 @@ task "images:compress" => :environment do
   end
 end
 
-desc "clean orphan uploaded files"
-task "images:clean_orphans" => :environment do
-  RailsMultisite::ConnectionManagement.each_connection do |db|
-    puts "Cleaning up #{db}"
-    # ligthweight safety net to prevent users from wiping all their uploads out
-    if PostUpload.count == 0 && Upload.count > 0
-      puts "The reverse index is empty. Make sure you run the `images:reindex` task"
-      next
-    end
-    Upload.joins("LEFT OUTER JOIN post_uploads ON uploads.id = post_uploads.upload_id")
-          .where("post_uploads.upload_id IS NULL")
-          .find_each do |u|
-      u.destroy
-      putc "."
-    end
-  end
-  puts "\ndone."
-end
-
 desc "download all hotlinked images"
 task "images:pull_hotlinked" => :environment do
   RailsMultisite::ConnectionManagement.each_connection do |db|

spec/components/cooked_post_processor_spec.rb

@@ -10,6 +10,7 @@ describe CookedPostProcessor do
     let(:post_process) { sequence("post_process") }
 
     it "post process in sequence" do
+      cpp.expects(:clean_up_reverse_index).in_sequence(post_process)
       cpp.expects(:post_process_attachments).in_sequence(post_process)
       cpp.expects(:post_process_images).in_sequence(post_process)
       cpp.expects(:post_process_oneboxes).in_sequence(post_process)
@@ -18,6 +19,18 @@ describe CookedPostProcessor do
 
   end
 
+  context "clean_up_reverse_index" do
+
+    let(:post) { build(:post) }
+    let(:cpp) { CookedPostProcessor.new(post) }
+
+    it "cleans the reverse index up for the current post" do
+      PostUpload.expects(:delete_all).with(post_id: post.id)
+      cpp.clean_up_reverse_index
+    end
+
+  end
+
   context "post_process_attachments" do
 
     context "with attachment" do