From 15de4ac8904fd8de9499a1884f1c84369ab2a003 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?R=C3=A9gis=20Hanol?= <regis@hanol.fr>
Date: Thu, 10 Oct 2013 04:01:01 +0200
Subject: [PATCH] add a rake task to pull hotlinked images

---
 lib/tasks/images.rake | 98 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 98 insertions(+)

diff --git a/lib/tasks/images.rake b/lib/tasks/images.rake
index 74c85642110..a6d4e60e222 100644
--- a/lib/tasks/images.rake
+++ b/lib/tasks/images.rake
@@ -28,3 +28,101 @@ task "images:clean_orphans" => :environment do
   end
   puts "\ndone."
 end
+
+desc "download all hotlinked images"
+task "images:pull_hotlinked" => :environment do
+  RailsMultisite::ConnectionManagement.each_connection do |db|
+    # currently only works when using the local storage
+    next if Discourse.store.external?
+
+    puts "Pulling hotlinked images for: #{db}"
+
+    # shorthand to the asset host
+    asset_host = Rails.configuration.action_controller.asset_host
+    # maximum size of the file in bytes
+    max_size = SiteSetting.max_image_size_kb * 1024
+    # will hold the urls of the already downloaded images
+    upload_urls = {}
+
+    Post.find_each do |post|
+      has_changed = false
+
+      extract_images_from(post.cooked).each do |image|
+        src = image['src']
+        if src.present? &&
+           src !~ /^\/[^\/]/ &&
+           !src.starts_with?(Discourse.base_url_no_prefix) &&
+           !(asset_host.present? && src.starts_with?(asset_host))
+          begin
+            # have we already downloaded that file?
+            if !upload_urls.include?(src)
+              # initialize
+              upload_urls[src] = nil
+              # download the file
+              hotlinked = download(src, max_size)
+              # if the hotlinked image is OK
+              if hotlinked.size <= max_size
+                file = ActionDispatch::Http::UploadedFile.new(tempfile: hotlinked, filename: File.basename(URI.parse(src).path))
+                upload_urls[src] = Upload.create_for(post.user_id, file, hotlinked.size).url
+              else
+                puts "\nFailed to pull: #{src} for post ##{post.id} - too large\n"
+              end
+            end
+            # if we have downloaded a file
+            if upload_urls[src].present?
+              src_for_regexp = src.gsub("?", "\\?").gsub(".", "\\.").gsub("+", "\\+")
+              # there are 4 ways to insert an image in a post
+              # HTML tag - <img src="http://...">
+              post.raw.gsub!(/src=["']#{src_for_regexp}["']/i, "src='#{upload_urls[src]}'")
+              # BBCode tag - [img]http://...[/img]
+              post.raw.gsub!(/\[img\]#{src_for_regexp}\[\/img\]/i, "[img]#{upload_urls[src]}[/img]")
+              # Markdown - ![alt](http://...)
+              post.raw.gsub!(/!\[([^\]]*)\]\(#{src_for_regexp}\)/) { "![#{$1}](#{upload_urls[src]})" }
+              # Direct link
+              post.raw.gsub!(src, "<img src='#{upload_urls[src]}'>")
+              # mark the post as changed
+              has_changed = true
+            end
+          rescue => e
+            puts "\nFailed to pull: #{src} for post ##{post.id} - #{e}\n"
+          ensure
+            # close & delete the temporary file
+            hotlinked && hotlinked.close!
+          end
+        end
+      end
+
+      if has_changed
+        # since the raw has changed, we cook the post once again
+        post.cooked = post.cook(post.raw, topic_id: post.topic_id, invalidate_oneboxes: true)
+        # update both raw & cooked version of the post
+        Post.exec_sql('update posts set cooked = ?, raw = ? where id = ?', post.cooked, post.raw, post.id)
+        # trigger the post processing
+        post.trigger_post_process
+        putc "#"
+      else
+        putc "."
+      end
+    end
+  end
+  puts "\ndone."
+end
+
+def extract_images_from(html)
+  doc = Nokogiri::HTML::fragment(html)
+  doc.css("img") - doc.css(".onebox-result img") - doc.css("img.avatar")
+end
+
+def download(url, max_size)
+  # create a temporary file
+  temp_file = Tempfile.new(["discourse-hotlinked", File.extname(URI.parse(url).path)])
+  # download the hotlinked image
+  File.open(temp_file.path, "wb") do |f|
+    hotlinked = open(url, "rb", read_timeout: 5)
+    while f.size <= max_size && data = hotlinked.read(max_size)
+      f.write(data)
+    end
+    hotlinked.close
+  end
+  temp_file
+end