BUGFIX: errors when post-processing 'data images'

app/assets/javascripts/discourse/lib/markdown.js

@@ -238,6 +238,8 @@ Discourse.Markdown.whiteListTag('a', 'class', 'mention');
 
 Discourse.Markdown.whiteListTag('a', 'data-bbcode');
 
+Discourse.Markdown.whiteListTag('img', 'src', /^data:image.*/i);
+
 Discourse.Markdown.whiteListTag('div', 'class', 'title');
 Discourse.Markdown.whiteListTag('div', 'class', 'quote-controls');
 

app/jobs/regular/pull_hotlinked_images.rb

@@ -90,7 +90,7 @@ module Jobs
 
     def extract_images_from(html)
       doc = Nokogiri::HTML::fragment(html)
-      doc.css("img") - doc.css(".onebox-result img") - doc.css("img.avatar")
+      doc.css("img[src]") - doc.css(".onebox-result img") - doc.css("img.avatar")
     end
 
     def is_valid_image_url(src)

app/models/upload.rb

@@ -112,6 +112,7 @@ class Upload < ActiveRecord::Base
   end
 
   def self.get_from_url(url)
+    return if url.blank?
     # we store relative urls, so we need to remove any host/cdn
     url = url.gsub(/^#{Discourse.asset_host}/i, "") if Discourse.asset_host.present?
     Upload.find_by(url: url) if Discourse.store.has_been_uploaded?(url)

lib/cooked_post_processor.rb

@@ -27,14 +27,14 @@ class CookedPostProcessor
   def keep_reverse_index_up_to_date
     upload_ids = Set.new
 
-    @doc.search("a").each do |a|
+    @doc.css("a[href]").each do |a|
       href = a["href"].to_s
       if upload = Upload.get_from_url(href)
         upload_ids << upload.id
       end
     end
 
-    @doc.search("img").each do |img|
+    @doc.css("img[src]").each do |img|
       src = img["src"].to_s
       if upload = Upload.get_from_url(src)
         upload_ids << upload.id
@@ -64,8 +64,14 @@ class CookedPostProcessor
   end
 
   def extract_images
-    # do not extract images inside oneboxes or quotes
-    @doc.css("img") - @doc.css(".onebox-result img, .onebox img") - @doc.css(".quote img")
+    # all image with a src attribute
+    @doc.css("img[src]") -
+    # minus, data images
+    @doc.css("img[src^='data']") -
+    # minus, image inside oneboxes
+    @doc.css(".onebox-result img, .onebox img") -
+    # minux, images inside quotes
+    @doc.css(".quote img")
   end
 
   def limit_size!(img)
@@ -207,12 +213,12 @@ class CookedPostProcessor
   end
 
   def optimize_urls
-    @doc.search("a").each do |a|
+    @doc.css("a[href]").each do |a|
       href = a["href"].to_s
       a["href"] = schemaless absolute(href) if is_local(href)
     end
 
-    @doc.search("img").each do |img|
+    @doc.css("img[src]").each do |img|
       src = img["src"].to_s
       img["src"] = schemaless absolute(src) if is_local(src)
     end

lib/file_store/local_store.rb

@@ -28,7 +28,7 @@ module FileStore
     end
 
     def has_been_uploaded?(url)
-      is_relative?(url) || is_local?(url)
+      url.present? && (is_relative?(url) || is_local?(url))
     end
 
     def absolute_base_url
@@ -120,10 +120,11 @@ module FileStore
     end
 
     def is_relative?(url)
-      url.start_with?(relative_base_url)
+      url.present? && url.start_with?(relative_base_url)
     end
 
     def is_local?(url)
+      return false if url.blank?
       absolute_url = url.start_with?("//") ? SiteSetting.scheme + ":" + url : url
       absolute_url.start_with?(absolute_base_url) || absolute_url.start_with?(absolute_base_cdn_url)
     end

lib/file_store/s3_store.rb

@@ -30,7 +30,7 @@ module FileStore
     end
 
     def has_been_uploaded?(url)
-      url.start_with?(absolute_base_url)
+      url.present? && url.start_with?(absolute_base_url)
     end
 
     def absolute_base_url

lib/url_helper.rb

@@ -1,10 +1,12 @@
 module UrlHelper
 
   def is_local(url)
-    Discourse.store.has_been_uploaded?(url) ||
-    url =~ /^\/assets\// ||
-    url =~ /^\/plugins\// ||
-    url.start_with?(Discourse.asset_host || Discourse.base_url_no_prefix)
+    url.present? && (
+      Discourse.store.has_been_uploaded?(url) ||
+      url =~ /^\/assets\// ||
+      url =~ /^\/plugins\// ||
+      url.start_with?(Discourse.asset_host || Discourse.base_url_no_prefix)
+    )
   end
 
   def absolute(url, cdn = Discourse.asset_host)