nokogiri: switch to the nokogumbo-based html5 parser.

https://github.com/sparklemotion/nokogiri/blob/main/CHANGELOG.md#1120--2021-08-02
danbooru · Aug 31, 2021 · 38c9559 · 38c9559
1 parent 49d18e6
commit 38c9559
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 7 deletions.
diff --git a/app/logical/d_text.rb b/app/logical/d_text.rb
@@ -49,7 +49,7 @@ def self.preprocess(dtext_messages)
   # @param artists [Array<Artist>]
   # @return [String] the HTML output
   def self.postprocess(html, wiki_pages, tags, artists)
-    fragment = Nokogiri::HTML.fragment(html)
+    fragment = Nokogiri::HTML5.fragment(html)
 
     fragment.css("a.dtext-wiki-link").each do |node|
       path = Addressable::URI.parse(node["href"]).path
@@ -174,7 +174,7 @@ def self.parse_mentions(text)
   # @return [Array<String>] the list of wiki page names
   def self.parse_wiki_titles(text)
     html = DTextRagel.parse(text)
-    fragment = Nokogiri::HTML.fragment(html)
+    fragment = Nokogiri::HTML5.fragment(html)
 
     titles = fragment.css("a.dtext-wiki-link").map do |node|
       title = node["href"][%r{\A/wiki_pages/(.*)\z}i, 1]
@@ -191,7 +191,7 @@ def self.parse_wiki_titles(text)
   # @return [Array<String>] the list of external URLs
   def self.parse_external_links(text)
     html = DTextRagel.parse(text)
-    fragment = Nokogiri::HTML.fragment(html)
+    fragment = Nokogiri::HTML5.fragment(html)
 
     links = fragment.css("a.dtext-external-link").map { |node| node["href"] }
     links.uniq
@@ -326,7 +326,7 @@ def self.to_markdown(dtext)
   # @param html [String] the HTML input
   # @return [String] the Markdown output
   def self.html_to_markdown(html)
-    html = Nokogiri::HTML.fragment(html)
+    html = Nokogiri::HTML5.fragment(html)
 
     html.children.map do |node|
       case node.name
@@ -349,7 +349,7 @@ def self.html_to_markdown(html)
   # @param inline [Boolean] if true, convert <img> tags to plaintext
   # @return [String] the DText output
   def self.from_html(text, inline: false, &block)
-    html = Nokogiri::HTML.fragment(text)
+    html = Nokogiri::HTML5.fragment(text)
 
     dtext = html.children.map do |element|
       block.call(element) if block.present?

diff --git a/app/logical/sources/strategies/deviant_art.rb b/app/logical/sources/strategies/deviant_art.rb
@@ -269,7 +269,7 @@ def page
         resp = http.cache(1.minute).get(page_url_from_image_url, follow: {max_hops: 1})
 
         if resp.status.success?
-          Nokogiri::HTML(resp.body.to_s)
+          resp.parse
         # the work was deleted
         elsif resp.code == 404
           nil

diff --git a/app/logical/sources/strategies/tumblr.rb b/app/logical/sources/strategies/tumblr.rb
@@ -183,7 +183,7 @@ def find_largest(url, sizes: SIZES)
     end
 
     def inline_images
-      html = Nokogiri::HTML.fragment(artist_commentary_desc)
+      html = Nokogiri::HTML5.fragment(artist_commentary_desc)
       html.css("img").map { |node| node["src"] }
     end