Add a new Liquid filter rebase_html

huginn · Oct 18, 2016 · 2890c9e · 2890c9e
1 parent 3099405
commit 2890c9e
Show file tree

Hide file tree

Showing 3 changed files with 132 additions and 0 deletions.
diff --git a/app/concerns/liquid_interpolatable.rb b/app/concerns/liquid_interpolatable.rb
@@ -189,6 +189,11 @@ def uri_expand(url, limit = 5)
       url
     end
 
+    # Rebase URIs contained in attributes in a given HTML fragment
+    def rebase_html(input, base_uri)
+      Utils.rebase_html(input, base_uri) rescue input
+    end
+
     # Unescape (basic) HTML entities in a string
     #
     # This currently decodes the following entities only: "&apos;",

diff --git a/lib/utils.rb b/lib/utils.rb
@@ -170,4 +170,93 @@ def self.if_present(string, method)
       nil
     end
   end
+
+  module HTMLTransformer
+    SINGLE = 1
+    MULTIPLE = 2
+    COMMA_SEPARATED = 3
+    SRCSET = 4
+
+    URI_ATTRIBUTES = {
+      'a' => { 'href' => SINGLE },
+      'applet' => { 'archive' => COMMA_SEPARATED, 'codebase' => SINGLE },
+      'area' => { 'href' => SINGLE },
+      'audio' => { 'src' => SINGLE },
+      'base' => { 'href' => SINGLE },
+      'blockquote' => { 'cite' => SINGLE },
+      'body' => { 'background' => SINGLE },
+      'button' => { 'formaction' => SINGLE },
+      'command' => { 'icon' => SINGLE },
+      'del' => { 'cite' => SINGLE },
+      'embed' => { 'src' => SINGLE },
+      'form' => { 'action' => SINGLE },
+      'frame' => { 'longdesc' => SINGLE, 'src' => SINGLE },
+      'head' => { 'profile' => SINGLE },
+      'html' => { 'manifest' => SINGLE },
+      'iframe' => { 'longdesc' => SINGLE, 'src' => SINGLE },
+      'img' => { 'longdesc' => SINGLE, 'src' => SINGLE, 'srcset' => SRCSET, 'usemap' => SINGLE },
+      'input' => { 'formaction' => SINGLE, 'src' => SINGLE, 'usemap' => SINGLE },
+      'ins' => { 'cite' => SINGLE },
+      'link' => { 'href' => SINGLE },
+      'object' => { 'archive' => MULTIPLE, 'classid' => SINGLE, 'codebase' => SINGLE, 'data' => SINGLE, 'usemap' => SINGLE },
+      'q' => { 'cite' => SINGLE },
+      'script' => { 'src' => SINGLE },
+      'source' => { 'src' => SINGLE, 'srcset' => SRCSET },
+      'video' => { 'poster' => SINGLE, 'src' => SINGLE },
+    }
+
+    URI_ELEMENTS_XPATH = '//*[%s]' % URI_ATTRIBUTES.keys.map { |name| "name()='#{name}'" }.join(' or ')
+
+    module_function
+
+    def transform(html, &block)
+      block or raise ArgumentError, 'block must be given'
+
+      case html
+      when /\A\s*(?:<\?xml[\s?]|<!DOCTYPE\s)/i
+        doc = Nokogiri.parse(html)
+        yield doc
+        doc.to_s
+      when /\A\s*<(html|head|body)[\s>]/i
+        # Libxml2 automatically adds DOCTYPE and <html>, so we need to
+        # skip them.
+        element_name = $1
+        doc = Nokogiri::HTML::Document.parse(html)
+        yield doc
+        doc.at_xpath("//#{element_name}").xpath('self::node() | following-sibling::node()').to_s
+      else
+        doc = Nokogiri::HTML::Document.parse("<html><body>#{html}")
+        yield doc
+        doc.xpath("/html/body/node()").to_s
+      end
+    end
+
+    def replace_uris(html, &block)
+      block or raise ArgumentError, 'block must be given'
+
+      transform(html) { |doc|
+        doc.xpath(URI_ELEMENTS_XPATH).each { |element|
+          uri_attrs = URI_ATTRIBUTES[element.name] or next
+          uri_attrs.each { |name, format|
+            attr = element.attribute(name) or next
+            case format
+            when SINGLE
+              attr.value = block.call(attr.value.strip)
+            when MULTIPLE
+              attr.value = attr.value.gsub(/(\S+)/) { block.call($1) }
+            when COMMA_SEPARATED, SRCSET
+              attr.value = attr.value.gsub(/((?:\A|,)\s*)(\S+)/) { $1 + block.call($2) }
+            end
+          }
+        }
+      }
+    end
+  end
+
+  def self.rebase_html(html, base_uri)
+    base_uri = normalize_uri(base_uri)
+    HTMLTransformer.replace_uris(html) { |url|
+      base_uri.merge(normalize_uri(url)).to_s
+    }
+  end
 end
diff --git a/spec/concerns/liquid_interpolatable_spec.rb b/spec/concerns/liquid_interpolatable_spec.rb
@@ -323,4 +323,42 @@ def ensure_safety(obj)
       end
     end
   end
+
+  describe 'rebase_html' do
+    let(:agent) { Agents::InterpolatableAgent.new(name: "test") }
+
+    let(:fragment) { <<HTML }
+<ul>
+  <li>
+    <a href="downloads/file1"><img src="/images/iconA.png" srcset="/images/iconA.png 1x, /images/iconA@2x.png 2x">file1</a>
+  </li>
+  <li>
+    <a href="downloads/file2"><img src="/images/iconA.png" srcset="/images/iconA.png 1x, /images/iconA@2x.png 2x">file2</a>
+  </li>
+  <li>
+    <a href="downloads/file3"><img src="/images/iconB.png" srcset="/images/iconB.png 1x, /images/iconB@2x.png 2x">file3</a>
+  </li>
+</ul>
+HTML
+
+    let(:replaced_fragment) { <<HTML }
+<ul>
+  <li>
+    <a href="http://example.com/support/downloads/file1"><img src="http://example.com/images/iconA.png" srcset="http://example.com/images/iconA.png 1x, http://example.com/images/iconA@2x.png 2x">file1</a>
+  </li>
+  <li>
+    <a href="http://example.com/support/downloads/file2"><img src="http://example.com/images/iconA.png" srcset="http://example.com/images/iconA.png 1x, http://example.com/images/iconA@2x.png 2x">file2</a>
+  </li>
+  <li>
+    <a href="http://example.com/support/downloads/file3"><img src="http://example.com/images/iconB.png" srcset="http://example.com/images/iconB.png 1x, http://example.com/images/iconB@2x.png 2x">file3</a>
+  </li>
+</ul>
+HTML
+
+    it 'rebases relative URLs in a fragment' do
+      agent.interpolation_context['content'] = fragment
+      agent.options['template'] = "{{ content | rebase_html: 'http://example.com/support/files.html' }}"
+      expect(agent.interpolated['template']).to eq(replaced_fragment)
+    end
+  end
 end