Merge 9eb30dfb1c into 1959365c2f

lib/sanitize_ext/sanitize_config.rb

@@ -64,6 +64,43 @@ class Sanitize
       current_node.wrap('<p></p>')
     end
 
+    # We assume that incomming <math> nodes are of the form
+    # <math><semantics>...<annotation>...</annotation></semantics></math>
+    # according to the [FEP]. We try to grab the most relevant plain-text
+    # annotation from the semantics node, and use it to display a representation
+    # of the mathematics.
+    #
+    # FEP: https://codeberg.org/fediverse/fep/src/branch/main/fep/dc88/fep-dc88.md
+    MATH_TRANSFORMER = lambda do |env|
+      math = env[:node]
+      return if env[:is_allowlisted]
+      return unless math.element? && env[:node_name] == 'math'
+
+      semantics = math.element_children[0]
+      return if semantics.nil? || semantics.name != 'semantics'
+
+      # next, we find the plain-text description
+      is_annotation_with_encoding = lambda do |encoding, node|
+        return false unless node.name == 'annotation'
+
+        node.attributes['encoding'].value == encoding
+      end
+
+      annotation = semantics.children.find(&is_annotation_with_encoding.curry['application/x-tex'])
+      if annotation
+        if math.attributes['display']&.value == 'block'
+          math.swap("$$#{annotation.content}$$")
+        else
+          math.swap("$#{annotation.content}$")
+        end
+        return
+      end
+      # Don't bother surrounding 'text/plain' annotations with dollar signs,
+      # since it isn't LaTeX
+      annotation = semantics.children.find(&is_annotation_with_encoding.curry['text/plain'])
+      math.swap(annotation.content) unless annotation.nil?
+    end
+
     MASTODON_STRICT = freeze_config(
       elements: %w(p br span a del pre blockquote code b strong u i em ul ol li),
 
@@ -86,6 +123,7 @@ class Sanitize
       transformers: [
         CLASS_WHITELIST_TRANSFORMER,
         TRANSLATE_TRANSFORMER,
+        MATH_TRANSFORMER,
         UNSUPPORTED_ELEMENTS_TRANSFORMER,
         UNSUPPORTED_HREF_TRANSFORMER,
       ]

spec/lib/sanitize/config_spec.rb

@@ -53,5 +53,25 @@ describe Sanitize::Config do
     it 'keeps a with supported scheme and no host' do
       expect(Sanitize.fragment('<a href="dweb:/a/foo">Test</a>', subject)).to eq '<a href="dweb:/a/foo" rel="nofollow noopener noreferrer" target="_blank">Test</a>'
     end
+
+    it 'sanitizes math to LaTeX' do
+      mathml = '<math><semantics><mrow><msup><mi>x</mi><mi>n</mi></msup><mo>+</mo><mi>y</mi></mrow><annotation encoding="application/x-tex">x^n+y</annotation></semantics></math>'
+      expect(Sanitize.fragment(mathml, subject)).to eq '$x^n+y$'
+    end
+
+    it 'sanitizes math blocks to LaTeX' do
+      mathml = '<math display="block"><semantics><mrow><msup><mi>x</mi><mi>n</mi></msup><mo>+</mo><mi>y</mi></mrow><annotation encoding="application/x-tex">x^n+y</annotation></semantics></math>'
+      expect(Sanitize.fragment(mathml, subject)).to eq '$$x^n+y$$'
+    end
+
+    it 'math sanitizer falls back to plaintext' do
+      mathml = '<math><semantics><msqrt><mi>x</mi></msqrt><annotation encoding="text/plain">sqrt(x)</annotation></semantics></math>'
+      expect(Sanitize.fragment(mathml, subject)).to eq 'sqrt(x)'
+    end
+
+    it 'prefers latex' do
+      mathml = '<math><semantics><msqrt><mi>x</mi></msqrt><annotation encoding="text/plain">sqrt(x)</annotation><annotation encoding="application/x-tex">\\sqrt x</annotation></semantics></math>'
+      expect(Sanitize.fragment(mathml, subject)).to eq '$\sqrt x$'
+    end
   end
 end