Merge pull request #129 from metanorma/fix/nokogiri-1.13

Nokogiri 1.13-safe processing: https://github.com/metanorma/isodoc/is…
metanorma · May 16, 2022 · 490aae8 · 490aae8
2 parents 51c6bac + bc00d0e
commit 490aae8
Show file tree

Hide file tree

Showing 4 changed files with 10 additions and 3 deletions.
diff --git a/lib/isodoc/ribose/word_convert.rb b/lib/isodoc/ribose/word_convert.rb
@@ -22,7 +22,7 @@ def make_body2(body, docxml)
           introduction docxml, div2
           preface docxml, div2
           acknowledgements docxml, div2
-          div2.p { |p| p << "&nbsp;" } # placeholder
+          div2.p { |p| p << "&#xa0;" } # placeholder
         end
         section_break(body)
       end

diff --git a/spec/isodoc/html_convert_spec.rb b/spec/isodoc/html_convert_spec.rb
@@ -746,7 +746,7 @@
     output = xmlpp(<<~"OUTPUT")
       <div id='H'>
         <h1 id='toc0'>1.&#xA0; Terms, Definitions, Symbols and Abbreviated Terms</h1>
-        <p class='Terms' style='text-align:left;' id='J'><strong>1.1.</strong>&#xA0;Term2</p>
+        <p class='Terms' style='text-align:left;' id='J'><strong>1.1.</strong>Term2</p>
         <p class='AltTerms' style='text-align:left;'>Term2A</p>
         <p class='AltTerms' style='text-align:left;'>Term2B</p>
         <p class='DeprecatedTerms' style='text-align:left;'>DEPRECATED: Term2C</p>

diff --git a/spec/metanorma/processor_spec.rb b/spec/metanorma/processor_spec.rb
@@ -58,7 +58,7 @@
         <p class="zzSTDTitle1"></p>
         <div id="H">
           <h1 id="toc0">1&#xA0; Terms, Definitions, Symbols and Abbreviated Terms</h1>
-          <p class='Terms' style='text-align:left;' id='J'><strong>1.1</strong>&#xA0;Term2</p>
+          <p class='Terms' style='text-align:left;' id='J'><strong>1.1</strong>Term2</p>
         </div>
       </main>
     OUTPUT

diff --git a/spec/spec_helper.rb b/spec/spec_helper.rb
@@ -52,6 +52,13 @@ def htmlencode(html)
 end
 
 def xmlpp(xml)
+  c = HTMLEntities.new
+  xml &&= xml.split(/(&\S+?;)/).map do |n|
+    if /^&\S+?;$/.match?(n)
+      c.encode(c.decode(n), :hexadecimal)
+    else n
+    end
+  end.join
   s = ""
   f = REXML::Formatters::Pretty.new(2)
   f.compact = true