add RTL and LTR markers if i18n text has different directionality fro…

…m document text: #166
metanorma · Dec 13, 2021 · 42dcc6a · 42dcc6a
1 parent 8ab4450
commit 42dcc6a
Show file tree

Hide file tree

Showing 2 changed files with 68 additions and 9 deletions.
diff --git a/lib/isodoc/i18n.rb b/lib/isodoc/i18n.rb
@@ -55,7 +55,7 @@ def initialize(lang, script, i18nyaml = nil)
       @labels = y
       @labels["language"] = @lang
       @labels["script"] = @script
-      @labels.each do |k, v|
+      @labels.each do |k, _v|
         self.class.send(:define_method, k.downcase) { get[k] }
       end
     end
@@ -68,19 +68,40 @@ def self.l10n(text, lang = @lang, script = @script)
     # function localising spaces and punctuation.
     # Not clear if period needs to be localised for zh
     def l10n(text, lang = @lang, script = @script)
-      if lang == "zh" && script == "Hans"
-        xml = Nokogiri::HTML::DocumentFragment.parse(text)
-        xml.traverse do |n|
-          next unless n.text?
+      if lang == "zh" && script == "Hans" then l10n_zh(text)
+      else bidiwrap(text, lang, script)
+      end
+    end
 
-          n.replace(n.text.gsub(/ /, "").gsub(/:/, "：").gsub(/,/, "、")
-            .gsub(/\(/, "（").gsub(/\)/, "）").gsub(/\[/, "【").gsub(/\]/, "】"))
-        end
-        xml.to_xml.gsub(/<b>/, "").gsub("</b>", "").gsub(/<\?[^>]+>/, "")
+    def bidiwrap(text, lang, script)
+      my_script, my_rtl, outer_rtl = bidiwrap_vars(lang, script)
+      if my_rtl && !outer_rtl
+        mark = %w(Arab Aran).include?(my_script) ? "&#x61c;" : "&#x200f;"
+        "#{mark}#{text}#{mark}"
+      elsif !my_rtl && outer_rtl then "&#x200e;#{text}&#x200e;"
       else text
       end
     end
 
+    def bidiwrap_vars(lang, script)
+      my_script = script || Metanorma::Utils.default_script(lang)
+      [my_script,
+       Metanorma::Utils.rtl_script?(my_script),
+       Metanorma::Utils.rtl_script?(@script || Metanorma::Utils
+         .default_script(@lang))]
+    end
+
+    def l10n_zh(text)
+      xml = Nokogiri::HTML::DocumentFragment.parse(text)
+      xml.traverse do |n|
+        next unless n.text?
+
+        n.replace(n.text.gsub(/ /, "").gsub(/:/, "：").gsub(/,/, "、")
+          .gsub(/\(/, "（").gsub(/\)/, "）").gsub(/\[/, "【").gsub(/\]/, "】"))
+      end
+      xml.to_xml.gsub(/<b>/, "").gsub("</b>", "").gsub(/<\?[^>]+>/, "")
+    end
+
     def multiple_and(names, andword)
       return "" if names.empty?
       return names[0] if names.length == 1

diff --git a/spec/isodoc/i18n_spec.rb b/spec/isodoc/i18n_spec.rb
@@ -1082,6 +1082,44 @@
       .to be_equivalent_to xmlpp(output)
   end
 
+  it "processes LTR within RTL" do
+    c = IsoDoc::Convert.new({})
+    c.convert_init(<<~"INPUT", "test", false)
+      <iso-standard xmlns="http://riboseinc.com/isoxml">
+        <bibdata type="standard">
+      <language>fa</language>
+      <script>Arab</script>
+      </bibdata>
+      </iso-standard>
+    INPUT
+    expect(c.i18n.l10n("hello!", "en", "Latn")).to eq "&#x200e;hello!&#x200e;"
+  end
+
+  it "processes Hebrew RTL within LTR" do
+    c = IsoDoc::Convert.new({})
+    c.convert_init(<<~"INPUT", "test", false)
+      <iso-standard xmlns="http://riboseinc.com/isoxml">
+        <bibdata type="standard">
+      <language>en</language>
+      </bibdata>
+      </iso-standard>
+    INPUT
+    expect(c.i18n.l10n("hello!", "he", "Hebr")).to eq "&#x200f;hello!&#x200f;"
+  end
+
+  it "processes Arabic RTL within LTR" do
+    c = IsoDoc::Convert.new({})
+    c.convert_init(<<~"INPUT", "test", false)
+      <iso-standard xmlns="http://riboseinc.com/isoxml">
+        <bibdata type="standard">
+      <language>en</language>
+      </bibdata>
+      </iso-standard>
+    INPUT
+    expect(c.i18n.l10n("hello!", "fa", "Arab")).to eq "&#x61c;hello!&#x61c;"
+  end
+
+
   private
 
   def mock_i18n