backport fix for #2263 apply text-tranform from custom role on phrase after attributes have been resolved

7 files changed, 89 insertions, 33 deletions

diff --git a/CHANGELOG.adoc b/CHANGELOG.adoc
index f82031fc..15337e28 100644
--- a/CHANGELOG.adoc
+++ b/CHANGELOG.adoc
@@ -14,6 +14,7 @@ Improvements::
 Bug Fixes::
 
 * allow alt text for block image, video, and audio to wrap to next line on same page (#2258)
+* apply text-tranform from custom role on phrase after attributes have been resolved (#2263)
 * make URL check more strict so image target containing a colon is not mistaken as a URL
 
 == 2.1.3 (2022-06-23) - @mojavelinux
diff --git a/lib/asciidoctor/pdf/converter.rb b/lib/asciidoctor/pdf/converter.rb
index ac0ae759..77934032 100644
--- a/lib/asciidoctor/pdf/converter.rb
+++ b/lib/asciidoctor/pdf/converter.rb
@@ -2688,13 +2688,7 @@ module Asciidoctor
         end
 
         if (roles = node.role)
-          theme = load_theme node.document
-          roles.split.each do |role|
-            if (text_transform = theme[%(role_#{role}_text_transform)])
-              inner_text = transform_text inner_text, text_transform
-            end
-            inner_text = inner_text.gsub DoubleSpaceRx, ' ' + ZeroWidthSpace if role == 'pre-wrap' && (inner_text.include? DoubleSpace)
-          end
+          inner_text = inner_text.gsub DoubleSpaceRx, ' ' + ZeroWidthSpace if (node.has_role? 'pre-wrap') && (inner_text.include? DoubleSpace)
           quoted_text = is_tag ? %(#{open.chop} class="#{roles}">#{inner_text}#{close}) : %(<span class="#{roles}">#{open}#{inner_text}#{close}</span>)
         else
           quoted_text = %(#{open}#{inner_text}#{close})
diff --git a/lib/asciidoctor/pdf/ext/prawn/extensions.rb b/lib/asciidoctor/pdf/ext/prawn/extensions.rb
index ae024e9f..74c6d11f 100644
--- a/lib/asciidoctor/pdf/ext/prawn/extensions.rb
+++ b/lib/asciidoctor/pdf/ext/prawn/extensions.rb
@@ -528,29 +528,6 @@ module Asciidoctor
         end
       end
 
-      # Apply the text transform to the specified text.
-      #
-      # Supported transform values are "uppercase", "lowercase", or "none" (passed
-      # as either a String or a Symbol). When the uppercase transform is applied to
-      # the text, it correctly uppercases visible text while leaving markup and
-      # named character entities unchanged. The none transform returns the text
-      # unmodified.
-      #
-      def transform_text text, transform
-        case transform
-        when :uppercase, 'uppercase'
-          uppercase_pcdata text
-        when :lowercase, 'lowercase'
-          lowercase_pcdata text
-        when :capitalize, 'capitalize'
-          capitalize_words_pcdata text
-        when :smallcaps, 'smallcaps'
-          smallcaps_pcdata text
-        else
-          text
-        end
-      end
-
       def hyphenate_text text, hyphenator
         hyphenate_words_pcdata text, hyphenator
       end
diff --git a/lib/asciidoctor/pdf/formatted_text/transform.rb b/lib/asciidoctor/pdf/formatted_text/transform.rb
index d5fd582b..30f47761 100644
--- a/lib/asciidoctor/pdf/formatted_text/transform.rb
+++ b/lib/asciidoctor/pdf/formatted_text/transform.rb
@@ -4,6 +4,8 @@ module Asciidoctor
   module PDF
     module FormattedText
       class Transform
+        include TextTransformer
+
         LF = ?\n
         ZeroWidthSpace = ?\u200b
         CharEntityTable = { amp: '&', apos: ?', gt: '>', lt: '<', nbsp: ?\u00a0, quot: '"' }
@@ -21,8 +23,8 @@ module Asciidoctor
           'font_size' => :size,
           'text_decoration_color' => :text_decoration_color,
           'text_decoration_width' => :text_decoration_width,
+          'text_transform' => :text_transform,
         }
-        #DummyText = ?\u0000
 
         def initialize options = {}
           @merge_adjacent_text_nodes = options[:merge_adjacent_text_nodes]
@@ -172,6 +174,11 @@ module Asciidoctor
                       previous_fragment_is_text && ((previous_fragment_text = fragments[-1][:text]).end_with? ' ')
                     fragments[-1][:text] = previous_fragment_text.chop
                   end
+                  if (text_transform = fragment.delete :text_transform)
+                    text = (text_chunks = extract_text pcdata).join
+                    text_io = StringIO.new transform_text text, text_transform
+                    restore_text pcdata, text_chunks.each_with_object([]) {|chunk, accum| accum << (text_io.read chunk.length) }
+                  end
                   # NOTE: decorate child fragments with inherited properties from this element
                   apply pcdata, fragments, fragment
                   previous_fragment_is_text = false
@@ -251,6 +258,8 @@ module Asciidoctor
           fragments
         end
 
+        private
+
         def build_fragment fragment, tag_name, attrs
           styles = (fragment[:styles] ||= ::Set.new)
           case tag_name
@@ -324,7 +333,6 @@ module Asciidoctor
             # NOTE: spaces in style value are superfluous for our purpose; split drops record after trailing ;
             attrs[:style].tr(' ', '').split(';').each do |style|
               pname, pvalue = style.split ':', 2
-              # TODO: text-transform
               case pname
               when 'color' # color needed to support syntax highlighters
                 fragment[:color] = pvalue.length == 7 ? (pvalue.slice 1, 6) : (pvalue.slice 1, 3).each_char.map {|c| c * 2 }.join if (pvalue.start_with? '#') && (HexColorRx.match? pvalue)
@@ -409,6 +417,29 @@ module Asciidoctor
             end
           end
         end
+
+        def extract_text pcdata
+          pcdata.reduce [] do |accum, it|
+            case it[:type]
+            when :text
+              accum << it[:value]
+            when :element
+              accum += (extract_text it[:pcdata]) if it.key? :pcdata
+            end
+            accum
+          end
+        end
+
+        def restore_text pcdata, text_chunks
+          pcdata.each do |it|
+            case it[:type]
+            when :text
+              it[:value] = text_chunks.shift
+            when :element
+              restore_text it[:pcdata], text_chunks if it.key? :pcdata
+            end
+          end
+        end
       end
     end
   end
diff --git a/lib/asciidoctor/pdf/index_catalog.rb b/lib/asciidoctor/pdf/index_catalog.rb
index 6ab022eb..531d68ce 100644
--- a/lib/asciidoctor/pdf/index_catalog.rb
+++ b/lib/asciidoctor/pdf/index_catalog.rb
@@ -3,7 +3,7 @@
 module Asciidoctor
   module PDF
     class IndexCatalog
-      include ::Asciidoctor::PDF::TextTransformer
+      include TextTransformer
 
       LeadingAlphaRx = /^\p{Alpha}/
 
diff --git a/lib/asciidoctor/pdf/text_transformer.rb b/lib/asciidoctor/pdf/text_transformer.rb
index 18dddb9b..61e2e559 100644
--- a/lib/asciidoctor/pdf/text_transformer.rb
+++ b/lib/asciidoctor/pdf/text_transformer.rb
@@ -64,6 +64,29 @@ module Asciidoctor
           string.tr LowerAlphaChars, SmallCapsChars
         end
       end
+
+      # Apply the text transform to the specified text.
+      #
+      # Supported transform values are "uppercase", "lowercase", or "none" (passed
+      # as either a String or a Symbol). When the uppercase transform is applied to
+      # the text, it correctly uppercases visible text while leaving markup and
+      # named character entities unchanged. The none transform returns the text
+      # unmodified.
+      #
+      def transform_text text, transform
+        case transform
+        when :uppercase, 'uppercase'
+          uppercase_pcdata text
+        when :lowercase, 'lowercase'
+          lowercase_pcdata text
+        when :capitalize, 'capitalize'
+          capitalize_words_pcdata text
+        when :smallcaps, 'smallcaps'
+          smallcaps_pcdata text
+        else
+          text
+        end
+      end
     end
   end
 end
diff --git a/spec/formatted_text_formatter_spec.rb b/spec/formatted_text_formatter_spec.rb
index 530dd24c..5dd1074d 100644
--- a/spec/formatted_text_formatter_spec.rb
+++ b/spec/formatted_text_formatter_spec.rb
@@ -887,6 +887,36 @@ describe Asciidoctor::PDF::FormattedText::Formatter do
       (expect shout_text[:string]).to eql 'SHOUT'
     end
 
+    it 'should apply text transform to value of attribute reference' do
+      pdf = to_pdf <<~'EOS', pdf_theme: { role_upper_text_transform: 'uppercase' }, analyze: true
+      :brandname: acme
+
+      [.upper]#{brandname}#
+      EOS
+
+      lines = pdf.lines
+      (expect lines).to have_size 1
+      (expect lines[0]).to eql 'ACME'
+    end
+
+    it 'should apply text transform to enclosed formatted text' do
+      pdf = to_pdf <<~'EOS', pdf_theme: { role_upper_text_transform: 'uppercase' }, analyze: true
+      [.upper]#_please_ transform *bob &amp; carl* +
+      to `uppercase`#
+      EOS
+
+      lines = pdf.lines
+      (expect lines).to have_size 2
+      (expect lines).to eql ['PLEASE TRANSFORM BOB & CARL', 'TO UPPERCASE']
+      text = pdf.text
+      (expect text).to have_size 5
+      (expect text[0][:font_name]).to eql 'NotoSerif-Italic'
+      (expect text[1][:font_name]).to eql 'NotoSerif'
+      (expect text[2][:font_name]).to eql 'NotoSerif-Bold'
+      (expect text[3][:font_name]).to eql 'NotoSerif'
+      (expect text[4][:font_name]).to eql 'mplus1mn-regular'
+    end
+
     it 'should allow custom role to specify relative font size' do
       pdf_theme = {
         heading_h2_font_size: 24,