coradoc-adoc 2.0.10 → 2.0.12

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 78eaf011433933b47142179dd46e5e1176fe7f0686b2e803b1c35a0ed8f0fc93
-  data.tar.gz: 39f23d77a5df2807f35e895e751bbfa0e4030e34a6d1c6372007061fa211f07d
+  metadata.gz: 8e7f028d984f0019578e97882169ce2be502e17fcfdeb376766f088d1af23958
+  data.tar.gz: 68533e4e500d008b4fc37f1beecc1fd160318adb6f9a6ed43fdf87a97ffedb31
 SHA512:
-  metadata.gz: 358a5c0d2562840abc484421446110c917112f68515eb22c9bd1d69c6cafe18832a1ffb3a71c37df8640b309e4aa9ec7b7e544393ed795ee63b894c7a1bd504d
-  data.tar.gz: aed852abae11e22ef32fbb764f9c09d2334c24ef3f1e0c9bd23fe4c5295ddf2b15adeca603713f732a3ffdb2250bf3a9def286fd6ecf563c9dc5f840cc251f23
+  metadata.gz: 6c386bcee2e9a2bc1f3e47f2fca189b92bb7ef568fdb5105c2b59da916a43cc27bf912699c92e850c4f782e3e7575c5472a42262cb57a8da63ba14a5d1de88a3
+  data.tar.gz: 45e2116e5532d41b0f76a855c12e51066d26b648b947f3b6a95f817c7772794d0af57ee1b57fb402e7579d9b620e2eb61d44c99962b1030d6de74319f27920ff

data/lib/coradoc/asciidoc/model/admonition.rb

@@ -28,7 +28,13 @@ module Coradoc
       #   admonition.content = "Be careful!"
       #
       class Admonition < Attached
-        attribute :content, :string
+        attribute :content,
+                  Lutaml::Model::Serializable,
+                  collection: true,
+                  polymorphic: [
+                    Lutaml::Model::Type::String,
+                    Coradoc::AsciiDoc::Model::TextElement
+                  ]
         attribute :type, :string
         attribute :line_break, :string, default: -> { '' }
       end

data/lib/coradoc/asciidoc/model/bibliography_entry.rb

@@ -36,9 +36,12 @@ module Coradoc
         # Coerce a raw parser AST value into the canonical ref_text string.
         # Accepts the shapes produced by Parser::Bibliography for `:ref_text`:
         # nil, Parslet::Slice, plain String, single Model::Base, or an Array
-        # of any of these. Keeping this coercion on the model that owns
-        # ref_text (rather than in a transformer rule) keeps the transformer
-        # declarative and lets callers build entries from any source shape.
+        # of any of these. Model objects (TextElement, Inline::Italic, etc.)
+        # are flattened via TextExtractVisitor so their text content is
+        # preserved instead of leaking `#<Class:0x...>` inspect strings.
+        # Keeping this coercion on the model that owns ref_text (rather than
+        # in a transformer rule) keeps the transformer declarative and lets
+        # callers build entries from any source shape.
         # @param raw [Object, nil]
         # @return [String]
         def self.coerce_ref_text(raw)
@@ -47,6 +50,8 @@ module Coradoc
           case raw
           when Array then raw.map { |e| coerce_ref_text(e) }.join
           when String then raw
+          when Coradoc::AsciiDoc::Model::Base
+            Coradoc::AsciiDoc::Transform::TextExtractVisitor.new.extract(raw).to_s
           else raw.to_s
           end
         end

data/lib/coradoc/asciidoc/parser/admonition.rb

@@ -15,7 +15,7 @@ module Coradoc
 
         def admonition_line
           admonition_type.as(:admonition_type) >> str(': ') >>
-            (text.as(:text) >>
+            (text_any.as(:text) >>
             line_ending.as(:line_break)
             ).repeat(1)
             .as(:content)

data/lib/coradoc/asciidoc/parser/block.rb

@@ -104,11 +104,22 @@ module Coradoc
             delim_str = c.captures[capture_key].to_s.strip
             closing_pattern = str(delim_str) >> newline
 
-            content = block_image
-            content |= block(n_deep - 1) if n_deep.positive?
-            content |= list
-            content |= text_line(false, unguarded: true, verbatim: verbatim)
-            content |= empty_line.as(:line_break)
+            # Verbatim blocks (source/listing) treat their body as literal
+            # text per the AsciiDoc spec — no substitutions, no nested
+            # blocks. Allowing nested block parsing here would consume
+            # shorter inner delimiters (e.g. `----` inside `------`) and
+            # strip the original structure when serializing back.
+            content = if verbatim
+                        text_line(false, unguarded: true, verbatim: true) |
+                          empty_line.as(:line_break)
+                      else
+                        c = block_image
+                        c |= block(n_deep - 1) if n_deep.positive?
+                        c |= list
+                        c |= text_line(false, unguarded: true)
+                        c |= empty_line.as(:line_break)
+                        c
+                      end
 
             (closing_pattern.absent? >> content).repeat(1)
           end

data/lib/coradoc/asciidoc/parser/content.rb

@@ -24,7 +24,7 @@ module Coradoc
         # :zero :one :many
         def text_line(many_breaks = false, unguarded: false, verbatim: false)
           tl = if verbatim
-                 text_any.as(:text)
+                 raw_verbatim_line.as(:text)
                elsif unguarded
                  literal_space? >> text_any.as(:text)
                else
@@ -38,6 +38,15 @@ module Coradoc
           end
         end
 
+        # A verbatim source/listing line: capture every character up to the
+        # next newline without applying any inline substitutions. Source and
+        # listing blocks must round-trip their text untouched — otherwise
+        # sequences like Liquid `{{ var }}` collide with AsciiDoc attribute
+        # references and get rewritten.
+        def raw_verbatim_line
+          (line_ending.absent? >> any).repeat(1)
+        end
+
         def asciidoc_char
           line_start? >> match['*_:+=\-']
         end

data/lib/coradoc/asciidoc/transform/callout_merger.rb

@@ -0,0 +1,94 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module AsciiDoc
+    module Transform
+      # Post-processing pass that merges AsciiDoc callout annotation
+      # paragraphs into the verbatim block they annotate.
+      #
+      # AsciiDoc callouts look like:
+      #
+      #   [source,ruby]
+      #   ----
+      #   get '/hi' do <1>
+      #   ----
+      #   <1> Returns hello world
+      #
+      # The parser emits the source block and the annotation as two
+      # independent children. The CoreModel representation should attach
+      # the annotation to the block as a typed Callout, so downstream
+      # serializers can render them appropriately for each format.
+      #
+      # Single responsibility: take a flat list of transformed CoreModel
+      # children, return a flat list with `<N>` paragraphs adjacent to a
+      # SourceBlock / ListingBlock folded into that block's `callouts`.
+      # Anything else is passed through untouched.
+      class CalloutMerger
+        ANNOTATION_LINE = /<(\d+)>\s*(.*?)\s*\z/
+        ANNOTATION_SPLIT = /(?=<\d+>)/
+
+        class << self
+          def call(children)
+            new.merge(Array(children))
+          end
+        end
+
+        # Walks the input children left-to-right. When a paragraph whose
+        # content is composed entirely of `<N> text` lines follows a
+        # verbatim block (SourceBlock or ListingBlock), each `<N>` line
+        # becomes a Callout attached to that block instead of a separate
+        # paragraph.
+        #
+        # Annotations that do not follow a verbatim block are preserved
+        # verbatim — they may be legitimate prose.
+        def merge(children)
+          children.each.with_object([]) do |child, result|
+            annotations = extract_annotations(child)
+            target = annotations && preceding_verbatim_block(result)
+            if target
+              target.callouts.concat(annotations)
+            else
+              result << child
+            end
+          end
+        end
+
+        private
+
+        # Returns an Array of Callout if the paragraph is entirely
+        # callout annotations, otherwise nil.
+        def extract_annotations(child)
+          return nil unless child.is_a?(Coradoc::CoreModel::ParagraphBlock)
+
+          lines = split_annotation_lines(child.flat_text)
+          return nil if lines.empty?
+
+          callouts = lines.map do |line|
+            match = line.match(ANNOTATION_LINE)
+            match ? Coradoc::CoreModel::Callout.new(index: match[1].to_i, content: match[2]) : nil
+          end
+          return nil if callouts.any?(&:nil?)
+
+          callouts
+        end
+
+        # Splits a paragraph into candidate annotation lines. Returns []
+        # if any non-blank segment fails to look like an annotation.
+        def split_annotation_lines(text)
+          return [] if text.nil? || text.strip.empty?
+
+          chunks = text.strip.split(ANNOTATION_SPLIT)
+          chunks.map(&:strip).reject(&:empty?)
+        end
+
+        def preceding_verbatim_block(result)
+          last = result.last
+          return nil unless last.is_a?(Coradoc::CoreModel::SourceBlock) ||
+                            last.is_a?(Coradoc::CoreModel::ListingBlock)
+
+          last
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/asciidoc/transform/element_transformers/document_transformer.rb

@@ -10,6 +10,7 @@ module Coradoc
               title_text = ToCoreModel.extract_title_text(doc.header&.title)
               attributes = ToCoreModel.extract_document_attributes(doc)
               children = ToCoreModel.transform(doc.sections || doc.contents || [])
+              children = CalloutMerger.call(children)
               children = prepend_frontmatter(children, doc.frontmatter)
 
               Coradoc::CoreModel::DocumentElement.new(
@@ -38,6 +39,7 @@ module Coradoc
               )
 
               content_children = ToCoreModel.transform(section.contents || [])
+              content_children = CalloutMerger.call(content_children)
               nested_sections = (section.sections || []).map do |child|
                 transform_section(child, parent_id: section_id)
               end

data/lib/coradoc/asciidoc/transform/from_core_model.rb

@@ -45,7 +45,7 @@ module Coradoc
               Coradoc::AsciiDoc::Model::Document.new(
                 id: element.id,
                 header: header,
-                sections: transform(sections),
+                sections: flatten_children(sections),
                 frontmatter: frontmatter
               )
             when CoreModel::SectionElement
@@ -53,17 +53,29 @@ module Coradoc
                 id: element.id,
                 level: element.level,
                 title: create_title(element.title, element.level),
-                contents: transform(element.children)
+                contents: flatten_children(element.children)
               )
             else
               Coradoc::AsciiDoc::Model::Section.new(
                 id: element.id,
                 title: create_title(element.title, 1),
-                contents: transform(element.children)
+                contents: flatten_children(element.children)
               )
             end
           end
 
+          # Transforms each CoreModel child and flattens one level so a
+          # transform that returns multiple siblings (e.g. a source block
+          # followed by its re-expanded callout paragraphs) stays in
+          # document order.
+          def flatten_children(children)
+            Array(children).flat_map { |child| flatten_one(transform(child)) }
+          end
+
+          def flatten_one(result)
+            result.is_a?(Array) ? result : [result]
+          end
+
           def transform_block(block)
             content = block.renderable_content
 
@@ -82,7 +94,13 @@ module Coradoc
             end
 
             content_text = safe_content_to_string(content)
+            result = build_verbatim_block(semantic, block, content_text)
+            return result unless verbatim_with_callouts?(semantic, block)
 
+            [result, *build_callout_paragraphs(block.callouts)]
+          end
+
+          def build_verbatim_block(semantic, block, content_text)
             case semantic
             when :source_code
               Coradoc::AsciiDoc::Model::Block::SourceCode.new(
@@ -163,6 +181,23 @@ module Coradoc
             end
           end
 
+          def verbatim_with_callouts?(semantic, block)
+            return false unless %i[source_code listing].include?(semantic)
+            return false if block.callouts.nil? || block.callouts.empty?
+
+            true
+          end
+
+          # Re-expands typed Callouts back into the AsciiDoc `<N> text`
+          # paragraph form so the round-trip is faithful.
+          def build_callout_paragraphs(callouts)
+            callouts.sort_by { |c| c.index || Float::INFINITY }.map do |callout|
+              Coradoc::AsciiDoc::Model::Paragraph.new(
+                content: create_text_elements("<#{callout.index}> #{callout.content}")
+              )
+            end
+          end
+
           def transform_table(table)
             rows = Array(table.rows).map do |row|
               columns = Array(row.cells).map do |cell|

data/lib/coradoc/asciidoc/transform.rb

@@ -13,6 +13,7 @@ module Coradoc
       autoload :InlineTransformVisitor, "#{__dir__}/transform/inline_transform_visitor"
       autoload :ElementTransformers, "#{__dir__}/transform/element_transformers"
       autoload :FrontmatterAttributeMap, "#{__dir__}/transform/frontmatter_attribute_map"
+      autoload :CalloutMerger, "#{__dir__}/transform/callout_merger"
     end
   end
 end

data/lib/coradoc/asciidoc/version.rb

@@ -2,6 +2,6 @@
 
 module Coradoc
   module AsciiDoc
-    VERSION = '2.0.10'
+    VERSION = '2.0.12'
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: coradoc-adoc
 version: !ruby/object:Gem::Version
-  version: 2.0.10
+  version: 2.0.12
 platform: ruby
 authors:
 - Ribose Inc.
@@ -313,6 +313,7 @@ files:
 - lib/coradoc/asciidoc/serializer/serializers/video.rb
 - lib/coradoc/asciidoc/serializer/spacing_strategy.rb
 - lib/coradoc/asciidoc/transform.rb
+- lib/coradoc/asciidoc/transform/callout_merger.rb
 - lib/coradoc/asciidoc/transform/element_transformers.rb
 - lib/coradoc/asciidoc/transform/element_transformers/block_transformer.rb
 - lib/coradoc/asciidoc/transform/element_transformers/document_transformer.rb