Basic Footnotes implementation

src/markd/node.cr

@@ -2,54 +2,56 @@ module Markd
   class Node
     # Node Type
     enum Type
-      Document
-      Paragraph
-      Text
-      Strong
-      Emphasis
-      Strikethrough
-      Link
-      Image
-      Heading
-      List
-      Item
       BlockQuote
-      ThematicBreak
       Code
       CodeBlock
+      CustomBlock
+      CustomInLine
+      Document
+      Emphasis
+      Footnote
+      FootnoteDefinition
+      Heading
       HTMLBlock
       HTMLInline
+      Image
+      Item
       LineBreak
+      Link
+      List
+      Paragraph
       SoftBreak
-
-      CustomInLine
-      CustomBlock
+      Strikethrough
+      Strong
       Table
       TableCell
       TableRow
+      Text
+      ThematicBreak
 
       def container?
         CONTAINER_TYPES.includes?(self)
       end
     end
 
     CONTAINER_TYPES = {
+      Type::BlockQuote,
+      Type::CustomBlock,
+      Type::CustomInLine,
       Type::Document,
-      Type::Paragraph,
-      Type::Strong,
       Type::Emphasis,
-      Type::Strikethrough,
-      Type::Link,
-      Type::Image,
+      Type::FootnoteDefinition,
       Type::Heading,
-      Type::List,
+      Type::Image,
       Type::Item,
-      Type::BlockQuote,
-      Type::CustomInLine,
-      Type::CustomBlock,
+      Type::Link,
+      Type::List,
+      Type::Paragraph,
+      Type::Strikethrough,
+      Type::Strong,
       Type::Table,
-      Type::TableRow,
       Type::TableCell,
+      Type::TableRow,
     }
 
     alias DataValue = String | Int32 | Bool

src/markd/parsers/block.cr

@@ -7,16 +7,17 @@ module Markd::Parser
     end
 
     RULES = {
-      Node::Type::Document      => Rule::Document.new,
-      Node::Type::BlockQuote    => Rule::BlockQuote.new,
-      Node::Type::Heading       => Rule::Heading.new,
-      Node::Type::CodeBlock     => Rule::CodeBlock.new,
-      Node::Type::HTMLBlock     => Rule::HTMLBlock.new,
-      Node::Type::ThematicBreak => Rule::ThematicBreak.new,
-      Node::Type::List          => Rule::List.new,
-      Node::Type::Item          => Rule::Item.new,
-      Node::Type::Paragraph     => Rule::Paragraph.new,
-      Node::Type::Table         => Rule::Table.new,
+      Node::Type::Document           => Rule::Document.new,
+      Node::Type::BlockQuote         => Rule::BlockQuote.new,
+      Node::Type::Heading            => Rule::Heading.new,
+      Node::Type::CodeBlock          => Rule::CodeBlock.new,
+      Node::Type::HTMLBlock          => Rule::HTMLBlock.new,
+      Node::Type::ThematicBreak      => Rule::ThematicBreak.new,
+      Node::Type::List               => Rule::List.new,
+      Node::Type::Item               => Rule::Item.new,
+      Node::Type::Paragraph          => Rule::Paragraph.new,
+      Node::Type::Table              => Rule::Table.new,
+      Node::Type::FootnoteDefinition => Rule::FootnoteDefinition.new,
     }
 
     property! tip : Node?
@@ -62,6 +63,55 @@ module Markd::Parser
         process_inlines
       end
 
+      if @options.gfm
+        # Extract all footnotes and footnote definitions
+        walker = @document.walker
+        footnotes = {} of String => Node
+        footnote_definitions = {} of String => Node
+        while (event = walker.next)
+          node, entering = event
+          if node.type.footnote?
+            footnotes[node.data["title"].to_s] = node
+          elsif !entering && node.type.footnote_definition?
+            footnote_definitions[node.data["title"].to_s] = node
+          end
+        end
+
+        # footnotes without definitions are converted to text
+        # and removed from our hash
+        footnotes.each do |title, _node|
+          if !footnote_definitions.keys.includes? title
+            _node.type = Node::Type::Text
+            _node.text = "[^#{title}]"
+            footnotes.delete title
+          end
+        end
+        # definitions without footnotes are removed
+        # and popped from our hash
+        footnote_definitions.each do |title, _node|
+          if !footnotes.keys.includes? title
+            _node.unlink
+            footnote_definitions.delete title
+          end
+        end
+        # Footnote numbers are normalized to 1...n
+        # Footnotes are always ordered because the important thing is
+        # appearing in the right order in the document, but now there
+        # may be holes in the numbering.
+        # Also, definitions get the matching number in their own
+        # metadata.
+        footnotes.each_with_index do |(title, node), index|
+          node.data["number"] = index + 1
+          footnote_definitions[title].data["number"] = index + 1
+        end
+
+        # Footnote definitionss are moved to the end of the document
+        footnotes.each do |(title, _)|
+          node = footnote_definitions[title]
+          node.unlink
+          @document.append_child(node)
+        end
+      end
       @document
     end
 
@@ -197,7 +247,7 @@ module Markd::Parser
       @inline_lexer.refmap = @refmap
       while (event = walker.next)
         node, entering = event
-        if !entering && (node.type.paragraph? || node.type.heading? || node.type.table_cell?)
+        if !entering && (node.type.paragraph? || node.type.heading? || node.type.table_cell? || node.type.footnote_definition?)
           @inline_lexer.parse(node)
         end
       end

src/markd/parsers/inline.cr

@@ -13,6 +13,7 @@ module Markd::Parser
       @text = ""
       @pos = 0
       @refmap = {} of String => Hash(String, String) | String
+      @footnote_counter = 0
     end
 
     def parse(node : Node)
@@ -151,22 +152,29 @@ module Markd::Parser
     private def bang(node : Node)
       start_pos = @pos
       @pos += 1
-      if char_at?(@pos) == '['
+      # It's an image if the next character is a [
+      # And it's not a footnote OR we are not in GFM mode
+      #
+      # * This IS an image: `![...]`
+      # * This IS an image in non-GFM mode: `![^...]`
+      # * This IS NOT an image in GFM mode: `![^...]`
+
+      if char_at?(@pos) == '[' && (char_at?(@pos + 1) != '^' || !@options.gfm)
         @pos += 1
         child = text("![")
         node.append_child(child)
 
-        add_bracket(child, start_pos + 1, true)
+        add_bracket(child, start_pos + 1, image: true)
       else
         node.append_child(text("!"))
       end
 
       true
     end
 
-    private def add_bracket(node : Node, index : Int32, image = false)
+    private def add_bracket(node : Node, index : Int32, *, image = false, footnote = false)
       brackets.bracket_after = true if brackets?
-      @brackets = Bracket.new(node, @brackets, @delimiters, index, image, true)
+      @brackets = Bracket.new(node, @brackets, @delimiters, index, image: image, active: true, footnote: footnote)
     end
 
     private def remove_bracket
@@ -180,7 +188,11 @@ module Markd::Parser
       child = text("[")
       node.append_child(child)
 
-      add_bracket(child, start_pos, false)
+      if char_at(@pos) == '^' && @options.gfm
+        add_bracket(child, start_pos, footnote: true)
+      else
+        add_bracket(child, start_pos)
+      end
 
       true
     end
@@ -210,6 +222,7 @@ module Markd::Parser
 
       # If we got here, open is a potential opener
       is_image = opener.image
+      is_footnote = opener.footnote
 
       # Check to see if we have a link/image
       save_pos = @pos
@@ -228,6 +241,21 @@ module Markd::Parser
         end
       end
 
+      # Is it a footnote?
+      if is_footnote
+        # If the 1st char after the closing bracket is a ":" then it's NOT
+        # a footnote, it's a footnote definition.
+        if char_at?(@pos) == ':'
+          @pos = start_pos
+          node.append_child(text("]"))
+          remove_bracket
+          return true
+        end
+        title = @text[opener.@index + 2...@pos - 1]
+        @pos += 1
+        matched = true
+      end
+
       ref_label = nil
       unless matched
         # Next, see if there's a link label
@@ -257,7 +285,15 @@ module Markd::Parser
       end
 
       if matched
-        child = Node.new(is_image ? Node::Type::Image : Node::Type::Link)
+        if is_image
+          child = Node.new(Node::Type::Image)
+        elsif is_footnote
+          child = Node.new(Node::Type::Footnote)
+          child.data["number"] = @footnote_counter += 1
+        else
+          child = Node.new(Node::Type::Link)
+        end
+
         child.data["destination"] = dest.not_nil!
         child.data["title"] = title || ""
 
@@ -907,8 +943,9 @@ module Markd::Parser
       property image : Bool
       property active : Bool
       property bracket_after : Bool
+      property footnote : Bool
 
-      def initialize(@node, @previous, @previous_delimiter, @index, @image, @active = true)
+      def initialize(@node, @previous, @previous_delimiter, @index, *, @image = false, @active = true, @footnote = false)
         @bracket_after = false
       end
     end

src/markd/renderer.cr

@@ -49,27 +49,29 @@ module Markd
       false
     end
 
-    abstract def heading(node : Node, entering : Bool) : Nil
-    abstract def list(node : Node, entering : Bool) : Nil
-    abstract def item(node : Node, entering : Bool) : Nil
     abstract def block_quote(node : Node, entering : Bool) : Nil
-    abstract def thematic_break(node : Node, entering : Bool) : Nil
     abstract def code_block(node : Node, entering : Bool, formatter : T?) : Nil forall T
     abstract def code(node : Node, entering : Bool) : Nil
+    abstract def emphasis(node : Node, entering : Bool) : Nil
+    abstract def footnote(node : Node, entering : Bool) : Nil
+    abstract def footnote_definition(node : Node, entering : Bool) : Nil
+    abstract def heading(node : Node, entering : Bool) : Nil
     abstract def html_block(node : Node, entering : Bool) : Nil
     abstract def html_inline(node : Node, entering : Bool) : Nil
+    abstract def image(node : Node, entering : Bool) : Nil
+    abstract def item(node : Node, entering : Bool) : Nil
+    abstract def line_break(node : Node, entering : Bool) : Nil
+    abstract def link(node : Node, entering : Bool) : Nil
+    abstract def list(node : Node, entering : Bool) : Nil
     abstract def paragraph(node : Node, entering : Bool) : Nil
-    abstract def emphasis(node : Node, entering : Bool) : Nil
     abstract def soft_break(node : Node, entering : Bool) : Nil
-    abstract def line_break(node : Node, entering : Bool) : Nil
-    abstract def strong(node : Node, entering : Bool) : Nil
     abstract def strikethrough(node : Node, entering : Bool) : Nil
-    abstract def link(node : Node, entering : Bool) : Nil
-    abstract def image(node : Node, entering : Bool) : Nil
-    abstract def text(node : Node, entering : Bool) : Nil
-    abstract def table(node : Node, entering : Bool) : Nil
-    abstract def table_row(node : Node, entering : Bool) : Nil
+    abstract def strong(node : Node, entering : Bool) : Nil
     abstract def table_cell(node : Node, entering : Bool) : Nil
+    abstract def table_row(node : Node, entering : Bool) : Nil
+    abstract def table(node : Node, entering : Bool) : Nil
+    abstract def text(node : Node, entering : Bool) : Nil
+    abstract def thematic_break(node : Node, entering : Bool) : Nil
 
     def render(document : Node, formatter : T?) forall T
       Utils.timer("rendering", @options.time) do
@@ -118,6 +120,10 @@ module Markd
             table_row(node, entering)
           when Node::Type::TableCell
             table_cell(node, entering)
+          when Node::Type::Footnote
+            footnote(node, entering)
+          when Node::Type::FootnoteDefinition
+            footnote_definition(node, entering)
           else
             text(node, entering)
           end