diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml index fc1bd28..06cee60 100644 --- a/.github/workflows/ci.yml +++ b/.github/workflows/ci.yml @@ -13,7 +13,7 @@ jobs: runs-on: ubuntu-latest strategy: matrix: - ameba-version: [v1.6.4] + ameba-version: [v1.7.0] name: Ameba ${{ matrix.ameba-version }} linter check steps: - name: Install latest Crystal @@ -64,7 +64,7 @@ jobs: crystal: ${{ matrix.crystal }} - name: Install dependencies - run: shards install --without-development + run: shards install - name: Run specs run: crystal spec --error-on-warnings --error-trace diff --git a/.gitignore b/.gitignore index 1bad4d9..98023d3 100644 --- a/.gitignore +++ b/.gitignore @@ -4,10 +4,18 @@ /.shards/ /src/main.cr +# Compiled spec binary +/markd_spec + # Libraries don't need dependency lock # Dependencies will be locked in application that uses them /shard.lock # vscode /.history/ -/.vscode/settings.json +/.vscode/ + + +# claude +CLAUDE.md +/.claude/ diff --git a/.vscode/launch.json b/.vscode/launch.json deleted file mode 100644 index 3ab15c3..0000000 --- a/.vscode/launch.json +++ /dev/null @@ -1,13 +0,0 @@ -{ - "version": "0.2.0", - "configurations": [ - { - "type": "lldb", - "request": "launch", - "name": "Launch", - "program": "${workspaceRoot}/bin/main", - "args": [], - "cwd": "${workspaceRoot}" - } - ] -} diff --git a/README.md b/README.md index d3a99f1..eeef57b 100644 --- a/README.md +++ b/README.md @@ -48,8 +48,8 @@ Markd.to_html(markdown, options) | source_pos | `Bool` | false | if **true**, source position information for block-level elements
will be rendered in the data-sourcepos attribute (for HTML) | | safe | `Bool` | false | if **true**, raw HTML will not be passed through to HTML output (it will be replaced by comments) | | prettyprint | `Bool` | false | if **true**, code tags generated by code blocks will have a `prettyprint` class added to them, to be used by [Google code-prettify](https://github.com/google/code-prettify). | -| gfm | `Bool` | false | **Partial support** | -| autolink | `Bool` | false | if **true**, more autolinks are detected, like bare email addresses or http links | +| gfm | `Bool` | false | **Partial support** | +| autolink | `Bool` | false | if **true**, more autolinks are detected, like bare email addresses or http links | | toc | `Bool` | false | **Not supported for now** | | emoji | `Bool` | false | if **true**, convert Github style emoji chars. | | tagfilter | `Bool` | false | if **true**, escape certain HTML tags. | diff --git a/shard.yml b/shard.yml index 7f06234..709771f 100644 --- a/shard.yml +++ b/shard.yml @@ -2,7 +2,7 @@ name: markd version: 0.6.0 development_dependencies: - tartrazine : + tartrazine: github: ralsina/tartrazine authors: diff --git a/spec/fixtures/gfm-extensions.txt b/spec/fixtures/gfm-extensions.txt index a5ff96d..dbb876d 100644 --- a/spec/fixtures/gfm-extensions.txt +++ b/spec/fixtures/gfm-extensions.txt @@ -704,7 +704,7 @@ Even with {"x":"y"} or 1 > 2 or whatever. Even **markdown**. ## Footnotes -```````````````````````````````` example pending +```````````````````````````````` example This is some text![^1]. Other text.[^footnote]. Here's a thing[^other-note]. @@ -762,9 +762,54 @@ Hi! ```````````````````````````````` +Multibyte characters before a reference must not shift its label: +the parser tracks byte offsets but the label used to be cut with +character-based slicing. + +```````````````````````````````` example +Emoji ✅[^emoji] and accented é[^accent]. + +| Tables | Cells | +| ------ | ----- | +| emoji | ✅[^t-emoji] | + +[^emoji]: emoji note +[^accent]: accent note +[^t-emoji]: table note +. +

Emoji ✅1 and accented é2.

+ + + + + + + + + + + + + +
TablesCells
emoji✅3
+
+
    +
  1. +

    emoji note ↩

    +
    2. +
  2. +

    accent note ↩

    +
    3. +
  3. +

    table note ↩

    +
    4. +
+
+```````````````````````````````` + ## When a footnote is used multiple times, we insert multiple backrefs. -```````````````````````````````` example pending +```````````````````````````````` example This is some text. It has a footnote[^a-footnote]. This footnote is referenced[^a-footnote] multiple times, in lots of different places.[^a-footnote] @@ -784,7 +829,7 @@ This footnote is referenced[^a-footnote] multiple times, in lots of different pl ## Footnote reference labels are href escaped -```````````````````````````````` example pending +```````````````````````````````` example Hello[^">] [^">]: pwned diff --git a/spec/fixtures/gfm-regression.txt b/spec/fixtures/gfm-regression.txt index b611a0c..302b3ab 100644 --- a/spec/fixtures/gfm-regression.txt +++ b/spec/fixtures/gfm-regression.txt @@ -167,7 +167,7 @@ This ~text~ ~~is~~ ~~~curious~~~. Footnotes in tables -```````````````````````````````` example table footnotes pending +```````````````````````````````` example table footnotes A footnote in a paragraph[^1] | Column1 | Column2 | @@ -273,7 +273,7 @@ Pull request #128 - Buffer overread in tables extension Footnotes may be nested inside other footnotes. -```````````````````````````````` example footnotes pending +```````````````````````````````` example footnotes This is some text. It has a citation.[^citation] [^another-citation]: My second citation. @@ -295,7 +295,7 @@ This is some text. It has a citation.[^citation] Footnotes are similar to, but should not be confused with, link references -```````````````````````````````` example footnotes pending +```````````````````````````````` example footnotes This is some text. It has two footnotes references, side-by-side without any spaces,[^footnote1][^footnote2] which are definitely not link references. [^footnote1]: Hello. @@ -317,7 +317,7 @@ This is some text. It has two footnotes references, side-by-side without any spa Footnotes may begin with or have a 'w' or a '_' in their reference label. -```````````````````````````````` example footnotes autolink pending +```````````````````````````````` example footnotes autolink This is some text. Sometimes the autolinker splits up text into multiple nodes, hoping it will find a hyperlink, so this text has a footnote whose reference label begins with a `w`.[^widely-cited] It has another footnote that contains many different characters (the autolinker was also breaking on `_`).[^sphinx-of-black-quartz_judge-my-vow-0123456789] @@ -342,7 +342,8 @@ It has another footnote that contains many different characters (the autolinker Footnotes interacting with strikethrough should not lead to a use-after-free -```````````````````````````````` example footnotes autolink strikethrough table pending + +```````````````````````````````` example footnotes autolink strikethrough table |Tot.....[^_a_]| .

|Tot.....[^_a_]|

@@ -350,7 +351,8 @@ Footnotes interacting with strikethrough should not lead to a use-after-free Footnotes interacting with strikethrough should not lead to a use-after-free pt2 -```````````````````````````````` example footnotes autolink strikethrough table pending + +```````````````````````````````` example footnotes autolink strikethrough table [^~~is~~1] .

[^~~is~~1]

@@ -374,3 +376,52 @@ Issue #424 - emphasis before links .

text link

```````````````````````````````` + +A footnote reference followed by (parentheses) is not parsed as a link + +```````````````````````````````` example footnotes +Text with a footnote[^a], and [^a](url) with a link-ish suffix. + +[^a]: note one +. +

Text with a footnote1, and 1(url) with a link-ish suffix.

+
+
    +
  1. +

    note one ↩ ↩2

    +
    2. +
+
+```````````````````````````````` + +Footnote definitions must start at the beginning of the line + +```````````````````````````````` example footnotes +foo [^b]: this is mid-line, so not a definition + +[^b]: real note +. +

foo [^b]: this is mid-line, so not a definition

+```````````````````````````````` + +Nested footnote references may contain arbitrary characters + +```````````````````````````````` example footnotes +See[^one]. + +[^one]: outer [^">] + +[^">]: inner note +. +

See1.

+
+
    +
  1. +

    outer 2 ↩

    +
    2. +
  2. +

    inner note ↩

    +
    3. +
+
+```````````````````````````````` diff --git a/spec/fixtures/tartrazine.cr b/spec/fixtures/tartrazine.cr index 5b1cb9f..f3473bf 100644 --- a/spec/fixtures/tartrazine.cr +++ b/spec/fixtures/tartrazine.cr @@ -2,9 +2,9 @@ require "tartrazine" require "../../src/markd" markdown = <<-MD -# Hello Markd + # Hello Markd -> Yet another markdown parser built for speed, written in Crystal, Compliant to CommonMark specification. -MD + > Yet another markdown parser built for speed, written in Crystal, Compliant to CommonMark specification. + MD -html = Markd.to_html(markdown) +Markd.to_html(markdown) diff --git a/src/markd/node.cr b/src/markd/node.cr index 31223b3..783c29b 100644 --- a/src/markd/node.cr +++ b/src/markd/node.cr @@ -2,32 +2,33 @@ module Markd class Node # Node Type enum Type - Document - Paragraph - Text - Strong - Emphasis - Strikethrough - Link - Image - Heading - List - Item BlockQuote Alert 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 def container? CONTAINER_TYPES.includes?(self) @@ -35,23 +36,24 @@ module Markd end CONTAINER_TYPES = { + Type::BlockQuote, + Type::Alert, + 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::Alert, - 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 @@ -132,7 +134,7 @@ module Markd end def to_s(io : IO) - io << "#<" << {{@type.name.id.stringify}} << ":0x" + io << "#<" << {{ @type.name.id.stringify }} << ":0x" object_id.to_s(16, io) io << " @type=" << @type io << " @parent=" << @parent if @parent diff --git a/src/markd/parsers/block.cr b/src/markd/parsers/block.cr index 2f0d459..0c8a2b5 100644 --- a/src/markd/parsers/block.cr +++ b/src/markd/parsers/block.cr @@ -7,17 +7,18 @@ module Markd::Parser end RULES = { - Node::Type::Document => Rule::Document.new, - Node::Type::BlockQuote => Rule::BlockQuote.new, - Node::Type::Alert => Rule::BlockQuote.new, # Alerts and BlockQuotes are the same - 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::Alert => Rule::BlockQuote.new, # Alerts and BlockQuotes are the same + 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? @@ -59,13 +60,142 @@ module Markd::Parser parse_blocks(source) end + # Process footnote definitions - parse their content as blocks + if @options.gfm? + process_footnote_definitions + end + Utils.timer("inline parsing", @options.time?) do process_inlines end + if @options.gfm? + process_footnotes + end + @document end + # Process footnotes: extract, resolve nested references, number them, and move definitions to end + private def process_footnotes + # Extract all footnotes and footnote definitions + walker = @document.walker + footnotes = {} of String => Array(Node) + footnote_definitions = {} of String => Node + while (event = walker.next) + node, entering = event + if node.type.footnote? + title = node.data["title"].to_s + footnotes[title] ||= [] of Node + footnotes[title] << node + elsif !entering && node.type.footnote_definition? + footnote_definitions[node.data["title"].to_s] = node + end + end + + # Recursively resolve nested footnotes in definitions before numbering + loop do + new_nodes = 0 + footnotes.keys.each do |footnote_title| + nodes = footnotes[footnote_title]? + next unless nodes + if !footnote_definitions.has_key?(footnote_title) + nodes.each do |fn_node| + fn_node.type = Node::Type::Text + fn_node.text = "[^#{footnote_title}]" + end + footnotes.delete footnote_title + else + def_node = footnote_definitions[footnote_title] + walker = def_node.walker + while (event = walker.next) + n, entering = event + if entering && n.type.text? + replaced = false + n.text = n.text.gsub(/\[\^([^\]]+)\]/) do |m| + nested_label = $1 + if footnote_definitions[nested_label]? + fn = Node.new(Node::Type::Footnote) + fn.data["title"] = nested_label + n.insert_after(fn) + footnotes[nested_label] ||= [] of Node + footnotes[nested_label] << fn + replaced = true + new_nodes += 1 + "" + else + m + end + end + if replaced && n.parent? + @inline_lexer.parse(n.parent) + end + n.unlink if n.text.empty? + end + end + end + end + break if new_nodes == 0 + end + + # Ensure all Footnote nodes are present in the footnotes hash + walker = @document.walker + while (event = walker.next) + node, entering = event + if node.type.footnote? + title = node.data["title"].to_s + footnotes[title] ||= [] of Node + unless footnotes[title].includes?(node) + footnotes[title] << node + end + end + end + + # Remove definitions without footnotes + footnote_definitions.keys.each do |footnote_title| + def_node = footnote_definitions[footnote_title]? + next unless def_node + unless footnotes.has_key?(footnote_title) + def_node.unlink + footnote_definitions.delete footnote_title + end + end + + # Footnote numbers are normalized to 1...n + # Each reference gets a ref_index (1, 2, 3...) and the definition gets a ref_count + footnote_number = 0 + footnotes.each do |footnote_title, nodes| + footnote_number += 1 + nodes.each_with_index do |fn_node, index| + fn_node.data["number"] = footnote_number + fn_node.data["ref_index"] = index + 1 + end + footnote_definitions[footnote_title].data["number"] = footnote_number + footnote_definitions[footnote_title].data["ref_count"] = nodes.size + end + + # Footnote definitions are moved to the end of the document + footnotes.each do |footnote_title, _| + def_node = footnote_definitions[footnote_title] + def_node.unlink + @document.append_child(def_node) + end + + # After all footnote definitions are attached, re-run inline parsing on their children + footnote_definitions.each_value do |def_node| + child = def_node.first_child? + while child + next_child = child.next? + if child.type.paragraph? || child.type.heading? || child.type.table_cell? + @inline_lexer.parse(child) + end + child = next_child + end + end + + nil + end + private def parse_blocks(source) lines_size = 0 source.each_line do |line| @@ -198,6 +328,8 @@ module Markd::Parser @inline_lexer.refmap = @refmap while (event = walker.next) node, entering = event + # Note: footnote_definition is not parsed here because its content + # has already been parsed as blocks in process_footnote_definitions if !entering && (node.type.paragraph? || node.type.heading? || node.type.table_cell?) @inline_lexer.parse(node) end @@ -206,6 +338,34 @@ module Markd::Parser nil end + # Parse footnote definition contents as block-level content + private def process_footnote_definitions + walker = @document.walker + while (event = walker.next) + node, entering = event + if entering && node.type.footnote_definition? && !node.text.empty? + # Parse the footnote definition content as blocks + content = node.text + node.text = "" + + # Create a temporary sub-parser for the footnote content + sub_parser = Block.new(@options) + sub_doc = sub_parser.parse(content) + + # Move all children from the sub-document to the footnote definition + child = sub_doc.first_child? + while child + next_child = child.next? + child.unlink + node.append_child(child) + child = next_child + end + end + end + + nil + end + def token(container : Node, line_number : Int32) container_parent = container.parent? diff --git a/src/markd/parsers/inline.cr b/src/markd/parsers/inline.cr index 8854318..78c34dc 100644 --- a/src/markd/parsers/inline.cr +++ b/src/markd/parsers/inline.cr @@ -14,6 +14,7 @@ module Markd::Parser @text = "" @pos = 0 @refmap = {} of String => Hash(String, String) | String + @footnote_counter = 0 end def parse(node : Node) @@ -194,12 +195,19 @@ 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 @@ -207,9 +215,9 @@ module Markd::Parser 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 @@ -223,7 +231,11 @@ 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 @@ -253,12 +265,15 @@ 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 # Inline link? - if char_at?(@pos) == '(' + # Footnote openers ([^...]) are not links, so don't try to parse a + # `(destination)` after the closing bracket for them. + if !is_footnote && char_at?(@pos) == '(' @pos += 1 if spnl && (dest = link_destination) && spnl && (char_at?(@pos - 1).try(&.whitespace?) && @@ -271,6 +286,22 @@ 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 + # @pos and opener.@index are byte offsets: slice bytes, or any + # multi-byte character earlier in the line shifts the label. + title = @text.byte_slice(opener.@index + 2, @pos - opener.@index - 3) + matched = true + end + ref_label = nil unless matched # Next, see if there's a link label @@ -300,7 +331,15 @@ 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 || "" @@ -904,7 +943,7 @@ module Markd::Parser text = @text.byte_slice(@pos) if (match = text.match(regex)) @pos += match.byte_end.not_nil! - return match[0] + match[0] end end @@ -1101,8 +1140,9 @@ 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 diff --git a/src/markd/renderer.cr b/src/markd/renderer.cr index 2283981..1578e4c 100644 --- a/src/markd/renderer.cr +++ b/src/markd/renderer.cr @@ -49,28 +49,30 @@ 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 alert(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 def render(document : Node, formatter : T? = nil) forall T Utils.timer("rendering", @options.time?) do @@ -121,6 +123,10 @@ 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 diff --git a/src/markd/renderers/html_renderer.cr b/src/markd/renderers/html_renderer.cr index a6f65d6..912ca12 100644 --- a/src/markd/renderers/html_renderer.cr +++ b/src/markd/renderers/html_renderer.cr @@ -257,11 +257,76 @@ module Markd newline tag("p", attrs(node)) else + # If this is the last paragraph in a footnote definition, append backrefs + if last_paragraph_in_footnote?(node) + append_footnote_backrefs(node.parent) + end tag("p", end_tag: true) newline end end + # Check if this paragraph is the last paragraph child of a footnote definition + private def last_paragraph_in_footnote?(node : Node) : Bool + parent = node.parent? + return false unless parent && parent.type.footnote_definition? + + # Check if this is the last paragraph (there might be no more paragraphs after this) + # but there could be non-paragraph nodes after, so we check if there's no next sibling + # that is a paragraph + sibling = node.next? + while sibling + return false if sibling.type.paragraph? + sibling = sibling.next? + end + true + end + + # Encode a string for use in HTML id attributes + # This matches the old URI.encode behavior for footnote IDs + private def encode_id_component(string : String) : String + String.build do |io| + URI.encode(string, io) do |byte| + # Allow unreserved chars plus some additional chars that are safe in HTML IDs + URI.unreserved?(byte) || ['/', '(', ')', '*', '!', '$', '\'', ',', ';', ':', '@', '&', '=', '+'].includes?(byte.chr) + end + end + end + + # Append footnote back-reference links + # @param in_paragraph If true, adds a space before the backrefs + private def append_footnote_backrefs(footnote_def : Node, in_paragraph : Bool = true) : Nil + encoded_title = encode_id_component(footnote_def.data["title"].to_s) + footnote_number = footnote_def.data["number"].as(Int32) + ref_count = footnote_def.data["ref_count"].as(Int32) + + # Add a space before the backrefs if we're inside a paragraph + literal " " if in_paragraph + + # Generate backref(s) - multiple if referenced multiple times + (1..ref_count).each do |ref_index| + backref_id = ref_index == 1 ? "fnref-#{encoded_title}" : "fnref-#{encoded_title}-#{ref_index}" + backref_idx = ref_index == 1 ? footnote_number.to_s : "#{footnote_number}-#{ref_index}" + + tag("a", { + "href" => "##{backref_id}", + "class" => "footnote-backref", + "data-footnote-backref" => nil, + "data-footnote-backref-idx" => backref_idx, + "aria-label" => "Back to reference #{backref_idx}", + }) + literal "↩" + # For 2nd and later backrefs, add a superscript with the index + if ref_index > 1 + tag("sup", {"class" => "footnote-ref"}) + literal ref_index.to_s + tag("sup", end_tag: true) + end + tag("a", end_tag: true) + literal " " if ref_index < ref_count + end + end + def emphasis(node : Node, entering : Bool) : Nil if entering node.data["strong_stack"] = @strong_stack @@ -300,6 +365,79 @@ module Markd output(node.text) end + def footnote(node : Node, entering : Bool) : Nil + # Spec says `[^1]` should generate: + # 1 + # For multiple references to the same footnote: + # First: id="fnref-label", Second: id="fnref-label-2", etc. + if entering + tag("sup", { + "class" => "footnote-ref", + }) + ref_index = node.data["ref_index"].as(Int32) + encoded_title = encode_id_component(node.data["title"].to_s) + id = ref_index == 1 ? "fnref-#{encoded_title}" : "fnref-#{encoded_title}-#{ref_index}" + tag("a", { + "href" => "#fn-#{encoded_title}", + "id" => id, + "data-footnote-ref" => nil, + }) + # GFM spec says to output the number of the footnote + output node.data["number"].to_s + tag("a", end_tag: true) + tag("sup", end_tag: true) + end + end + + def footnote_definition(node : Node, entering : Bool) : Nil + # A footnote definition by spec should render something like: + #
  • + #

    The actual content of the footnote + # ↩

    + #
    • + # For multiple references, multiple backref links are generated: + # ↩ + # ↩2 + if entering + if !node.prev.type.footnote_definition? + newline + tag("section", {"class" => "footnotes", "data-footnotes" => nil}) + newline + tag("ol") + end + newline + tag("li", { + "id" => "fn-#{encode_id_component(node.data["title"].to_s)}", + }) + newline + else + # If there's no paragraph child, output backrefs here + # (e.g., for footnote definitions that only contain code blocks) + unless has_paragraph_child?(node) + append_footnote_backrefs(node, in_paragraph: false) + newline + end + tag("li", end_tag: true) + if node == node.parent.last_child + newline + tag("ol", end_tag: true) + newline + tag("section", end_tag: true) + newline + end + end + end + + # Check if a node has a paragraph child + private def has_paragraph_child?(node : Node) : Bool + child = node.first_child? + while child + return true if child.type.paragraph? + child = child.next? + end + false + end + private def tag(name : String, attrs = nil, self_closing = false, end_tag = false) return if @disable_tag > 0 @@ -307,7 +445,11 @@ module Markd @output_io << "/" if end_tag @output_io << name attrs.try &.each do |key, value| - @output_io << ' ' << key << '=' << '"' << value << '"' + if value.nil? + @output_io << ' ' << key + else + @output_io << ' ' << key << '=' << '"' << value << '"' + end end @output_io << " /" if self_closing diff --git a/src/markd/rule.cr b/src/markd/rule.cr index 5f57db0..1efaaa2 100644 --- a/src/markd/rule.cr +++ b/src/markd/rule.cr @@ -15,7 +15,7 @@ module Markd ATTRIBUTE_VALUE_SPEC_STRING = "(?:" + "\\s*=" + "\\s*" + ATTRIBUTE_VALUE_STRING + ")" ATTRIBUTE = "(?:" + "\\s+" + ATTRIBUTE_NAME_STRING + ATTRIBUTE_VALUE_SPEC_STRING + "?)" - MAYBE_SPECIAL = {'#', '`', '~', '*', '+', '_', '=', '<', '>', '-', '|'} + MAYBE_SPECIAL = {'#', '`', '~', '*', '+', '_', '=', '<', '>', '-', '|', '['} THEMATIC_BREAK = /^(?:(?:\*[ \t]*){3,}|(?:_[ \t]*){3,}|(?:-[ \t]*){3,})[ \t]*$/ ESCAPABLE = /^#{ESCAPABLE_STRING}/ @@ -107,6 +107,8 @@ module Markd ADMONITION_START = /^> \[!((?:NOTE|TIP|IMPORTANT|CAUTION|WARNING)+)](\s*.*)?$/ + FOOTNOTE_DEFINITION_START = /^\[\^[^\]]+\]:(\s|$)/ + # Match Value # # - None: no match diff --git a/src/markd/rules/footnote_definition.cr b/src/markd/rules/footnote_definition.cr new file mode 100644 index 0000000..db0ada4 --- /dev/null +++ b/src/markd/rules/footnote_definition.cr @@ -0,0 +1,64 @@ +module Markd::Rule + struct FootnoteDefinition + include Rule + + def match(parser : Parser, container : Node) : MatchValue + if match?(parser) && parser.gfm? + parser.close_unmatched_blocks + parser.add_child(Node::Type::FootnoteDefinition, 0) + MatchValue::Leaf + else + MatchValue::None + end + end + + # Footnote definitions continue as long as lines are indented (4+ spaces) + # Blank lines are allowed (they separate paragraphs inside footnotes) + def continue(parser : Parser, container : Node) : ContinueStatus + # Blank lines are allowed (separate paragraphs) + return ContinueStatus::Continue if parser.blank + + # Indented content continues the definition + return ContinueStatus::Continue if parser.indent >= Rule::CODE_INDENT + + ContinueStatus::Stop + end + + def token(parser : Parser, container : Node) : Nil + lines = container.text.split "\n" + lines.each_with_index do |line, i| + if i == 0 + # First line has the footnote definition label removed and + # leading spaces removed. + container.data["title"] = line.split("]:")[0].lstrip("[^") + lines[i] = line.split("]:", 2)[1].lstrip + elsif line == "" + # Empty lines go as-is + lines[i] = "" + else + # Subsequent lines have the leading 4 spaces removed + lines[i] = line[4..-1] + end + end + # Store the processed text for later block parsing + container.text = lines.join("\n") + end + + def can_contain?(type) + true + end + + # Footnote definitions can be multiline + def accepts_lines? : Bool + true + end + + # Match only lines that look like the first line of a footnote definition: + # Start with [^something]: + + private def match?(parser) + !parser.indented && \ + parser.line.match(FOOTNOTE_DEFINITION_START) + end + end +end