From 1fb5f15eadce697ead8f131864ae56edf22cb7fb Mon Sep 17 00:00:00 2001
From: Yessiest <yessiest@memeware.net>
Date: Fri, 7 Mar 2025 21:29:24 +0000
Subject: [PATCH] HTML renderer fixes, additional compliance
---
lib/mmmd.rb | 14 +++
lib/mmmd/blankshell.rb | 35 ++++++--
lib/mmmd/renderers.rb | 8 +-
lib/mmmd/renderers/html.rb | 149 ++++++++++++++++++++++++++------
lib/mmmd/renderers/plainterm.rb | 3 +
lib/mmmd/util.rb | 18 +++-
lib/rubymark | 1 -
7 files changed, 188 insertions(+), 40 deletions(-)
create mode 100644 lib/mmmd.rb
delete mode 120000 lib/rubymark
diff --git a/lib/mmmd.rb b/lib/mmmd.rb
new file mode 100644
index 0000000..8eada91
--- /dev/null
+++ b/lib/mmmd.rb
@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+
+require_relative 'mmmd/blankshell'
+require_relative 'mmmd/renderers'
+
+# Extensible, multi-format markdown processor
+module MMMD
+ # Parse a Markdown document into a DOM form
+ # @param doc [String]
+ # @return [::PointBlank::DOM::Document]
+ def self.parse(doc)
+ ::PointBlank::DOM::Document.parse(doc)
+ end
+end
diff --git a/lib/mmmd/blankshell.rb b/lib/mmmd/blankshell.rb
index c473e35..01a57f6 100644
--- a/lib/mmmd/blankshell.rb
+++ b/lib/mmmd/blankshell.rb
@@ -1,6 +1,7 @@
# frozen_string_literal: true
require_relative 'util'
+require 'uri'
# Modular, extensible approach to parsing markdown as close as
# it gets to CommonMark spec (as of version 0.31.2).
@@ -75,13 +76,13 @@ module PointBlank
def read_destination(text)
if (result = text.match(/\A<.*?(?<![^\\]\\)>/m)) &&
!result[0][1..].match?(/(?<![^\\]\\)</)
- [result[0].gsub(/\\(?=[><])/, '')[1..-2],
+ [process_destination(result[0].gsub(/\\(?=[><])/, '')[1..-2]),
text.delete_prefix(result[0]).lstrip]
elsif (result = text.match(/\A\S+/)) &&
!result[0].start_with?('<') &&
result &&
balanced?(result[0])
- [result[0],
+ [process_destination(result[0]),
text.delete_prefix(result[0]).lstrip]
else
[nil, text]
@@ -95,15 +96,15 @@ module PointBlank
def read_title(text)
if text.start_with?("'") &&
(result = text.match(/\A'.*?(?<!\\)'/m))
- [result[0][1..-2],
+ [process_title(result[0][1..-2]),
text.delete_prefix(result[0]).lstrip]
elsif text.start_with?('"') &&
(result = text.match(/\A".*?(?<!\\)"/m))
- [result[0][1..-2],
+ [process_title(result[0][1..-2]),
text.delete_prefix(result[0]).lstrip]
elsif text.start_with?('(') &&
(result = find_balanced_end(text))
- [text[1..(result - 1)],
+ [process_title(text[1..(result - 1)]),
text.delete_prefix(text[..result]).lstrip]
else
[nil, text]
@@ -125,7 +126,7 @@ module PointBlank
return [nil, text] unless destination
title, remaining = read_title(remaining)
- properties[:destination] = destination
+ properties[:uri] = destination
properties[:title] = title
close_bracket = true
end
@@ -169,6 +170,28 @@ module PointBlank
end
nil
end
+
+ # Process destination string
+ # @param string [String]
+ # @return [String]
+ def process_destination(string)
+ string = string.gsub(/\\([!"\#$%&'()*+,\-.\/:;<=>?@\[\\\]\^_`{|}~])/,
+ '\\1')
+ string = string.gsub("\n", " ")
+ URI.encode_uri_component(
+ MMMD::EntityUtils.decode_entities(string)
+ )
+ end
+
+ # Process title string
+ # @param string [String]
+ # @return [String]
+ def process_title(string)
+ string = string.gsub(/\\([!"\#$%&'()*+,\-.\/:;<=>?@\[\\\]\^_`{|}~])/,
+ '\\1')
+ string = string.gsub("\n", " ")
+ MMMD::EntityUtils.decode_entities(string)
+ end
end
class LineScanner
diff --git a/lib/mmmd/renderers.rb b/lib/mmmd/renderers.rb
index 7cbe4b5..b6a8b9d 100644
--- a/lib/mmmd/renderers.rb
+++ b/lib/mmmd/renderers.rb
@@ -1,9 +1,11 @@
# frozen_string_literal: true
-module RBMark
+$LOAD_PATH.append(__dir__)
+
+module MMMD
# Renderers from Markdown to expected output format
module Renderers
+ autoload :HTML, 'renderers/html'
+ autoload :PlainTerm, 'renderers/plainterm'
end
end
-
-require_relative 'renderers/html'
diff --git a/lib/mmmd/renderers/html.rb b/lib/mmmd/renderers/html.rb
index 4f771c9..9aa5f49 100644
--- a/lib/mmmd/renderers/html.rb
+++ b/lib/mmmd/renderers/html.rb
@@ -7,7 +7,8 @@ module MMMD
module HTMLConstants
ELEMENT_MAP = {
"PointBlank::DOM::InlinePre" => {
- tag: "pre"
+ tag: "code",
+ style: "white-space: pre;"
},
"PointBlank::DOM::InlineBreak" => {
tag: "br"
@@ -26,12 +27,14 @@ module MMMD
tag: "s"
},
"PointBlank::DOM::InlineLink" => {
- tag: "link",
+ tag: "a",
href: true
},
"PointBlank::DOM::InlineImage" => {
tag: "img",
- src: true
+ src: true,
+ inline: true,
+ alt: true
},
"PointBlank::DOM::ULBlock" => {
tag: "ul"
@@ -88,7 +91,8 @@ module MMMD
tag: "blockquote"
},
"PointBlank::DOM::HorizontalRule" => {
- tag: "hr"
+ tag: "hr",
+ inline: true
},
"PointBlank::DOM::Text" => {
sanitize: true
@@ -128,41 +132,80 @@ module MMMD
def initialize(dom, options)
@document = dom
@options = options
+ @options["linewrap"] ||= 80
@options["init_level"] ||= 2
@options["indent"] ||= 2
mapmanager = HTMLConstants::MapManager.new(options)
@mapping = mapmanager.mapping
+ return unless @options["nowrap"]
+
+ @options["init_level"] = 0
+ @mapping.delete("PointBlank::DOM::Document")
end
# Render document to HTML
def render
text = _render(@document, @options, level: @options["init_level"])
@options["init_level"].times { text = indent(text) }
- [
- preambule,
- text,
- postambule
- ].join("\n")
+ if @options["nowrap"]
+ text
+ else
+ [
+ preambule,
+ remove_pre_spaces(text),
+ postambule
+ ].join("\n")
+ end
end
private
+ # Find and remove extra spaces inbetween preformatted text
+ # @param string [String]
+ # @return [String]
+ def remove_pre_spaces(string)
+ output = []
+ buffer = []
+ open = nil
+ string.lines.each do |line|
+ opentoken = line.match?(/<pre>/)
+ closetoken = line.match?(/<\/pre>/)
+ if closetoken
+ open = false
+ buffer = strip_leading_spaces_in_buffer(buffer)
+ output.append(*buffer)
+ buffer = []
+ end
+ (open ? buffer : output).append(line)
+ open = true if opentoken && !closetoken
+ end
+ output.append(*buffer) unless buffer.empty?
+ output.join('')
+ end
+
+ # Strip leading spaces in the buffer
+ # @param lines [Array<String>]
+ # @return [Array<String>]
+ def strip_leading_spaces_in_buffer(buffer)
+ minprefix = buffer.map { |x| x.match(/^ */)[0] }
+ .min_by(&:length)
+ buffer.map do |line|
+ line.delete_prefix(minprefix)
+ end
+ end
+
# Word wrapping algorithm
# @param text [String]
# @param width [Integer]
# @return [String]
def wordwrap(text, width)
- words = text.split(/( +)/)
+ words = text.split(/( +|<[^>]+>)/)
output = []
line = ""
length = 0
until words.empty?
word = words.shift
wordlength = word.length
- if wordlength > width
- words.prepend(word[width..])
- word = word[..width - 1]
- end
if length + wordlength + 1 > width
output.append(line.lstrip)
line = word
@@ -176,7 +219,7 @@ module MMMD
output.join("\n")
end
- def _render(element, options, inline: false, level: 0)
+ def _render(element, options, inline: false, level: 0, literaltext: false)
modeswitch = element.is_a?(::PointBlank::DOM::LeafBlock) ||
element.is_a?(::PointBlank::DOM::Paragraph)
inline ||= modeswitch
@@ -184,24 +227,39 @@ module MMMD
text = if element.children.empty?
element.content
else
+ literal = @mapping[element.class.name][:inline] || literaltext
element.children.map do |child|
_render(child, options, inline: inline,
- level: level)
+ level: level,
+ literaltext: literal)
end.join(inline ? '' : "\n")
end
run_filters(text, element, level: level,
inline: inline,
- modeswitch: modeswitch)
+ modeswitch: modeswitch,
+ literaltext: literaltext)
end
- def run_filters(text, element, level:, inline:, modeswitch:)
+ def run_filters(text, element, level:, inline:, modeswitch:,
+ literaltext:)
element_style = @mapping[element.class.name]
- hsize = 80 - (level * @options["indent"])
+ return text unless element_style
+ return text if literaltext
+
+ hsize = @options["linewrap"] - (level * @options["indent"])
text = wordwrap(text, hsize) if modeswitch
if element_style[:sanitize]
text = MMMD::EntityUtils.encode_entities(text)
end
- opentag, closetag = construct_tags(element_style)
+ if element_style[:inline]
+ innerclose(element, element_style, text)
+ else
+ openclose(text, element, element_style, inline)
+ end
+ end
+
+ def openclose(text, element, element_style, inline)
+ opentag, closetag = construct_tags(element_style, element)
if inline
opentag + text + closetag
else
@@ -211,21 +269,54 @@ module MMMD
end
end
- def construct_tags(style)
+ def innerclose(element, style, text)
+ props = element.properties
+ tag = "<#{style[:tag]}"
+ tag += " style=#{style[:style].inspect}" if style[:style]
+ tag += " href=#{read_link(element)}" if style[:href]
+ tag += " alt=#{text.inspect}" if style[:alt]
+ tag += " src=#{read_link(element)}" if style[:src]
+ tag += " title=#{read_title(element)}" if style[:title] && props[:title]
+ tag += ">"
+ if style[:outer]
+ outeropen, outerclose = construct_tags(style[:outer], element)
+ tag = outeropen + tag + outerclose
+ end
+ tag
+ end
+
+ def construct_tags(style, element)
return ["", ""] unless style && style[:tag]
+ props = element.properties
opentag = "<#{style[:tag]}"
closetag = "</#{style[:tag]}>"
- opentag += " style=#{style[:style].dump}" if style["style"]
+ opentag += " style=#{style[:style].inspect}" if style[:style]
+ opentag += " href=#{read_link(element)}" if style[:href]
+ opentag += " src=#{read_link(element)}" if style[:src]
+ opentag += " title=#{read_title(element)}" if style[:title] &&
+ props[:title]
opentag += ">"
if style[:outer]
- outeropen, outerclose = construct_tags(style[:outer])
+ outeropen, outerclose = construct_tags(style[:outer], element)
opentag = outeropen + opentag
closetag += outerclose
end
[opentag, closetag]
end
+ def read_title(element)
+ title = element.properties[:title]
+ title = MMMD::EntityUtils.encode_entities(title)
+ title.dump
+ end
+
+ def read_link(element)
+ link = element.properties[:uri]
+ link = MMMD::EntityUtils.encode_entities(link)
+ link.dump
+ end
+
def indent(text)
text.lines.map do |line|
"#{' ' * @options["indent"]}#{line}"
@@ -233,13 +324,17 @@ module MMMD
end
def preambule
+ head = @options['head']
+ headinfo = "#{indent(<<~HEAD.rstrip)}\n " if head
+ <head>
+ #{head.is_a?(Array) ? head.join("\n") : head}
+ </head>
+ HEAD
+ headinfo ||= " "
@options['preambule'] or <<~TEXT.rstrip
<!DOCTYPE HTML>
<html>
- <head>
- #{@options['head']}
- </head>
- <body>
+ #{headinfo}<body>
TEXT
end
diff --git a/lib/mmmd/renderers/plainterm.rb b/lib/mmmd/renderers/plainterm.rb
index 27dec04..8ad1f48 100644
--- a/lib/mmmd/renderers/plainterm.rb
+++ b/lib/mmmd/renderers/plainterm.rb
@@ -312,6 +312,9 @@ module MMMD
"PointBlank::DOM::QuoteBlock" => {
leftline: true,
increase_level: true
+ },
+ "PointBlank::DOM::HorizontalRule" => {
+ underline_full_block: true
}
}.freeze
diff --git a/lib/mmmd/util.rb b/lib/mmmd/util.rb
index 5b15cf0..f9efe09 100644
--- a/lib/mmmd/util.rb
+++ b/lib/mmmd/util.rb
@@ -22,6 +22,21 @@ module MMMD
end
end
+ # Encode unsafe html entities in string (ASCII-compatible)
+ # @param string [String]
+ # @return [String]
+ # @sg-ignore
+ def self.encode_entities_ascii(string)
+ string.gsub("&", "&")
+ .gsub("<", "<")
+ .gsub(">", ">")
+ .gsub('"', """)
+ .gsub("'", "'")
+ .gsub(/[^\x00-\x7F]/) do |match|
+ "&#x#{match.codepoints[0]};"
+ end
+ end
+
# Encode unsafe html entities in string
# @param string [String]
# @return [String]
@@ -32,9 +47,6 @@ module MMMD
.gsub(">", ">")
.gsub('"', """)
.gsub("'", "'")
- .gsub(/[^\x00-\x7F]/) do |match|
- "&#x#{match.codepoints[0]};"
- end
end
end
end
diff --git a/lib/rubymark b/lib/rubymark
deleted file mode 120000
index 5f5df8f..0000000
--- a/lib/rubymark
+++ /dev/null
@@ -1 +0,0 @@
-mmmd
\ No newline at end of file