discourse/lib/html_to_markdown.rb
Régis Hanol cd93d1b5f7
FEATURE: new 'trim_incoming_emails' site setting (#12874)
This setting allows admin to de/activate automatic trimming of incoming email.
There are instances where it does wonders in trimming all the garbage content and other
instances where it's so bad that it trims the most important part of the email.

FIX: don't remove hidden content using the style attribute when converting HTML to Markdown.
The regexp used was doing more harm than good. It was way too broad.

FIX: properly elide signatures from emails sent with Front App.
This is fairly safe as Front App nicely identifies signatures in the HTML part.
2021-04-28 17:08:48 +02:00

363 lines
8.8 KiB
Ruby

# frozen_string_literal: true
require "nokogumbo"
require "securerandom"
class HtmlToMarkdown
def initialize(html, opts = {})
@opts = opts
# we're only interested in <body>
@doc = Nokogiri::HTML5(html).at("body")
remove_not_allowed!(@doc)
remove_hidden!(@doc)
hoist_line_breaks!(@doc)
remove_whitespaces!(@doc)
end
def to_markdown
traverse(@doc)
.gsub(/\n{2,}/, "\n\n")
.strip
end
private
def remove_not_allowed!(doc)
allowed = Set.new
HtmlToMarkdown.private_instance_methods.each do |m|
if tag = m.to_s[/^visit_(.+)/, 1]
allowed << tag
end
end
@doc.traverse { |node| node.remove if !allowed.include?(node.name) }
end
def remove_hidden!(doc)
@doc.css("[hidden]").remove
@doc.css("img[width]").each { |n| n.remove if n["width"].to_i <= 0 }
@doc.css("img[height]").each { |n| n.remove if n["height"].to_i <= 0 }
end
# When there's a <br> inside an inline element, split the inline element around the <br>
def hoist_line_breaks!(doc)
klass = "_" + SecureRandom.hex
doc.css("br").each { |br| br.add_class(klass) }
loop do
changed = false
doc.css("br.#{klass}").each do |br|
parent = br.parent
if parent.description.block?
br.remove_class(klass)
else
before, after = parent.children.slice_when { |n| n == br }.to_a
if before.size > 1
b = Nokogiri::XML::Node.new(parent.name, doc)
before[0...-1].each { |c| b.add_child(c) }
parent.previous = b if b.inner_html.present?
end
if after.present?
a = Nokogiri::XML::Node.new(parent.name, doc)
after.each { |c| a.add_child(c) }
parent.next = a if a.inner_html.present?
end
parent.replace(br)
changed = true
end
end
break if !changed
end
end
# Removes most of the unnecessary white spaces for better markdown conversion
# Loosely based on the CSS' White Space Processing Rules (https://www.w3.org/TR/css-text-3/#white-space-rules)
def remove_whitespaces!(node)
return true if "pre" == node.name
node.children.chunk { |n| is_inline?(n) }.each do |inline, nodes|
if inline
collapse_spaces!(nodes) && remove_trailing_space!(nodes)
else
nodes.each { |n| remove_whitespaces!(n) }
end
end
end
def is_inline?(node)
node.text? || ("br" != node.name && node.description&.inline? && node.children.all? { |n| is_inline?(n) })
end
def collapse_spaces!(nodes, was_space = true)
nodes.each do |node|
if node.text?
text = String.new
node.text.chars.each do |c|
if c[/[[:space:]]/]
text << " " if !was_space
was_space = true
else
text << c
was_space = false
end
end
node.content = text
else
node.children.each { |n| was_space = collapse_spaces!([n], was_space) }
end
end
was_space
end
def remove_trailing_space!(nodes)
last = nodes[-1]
if last.text?
last.content = last.content[0...-1] if last.content[-1] == " "
elsif last.children.present?
remove_trailing_space!(last.children)
end
end
def traverse(node)
node.children.map { |n| visit(n) }.join
end
def visit(node)
visitor = "visit_#{node.name}"
send(visitor, node) if respond_to?(visitor, true)
end
ALLOWED_IMG_SRCS ||= %w{http:// https:// www.}
def allowed_hrefs
@allowed_hrefs ||= begin
hrefs = SiteSetting.allowed_href_schemes.split("|").map { |scheme| "#{scheme}:" }.to_set
ALLOWED_IMG_SRCS.each { |src| hrefs << src }
hrefs << "mailto:"
hrefs.to_a
end
end
def visit_a(node)
if node["href"].present? && node["href"].starts_with?(*allowed_hrefs)
"[#{traverse(node)}](#{node["href"]})"
else
traverse(node)
end
end
def visit_img(node)
return if node["src"].blank?
if @opts[:keep_img_tags]
node.to_html
elsif @opts[:keep_cid_imgs] && node["src"].starts_with?("cid:")
node.to_html
elsif node["src"].starts_with?(*ALLOWED_IMG_SRCS)
title = node["alt"].presence || node["title"].presence
width = node["width"].to_i
height = node["height"].to_i
dimensions = "|#{width}x#{height}" if width > 0 && height > 0
"![#{title}#{dimensions}](#{node["src"]})"
end
end
ALLOWED ||= %w{kbd del ins small big sub sup dl dd dt mark}
ALLOWED.each do |tag|
define_method("visit_#{tag}") do |node|
"<#{tag}>#{traverse(node)}</#{tag}>"
end
end
def visit_blockquote(node)
text = traverse(node)
text.strip!
text.gsub!(/\n{2,}/, "\n\n")
text.gsub!(/^/, "> ")
"\n\n#{text}\n\n"
end
BLOCKS ||= %w{div tr}
BLOCKS.each do |tag|
define_method("visit_#{tag}") do |node|
prefix = node.previous_element&.description&.block? ? "" : "\n"
"#{prefix}#{traverse(node)}\n"
end
end
def visit_p(node)
"\n\n#{traverse(node)}\n\n"
end
TRAVERSABLES ||= %w{aside font span thead tbody tfooter u}
TRAVERSABLES.each do |tag|
define_method("visit_#{tag}") do |node|
traverse(node)
end
end
def visit_tt(node)
"`#{traverse(node)}`"
end
def visit_code(node)
node.ancestors("pre").present? ? traverse(node) : visit_tt(node)
end
def visit_pre(node)
text = traverse(node)
fence = text["`"] ? "~~~" : "```"
code = node.at("code")
code_class = code ? code["class"] : ""
lang = code_class ? code_class[/lang-(\w+)/, 1] : ""
"\n\n#{fence}#{lang}\n#{traverse(node)}\n#{fence}\n\n"
end
def visit_br(node)
"\n"
end
def visit_hr(node)
"\n\n---\n\n"
end
def visit_abbr(node)
title = node["title"].presence
title_attr = title ? %[ title="#{title}"] : ""
"<abbr#{title_attr}>#{traverse(node)}</abbr>"
end
def visit_acronym(node)
visit_abbr(node)
end
(1..6).each do |n|
define_method("visit_h#{n}") do |node|
"#{"#" * n} #{traverse(node)}"
end
end
CELLS ||= %w{th td}
CELLS.each do |tag|
define_method("visit_#{tag}") do |node|
"#{traverse(node)} "
end
end
def visit_table(node)
if rows = extract_rows(node)
headers = rows[0].css("td, th")
text = "| " + headers.map { |td| traverse(td).gsub(/\n/, "<br>") }.join(" | ") + " |\n"
text << "| " + (["-"] * headers.size).join(" | ") + " |\n"
rows[1..-1].each do |row|
text << "| " + row.css("td").map { |td| traverse(td).gsub(/\n/, "<br>") }.join(" | ") + " |\n"
end
"\n\n#{text}\n\n"
else
traverse(node)
end
end
def extract_rows(table)
return if table.ancestors("table").present?
return if (rows = table.css("tr")).empty?
headers_count = rows[0].css("td, th").size
return if rows[1..-1].any? { |row| row.css("td").size != headers_count }
rows
end
LISTS ||= %w{ul ol}
LISTS.each do |tag|
define_method("visit_#{tag}") do |node|
prefix = node.previous_element&.description&.block? ? "" : "\n"
suffix = node.ancestors("ul, ol, li").size > 0 ? "" : "\n"
"#{prefix}#{traverse(node)}#{suffix}"
end
end
def visit_li(node)
text = traverse(node)
lists = node.ancestors("ul, ol")
marker = "ol" == lists[0]&.name ? "1. " : "- "
indent = (" " * marker.size) * [1, lists.size].max
suffix = node == node.parent.elements[-1] ? "" : "\n"
text.gsub!(/\n{2,}/, "\n\n")
text.gsub!(/^(?!\s*$)/, indent)
text.lstrip!
"#{marker}#{text}#{suffix}"
end
EMPHASES ||= %w{i em}
EMPHASES.each do |tag|
define_method("visit_#{tag}") do |node|
text = traverse(node)
return "" if text.empty?
return " " if text.blank?
return "<#{tag}>#{text}</#{tag}>" if text["\n"] || (text["*"] && text["_"])
prefix = text[0][" "]
suffix = text[-1][" "] if text.size > 1
wrap = text["*"] ? "_" : "*"
"#{prefix}#{wrap}#{text.strip}#{wrap}#{suffix}"
end
end
STRONGS ||= %w{b strong}
STRONGS.each do |tag|
define_method("visit_#{tag}") do |node|
text = traverse(node)
return "" if text.empty?
return " " if text.blank?
return "<#{tag}>#{text}</#{tag}>" if text["\n"] || (text["*"] && text["_"])
prefix = text[0][" "]
suffix = text[-1][" "] if text.size > 1
wrap = text["*"] ? "__" : "**"
"#{prefix}#{wrap}#{text.strip}#{wrap}#{suffix}"
end
end
STRIKES ||= %w{s strike}
STRIKES.each do |tag|
define_method("visit_#{tag}") do |node|
text = traverse(node)
return "" if text.empty?
return " " if text.blank?
return "<#{tag}>#{text}</#{tag}>" if text["\n"] || text["~~"]
prefix = text[0][" "]
suffix = text[-1][" "] if text.size > 1
"#{prefix}~~#{text.strip}~~#{suffix}"
end
end
def visit_text(node)
node.text
end
end