discourse/lib/pretty_text.rb

299 lines
9.0 KiB
Ruby
Raw Normal View History

2013-02-05 14:16:51 -05:00
require 'v8'
require 'nokogiri'
2013-05-27 19:48:47 -04:00
require_dependency 'excerpt_parser'
2013-02-05 14:16:51 -05:00
module PrettyText
def self.whitelist
{
elements: %w[
2013-02-05 14:16:51 -05:00
a abbr aside b bdo blockquote br caption cite code col colgroup dd div del dfn dl
dt em hr figcaption figure h1 h2 h3 h4 h5 h6 hgroup i img ins kbd li mark
ol p pre q rp rt ruby s samp small span strike strong sub sup table tbody td
tfoot th thead time tr u ul var wbr
],
attributes: {
2013-02-05 14:16:51 -05:00
:all => ['dir', 'lang', 'title', 'class'],
'aside' => ['data-post', 'data-full', 'data-topic'],
'a' => ['href'],
'blockquote' => ['cite'],
'col' => ['span', 'width'],
'colgroup' => ['span', 'width'],
'del' => ['cite', 'datetime'],
'img' => ['align', 'alt', 'height', 'src', 'width'],
'ins' => ['cite', 'datetime'],
'ol' => ['start', 'reversed', 'type'],
'q' => ['cite'],
'span' => ['style'],
'table' => ['summary', 'width', 'style', 'cellpadding', 'cellspacing'],
'td' => ['abbr', 'axis', 'colspan', 'rowspan', 'width', 'style'],
'th' => ['abbr', 'axis', 'colspan', 'rowspan', 'scope', 'width', 'style'],
'time' => ['datetime', 'pubdate'],
'ul' => ['type']
},
protocols: {
2013-02-05 14:16:51 -05:00
'a' => {'href' => ['ftp', 'http', 'https', 'mailto', :relative]},
'blockquote' => {'cite' => ['http', 'https', :relative]},
'del' => {'cite' => ['http', 'https', :relative]},
'img' => {'src' => ['http', 'https', :relative]},
'ins' => {'cite' => ['http', 'https', :relative]},
'q' => {'cite' => ['http', 'https', :relative]}
}
2013-02-25 11:42:20 -05:00
}
2013-02-05 14:16:51 -05:00
end
class Helpers
def t(key, opts)
str = I18n.t("js." + key)
if opts
# TODO: server localisation has no parity with client
# should be fixed
opts.each do |k,v|
str.gsub!("{{#{k}}}", v)
end
end
str
end
2013-02-25 11:42:20 -05:00
# function here are available to v8
2013-02-05 14:16:51 -05:00
def avatar_template(username)
return "" unless username
user = User.where(username_lower: username.downcase).first
2013-08-13 16:08:29 -04:00
if user.present?
2013-02-05 14:16:51 -05:00
user.avatar_template
end
end
def is_username_valid(username)
return false unless username
username = username.downcase
return User.exec_sql('select 1 from users where username_lower = ?', username).values.length == 1
end
end
@mutex = Mutex.new
@ctx_init = Mutex.new
2013-02-05 14:16:51 -05:00
def self.mention_matcher
Regexp.new("(\@[a-zA-Z0-9_]{#{User.username_length.begin},#{User.username_length.end}})")
2013-02-25 11:42:20 -05:00
end
2013-02-05 14:16:51 -05:00
def self.app_root
Rails.root
end
def self.create_new_context
ctx = V8::Context.new
2013-02-25 11:42:20 -05:00
ctx["helpers"] = Helpers.new
2013-02-05 14:16:51 -05:00
ctx_load(ctx,
"app/assets/javascripts/external/md5.js",
"app/assets/javascripts/external/lodash.js",
2013-05-27 19:48:47 -04:00
"app/assets/javascripts/external/Markdown.Converter.js",
"lib/headless-ember.js",
"app/assets/javascripts/external/rsvp.js",
Rails.configuration.ember.handlebars_location)
2013-02-05 14:16:51 -05:00
ctx.eval("var Discourse = {}; Discourse.SiteSettings = #{SiteSetting.client_settings_json};")
ctx.eval("var window = {}; window.devicePixelRatio = 2;") # hack to make code think stuff is retina
ctx.eval("var I18n = {}; I18n.t = function(a,b){ return helpers.t(a,b); }");
2013-02-05 14:16:51 -05:00
decorate_context(ctx)
ctx_load(ctx,
2013-08-27 12:24:17 -04:00
"app/assets/javascripts/external/better_markdown.js",
2013-08-08 18:14:12 -04:00
"app/assets/javascripts/discourse/dialects/dialect.js",
2013-05-27 19:48:47 -04:00
"app/assets/javascripts/discourse/components/utilities.js",
"app/assets/javascripts/discourse/components/markdown.js")
2013-02-05 14:16:51 -05:00
2013-08-08 18:14:12 -04:00
Dir["#{Rails.root}/app/assets/javascripts/discourse/dialects/**.js"].each do |dialect|
unless dialect =~ /\/dialect\.js$/
ctx.load(dialect)
end
end
2013-02-05 14:16:51 -05:00
# Load server side javascripts
if DiscoursePluginRegistry.server_side_javascripts.present?
DiscoursePluginRegistry.server_side_javascripts.each do |ssjs|
ctx.load(ssjs)
2013-02-05 14:16:51 -05:00
end
end
ctx['quoteTemplate'] = File.open(app_root + 'app/assets/javascripts/discourse/templates/quote.js.shbrs') {|f| f.read}
ctx['quoteEmailTemplate'] = File.open(app_root + 'lib/assets/quote_email.js.shbrs') {|f| f.read}
ctx.eval("HANDLEBARS_TEMPLATES = {
2013-02-05 14:16:51 -05:00
'quote': Handlebars.compile(quoteTemplate),
'quote_email': Handlebars.compile(quoteEmailTemplate),
};")
ctx
end
def self.v8
return @ctx if @ctx
# ensure we only init one of these
@ctx_init.synchronize do
return @ctx if @ctx
@ctx = create_new_context
end
2013-02-05 14:16:51 -05:00
@ctx
end
def self.decorate_context(context)
context.eval("Discourse.SiteSettings = #{SiteSetting.client_settings_json};")
context.eval("Discourse.CDN = '#{Rails.configuration.action_controller.asset_host}';")
context.eval("Discourse.BaseUrl = 'http://#{RailsMultisite::ConnectionManagement.current_hostname}';")
context.eval("Discourse.getURL = function(url) {return '#{Discourse::base_uri}' + url};")
end
2013-02-05 14:16:51 -05:00
def self.markdown(text, opts=nil)
# we use the exact same markdown converter as the client
2013-02-25 11:42:20 -05:00
# TODO: use the same extensions on both client and server (in particular the template for mentions)
2013-02-05 14:16:51 -05:00
baked = nil
2013-02-25 11:42:20 -05:00
@mutex.synchronize do
context = v8
2013-02-05 14:16:51 -05:00
# we need to do this to work in a multi site environment, many sites, many settings
decorate_context(context)
context['opts'] = opts || {}
context['raw'] = text
context.eval('opts["mentionLookup"] = function(u){return helpers.is_username_valid(u);}')
context.eval('opts["lookupAvatar"] = function(p){return Discourse.Utilities.avatarImg({size: "tiny", avatarTemplate: helpers.avatar_template(p)});}')
baked = context.eval('Discourse.Markdown.markdownConverter(opts).makeHtml(raw)')
2013-02-05 14:16:51 -05:00
end
# we need some minimal server side stuff, apply CDN and TODO filter disallowed markup
2013-02-25 11:42:20 -05:00
baked = apply_cdn(baked, Rails.configuration.action_controller.asset_host)
2013-02-05 14:16:51 -05:00
baked
end
# leaving this here, cause it invokes v8, don't want to implement twice
2013-08-13 16:08:29 -04:00
def self.avatar_img(avatar_template, size)
2013-02-05 14:16:51 -05:00
r = nil
2013-02-25 11:42:20 -05:00
@mutex.synchronize do
2013-08-13 16:08:29 -04:00
v8['avatarTemplate'] = avatar_template
2013-02-05 14:16:51 -05:00
v8['size'] = size
decorate_context(v8)
2013-08-13 16:08:29 -04:00
r = v8.eval("Discourse.Utilities.avatarImg({ avatarTemplate: avatarTemplate, size: size });")
2013-02-05 14:16:51 -05:00
end
r
end
def self.apply_cdn(html, url)
return html unless url
image = /\.(jpg|jpeg|gif|png|tiff|tif|bmp)$/
2013-07-16 16:16:33 -04:00
relative = /^\/[^\/]/
2013-02-05 14:16:51 -05:00
doc = Nokogiri::HTML.fragment(html)
2013-02-05 14:16:51 -05:00
doc.css("a").each do |l|
href = l["href"].to_s
2013-07-16 16:16:33 -04:00
l["href"] = url + href if href =~ relative && href =~ image
2013-02-05 14:16:51 -05:00
end
2013-02-05 14:16:51 -05:00
doc.css("img").each do |l|
src = l["src"].to_s
2013-07-16 16:16:33 -04:00
l["src"] = url + src if src =~ relative
2013-02-05 14:16:51 -05:00
end
doc.to_s
end
def self.cook(text, opts={})
cloned = opts.dup
# we have a minor inconsistency
cloned[:topicId] = opts[:topic_id]
2013-02-25 11:42:20 -05:00
sanitized = Sanitize.clean(markdown(text.dup, cloned), PrettyText.whitelist)
if SiteSetting.add_rel_nofollow_to_user_content
2013-02-25 11:42:20 -05:00
sanitized = add_rel_nofollow_to_user_content(sanitized)
end
sanitized
end
2013-02-25 11:42:20 -05:00
def self.add_rel_nofollow_to_user_content(html)
whitelist = []
2013-02-11 03:01:33 -05:00
l = SiteSetting.exclude_rel_nofollow_domains
if l.present?
2013-02-25 11:42:20 -05:00
whitelist = l.split(",")
end
2013-02-11 03:01:33 -05:00
site_uri = nil
doc = Nokogiri::HTML.fragment(html)
doc.css("a").each do |l|
href = l["href"].to_s
2013-02-25 11:42:20 -05:00
begin
uri = URI(href)
site_uri ||= URI(Discourse.base_url)
2013-02-25 11:42:20 -05:00
if !uri.host.present? ||
uri.host.ends_with?(site_uri.host) ||
2013-02-11 03:01:33 -05:00
whitelist.any?{|u| uri.host.ends_with?(u)}
# we are good no need for nofollow
else
l["rel"] = "nofollow"
end
rescue URI::InvalidURIError
2013-02-25 11:42:20 -05:00
# add a nofollow anyway
l["rel"] = "nofollow"
end
end
doc.to_html
2013-02-05 14:16:51 -05:00
end
def self.extract_links(html)
links = []
2013-06-05 14:53:07 -04:00
doc = Nokogiri::HTML.fragment(html)
# remove href inside quotes
doc.css("aside.quote a").each { |l| l["href"] = "" }
# extract all links from the post
doc.css("a").each { |l| links << l["href"] unless l["href"].blank? }
2013-06-05 14:53:07 -04:00
# extract links to quotes
doc.css("aside.quote").each do |a|
2013-06-05 14:53:07 -04:00
topic_id = a['data-topic']
2013-02-25 11:42:20 -05:00
url = "/t/topic/#{topic_id}"
2013-06-05 14:53:07 -04:00
if post_number = a['data-post']
url << "/#{post_number}"
end
links << url
end
2013-02-05 14:16:51 -05:00
links
end
2013-05-27 19:48:47 -04:00
def self.excerpt(html, max_length, options={})
ExcerptParser.get_excerpt(html, max_length, options)
end
2013-02-05 14:16:51 -05:00
def self.strip_links(string)
return string if string.blank?
# If the user is not basic, strip links from their bio
fragment = Nokogiri::HTML.fragment(string)
fragment.css('a').each {|a| a.replace(a.text) }
fragment.to_html
end
2013-05-27 19:48:47 -04:00
protected
2013-02-05 14:16:51 -05:00
def self.ctx_load(ctx, *files)
2013-05-27 19:48:47 -04:00
files.each do |file|
ctx.load(app_root + file)
2013-02-05 14:16:51 -05:00
end
end
end