basic support for Redcarpet Markdown

This commit is contained in:
Brian Wisti 2011-08-16 10:28:00 -07:00
parent 97ab077e04
commit 03127cd592
7 changed files with 257 additions and 152 deletions

View file

@ -50,6 +50,7 @@ gem 'json', '1.4.6'
gem 'http_accept_language', :git => 'git://github.com/iain/http_accept_language.git', :ref => '0b78aa7849fc90cf9e12'
gem 'thin', '1.2.11', :require => false
gem 'redcarpet', :git => 'git://github.com/tanoku/redcarpet'
#Websocket
gem 'em-websocket', :git => 'git://github.com/igrigorik/em-websocket', :ref => 'e278f5a1c4db60be7485'

View file

@ -56,6 +56,12 @@ GIT
addressable (>= 2.1.1)
eventmachine (>= 0.12.9)
GIT
remote: git://github.com/tanoku/redcarpet
revision: 6b43f042305eb63620e3771a977f45e24ea7b576
specs:
redcarpet (2.0.0b3)
GIT
remote: git://github.com/zhitomirskiyi/ruby-jwt.git
revision: fa7f46b5ac3653e30cf60abc78de9ffb3319dc0c
@ -500,6 +506,7 @@ DEPENDENCIES
rails (= 3.0.9)
rails-i18n
rcov
redcarpet!
resque (= 1.10.0)
resque-ensure-connected
rest-client (= 1.6.1)

View file

@ -2,134 +2,27 @@
# licensed under the Affero General Public License version 3 or later. See
# the COPYRIGHT file.
require 'lib/diaspora/markdownify'
module MarkdownifyHelper
def markdownify(message, options={})
message = h(message).to_str
def markdownify(message, render_options={})
markdown_options = {
:autolink => true,
}
options[:newlines] = true if !options.has_key?(:newlines)
options[:specialchars] = true if !options.has_key?(:specialchars)
render_options[:filter_html] = true
message = process_links(message)
message = process_autolinks(message)
message = process_emphasis(message)
message = process_youtube(message, options[:youtube_maps])
message = process_vimeo(message, options[:vimeo_maps])
message = process_specialchars(message) if options[:specialchars]
message = process_newlines(message) if options[:newlines]
message.html_safe
renderer = Diaspora::Markdownify::HTML.new(render_options)
markdown = Redcarpet::Markdown.new(renderer, markdown_options)
message = markdown.render(message)
return message.html_safe
end
def process_newlines(message)
message.gsub!(/\n+/, '<br />')
message
end
def process_links(message)
message.gsub!(/\[\s*([^\[]+?)\s*\]\(\s*([^ ]+\s*) \&quot;(([^&]|(&[^q])|(&q[^u])|(&qu[^o])|(&quo[^t])|(&quot[^;]))+)\&quot;\s*\)/) do |m|
escape = "\\"
link = $1
url = $2
title = $3
url.gsub!("_", "\\_")
url.gsub!("*", "\\*")
protocol = (url =~ /^\w+:\/\//) ? '' :'http://'
res = "<a target=\"#{escape}_blank\" href=\"#{protocol}#{url}\" title=\"#{title}\">#{link}</a>"
res
# in very clear cases, let newlines become <br /> tags
# Grabbed from Github flavored Markdown
message.gsub(/^[\w\<][^\n]*\n+/) do |x|
x =~ /\n{2}/ ? x : (x.strip!; x << " \n")
end
message.gsub!(/\[\s*([^\[]+?)\s*\]\(\s*([^ ]+)\s*\)/) do |m|
escape = "\\"
link = $1
url = $2
url.gsub!("_", "\\_")
url.gsub!("*", "\\*")
protocol = (url =~ /^\w+:\/\//) ? '' :'http://'
res = "<a target=\"#{escape}_blank\" href=\"#{protocol}#{url}\">#{link}</a>"
res
end
message
end
def process_youtube(message, youtube_maps)
processed_message = message.gsub(YoutubeTitles::YOUTUBE_ID_REGEX) do |matched_string|
match_data = matched_string.match(YoutubeTitles::YOUTUBE_ID_REGEX)
video_id = match_data[1]
anchor = match_data[2]
anchor ||= ''
if youtube_maps && youtube_maps[video_id]
title = h(CGI::unescape(youtube_maps[video_id]))
else
title = I18n.t 'application.helper.video_title.unknown'
end
' <a class="video-link" data-host="youtube.com" data-video-id="' + video_id + '" data-anchor="' + anchor + '" href="'+ match_data[0].strip + '" target="_blank">Youtube: ' + title + '</a>'
end
processed_message
end
def process_autolinks(message)
message.gsub!(/( |^)(www\.[^\s]+\.[^\s])/, '\1http://\2')
message.gsub!(/(<a target="\\?_blank" href=")?(https|http|ftp):\/\/([^\s]+)/) do |m|
captures = [$1,$2,$3]
if !captures[0].nil?
m
elsif m.match(/(youtu.?be|vimeo)/)
m.gsub(/(\*|_)/) { |m| "\\#{$1}" } #remove markers on markdown chars to not markdown inside links
else
res = %{<a target="_blank" href="#{captures[1]}://#{captures[2]}">#{captures[2]}</a>}
res.gsub!(/(\*|_)/) { |m| "\\#{$1}" }
res
end
end
message
end
def process_emphasis(message)
message.gsub!("\\**", "-^doublestar^-")
message.gsub!("\\__", "-^doublescore^-")
message.gsub!("\\*", "-^star^-")
message.gsub!("\\_", "-^score^-")
message.gsub!(/(\*\*\*|___)(.+?)\1/m, '<em><strong>\2</strong></em>')
message.gsub!(/(\*\*|__)(.+?)\1/m, '<strong>\2</strong>')
message.gsub!(/(\*|_)(.+?)\1/m, '<em>\2</em>')
message.gsub!("-^doublestar^-", "**")
message.gsub!("-^doublescore^-", "__")
message.gsub!("-^star^-", "*")
message.gsub!("-^score^-", "_")
message
end
def process_vimeo(message, vimeo_maps)
regex = /https?:\/\/(?:w{3}\.)?vimeo.com\/(\d{6,})\/?/
processed_message = message.gsub(regex) do |matched_string|
match_data = message.match(regex)
video_id = match_data[1]
if vimeo_maps && vimeo_maps[video_id]
title = h(CGI::unescape(vimeo_maps[video_id]))
else
title = I18n.t 'application.helper.video_title.unknown'
end
' <a class="video-link" data-host="vimeo.com" data-video-id="' + video_id + '" href="' + match_data[0] + '" target="_blank">Vimeo: ' + title + '</a>'
end
processed_message
end
def process_specialchars(message)
map = [
["&lt;3", "&hearts;"],
["&lt;-&gt;", "&#8596;"],
["-&gt;", "&rarr;"],
["&lt;-", "&larr;"],
["...", "&hellip;"],
["(tm)", "&trade;"],
["(r)", "&reg;"],
["(c)", "&copy;"]
]
map.each do |mapping|
message.gsub!(mapping[0], mapping[1])
end
message
end
end

View file

@ -25,7 +25,7 @@ Feature: posting
Scenario: see a tag that I am following
When I go to the home page
And I follow "#boss"
Then I should see "I am da #boss"
Then I should see "I am da #boss" within "body"
Scenario: can stop following a particular tag
When I hover over the ".button.tag_following"

169
lib/diaspora/markdownify.rb Normal file
View file

@ -0,0 +1,169 @@
require 'uri'
module Diaspora
module Markdownify
class HTML < Redcarpet::Render::HTML
attr_accessor :newlines, :specialchars, :youtube_maps, :vimeo_maps
def initialize(options={})
super
@expand_tags = options.fetch(:expand_tabs, true)
@newlines = options.fetch(:newlines, true)
@specialchars = options.fetch(:specialchars, true)
@youtube_maps = options.fetch(:youtube_maps, {})
@vimeo_maps = options.fetch(:vimeo_maps, {})
end
def autolink(link, type)
autolink_youtube(link) || autolink_vimeo(link) || autolink_simple(link)
end
def autolink_simple(link)
# If there isn't *some* protocol, assume http
if link !~ %r{^\w+://}
link = "http://#{link}"
end
content = link.gsub(%r{^\w+://}, '')
%Q{<a target="_blank" href="#{link}">#{content}</a>}
end
def autolink_vimeo(link)
regex = %r{https?://(?:w{3}\.)?vimeo.com/(\d{6,})/?}
if link =~ regex
video_id = $1
if @vimeo_maps[video_id]
title = h(CGI::unescape(@vimeo_maps[video_id]))
else
title = I18n.t 'application.helper.video_title.unknown'
end
return ' <a class="video-link" data-host="vimeo.com" data-video-id="' +
video_id + '" href="' + link + '" target="_blank">Vimeo: ' + title + '</a>'
end
end
def autolink_youtube(link)
if link =~ YoutubeTitles::YOUTUBE_ID_REGEX
video_id = $1
anchor = $2 || ''
if @youtube_maps[video_id]
title = h(CGI::unescape(@youtube_maps[video_id]))
else
title = I18n.t 'application.helper.video_title.unknown'
end
return ' <a class="video-link" data-host="youtube.com" data-video-id="' +
video_id + '" data-anchor="' + anchor +
'" href="'+ link + '" target="_blank">Youtube: ' + title + '</a>'
end
end
def double_emphasis(text)
"<strong>#{text}</strong>"
end
def linebreak()
"<br />"
end
def link(link, title, content)
return autolink(link, 'url') if link == content
if link =~ Regexp.new(Regexp.escape(content))
return autolink(link, 'url')
end
if link !~ %r{^\w+://}
link = "http://#{link}"
end
tag = if title and content
%Q{<a target="_blank" href="#{link}" title="#{title}">#{content}</a>}
elsif content
%Q{<a target="_blank" href="#{link}">#{content}</a>}
else
autolink(link, 'url')
end
return tag
end
def paragraph(text)
if @expand_tags
text = Diaspora::Taggable.format_tags(text, :no_escape => true)
end
return "<p>#{text}</p>"
end
def preprocess(full_document)
if @specialchars
full_document = specialchars(full_document)
end
our_unsafe_chars = '()'
full_document = full_document.gsub(%r{
\[ \s*? ( [^ \] ]+ ) \s*? \]
(?:
\( \s*? (\S+) \s*? (?: "([^"]+)" )? \) \s*?
)
}xm) do |m|
content = $1
link = URI.escape($2, our_unsafe_chars)
title = $3
title_chunk = if title
%W{" #{title}"}
else
''
end
%Q{[#{content}](#{link}#{title_chunk})}
end
if @newlines
# in very clear cases, let newlines become <br /> tags
# Grabbed from Github flavored Markdown
full_document = full_document.gsub(/^[\w\<][^\n]*\n+/) do |x|
x =~ /\n{2}/ ? x : (x.strip!; x << " \n")
end
end
return full_document
end
def single_emphasis(text)
"<em>#{text}</em>"
end
def specialchars(text)
if @specialchars
map = [
["<3", "&hearts;"],
["<->", "&#8596;"],
["->", "&rarr;"],
["<-", "&larr;"],
["\.\.\.", "&hellip;"],
["(tm)", "&trade;"],
["(r)", "&reg;"],
["(c)", "&copy;"]
]
end
map.each do |mapping|
text = text.gsub(mapping[0], mapping[1])
end
return text
end
def triple_emphasis(text)
single_emphasis(double_emphasis(text))
end
end
end
end

View file

@ -38,6 +38,9 @@ h1
ul > li
:list-style none
.content ul li
:list-style disc
form
:position relative
input,

View file

@ -8,31 +8,35 @@ describe MarkdownifyHelper do
describe "#markdownify" do
describe "autolinks" do
it "should not allow basic XSS/HTML" do
markdownify("<script>alert('XSS is evil')</script>").should == "&lt;script&gt;alert('XSS is evil')&lt;/script&gt;"
markdownify("<script>alert('XSS is evil')</script>").should == "<p>alert('XSS is evil')</p>"
end
it "should recognize basic http links (1/3)" do
proto="http"
url="bugs.joindiaspora.com/issues/332"
markdownify(proto+"://"+url).should == "<a target=\"_blank\" href=\""+proto+"://"+url+"\">"+url+"</a>"
full_url = "#{proto}://#{url}"
markdownify(full_url).should == %Q{<p><a target="_blank" href="#{full_url}">#{url}</a></p>}
end
it "should recognize basic http links (2/3)" do
proto="http"
url="webmail.example.com?~()!*/"
markdownify(proto+"://"+url).should == "<a target=\"_blank\" href=\""+proto+"://"+url+"\">"+url+"</a>"
full_url = "#{proto}://#{url}"
markdownify(full_url).should == %Q{<p><a target="_blank" href="#{full_url}">#{url}</a></p>}
end
it "should recognize basic http links (3/3)" do
proto="http"
url="127.0.0.1:3000/users/sign_in"
markdownify(proto+"://"+url).should == "<a target=\"_blank\" href=\""+proto+"://"+url+"\">"+url+"</a>"
full_url = "#{proto}://#{url}"
markdownify(full_url).should == %Q{<p><a target="_blank" href="#{full_url}">#{url}</a></p>}
end
it "should recognize secure https links" do
proto="https"
url="127.0.0.1:3000/users/sign_in"
markdownify(proto+"://"+url).should == "<a target=\"_blank\" href=\""+proto+"://"+url+"\">"+url+"</a>"
full_url = "#{proto}://#{url}"
markdownify(full_url).should == %Q{<p><a target="_blank" href="#{full_url}">#{url}</a></p>}
end
it "doesn't double parse video links" do
@ -40,8 +44,8 @@ describe MarkdownifyHelper do
http://www.youtube.com/watch?v=0x__dDWdf23&a=GxdCwVVULXdvEBKmx_f5ywvZ0zZHHHDU&list=ML&playnext=1
http://youtu.be/x_CzD0GBD-4"
res = markdownify(message)
res.should =~ /href.+href.+href/
res.should_not =~ /href.+href.+href.+href/
res.should =~ /href.+href.+href/m
res.should_not =~ /href.+href.+href.+href/m
end
describe "video links" do
@ -69,6 +73,12 @@ describe MarkdownifyHelper do
res.should =~ /Youtube:/
res.should =~ /data-host="youtube.com"/
res.should =~ /data-video-id="#{video_id}"/
url = "www.youtube.com/watch?foo=bar&v=BARFOO-----&whatever=related"
res = markdownify(url)
res.should =~ /Youtube:/
res.should =~ /data-host="youtube.com"/
res.should =~ /data-video-id="BARFOO-----"/
end
it "recognizes youtu.be links" do
@ -141,113 +151,135 @@ describe MarkdownifyHelper do
proto="ftp"
url="ftp.uni-kl.de/CCC/26C3/mp4/26c3-3540-en-a_hackers_utopia.mp4"
# I did not watch that one, but the title sounds nice :P
markdownify(proto+"://"+url).should == "<a target=\"_blank\" href=\""+proto+"://"+url+"\">"+url+"</a>"
markdownify(proto+"://"+url).should == "<p><a target=\"_blank\" href=\""+proto+"://"+url+"\">"+url+"</a></p>"
end
it "should recognize www links" do
url="www.joindiaspora.com"
markdownify(url).should == "<a target=\"_blank\" href=\"http://"+url+"\">"+url+"</a>"
markdownify(url).should == %Q{<p><a target="_blank" href="http://#{url}">#{url}</a></p>}
end
end
describe "specialchars" do
it "replaces &lt;3 with &hearts;" do
message = "i <3 you"
markdownify(message).should == "i &hearts; you"
markdownify(message).should == "<p>i &hearts; you</p>"
end
it "replaces various things with (their) HTML entities" do
message = "... <-> -> <- (tm) (r) (c)"
markdownify(message).should == "&hellip; &#8596; &rarr; &larr; &trade; &reg; &copy;"
markdownify(message).should == "<p>&hellip; &#8596; &rarr; &larr; &trade; &reg; &copy;</p>"
end
it "skips doing it if you say so" do
message = "... -> <-"
markdownify(message, :specialchars => false).should == "... -&gt; &lt;-"
markdownify(message, :specialchars => false).should == "<p>... -&gt; &lt;-</p>"
end
end
describe "weak emphasis" do
it "should be recognized (1/2)" do
message = "*some text* some text *some text* some text"
markdownify(message).should == "<em>some text</em> some text <em>some text</em> some text"
markdownify(message).should == "<p><em>some text</em> some text <em>some text</em> some text</p>"
end
it "should be recognized (2/2)" do
message = "_some text_ some text _some text_ some text"
markdownify(message).should == "<em>some text</em> some text <em>some text</em> some text"
markdownify(message).should == "<p><em>some text</em> some text <em>some text</em> some text</p>"
end
end
describe "strong emphasis" do
it "should be recognized (1/2)" do
message = "**some text** some text **some text** some text"
markdownify(message).should == "<strong>some text</strong> some text <strong>some text</strong> some text"
markdownify(message).should == "<p><strong>some text</strong> some text <strong>some text</strong> some text</p>"
end
it "should be recognized (2/2)" do
message = "__some text__ some text __some text__ some text"
markdownify(message).should == "<strong>some text</strong> some text <strong>some text</strong> some text"
markdownify(message).should == "<p><strong>some text</strong> some text <strong>some text</strong> some text</p>"
end
end
describe "nested weak and strong emphasis" do
it "should be rendered correctly" do
message = "__this is _some_ text__"
markdownify(message).should == "<strong>this is <em>some</em> text</strong>"
markdownify(message).should == "<p><strong>this is <em>some</em> text</strong></p>"
message = "*this is **some** text*"
markdownify(message).should == "<em>this is <strong>some</strong> text</em>"
markdownify(message).should == "<p><em>this is <strong>some</strong> text</em></p>"
message = "___some text___"
markdownify(message).should == "<em><strong>some text</strong></em>"
markdownify(message).should == "<p><em><strong>some text</strong></em></p>"
end
end
describe "links" do
it "should be recognized without title attribute" do
message = "[link text](http://someurl.com) [link text](http://someurl.com)"
markdownify(message).should == '<a target="_blank" href="http://someurl.com">link text</a> <a target="_blank" href="http://someurl.com">link text</a>'
markdownify(message).should == '<p><a target="_blank" href="http://someurl.com">link text</a> <a target="_blank" href="http://someurl.com">link text</a></p>'
end
it "should be recognized with title attribute" do
message = '[link text](http://someurl.com "some title") [link text](http://someurl.com "some title")'
markdownify(message).should == '<a target="_blank" href="http://someurl.com" title="some title">link text</a> <a target="_blank" href="http://someurl.com" title="some title">link text</a>'
markdownify(message).should == '<p><a target="_blank" href="http://someurl.com" title="some title">link text</a> <a target="_blank" href="http://someurl.com" title="some title">link text</a></p>'
end
it "should have a robust link parsing" do
message = "[wikipedia](http://en.wikipedia.org/wiki/Text_(literary_theory))"
link = markdownify(message)
link.should =~ %r{href="http://en.wikipedia.org/wiki/Text_%28literary_theory%29"}
message = "[ links]( google.com)"
markdownify(message).should == %Q{<p><a target="_blank" href="http://google.com">links</a></p>}
message = "[_http_](http://google.com/search?q=with_multiple__underscores*and**asterisks )"
markdownify(message).should == %Q{<p><a target="_blank" href="http://google.com/search?q=with_multiple__underscores*and**asterisks"><em>http</em></a></p>}
message = %{[___FTP___]( ftp://ftp.uni-kl.de/CCC/26C3/mp4/26c3-3540-en-a_hackers_utopia.mp4 \"File Transfer Protocol\")}
markdownify(message).should == %{<p><a target="_blank" href="ftp://ftp.uni-kl.de/CCC/26C3/mp4/26c3-3540-en-a_hackers_utopia.mp4" title="File Transfer Protocol"><em><strong>FTP</strong></em></a></p>}
message = %{[**any protocol**](foo://bar.example.org/yes_it*makes*no_sense)}
markdownify(message).should == %{<p><a target="_blank" href="foo://bar.example.org/yes_it*makes*no_sense"><strong>any protocol</strong></a></p>}
message = "This [ *text* ]( http://en.wikipedia.org/wiki/Text_(literary_theory) ) with many [ links]( google.com) tests [_http_](http://google.com/search?q=with_multiple__underscores*and**asterisks ), [___FTP___]( ftp://ftp.uni-kl.de/CCC/26C3/mp4/26c3-3540-en-a_hackers_utopia.mp4 \"File Transfer Protocol\"), [**any protocol**](foo://bar.example.org/yes_it*makes*no_sense)"
markdownify(message).should == 'This <a target="_blank" href="http://en.wikipedia.org/wiki/Text_(literary_theory)"><em>text</em></a> with many <a target="_blank" href="http://google.com">links</a> tests <a target="_blank" href="http://google.com/search?q=with_multiple__underscores*and**asterisks"><em>http</em></a>, <a target="_blank" href="ftp://ftp.uni-kl.de/CCC/26C3/mp4/26c3-3540-en-a_hackers_utopia.mp4" title="File Transfer Protocol"><em><strong>FTP</strong></em></a>, <a target="_blank" href="foo://bar.example.org/yes_it*makes*no_sense"><strong>any protocol</strong></a>'
markdownify(message).should == '<p>This <a target="_blank" href="http://en.wikipedia.org/wiki/Text_%28literary_theory%29"><em>text</em></a> with many <a target="_blank" href="http://google.com">links</a> tests <a target="_blank" href="http://google.com/search?q=with_multiple__underscores*and**asterisks"><em>http</em></a>, <a target="_blank" href="ftp://ftp.uni-kl.de/CCC/26C3/mp4/26c3-3540-en-a_hackers_utopia.mp4" title="File Transfer Protocol"><em><strong>FTP</strong></em></a>, <a target="_blank" href="foo://bar.example.org/yes_it*makes*no_sense"><strong>any protocol</strong></a></p>'
end
it "should leave #tag links intact" do
message = %{<a href="/tags/tagged" class="tag">#tagged</a>}
markdownify(message).should == "<p>#{message}</p>"
message = %{alice - 1 - <a href="/tags/tagged" class="tag">#tagged</a>}
markdownify(message).should == "<p>#{message}</p>"
end
end
describe "nested emphasis and links tags" do
it "should be rendered correctly" do
message = '[**some *link* text**](someurl.com "some title")'
markdownify(message).should == '<a target="_blank" href="http://someurl.com" title="some title"><strong>some <em>link</em> text</strong></a>'
markdownify(message).should == '<p><a target="_blank" href="http://someurl.com" title="some title"><strong>some <em>link</em> text</strong></a></p>'
end
end
it "should allow escaping" do
message = '*some text* \\*some text* \\**some text* _some text_ \\_some text_ \\__some text_'
markdownify(message).should == "<em>some text</em> *some text<em> **some text</em> <em>some text</em> _some text<em> __some text</em>"
message = '*some text* \*some text* \**some text* _some text_ \_some text_ \__some text_'
markdownify(message).should == "<p><em>some text</em> *some text* *<em>some text</em> <em>some text</em> _some text_ _<em>some text</em></p>"
end
describe "newlines" do
it 'skips inserting newlines if you pass the newlines option' do
message = "These\nare\n\some\nnew\lines"
res = markdownify(message, :newlines => false)
res.should == message
res.should == "<p>#{message}</p>"
end
it 'generates breaklines' do
message = "These\nare\nsome\nnew\nlines"
res = markdownify(message)
res.should == "These<br /\>are<br /\>some<br /\>new<br /\>lines"
res.should == "<p>These<br /\>are<br /\>some<br /\>new<br /\>lines</p>"
end
it 'should render newlines and basic http links correctly' do
message = "Some text, then a line break and a link\nhttp://joindiaspora.com\nsome more text"
res = markdownify(message)
res.should == 'Some text, then a line break and a link<br /><a target="_blank" href="http://joindiaspora.com">joindiaspora.com</a><br />some more text'
res.should == '<p>Some text, then a line break and a link<br /><a target="_blank" href="http://joindiaspora.com">joindiaspora.com</a><br />some more text</p>'
end
end