twitter-text 1.4.15 → 1.4.16

data/README.rdoc

@@ -55,9 +55,16 @@ numbers but cannot be solely numbers and cannot contain punctuation.
 
 === URLs
 
-Auto-linking and extraction of URLs differs from the Rails default so that it
-will work correctly in Tweets written in languages that do not include spaces
-between words.
+Asian languages like Chinese, Japanese or Korean may not use a delimiter such as
+a space to separate normal text from URLs making it difficult to identify where
+the URL ends and the text starts.
+
+For this reason twitter-text currently does not support extracting or auto-linking
+of URLs immediately followed by non-Latin characters.
+
+Example: "http://twitter.com/は素晴らしい" .
+The normal text is "は素晴らしい" and is not part of the URL even though
+it isn't space separated.
 
 === International
 
@@ -112,4 +119,4 @@ Thanks to everybody who has filed issues, provided feedback or contributed patch
     distributed under the License is distributed on an "AS IS" BASIS,
     WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
     See the License for the specific language governing permissions and
-    limitations under the License.
+    limitations under the License.

data/lib/autolink.rb

@@ -20,7 +20,7 @@ module Twitter
     OPTIONS_NOT_ATTRIBUTES = [:url_class, :list_class, :username_class, :hashtag_class,
                               :username_url_base, :list_url_base, :hashtag_url_base,
                               :username_url_block, :list_url_block, :hashtag_url_block, :link_url_block,
-                              :suppress_lists, :suppress_no_follow, :url_entities]
+                              :username_include_symbol, :suppress_lists, :suppress_no_follow, :url_entities]
 
     HTML_ENTITIES = {
       '&' => '&',
@@ -47,6 +47,7 @@ module Twitter
     # <tt>:username_url_base</tt>::      the value for <tt>href</tt> attribute on username links. The <tt>@username</tt> (minus the <tt>@</tt>) will be appended at the end of this.
     # <tt>:list_url_base</tt>::      the value for <tt>href</tt> attribute on list links. The <tt>@username/list</tt> (minus the <tt>@</tt>) will be appended at the end of this.
     # <tt>:hashtag_url_base</tt>::      the value for <tt>href</tt> attribute on hashtag links. The <tt>#hashtag</tt> (minus the <tt>#</tt>) will be appended at the end of this.
+    # <tt>:username_include_symbol</tt>::    place the <tt>@</tt> symbol within username and list links
     # <tt>:suppress_lists</tt>::    disable auto-linking to lists
     # <tt>:suppress_no_follow</tt>::   Do not add <tt>rel="nofollow"</tt> to auto-linked items
     # <tt>:target</tt>::   add <tt>target="window_name"</tt> to auto-linked items
@@ -66,6 +67,7 @@ module Twitter
     # <tt>:list_class</tt>::    class to add to list <tt><a></tt> tags
     # <tt>:username_class</tt>::    class to add to username <tt><a></tt> tags
     # <tt>:username_url_base</tt>::      the value for <tt>href</tt> attribute on username links. The <tt>@username</tt> (minus the <tt>@</tt>) will be appended at the end of this.
+    # <tt>:username_include_symbol</tt>::    place the <tt>@</tt> symbol within username and list links
     # <tt>:list_url_base</tt>::      the value for <tt>href</tt> attribute on list links. The <tt>@username/list</tt> (minus the <tt>@</tt>) will be appended at the end of this.
     # <tt>:suppress_lists</tt>::    disable auto-linking to lists
     # <tt>:suppress_no_follow</tt>::   Do not add <tt>rel="nofollow"</tt> to auto-linked items
@@ -75,13 +77,16 @@ module Twitter
       options[:url_class] ||= DEFAULT_URL_CLASS
       options[:list_class] ||= DEFAULT_LIST_CLASS
       options[:username_class] ||= DEFAULT_USERNAME_CLASS
-      options[:username_url_base] ||= "http://twitter.com/"
-      options[:list_url_base] ||= "http://twitter.com/"
+      options[:username_url_base] ||= "https://twitter.com/"
+      options[:list_url_base] ||= "https://twitter.com/"
       options[:target] ||= DEFAULT_TARGET
 
       extra_html = HTML_ATTR_NO_FOLLOW unless options[:suppress_no_follow]
 
       Twitter::Rewriter.rewrite_usernames_or_lists(text) do |at, username, slash_listname|
+        at_before_user = options[:username_include_symbol] ? at : ''
+        at = options[:username_include_symbol] ? '' : at
+
         name = "#{username}#{slash_listname}"
         chunk = block_given? ? yield(name) : name
 
@@ -89,16 +94,16 @@ module Twitter
           href = if options[:list_url_block]
             options[:list_url_block].call(name.downcase)
           else
-            "#{html_escape(options[:list_url_base])}#{html_escape(name.downcase)}"
+            "#{html_escape(options[:list_url_base] + name.downcase)}"
           end
-          %(#{at}<a class="#{options[:url_class]} #{options[:list_class]}" #{target_tag(options)}href="#{href}"#{extra_html}>#{html_escape(chunk)}</a>)
+          %(#{at}<a class="#{options[:url_class]} #{options[:list_class]}" #{target_tag(options)}href="#{href}"#{extra_html}>#{html_escape(at_before_user + chunk)}</a>)
         else
           href = if options[:username_url_block]
             options[:username_url_block].call(chunk)
           else
-            "#{html_escape(options[:username_url_base])}#{html_escape(chunk)}"
+            "#{html_escape(options[:username_url_base] + chunk)}"
           end
-          %(#{at}<a class="#{options[:url_class]} #{options[:username_class]}" #{target_tag(options)}href="#{href}"#{extra_html}>#{html_escape(chunk)}</a>)
+          %(#{at}<a class="#{options[:url_class]} #{options[:username_class]}" #{target_tag(options)}href="#{href}"#{extra_html}>#{html_escape(at_before_user + chunk)}</a>)
         end
       end
     end
@@ -116,7 +121,7 @@ module Twitter
       options = options.dup
       options[:url_class] ||= DEFAULT_URL_CLASS
       options[:hashtag_class] ||= DEFAULT_HASHTAG_CLASS
-      options[:hashtag_url_base] ||= "http://twitter.com/search?q=%23"
+      options[:hashtag_url_base] ||= "https://twitter.com/#!/search?q=%23"
       options[:target] ||= DEFAULT_TARGET
       extra_html = HTML_ATTR_NO_FOLLOW unless options[:suppress_no_follow]
 
@@ -179,10 +184,10 @@ module Twitter
     BOOLEAN_ATTRIBUTES = Set.new([:disabled, :readonly, :multiple, :checked]).freeze
 
     def html_attrs_for_options(options)
-      html_attrs options.reject{|k, v| OPTIONS_NOT_ATTRIBUTES.include?(k)}
+      autolink_html_attrs options.reject{|k, v| OPTIONS_NOT_ATTRIBUTES.include?(k)}
     end
 
-    def html_attrs(options)
+    def autolink_html_attrs(options)
       options.inject("") do |attrs, (key, value)|
         if BOOLEAN_ATTRIBUTES.include?(key)
           value = value ? key : nil

data/lib/regex.rb

@@ -109,17 +109,17 @@ module Twitter
     REGEXEN[:end_hashtag_match] = /\A(?:[#＃]|:\/\/)/o
 
     REGEXEN[:at_signs] = /[@＠]/
-    REGEXEN[:extract_mentions] = /(^|[^a-zA-Z0-9_])#{REGEXEN[:at_signs]}([a-zA-Z0-9_]{1,20})/o
-    REGEXEN[:extract_mentions_or_lists] = /(^|[^a-zA-Z0-9_])#{REGEXEN[:at_signs]}([a-zA-Z0-9_]{1,20})(\/[a-zA-Z][a-zA-Z0-9_\-]{0,24})?/o
+    REGEXEN[:extract_mentions] = /(^|[^a-zA-Z0-9_!#\$%&*@＠])#{REGEXEN[:at_signs]}([a-zA-Z0-9_]{1,20})/o
+    REGEXEN[:extract_mentions_or_lists] = /(^|[^a-zA-Z0-9_!#\$%&*@＠])#{REGEXEN[:at_signs]}([a-zA-Z0-9_]{1,20})(\/[a-zA-Z][a-zA-Z0-9_\-]{0,24})?/o
     REGEXEN[:extract_reply] = /^(?:#{REGEXEN[:spaces]})*#{REGEXEN[:at_signs]}([a-zA-Z0-9_]{1,20})/o
     # Used in Extractor and Rewriter for final filtering
     REGEXEN[:end_screen_name_match] = /\A(?:#{REGEXEN[:at_signs]}|#{REGEXEN[:latin_accents]}|:\/\/)/o
 
-    REGEXEN[:auto_link_usernames_or_lists] = /([^a-zA-Z0-9_]|^|RT:?)([@＠]+)([a-zA-Z0-9_]{1,20})(\/[a-zA-Z][a-zA-Z0-9_\-]{0,24})?/o
+    REGEXEN[:auto_link_usernames_or_lists] = /([^a-zA-Z0-9_!#\$%&*@＠]|^|RT:?)([@＠]+)([a-zA-Z0-9_]{1,20})(\/[a-zA-Z][a-zA-Z0-9_\-]{0,24})?/o
     REGEXEN[:auto_link_emoticon] = /(8\-\#|8\-E|\+\-\(|\`\@|\`O|\<\|:~\(|\}:o\{|:\-\[|\>o\<|X\-\/|\[:-\]\-I\-|\/\/\/\/Ö\\\\\\\\|\(\|:\|\/\)|∑:\*\)|\( \| \))/
 
     # URL related hash regex collection
-    REGEXEN[:valid_preceding_chars] = /(?:[^-\/"'!=A-Z0-9_@＠#＃\.#{INVALID_CHARACTERS.join('')}]|^)/io
+    REGEXEN[:valid_preceding_chars] = /(?:[^-\/"'!=A-Z0-9_@＠\$#＃\.#{INVALID_CHARACTERS.join('')}]|^)/io
 
     DOMAIN_VALID_CHARS = "[^[:punct:][:space:][:blank:][:cntrl:]#{INVALID_CHARACTERS.join('')}#{UNICODE_SPACES.join('')}]"
     REGEXEN[:valid_subdomain] = /(?:(?:#{DOMAIN_VALID_CHARS}(?:[_-]|#{DOMAIN_VALID_CHARS})*)?#{DOMAIN_VALID_CHARS}\.)/io

data/spec/autolinking_spec.rb

@@ -298,7 +298,7 @@ describe Twitter::Autolink do
         it "should be linked" do
           link = Nokogiri::HTML(@autolinked_text).search('a')
           (link.inner_text.respond_to?(:force_encoding) ? link.inner_text.force_encoding("utf-8") : link.inner_text).should == "#{[0xFF03].pack('U')}twj_dev"
-          link.first['href'].should == 'http://twitter.com/search?q=%23twj_dev'
+          link.first['href'].should == 'https://twitter.com/#!/search?q=%23twj_dev'
         end
       end
 
@@ -309,7 +309,7 @@ describe Twitter::Autolink do
         end
 
         it "should be linked" do
-          @autolinked_text.should == "<a href=\"http://twitter.com/search?q=%23éhashtag\" title=\"#éhashtag\" class=\"tweet-url hashtag\" rel=\"nofollow\">#éhashtag</a>"
+          @autolinked_text.should == "<a href=\"https://twitter.com/#!/search?q=%23éhashtag\" title=\"#éhashtag\" class=\"tweet-url hashtag\" rel=\"nofollow\">#éhashtag</a>"
         end
       end
 
@@ -530,13 +530,13 @@ describe Twitter::Autolink do
       end
 
       it "should allow url/hashtag overlap" do
-        auto_linked = @linker.auto_link("http://twitter.com/#search")
-        auto_linked.should have_autolinked_url('http://twitter.com/#search')
+        auto_linked = @linker.auto_link("https://twitter.com/#search")
+        auto_linked.should have_autolinked_url('https://twitter.com/#search')
       end
 
       it "should not add invalid option in HTML tags" do
-        auto_linked = @linker.auto_link("http://twitter.com/ is a URL, not a hashtag", :hashtag_class => 'hashtag_classname')
-        auto_linked.should have_autolinked_url('http://twitter.com/')
+        auto_linked = @linker.auto_link("https://twitter.com/ is a URL, not a hashtag", :hashtag_class => 'hashtag_classname')
+        auto_linked.should have_autolinked_url('https://twitter.com/')
         auto_linked.should_not include('hashtag_class')
         auto_linked.should_not include('hashtag_classname')
       end
@@ -558,6 +558,16 @@ describe Twitter::Autolink do
       linked.should match(/nofollow/)
     end
 
+    it "should include the '@' symbol in a username when passed :username_include_symbol" do
+      linked = TestAutolink.new.auto_link("@user", :username_include_symbol => true)
+      linked.should link_to_screen_name('user', '@user')
+    end
+
+    it "should include the '@' symbol in a list when passed :username_include_symbol" do
+      linked = TestAutolink.new.auto_link("@user/list", :username_include_symbol => true)
+      linked.should link_to_list_path('user/list', '@user/list')
+    end
+
     it "should not add rel=nofollow when passed :suppress_no_follow" do
       linked = TestAutolink.new.auto_link("http://example.com/", :suppress_no_follow => true)
       linked.should have_autolinked_url('http://example.com/')
@@ -610,7 +620,7 @@ describe Twitter::Autolink do
       @linker.html_escape("&<>\"").should == "&amp;&lt;&gt;&quot;"
       @linker.html_escape("<div>").should == "&lt;div&gt;"
       @linker.html_escape("a&b").should == "a&amp;b"
-      @linker.html_escape("<a href=\"http://twitter.com\" target=\"_blank\">twitter & friends</a>").should == "&lt;a href=&quot;http://twitter.com&quot; target=&quot;_blank&quot;&gt;twitter &amp; friends&lt;/a&gt;"
+      @linker.html_escape("<a href=\"https://twitter.com\" target=\"_blank\">twitter & friends</a>").should == "&lt;a href=&quot;https://twitter.com&quot; target=&quot;_blank&quot;&gt;twitter &amp; friends&lt;/a&gt;"
       @linker.html_escape("&amp;").should == "&amp;amp;"
       @linker.html_escape(nil).should == nil
     end

data/spec/extractor_spec.rb

@@ -39,6 +39,13 @@ describe Twitter::Extractor do
       it "should be linked in Japanese text" do
         @extractor.extract_mentioned_screen_names("の@aliceに到着を待っている").should == ["alice"]
       end
+
+      it "should ignore mentions preceded by !, @, #, $, %, & or *" do
+        invalid_chars = ['!', '@', '#', '$', '%', '&', '*']
+        invalid_chars.each do |c|
+          @extractor.extract_mentioned_screen_names("f#{c}@kn").should == []
+        end
+      end
     end
 
     it "should accept a block arugment and call it in order" do

data/spec/spec_helper.rb

@@ -51,18 +51,26 @@ RSpec::Matchers.define :have_autolinked_url do |url, inner_text|
   end
 end
 
-RSpec::Matchers.define :link_to_screen_name do |screen_name|
+RSpec::Matchers.define :link_to_screen_name do |screen_name, inner_text|
+  expected = inner_text ? inner_text : screen_name
+
   match do |text|
     @link = Nokogiri::HTML(text).search("a.username")
-    @link && @link.inner_text == screen_name && "http://twitter.com/#{screen_name}".downcase.should == @link.first['href']
+    @link &&
+    @link.inner_text == expected &&
+    "https://twitter.com/#{screen_name}".downcase.should == @link.first['href']
   end
 
   failure_message_for_should do |text|
-    "expected link #{@link.inner_text} with href #{@link['href']} to match screen_name #{@screen_name}, but it does not"
+    if @link.first
+      "Expected link '#{@link.inner_text}' with href '#{@link.first['href']}' to match screen_name '#{expected}', but it does not."
+    else
+      "Expected screen name '#{screen_name}' to be autolinked in '#{text}', but no link was found."
+    end
   end
 
   failure_message_for_should_not do |text|
-    "expected link #{@link.inner_text} with href #{@link['href']} not to match screen_name #{@screen_name}, but it does"
+    "Expected link '#{@link.inner_text}' with href '#{@link.first['href']}' not to match screen_name '#{expected}', but it does."
   end
 
   description do
@@ -70,18 +78,26 @@ RSpec::Matchers.define :link_to_screen_name do |screen_name|
   end
 end
 
-RSpec::Matchers.define :link_to_list_path do |list_path|
+RSpec::Matchers.define :link_to_list_path do |list_path, inner_text|
+  expected = inner_text ? inner_text : list_path
+
   match do |text|
     @link = Nokogiri::HTML(text).search("a.list-slug")
-    !@link.nil? && @link.inner_text == list_path && "http://twitter.com/#{list_path}".downcase.should == @link.first['href']
+    @link &&
+    @link.inner_text == expected &&
+    "https://twitter.com/#{list_path}".downcase.should == @link.first['href']
   end
 
   failure_message_for_should do |text|
-    "expected link #{@link.inner_text} with href #{@link['href']} to match the list path #{list_path}, but it does not"
+    if @link.first
+      "Expected link '#{@link.inner_text}' with href '#{@link.first['href']}' to match the list path '#{expected}', but it does not."
+    else
+      "Expected list path '#{list_path}' to be autolinked in '#{text}', but no link was found."
+    end
   end
 
   failure_message_for_should_not do |text|
-    "expected link #{@link.inner_text} with href #{@link['href']} not to match the list path #{@list_path}, but it does"
+    "Expected link '#{@link.inner_text}' with href '#{@link.first['href']}' not to match the list path '#{expected}', but it does."
   end
 
   description do
@@ -91,17 +107,21 @@ end
 
 RSpec::Matchers.define :have_autolinked_hashtag do |hashtag|
   match do |text|
-    @link = Nokogiri::HTML(text).search("a[@href='http://twitter.com/search?q=#{hashtag.sub(/^#/, '%23')}']")
+    @link = Nokogiri::HTML(text).search("a[@href='https://twitter.com/#!/search?q=#{hashtag.sub(/^#/, '%23')}']")
     @link &&
     @link.inner_text &&
     @link.inner_text == hashtag
   end
 
   failure_message_for_should do |text|
-    if @link
+    if @link.first
       "Expected link text to be [#{hashtag}], but it was [#{@link.inner_text}] in #{text}"
     else
       "Expected hashtag #{hashtag} to be autolinked in '#{text}', but no link was found."
     end
   end
+
+  failure_message_for_should_not do |text|
+    "Expected link '#{@link.inner_text}' with href '#{@link.first['href']}' not to match the hashtag '#{hashtag}', but it does."
+  end
 end

data/twitter-text.gemspec

@@ -2,7 +2,7 @@
 
 Gem::Specification.new do |s|
   s.name = "twitter-text"
-  s.version = "1.4.15"
+  s.version = "1.4.16"
   s.authors = ["Matt Sanford", "Patrick Ewing", "Ben Cherry", "Britt Selvitelle",
                "Raffi Krikorian", "J.P. Cummins", "Yoshimasa Niwa", "Keita Fujii"]
   s.email = ["matt@twitter.com", "patrick.henry.ewing@gmail.com", "bcherry@gmail.com", "bs@brittspace.com",

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: twitter-text
 version: !ruby/object:Gem::Version 
-  hash: 25
+  hash: 39
   prerelease: 
   segments: 
   - 1
   - 4
-  - 15
-  version: 1.4.15
+  - 16
+  version: 1.4.16
 platform: ruby
 authors: 
 - Matt Sanford
@@ -22,7 +22,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-12-02 00:00:00 -08:00
+date: 2012-02-07 00:00:00 -08:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency