Change regexp for hashtags

Fixes #5765, fixes #5815, supersedes #6099.

closes #7350

Changelog.md

@@ -14,6 +14,7 @@
 * Remove whitespace from author link [#7330](https://github.com/diaspora/diaspora/pull/7330)
 * Fix autosize in modals [#7339](https://github.com/diaspora/diaspora/pull/7339)
 * Only display invite link on contacts page if invitations are enabled [#7342](https://github.com/diaspora/diaspora/pull/7342)
+* Fix regex for hashtags for some languages [#7350](https://github.com/diaspora/diaspora/pull/7350)
 
 ## Features
 * Add support for [Liberapay](https://liberapay.com) donations [#7290](https://github.com/diaspora/diaspora/pull/7290)

app/assets/javascripts/app/helpers/text_formatter.js

@@ -37,7 +37,14 @@
     var hashtagPlugin = window.markdownitHashtag;
     md.use(hashtagPlugin, {
       // compare tag_text_regexp in app/models/acts_as_taggable_on-tag.rb
-      hashtagRegExp: "[" + PosixBracketExpressions.alnum + "_\\-]+|<3",
+      hashtagRegExp: "[" + PosixBracketExpressions.word +
+                           "\\u055b" + // Armenian emphasis mark
+                           "\\u055c" + // Armenian exclamation mark
+                           "\\u055e" + // Armenian question mark
+                           "\\u058a" + // Armenian hyphen
+                           "_" +
+                           "\\-" +
+                     "]+|<3",
       // compare tag_strings in lib/diaspora/taggabe.rb
       preceding: "^|\\s"
     });

app/models/acts_as_taggable_on-tag.rb

@@ -4,7 +4,7 @@ module ActsAsTaggableOn
     self.include_root_in_json = false
 
     def self.tag_text_regexp
-      @@tag_text_regexp ||= "[[:alnum:]]_-"
+      @tag_text_regexp ||= "[[:word:]]\u055b\u055c\u055e\u058a_-"
     end
 
     def self.autocomplete(name)

spec/javascripts/app/helpers/text_formatter_spec.js

@@ -9,21 +9,46 @@ describe("app.helpers.textFormatter", function(){
   // https://github.com/svbergerem/markdown-it-hashtag/tree/master/test
   context("hashtags", function() {
     beforeEach(function() {
-      this.tags = [
+      this.goodTags = [
         "tag",
         "diaspora",
         "PARTIES",
-        "<3"
+        "<3",
+        "diaspora-dev",
+        "diaspora_dev",
+        // issue #5765
+        "മലയാണ്മ",
+        // issue #5815
+        "ինչո՞ւ",
+        "այո՜ո",
+        "սեւ֊սպիտակ",
+        "գժանո՛ց"
+      ];
+
+      this.badTags = [
+        "tag.tag",
+        "hash:tag",
+        "hash*tag"
       ];
     });
 
-    it("renders tags as links", function() {
-      var formattedText = this.formatter('#'+this.tags.join(" #"));
-      _.each(this.tags, function(tag) {
-        var link ='<a href="/tags/'+tag.toLowerCase()+'" class="tag">#'+tag.replace("<","&lt;")+'</a>';
+    it("renders good tags as links", function() {
+      var self = this;
+      this.goodTags.forEach(function(tag) {
+        var formattedText = self.formatter("#newhashtag #" + tag + " test");
+        var link = "<a href=\"/tags/" + tag.toLowerCase() + "\" class=\"tag\">#" + tag.replace("<", "&lt;") + "</a>";
         expect(formattedText).toContain(link);
       });
     });
+
+    it("doesn't render bad tags as links", function() {
+      var self = this;
+      this.badTags.forEach(function(tag) {
+        var formattedText = self.formatter("#newhashtag #" + tag + " test");
+        var link = "<a href=\"/tags/" + tag.toLowerCase() + "\" class=\"tag\">#" + tag.replace("<", "&lt;") + "</a>";
+        expect(formattedText).not.toContain(link);
+      });
+    });
   });
 
   // Some basic specs. For more detailed specs see

spec/lib/diaspora/taggable_spec.rb

@@ -27,6 +27,42 @@ describe Diaspora::Taggable do
         expect(text).to eq("<a class=\"tag\" href=\"/tags/l\">#l</a> <a class=\"tag\" href=\"/tags/lol\">#lol</a>")
       end
     end
+
+    context "good tags" do
+      it "autolinks" do
+        good_tags = [
+          "tag",
+          "diaspora",
+          "PARTIES",
+          "diaspora-dev",
+          "diaspora_dev",
+          # issue #5765
+          "മലയാണ്മ",
+          # issue #5815
+          "ինչո՞ւ",
+          "այո՜ո",
+          "սեւ֊սպիտակ",
+          "գժանո՛ց"
+        ]
+        good_tags.each do |tag|
+          text = Diaspora::Taggable.format_tags("#newhashtag ##{tag} #newhashtag")
+          expect(text).to match("<a class=\"tag\" href=\"/tags/#{tag}\">##{tag}</a>")
+        end
+      end
+    end
+
+    context "bad tags" do
+      it "doesn't autolink" do
+        bad_tags = [
+          "tag.tag",
+          "hash:tag"
+        ]
+        bad_tags.each do |tag|
+          text = Diaspora::Taggable.format_tags("#newhashtag ##{tag} #newhashtag")
+          expect(text).not_to match("<a class=\"tag\" href=\"/tags/#{tag}\">##{tag}</a>")
+        end
+      end
+    end
   end
 
   describe "#format_tags_for_mail" do

spec/shared_behaviors/taggable.rb

@@ -10,23 +10,35 @@ shared_examples_for "it is taggable" do
     link_to  "##{s}", "/tags/#{s}", :class => 'tag'
   end
 
-  describe '.format_tags' do
+  describe ".format_tags" do
+    let(:tag_list) {
+      [
+        "what",
+        "hey",
+        "vöglein",
+        "മലയാണ്മ",
+        "գժանո՛ց"
+      ]
+    }
+
     before do
-      @str = '#what #hey #vöglein'
+      @str = tag_list.map {|tag| "##{tag}" }.join(" ")
       @object.send(@object.class.field_with_tags_setter, @str)
       @object.build_tags
       @object.save!
     end
 
     it "supports non-ascii characters" do
-      expect(@object.tags(true).map(&:name)).to include('vöglein')
+      tag_list.each do |tag|
+        expect(@object.tags(true).map(&:name)).to include(tag)
+      end
     end
 
-    it 'links each tag' do
+    it "links each tag" do
       formatted_string = Diaspora::Taggable.format_tags(@str)
-      expect(formatted_string).to include(tag_link('what'))
-      expect(formatted_string).to include(tag_link('hey'))
-      expect(formatted_string).to include(tag_link('vöglein'))
+      tag_list.each do |tag|
+        expect(formatted_string).to include(tag_link(tag))
+      end
     end
 
     it 'responds to plain_text' do