Fix some link anchors being recognized as hashtags (mastodon#27271)

app/models/tag.rb

@@ -35,7 +35,7 @@ class Tag < ApplicationRecord
   HASHTAG_LAST_SEQUENCE = '([[:word:]_]*[[:alpha:]][[:word:]_]*)'
   HASHTAG_NAME_PAT = "#{HASHTAG_FIRST_SEQUENCE}|#{HASHTAG_LAST_SEQUENCE}"
 
-  HASHTAG_RE = %r{(?:^|[^/)\w])#(#{HASHTAG_NAME_PAT})}i
+  HASHTAG_RE = %r{(?<![=/)[:word]])#(#{HASHTAG_NAME_PAT})}i
   HASHTAG_NAME_RE = /\A(#{HASHTAG_NAME_PAT})\z/i
   HASHTAG_INVALID_CHARS_RE = /[^[:alnum:]#{HASHTAG_SEPARATORS}]/
 

spec/models/tag_spec.rb

@@ -32,52 +32,56 @@
       expect(subject.match('https://en.wikipedia.org/wiki/Ghostbusters_(song)#Lawsuit')).to be_nil
     end
 
+    it 'does not match URLs with hashtag-like anchors after an empty query parameter' do
+      expect(subject.match('https://en.wikipedia.org/wiki/Ghostbusters_(song)?foo=#Lawsuit')).to be_nil
+    end
+
     it 'matches #ａｅｓｔｈｅｔｉｃ' do
-      expect(subject.match('this is #ａｅｓｔｈｅｔｉｃ').to_s).to eq ' #ａｅｓｔｈｅｔｉｃ'
+      expect(subject.match('this is #ａｅｓｔｈｅｔｉｃ').to_s).to eq '#ａｅｓｔｈｅｔｉｃ'
     end
 
     it 'matches digits at the start' do
-      expect(subject.match('hello #3d').to_s).to eq ' #3d'
+      expect(subject.match('hello #3d').to_s).to eq '#3d'
     end
 
     it 'matches digits in the middle' do
-      expect(subject.match('hello #l33ts35k').to_s).to eq ' #l33ts35k'
+      expect(subject.match('hello #l33ts35k').to_s).to eq '#l33ts35k'
     end
 
     it 'matches digits at the end' do
-      expect(subject.match('hello #world2016').to_s).to eq ' #world2016'
+      expect(subject.match('hello #world2016').to_s).to eq '#world2016'
     end
 
     it 'matches underscores at the beginning' do
-      expect(subject.match('hello #_test').to_s).to eq ' #_test'
+      expect(subject.match('hello #_test').to_s).to eq '#_test'
     end
 
     it 'matches underscores at the end' do
-      expect(subject.match('hello #test_').to_s).to eq ' #test_'
+      expect(subject.match('hello #test_').to_s).to eq '#test_'
     end
 
     it 'matches underscores in the middle' do
-      expect(subject.match('hello #one_two_three').to_s).to eq ' #one_two_three'
+      expect(subject.match('hello #one_two_three').to_s).to eq '#one_two_three'
     end
 
     it 'matches middle dots' do
-      expect(subject.match('hello #one·two·three').to_s).to eq ' #one·two·three'
+      expect(subject.match('hello #one·two·three').to_s).to eq '#one·two·three'
     end
 
     it 'matches ・unicode in ぼっち・ざ・ろっく correctly' do
-      expect(subject.match('testing #ぼっち・ざ・ろっく').to_s).to eq ' #ぼっち・ざ・ろっく'
+      expect(subject.match('testing #ぼっち・ざ・ろっく').to_s).to eq '#ぼっち・ざ・ろっく'
     end
 
     it 'matches ZWNJ' do
-      expect(subject.match('just add #نرم‌افزار and').to_s).to eq ' #نرم‌افزار'
+      expect(subject.match('just add #نرم‌افزار and').to_s).to eq '#نرم‌افزار'
     end
 
     it 'does not match middle dots at the start' do
       expect(subject.match('hello #·one·two·three')).to be_nil
     end
 
     it 'does not match middle dots at the end' do
-      expect(subject.match('hello #one·two·three·').to_s).to eq ' #one·two·three'
+      expect(subject.match('hello #one·two·three·').to_s).to eq '#one·two·three'
     end
 
     it 'does not match purely-numeric hashtags' do