FIX: Make replace watched words work with wildcard (#13084)

Watched words are always regular expressions, despite watched_words_ _regular_expressions being enabled or not. Internally, wildcard characters are replaced with a regular expression that matches any non whitespace character.
discourse · May 18, 2021 · c1dfd76 · c1dfd76
1 parent a21700a
commit c1dfd76
Show file tree

Hide file tree

Showing 5 changed files with 36 additions and 49 deletions.
diff --git a/app/assets/javascripts/pretty-text/engines/discourse-markdown/watched-words-replace.js b/app/assets/javascripts/pretty-text/engines/discourse-markdown/watched-words-replace.js
@@ -6,60 +6,38 @@ function isLinkClose(str) {
   return /^<\/a\s*>/i.test(str);
 }
 
-function findAllMatches(text, matchers, useRegExp) {
+function findAllMatches(text, matchers) {
   const matches = [];
 
-  if (useRegExp) {
-    const maxMatches = 100;
-    let count = 0;
-
-    matchers.forEach((matcher) => {
-      let match;
-      while (
-        (match = matcher.pattern.exec(text)) !== null &&
-        count++ < maxMatches
-      ) {
-        matches.push({
-          index: match.index,
-          text: match[0],
-          replacement: matcher.replacement,
-        });
-      }
-    });
-  } else {
-    const lowerText = text.toLowerCase();
-    matchers.forEach((matcher) => {
-      const lowerPattern = matcher.pattern.toLowerCase();
-      let index = -1;
-      while ((index = lowerText.indexOf(lowerPattern, index + 1)) !== -1) {
-        matches.push({
-          index,
-          text: text.substr(index, lowerPattern.length),
-          replacement: matcher.replacement,
-        });
-      }
-    });
-  }
+  const maxMatches = 100;
+  let count = 0;
+
+  matchers.forEach((matcher) => {
+    let match;
+    while (
+      (match = matcher.pattern.exec(text)) !== null &&
+      count++ < maxMatches
+    ) {
+      matches.push({
+        index: match.index,
+        text: match[0],
+        replacement: matcher.replacement,
+      });
+    }
+  });
 
   return matches.sort((a, b) => a.index - b.index);
 }
 
 export function setup(helper) {
-  helper.registerOptions((opts, siteSettings) => {
-    opts.watchedWordsRegularExpressions =
-      siteSettings.watched_words_regular_expressions;
-  });
-
   helper.registerPlugin((md) => {
     const replacements = md.options.discourse.watchedWordsReplacements;
     if (!replacements) {
       return;
     }
 
     const matchers = Object.keys(replacements).map((word) => ({
-      pattern: md.options.discourse.watchedWordsRegularExpressions
-        ? new RegExp(word, "gi")
-        : word,
+      pattern: new RegExp(word, "gi"),
       replacement: replacements[word],
     }));
 
@@ -110,12 +88,7 @@ export function setup(helper) {
           if (currentToken.type === "text") {
             const text = currentToken.content;
             const matches = (cache[text] =
-              cache[text] ||
-              findAllMatches(
-                text,
-                matchers,
-                md.options.discourse.watchedWordsRegularExpressions
-              ));
+              cache[text] || findAllMatches(text, matchers));
 
             // Now split string to nodes
             const nodes = [];

diff --git a/app/serializers/site_serializer.rb b/app/serializers/site_serializer.rb
@@ -187,7 +187,7 @@ def include_shared_drafts_category_id?
   end
 
   def watched_words_replace
-    WordWatcher.get_cached_words(:replace)
+    WordWatcher.word_matcher_regexps(:replace)
   end
 
   private

diff --git a/app/services/word_watcher.rb b/app/services/word_watcher.rb
@@ -51,6 +51,12 @@ def self.word_matcher_regexp(action, raise_errors: false)
     nil # Admin will be alerted via admin_dashboard_data.rb
   end
 
+  def self.word_matcher_regexps(action)
+    if words = get_cached_words(action)
+      words.map { |w, r| [word_to_regexp(w), r] }.to_h
+    end
+  end
+
   def self.word_to_regexp(word)
     if SiteSetting.watched_words_regular_expressions?
       # Strip ruby regexp format if present, we're going to make the whole thing

diff --git a/lib/pretty_text.rb b/lib/pretty_text.rb
@@ -173,7 +173,7 @@ def self.markdown(text, opts = {})
         __optInput.emojiUnicodeReplacer = __emojiUnicodeReplacer;
         __optInput.lookupUploadUrls = __lookupUploadUrls;
         __optInput.censoredRegexp = #{WordWatcher.word_matcher_regexp(:censor)&.source.to_json};
-        __optInput.watchedWordsReplacements = #{WordWatcher.get_cached_words(:replace).to_json};
+        __optInput.watchedWordsReplacements = #{WordWatcher.word_matcher_regexps(:replace).to_json};
       JS
 
       if opts[:topicId]

diff --git a/spec/components/pretty_text_spec.rb b/spec/components/pretty_text_spec.rb
@@ -1401,11 +1401,19 @@ def expect_cooked_match(raw, expected_cooked)
     after(:all) { Discourse.redis.flushdb }
 
     it "replaces words with other words" do
-      Fabricate(:watched_word, action: WatchedWord.actions[:replace], word: "dolor sit", replacement: "something else")
+      Fabricate(:watched_word, action: WatchedWord.actions[:replace], word: "dolor sit*", replacement: "something else")
 
       expect(PrettyText.cook("Lorem ipsum dolor sit amet")).to match_html(<<~HTML)
         <p>Lorem ipsum something else amet</p>
       HTML
+
+      expect(PrettyText.cook("Lorem ipsum dolor sits amet")).to match_html(<<~HTML)
+        <p>Lorem ipsum something else amet</p>
+      HTML
+
+      expect(PrettyText.cook("Lorem ipsum dolor sittt amet")).to match_html(<<~HTML)
+        <p>Lorem ipsum something else amet</p>
+      HTML
     end
 
     it "replaces words with links" do