FIX: Split link watched words from replace (#13196)

It was not clear that replace watched words can be used to replace text
with URLs. This introduces a new watched word type that makes it easier
to understand.

app/assets/javascripts/admin/addon/components/admin-watched-word.js

@@ -9,6 +9,7 @@ export default Component.extend({
 
   isReplace: equal("actionKey", "replace"),
   isTag: equal("actionKey", "tag"),
+  isLink: equal("actionKey", "link"),
 
   @discourseComputed("word.replacement")
   tags(replacement) {

app/assets/javascripts/admin/addon/components/watched-word-form.js

@@ -15,9 +15,16 @@ export default Component.extend({
   formSubmitted: false,
   actionKey: null,
   showMessage: false,
+  selectedTags: null,
 
   canReplace: equal("actionKey", "replace"),
   canTag: equal("actionKey", "tag"),
+  canLink: equal("actionKey", "link"),
+
+  didInsertElement() {
+    this._super(...arguments);
+    this.set("selectedTags", []);
+  },
 
   @discourseComputed("siteSettings.watched_words_regular_expressions")
   placeholderKey(watchedWordsRegularExpressions) {
@@ -47,6 +54,13 @@ export default Component.extend({
   },
 
   actions: {
+    changeSelectedTags(tags) {
+      this.setProperties({
+        selectedTags: tags,
+        replacement: tags.join(","),
+      });
+    },
+
     submit() {
       if (!this.isUniqueWord) {
         this.setProperties({
@@ -61,7 +75,10 @@ export default Component.extend({
 
         const watchedWord = WatchedWord.create({
           word: this.word,
-          replacement: this.canReplace || this.canTag ? this.replacement : null,
+          replacement:
+            this.canReplace || this.canTag || this.canLink
+              ? this.replacement
+              : null,
           action: this.actionKey,
         });
 

app/assets/javascripts/admin/addon/controllers/modals/admin-watched-word-test.js

@@ -6,15 +6,17 @@ import { equal } from "@ember/object/computed";
 export default Controller.extend(ModalFunctionality, {
   isReplace: equal("model.nameKey", "replace"),
   isTag: equal("model.nameKey", "tag"),
+  isLink: equal("model.nameKey", "link"),
 
   @discourseComputed(
     "value",
     "model.compiledRegularExpression",
     "model.words",
     "isReplace",
-    "isTag"
+    "isTag",
+    "isLink"
   )
-  matches(value, regexpString, words, isReplace, isTag) {
+  matches(value, regexpString, words, isReplace, isTag, isLink) {
     if (!value || !regexpString) {
       return;
     }
@@ -22,7 +24,7 @@ export default Controller.extend(ModalFunctionality, {
     const regexp = new RegExp(regexpString, "ig");
     const matches = value.match(regexp) || [];
 
-    if (isReplace) {
+    if (isReplace || isLink) {
       return matches.map((match) => ({
         match,
         replacement: words.find((word) =>

app/assets/javascripts/admin/addon/templates/components/admin-watched-word.hbs

@@ -1,5 +1,5 @@
 {{d-icon "times"}} {{word.word}}
-{{#if isReplace}}
+{{#if (or isReplace isLink)}}
   &rarr; <span class="replacement">{{word.replacement}}</span>
 {{else if isTag}}
   &rarr;

app/assets/javascripts/admin/addon/templates/components/watched-word-form.hbs

@@ -5,15 +5,29 @@
 
 {{#if canReplace}}
   <div class="watched-word-input">
-    <label for="watched-replacement">{{i18n "admin.watched_words.form.replacement_label"}}</label>
-    {{text-field id="watched-replacement" value=replacement disabled=formSubmitted class="watched-word-input-field" autocorrect="off" autocapitalize="off" placeholderKey="admin.watched_words.form.replacement_placeholder"}}
+    <label for="watched-replacement">{{i18n "admin.watched_words.form.replace_label"}}</label>
+    {{text-field id="watched-replacement" value=replacement disabled=formSubmitted class="watched-word-input-field" autocorrect="off" autocapitalize="off" placeholderKey="admin.watched_words.form.replace_placeholder"}}
   </div>
 {{/if}}
 
 {{#if canTag}}
   <div class="watched-word-input">
     <label for="watched-tag">{{i18n "admin.watched_words.form.tag_label"}}</label>
-    {{text-field id="watched-tag" value=replacement disabled=formSubmitted class="watched-word-input-field" autocorrect="off" autocapitalize="off" placeholderKey="admin.watched_words.form.tag_placeholder"}}
+    {{tag-chooser
+      id="watched-tag"
+      class="watched-word-input-field"
+      allowCreate=true
+      disabled=formSubmitted
+      tags=selectedTags
+      onChange=(action "changeSelectedTags")
+    }}
+  </div>
+{{/if}}
+
+{{#if canLink}}
+  <div class="watched-word-input">
+    <label for="watched-replacement">{{i18n "admin.watched_words.form.link_label"}}</label>
+    {{text-field id="watched-replacement" value=replacement disabled=formSubmitted class="watched-word-input-field" autocorrect="off" autocapitalize="off" placeholderKey="admin.watched_words.form.link_placeholder"}}
   </div>
 {{/if}}
 

app/assets/javascripts/admin/addon/templates/modal/admin-watched-word-test.hbs

@@ -5,7 +5,7 @@
     <p>
       {{i18n "admin.watched_words.test.found_matches"}}
       <ul>
-        {{#if isReplace}}
+        {{#if (or isReplace isLink)}}
           {{#each matches as |match|}}
             <li>
               <span class="match">{{match.match}}</span>

app/assets/javascripts/admin/addon/templates/watched-words-action.hbs

@@ -40,7 +40,7 @@
 {{/if}}
 
 {{#if showWordsList}}
-  <div class="watched-words-list">
+  <div class="watched-words-list watched-words-{{actionNameKey}}">
     {{#each currentAction.words as |word| }}
       <div class="watched-word-box">{{admin-watched-word actionKey=actionNameKey word=word action=(action "recordRemoved")}}</div>
     {{/each}}

app/assets/javascripts/discourse/app/lib/text.js

@@ -21,7 +21,8 @@ function getOpts(opts) {
       customEmojiTranslation: context.site.custom_emoji_translation,
       siteSettings: context.siteSettings,
       formatUsername,
-      watchedWordsReplacements: context.site.watched_words_replace,
+      watchedWordsReplace: context.site.watched_words_replace,
+      watchedWordsLink: context.site.watched_words_link,
     },
     opts
   );

app/assets/javascripts/discourse/tests/unit/lib/pretty-text-test.js

@@ -1675,17 +1675,29 @@ var bar = 'bar';
 
   test("watched words replace", function (assert) {
     const opts = {
-      watchedWordsReplacements: { fun: "times" },
+      watchedWordsReplace: { fun: "times" },
     };
 
     assert.cookedOptions("test fun", opts, "<p>test times</p>");
   });
 
+  test("watched words link", function (assert) {
+    const opts = {
+      watchedWordsLink: { fun: "https://discourse.org" },
+    };
+
+    assert.cookedOptions(
+      "test fun",
+      opts,
+      '<p>test <a href="https://discourse.org">fun</a></p>'
+    );
+  });
+
   test("watched words replace with bad regex", function (assert) {
     const maxMatches = 100; // same limit as MD watched-words-replace plugin
     const opts = {
       siteSettings: { watched_words_regular_expressions: true },
-      watchedWordsReplacements: { "\\bu?\\b": "you" },
+      watchedWordsReplace: { "\\bu?\\b": "you" },
     };
 
     assert.cookedOptions(

app/assets/javascripts/pretty-text/addon/pretty-text.js

@@ -33,7 +33,8 @@ export function buildOptions(state) {
     censoredRegexp,
     disableEmojis,
     customEmojiTranslation,
-    watchedWordsReplacements,
+    watchedWordsReplace,
+    watchedWordsLink,
   } = state;
 
   let features = {
@@ -83,7 +84,8 @@ export function buildOptions(state) {
       siteSettings.enable_advanced_editor_preview_sync,
     previewing,
     disableEmojis,
-    watchedWordsReplacements,
+    watchedWordsReplace,
+    watchedWordsLink,
   };
 
   // note, this will mutate options due to the way the API is designed

app/assets/javascripts/pretty-text/engines/discourse-markdown/watched-words.js

@@ -23,6 +23,7 @@ function findAllMatches(text, matchers) {
         index: match.index,
         text: match[0],
         replacement: matcher.replacement,
+        link: matcher.link,
       });
     }
   });
@@ -32,19 +33,39 @@ function findAllMatches(text, matchers) {
 
 export function setup(helper) {
   helper.registerPlugin((md) => {
-    const replacements = md.options.discourse.watchedWordsReplacements;
-    if (!replacements) {
+    const matchers = [];
+
+    if (md.options.discourse.watchedWordsReplace) {
+      Object.entries(md.options.discourse.watchedWordsReplace).map(
+        ([word, replacement]) => {
+          matchers.push({
+            pattern: new RegExp(word, "gi"),
+            replacement,
+            link: false,
+          });
+        }
+      );
+    }
+
+    if (md.options.discourse.watchedWordsLink) {
+      Object.entries(md.options.discourse.watchedWordsLink).map(
+        ([word, replacement]) => {
+          matchers.push({
+            pattern: new RegExp(word, "gi"),
+            replacement,
+            link: true,
+          });
+        }
+      );
+    }
+
+    if (matchers.length === 0) {
       return;
     }
 
-    const matchers = Object.keys(replacements).map((word) => ({
-      pattern: new RegExp(word, "gi"),
-      replacement: replacements[word],
-    }));
-
     const cache = {};
 
-    md.core.ruler.push("watched-words-replace", (state) => {
+    md.core.ruler.push("watched-words", (state) => {
       for (let j = 0, l = state.tokens.length; j < l; j++) {
         if (state.tokens[j].type !== "inline") {
           continue;
@@ -82,10 +103,6 @@ export function setup(helper) {
             }
           }
 
-          if (htmlLinkLevel > 0) {
-            continue;
-          }
-
           if (currentToken.type === "text") {
             const text = currentToken.content;
             const matches = (cache[text] =
@@ -109,25 +126,27 @@ export function setup(helper) {
                 nodes.push(token);
               }
 
-              let url = state.md.normalizeLink(matches[ln].replacement);
-              if (state.md.validateLink(url) && /^https?/.test(url)) {
-                token = new state.Token("link_open", "a", 1);
-                token.attrs = [["href", url]];
-                token.level = level++;
-                token.markup = "linkify";
-                token.info = "auto";
-                nodes.push(token);
+              if (matches[ln].link) {
+                const url = state.md.normalizeLink(matches[ln].replacement);
+                if (htmlLinkLevel === 0 && state.md.validateLink(url)) {
+                  token = new state.Token("link_open", "a", 1);
+                  token.attrs = [["href", url]];
+                  token.level = level++;
+                  token.markup = "linkify";
+                  token.info = "auto";
+                  nodes.push(token);
 
-                token = new state.Token("text", "", 0);
-                token.content = matches[ln].text;
-                token.level = level;
-                nodes.push(token);
+                  token = new state.Token("text", "", 0);
+                  token.content = matches[ln].text;
+                  token.level = level;
+                  nodes.push(token);
 
-                token = new state.Token("link_close", "a", -1);
-                token.level = --level;
-                token.markup = "linkify";
-                token.info = "auto";
-                nodes.push(token);
+                  token = new state.Token("link_close", "a", -1);
+                  token.level = --level;
+                  token.markup = "linkify";
+                  token.info = "auto";
+                  nodes.push(token);
+                }
               } else {
                 token = new state.Token("text", "", 0);
                 token.content = matches[ln].replacement;

app/assets/stylesheets/common/admin/staff_logs.scss

@@ -332,6 +332,19 @@ table.screened-ip-addresses {
   vertical-align: top;
 }
 
+.watched-words-link {
+  .watched-word-box {
+    min-width: 100%;
+  }
+}
+
+.watched-words-replace,
+.watched-words-tag {
+  .watched-word-box {
+    min-width: calc(50% - 5px);
+  }
+}
+
 .watched-word-box,
 .watched-words-test-modal {
   .replacement {
@@ -354,6 +367,7 @@ table.screened-ip-addresses {
 .watched-words-list {
   margin-top: 20px;
   display: inline-block;
+  width: 100%;
 }
 
 .watched-word {
@@ -396,6 +410,9 @@ table.screened-ip-addresses {
   input.watched-word-input-field {
     min-width: 300px;
   }
+  .select-kit.multi-select.watched-word-input-field {
+    width: 300px;
+  }
 }
 
 // Search logs

app/models/watched_word.rb

@@ -8,9 +8,10 @@ class WatchedWord < ActiveRecord::Base
       censor: 2,
       require_approval: 3,
       flag: 4,
+      link: 8,
       replace: 5,
       tag: 6,
-      silence: 7
+      silence: 7,
     )
   end
 
@@ -18,10 +19,16 @@ class WatchedWord < ActiveRecord::Base
 
   before_validation do
     self.word = self.class.normalize_word(self.word)
+    if self.action == WatchedWord.actions[:link] && !(self.replacement =~ /^https?:\/\//)
+      self.replacement = "#{Discourse.base_url}#{self.replacement.starts_with?("/") ? "" : "/"}#{self.replacement}"
+    end
   end
 
   validates :word,   presence: true, uniqueness: true, length: { maximum: 100 }
   validates :action, presence: true
+
+  validate :replacement_is_url, if: -> { action == WatchedWord.actions[:link] }
+
   validates_each :word do |record, attr, val|
     if WatchedWord.where(action: record.action).count >= MAX_WORDS_PER_ACTION
       record.errors.add(:word, :too_many)
@@ -37,6 +44,12 @@ class WatchedWord < ActiveRecord::Base
     w.strip.squeeze('*')
   end
 
+  def replacement_is_url
+    if !(replacement =~ URI::regexp)
+      errors.add(:base, :invalid_url)
+    end
+  end
+
   def self.create_or_update_word(params)
     new_word = normalize_word(params[:word])
     w = WatchedWord.where("word ILIKE ?", new_word).first || WatchedWord.new(word: new_word)
@@ -48,7 +61,7 @@ class WatchedWord < ActiveRecord::Base
   end
 
   def self.has_replacement?(action)
-    action == :replace || action == :tag
+    action == :replace || action == :tag || action == :link
   end
 
   def action_key=(arg)

app/serializers/site_serializer.rb

@@ -29,7 +29,8 @@ class SiteSerializer < ApplicationSerializer
     :censored_regexp,
     :shared_drafts_category_id,
     :custom_emoji_translation,
-    :watched_words_replace
+    :watched_words_replace,
+    :watched_words_link
   )
 
   has_many :categories, serializer: SiteCategorySerializer, embed: :objects
@@ -185,6 +186,10 @@ class SiteSerializer < ApplicationSerializer
     WordWatcher.word_matcher_regexps(:replace)
   end
 
+  def watched_words_link
+    WordWatcher.word_matcher_regexps(:link)
+  end
+
   private
 
   def ordered_flags(flags)

app/services/word_watcher.rb

@@ -8,7 +8,7 @@ class WordWatcher
 
   def self.words_for_action(action)
     words = WatchedWord.where(action: WatchedWord.actions[action.to_sym]).limit(1000)
-    if action.to_sym == :replace || action.to_sym == :tag
+    if WatchedWord.has_replacement?(action.to_sym)
       words.pluck(:word, :replacement).to_h
     else
       words.pluck(:word)
@@ -31,7 +31,7 @@ class WordWatcher
   def self.word_matcher_regexp(action, raise_errors: false)
     words = get_cached_words(action)
     if words
-      if action.to_sym == :replace || action.to_sym == :tag
+      if WatchedWord.has_replacement?(action.to_sym)
         words = words.keys
       end
       words = words.map do |w|

config/locales/client.en.yml

@@ -4741,22 +4741,25 @@ en:
           replace: "Replace"
           tag: "Tag"
           silence: "Silence"
+          link: "Link"
         action_descriptions:
           block: "Prevent posts containing these words from being posted. The user will see an error message when they try to submit their post."
           censor: "Allow posts containing these words, but replace them with characters that hide the censored words."
           require_approval: "Posts containing these words will require approval by staff before they can be seen."
           flag: "Allow posts containing these words, but flag them as inappropriate so moderators can review them."
-          replace: "Replace words in posts with other words or links"
+          replace: "Replace words in posts with other words"
           tag: "Automatically tag topics based on first post"
           silence: "First posts of users containing these words will require approval by staff before they can be seen and the user will be automatically silenced."
+          link: "Replace words in posts with links"
         form:
           label: "Has word or phrase"
           placeholder: "Enter word or phrase (* is a wildcard)"
           placeholder_regexp: "regular expression"
-          replacement_label: "Replacement"
-          replacement_placeholder: "example or https://example.com"
+          replace_label: "Replacement"
+          replace_placeholder: "example"
           tag_label: "Tag"
-          tag_placeholder: "tag1,tag2,tag3"
+          link_label: "Link"
+          link_placeholder: "https://example.com"
           add: "Add"
           success: "Success"
           exists: "Already exists"

config/locales/server.en.yml

@@ -637,6 +637,8 @@ en:
           attributes:
             word:
               too_many: "Too many words for that action"
+            base:
+              invalid_url: "Replacement URL is invalid"
 
   uncategorized_category_name: "Uncategorized"
 

db/migrate/20210528144647_migrate_watched_words_from_replace_to_link.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+class MigrateWatchedWordsFromReplaceToLink < ActiveRecord::Migration[6.1]
+  def up
+    execute <<~SQL
+      UPDATE watched_words
+      SET action = 8
+      WHERE action = 5 AND replacement ILIKE 'http%'
+    SQL
+  end
+
+  def down
+    execute("UPDATE watched_words SET action = 5 WHERE action = 8")
+  end
+end

lib/pretty_text.rb

@@ -173,7 +173,8 @@ module PrettyText
         __optInput.emojiUnicodeReplacer = __emojiUnicodeReplacer;
         __optInput.lookupUploadUrls = __lookupUploadUrls;
         __optInput.censoredRegexp = #{WordWatcher.word_matcher_regexp(:censor)&.source.to_json};
-        __optInput.watchedWordsReplacements = #{WordWatcher.word_matcher_regexps(:replace).to_json};
+        __optInput.watchedWordsReplace = #{WordWatcher.word_matcher_regexps(:replace).to_json};
+        __optInput.watchedWordsLink = #{WordWatcher.word_matcher_regexps(:link).to_json};
       JS
 
       if opts[:topicId]

spec/components/pretty_text_spec.rb

@@ -1403,7 +1403,7 @@ HTML
     end
   end
 
-  describe "watched words - replace" do
+  describe "watched words - replace & link" do
     after(:all) { Discourse.redis.flushdb }
 
     it "replaces words with other words" do
@@ -1423,7 +1423,7 @@ HTML
     end
 
     it "replaces words with links" do
-      Fabricate(:watched_word, action: WatchedWord.actions[:replace], word: "meta", replacement: "https://meta.discourse.org")
+      Fabricate(:watched_word, action: WatchedWord.actions[:link], word: "meta", replacement: "https://meta.discourse.org")
 
       expect(PrettyText.cook("Meta is a Discourse forum")).to match_html(<<~HTML)
         <p>
@@ -1446,14 +1446,14 @@ HTML
     end
 
     it "supports overlapping words" do
-      Fabricate(:watched_word, action: WatchedWord.actions[:replace], word: "discourse", replacement: "https://discourse.org")
-      Fabricate(:watched_word, action: WatchedWord.actions[:replace], word: "is", replacement: "https://example.com")
+      Fabricate(:watched_word, action: WatchedWord.actions[:link], word: "meta", replacement: "https://meta.discourse.org")
+      Fabricate(:watched_word, action: WatchedWord.actions[:replace], word: "iz", replacement: "is")
+      Fabricate(:watched_word, action: WatchedWord.actions[:link], word: "discourse", replacement: "https://discourse.org")
 
-      expect(PrettyText.cook("Meta is a Discourse forum")).to match_html(<<~HTML)
+      expect(PrettyText.cook("Meta iz a Discourse forum")).to match_html(<<~HTML)
         <p>
-          Meta
-          <a href="https://example.com" rel="noopener nofollow ugc">is</a>
-          a
+          <a href="https://meta.discourse.org" rel="noopener nofollow ugc">Meta</a>
+          is a
           <a href="https://discourse.org" rel="noopener nofollow ugc">Discourse</a>
           forum
         </p>