Improve content search (#7028)

* Improve content search * Fix search with short word * Improve URL highlighting on content search * Test for mysql keyword search
humhub · May 27, 2024 · 7415609 · 7415609
1 parent 873d856
commit 7415609
Show file tree

Hide file tree

Showing 4 changed files with 34 additions and 5 deletions.
diff --git a/protected/humhub/modules/content/search/driver/MysqlDriver.php b/protected/humhub/modules/content/search/driver/MysqlDriver.php
@@ -138,7 +138,10 @@ private function createMysqlFullTextQuery(SearchQuery $query, array $matchFields
         $againstQuery = '';
 
         foreach ($query->terms as $term) {
-            $againstQuery .= '+' . $this->prepareTerm($term) . ' ';
+            if (strlen($term) >= $this->minAndTermLength) {
+                $againstQuery .= '+';// Search with "AND" condition
+            }
+            $againstQuery .= $this->prepareTerm($term) . ' ';
         }
         foreach ($query->notTerms as $term) {
             $againstQuery .= '-' . $this->prepareTerm($term) . ' ';

diff --git a/protected/humhub/modules/content/tests/codeception/unit/search/AbstractDriverTestSuite.php b/protected/humhub/modules/content/tests/codeception/unit/search/AbstractDriverTestSuite.php
@@ -5,7 +5,6 @@
 use humhub\modules\content\models\Content;
 use humhub\modules\content\Module;
 use humhub\modules\content\search\driver\AbstractDriver;
-use humhub\modules\content\search\driver\MysqlDriver;
 use humhub\modules\content\search\ResultSet;
 use humhub\modules\content\search\SearchRequest;
 use humhub\modules\content\services\ContentSearchService;
@@ -91,7 +90,7 @@ private function getSearchRequest(): SearchRequest
         return new SearchRequest();
     }
 
-    private function getSearchResultByKeyword(string $keyword): ResultSet
+    protected function getSearchResultByKeyword(string $keyword): ResultSet
     {
         $request = $this->getSearchRequest();
         $request->keyword = $keyword;

diff --git a/protected/humhub/modules/content/tests/codeception/unit/search/MysqlDriverTest.php b/protected/humhub/modules/content/tests/codeception/unit/search/MysqlDriverTest.php
@@ -2,13 +2,39 @@
 
 namespace humhub\modules\content\tests\codeception\unit\search;
 
+use humhub\modules\content\models\Content;
 use humhub\modules\content\search\driver\AbstractDriver;
 use humhub\modules\content\search\driver\MysqlDriver;
+use humhub\modules\post\models\Post;
+use humhub\modules\space\models\Space;
 
 class MysqlDriverTest extends AbstractDriverTestSuite
 {
     protected function createDriver(): AbstractDriver
     {
         return new MysqlDriver();
     }
+
+    public function testKeywords()
+    {
+        parent::testKeywords();
+
+        $space = Space::findOne(['id' => 1]);
+
+        (new Post($space, Content::VISIBILITY_PUBLIC, ['message' => 'Apple & Banana']))->save();
+        $this->assertCount(1, $this->getSearchResultByKeyword('Apple & Banana')->results);
+        $this->assertCount(1, $this->getSearchResultByKeyword('"Apple & Banana"')->results);
+
+        (new Post($space, Content::VISIBILITY_PUBLIC, ['message' => 'Here\'s a sentence']))->save();
+        $this->assertCount(1, $this->getSearchResultByKeyword('Here\'s a sentence')->results);
+        $this->assertCount(1, $this->getSearchResultByKeyword('"Here\'s a sentence"')->results);
+
+        (new Post($space, Content::VISIBILITY_PUBLIC, ['message' => 'a ab bc']))->save();
+        $this->assertCount(0, $this->getSearchResultByKeyword('a ab bc')->results);
+        $this->assertCount(0, $this->getSearchResultByKeyword('"a ab bc"')->results);
+
+        (new Post($space, Content::VISIBILITY_PUBLIC, ['message' => 'Test with `code1234` in text']))->save();
+        $this->assertCount(0, $this->getSearchResultByKeyword('with code1234')->results);
+        $this->assertCount(1, $this->getSearchResultByKeyword('"with code1234"')->results);
+    }
 }
diff --git a/static/js/humhub/humhub.ui.additions.js b/static/js/humhub/humhub.ui.additions.js
@@ -304,14 +304,15 @@ humhub.module('ui.additions', function (module, require, $) {
         }
 
         if (typeof words === 'string' && words !== '') {
-            words = words.split(/[^\p{Script=Latin}\d\-’']+/u)
+            words = words.match(/[^\s]+\/[^\s]+|"[^"]+"|[\p{L}\d]+(?:['’`]\p{L}+)?/gu)
+                .map(item => item.replace(/"/g, ''));
+            words = [...new Set(words)].sort((a, b) => b.length - a.length);
         }
         if (!Array.isArray(words)) {
             return;
         }
 
         words.forEach(function (word) {
-            word = word.replace(/^([^a-z\d]*)(.+?)([^a-z\d]*)$/i, '$2');
             $node.highlight(word);
             word.indexOf("'") > -1 && $node.highlight(word.replace("'", '’'));
             word.indexOf("’") > -1 && $node.highlight(word.replace('’', "'"));