Fixed a bug with unrelated search results for a '..' query.

src/S2/Rose/Entity/Metadata/SentenceCollection.php

@@ -1,6 +1,6 @@
 <?php declare(strict_types=1);
 /**
- * @copyright 2023 Roman Parpalak
+ * @copyright 2023-2024 Roman Parpalak
  * @license   MIT
  */
 
@@ -104,13 +104,13 @@ public static function breakIntoWords(string $content): array
         // We allow letters, digits and some punctuation: ".,-"
         $content = preg_replace('#[^\\-.,0-9\\p{L}^_]+#u', ' ', $content);
         $content = mb_strtolower($content);
-        $content = str_replace([", ", ". ", "- ", 'ё'], [' ', ' ', ' ', 'е'], $content);
+        $content = str_replace(['ё'], ['е'], $content);
 
         // These punctuation characters are meant to be inside words and numbers.
         // Remove trailing characters when splitting the words.
         $content = rtrim($content, '-.,');
 
-        $words = explode(' ', $content);
+        $words = preg_split('#[\\-.,]*?[ ]+#S', $content);
         StringHelper::removeLongWords($words);
 
         return $words;

src/S2/Rose/Entity/SnippetLine.php

@@ -1,6 +1,6 @@
 <?php declare(strict_types=1);
 /**
- * @copyright 2017-2023 Roman Parpalak
+ * @copyright 2017-2024 Roman Parpalak
  * @license   MIT
  */
 
@@ -87,7 +87,7 @@ public function getHighlighted(string $highlightTemplate, bool $includeFormattin
             // TODO: After implementing formatting this regex became a set of crutches.
             // One has to break the snippets into words, clear formatting, convert words to stems
             // and detect what stems has been found. Then highlight the original text based on words source offset.
-            $wordPattern               = implode('|', $this->foundWords);
+            $wordPattern               = implode('|', array_map(static fn(string $word) => preg_quote($word, '#'), $this->foundWords));
             $wordPatternWithFormatting = '(?:\\\\[' . StringHelper::FORMATTING_SYMBOLS . '])*(?:' . $wordPattern . ')(?:\\\\[' . strtoupper(StringHelper::FORMATTING_SYMBOLS) . '])*';
             $replacedLine              = preg_replace_callback(
                 '#(?:\\s|^|\p{P})\\K' . $wordPatternWithFormatting . '(?:\\s+(?:' . $wordPatternWithFormatting . '))*\\b#su',

src/S2/Rose/Indexer.php

@@ -2,7 +2,7 @@
 /**
  * Creates search index
  *
- * @copyright 2010-2023 Roman Parpalak
+ * @copyright 2010-2024 Roman Parpalak
  * @license   MIT
  */
 
@@ -207,8 +207,8 @@ private function getStemsWithComponents(array $words): array
 
             // If the word contains punctuation marks like hyphen, add a variant without it
             if (false !== strpbrk($stemmedWord, '-.,')) {
-                foreach (preg_split('#[\-.,]#', $word) as $k => $subWord) {
-                    if ($subWord) {
+                foreach (preg_split('#[\p{L}\d]\K[\-.,]+|[\-.,]+(?=[\p{L}\d])#u', $word) as $k => $subWord) {
+                    if ($subWord !== '' && $subWord !== $word) {
                         $componentsOfCompoundWords[(string)($i + 0.001 * ($k + 1))] = $this->stemmer->stemWord($subWord, false);
                     }
                 }

src/S2/Rose/Snippet/SnippetBuilder.php

@@ -1,6 +1,6 @@
 <?php declare(strict_types=1);
 /**
- * @copyright 2011-2023 Roman Parpalak
+ * @copyright 2011-2024 Roman Parpalak
  * @license   MIT
  */
 
@@ -84,6 +84,7 @@ public function buildSnippet(array $foundPositionsByStems, string $highlightTemp
             $stems = array_merge($stems, $this->stemmer->irregularWordsFromStems($stems));
 
             $regexRules = $this->stemmer->getRegexTransformationRules();
+            $regexRules['#\\.#'] = '\\.'; // escaping dot in the following preg_match_all() call
             $stemsForRegex = array_map(static fn(string $stem): string => preg_replace(
                 array_keys($regexRules),
                 array_values($regexRules),

tests/unit/Rose/Entity/QueryTest.php

@@ -1,6 +1,6 @@
 <?php
 /**
- * @copyright 2016-2020 Roman Parpalak
+ * @copyright 2016-2024 Roman Parpalak
  * @license   MIT
  */
 
@@ -14,11 +14,14 @@
  */
 class QueryTest extends Unit
 {
-    public function testFilterInput()
+    public function testFilterInput(): void
     {
         $this->assertEquals([1, 2], (new Query('1|||2'))->valueToArray());
         $this->assertEquals([1, 2], (new Query('1\\\\\\2'))->valueToArray());
         $this->assertEquals(['a', 'b'], (new Query('a/b'))->valueToArray());
         $this->assertEquals(['a', 'b'], (new Query(' a   b   '))->valueToArray());
+        $this->assertEquals(['..'], (new Query('..'))->valueToArray());
+        $this->assertEquals(['...'], (new Query('...'))->valueToArray());
+        $this->assertEquals(['a..b'], (new Query('a..b'))->valueToArray());
     }
 }

tests/unit/Rose/IntegrationTest.php

@@ -2,7 +2,7 @@
 /** @noinspection PhpComposerExtensionStubsInspection */
 
 /**
- * @copyright 2016-2023 Roman Parpalak
+ * @copyright 2016-2024 Roman Parpalak
  * @license   MIT
  */
 
@@ -90,7 +90,7 @@ public function testFeatures(
 
         $this->assertEquals([
             '20:id_2' => 2.5953804134970615,
-            '20:id_1' => 0.12828323517212156,
+            '20:id_1' => 0.12932092968696407,
             '10:id_1' => 0.08569157515491249,
         ], $resultSet2->getSortedRelevanceByExternalId());
 
@@ -112,7 +112,7 @@ public function testFeatures(
 
         $this->assertEquals([
             '20:id_2' => 2.5953804134970615,
-            '20:id_1' => 0.12828323517212156
+            '20:id_1' => 0.12932092968696407
         ], $resultSet2->getSortedRelevanceByExternalId());
 
         $this->assertEquals(3, $resultSet2->getTotalCount());
@@ -265,6 +265,10 @@ public function testFeatures(
         $this->assertEquals('25', $img1->getHeight());
         $this->assertEquals('Alternative text', $img1->getAlt());
 
+        // Empty result
+        $this->assertCount(0, $finder->find(new Query('..'))->getItems());
+        $this->assertCount(0, $finder->find(new Query('...'))->getItems());
+
         if ($readStorage instanceof PdoStorage && strpos($GLOBALS['s2_rose_test_db']['dsn'], 'sqlite') !== 0) {
             $indexer->index(new Indexable('dummy', 'Dummy new', ''));
             $similarItems = $readStorage->getSimilar(new ExternalId('id_2', 20), false);
@@ -418,7 +422,7 @@ public function indexableProvider()
                 ->setDate(new \DateTime('2016-08-24 00:00:00'))
                 ->setUrl('url1')
             ,
-            (new Indexable('id_1', 'Another instance', 'The same id but another instance. Word "content" is present here. Twice: content.', 20))
+            (new Indexable('id_1', 'Another instance', 'The same id but another instance. Word "content" is present here. Twice: content. Delimiters must be $...$ or  \[...\]', 20))
             ,
             (new Indexable('id_4', 'Another instance', 'Nothing is here but images: <img src="1.jpg" width="10" height="15"> <img src="2.jpg" width="20" height="25" alt="Alternative text" />', 20))
             ,