Merge branch 'fixTitleError-888' into 'main'

Corrige problema de detecção de título em inglês

See merge request softwares-pkp/plugins_ojs/verificacao-metadados-documento!35

Author: JhonathanLepidus

.gitlab-ci.yml

@@ -3,7 +3,7 @@ variables:
 
 include:
   - project: 'documentacao-e-tarefas/modelosparaintegracaocontinua'
-    ref: main
+    ref: stable-3_4_0
     file:
       - 'templates/groups/pkp_plugin.yml'
       - 'templates/groups/ops/unit_tests.yml'

classes/ContentParser.php

@@ -13,11 +13,16 @@
 
 class ContentParser
 {
+    private const ZERO_WIDTH_SPACE = "\x{200B}";
+    private const MIN_WORD_LENGTH = 2;
+
     private function cleanWord($word)
     {
         $patternsToReplace = [
             '“' => '"',
-            '”' => '"'
+            '”' => '"',
+            '‘' => "'",
+            '’' => "'",
         ];
 
         return $this->replacePatternsInText($word, $patternsToReplace);
@@ -40,8 +45,10 @@ private function parseWordsFromString($string)
                 }
 
                 $word = mb_strtolower(substr($string, $wordStart, $wordEnd - $wordStart));
+                if (strlen($word) >= self::MIN_WORD_LENGTH) {
+                    $words[] = $this->cleanWord($word);
+                }
 
-                $words[] = $this->cleanWord($word);
                 $i = $wordEnd;
             }
         }
@@ -51,7 +58,7 @@ private function parseWordsFromString($string)
 
     private function parseLine($line)
     {
-        $zeroWidthSpacePattern = '/\x{200B}/u';
+        $zeroWidthSpacePattern = '/' . self::ZERO_WIDTH_SPACE . '/u';
         $line = preg_replace($zeroWidthSpacePattern, '', $line);
         $lineWords = $this->parseWordsFromString($line);
 
@@ -95,7 +102,11 @@ public function createPatternFromString($string)
                     $end++;
                 }
 
-                $pattern[] = mb_strtolower(substr($string, $start, $end - $start));
+                $word = mb_strtolower(substr($string, $start, $end - $start));
+                if (strlen($word) >= self::MIN_WORD_LENGTH) {
+                    $pattern[] = $word;
+                }
+
                 $i = $end;
             }
         }
@@ -113,7 +124,9 @@ public function cleanStyledText($text)
             '<u>' => '',
             '</u>' => '',
             '“' => '"',
-            '”' => '"'
+            '”' => '"',
+            '‘' => "'",
+            '’' => "'",
         ];
 
         return $this->replacePatternsInText($text, $patternsToReplace);

classes/DocumentChecker.php

@@ -138,7 +138,7 @@ private function checkForPattern($patterns, $limit, $limiarForWord, $limiarForPa
 
                     if ($similarity < $limiarForWord && $similarWords == 0) {
                         break;
-                    } else if ($similarity >= $limiarForWord) {
+                    } elseif ($similarity >= $limiarForWord) {
                         $similarWords++;
                     }
                 }

tests/ContentParserTest.php

@@ -23,7 +23,7 @@ public function testParsingOfDocument(): void
             'dolor',
             'sit',
             'amet",',
-            'consectetur',
+            "'consectetur'",
             'adipiscing',
             'elit.',
             'proin',
@@ -62,7 +62,6 @@ public function testParsingOfLineNumberedDocument(): void
             'ex',
             'libero,',
             'porttitor',
-            'a',
             'elit',
             'eget,',
             'maximus',
@@ -78,7 +77,7 @@ public function testParsingOfLineNumberedDocument(): void
     public function testCreatePatternFromString(): void
     {
         $string = 'Innovations and new advances for this world: a survey';
-        $expectedPattern = ['innovations', 'and', 'new', 'advances', 'for', 'this', 'world:', 'a', 'survey'];
+        $expectedPattern = ['innovations', 'and', 'new', 'advances', 'for', 'this', 'world:', 'survey'];
 
         $patternCreated = $this->contentParser->createPatternFromString($string);
         $this->assertEquals($expectedPattern, $patternCreated);
@@ -97,7 +96,11 @@ public function testCleansOtherCharactersFromTitle(): void
     {
         $title = 'Reflections on “Arrival” and brazilian sign language (LIBRAS)';
         $expectedCleanedTitle = 'Reflections on "Arrival" and brazilian sign language (LIBRAS)';
+        $cleanedTitle = $this->contentParser->cleanStyledText($title);
+        $this->assertEquals($expectedCleanedTitle, $cleanedTitle);
 
+        $title = 'Schindler’s List: ‘absolut cinema’';
+        $expectedCleanedTitle = "Schindler's List: 'absolut cinema'";
         $cleanedTitle = $this->contentParser->cleanStyledText($title);
         $this->assertEquals($expectedCleanedTitle, $cleanedTitle);
     }

tests/fixtures/dummy_document.pdf

@@ -12,8 +12,8 @@
 <version>
 	<application>contentAnalysis</application>
 	<type>plugins.generic</type>
-	<release>2.2.4.0</release>
-	<date>2026-01-06</date>
+	<release>2.2.5.0</release>
+	<date>2026-02-27</date>
 	<lazy-load>1</lazy-load>
 	<class>ContentAnalysisPlugin</class>
 </version>