Merge branch 'fixDetection-873' into 'main'

Resolve problemas de detecção

See merge request softwares-pkp/plugins_ojs/verificacao-metadados-documento!34

Author: JhonathanLepidus

classes/ContentParser.php

@@ -51,6 +51,7 @@ private function parseWordsFromString($string)
 
     private function parseLine($line)
     {
+        $line = preg_replace('/\x{200B}/u', '', $line);
         $lineWords = $this->parseWordsFromString($line);
 
         if (!empty($lineWords) && is_numeric($lineWords[0])) {
@@ -63,7 +64,7 @@ private function parseLine($line)
     public function parseDocument($pathFile)
     {
         $pathTxt = substr($pathFile, 0, -3) . 'txt';
-        shell_exec("pdftotext " . $pathFile . " " . $pathTxt . " -layout 2>/dev/null");
+        shell_exec("pdftotext " . $pathFile . " " . $pathTxt . " -raw 2>/dev/null");
 
         $docText = file_get_contents($pathTxt);
         $docLines = preg_split("/\r\n|\n|\r/", $docText);

classes/DocumentChecker.php

@@ -129,19 +129,21 @@ private function checkForPattern($patterns, $limit, $limiarForWord, $limiarForPa
     {
         for ($i = 0; $i < count($this->words) - $limit; $i++) {
             for ($j = 0; $j < count($patterns); $j++) {
-                $depth = 0;
+                $depth = $similarWords = 0;
 
                 foreach ($patterns[$j] as $wordPattern) {
                     $wordFromText = $this->words[$i + $depth];
                     similar_text($wordFromText, $wordPattern, $similarity);
-                    if ($similarity < $limiarForWord) {
+                    $depth++;
+
+                    if ($similarity < $limiarForWord && $similarWords == 0) {
                         break;
-                    } else {
-                        $depth++;
+                    } else if ($similarity >= $limiarForWord) {
+                        $similarWords++;
                     }
                 }
 
-                if ($depth / count($patterns[$j]) >= $limiarForPattern) {
+                if ($similarWords / count($patterns[$j]) >= $limiarForPattern) {
                     return 'Success';
                 }
             }

classes/DocumentChecklist.php

@@ -102,5 +102,4 @@ private function getTitleInEnglishStatus($submission)
             'titleInEnglish' => $titleInEnglish
         ];
     }
-
 }

cypress/tests/Test3_nonArticleChecklist.cy.js

@@ -72,7 +72,7 @@ describe('Content Analysis Plugin - Non-article checklist execution', function()
         });
 
         cy.get('#userDetailsForm .submitFormButton').click();
-        cy.waitJQuery();
+        cy.wait(2000);
     });
     it('Non-article checklist execution on PDF without any patterns', function() {
         cy.login('eostrom', null, 'publicknowledge');

tests/DetectionOnDocumentTest.php

@@ -19,7 +19,7 @@ public function setUp(): void
 
     protected function insertWordsIntoDocWordList($words, $docWordList)
     {
-        $middlePosition = (int) count($docWordList) / 2;
+        $middlePosition = (int) (count($docWordList) / 2);
 
         return array_merge(
             array_slice($docWordList, 0, $middlePosition),

version.xml

@@ -12,8 +12,8 @@
 <version>
 	<application>contentAnalysis</application>
 	<type>plugins.generic</type>
-	<release>2.2.3.2</release>
-	<date>2025-12-10</date>
+	<release>2.2.4.0</release>
+	<date>2026-01-06</date>
 	<lazy-load>1</lazy-load>
 	<class>ContentAnalysisPlugin</class>
 </version>