Merge branch 'fixDataDetection340-901' into 'stable-3_4_0'

Estende a verificação do PDF para utilizar dupla-conversão do PDF - 3.4.0

See merge request softwares-pkp/plugins_ojs/verificacao-metadados-documento!39

Author: JhonathanLepidus

classes/ContentParser.php

@@ -69,10 +69,11 @@ private function parseLine($line)
         return $lineWords;
     }
 
-    public function parseDocument($pathFile)
+    public function parseDocument($pathFile, $useRawMode = true)
     {
         $pathTxt = substr($pathFile, 0, -3) . 'txt';
-        shell_exec("pdftotext " . $pathFile . " " . $pathTxt . " -raw 2>/dev/null");
+        $command = "pdftotext $pathFile $pathTxt" . ($useRawMode ? ' -raw ' : '') . " 2>/dev/null";
+        shell_exec($command);
 
         $docText = file_get_contents($pathTxt);
         $docLines = preg_split("/\r\n|\n|\r/", $docText);

classes/DocumentChecker.php

@@ -17,12 +17,14 @@ class DocumentChecker
 {
     private $pathFile;
     public $words;
+    public $secondaryWords;
 
     public function __construct($path)
     {
         $this->pathFile = $path;
         $parser = new ContentParser();
         $this->words = $parser->parseDocument($path);
+        $this->secondaryWords = $parser->parseDocument($path, false);
     }
 
     private $patternsContribution = [
@@ -125,14 +127,14 @@ private function checkOrcid($text)
         ["dados", "da", "pesquisa"]
     ];
 
-    private function checkForPattern($patterns, $limit, $limiarForWord, $limiarForPattern)
+    private function checkForPatternsInWordList($wordList, $patterns, $limit, $limiarForWord, $limiarForPattern)
     {
-        for ($i = 0; $i < count($this->words) - $limit; $i++) {
+        for ($i = 0; $i < count($wordList) - $limit; $i++) {
             for ($j = 0; $j < count($patterns); $j++) {
                 $depth = $similarWords = 0;
 
                 foreach ($patterns[$j] as $wordPattern) {
-                    $wordFromText = $this->words[$i + $depth];
+                    $wordFromText = $wordList[$i + $depth];
                     similar_text($wordFromText, $wordPattern, $similarity);
                     $depth++;
 
@@ -152,9 +154,20 @@ private function checkForPattern($patterns, $limit, $limiarForWord, $limiarForPa
         return 'Error';
     }
 
+    public function checkForPatterns($patterns, $limit, $limiarForWord, $limiarForPattern)
+    {
+        $result = $this->checkForPatternsInWordList($this->words, $patterns, $limit, $limiarForWord, $limiarForPattern);
+
+        if ($result == 'Error') {
+            $result = $this->checkForPatternsInWordList($this->secondaryWords, $patterns, $limit, $limiarForWord, $limiarForPattern);
+        }
+
+        return $result;
+    }
+
     public function checkAuthorsContribution()
     {
-        return $this->checkForPattern($this->patternsContribution, 3, 75, 1);
+        return $this->checkForPatterns($this->patternsContribution, 3, 75, 1);
     }
 
     public function checkTextOrcidsNumber()
@@ -176,17 +189,17 @@ public function checkTextOrcidsNumber()
 
     public function checkConflictInterest()
     {
-        return $this->checkForPattern($this->patternsConflictInterest, 3, 75, 1);
+        return $this->checkForPatterns($this->patternsConflictInterest, 3, 75, 1);
     }
 
     public function checkKeywordsInEnglish()
     {
-        return $this->checkForPattern($this->patternsKeywordsEnglish, 2, 92, 1);
+        return $this->checkForPatterns($this->patternsKeywordsEnglish, 2, 92, 1);
     }
 
     public function checkAbstractInEnglish()
     {
-        return $this->checkForPattern($this->patternsAbstractEnglish, 2, 92, 1);
+        return $this->checkForPatterns($this->patternsAbstractEnglish, 2, 92, 1);
     }
 
     public function checkTitleInEnglish($title)
@@ -199,16 +212,16 @@ public function checkTitleInEnglish($title)
         $cleanedTitle = $parser->cleanStyledText($title);
         $patternTitle = $parser->createPatternFromString($cleanedTitle);
 
-        return $this->checkForPattern(array($patternTitle), count($patternTitle), 75, 0.75);
+        return $this->checkForPatterns(array($patternTitle), count($patternTitle), 75, 0.75);
     }
 
     public function checkEthicsCommittee()
     {
-        return $this->checkForPattern($this->patternsEthicsCommittee, 2, 75, 1);
+        return $this->checkForPatterns($this->patternsEthicsCommittee, 2, 75, 1);
     }
 
     public function checkDataStatement()
     {
-        return $this->checkForPattern($this->patternsDataStatement, 3, 90, 1);
+        return $this->checkForPatterns($this->patternsDataStatement, 3, 90, 1);
     }
 }

tests/DetectionOnDocumentTest.php

@@ -5,7 +5,7 @@
 use PHPUnit\Framework\TestCase;
 use APP\plugins\generic\contentAnalysis\classes\DocumentChecker;
 
-abstract class DetectionOnDocumentTest extends TestCase
+class DetectionOnDocumentTest extends TestCase
 {
     protected const FIXTURES_PATH = __DIR__ . DIRECTORY_SEPARATOR . 'fixtures' . DIRECTORY_SEPARATOR;
     protected $documentChecker;
@@ -28,8 +28,30 @@ protected function insertWordsIntoDocWordList($words, $docWordList)
         );
     }
 
-    protected function insertStringIntoTextHtml($string, $textHtml)
+    public function testCheckerHasWords(): void
     {
-        return $textHtml . " " . $string;
+        $this->assertNotNull($this->documentChecker->words);
+        $this->assertNotEmpty($this->documentChecker->words);
+
+        $this->assertNotNull($this->documentChecker->secondaryWords);
+        $this->assertNotEmpty($this->documentChecker->secondaryWords);
+    }
+
+    public function testGeneralPatternDetectionOnPrimaryWords(): void
+    {
+        $pattern = ['expected', 'pattern', 'to', 'be', 'found'];
+        $backupSecondaryWords = $this->documentChecker->secondaryWords;
+
+        $checkResult = $this->documentChecker->checkForPatterns([$pattern], 5, 50, 1);
+        $this->assertEquals('Error', $checkResult);
+
+        $this->documentChecker->secondaryWords = $this->insertWordsIntoDocWordList($pattern, $this->documentChecker->secondaryWords);
+        $checkResult = $this->documentChecker->checkForPatterns([$pattern], 5, 50, 1);
+        $this->assertEquals('Success', $checkResult);
+
+        $this->documentChecker->secondaryWords = $backupSecondaryWords;
+        $this->documentChecker->words = $this->insertWordsIntoDocWordList($pattern, $this->documentChecker->words);
+        $checkResult = $this->documentChecker->checkForPatterns([$pattern], 5, 50, 1);
+        $this->assertEquals('Success', $checkResult);
     }
 }

version.xml

@@ -12,8 +12,8 @@
 <version>
 	<application>contentAnalysis</application>
 	<type>plugins.generic</type>
-	<release>2.2.5.0</release>
-	<date>2026-02-27</date>
+	<release>2.2.6.0</release>
+	<date>2026-04-23</date>
 	<lazy-load>1</lazy-load>
 	<class>ContentAnalysisPlugin</class>
 </version>