Merge branch 'master' into feature/php-8.4-support

smalot · Oct 24, 2024 · 304cd3d · 304cd3d
2 parents fabd628 + 15da82a
commit 304cd3d
Show file tree

Hide file tree

Showing 6 changed files with 34 additions and 11 deletions.
diff --git a/.php-cs-fixer.php b/.php-cs-fixer.php
@@ -14,10 +14,8 @@
 $config = new Config();
 $config
     ->setFinder($finder)
-    ->setRiskyAllowed(true)
     ->setRules([
-        '@Symfony' => true,
-        '@Symfony:risky' => true,
+        '@PSR12' => true,
         'array_syntax' => ['syntax' => 'short'],
         'no_empty_phpdoc' => true,
         'no_unused_imports' => true,

diff --git a/samples/bugs/Issue727.pdf b/samples/bugs/Issue727.pdf
diff --git a/src/Smalot/PdfParser/Document.php b/src/Smalot/PdfParser/Document.php
@@ -255,7 +255,7 @@ public function extractXMPMetadata(string $content): void
                             if ('rdf:li' == $val['tag']) {
                                 $metadata[] = $val['value'];
 
-                            // Else assign a value to this property
+                                // Else assign a value to this property
                             } else {
                                 $metadata[$val['tag']] = $val['value'];
                             }

diff --git a/src/Smalot/PdfParser/PDFObject.php b/src/Smalot/PdfParser/PDFObject.php
@@ -451,7 +451,7 @@ public function getSectionsText(?string $content): array
                 $inTextBlock = true;
                 $sections[] = $line;
 
-            // If an 'ET' is encountered, unset the $inTextBlock flag
+                // If an 'ET' is encountered, unset the $inTextBlock flag
             } elseif ('ET' == $line) {
                 $inTextBlock = false;
                 $sections[] = $line;

diff --git a/src/Smalot/PdfParser/RawData/RawDataParser.php b/src/Smalot/PdfParser/RawData/RawDataParser.php
@@ -214,8 +214,11 @@ protected function decodeXref(string $pdfData, int $startxref, array $xref = [])
                 }
             }
             if (preg_match('/Prev[\s]+([0-9]+)/i', $trailer_data, $matches) > 0) {
-                // get previous xref
-                $xref = $this->getXrefData($pdfData, (int) $matches[1], $xref);
+                $offset = (int) $matches[1];
+                if (0 != $offset) {
+                    // get previous xref
+                    $xref = $this->getXrefData($pdfData, $offset, $xref);
+                }
             }
         } else {
             throw new \Exception('Unable to find trailer');
@@ -264,7 +267,8 @@ protected function decodeXrefStream(string $pdfData, int $startxref, array $xref
             if (
                 ('/' == $v[0])
                 && ('Type' == $v[1])
-                && (isset($sarr[$k + 1])
+                && (
+                    isset($sarr[$k + 1])
                     && '/' == $sarr[$k + 1][0]
                     && 'XRef' == $sarr[$k + 1][1]
                 )
@@ -290,15 +294,17 @@ protected function decodeXrefStream(string $pdfData, int $startxref, array $xref
                     if (
                         '/' == $vdc[0]
                         && 'Columns' == $vdc[1]
-                        && (isset($decpar[$kdc + 1])
+                        && (
+                            isset($decpar[$kdc + 1])
                             && 'numeric' == $decpar[$kdc + 1][0]
                         )
                     ) {
                         $columns = (int) $decpar[$kdc + 1][1];
                     } elseif (
                         '/' == $vdc[0]
                         && 'Predictor' == $vdc[1]
-                        && (isset($decpar[$kdc + 1])
+                        && (
+                            isset($decpar[$kdc + 1])
                             && 'numeric' == $decpar[$kdc + 1][0]
                         )
                     ) {
@@ -404,7 +410,7 @@ protected function decodeXrefStream(string $pdfData, int $startxref, array $xref
                     }
                     $prev_row = $ddata[$k];
                 } // end for each row
-            // complete decoding
+                // complete decoding
             } else {
                 // number of bytes in a row
                 $rowlen = array_sum($wb);

diff --git a/tests/PHPUnit/Integration/RawData/RawDataParserTest.php b/tests/PHPUnit/Integration/RawData/RawDataParserTest.php
@@ -194,4 +194,23 @@ public function testGetXrefDataIssue673(): void
 
         self::assertStringContainsString('6 rue des Goutais', $text);
     }
+
+    /**
+     * Handle self referencing xref
+     *
+     * It seems that some PDF creators output `Prev 0` when there is no previous xref.
+     *
+     * @see https://github.com/smalot/pdfparser/pull/727
+     */
+    public function testDecodeXrefIssue727(): void
+    {
+        $filename = $this->rootDir.'/samples/bugs/Issue727.pdf';
+
+        // Parsing this document would previously cause an infinite loop
+        $parser = $this->getParserInstance();
+        $document = $parser->parseFile($filename);
+        $text = $document->getText();
+
+        self::assertStringContainsString('', $text);
+    }
 }