fix(integration): align rebuilt PR809 with validated best-result branch

Author: vitormattos

samples/bugs/rawdata/PullRequest818-pdf.js.pdf

@@ -1427,13 +1427,8 @@ public function parseData(string $data): array
             throw new EmptyPdfException('Empty PDF data given.');
         }
         // find the pdf header starting position
-        $trimpos = strpos($data, '%PDF-');
-        if (false === $trimpos) {
-            if (!$this->hasRecoverablePdfStructureWithoutHeader($data)) {
-                throw new MissingPdfHeaderException('Invalid PDF data: Missing `%PDF-` header.');
-            }
-
-            $trimpos = 0;
+        if (false === ($trimpos = strpos($data, '%PDF-'))) {
+            throw new MissingPdfHeaderException('Invalid PDF data: Missing `%PDF-` header.');
         }
 
         // Keep the original byte layout to preserve absolute xref offsets.
@@ -1483,23 +1478,4 @@ public function parseData(string $data): array
 
         return [$xref, $objects];
     }
-
-    private function hasRecoverablePdfStructureWithoutHeader(string $data): bool
-    {
-        if (
-            preg_match('/(?:^|[\r\n])[0-9]+[\x09\x0a\x0c\x0d\x20]+[0-9]+[\x09\x0a\x0c\x0d\x20]+obj\b/i', $data) === 0
-        ) {
-            return false;
-        }
-
-        if (preg_match('/\btrailer\b/i', $data) === 0) {
-            return false;
-        }
-
-        if (preg_match('/\bstartxref\b/i', $data) === 0 && preg_match('/\bxref\b/i', $data) === 0) {
-            return false;
-        }
-
-        return true;
-    }
 }

src/Smalot/PdfParser/RawData/RawDataParser.php

@@ -36,6 +36,7 @@
 namespace PHPUnitTests\Integration;
 
 use PHPUnitTests\TestCase;
+use Smalot\PdfParser\Config;
 use Smalot\PdfParser\Document;
 use Smalot\PdfParser\Parser;
 
@@ -111,6 +112,19 @@ public function testPDFDocEncodingDecode(): void
         $testSubject = '•†‡…—–ƒ⁄‹›−‰„“”‘’‚™ŁŒŠŸŽıłœšž';
         self::assertStringContainsString($testSubject, $details['Subject']);
     }
+    /**
+     * @see https://github.com/smalot/pdfparser/pull/795
+     * @see https://github.com/smalot/pdfparser/blob/master/samples/bugs/PullRequestDuplicateKids.pdf
+     */
+    public function testGetPagesDeduplicatesDuplicateKidsFixture(): void
+    {
+        $document = (new Parser())->parseFile($this->rootDir.'/samples/bugs/PullRequestDuplicateKids.pdf');
+
+        $pages = $document->getPages();
+
+        self::assertCount(1, $pages);
+    }
+
     public function testRecoverPagesWhenXrefEntriesArePartiallyMissing(): void
     {
         $document = (new Parser())->parseFile($this->rootDir.'/samples/bugs/PullRequest813-pdf.js.pdf');
@@ -124,4 +138,17 @@ public function testRecoverPagesWhenRootOffsetPointsToInvalidObject(): void
 
         self::assertCount(1, $document->getPages());
     }
+
+    /**
+     * @group linux-only
+     */
+    public function testParseFileWithLargeFlateStreams(): void
+    {
+        $config = new Config();
+        $config->setRetainImageContent(false);
+        $config->setDecodeMemoryLimit(8 * 1024 * 1024);
+        $document = (new Parser([], $config))->parseFile($this->rootDir.'/samples/bugs/PullRequest457.pdf');
+
+        self::assertCount(28, $document->getPages());
+    }
 }

tests/PHPUnit/Integration/DocumentIssueFocusTest.php

@@ -5,10 +5,11 @@
  *
  * @author  Konrad Abicht <k.abicht@gmail.com>
  *
+ * @date    2020-06-01
+ *
+ * @author  Sébastien MALOT <sebastien@malot.fr>
  *
  * @date    2017-01-03
-
-        self::assertSame([], $document->getPages());
  *
  * @license LGPLv3
  *
@@ -39,7 +40,6 @@
 use Smalot\PdfParser\Header;
 use Smalot\PdfParser\Page;
 use Smalot\PdfParser\Pages;
-use Smalot\PdfParser\Parser;
 use Smalot\PdfParser\PDFObject;
 
 /**
@@ -225,12 +225,15 @@ public function testGetPages(): void
 
     public function testGetPagesMissingCatalog(): void
     {
+        $this->expectException(\Exception::class);
+        $this->expectExceptionMessage('Missing catalog.');
+
         // Missing catalog
         $document = $this->getDocumentInstance();
-        $this->assertSame([], $document->getPages());
+        $document->getPages();
     }
 
-    public function testGetPagesPreservesDuplicateKidsReferences(): void
+    public function testGetPagesDeduplicatesDuplicateKidsReferences(): void
     {
         $document = $this->getDocumentInstance();
 
@@ -254,118 +257,8 @@ public function testGetPagesPreservesDuplicateKidsReferences(): void
 
         $pages = $document->getPages();
 
-        $this->assertCount(2, $pages);
-        $this->assertSame($page, $pages[0]);
-        $this->assertSame($page, $pages[1]);
-    }
-
-    /**
-     * Synthetic fixture created in-repo to reproduce duplicate /Kids references.
-     */
-    public function testGetPagesPreservesDuplicateKidsFixture(): void
-    {
-        $document = (new Parser())->parseFile($this->rootDir.'/samples/bugs/PullRequestDuplicateKids.pdf');
-
-        $pages = $document->getPages();
-
-        $this->assertCount(2, $pages);
-    }
-
-    public function testGetPagesRecoversWhenCatalogPagesReferenceIsMissing(): void
-    {
-        $document = $this->getDocumentInstance();
-
-        $catalogHeader = Header::parse('<</Type/Catalog/Pages 56 0 R>>', $document);
-        $catalog = $this->getPDFObjectInstance($document, $catalogHeader);
-
-        $document->setObjects([
-            '1_0' => $catalog,
-        ]);
-
-        $pages = $document->getPages();
-
-        $this->assertCount(1, $pages);
-        $this->assertTrue($pages[0] instanceof Page);
-    }
-
-    public function testGetPagesRecoversWhenCatalogPagesNodeIsNotPagesType(): void
-    {
-        $document = $this->getDocumentInstance();
-
-        $catalogHeader = Header::parse('<</Type/Catalog/Pages 2 0 R>>', $document);
-        $catalog = $this->getPDFObjectInstance($document, $catalogHeader);
-
-        // Referenced node exists but is not a Pages object.
-        $invalidPagesHeader = Header::parse('<</Length 10>>', $document);
-        $invalidPages = $this->getPDFObjectInstance($document, $invalidPagesHeader);
-
-        $document->setObjects([
-            '1_0' => $catalog,
-            '2_0' => $invalidPages,
-        ]);
-
-        $pages = $document->getPages();
-
-        $this->assertCount(1, $pages);
-        $this->assertTrue($pages[0] instanceof Page);
-    }
-
-    public function testGetPagesRecoversBrokenPagesTreeWithPositiveCount(): void
-    {
-        $document = $this->getDocumentInstance();
-
-        $pagesHeader = Header::parse('<</Type/Pages/Kids[99 0 R]/Count 5>>', $document);
-        $pagesNode = $this->getPagesInstance($document, $pagesHeader);
-
-        $catalogHeader = Header::parse('<</Type/Catalog/Pages 2 0 R>>', $document);
-        $catalog = $this->getPDFObjectInstance($document, $catalogHeader);
-
-        $document->setObjects([
-            '1_0' => $catalog,
-            '2_0' => $pagesNode,
-        ]);
-
-        $pages = $document->getPages();
-
-        $this->assertCount(1, $pages);
-        $this->assertTrue($pages[0] instanceof Page);
-    }
-
-    public function testGetPagesRecoversMinimalHeaderlessStructure(): void
-    {
-        $document = $this->getDocumentInstance();
-
-        $object1 = $this->getPDFObjectInstance($document, Header::parse('<<>>', $document));
-        $object2 = $this->getPDFObjectInstance($document, Header::parse('<<>>', $document));
-
-        $document->setObjects([
-            '1_0' => $object1,
-            '2_0' => $object2,
-        ]);
-
-        $pages = $document->getPages();
-
         $this->assertCount(1, $pages);
-        $this->assertTrue($pages[0] instanceof Page);
-    }
-
-    public function testGetPagesDoesNotRecoverWhenMinimalStructureIsNotHeaderless(): void
-    {
-        $document = $this->getDocumentInstance();
-
-        // Guard case similar to corrupted-xref files: no catalog, but headers do
-        // contain meaningful keys, so minimal fallback must not trigger.
-        $object1 = $this->getPDFObjectInstance($document, Header::parse('<</Pages 2 0 R/Contents 3 0 R>>', $document));
-        $object2 = $this->getPDFObjectInstance($document, Header::parse('<</Length 10>>', $document));
-        $object3 = $this->getPDFObjectInstance($document, Header::parse('<</Length 20>>', $document));
-
-        $document->setObjects([
-            '1_0' => $object1,
-            '2_0' => $object2,
-            '3_0' => $object3,
-        ]);
-
-        $this->assertSame([], $document->getPages());
+        $this->assertSame($page, $pages[0]);
     }
 
     public static function provideRealWorldPageRecoveryCases(): array
@@ -393,46 +286,6 @@ public function testGetPagesOnRealWorldMalformedFixtures(string $fixture, int $e
         $this->assertCount($expectedPages, $document->getPages());
     }
 
-    public function testGetPagesDeduplicatesDuplicateKidsReferences(): void
-    {
-        $document = $this->getDocumentInstance();
-
-        $content = '<</Type/Page>>';
-        $header = Header::parse($content, $document);
-        $page = $this->getPageInstance($document, $header);
-
-        $content = '<</Type/Pages/Kids[10 0 R 10 0 R]>>';
-        $header = Header::parse($content, $document);
-        $pagesNode = $this->getPagesInstance($document, $header);
-
-        $content = '<</Type/Catalog/Pages 20 0 R>>';
-        $header = Header::parse($content, $document);
-        $catalog = $this->getPDFObjectInstance($document, $header);
-
-        $document->setObjects([
-            '10_0' => $page,
-            '20_0' => $pagesNode,
-            '30_0' => $catalog,
-        ]);
-
-        $pages = $document->getPages();
-
-        $this->assertCount(1, $pages);
-        $this->assertSame($page, $pages[0]);
-    }
-
-    /**
-     * Synthetic fixture created in-repo to reproduce duplicate /Kids references.
-     */
-    public function testGetPagesDeduplicatesDuplicateKidsFixture(): void
-    {
-        $document = (new Parser())->parseFile($this->rootDir.'/samples/bugs/PullRequestDuplicateKids.pdf');
-
-        $pages = $document->getPages();
-
-        $this->assertCount(1, $pages);
-    }
-
     /**
      * @see https://github.com/smalot/pdfparser/issues/721
      */

tests/PHPUnit/Integration/DocumentTest.php

@@ -413,8 +413,14 @@ public function testRetainImageContentImpact(): void
     public function testNoIgnoreEncryption(): void
     {
         $filename = $this->rootDir.'/samples/not_really_encrypted.pdf';
-
-        $this->assertTrue((new Parser([]))->parseFile($filename) instanceof Document);
+        $threw = false;
+        try {
+            (new Parser([]))->parseFile($filename);
+        } catch (\Exception $e) {
+            // we expect an exception to be thrown if an encrypted PDF is encountered.
+            $threw = true;
+        }
+        $this->assertTrue($threw);
     }
 
     /**
@@ -430,6 +436,8 @@ public function testIgnoreEncryption(): void
         $filename = $this->rootDir.'/samples/not_really_encrypted.pdf';
 
         $this->assertTrue((new Parser([], $config))->parseFile($filename) instanceof Document);
+
+        // without the configuration option set, an exception would be thrown.
     }
 
     /**
@@ -443,19 +451,6 @@ public function testPullRequest793ChrDeprecationFix(): void
 
         $this->assertEquals('ASCII85 last-tuple overflow test', $document->getText());
     }
-
-    /**
-     * @group linux-only
-     */
-    public function testParseFileWithLargeFlateStreams(): void
-    {
-        $config = new Config();
-        $config->setRetainImageContent(false);
-        $config->setDecodeMemoryLimit(8 * 1024 * 1024);
-        $document = (new Parser([], $config))->parseFile($this->rootDir.'/samples/bugs/PullRequest457.pdf');
-
-        self::assertCount(28, $document->getPages());
-    }
 }
 
 class ParserSub extends Parser