chore: use pdfinfo as fallback to get page dimension

Signed-off-by: Vitor Mattos <[email protected]>
LibreSign · Jan 25, 2025 · f0cffa5 · f0cffa5
1 parent 82d217f
commit f0cffa5
Show file tree

Hide file tree

Showing 4 changed files with 142 additions and 16 deletions.
diff --git a/lib/Helper/ConfigureCheckHelper.php b/lib/Helper/ConfigureCheckHelper.php
@@ -27,6 +27,9 @@
 use JsonSerializable;
 
 /**
+ * @method ConfigureCheckHelper setSuccessMessage(string $value)
+ * @method ConfigureCheckHelper setInfoMessage(string $value)
+ * @method ConfigureCheckHelper setErrorMessage(string $value)
  * @method ConfigureCheckHelper setStatus(string $value)
  * @method string getStatus()
  * @method ConfigureCheckHelper setMessage(string $value)

diff --git a/lib/Service/Install/ConfigureCheckService.php b/lib/Service/Install/ConfigureCheckService.php
@@ -82,11 +82,17 @@ public function checkSign(): array {
 	}
 
 	public function checkPoppler(): array {
+		$return = $this->checkPdfSig();
+		$return = array_merge($return, $this->checkPdfinfo());
+		return $return;
+	}
+
+	public function checkPdfSig(): array {
 		if (shell_exec('which pdfsig') === null) {
 			return [
 				(new ConfigureCheckHelper())
 					->setInfoMessage('Poppler utils not installed')
-					->setResource('poppler-utils')
+					->setResource('pdfsig')
 					->setTip('Install the package poppler-utils at your operational system to be possible get more details about validation of signatures.'),
 			];
 		}
@@ -96,17 +102,17 @@ public function checkPoppler(): array {
 		if (!$version) {
 			return [
 				(new ConfigureCheckHelper())
-					->setInfoMessage('Fail to retrieve pdfsig version')
-					->setResource('poppler-utils')
+					->setErrorMessage('Fail to retrieve pdfsig version')
+					->setResource('pdfsig')
 					->setTip("The command <pdfsig -v> executed by PHP haven't any output."),
 			];
 		}
 		$version = preg_match('/pdfsig version (?<version>.*)/', $version, $matches);
 		if (!$version) {
 			return [
 				(new ConfigureCheckHelper())
-					->setInfoMessage('Fail to retrieve pdfsig version')
-					->setResource('poppler-utils')
+					->setErrorMessage('Fail to retrieve pdfsig version')
+					->setResource('pdfsig')
 					->setTip("This is a poppler-utils dependency and wasn't possible to parse the output of command pdfsig -v"),
 			];
 		}
@@ -116,6 +122,41 @@ public function checkPoppler(): array {
 		];
 	}
 
+	public function checkPdfinfo(): array {
+		if (shell_exec('which pdfinfo') === null) {
+			return [
+				(new ConfigureCheckHelper())
+					->setInfoMessage('Poppler utils not installed')
+					->setResource('pdfinfo')
+					->setTip('Install the package poppler-utils at your operational system have a fallback to fetch page dimensions.'),
+			];
+		}
+		// The output of this command go to STDERR and shell_exec get the STDOUT
+		// With 2>&1 the STRERR is redirected to STDOUT
+		$version = shell_exec('pdfinfo -v 2>&1');
+		if (!$version) {
+			return [
+				(new ConfigureCheckHelper())
+					->setErrorMessage('Fail to retrieve pdfinfo version')
+					->setResource('pdfinfo')
+					->setTip("The command <pdfinfo -v> executed by PHP haven't any output."),
+			];
+		}
+		$version = preg_match('/pdfinfo version (?<version>.*)/', $version, $matches);
+		if (!$version) {
+			return [
+				(new ConfigureCheckHelper())
+					->setErrorMessage('Fail to retrieve pdfinfo version')
+					->setResource('pdfinfo')
+					->setTip("This is a poppler-utils dependency and wasn't possible to parse the output of command pdfinfo -v"),
+			];
+		}
+		return [(new ConfigureCheckHelper())
+			->setSuccessMessage('pdfinfo version: ' . $matches['version'])
+			->setResource('pdfinfo')
+		];
+	}
+
 	/**
 	 * Check all requirements to use JSignPdf
 	 *

diff --git a/lib/Service/PdfParserService.php b/lib/Service/PdfParserService.php
@@ -57,14 +57,19 @@ public function setFile(File|string $file): self {
 		return $this;
 	}
 
+	private function getContent(): string {
+		if (!$this->content) {
+			throw new LibresignException('File not defined to be parsed.');
+		}
+		return $this->content;
+	}
+
 	private function getDocument(): Document {
 		if (!$this->document) {
-			if (!$this->content) {
-				throw new LibresignException('File not defined to be parsed.');
-			}
+			$content = $this->getContent();
 			try {
 				$parser = new \Smalot\PdfParser\Parser();
-				$this->document = $parser->parseContent($this->content);
+				$this->document = $parser->parseContent($content);
 				return $this->document;
 			} catch (\Throwable $th) {
 				if ($th->getMessage() === 'Secured pdf file are currently not supported.') {
@@ -83,6 +88,13 @@ private function getDocument(): Document {
 	 * @psalm-return array{p: int, d?: non-empty-list<array{w: mixed, h: mixed}>}
 	 */
 	public function getPageDimensions(): array {
+		if ($return = $this->getPageDimensionsWithPdfInfo()) {
+			return $return;
+		}
+		return $this->getPageDimensionsWithSmalotPdfParser();
+	}
+
+	private function getPageDimensionsWithSmalotPdfParser(): array {
 		$document = $this->getDocument();
 		$pages = $document->getPages();
 		$output = [
@@ -94,15 +106,49 @@ public function getPageDimensions(): array {
 				$pages = $document->getObjectsByType('Pages');
 				$details = reset($pages)->getHeader()->getDetails();
 			}
-			$widthAndHeight = [
-				'w' => $details['MediaBox'][2],
-				'h' => $details['MediaBox'][3]
-			];
-			if (!is_numeric($widthAndHeight['w']) || !is_numeric($widthAndHeight['h'])) {
+			if (!isset($details['MediaBox']) || !is_numeric($details['MediaBox'][2]) || !is_numeric($details['MediaBox'][3])) {
 				$this->logger->error('Impossible get metadata from this file: Error to get page width and height. If possible, open an issue at github.com/libresign/libresign with the file that you used.');
 				throw new LibresignException('Impossible get metadata from this file.');
 			}
-			$output['d'][] = $widthAndHeight;
+			$output['d'][] = [
+				'w' => $details['MediaBox'][2],
+				'h' => $details['MediaBox'][3],
+			];
+		}
+		$pending = $output['p'] - count($output['d']);
+		if ($pending) {
+			for ($i = 0; $i < $pending; $i++) {
+				$output['d'][] = $output['d'][0];
+			}
+		}
+		return $output;
+	}
+
+	private function getPageDimensionsWithPdfInfo(): array {
+		if (shell_exec('which pdfinfo') === null) {
+			return [];
+		}
+		$content = $this->getContent();
+		$filename = $this->tempManager->getTemporaryFile('.pdf');
+		file_put_contents($filename, $content);
+
+		// The output of this command go to STDERR and shell_exec get the STDOUT
+		// With 2>&1 the STRERR is redirected to STDOUT
+		$pdfinfo = shell_exec('pdfinfo ' . $filename . ' -l -1 2>&1');
+		if (!$pdfinfo) {
+			return [];
+		}
+		if (!preg_match_all('/Page +\d+ +size: +(\d+\.?\d*) x (\d+\.?\d*)/', $pdfinfo, $pages)) {
+			return [];
+		}
+		$output = [
+			'p' => count($pages[1]),
+		];
+		foreach ($pages[1] as $page => $width) {
+			$output['d'][] = [
+				'w' => (float)$width,
+				'h' => (float)$pages[2][$page],
+			];
 		}
 		return $output;
 	}

diff --git a/tests/Unit/Service/PdfParseServiceTest.php b/tests/Unit/Service/PdfParseServiceTest.php
@@ -2,6 +2,18 @@
 
 declare(strict_types=1);
 
+namespace OCA\Libresign\Service;
+
+/**
+ * Overwrite shell_exec in the OCA\Libresign\Service namespace.
+ */
+function shell_exec($command) {
+	if (\OCA\Libresign\Tests\Unit\Service\PdfParseServiceTest::$disablePdfInfo) {
+		return null;
+	}
+	return \shell_exec($command);
+}
+
 namespace OCA\Libresign\Tests\Unit\Service;
 
 use OCA\Libresign\Exception\LibresignException;
@@ -17,6 +29,7 @@
 final class PdfParseServiceTest extends \OCA\Libresign\Tests\Unit\TestCase {
 	private ITempManager $tempManager;
 	private LoggerInterface|MockObject $loggerInterface;
+	public static $disablePdfInfo = false;
 
 	public function setUp(): void {
 		parent::setUp();
@@ -58,7 +71,8 @@ public function dataGetMetadataWithFail(): array {
 	/**
 	 * @dataProvider providerGetMetadataWithSuccess
 	 */
-	public function testGetMetadataWithSuccess(string $path, array $expected): void {
+	public function testGetMetadataWithSuccess(bool $disablePdfInfo, string $path, array $expected): void {
+		self::$disablePdfInfo = $disablePdfInfo;
 		/** @var File|MockObject */
 		$file = $this->createMock(File::class);
 		$file->method('getContent')
@@ -72,6 +86,7 @@ public function testGetMetadataWithSuccess(string $path, array $expected): void
 	public function providerGetMetadataWithSuccess(): array {
 		return [
 			[
+				'disablePdfInfo' => true,
 				'tests/fixtures/small_valid.pdf',
 				[
 					'p' => 1,
@@ -81,6 +96,7 @@ public function providerGetMetadataWithSuccess(): array {
 				]
 			],
 			[
+				'disablePdfInfo' => true,
 				'tests/fixtures/small_valid-signed.pdf',
 				[
 					'p' => 1,
@@ -89,6 +105,26 @@ public function providerGetMetadataWithSuccess(): array {
 					],
 				]
 			],
+			[
+				'disablePdfInfo' => false,
+				'tests/fixtures/small_valid.pdf',
+				[
+					'p' => 1,
+					'd' => [
+						['w' => 595.276, 'h' => 841.89],
+					],
+				]
+			],
+			[
+				'disablePdfInfo' => false,
+				'tests/fixtures/small_valid-signed.pdf',
+				[
+					'p' => 1,
+					'd' => [
+						['w' => 595.276, 'h' => 841.89],
+					],
+				]
+			],
 		];
 	}
 }