Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
53.85% |
35 / 65 |
|
50.00% |
4 / 8 |
CRAP | |
0.00% |
0 / 1 |
| PdfMetadataExtractor | |
53.85% |
35 / 65 |
|
50.00% |
4 / 8 |
80.63 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setFile | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
3 | |||
| getContent | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| getDocument | |
0.00% |
0 / 4 |
|
0.00% |
0 / 1 |
6 | |||
| getPageDimensions | |
66.67% |
2 / 3 |
|
0.00% |
0 / 1 |
2.15 | |||
| getPageDimensionsWithSmalotPdfParser | |
0.00% |
0 / 22 |
|
0.00% |
0 / 1 |
72 | |||
| getPageDimensionsWithPdfInfo | |
84.21% |
16 / 19 |
|
0.00% |
0 / 1 |
5.10 | |||
| getPdfVersion | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | |
| 3 | declare(strict_types=1); |
| 4 | |
| 5 | /** |
| 6 | * SPDX-FileCopyrightText: 2020-2026 LibreCode coop and contributors |
| 7 | * SPDX-License-Identifier: AGPL-3.0-or-later |
| 8 | */ |
| 9 | |
| 10 | namespace OCA\Libresign\Service\File\Pdf; |
| 11 | |
| 12 | use OCA\Libresign\Exception\LibresignException; |
| 13 | use OCA\Libresign\Vendor\Smalot\PdfParser\Document; |
| 14 | use OCP\Files\File; |
| 15 | use OCP\ITempManager; |
| 16 | use Psr\Log\LoggerInterface; |
| 17 | |
| 18 | class PdfMetadataExtractor { |
| 19 | private string $content = ''; |
| 20 | private string $fileName = ''; |
| 21 | private ?Document $document = null; |
| 22 | |
| 23 | public function __construct( |
| 24 | private PdfParser $pdfParser, |
| 25 | private ITempManager $tempManager, |
| 26 | private LoggerInterface $logger, |
| 27 | ) { |
| 28 | } |
| 29 | |
| 30 | /** |
| 31 | * @throws LibresignException |
| 32 | */ |
| 33 | public function setFile(File $file): self { |
| 34 | $fileName = $file->getName(); |
| 35 | try { |
| 36 | $this->content = $file->getContent(); |
| 37 | } catch (\Throwable $e) { |
| 38 | throw new LibresignException( |
| 39 | sprintf('Unable to read file "%s": %s', $fileName, $e->getMessage()) |
| 40 | ); |
| 41 | } |
| 42 | if (!$this->content) { |
| 43 | throw new LibresignException(sprintf('The file "%s" is empty.', $fileName)); |
| 44 | } |
| 45 | $this->fileName = $fileName; |
| 46 | $this->document = null; |
| 47 | return $this; |
| 48 | } |
| 49 | |
| 50 | private function getContent(): string { |
| 51 | if (!$this->content) { |
| 52 | throw new LibresignException('File not defined to be parsed.'); |
| 53 | } |
| 54 | return $this->content; |
| 55 | } |
| 56 | |
| 57 | private function getDocument(): Document { |
| 58 | if (!$this->document) { |
| 59 | $content = $this->getContent(); |
| 60 | $this->document = $this->pdfParser->parse($content, $this->fileName); |
| 61 | } |
| 62 | return $this->document; |
| 63 | } |
| 64 | |
| 65 | /** |
| 66 | * @return (array[]|int)[] |
| 67 | * @throws LibresignException |
| 68 | * @psalm-return array{p: int, d?: non-empty-list<array{w: mixed, h: mixed}>} |
| 69 | */ |
| 70 | public function getPageDimensions(): array { |
| 71 | if ($return = $this->getPageDimensionsWithPdfInfo()) { |
| 72 | return $return; |
| 73 | } |
| 74 | return $this->getPageDimensionsWithSmalotPdfParser(); |
| 75 | } |
| 76 | |
| 77 | private function getPageDimensionsWithSmalotPdfParser(): array { |
| 78 | $document = $this->getDocument(); |
| 79 | $pages = $document->getPages(); |
| 80 | $output = [ |
| 81 | 'p' => count($pages), |
| 82 | ]; |
| 83 | foreach ($pages as $page) { |
| 84 | $details = $page->getDetails(); |
| 85 | if (!isset($details['MediaBox'])) { |
| 86 | $pages = $document->getObjectsByType('Pages'); |
| 87 | $details = reset($pages)->getHeader()->getDetails(); |
| 88 | } |
| 89 | if (!isset($details['MediaBox']) || !is_numeric($details['MediaBox'][2]) || !is_numeric($details['MediaBox'][3])) { |
| 90 | $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.'); |
| 91 | throw new LibresignException('Impossible get metadata from this file.'); |
| 92 | } |
| 93 | $output['d'][] = [ |
| 94 | 'w' => $details['MediaBox'][2], |
| 95 | 'h' => $details['MediaBox'][3], |
| 96 | ]; |
| 97 | } |
| 98 | $pending = $output['p'] - count($output['d']); |
| 99 | if ($pending) { |
| 100 | for ($i = 0; $i < $pending; $i++) { |
| 101 | $output['d'][] = $output['d'][0]; |
| 102 | } |
| 103 | } |
| 104 | return $output; |
| 105 | } |
| 106 | |
| 107 | private function getPageDimensionsWithPdfInfo(): array { |
| 108 | if (shell_exec('which pdfinfo') === null) { |
| 109 | return []; |
| 110 | } |
| 111 | $content = $this->getContent(); |
| 112 | $filename = $this->tempManager->getTemporaryFile('.pdf'); |
| 113 | file_put_contents($filename, $content); |
| 114 | |
| 115 | // The output of this command go to STDERR and shell_exec get the STDOUT |
| 116 | // With 2>&1 the STRERR is redirected to STDOUT |
| 117 | $pdfinfo = shell_exec('pdfinfo ' . $filename . ' -l -1 2>&1'); |
| 118 | if (!$pdfinfo) { |
| 119 | return []; |
| 120 | } |
| 121 | if (!preg_match_all('/Page +\d+ +size: +(\d+\.?\d*) x (\d+\.?\d*)/', (string)$pdfinfo, $pages)) { |
| 122 | return []; |
| 123 | } |
| 124 | $output = [ |
| 125 | 'p' => count($pages[1]), |
| 126 | ]; |
| 127 | foreach ($pages[1] as $page => $width) { |
| 128 | $output['d'][] = [ |
| 129 | 'w' => (float)$width, |
| 130 | 'h' => (float)$pages[2][$page], |
| 131 | ]; |
| 132 | } |
| 133 | return $output; |
| 134 | } |
| 135 | |
| 136 | public function getPdfVersion(): string { |
| 137 | preg_match('/^%PDF-(?<version>\d+(\.\d+)?)/', $this->getContent(), $match); |
| 138 | return $match['version']; |
| 139 | } |
| 140 | } |