/**
* Build details array.
*/
protected function buildDetails()
{
// Build details array.
$details = array();
// Extract document info
if ($this->trailer->has('Info')) {
/** @var Object $info */
$info = $this->trailer->get('Info');
$details = $info->getHeader()->getDetails();
}
// Retrieve the page count
try {
$pages = $this->getPages();
$details['Pages'] = count($pages);
} catch (\Exception $e) {
$details['Pages'] = 0;
}
$this->details = $details;
}
/**
* @param $document Document
* @param $header Header
* @param $content string
*
* @return Object
*/
public static function factory(Document $document, Header $header, $content)
{
switch ($header->get('Type')->getContent()) {
case 'XObject':
switch ($header->get('Subtype')->getContent()) {
case 'Image':
return new Image($document, $header, $content);
case 'Form':
return new Form($document, $header, $content);
default:
return new Object($document, $header, $content);
}
break;
case 'Pages':
return new Pages($document, $header, $content);
case 'Page':
return new Page($document, $header, $content);
case 'Encoding':
return new Encoding($document, $header, $content);
case 'Font':
$subtype = $header->get('Subtype')->getContent();
$classname = '\\Smalot\\PdfParser\\Font\\Font' . $subtype;
if (class_exists($classname)) {
return new $classname($document, $header, $content);
} else {
return new Font($document, $header, $content);
}
default:
return new Object($document, $header, $content);
}
}
/**
* @param string $id
* @param array $structure
* @param Document $document
*/
protected function parseObject($id, $structure, $document)
{
$header = new Header(array(), $document);
$content = '';
foreach ($structure as $position => $part) {
switch ($part[0]) {
case '[':
$elements = array();
foreach ($part[1] as $sub_element) {
$sub_type = $sub_element[0];
$sub_value = $sub_element[1];
$elements[] = $this->parseHeaderElement($sub_type, $sub_value, $document);
}
$header = new Header($elements, $document);
break;
case '<<':
$header = $this->parseHeader($part[1], $document);
break;
case 'stream':
$content = isset($part[3][0]) ? $part[3][0] : $part[1];
if ($header->get('Type')->equals('ObjStm')) {
$match = array();
// Split xrefs and contents.
preg_match('/^((\\d+\\s+\\d+\\s*)*)(.*)$/s', $content, $match);
$content = $match[3];
// Extract xrefs.
$xrefs = preg_split('/(\\d+\\s+\\d+\\s*)/s', $match[1], -1, PREG_SPLIT_NO_EMPTY | PREG_SPLIT_DELIM_CAPTURE);
$table = array();
foreach ($xrefs as $xref) {
list($id, $position) = explode(' ', trim($xref));
$table[$position] = $id;
}
ksort($table);
$ids = array_values($table);
$positions = array_keys($table);
foreach ($positions as $index => $position) {
$id = $ids[$index] . '_0';
$next_position = isset($positions[$index + 1]) ? $positions[$index + 1] : strlen($content);
$sub_content = substr($content, $position, $next_position - $position);
$sub_header = Header::parse($sub_content, $document);
$object = Object::factory($document, $sub_header, '');
$this->objects[$id] = $object;
}
// It is not necessary to store this content.
$content = '';
return;
}
break;
default:
if ($part != 'null') {
$element = $this->parseHeaderElement($part[0], $part[1], $document);
if ($element) {
$header = new Header(array($element), $document);
}
}
break;
}
}
if (!isset($this->objects[$id])) {
$this->objects[$id] = Object::factory($document, $header, $content);
}
}
public function testGetDetails()
{
// // Document with text.
// $filename = __DIR__ . '/../../../../../../samples/Document1_pdfcreator_nocompressed.pdf';
// $parser = new \Smalot\PdfParser\Parser();
// $document = $parser->parseFile($filename);
// $object = $document->getObjectById('3_0');
// /** @var \Smalot\PdfParser\Element\ElementArray $kids */
// $kids = $object->get('Kids');
// $details = $kids->getDetails();
//
// $this->assert->array($details)->hasSize(1);
// $this->assert->string($details[0]['Type'])->isEqualTo('Page');
$document = new Document();
$content = '<</Type/Page/Types[8]/Sizes[1 2 3 4 5 <</Subtype/XObject>> [8 [9 <</FontSize 10>>]]]>>';
$details_reference = array('Type' => 'Page', 'Types' => array(8), 'Sizes' => array(1, 2, 3, 4, 5, array('Subtype' => 'XObject'), array(8, array(9, array('FontSize' => 10)))));
$header = Header::parse($content, $document);
$details = $header->getDetails();
$this->assert->array($details)->hasSize(3);
$this->assert->array($details)->isEqualTo($details_reference);
}
public function testGetPages()
{
// Missing catalog
$document = new \Smalot\PdfParser\Document();
try {
$pages = $document->getPages();
$this->assert->boolean($pages)->isEqualTo(false);
} catch (\Exception $e) {
$this->assert->object($e)->isInstanceOf('\\Exception');
}
// Listing pages from type Page
$content = '<</Type/Page>>';
$header = \Smalot\PdfParser\Header::parse($content, $document);
$object1 = new \Smalot\PdfParser\Page($document, $header);
$header = \Smalot\PdfParser\Header::parse($content, $document);
$object2 = new \Smalot\PdfParser\Page($document, $header);
$document->setObjects(array(1 => $object1, 2 => $object2));
$pages = $document->getPages();
$this->assert->integer(count($pages))->isEqualTo(2);
$this->assert->object($pages[0])->isInstanceOf('\\Smalot\\PdfParser\\Page');
$this->assert->object($pages[1])->isInstanceOf('\\Smalot\\PdfParser\\Page');
// Listing pages from type Pages (kids)
$content = '<</Type/Page>>';
$header = \Smalot\PdfParser\Header::parse($content, $document);
$object1 = new \Smalot\PdfParser\Page($document, $header);
$header = \Smalot\PdfParser\Header::parse($content, $document);
$object2 = new \Smalot\PdfParser\Page($document, $header);
$header = \Smalot\PdfParser\Header::parse($content, $document);
$object3 = new \Smalot\PdfParser\Page($document, $header);
$content = '<</Type/Pages/Kids[1 0 R 2 0 R]>>';
$header = \Smalot\PdfParser\Header::parse($content, $document);
$object4 = new \Smalot\PdfParser\Pages($document, $header);
$content = '<</Type/Pages/Kids[3 0 R]>>';
$header = \Smalot\PdfParser\Header::parse($content, $document);
$object5 = new \Smalot\PdfParser\Pages($document, $header);
$document->setObjects(array('1_0' => $object1, '2_0' => $object2, '3_0' => $object3, '4_0' => $object4, '5_0' => $object5));
$pages = $document->getPages();
$this->assert->integer(count($pages))->isEqualTo(3);
$this->assert->object($pages[0])->isInstanceOf('\\Smalot\\PdfParser\\Page');
$this->assert->object($pages[1])->isInstanceOf('\\Smalot\\PdfParser\\Page');
$this->assert->object($pages[2])->isInstanceOf('\\Smalot\\PdfParser\\Page');
// Listing pages from type Catalog
$content = '<</Type/Page>>';
$header = \Smalot\PdfParser\Header::parse($content, $document);
$object1 = new \Smalot\PdfParser\Page($document, $header);
$header = \Smalot\PdfParser\Header::parse($content, $document);
$object2 = new \Smalot\PdfParser\Page($document, $header);
$header = \Smalot\PdfParser\Header::parse($content, $document);
$object3 = new \Smalot\PdfParser\Page($document, $header);
$content = '<</Type/Pages/Kids[1 0 R 2 0 R]>>';
$header = \Smalot\PdfParser\Header::parse($content, $document);
$object4 = new \Smalot\PdfParser\Pages($document, $header);
$content = '<</Type/Pages/Kids[4 0 R 3 0 R]>>';
$header = \Smalot\PdfParser\Header::parse($content, $document);
$object5 = new \Smalot\PdfParser\Pages($document, $header);
$content = '<</Type/Catalog/Pages 5 0 R >>';
$header = \Smalot\PdfParser\Header::parse($content, $document);
$object6 = new \Smalot\PdfParser\Pages($document, $header);
$document->setObjects(array('1_0' => $object1, '2_0' => $object2, '3_0' => $object3, '4_0' => $object4, '5_0' => $object5, '6_0' => $object6));
$pages = $document->getPages();
$this->assert->integer(count($pages))->isEqualTo(3);
$this->assert->object($pages[0])->isInstanceOf('\\Smalot\\PdfParser\\Page');
$this->assert->object($pages[1])->isInstanceOf('\\Smalot\\PdfParser\\Page');
$this->assert->object($pages[2])->isInstanceOf('\\Smalot\\PdfParser\\Page');
}
public function testResolveXRef()
{
$document = new \Smalot\PdfParser\Document();
$content = '<</Type/Page/SubType/Text/Font 5 0 R/Resources 8 0 R>>foo';
$position = 0;
$header = \Smalot\PdfParser\Header::parse($content, $document, $position);
$object = new \Smalot\PdfParser\Page($document, $header);
$document->setObjects(array('5_0' => $object));
$this->assert->object($header->get('Font'))->isInstanceOf('\\Smalot\\PdfParser\\Object');
try {
$this->assert->object($header->get('Resources'))->isInstanceOf('\\Smalot\\PdfParser\\Element\\ElementMissing');
$this->assert->boolean(true)->isEqualTo(false);
} catch (\Exception $e) {
$this->assert->exception($e)->hasMessage('Missing object reference #8_0.');
}
}
