Smalot\PdfParser\RawData\RawDataParser::decodeXref()

Complexity

Conditions	15
Paths	134

Size

Total Lines	64
Code Lines	37

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	34
CRAP Score	15.0386

Importance

Changes	3
Bugs	1	Features	1

<?php
    protected function decodeXref(string $pdfData, int $startxref, array $xref = []): array
    {
        $startxref += 4; // 4 is the length of the word 'xref'
        // skip initial white space chars
        $offset = $startxref + strspn($pdfData, $this->config->getPdfWhitespaces(), $startxref);
        // initialize object number
        $obj_num = 0;
        // search for cross-reference entries or subsection
        while (preg_match('/([0-9]+)[\x20]([0-9]+)[\x20]?([nf]?)(\r\n|[\x20]?[\r\n])/', $pdfData, $matches, \PREG_OFFSET_CAPTURE, $offset) > 0) {
            if ($matches[0][1] != $offset) {
                // we are on another section
                break;
            }
            $offset += \strlen($matches[0][0]);
            if ('n' == $matches[3][0]) {
                // create unique object index: [object number]_[generation number]
                $index = $obj_num.'_'.(int) ($matches[2][0]);
                // check if object already exist
                if (!isset($xref['xref'][$index])) {
                    // store object offset position
                    $xref['xref'][$index] = (int) ($matches[1][0]);
                }
                ++$obj_num;
            } elseif ('f' == $matches[3][0]) {
                ++$obj_num;
            } else {
                // object number (index)
                $obj_num = (int) ($matches[1][0]);
            }
        }
        // get trailer data
        if (preg_match('/trailer[\s]*<<(.*)>>/isU', $pdfData, $matches, \PREG_OFFSET_CAPTURE, $offset) > 0) {
            $trailer_data = $matches[1][0];
            if (!isset($xref['trailer']) || empty($xref['trailer'])) {
                // get only the last updated version
                $xref['trailer'] = [];
                // parse trailer_data
                if (preg_match('/Size[\s]+([0-9]+)/i', $trailer_data, $matches) > 0) {
                    $xref['trailer']['size'] = (int) ($matches[1]);
                }
                if (preg_match('/Root[\s]+([0-9]+)[\s]+([0-9]+)[\s]+R/i', $trailer_data, $matches) > 0) {
                    $xref['trailer']['root'] = (int) ($matches[1]).'_'.(int) ($matches[2]);
                }
                if (preg_match('/Encrypt[\s]+([0-9]+)[\s]+([0-9]+)[\s]+R/i', $trailer_data, $matches) > 0) {
                    $xref['trailer']['encrypt'] = (int) ($matches[1]).'_'.(int) ($matches[2]);
                }
                if (preg_match('/Info[\s]+([0-9]+)[\s]+([0-9]+)[\s]+R/i', $trailer_data, $matches) > 0) {
                    $xref['trailer']['info'] = (int) ($matches[1]).'_'.(int) ($matches[2]);
                }
                if (preg_match('/ID[\s]*[\[][\s]*[<]([^>]*)[>][\s]*[<]([^>]*)[>]/i', $trailer_data, $matches) > 0) {
                    $xref['trailer']['id'] = [];
                    $xref['trailer']['id'][0] = $matches[1];
                    $xref['trailer']['id'][1] = $matches[2];
                }
            }
            if (preg_match('/Prev[\s]+([0-9]+)/i', $trailer_data, $matches) > 0) {
                // get previous xref
                $xref = $this->getXrefData($pdfData, (int) ($matches[1]), $xref);
            }
        } else {
            throw new Exception('Unable to find trailer');
        }

        return $xref;
    }