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

Complexity

Conditions	9
Paths	4

Size

Total Lines	53
Code Lines	23

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	22
CRAP Score	9.0066

Importance

Changes	5
Bugs	3	Features	1

<?php
    protected function getIndirectObject($pdfData, $xref, $objRef, $offset = 0, $decoding = true)
    {
        /*
         * build indirect object header
         */
        // $objHeader = "[object number] [generation number] obj"
        $objRefArr = explode('_', $objRef);
        if (2 !== \count($objRefArr)) {
            throw new Exception('Invalid object reference for $obj.');
        }
        $objHeader = $objRefArr[0].' '.$objRefArr[1].' obj';

        /*
         * check if we are in position
         */
        // ignore whitespace characters at offset (NUL, HT, LF, FF, CR, SP)
        $offset += strspn($pdfData, "\0\t\n\f\r ", $offset);
        // ignore leading zeros for object number
        $offset += strspn($pdfData, '0', $offset);

        // Newer pdf may use \n instead of whitespace
        $objHeaderPattern = '/' . $objRefArr[0] . '[ \n]' . $objRefArr[1].'[ \n]obj' . '/';
        if (preg_match($objHeaderPattern, substr($pdfData, $offset, \strlen($objHeader))) == 0) {
            // an indirect reference to an undefined object shall be considered a reference to the null object
            return ['null', 'null', $offset];
        }

        /*
         * get content
         */
        // starting position of object content
        $offset += \strlen($objHeader);
        $objContentArr = [];
        $i = 0; // object main index
        do {
            $oldOffset = $offset;
            // get element
            $element = $this->getRawObject($pdfData, $offset);
            $offset = $element[2];
            // decode stream using stream's dictionary information
            if ($decoding && ('stream' === $element[0]) && (isset($objContentArr[($i - 1)][0])) && ('<<' === $objContentArr[($i - 1)][0])) {
                $element[3] = $this->decodeStream($pdfData, $xref, $objContentArr[($i - 1)][1], $element[1]);
            }
            $objContentArr[$i] = $element;
            ++$i;
        } while (('endobj' !== $element[0]) && ($offset !== $oldOffset));
        // remove closing delimiter
        array_pop($objContentArr);

        /*
         * return raw object content
         */
        return $objContentArr;
    }