FileParser.php

Go to the documentation of this file.

<?php
abstract class Zend_Pdf_FileParser
{
  /**** Class Constants ****/

    const BYTE_ORDER_LITTLE_ENDIAN = 0;

    const BYTE_ORDER_BIG_ENDIAN    = 1;



  /**** Instance Variables ****/


    protected $_isScreened = false;

    protected $_isParsed = false;

    protected $_dataSource = null;



  /**** Public Interface ****/


  /* Abstract Methods */

    abstract public function screen();

    abstract public function parse();


  /* Object Lifecycle */

    public function __construct(Zend_Pdf_FileParserDataSource $dataSource)
    {
        if ($dataSource->getSize() == 0) {
            #require_once 'Zend/Pdf/Exception.php';
            throw new Zend_Pdf_Exception('The data source has not been properly initialized',
                                         Zend_Pdf_Exception::BAD_DATA_SOURCE);
        }
        $this->_dataSource = $dataSource;
    }

    public function __destruct()
    {
        $this->_dataSource = null;
    }


  /* Accessors */

    public function isScreened()
    {
        return $this->_isScreened;
    }

    public function isParsed()
    {
        return $this->_isParsed;
    }

    public function getDataSource()
    {
        return $this->_dataSource;
    }


  /* Primitive Methods */

    public function moveToOffset($offset)
    {
        $this->_dataSource->moveToOffset($offset);
    }

    public function getOffset() {
       return $this->_dataSource->getOffset();
    }

    public function getSize() {
       return $this->_dataSource->getSize();
    }

    public function readBytes($byteCount)
    {
        return $this->_dataSource->readBytes($byteCount);
    }

    public function skipBytes($byteCount)
    {
        $this->_dataSource->skipBytes($byteCount);
    }


  /* Parser Methods */

    public function readInt($size, $byteOrder = Zend_Pdf_FileParser::BYTE_ORDER_BIG_ENDIAN)
    {
        if (($size < 1) || ($size > 4)) {
            #require_once 'Zend/Pdf/Exception.php';
            throw new Zend_Pdf_Exception("Invalid signed integer size: $size",
                                         Zend_Pdf_Exception::INVALID_INTEGER_SIZE);
        }
        $bytes = $this->_dataSource->readBytes($size);
        /* unpack() will not work for this method because it always works in
         * the host byte order for signed integers. It also does not allow for
         * variable integer sizes.
         */
        if ($byteOrder == Zend_Pdf_FileParser::BYTE_ORDER_BIG_ENDIAN) {
            $number = ord($bytes[0]);
            if (($number & 0x80) == 0x80) {
                /* This number is negative. Extract the positive equivalent.
                 */
                $number = (~ $number) & 0xff;
                for ($i = 1; $i < $size; $i++) {
                    $number = ($number << 8) | ((~ ord($bytes[$i])) & 0xff);
                }
                /* Now turn this back into a negative number by taking the
                 * two's complement (we didn't add one above so won't
                 * subtract it below). This works reliably on both 32- and
                 * 64-bit systems.
                 */
                $number = ~$number;
            } else {
                for ($i = 1; $i < $size; $i++) {
                    $number = ($number << 8) | ord($bytes[$i]);
                }
            }
        } else if ($byteOrder == Zend_Pdf_FileParser::BYTE_ORDER_LITTLE_ENDIAN) {
            $number = ord($bytes[$size - 1]);
            if (($number & 0x80) == 0x80) {
                /* Negative number. See discussion above.
                 */
                $number = 0;
                for ($i = --$size; $i >= 0; $i--) {
                    $number |= ((~ ord($bytes[$i])) & 0xff) << ($i * 8);
                }
                $number = ~$number;
            } else {
                $number = 0;
                for ($i = --$size; $i >= 0; $i--) {
                    $number |= ord($bytes[$i]) << ($i * 8);
                }
            }
        } else {
            #require_once 'Zend/Pdf/Exception.php';
            throw new Zend_Pdf_Exception("Invalid byte order: $byteOrder",
                                         Zend_Pdf_Exception::INVALID_BYTE_ORDER);
        }
        return $number;
    }

    public function readUInt($size, $byteOrder = Zend_Pdf_FileParser::BYTE_ORDER_BIG_ENDIAN)
    {
        if (($size < 1) || ($size > 4)) {
            #require_once 'Zend/Pdf/Exception.php';
            throw new Zend_Pdf_Exception("Invalid unsigned integer size: $size",
                                         Zend_Pdf_Exception::INVALID_INTEGER_SIZE);
        }
        $bytes = $this->_dataSource->readBytes($size);
        /* unpack() is a bit heavyweight for this simple conversion. Just
         * work the bytes directly.
         */
        if ($byteOrder == Zend_Pdf_FileParser::BYTE_ORDER_BIG_ENDIAN) {
            $number = ord($bytes[0]);
            for ($i = 1; $i < $size; $i++) {
                $number = ($number << 8) | ord($bytes[$i]);
            }
        } else if ($byteOrder == Zend_Pdf_FileParser::BYTE_ORDER_LITTLE_ENDIAN) {
            $number = 0;
            for ($i = --$size; $i >= 0; $i--) {
                $number |= ord($bytes[$i]) << ($i * 8);
            }
        } else {
            #require_once 'Zend/Pdf/Exception.php';
            throw new Zend_Pdf_Exception("Invalid byte order: $byteOrder",
                                         Zend_Pdf_Exception::INVALID_BYTE_ORDER);
        }
        return $number;
    }

    public function isBitSet($bit, $bitField)
    {
        $bitMask = 1 << $bit;
        $isSet = (($bitField & $bitMask) == $bitMask);
        return $isSet;
    }

    public function readFixed($mantissaBits, $fractionBits,
                              $byteOrder = Zend_Pdf_FileParser::BYTE_ORDER_BIG_ENDIAN)
    {
        $bitsToRead = $mantissaBits + $fractionBits;
        if (($bitsToRead % 8) !== 0) {
            #require_once 'Zend/Pdf/Exception.php';
            throw new Zend_Pdf_Exception('Fixed-point numbers are whole bytes',
                                         Zend_Pdf_Exception::BAD_FIXED_POINT_SIZE);
        }
        $number = $this->readInt(($bitsToRead >> 3), $byteOrder) / (1 << $fractionBits);
        return $number;
    }

    public function readStringUTF16($byteCount,
                                    $byteOrder = Zend_Pdf_FileParser::BYTE_ORDER_BIG_ENDIAN,
                                    $characterSet = '')
    {
        if ($byteCount == 0) {
            return '';
        }
        $bytes = $this->_dataSource->readBytes($byteCount);
        if ($byteOrder == Zend_Pdf_FileParser::BYTE_ORDER_BIG_ENDIAN) {
            if ($characterSet == 'UTF-16BE') {
                return $bytes;
            }
            return iconv('UTF-16BE', $characterSet, $bytes);
        } else if ($byteOrder == Zend_Pdf_FileParser::BYTE_ORDER_LITTLE_ENDIAN) {
            if ($characterSet == 'UTF-16LE') {
                return $bytes;
            }
            return iconv('UTF-16LE', $characterSet, $bytes);
        } else {
            #require_once 'Zend/Pdf/Exception.php';
            throw new Zend_Pdf_Exception("Invalid byte order: $byteOrder",
                                         Zend_Pdf_Exception::INVALID_BYTE_ORDER);
        }
    }

    public function readStringMacRoman($byteCount, $characterSet = '')
    {
        if ($byteCount == 0) {
            return '';
        }
        $bytes = $this->_dataSource->readBytes($byteCount);
        if ($characterSet == 'MacRoman') {
            return $bytes;
        }
        return iconv('MacRoman', $characterSet, $bytes);
    }

    public function readStringPascal($characterSet = '', $lengthBytes = 1)
    {
        $byteCount = $this->readUInt($lengthBytes);
        if ($byteCount == 0) {
            return '';
        }
        $bytes = $this->_dataSource->readBytes($byteCount);
        if ($characterSet == 'ASCII') {
            return $bytes;
        }
        return iconv('ASCII', $characterSet, $bytes);
    }

}