stringparser.php

来自「Bug tracker, and reporter.」· PHP 代码 · 共 710 行 · 第 1/2 页

        $openedBrackets = 1;

        while ($this->offset < strlen($this->data)) {
            switch (ord( $this->data[$this->offset] )) {
                case 0x28: // '(' - opened bracket in the string, needs balanced pair.
                    $openedBrackets++;
                    break;

                case 0x29: // ')' - pair to the opened bracket
                    $openedBrackets--;
                    break;

                case 0x5C: // '\\' - escape sequence, skip next char from a check
                    $this->offset++;
            }

            $this->offset++;
            if ($openedBrackets == 0) {
                break; // end of string
            }
        }
        if ($openedBrackets != 0) {
            throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while string reading. Offset - 0x%X. \')\' expected.', $start));
        }

        return new Zend_Pdf_Element_String(Zend_Pdf_Element_String::unescape( substr($this->data,
                                                                 $start,
                                                                 $this->offset - $start - 1) ));
    }


    /**
     * Read binary string PDF object
     * Also reads trailing '>' from a pdf stream
     *
     * @return Zend_Pdf_Element_String_Binary
     * @throws Zend_Pdf_Exception
     */
    private function _readBinaryString()
    {
        $start = $this->offset;

        while ($this->offset < strlen($this->data)) {
            if (self::isWhiteSpace( ord($this->data[$this->offset]) ) ||
                ctype_xdigit( $this->data[$this->offset] ) ) {
                $this->offset++;
            } else if ($this->data[$this->offset] == '>') {
                $this->offset++;
                return new Zend_Pdf_Element_String_Binary(
                               Zend_Pdf_Element_String_Binary::unescape( substr($this->data,
                                                                    $start,
                                                                    $this->offset - $start - 1) ));
            } else {
                throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected character while binary string reading. Offset - 0x%X.', $this->offset));
            }
        }
        throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while binary string reading. Offset - 0x%X. \'>\' expected.', $start));
    }


    /**
     * Read array PDF object
     * Also reads trailing ']' from a pdf stream
     *
     * @return Zend_Pdf_Element_Array
     * @throws Zend_Pdf_Exception
     */
    private function _readArray()
    {
        $elements = array();

        while ( strlen($nextLexeme = $this->readLexeme()) != 0 ) {
            if ($nextLexeme != ']') {
                $elements[] = $this->readElement($nextLexeme);
            } else {
                return new Zend_Pdf_Element_Array($elements);
            }
        }

        throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while array reading. Offset - 0x%X. \']\' expected.', $this->offset));
    }


    /**
     * Read dictionary PDF object
     * Also reads trailing '>>' from a pdf stream
     *
     * @return Zend_Pdf_Element_Dictionary
     * @throws Zend_Pdf_Exception
     */
    private function _readDictionary()
    {
        $dictionary = new Zend_Pdf_Element_Dictionary();

        while ( strlen($nextLexeme = $this->readLexeme()) != 0 ) {
            if ($nextLexeme != '>>') {
                $nameStart = $this->offset - strlen($nextLexeme);

                $name  = $this->readElement($nextLexeme);
                $value = $this->readElement();

                if (!$name instanceof Zend_Pdf_Element_Name) {
                    throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Name object expected while dictionary reading. Offset - 0x%X.', $nameStart));
                }

                $dictionary->add($name, $value);
            } else {
                return $dictionary;
            }
        }

        throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Unexpected end of file while dictionary reading. Offset - 0x%X. \'>>\' expected.', $this->offset));
    }


    /**
     * Read reference PDF object
     *
     * @param string $nextLexeme
     * @return Zend_Pdf_Element_Reference
     */
    private function _readReference($nextLexeme = null)
    {
        $start = $this->offset;

        if ($nextLexeme === null) {
            $objNum = $this->readLexeme();
        } else {
            $objNum = $nextLexeme;
        }
        if (!ctype_digit($objNum)) { // it's not a reference
            $this->offset = $start;
            return null;
        }

        $genNum = $this->readLexeme();
        if (!ctype_digit($genNum)) { // it's not a reference
            $this->offset = $start;
            return null;
        }

        $rMark  = $this->readLexeme();
        if ($rMark != 'R') { // it's not a reference
            $this->offset = $start;
            return null;
        }

        $ref = new Zend_Pdf_Element_Reference((int)$objNum, (int)$genNum, $this->_context, $this->_objFactory->resolve());

        return $ref;
    }


    /**
     * Read numeric PDF object
     *
     * @param string $nextLexeme
     * @return Zend_Pdf_Element_Numeric
     */
    private function _readNumeric($nextLexeme = null)
    {
        if ($nextLexeme === null) {
            $nextLexeme = $this->readLexeme();
        }

        return new Zend_Pdf_Element_Numeric($nextLexeme);
    }


    /**
     * Read inderect object from a PDF stream
     *
     * @param integer $offset
     * @param Zend_Pdf_Element_Reference_Context $context
     * @return Zend_Pdf_Element_Object
     */
    public function getObject($offset, Zend_Pdf_Element_Reference_Context $context)
    {
        if ($offset === null ) {
            return new Zend_Pdf_Element_Null();
        }

        // Save current offset to make getObject() reentrant
        $offsetSave = $this->offset;

        $this->offset    = $offset;
        $this->_context  = $context;
        $this->_elements = array();

        $objNum = $this->readLexeme();
        if (!ctype_digit($objNum)) {
            throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Object number expected.', $this->offset - strlen($objNum)));
        }

        $genNum = $this->readLexeme();
        if (!ctype_digit($genNum)) {
            throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Object generation number expected.', $this->offset - strlen($genNum)));
        }

        $objKeyword = $this->readLexeme();
        if ($objKeyword != 'obj') {
            throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'obj\' keyword expected.', $this->offset - strlen($objKeyword)));
        }

        $objValue = $this->readElement();

        $nextLexeme = $this->readLexeme();

        if( $nextLexeme == 'endobj' ) {
            /**
             * Object is not generated by factory (thus it's not marked as modified object).
             * But factory is assigned to the obect.
             */
            $obj = new Zend_Pdf_Element_Object($objValue, (int)$objNum, (int)$genNum, $this->_objFactory->resolve());

            foreach ($this->_elements as $element) {
                $element->setParentObject($obj);
            }

            // Restore offset value
            $this->offset = $offsetSave;

            return $obj;
        }

        /**
         * It's a stream object
         */
        if ($nextLexeme != 'stream') {
            throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'endobj\' or \'stream\' keywords expected.', $this->offset - strlen($nextLexeme)));
        }

        if (!$objValue instanceof Zend_Pdf_Element_Dictionary) {
            throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. Stream extent must be preceded by stream dictionary.', $this->offset - strlen($nextLexeme)));
        }

        /**
         * References are automatically dereferenced at this moment.
         */
        $streamLength = $objValue->Length->value;

        /**
         * 'stream' keyword must be followed by either cr-lf sequence or lf character only.
         * This restriction gives the possibility to recognize all cases exactly
         */
        if ($this->data[$this->offset] == "\r" &&
            $this->data[$this->offset + 1] == "\n"    ) {
            $this->offset += 2;
        } else if ($this->data[$this->offset] == "\n"    ) {
            $this->offset++;
        } else {
            throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'stream\' must be followed by either cr-lf sequence or lf character only.', $this->offset - strlen($nextLexeme)));
        }

        $dataOffset = $this->offset;

        $this->offset += $streamLength;

        $nextLexeme = $this->readLexeme();
        if ($nextLexeme != 'endstream') {
            throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'endstream\' keyword expected.', $this->offset - strlen($nextLexeme)));
        }

        $nextLexeme = $this->readLexeme();
        if ($nextLexeme != 'endobj') {
            throw new Zend_Pdf_Exception(sprintf('PDF file syntax error. Offset - 0x%X. \'endobj\' keyword expected.', $this->offset - strlen($nextLexeme)));
        }

        $obj = new Zend_Pdf_Element_Object_Stream(substr($this->data,
                                                         $dataOffset,
                                                         $streamLength),
                                                  (int)$objNum,
                                                  (int)$genNum,
                                                  $this->_objFactory->resolve(),
                                                  $objValue);

        foreach ($this->_elements as $element) {
            $element->setParentObject($obj);
        }

        // Restore offset value
        $this->offset = $offsetSave;

        return $obj;
    }


    /**
     * Get length of source string
     *
     * @return integer
     */
    public function getLength()
    {
        return strlen($this->data);
    }

    /**
     * Get source string
     *
     * @return string
     */
    public function getString()
    {
        return $this->data;
    }


    /**
     * Parse integer value from a binary stream
     *
     * @param string $stream
     * @param integer $offset
     * @param integer $size
     * @return integer
     */
    public static function parseIntFromStream($stream, $offset, $size)
    {
        $value = 0;
        for ($count = 0; $count < $size; $count++) {
            $value *= 256;
            $value += ord($stream[$offset + $count]);
        }

        return $value;
    }



    /**
     * Set current context
     *
     * @param Zend_Pdf_Element_Reference_Context $context
     */
    public function setContext(Zend_Pdf_Element_Reference_Context $context)
    {
        $this->_context = $context;
    }

    /**
     * Object constructor
     *
     * Note: PHP duplicates string, which is sent by value, only of it's updated.
     * Thus we don't need to care about overhead
     *
     * @param string $pdfString
     * @param Zend_Pdf_ElementFactory_Interface $factory
     */
    public function __construct($source, Zend_Pdf_ElementFactory_Interface $factory)
    {
        $this->data         = $source;
        $this->_objFactory  = $factory;
    }
}