HTML.php

Go to the documentation of this file.

<?php

if (!defined('PHPEXCEL_ROOT')) {
    define('PHPEXCEL_ROOT', dirname(__FILE__) . '/../../');
    require(PHPEXCEL_ROOT . 'PHPExcel/Autoloader.php');
}

class PHPExcel_Reader_HTML extends PHPExcel_Reader_Abstract implements PHPExcel_Reader_IReader
{

    protected $_inputEncoding = 'ANSI';

    protected $_sheetIndex = 0;

    protected $_formats = array(
        'h1' => array('font' => array('bold' => true,
                'size' => 24,
            ),
        ), //   Bold, 24pt
        'h2' => array('font' => array('bold' => true,
                'size' => 18,
            ),
        ), //   Bold, 18pt
        'h3' => array('font' => array('bold' => true,
                'size' => 13.5,
            ),
        ), //   Bold, 13.5pt
        'h4' => array('font' => array('bold' => true,
                'size' => 12,
            ),
        ), //   Bold, 12pt
        'h5' => array('font' => array('bold' => true,
                'size' => 10,
            ),
        ), //   Bold, 10pt
        'h6' => array('font' => array('bold' => true,
                'size' => 7.5,
            ),
        ), //   Bold, 7.5pt
        'a' => array('font' => array('underline' => true,
                'color' => array('argb' => PHPExcel_Style_Color::COLOR_BLUE,
                ),
            ),
        ), //   Blue underlined
        'hr' => array('borders' => array('bottom' => array('style' => PHPExcel_Style_Border::BORDER_THIN,
                    'color' => array(\PHPExcel_Style_Color::COLOR_BLACK,
                    ),
                ),
            ),
        ), //   Bottom border
    );

    protected $rowspan = array();

    public function __construct()
    {
        $this->_readFilter = new PHPExcel_Reader_DefaultReadFilter();
    }

    protected function _isValidFormat()
    {
        //      Reading 2048 bytes should be enough to validate that the format is HTML
        $data = fread($this->_fileHandle, 2048);
        if ((strpos($data, '<') !== FALSE) &&
                (strlen($data) !== strlen(strip_tags($data)))) {
            return TRUE;
        }

        return FALSE;
    }

    public function load($pFilename)
    {
        // Create new PHPExcel
        $objPHPExcel = new PHPExcel();

        // Load into this instance
        return $this->loadIntoExisting($pFilename, $objPHPExcel);
    }

    public function setInputEncoding($pValue = 'ANSI')
    {
        $this->_inputEncoding = $pValue;

        return $this;
    }

    public function getInputEncoding()
    {
        return $this->_inputEncoding;
    }

    //  Data Array used for testing only, should write to PHPExcel object on completion of tests
    protected $_dataArray = array();
    protected $_tableLevel = 0;
    protected $_nestedColumn = array('A');

    protected function _setTableStartColumn($column)
    {
        if ($this->_tableLevel == 0)
            $column = 'A';
        ++$this->_tableLevel;
        $this->_nestedColumn[$this->_tableLevel] = $column;

        return $this->_nestedColumn[$this->_tableLevel];
    }

    protected function _getTableStartColumn()
    {
        return $this->_nestedColumn[$this->_tableLevel];
    }

    protected function _releaseTableStartColumn()
    {
        --$this->_tableLevel;

        return array_pop($this->_nestedColumn);
    }

    protected function _flushCell($sheet, $column, $row, &$cellContent)
    {
        if (is_string($cellContent)) {
            //  Simple String content
            if (trim($cellContent) > '') {
                //      Only actually write it if there's content in the string
//                              echo 'FLUSH CELL: ' , $column , $row , ' => ' , $cellContent , '<br />';
                //      Write to worksheet to be done here...
                //      ... we return the cell so we can mess about with styles more easily
                $sheet->setCellValue($column . $row, $cellContent, true);
                $this->_dataArray[$row][$column] = $cellContent;
            }
        } else {
            //  We have a Rich Text run
            //  TODO
            $this->_dataArray[$row][$column] = 'RICH TEXT: ' . $cellContent;
        }
        $cellContent = (string) '';
    }

    protected function _processDomElement(DOMNode $element, $sheet, &$row, &$column, &$cellContent, $format = null)
    {
        foreach ($element->childNodes as $child) {
            if ($child instanceof DOMText) {
                $domText = preg_replace('/\s+/u', ' ', trim($child->nodeValue));
                if (is_string($cellContent)) {
                    //  simply append the text if the cell content is a plain text string
                    $cellContent .= $domText;
                } else {
                    //  but if we have a rich text run instead, we need to append it correctly
                    //  TODO
                }
            } elseif ($child instanceof DOMElement) {
//                              echo '<b>DOM ELEMENT: </b>' , strtoupper($child->nodeName) , '<br />';

                $attributeArray = array();
                foreach ($child->attributes as $attribute) {
//                                      echo '<b>ATTRIBUTE: </b>' , $attribute->name , ' => ' , $attribute->value , '<br />';
                    $attributeArray[$attribute->name] = $attribute->value;
                }

                switch ($child->nodeName) {
                    case 'meta' :
                        foreach ($attributeArray as $attributeName => $attributeValue) {
                            switch ($attributeName) {
                                case 'content':
                                    //  TODO
                                    //  Extract character set, so we can convert to UTF-8 if required
                                    break;
                            }
                        }
                        $this->_processDomElement($child, $sheet, $row, $column, $cellContent);
                        break;
                    case 'title' :
                        $this->_processDomElement($child, $sheet, $row, $column, $cellContent);
                        $sheet->setTitle($cellContent);
                        $cellContent = '';
                        break;
                    case 'span' :
                    case 'div' :
                    case 'font' :
                    case 'i' :
                    case 'em' :
                    case 'strong':
                    case 'b' :
//                                              echo 'STYLING, SPAN OR DIV<br />';
                        if ($cellContent > '')
                            $cellContent .= ' ';
                        $this->_processDomElement($child, $sheet, $row, $column, $cellContent);
                        if ($cellContent > '')
                            $cellContent .= ' ';
//                                              echo 'END OF STYLING, SPAN OR DIV<br />';
                        break;
                    case 'hr' :
                        $this->_flushCell($sheet, $column, $row, $cellContent);
                        ++$row;
                        if (isset($this->_formats[$child->nodeName])) {
                            $sheet->getStyle($column . $row)->applyFromArray($this->_formats[$child->nodeName]);
                        } else {
                            $cellContent = '----------';
                            $this->_flushCell($sheet, $column, $row, $cellContent);
                        }
                        ++$row;
                    case 'br' :
                        if ($this->_tableLevel > 0) {
                            //  If we're inside a table, replace with a \n
                            $cellContent .= "\n";
                        } else {
                            //  Otherwise flush our existing content and move the row cursor on
                            $this->_flushCell($sheet, $column, $row, $cellContent);
                            ++$row;
                        }
//                                              echo 'HARD LINE BREAK: ' , '<br />';
                        break;
                    case 'a' :
//                                              echo 'START OF HYPERLINK: ' , '<br />';
                        foreach ($attributeArray as $attributeName => $attributeValue) {
                            switch ($attributeName) {
                                case 'href':
//                                                                      echo 'Link to ' , $attributeValue , '<br />';
                                    $sheet->getCell($column . $row)->getHyperlink()->setUrl($attributeValue);
                                    if (isset($this->_formats[$child->nodeName])) {
                                        $sheet->getStyle($column . $row)->applyFromArray($this->_formats[$child->nodeName]);
                                    }
                                    break;
                            }
                        }
                        $cellContent .= ' ';
                        $this->_processDomElement($child, $sheet, $row, $column, $cellContent);
//                                              echo 'END OF HYPERLINK:' , '<br />';
                        break;
                    case 'h1' :
                    case 'h2' :
                    case 'h3' :
                    case 'h4' :
                    case 'h5' :
                    case 'h6' :
                    case 'ol' :
                    case 'ul' :
                    case 'p' :
                        if ($this->_tableLevel > 0) {
                            //  If we're inside a table, replace with a \n
                            $cellContent .= "\n";
//                                                      echo 'LIST ENTRY: ' , '<br />';
                            $this->_processDomElement($child, $sheet, $row, $column, $cellContent);
//                                                      echo 'END OF LIST ENTRY:' , '<br />';
                        } else {
                            if ($cellContent > '') {
                                $this->_flushCell($sheet, $column, $row, $cellContent);
                                $row++;
                            }
//                                                      echo 'START OF PARAGRAPH: ' , '<br />';
                            $this->_processDomElement($child, $sheet, $row, $column, $cellContent);
//                                                      echo 'END OF PARAGRAPH:' , '<br />';
                            $this->_flushCell($sheet, $column, $row, $cellContent);

                            if (isset($this->_formats[$child->nodeName])) {
                                $sheet->getStyle($column . $row)->applyFromArray($this->_formats[$child->nodeName]);
                            }

                            $row++;
                            $column = 'A';
                        }
                        break;
                    case 'li' :
                        if ($this->_tableLevel > 0) {
                            //  If we're inside a table, replace with a \n
                            $cellContent .= "\n";
//                                                      echo 'LIST ENTRY: ' , '<br />';
                            $this->_processDomElement($child, $sheet, $row, $column, $cellContent);
//                                                      echo 'END OF LIST ENTRY:' , '<br />';
                        } else {
                            if ($cellContent > '') {
                                $this->_flushCell($sheet, $column, $row, $cellContent);
                            }
                            ++$row;
//                                                      echo 'LIST ENTRY: ' , '<br />';
                            $this->_processDomElement($child, $sheet, $row, $column, $cellContent);
//                                                      echo 'END OF LIST ENTRY:' , '<br />';
                            $this->_flushCell($sheet, $column, $row, $cellContent);
                            $column = 'A';
                        }
                        break;
                    case 'table' :
                        $this->_flushCell($sheet, $column, $row, $cellContent);
                        $column = $this->_setTableStartColumn($column);
//                                              echo 'START OF TABLE LEVEL ' , $this->_tableLevel , '<br />';
                        if ($this->_tableLevel > 1)
                            --$row;
                        $this->_processDomElement($child, $sheet, $row, $column, $cellContent);
//                                              echo 'END OF TABLE LEVEL ' , $this->_tableLevel , '<br />';
                        $column = $this->_releaseTableStartColumn();
                        if ($this->_tableLevel > 1) {
                            ++$column;
                        } else {
                            ++$row;
                        }
                        break;
                    case 'thead' :
                    case 'tbody' :
                        $this->_processDomElement($child, $sheet, $row, $column, $cellContent);
                        break;
                    case 'tr' :
                        $column = $this->_getTableStartColumn();
                        $cellContent = '';
//                                              echo 'START OF TABLE ' , $this->_tableLevel , ' ROW<br />';
                        $this->_processDomElement($child, $sheet, $row, $column, $cellContent);
                        ++$row;
//                                              echo 'END OF TABLE ' , $this->_tableLevel , ' ROW<br />';
                        break;
                    case 'th' :
                    case 'td' :
//                                              echo 'START OF TABLE ' , $this->_tableLevel , ' CELL<br />';
                        $this->_processDomElement($child, $sheet, $row, $column, $cellContent);
//                                              echo 'END OF TABLE ' , $this->_tableLevel , ' CELL<br />';

                        while (isset($this->rowspan[$column . $row])) {
                            ++$column;
                        }

                        $this->_flushCell($sheet, $column, $row, $cellContent);

//                        if (isset($attributeArray['style']) && !empty($attributeArray['style'])) {
//                            $styleAry = $this->getPhpExcelStyleArray($attributeArray['style']);
//
//                            if (!empty($styleAry)) {
//                                $sheet->getStyle($column . $row)->applyFromArray($styleAry);
//                            }
//                        }

                        if (isset($attributeArray['rowspan']) && isset($attributeArray['colspan'])) {
                            //create merging rowspan and colspan
                            $columnTo = $column;
                            for ($i = 0; $i < $attributeArray['colspan'] - 1; $i++) {
                                ++$columnTo;
                            }
                            $range = $column . $row . ':' . $columnTo . ($row + $attributeArray['rowspan'] - 1);
                            foreach (\PHPExcel_Cell::extractAllCellReferencesInRange($range) as $value) {
                                $this->rowspan[$value] = true;
                            }
                            $sheet->mergeCells($range);
                            $column = $columnTo;
                        } elseif (isset($attributeArray['rowspan'])) {
                            //create merging rowspan
                            $range = $column . $row . ':' . $column . ($row + $attributeArray['rowspan'] - 1);
                            foreach (\PHPExcel_Cell::extractAllCellReferencesInRange($range) as $value) {
                                $this->rowspan[$value] = true;
                            }
                            $sheet->mergeCells($range);
                        } elseif (isset($attributeArray['colspan'])) {
                            //create merging colspan
                            $columnTo = $column;
                            for ($i = 0; $i < $attributeArray['colspan'] - 1; $i++) {
                                ++$columnTo;
                            }
                            $sheet->mergeCells($column . $row . ':' . $columnTo . $row);
                            $column = $columnTo;
                        }
                        ++$column;
                        break;
                    case 'body' :
                        $row = 1;
                        $column = 'A';
                        $content = '';
                        $this->_tableLevel = 0;
                        $this->_processDomElement($child, $sheet, $row, $column, $cellContent);
                        break;
                    default:
                        $this->_processDomElement($child, $sheet, $row, $column, $cellContent);
                }
            }
        }
    }

    public function loadIntoExisting($pFilename, PHPExcel $objPHPExcel)
    {
        // Open file to validate
        $this->_openFile($pFilename);
        if (!$this->_isValidFormat()) {
            fclose($this->_fileHandle);
            throw new PHPExcel_Reader_Exception($pFilename . " is an Invalid HTML file.");
        }
        //      Close after validating
        fclose($this->_fileHandle);

        // Create new PHPExcel
        while ($objPHPExcel->getSheetCount() <= $this->_sheetIndex) {
            $objPHPExcel->createSheet();
        }
        $objPHPExcel->setActiveSheetIndex($this->_sheetIndex);

        //      Create a new DOM object
        $dom = new domDocument;
        //      Reload the HTML file into the DOM object
        $loaded = $dom->loadHTML($this->securityScanFile($pFilename));
        if ($loaded === FALSE) {
            throw new PHPExcel_Reader_Exception('Failed to load ', $pFilename, ' as a DOM Document');
        }

        //      Discard white space
        $dom->preserveWhiteSpace = false;

        $row = 0;
        $column = 'A';
        $content = '';
        $this->_processDomElement($dom, $objPHPExcel->getActiveSheet(), $row, $column, $content);

                // Return
        return $objPHPExcel;
    }

    public function getSheetIndex()
    {
        return $this->_sheetIndex;
    }

    public function setSheetIndex($pValue = 0)
    {
        $this->_sheetIndex = $pValue;

        return $this;
    }

        public function securityScan($xml)
        {
        $pattern = '/\\0?' . implode('\\0?', str_split('<!ENTITY')) . '\\0?/';
        if (preg_match($pattern, $xml)) { 
            throw new PHPExcel_Reader_Exception('Detected use of ENTITY in XML, spreadsheet file load() aborted to prevent XXE/XEE attacks');
        }
        return $xml;
    }

}