From 3783c7e0996a8d069b384794ba0656dc40d24f5a Mon Sep 17 00:00:00 2001 From: Mark Baker Date: Thu, 7 Jun 2012 19:43:19 +0000 Subject: [PATCH] Feature: Initial version of HTML Reader git-svn-id: https://phpexcel.svn.codeplex.com/svn/trunk@91447 2327b42d-5241-43d6-9e2a-de5ac946f064 --- Classes/PHPExcel/Reader/HTML.php | 495 +++++++++++++++++++++++++++++++ 1 file changed, 495 insertions(+) create mode 100644 Classes/PHPExcel/Reader/HTML.php diff --git a/Classes/PHPExcel/Reader/HTML.php b/Classes/PHPExcel/Reader/HTML.php new file mode 100644 index 0000000..dafc7c5 --- /dev/null +++ b/Classes/PHPExcel/Reader/HTML.php @@ -0,0 +1,495 @@ + 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 + ); + + /** + * PHPExcel_Reader_IReadFilter instance + * + * @var PHPExcel_Reader_IReadFilter + */ + private $_readFilter = null; + + + /** + * Create a new PHPExcel_Reader_HTML + */ + public function __construct() { + $this->_readFilter = new PHPExcel_Reader_DefaultReadFilter(); + } + + /** + * Can the current PHPExcel_Reader_IReader read the file? + * + * @param string $pFileName + * @return boolean + * @throws Exception + */ + public function canRead($pFilename) + { + // Check if file exists + if (!file_exists($pFilename)) { + throw new Exception("Could not open " . $pFilename . " for reading! File does not exist."); + } + + // Read sample data (first 2 KB will do) + $fh = fopen($pFilename, 'r'); + $data = fread($fh, 2048); + fclose($fh); + + return true; + } + + /** + * Loads PHPExcel from file + * + * @param string $pFilename + * @return PHPExcel + * @throws Exception + */ + public function load($pFilename) + { + // Create new PHPExcel + $objPHPExcel = new PHPExcel(); + + // Load into this instance + return $this->loadIntoExisting($pFilename, $objPHPExcel); + } + + /** + * Read filter + * + * @return PHPExcel_Reader_IReadFilter + */ + public function getReadFilter() { + return $this->_readFilter; + } + + /** + * Set read filter + * + * @param PHPExcel_Reader_IReadFilter $pValue + */ + public function setReadFilter(PHPExcel_Reader_IReadFilter $pValue) { + $this->_readFilter = $pValue; + return $this; + } + + /** + * Set input encoding + * + * @param string $pValue Input encoding + */ + public function setInputEncoding($pValue = 'ANSI') + { + $this->_inputEncoding = $pValue; + return $this; + } + + /** + * Get input encoding + * + * @return string + */ + public function getInputEncoding() + { + return $this->_inputEncoding; + } + + // Data Array used for testing only, should write to PHPExcel object on completion of tests + private $_dataArray = array(); + + private $_tableLevel = 0; + private $_nestedColumn = array('A'); + + private function _setTableStartColumn($column) { + if ($this->_tableLevel == 0) + $column = 'A'; + ++$this->_tableLevel; + $this->_nestedColumn[$this->_tableLevel] = $column; + + return $this->_nestedColumn[$this->_tableLevel]; + } + + private function _getTableStartColumn() { + return $this->_nestedColumn[$this->_tableLevel]; + } + + private function _releaseTableStartColumn() { + --$this->_tableLevel; + return array_pop($this->_nestedColumn); + } + + private 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 , '
'; + // Write to worksheet to be done here... + // ... we return the cell so we can mess about with styles more easily + $cell = $sheet->setCellValue($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) ''; + } + + private function _processDomElement(DOMNode $element, $sheet, &$row, &$column, &$cellContent){ + foreach($element->childNodes as $child){ + if ($child instanceOf DOMText) { + $domText = preg_replace('/\s+/',' ',trim($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 'DOM ELEMENT: ' , strtoupper($child->nodeName) , '
'; + + $attributeArray = array(); + foreach($child->attributes as $attribute) { + echo 'ATTRIBUTE: ' , $attribute->name , ' => ' , $attribute->value , '
'; + $attributeArray[$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); + $cellContent = ''; + break; + case 'span' : + case 'div' : + case 'font' : + case 'i' : + case 'em' : + case 'strong': + case 'b' : + echo 'STYLING, SPAN OR DIV
'; + if ($cellContent > '') + $cellContent .= ' '; + $this->_processDomElement($child,$sheet,$row,$column,$cellContent); + if ($cellContent > '') + $cellContent .= ' '; + echo 'END OF STYLING, SPAN OR DIV
'; + 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: ' , '
'; + break; + case 'a' : + echo 'START OF HYPERLINK: ' , '
'; + foreach($attributeArray as $attributeName => $attributeValue) { + switch($attributeName) { + case 'href': + echo 'Link to ' , $attributeValue , '
'; + $sheet->getCell($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:' , '
'; + 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: ' , '
'; + $this->_processDomElement($child,$sheet,$row,$column,$cellContent); + echo 'END OF LIST ENTRY:' , '
'; + } else { + if ($cellContent > '') { + $this->_flushCell($sheet,$column,$row,$cellContent); + $row += 2; + } + echo 'START OF PARAGRAPH: ' , '
'; + $this->_processDomElement($child,$sheet,$row,$column,$cellContent); + echo 'END OF PARAGRAPH:' , '
'; + $this->_flushCell($sheet,$column,$row,$cellContent); + + if (isset($this->_formats[$child->nodeName])) { + $sheet->getStyle($column.$row)->applyFromArray($this->_formats[$child->nodeName]); + } + + $row += 2; + $column = 'A'; + } + break; + case 'li' : + if ($this->_tableLevel > 0) { + // If we're inside a table, replace with a \n + $cellContent .= "\n"; + echo 'LIST ENTRY: ' , '
'; + $this->_processDomElement($child,$sheet,$row,$column,$cellContent); + echo 'END OF LIST ENTRY:' , '
'; + } else { + if ($cellContent > '') { + $this->_flushCell($sheet,$column,$row,$cellContent); + } + ++$row; + echo 'LIST ENTRY: ' , '
'; + $this->_processDomElement($child,$sheet,$row,$column,$cellContent); + echo 'END OF LIST ENTRY:' , '
'; + $this->_flushCell($sheet,$column,$row,$cellContent); + $column = 'A'; + } + break; + case 'table' : + $this->_flushCell($sheet,$column,$row,$cellContent); + $column = $this->_setTableStartColumn($column); + echo 'START OF TABLE LEVEL ' , $this->_tableLevel , '
'; + if ($this->_tableLevel > 1) + --$row; + $this->_processDomElement($child,$sheet,$row,$column,$cellContent); + echo 'END OF TABLE LEVEL ' , $this->_tableLevel , '
'; + $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' : + ++$row; + $column = $this->_getTableStartColumn(); + $cellContent = ''; + echo 'START OF TABLE ' , $this->_tableLevel , ' ROW
'; + $this->_processDomElement($child,$sheet,$row,$column,$cellContent); + echo 'END OF TABLE ' , $this->_tableLevel , ' ROW
'; + break; + case 'th' : + case 'td' : + echo 'START OF TABLE ' , $this->_tableLevel , ' CELL
'; + $this->_processDomElement($child,$sheet,$row,$column,$cellContent); + echo 'END OF TABLE ' , $this->_tableLevel , ' CELL
'; + $this->_flushCell($sheet,$column,$row,$cellContent); + ++$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); + } + } + } + } + + /** + * Loads PHPExcel from file into PHPExcel instance + * + * @param string $pFilename + * @param PHPExcel $objPHPExcel + * @return PHPExcel + * @throws Exception + */ + public function loadIntoExisting($pFilename, PHPExcel $objPHPExcel) + { + // Check if file exists + if (!file_exists($pFilename)) { + throw new Exception("Could not open " . $pFilename . " for reading! File does not exist."); + } + + // Create new PHPExcel + while ($objPHPExcel->getSheetCount() <= $this->_sheetIndex) { + $objPHPExcel->createSheet(); + } + $objPHPExcel->setActiveSheetIndex( $this->_sheetIndex ); + + // Create a new DOM object + $dom = new domDocument; + // Load the HTML file into the DOM object + $loaded = $dom->loadHTMLFile($pFilename); + if ($loaded === false) { + throw new Exception('Failed to load ',$pFilename,' as a DOM Document'); + } + + // Discard white space + $dom->preserveWhiteSpace = false; + + + $row = 0; + $column = 'A'; + $content = ''; + $this->_processDomElement($dom,$objPHPExcel->getActiveSheet(),$row,$column,$content); + + echo '
'; + var_dump($this->_dataArray); + + // Return + return $objPHPExcel; + } + + /** + * Get sheet index + * + * @return int + */ + public function getSheetIndex() { + return $this->_sheetIndex; + } + + /** + * Set sheet index + * + * @param int $pValue Sheet index + * @return PHPExcel_Reader_HTML + */ + public function setSheetIndex($pValue = 0) { + $this->_sheetIndex = $pValue; + return $this; + } + +}