(retrofir from trunk) N° 615: spreadsheet export enhancement to remove unneeded line breaks.

SVN:2.3[4550]

core/spreadsheetbulkexport.class.inc.php

@@ -31,6 +31,7 @@ class SpreadsheetBulkExport extends TabularBulkExport
 		$oP->p(" *\tfields: (mandatory) the comma separated list of field codes to export (e.g: name,org_id,service_name...).");
 		$oP->p(" *\tno_localize: (optional) pass 1 to retrieve the raw (untranslated) values for enumerated fields. Default: 0.");
 		$oP->p(" *\tdate_format: the format to use when exporting date and time fields (default = the SQL format). e.g. 'Y-m-d H:i:s'");
+		$oP->p(" *\tformatted_text: set to 1 to formatted text fields with their HTML markup, 0 to remove formatting. Default is 1 (= formatted text)");
 	}
 
 	public function EnumFormParts()
@@ -51,12 +52,19 @@ class SpreadsheetBulkExport extends TabularBulkExport
 				$oP->add('<fieldset><legend>'.Dict::S('Core:BulkExport:SpreadsheetOptions').'</legend>');
 				$oP->add('<table>');
 				$oP->add('<tr>');
-				$oP->add('<td><input type="checkbox" id="spreadsheet_no_localize" name="no_localize" value="1"'.$sChecked.'><label for="spreadsheet_no_localize"> '.Dict::S('Core:BulkExport:OptionNoLocalize').'</label></td>');
+
-				
+				$oP->add('<td style="vertical-align:top">');
+				$sChecked = (utils::ReadParam('formatted_text', 1) == 1) ? ' checked ' : '';
+				$oP->add('<h3>'.Dict::S('Core:BulkExport:TextFormat').'</h3>');
+				$oP->add('<input type="hidden" name="formatted_text" value="0">'); // Trick to pass the zero value if the checkbox below is unchecked, since we want the default value to be "1"
+				$oP->add('<input type="checkbox" id="spreadsheet_formatted_text" name="formatted_text" value="1"'.$sChecked.'><label for="spreadsheet_formatted_text"> '.Dict::S('Core:BulkExport:OptionFormattedText').'</label><br/><br/>');
+				$oP->add('<input type="checkbox" id="spreadsheet_no_localize" name="no_localize" value="1"'.$sChecked.'><label for="spreadsheet_no_localize"> '.Dict::S('Core:BulkExport:OptionNoLocalize').'</label>');
+				$oP->add('</td>');
+
 				$sDateTimeFormat = utils::ReadParam('date_format', (string)AttributeDateTime::GetFormat(), true, 'raw_data');
 				$sDefaultChecked = ($sDateTimeFormat == (string)AttributeDateTime::GetFormat()) ? ' checked' : '';
 				$sCustomChecked = ($sDateTimeFormat !== (string)AttributeDateTime::GetFormat()) ? ' checked' : '';
-				
+
 				$oP->add('<td>');
 				$oP->add('<h3>'.Dict::S('Core:BulkExport:DateTimeFormat').'</h3>');
 				$sDefaultFormat = htmlentities((string)AttributeDateTime::GetFormat(), ENT_QUOTES, 'UTF-8');
@@ -65,23 +73,23 @@ class SpreadsheetBulkExport extends TabularBulkExport
 				$sFormatInput = '<input type="text" size="15" name="date_format" id="spreadsheet_custom_date_time_format" title="" value="'.htmlentities($sDateTimeFormat, ENT_QUOTES, 'UTF-8').'"/>';
 				$oP->add('<input type="radio" id="spreadsheet_date_time_format_custom" name="spreadsheet_date_format_radio" value="custom"'.$sCustomChecked.'><label for="spreadsheet_date_time_format_custom"> '.Dict::Format('Core:BulkExport:DateTimeFormatCustom_Format', $sFormatInput).'</label>');
 				$oP->add('</td>');
-				
+
 				$oP->add('</tr>');
 				$oP->add('</table>');
 				$oP->add('</fieldset>');
 				$sJSTooltip = json_encode('<div class="date_format_tooltip">'.Dict::S('UI:CSVImport:CustomDateTimeFormatTooltip').'</div>');
 				$oP->add_ready_script(
-<<<EOF
+						<<<EOF
 $('#spreadsheet_custom_date_time_format').tooltip({content: function() { return $sJSTooltip; } });
 $('#form_part_spreadsheet_options').on('preview_updated', function() { FormatDatesInPreview('spreadsheet', 'spreadsheet'); });
 $('#spreadsheet_date_time_format_default').on('click', function() { FormatDatesInPreview('spreadsheet', 'spreadsheet'); });
 $('#spreadsheet_date_time_format_custom').on('click', function() { FormatDatesInPreview('spreadsheet', 'spreadsheet'); });
 $('#spreadsheet_custom_date_time_format').on('click', function() { $('#spreadsheet_date_time_format_custom').prop('checked', true); });
-$('#spreadsheet_custom_date_time_format').on('click', function() { $('#spreadsheet_date_time_format_custom').prop('checked', true); FormatDatesInPreview('spreadsheet', 'spreadsheet'); }).on('keyup', function() { FormatDatesInPreview('spreadsheet', 'spreadsheet'); });											
+$('#spreadsheet_custom_date_time_format').on('click', function() { $('#spreadsheet_date_time_format_custom').prop('checked', true); FormatDatesInPreview('spreadsheet', 'spreadsheet'); }).on('keyup', function() { FormatDatesInPreview('spreadsheet', 'spreadsheet'); });
 EOF
-				);
+						);
 				break;
-				
+
 			default:
 				return parent:: DisplayFormPart($oP, $sPartId);
 		}
@@ -90,26 +98,27 @@ EOF
 	public function ReadParameters()
 	{
 		parent::ReadParameters();
+		$this->aStatusInfo['formatted_text'] = (bool)utils::ReadParam('formatted_text', 1, true);
 
 		$sDateFormatRadio = utils::ReadParam('spreadsheet_date_format_radio', '');
 		switch($sDateFormatRadio)
 		{
 			case 'default':
-			// Export from the UI => format = same as is the UI
+				// Export from the UI => format = same as is the UI
-			$this->aStatusInfo['date_format'] = (string)AttributeDateTime::GetFormat();
+				$this->aStatusInfo['date_format'] = (string)AttributeDateTime::GetFormat();
-			break;
+				break;
-			
+					
 			case 'custom':
-			// Custom format specified from the UI
+				// Custom format specified from the UI
-			$this->aStatusInfo['date_format'] = utils::ReadParam('date_format', (string)AttributeDateTime::GetFormat(), true, 'raw_data');
+				$this->aStatusInfo['date_format'] = utils::ReadParam('date_format', (string)AttributeDateTime::GetFormat(), true, 'raw_data');
-			break;
+				break;
-			
+					
 			default:
-			// Export from the command line (or scripted) => default format is SQL, as in previous versions of iTop, unless specified otherwise
+				// Export from the command line (or scripted) => default format is SQL, as in previous versions of iTop, unless specified otherwise
-			$this->aStatusInfo['date_format'] = utils::ReadParam('date_format', (string)AttributeDateTime::GetSQLFormat(), true, 'raw_data');
+				$this->aStatusInfo['date_format'] = utils::ReadParam('date_format', (string)AttributeDateTime::GetSQLFormat(), true, 'raw_data');
 		}
 	}
-	
+
 	protected function GetSampleData($oObj, $sAttCode)
 	{
 		if ($sAttCode != 'id')
@@ -126,6 +135,7 @@ EOF
 
 	protected function GetValue($oObj, $sAttCode)
 	{
+		$bFormattedText =  (array_key_exists('formatted_text', $this->aStatusInfo) ? $this->aStatusInfo['formatted_text'] : false);
 		switch($sAttCode)
 		{
 			case 'id':
@@ -147,6 +157,18 @@ EOF
 				{
 					$sRet = '';
 				}
+				elseif ($oAttDef instanceof AttributeText)
+				{
+					if ($bFormattedText)
+					{
+						// Replace paragraphs (<p...>...</p>, etc) by line breaks (<br/>) since Excel (pre-2016) splits the cells when there is a paragraph
+						$sRet = static::HtmlToSpreadsheet($oObj->GetAsHTML($sAttCode));
+					}
+					else
+					{
+						$sRet = utils::HtmlToText($oObj->GetAsHTML($sAttCode));
+					}
+				}
 				elseif ($oAttDef instanceof AttributeString)
 				{
 					$sRet = $oObj->GetAsHTML($sAttCode);
@@ -175,7 +197,7 @@ EOF
 	{
 		// Integration within MS-Excel web queries + HTTPS + IIS:
 		// MS-IIS set these header values with no-cache... while Excel fails to do the job if using HTTPS
-		// Then the fix is to force the reset of header values Pragma and Cache-control 
+		// Then the fix is to force the reset of header values Pragma and Cache-control
 		$oPage->add_header("Pragma:", true);
 		$oPage->add_header("Cache-control:", true);
 	}
@@ -230,13 +252,14 @@ EOF
 		$oSet = new DBObjectSet($this->oSearch);
 		$oSet->SetLimit($this->iChunkSize, $this->aStatusInfo['position']);
 		$this->OptimizeColumnLoad($oSet);
-		
+
 		$sExportDateTimeFormat = $this->aStatusInfo['date_format'];
+		$bFormattedText =  (array_key_exists('formatted_text', $this->aStatusInfo) ? $this->aStatusInfo['formatted_text'] : false);
 		// Date & time formats
 		$oDateTimeFormat = new DateTimeFormat($sExportDateTimeFormat);
 		$oDateFormat = new DateTimeFormat($oDateTimeFormat->ToDateFormat());
 		$oTimeFormat = new DateTimeFormat($oDateTimeFormat->ToTimeFormat());
-		
+
 		$iCount = 0;
 		$sData = '';
 		$iPreviousTimeLimit = ini_get('max_execution_time');
@@ -258,55 +281,69 @@ EOF
 					$sData .= "<td x:str></td>";
 					continue;
 				}
-				
+
 				switch($sAttCode)
 				{
 					case 'id':
-					$sField = $oObj->GetKey();
+						$sField = $oObj->GetKey();
-					$sData .= "<td>$sField</td>";
+						$sData .= "<td>$sField</td>";
-					break;
+						break;
 							
 					default:
-					$oAttDef = MetaModel::GetAttributeDef(get_class($oObj), $sAttCode);
+						$oAttDef = MetaModel::GetAttributeDef(get_class($oObj), $sAttCode);
-					$oFinalAttDef = $oAttDef->GetFinalAttDef();
+						$oFinalAttDef = $oAttDef->GetFinalAttDef();
-					if (get_class($oFinalAttDef) == 'AttributeDateTime')
+						if (get_class($oFinalAttDef) == 'AttributeDateTime')
-					{
-						// Split the date and time in two columns
-						$sDate = $oDateFormat->Format($oObj->Get($sAttCode));
-						$sTime = $oTimeFormat->Format($oObj->Get($sAttCode));
-						$sData .= "<td>$sDate</td>";
-						$sData .= "<td>$sTime</td>";
-					}
-					else if (get_class($oFinalAttDef) == 'AttributeDate')
-					{
-						$sDate = $oDateFormat->Format($oObj->Get($sAttCode));
-						$sData .= "<td>$sDate</td>";
-					}
-					else if($oAttDef instanceof AttributeCaseLog)
-					{
-						$rawValue = $oObj->Get($sAttCode);
-						$sField = str_replace("\n", "<br/>", htmlentities($rawValue->__toString(), ENT_QUOTES, 'UTF-8'));
-						// Trick for Excel: treat the content as text even if it begins with an equal sign
-						$sData .= "<td x:str>$sField</td>";
-					}
-					else if($oAttDef instanceof AttributeString)
-					{
-						$sField = $oObj->GetAsHTML($sAttCode, $this->bLocalizeOutput);
-						$sData .= "<td x:str>$sField</td>";
-					}
-					else
-					{
-						$rawValue = $oObj->Get($sAttCode);
-						if ($this->bLocalizeOutput)
 						{
-							$sField = htmlentities($oFinalAttDef->GetEditValue($rawValue), ENT_QUOTES, 'UTF-8');
+							// Split the date and time in two columns
+							$sDate = $oDateFormat->Format($oObj->Get($sAttCode));
+							$sTime = $oTimeFormat->Format($oObj->Get($sAttCode));
+							$sData .= "<td>$sDate</td>";
+							$sData .= "<td>$sTime</td>";
+						}
+						else if (get_class($oFinalAttDef) == 'AttributeDate')
+						{
+							$sDate = $oDateFormat->Format($oObj->Get($sAttCode));
+							$sData .= "<td>$sDate</td>";
+						}
+						else if($oAttDef instanceof AttributeCaseLog)
+						{
+							$rawValue = $oObj->Get($sAttCode);
+							$sField = str_replace("\n", "<br/>", htmlentities($rawValue->__toString(), ENT_QUOTES, 'UTF-8'));
+							// Trick for Excel: treat the content as text even if it begins with an equal sign
+							$sData .= "<td x:str>$sField</td>";
+						}
+						elseif ($oAttDef instanceof AttributeText)
+						{
+							if ($bFormattedText)
+							{
+								// Replace paragraphs (<p...>...</p>, etc) by line breaks (<br/>) since Excel (pre-2016) splits the cells when there is a paragraph
+								$sField = static::HtmlToSpreadsheet($oObj->GetAsHTML($sAttCode));
+							}
+							else
+							{
+								// Convert to plain text
+								$sField = utils::HtmlToText($oObj->GetAsHTML($sAttCode));
+							}
+							$sData .= "<td x:str>$sField</td>";
+						}
+						else if($oAttDef instanceof AttributeString)
+						{
+							$sField = $oObj->GetAsHTML($sAttCode, $this->bLocalizeOutput);
+							$sData .= "<td x:str>$sField</td>";
 						}
 						else
 						{
-							$sField = htmlentities($rawValue, ENT_QUOTES, 'UTF-8');
+							$rawValue = $oObj->Get($sAttCode);
+							if ($this->bLocalizeOutput)
+							{
+								$sField = htmlentities($oFinalAttDef->GetEditValue($rawValue), ENT_QUOTES, 'UTF-8');
+							}
+							else
+							{
+								$sField = htmlentities($rawValue, ENT_QUOTES, 'UTF-8');
+							}
+							$sData .= "<td>$sField</td>";
 						}
-						$sData .= "<td>$sField</td>";
-					}
 				}
 
 			}
@@ -354,4 +391,51 @@ EOF
 	{
 		return 'html';
 	}
+
+	/**
+	 * Cleanup all markup displayed as line breaks (except <br> tags) since this
+	 * causes Excel (pre-2016) to generate extra lines in the table, thus breaking
+	 * the tabular disposition of the export
+	 * Note: Excel 2016 also refuses line breaks, so the only solution for this case is alas plain text
+	 * @param string $sHtml The HTML to cleanup
+	 * @return string The cleaned HTML
+	 */
+	public static function HtmlToSpreadsheet($sHtml)
+	{
+		if (trim(strip_tags($sHtml)) === '')
+		{
+			// Display this value as an empty cell in the table
+			return '&nbsp;';
+		}
+		// The tags listed here are a subset of the whitelist defined in HTMLDOMSanitizer
+		// Tags causing a visual "line break" in the displayed page (i.e. display: block) are to be replaced by a <span> followed by a <br/>
+		// in order to preserve any inline style/attribute of the removed tag
+		$aTagsToReplace = array(
+				'pre', 'div', 'p', 'hr', 'center', 'h1', 'h2', 'h3', 'h4', 'li', 'fieldset', 'legend', 'nav', 'section', 'tr', 'caption',
+		);
+		// Tags to completely remove from the markup
+		$aTagsToRemove = array(
+				'table', 'thead', 'tbody', 'ul', 'ol', 'td', 'th',
+		);
+
+		// Remove the englobing <div class="HTML" >...</div> to prevent an extra line break
+		$sHtml = preg_replace('|^<div class="HTML" >(.*)</div>$|s', '$1', $sHtml); // Must use the "s" (. matches newline) modifier
+		
+		foreach($aTagsToReplace as $sTag)
+		{
+			$sHtml = preg_replace("|<{$sTag} ?([^>]*)>|is", '<span $1>', $sHtml);
+			$sHtml = preg_replace("|</{$sTag}>|i", '</span><br/>', $sHtml);
+		}
+
+		foreach($aTagsToRemove as $sTag)
+		{
+			$sHtml = preg_replace("|<{$sTag} ?([^>]*)>|is", '', $sHtml);
+			$sHtml = preg_replace("|</{$sTag}>|i", '', $sHtml);
+		}
+
+		// Remove any trailing <br/>, if any, to prevent an extra line break
+		$sHtml = preg_replace("|<br/>$|", '', $sHtml);
+
+		return $sHtml;
+	}
 }