Bugfix: Work Item GH-554 - Whitespace after toRichTextObject()

Examples/42richText.php

@@ -34,12 +34,12 @@ date_default_timezone_set('Europe/London');
 define('EOL',(PHP_SAPI == 'cli') ? PHP_EOL : '<br />');
 
 /** Include PHPExcel */
-require_once dirname(__FILE__) . '/../Classes/PHPExcel.php';
+require_once dirname(__FILE__) . '/../src/Bootstrap.php';
 
 
 // Create new PHPExcel object
 echo date('H:i:s') , " Create new PHPExcel object" , EOL;
-$objPHPExcel = new PHPExcel();
+$objPHPExcel = new \PHPExcel\Spreadsheet();
 
 // Set document properties
 echo date('H:i:s') , " Set document properties" , EOL;
@@ -82,26 +82,34 @@ $html3='2<sup>3</sup> equals 8';
 
 $html4 = 'H<sub>2</sub>SO<sub>4</sub> is the chemical formula for Sulphuric acid';
 
+$html5 = '<strong>bold</strong>, <em>italic</em>, <strong><em>bold+italic</em></strong>';
 
-$wizard = new PHPExcel_Helper_HTML;
+$wizard = new \PHPExcel\Helper\HTML;
 $richText = $wizard->toRichTextObject($html1);
 
-$objPHPExcel->setActiveSheetIndex(0)
+$objPHPExcel->getActiveSheet()
     ->setCellValue('A1', $richText);
 
-$objPHPExcel->getActiveSheet()->getColumnDimension('A')->setWidth(48);
+$objPHPExcel->getActiveSheet()
-$objPHPExcel->getActiveSheet()->getRowDimension(1)->setRowHeight(-1);
+    ->getColumnDimension('A')
+    ->setWidth(48);
+$objPHPExcel->getActiveSheet()
+    ->getRowDimension(1)
+    ->setRowHeight(-1);
 $objPHPExcel->getActiveSheet()->getStyle('A1')
     ->getAlignment()
     ->setWrapText(true);
 
 $richText = $wizard->toRichTextObject($html2);
 
-$objPHPExcel->setActiveSheetIndex(0)
+$objPHPExcel->getActiveSheet()
     ->setCellValue('A2', $richText);
 
-$objPHPExcel->getActiveSheet()->getRowDimension(1)->setRowHeight(-1);
+$objPHPExcel->getActiveSheet()
-$objPHPExcel->getActiveSheet()->getStyle('A2')
+    ->getRowDimension(1)
+    ->setRowHeight(-1);
+$objPHPExcel->getActiveSheet()
+    ->getStyle('A2')
     ->getAlignment()
     ->setWrapText(true);
 
@@ -111,10 +119,14 @@ $objPHPExcel->setActiveSheetIndex(0)
 $objPHPExcel->setActiveSheetIndex(0)
     ->setCellValue('A4', $wizard->toRichTextObject($html4));
 
+$objPHPExcel->setActiveSheetIndex(0)
+    ->setCellValue('A5', $wizard->toRichTextObject($html5));
+
 
 // Rename worksheet
 echo date('H:i:s') , " Rename worksheet" , EOL;
-$objPHPExcel->getActiveSheet()->setTitle('Simple');
+$objPHPExcel->getActiveSheet()
+    ->setTitle('Rich Text Examples');
 
 
 // Set active sheet index to the first sheet, so Excel opens this as the first sheet
@@ -125,7 +137,7 @@ $objPHPExcel->setActiveSheetIndex(0);
 echo date('H:i:s') , " Write to Excel2007 format" , EOL;
 $callStartTime = microtime(true);
 
-$objWriter = PHPExcel_IOFactory::createWriter($objPHPExcel, 'Excel2007');
+$objWriter = \PHPExcel\IOFactory::createWriter($objPHPExcel, 'Excel2007');
 $objWriter->save(str_replace('.php', '.xlsx', __FILE__));
 $callEndTime = microtime(true);
 $callTime = $callEndTime - $callStartTime;
@@ -140,7 +152,7 @@ echo date('H:i:s') , ' Current memory usage: ' , (memory_get_usage(true) / 1024
 echo date('H:i:s') , " Write to Excel5 format" , EOL;
 $callStartTime = microtime(true);
 
-$objWriter = PHPExcel_IOFactory::createWriter($objPHPExcel, 'Excel5');
+$objWriter = \PHPExcel\IOFactory::createWriter($objPHPExcel, 'Excel5');
 $objWriter->save(str_replace('.php', '.xls', __FILE__));
 $callEndTime = microtime(true);
 $callTime = $callEndTime - $callStartTime;

changelog.txt

@@ -23,12 +23,12 @@
 **************************************************************************************
 
 
-Planned for 1.8.2
+Planned for 1.9
 - Bugfix:   (MBaker)                            - Fix to getCell() method when cell reference includes a worksheet reference
 - Bugfix:   (ncrypthic)       Work Item GH-570  - Ignore inlineStr type if formula element exists
+- Bugfix:   (MBaker)          Work Item GH-554  - Whitespace after toRichTextObject()
 - General:  (umpirsky)        Work Item GH-548  - Optimize vlookup() sort
 
-
 2015-04-30 (v1.8.1):
 - Bugfix:   (goncons)         Work Item GH-397  - Fix for Writing an Open Document cell with non-numeric formula
 - Bugfix:   (sarciszewski)    Work Item GH-329  - Avoid potential divide by zero in basedrawing

src/PhpSpreadsheet/Helper/HTML.php

@@ -615,7 +615,7 @@ class HTML
         $this->initialise();
 
         //    Create a new DOM object
-        $dom = new domDocument;
+        $dom = new \DOMDocument;
         //    Load the HTML file into the DOM object
         //  Note the use of error suppression, because typically this will be an html fragment, so not fully valid markup
         $loaded = @$dom->loadHTML($html);
@@ -625,9 +625,27 @@ class HTML
 
         $this->richTextObject = new \PHPExcel\RichText();
         $this->parseElements($dom);
+
+        // Clean any further spurious whitespace
+        $this->cleanWhitespace();
+
         return $this->richTextObject;
     }
 
+    protected function cleanWhitespace()
+    {
+        foreach($this->richTextObject->getRichTextElements() as $key => $element) {
+            $text = $element->getText();
+            // Trim any leading spaces on the first run
+            if ($key == 0) {
+                $text = ltrim($text);
+            }
+            // Trim any spaces immediately after a line break
+            $text = preg_replace('/\n */mu', "\n", $text);
+            $element->setText($text);
+        }
+    }
+
     protected function buildTextRun()
     {
         $text = $this->stringData;
@@ -767,12 +785,16 @@ class HTML
 
     protected function breakTag()
     {
-        $this->stringData .= PHP_EOL;
+        $this->stringData .= "\n";
     }
 
-    protected function parseTextNode(DOMText $textNode)
+    protected function parseTextNode(\DOMText $textNode)
     {
-        $domText = preg_replace('/\s+/u', ' ', ltrim($textNode->nodeValue));
+        $domText = preg_replace(
+            '/\s+/u',
+            ' ',
+            str_replace(["\r", "\n"], ' ', $textNode->nodeValue)
+        );
         $this->stringData .= $domText;
         $this->buildTextRun();
     }
@@ -787,7 +809,7 @@ class HTML
         }
     }
 
-    protected function parseElementNode(DOMElement $element)
+    protected function parseElementNode(\DOMElement $element)
     {
         $callbackTag = strtolower($element->nodeName);
         $this->stack[] = $callbackTag;
@@ -795,18 +817,17 @@ class HTML
         $this->handleCallback($element, $callbackTag, $this->startTagCallbacks);
 
         $this->parseElements($element);
-        $this->stringData .= ' ';
         array_pop($this->stack);
 
         $this->handleCallback($element, $callbackTag, $this->endTagCallbacks);
     }
 
-    protected function parseElements(DOMNode $element)
+    protected function parseElements(\DOMNode $element)
     {
         foreach ($element->childNodes as $child) {
-            if ($child instanceof DOMText) {
+            if ($child instanceof \DOMText) {
                 $this->parseTextNode($child);
-            } elseif ($child instanceof DOMElement) {
+            } elseif ($child instanceof \DOMElement) {
                 $this->parseElementNode($child);
             }
         }