GH-554 - Whitespace after toRichTextObject()

Classes/PHPExcel/Helper/HTML.php

@@ -3,7 +3,7 @@
 class PHPExcel_Helper_HTML
 {
     protected static $colourMap = array(
-        'aliceblue' => 'f0f8ff',
+        'aliceblue' => 'f0f8ff', 
         'antiquewhite' => 'faebd7',
         'antiquewhite1' => 'ffefdb',
         'antiquewhite2' => 'eedfcc',
@@ -526,12 +526,12 @@ class PHPExcel_Helper_HTML
     protected $size;
     protected $color;
 
-    protected $bold = false;
+	protected $bold = false;
-    protected $italic = false;
+	protected $italic = false;
-    protected $underline = false;
+	protected $underline = false;
-    protected $superscript = false;
+	protected $superscript = false;
-    protected $subscript = false;
+	protected $subscript = false;
-    protected $strikethrough = false;
+	protected $strikethrough = false;
 
     protected $startTagCallbacks = array(
         'font' => 'startFontTag',
@@ -573,8 +573,7 @@ class PHPExcel_Helper_HTML
 
     protected $richTextObject;
 
-    protected function initialise()
+    protected function initialise() {
-    {
         $this->face = $this->size = $this->color = null;
         $this->bold = $this->italic = $this->underline = $this->superscript = $this->subscript = $this->strikethrough = false;
 
@@ -583,30 +582,44 @@ class PHPExcel_Helper_HTML
         $this->stringData = '';
     }
 
-    public function toRichTextObject($html)
+    public function toRichTextObject($html) {
-    {
         $this->initialise();
 
-        //    Create a new DOM object
+        //	Create a new DOM object
         $dom = new domDocument;
-        //    Load the HTML file into the DOM object
+        //	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);
 
-        //    Discard excess white space
+        //	Discard excess white space
         $dom->preserveWhiteSpace = false;
 
-        $this->richTextObject = new PHPExcel_RichText();
+        $this->richTextObject = new PHPExcel_RichText();;
         $this->parseElements($dom);
+
+        // Clean any further spurious whitespace
+        $this->cleanWhitespace();
+
         return $this->richTextObject;
     }
 
-    protected function buildTextRun()
+    protected function cleanWhitespace() {
-    {
+        foreach($this->richTextObject->getRichTextElements() as $key => $element) {
-        $text = $this->stringData;
+            $text = $element->getText();
-        if (trim($text) === '') {
+            // Trim any leading spaces on the first run
-            return;
+            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 = $this->stringData;
+        if (trim($text) === '')
+            return;
 
         $richtextRun = $this->richTextObject->createTextRun($this->stringData);
         if ($this->face) {
@@ -616,7 +629,7 @@ class PHPExcel_Helper_HTML
             $richtextRun->getFont()->setSize($this->size);
         }
         if ($this->color) {
-            $richtextRun->getFont()->setColor(new PHPExcel_Style_Color('ff' . $this->color));
+            $richtextRun->getFont()->setColor( new PHPExcel_Style_Color( 'ff' . $this->color ) );
         }
         if ($this->bold) {
             $richtextRun->getFont()->setBold(true);
@@ -639,22 +652,19 @@ class PHPExcel_Helper_HTML
         $this->stringData = '';
     }
 
-    protected function rgbToColour($rgb)
+    protected function rgbToColour($rgb) {
-    {
         preg_match_all('/\d+/', $rgb, $values);
-        foreach ($values[0] as &$value) {
+        foreach($values[0] as &$value) {
             $value = str_pad(dechex($value), 2, '0', STR_PAD_LEFT);
         }
         return implode($values[0]);
     }
 
-    protected function colourNameLookup($rgb)
+    protected function colourNameLookup($rgb) {
-    {
         return self::$colourMap[$rgb];
     }
 
-    protected function startFontTag($tag)
+    protected function startFontTag($tag) {
-    {
         foreach ($tag->attributes as $attribute) {
             $attributeName = strtolower($attribute->name);
             $attributeValue = $attribute->value;
@@ -662,7 +672,7 @@ class PHPExcel_Helper_HTML
             if ($attributeName == 'color') {
                 if (preg_match('/rgb\s*\(/', $attributeValue)) {
                     $this->$attributeName = $this->rgbToColour($attributeValue);
-                } elseif (strpos(trim($attributeValue), '#') === 0) {
+                } elseif(strpos(trim($attributeValue), '#') === 0) {
                     $this->$attributeName = ltrim($attributeValue, '#');
                 } else {
                     $this->$attributeName = $this->colourNameLookup($attributeValue);
@@ -673,85 +683,69 @@ class PHPExcel_Helper_HTML
         }
     }
 
-    protected function endFontTag()
+    protected function endFontTag() {
-    {
         $this->face = $this->size = $this->color = null;
     }
 
-    protected function startBoldTag()
+    protected function startBoldTag() {
-    {
         $this->bold = true;
     }
 
-    protected function endBoldTag()
+    protected function endBoldTag() {
-    {
         $this->bold = false;
     }
 
-    protected function startItalicTag()
+    protected function startItalicTag() {
-    {
         $this->italic = true;
     }
 
-    protected function endItalicTag()
+    protected function endItalicTag() {
-    {
         $this->italic = false;
     }
 
-    protected function startUnderlineTag()
+    protected function startUnderlineTag() {
-    {
         $this->underline = true;
     }
 
-    protected function endUnderlineTag()
+    protected function endUnderlineTag() {
-    {
         $this->underline = false;
     }
 
-    protected function startSubscriptTag()
+    protected function startSubscriptTag() {
-    {
         $this->subscript = true;
     }
 
-    protected function endSubscriptTag()
+    protected function endSubscriptTag() {
-    {
         $this->subscript = false;
     }
 
-    protected function startSuperscriptTag()
+    protected function startSuperscriptTag() {
-    {
         $this->superscript = true;
     }
 
-    protected function endSuperscriptTag()
+    protected function endSuperscriptTag() {
-    {
         $this->superscript = false;
     }
 
-    protected function startStrikethruTag()
+    protected function startStrikethruTag() {
-    {
         $this->strikethrough = true;
     }
 
-    protected function endStrikethruTag()
+    protected function endStrikethruTag() {
-    {
         $this->strikethrough = false;
     }
 
-    protected function breakTag()
+    protected function breakTag() {
-    {
+        $this->stringData .= "\n";
-        $this->stringData .= PHP_EOL;
     }
 
-    protected function parseTextNode(DOMText $textNode)
+    protected function parseTextNode(DOMText $textNode) {
-    {
+        $domText = preg_replace('/\s+/u', ' ', str_replace(["\r", "\n"], ' ', $textNode->nodeValue));
-        $domText = preg_replace('/\s+/u', ' ', ltrim($textNode->nodeValue));
         $this->stringData .= $domText;
         $this->buildTextRun();
     }
 
-    protected function handleCallback($element, $callbackTag, $callbacks)
+    protected function handleCallback($element, $callbackTag, $callbacks) {
-    {
         if (isset($callbacks[$callbackTag])) {
             $elementHandler = $callbacks[$callbackTag];
             if (method_exists($this, $elementHandler)) {
@@ -760,22 +754,20 @@ class PHPExcel_Helper_HTML
         }
     }
 
-    protected function parseElementNode(DOMElement $element)
+    protected function parseElementNode(DOMElement $element) {
-    {
         $callbackTag = strtolower($element->nodeName);
         $this->stack[] = $callbackTag;
 
         $this->handleCallback($element, $callbackTag, $this->startTagCallbacks);
 
         $this->parseElements($element);
-        $this->stringData .= ' ';
+//        $this->stringData .= ' ';
         array_pop($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) {
                 $this->parseTextNode($child);

Examples/42richText.php

@@ -2,7 +2,7 @@
 /**
  * PHPExcel
  *
- * Copyright (c) 2006 - 2015 PHPExcel
+ * Copyright (C) 2006 - 2014 PHPExcel
  *
  * This library is free software; you can redistribute it and/or
  * modify it under the terms of the GNU Lesser General Public
@@ -20,7 +20,7 @@
  *
  * @category   PHPExcel
  * @package    PHPExcel
- * @copyright  Copyright (c) 2006 - 2015 PHPExcel (http://www.codeplex.com/PHPExcel)
+ * @copyright  Copyright (c) 2006 - 2014 PHPExcel (http://www.codeplex.com/PHPExcel)
  * @license    http://www.gnu.org/licenses/old-licenses/lgpl-2.1.txt	LGPL
  * @version    ##VERSION##, ##DATE##
  */
@@ -55,7 +55,7 @@ $objPHPExcel->getProperties()->setCreator("Maarten Balliauw")
 // Add some data
 echo date('H:i:s') , " Add some data" , EOL;
 
-$html1='<font color="#0000ff">
+$html1 = '<font color="#0000ff">
 <h1 align="center">My very first example of rich text<br />generated from html markup</h1>
 <p>
 <font size="14" COLOR="rgb(0,255,128)">
@@ -64,11 +64,11 @@ while this block uses an <u>underline</u>.
 </font>
 </p>
 <p align="right"><font size="9" color="red">
-I want to eat <ins><del>healthy food</del><strong>pizza</strong></ins>.
+I want to eat <ins><del>healthy food</del> <strong>pizza</strong></ins>.
 </font>
 ';
 
-$html2='<p>
+$html2 = '<p>
 <font color="#ff0000">
     100&deg;C is a hot temperature
 </font>
@@ -78,9 +78,11 @@ $html2='<p>
 </font>
 </p>';
 
-$html3='2<sup>3</sup> equals 8';
+$html3 = '2<sup>3</sup> equals 8';
 
-$html4='H<sub>2</sub>SO<sub>4</sub> is the chemical formula for Sulphuric acid';
+$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;
@@ -97,7 +99,7 @@ $objPHPExcel->getActiveSheet()->getStyle('A1')
 
 $richText = $wizard->toRichTextObject($html2);
 
-$objPHPExcel->setActiveSheetIndex(0)
+$objPHPExcel->getActiveSheet()
     ->setCellValue('A2', $richText);
 
 $objPHPExcel->getActiveSheet()->getRowDimension(1)->setRowHeight(-1);
@@ -105,12 +107,15 @@ $objPHPExcel->getActiveSheet()->getStyle('A2')
     ->getAlignment()
     ->setWrapText(true);
 
-$objPHPExcel->setActiveSheetIndex(0)
+$objPHPExcel->getActiveSheet()
-            ->setCellValue('A3', $wizard->toRichTextObject($html3));
+    ->setCellValue('A3', $wizard->toRichTextObject($html3));
 
-$objPHPExcel->setActiveSheetIndex(0)
+$objPHPExcel->getActiveSheet()
     ->setCellValue('A4', $wizard->toRichTextObject($html4));
 
+$objPHPExcel->getActiveSheet()
+    ->setCellValue('A5', $wizard->toRichTextObject($html5));
+
 
 // Rename worksheet
 echo date('H:i:s') , " Rename worksheet" , EOL;

changelog.txt

@@ -27,6 +27,7 @@ Planned for 1.8.2
 - 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
 - General:  (umpirsky)        Work Item GH-548  - Optimize vlookup() sort
+- Bugfix:   (MBaker)          Work Item GH-554  - Whitespace after toRichTextObject()
 
 
 2015-04-30 (v1.8.1):