InputfieldTinyMCE: Add support for new "remove all style attributes" option to the Markup Toggle settings. Plus refactoring of the pasteFilter JS in attempt to fix processwire/processwire-issues#1866 which should improve pasting from MS Word.

wire/modules/Inputfield/InputfieldTinyMCE/InputfieldTinyMCE.js

@@ -598,7 +598,7 @@ var InputfieldTinyMCE = {
 	
 		var t = InputfieldTinyMCE;
 		var allow = ',' + ProcessWire.config.InputfieldTinyMCE.pasteFilter + ',';
-		var regexTag = /<([a-z0-9]+)([^>]*)>/gi;
+		var regexTag = /<([a-z0-9:!\[\]]+)([^>]*)>/gi;
 		var regexAttr = /([-_a-z0-9]+)=["']([^"']*)["']/gi;
 		var html = args.content;
 		var matchTag, matchAttr;
@@ -607,13 +607,30 @@ var InputfieldTinyMCE = {
 		var replaces = [];
 		var startLength = html.length;
 		
+		allow = allow.toLowerCase();
+		
+		/*
+		 * HTML for testing MS word paste	
+		msWordHtml = 
+			"<p className=MsoNormal>This is <b>bold</b> text. <o:p></o:p></p>\n\n" + 
+			"<h2>This is headline 2. <o:p></o:p></h2>\n\n" + 
+			"<p className=MsoNormal>This is <I>italic</I> text<o:p></o:p></p>\n\n" + 
+			"<p className=MsoListParagraphCxSpFirst style='text-indent:-.25in;mso-list:10 level1 lfo1'>" + 
+			"<![if !supportsLists]><span style='iso-bidi-font-family:Aptos;mso-bidi-theme-font:minor-latin'>" + 
+			"<span style='so-list:Ignore'>1.<span style='font:7.0pt \"Times New Roman\"'>" +
+			"&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</span></span></span><![endif]>One <o:p></o:p></p>";
+		html = msWordHtml;
+		*/
+		
+		startLength = html.length;
+		
 		if(args.internal) {
 			t.log('Skipping pasteFilter for interal copy/paste'); 
 			return; // skip filtering for internal copy/paste operations
 		}
 		
 		if(allow === ',text,') {
-			t.log('Skipping pasteFilter since paste_as_text settingw will be used'); 
+			t.log('Skipping pasteFilter since paste_as_text settings will be used'); 
 			return; // will be processed by paste_as_text setting
 		}
 		
@@ -621,17 +638,19 @@ var InputfieldTinyMCE = {
 			
 			var tagOpen = matchTag[0]; // i.e. <strong>, <img src="..">, <h2>, etc.
 			var tagName = matchTag[1]; // i.e. 'strong', 'img', 'h2', etc.
+			var tagNameLower = tagName.toLowerCase();
 			var tagClose = '</' + tagName + '>'; // i.e. </strong>, </h2>
 			var tagAttrs = matchTag[2]; // i.e. 'src="a.jpg" alt="alt"'
 			var allowAttrs = false;
 	
 			// first see if we can match a tag replacement		
-			var find = ',' + tagName + '='; // i.e. ',b=strong'
+			var findTagEqual = ',' + tagNameLower + '='; // i.e. ',b=strong'
-			var pos = allow.indexOf(find);
+			var findTagAttr = ',' + tagNameLower + '['; // i.e. ',b[...]'
+			var findTagOnly = ',' + tagNameLower + ','; // i.e. ,b
+			var posTagEqual = allow.indexOf(findTagEqual);
 			
-			if(pos > -1) {
+			if(posTagEqual > -1) {
-				// tag replacement
+				var rule = allow.substring(posTagEqual + 1); // i.e. b=strong,and,more
-				var rule = allow.substring(pos + 1); // i.e. b=strong,and,more
 				rule = rule.substring(0, rule.indexOf(',')); // i.e. b=strong
 				rule = rule.split('=');
 				var replaceTag = rule[1];
@@ -641,11 +660,11 @@ var InputfieldTinyMCE = {
 				replaces.push('</' + replaceTag + '>');
 			}
 	
-			if(allow.indexOf(',' + tagName + '[') > -1) {
+			if(allow.indexOf(findTagAttr) > -1) {
 				// tag appears in whitelist with attributes
 				allowAttrs = true;
-			} else if(allow.indexOf(',' + tagName + ',') === -1) {
+			} else if(posTagEqual === -1 && allow.indexOf(findTagOnly) === -1) {
-				// tag does not appear in whitelist
+				// tag does not appear in whitelist at all
 				removals.push(tagOpen);
 				removals.push(tagClose);
 				continue;

wire/modules/Inputfield/InputfieldTinyMCE/InputfieldTinyMCE.module.php

@@ -74,7 +74,7 @@ class InputfieldTinyMCE extends InputfieldTextarea implements ConfigurableModule
 		return array(
 			'title' => 'TinyMCE',
 			'summary' => 'TinyMCE rich text editor version ' . self::mceVersion . '.',
-			'version' => 617,
+			'version' => 618,
 			'icon' => 'keyboard-o',
 			'requires' => 'ProcessWire>=3.0.200, MarkupHTMLPurifier',
 		);
@@ -89,6 +89,7 @@ class InputfieldTinyMCE extends InputfieldTextarea implements ConfigurableModule
 	const toggleCleanDiv = 2; // remove <div>s
 	const toggleCleanP = 4; // remove empty <p> tags	
 	const toggleCleanNbsp = 8; // remove &nbsp; entities
+	const toggleRemoveStyles = 16; // remove all style attributes
 	
 	/**
 	 * Default configuration for filtered paste

wire/modules/Inputfield/InputfieldTinyMCE/InputfieldTinyMCEConfigs.php

@@ -711,6 +711,7 @@ class InputfieldTinyMCEConfigs extends InputfieldTinyMCEClass {
 		$f->addOption(InputfieldTinyMCE::toggleCleanDiv, $this->_('Convert `<div>` tags to `<p>` tags on save?'));
 		$f->addOption(InputfieldTinyMCE::toggleCleanP, $this->_('Remove empty `<p>` tags on save?'));
 		$f->addOption(InputfieldTinyMCE::toggleCleanNbsp, $this->_('Remove non-breaking spaces on save?'));
+		$f->addOption(InputfieldTinyMCE::toggleRemoveStyles, $this->_('Remove `style` attributes from all elements'));
 		$f->attr('value', $this->inputfield->toggles);
 		$f->collapsed = Inputfield::collapsedYes;
 		$f->themeOffset = 1;

wire/modules/Inputfield/InputfieldTinyMCE/InputfieldTinyMCETools.php

@@ -163,22 +163,38 @@ class InputfieldTinyMCETools extends InputfieldTinyMCEClass {
 		$toggles = $this->inputfield->toggles;
 		if(!is_array($toggles)) return $value;
 		
-		if(in_array(InputfieldTinyMCE::toggleCleanDiv, $toggles) && strpos($value, '<div') !== false) {
+		foreach($toggles as $toggle) {
+			switch($toggle) {
+				case InputfieldTinyMCE::toggleCleanDiv: 
+					// convert <div> to <p>
+					if(strpos($value, '<div') !== false) {
 						$value = preg_replace('{\s*(</?)div[^><]*>\s*}is', '$1' . 'p>', $value);
 						while(strpos($value, '<p><p>') !== false) {
 							$value = str_replace(array('<p><p>', '</p></p>'), array('<p>', '</p>'), $value);
 						}
 					}
-
+					break;
-		// remove gratuitous whitespace
+				case InputfieldTinyMCE::toggleCleanP:
-		if(in_array(InputfieldTinyMCE::toggleCleanP, $toggles)) {
+					// remove empty paragraphs
 					$value = str_replace(array('<p><br /></p>', '<p>&nbsp;</p>', "<p>\xc2\xa0</p>", '<p></p>', '<p> </p>'), '', $value);
-		}
+					break;
-
+				case InputfieldTinyMCE::toggleCleanNbsp:
 					// convert non-breaking space to regular space
-		if(in_array(InputfieldTinyMCE::toggleCleanNbsp, $toggles)) {
 					$value = str_ireplace('&nbsp;', ' ', $value);
 					$value = str_replace("\xc2\xa0",' ', $value);
+					break;
+				case InputfieldTinyMCE::toggleRemoveStyles:
+					// remove all style attributes
+					if(strpos($value, 'style=')) {
+						if(preg_match_all('!(<.+?)\sstyle=(["\']).*?\2!i', $value, $matches)) {
+							foreach($matches[0] as $key => $fullMatch) {
+								$startMatch = $matches[1][$key];
+								$value = str_replace($fullMatch, $startMatch, $value);
+							}
+						}
+					}
+					break;
+			}
 		}
 
 		return $value;