add case when HTML tags contain attributes

htdocs/includes/odtphp/odf.php

@@ -46,8 +46,8 @@ class Odf
 	public $userdefined=array();
 
 	const PIXEL_TO_CM = 0.026458333;
-	const FIND_TAGS_REGEX = '/<([A-Za-z0-9]+)(?:\s([A-Za-z]+(?:\-[A-Za-z]+)?(?:=(?:".*?")|(?:[0-9]+))))*(?:(?:\s\/>)|(?:>(((?!<\1>).)*)<\/\1>))/s';
-	const FIND_ENCODED_TAGS_REGEX = '/&lt;([A-Za-z]+)(?:\s([A-Za-z]+(?:\-[A-Za-z]+)?(?:=(?:".*?")|(?:[0-9]+))))*(?:(?:\s\/&gt;)|(?:&gt;(((?!&lt;\1&gt;).)*)&lt;\/\1&gt;))/';
+	const FIND_TAGS_REGEX = '/<([A-Za-z0-9]+)(?:\s([A-Za-z]+(?:\-[A-Za-z]+)?(?:=(?:".*?")|(?:[0-9]+))))*(?:(?:\s\/>)|(?:>(((?!<\1(\s.*)?>).)*)<\/\1>))/s';
+	const FIND_ENCODED_TAGS_REGEX = '/&lt;([A-Za-z]+)(?:\s([A-Za-z]+(?:\-[A-Za-z]+)?(?:=(?:".*?")|(?:[0-9]+))))*(?:(?:\s\/&gt;)|(?:&gt;(((?!&lt;\1(\s.*)?&gt;).)*)&lt;\/\1&gt;))/';
 
 
 	/**

test/phpunit/ODFTest.php

@@ -302,33 +302,39 @@ public function testODFconvertVarToOdf()
 				'charset' => null,
 				'expected' => utf8_encode('text with <text:span text:style-name="boldText">two</text:span> (strong) <text:span text:style-name="boldText">tags</text:span>'),
 			],
+			25 => [
+				'to_convert' => "text with <strong class=\"whatever\">two</strong> (strong) <strong class=\"the weather\">tags and <u>intricated</u> underline </strong>",
+				'encode' => true,
+				'charset' => null,
+				'expected' => utf8_encode('text with <text:span text:style-name="boldText">two</text:span> (strong) <text:span text:style-name="boldText">tags and <text:span text:style-name="underlineText">intricated</text:span> underline </text:span>'),
+			],
 
 			// One can also pass html-encoded string to the method
-			25 => [
+			26 => [
 				'to_convert' => 'One&amp;two',
 				'encode' => true,
 				'charset' => null,
 				'expected' => 'One&amp;two',
 			],
-			26 => [
+			27 => [
 				'to_convert' => "text with &lt;strong&gt;strong, &lt;/strong&gt;&lt;em&gt;emphasis&lt;/em&gt; and &lt;u&gt;underlined&lt;/u&gt; words with &lt;i&gt;it@lic sp&amp;ciàlchärs éè l'&lt;/i&gt;",
 				'encode' => false,
 				'charset' => 'UTF-8',
 				'expected' => 'text with <text:span text:style-name="boldText">strong, </text:span><text:span text:style-name="italicText">emphasis</text:span> and <text:span text:style-name="underlineText">underlined</text:span> words with <text:span text:style-name="italicText">it@lic sp&ciàlchärs éè l\'</text:span>',
 			],
-			27 => [
+			28 => [
 				'to_convert' => "text with &lt;strong&gt;strong, &lt;/strong&gt;&lt;em&gt;emphasis&lt;/em&gt; and &lt;u&gt;underlined&lt;/u&gt; words with &lt;i&gt;it@lic sp&amp;ciàlchärs éè l'&lt;/i&gt;",
 				'encode' => true,
 				'charset' => 'UTF-8',
 				'expected' => 'text with <text:span text:style-name="boldText">strong, </text:span><text:span text:style-name="italicText">emphasis</text:span> and <text:span text:style-name="underlineText">underlined</text:span> words with <text:span text:style-name="italicText">it@lic sp&amp;ciàlchärs éè l&apos;</text:span>',
 			],
-			28 => [
+			29 => [
 				'to_convert' => "text with &lt;strong&gt;strong, &lt;/strong&gt;&lt;em&gt;emphasis&lt;/em&gt; and &lt;u&gt;underlined&lt;/u&gt; words with &lt;i&gt;it@lic sp&amp;ciàlchärs éè l'&lt;/i&gt;",
 				'encode' => false,
 				'charset' => null,
 				'expected' => utf8_encode('text with <text:span text:style-name="boldText">strong, </text:span><text:span text:style-name="italicText">emphasis</text:span> and <text:span text:style-name="underlineText">underlined</text:span> words with <text:span text:style-name="italicText">it@lic sp&ciàlchärs éè l\'</text:span>'),
 			],
-			29 => [
+			30 => [
 				'to_convert' => "text with &lt;strong&gt;strong, &lt;/strong&gt;&lt;em&gt;emphasis&lt;/em&gt; and &lt;u&gt;underlined&lt;/u&gt; words with &lt;i&gt;it@lic sp&amp;ciàlchärs éè l'&lt;/i&gt;",
 				'encode' => true,
 				'charset' => null,
@@ -347,20 +353,20 @@ public function testODFconvertVarToOdf()
 			// Following tests reflect the current behavior. They may evolve if the method behavior changes.
 
 			// The method removes hyperlinks and tags that are not dealt with.
-			30 => [
+			31 => [
 				'to_convert' => '123 <a href="/test.php">trucmachin > truc < troc > trac</a>bla bla',
 				'encode' => true,
 				'charset' => null,
 				'expected' => "123 trucmachin &gt; truc &lt; troc &gt; tracbla bla",
 			],
-			31 => [
+			32 => [
 				'to_convert' => '123 <h3>Title</h3> bla',
 				'encode' => true,
 				'charset' => null,
 				'expected' => "123 Title bla",
 			],
 			// HTML should not take \n into account, but only <br />.
-			32 => [
+			33 => [
 				'to_convert' => "text with <strong>strong text </strong>, a line\nbreak and <u>underlined</u> words with <i>it@lic sp&ciàlchärs éè l'</i>",
 				'encode' => false,
 				'charset' => 'UTF-8',