diff --git a/typo3/sysext/core/Classes/Html/HtmlParser.php b/typo3/sysext/core/Classes/Html/HtmlParser.php
index 15aab06f85909b1d1c36ece6eb83a2f5958ff89c..8053d1da8a175dc707f10b824aa400accccb5440 100644
--- a/typo3/sysext/core/Classes/Html/HtmlParser.php
+++ b/typo3/sysext/core/Classes/Html/HtmlParser.php
@@ -986,19 +986,24 @@ class HtmlParser
* @param string $tagList The comma separated list of tags to be stripped.
* If empty, all empty tags will be stripped
* @param bool $treatNonBreakingSpaceAsEmpty If TRUE tags containing only entities will be treated as empty.
+ * @param bool $keepTags If true, the provided tags will be kept instead of stripped.
* @return string the stripped content
*/
- public function stripEmptyTags($content, $tagList = null, $treatNonBreakingSpaceAsEmpty = false)
+ public function stripEmptyTags($content, $tagList = '', $treatNonBreakingSpaceAsEmpty = false, $keepTags = false)
{
- $tagRegEx = '[^ >]+'; // all characters until you reach a > or space;
- if ($tagList) {
- $tags = preg_split('/,/', $tagList);
- $tagRegEx = preg_replace('/ */', '', join('|', $tags));
+ if (!empty($tagList)) {
+ $tagRegEx = join('|', GeneralUtility::trimExplode(',', $tagList, true));
+ if ($keepTags) {
+ $tagRegEx = '(?!' . $tagRegEx . ')[^ >]+';
+ }
+ } else {
+ $tagRegEx = '[^ >]+'; // all characters until you reach a > or space;
}
$count = 1;
$nbspRegex = $treatNonBreakingSpaceAsEmpty ? '|( )' : '';
- while ($count != 0) {
- $content = preg_replace(sprintf('/<(%s)[^>]*>( %s)*<\/\\1[^>]*>/i', $tagRegEx, $nbspRegex), '', $content, -1, $count);
+ $finalRegex = sprintf('/<(%s)[^>]*>( %s)*<\/\\1[^>]*>/i', $tagRegEx, $nbspRegex);
+ while ($count !== 0) {
+ $content = preg_replace($finalRegex, '', $content, -1, $count);
}
return $content;
}
@@ -1012,26 +1017,21 @@ class HtmlParser
*/
protected function stripEmptyTagsIfConfigured($value, $configuration)
{
- if (isset($configuration['stripEmptyTags']) && $configuration['stripEmptyTags']) {
- $tags = null;
- if (
- isset($configuration['stripEmptyTags.']['tags'])
- && $configuration['stripEmptyTags.']['tags'] !== ''
- ) {
- $tags = $configuration['stripEmptyTags.']['tags'];
- }
-
- $treatNonBreakingSpaceAsEmpty = false;
- if (
- isset($configuration['stripEmptyTags.']['treatNonBreakingSpaceAsEmpty'])
- && $configuration['stripEmptyTags.']['treatNonBreakingSpaceAsEmpty']
- ) {
- $treatNonBreakingSpaceAsEmpty = (bool)$configuration['stripEmptyTags.']['treatNonBreakingSpaceAsEmpty'];
- }
+ if (empty($configuration['stripEmptyTags'])) {
+ return $value;
+ }
- $value = $this->stripEmptyTags($value, $tags, $treatNonBreakingSpaceAsEmpty);
+ $tags = null;
+ $keepTags = false;
+ if (!empty($configuration['stripEmptyTags.']['keepTags'])) {
+ $tags = $configuration['stripEmptyTags.']['keepTags'];
+ $keepTags = true;
+ } elseif (!empty($configuration['stripEmptyTags.']['tags'])) {
+ $tags = $configuration['stripEmptyTags.']['tags'];
}
- return $value;
+ $treatNonBreakingSpaceAsEmpty = !empty($configuration['stripEmptyTags.']['treatNonBreakingSpaceAsEmpty']);
+
+ return $this->stripEmptyTags($value, $tags, $treatNonBreakingSpaceAsEmpty, $keepTags);
}
}
diff --git a/typo3/sysext/core/Documentation/Changelog/master/Feature-72045-KeepTagsInHtmlParserWhenStrippingEmptyTags.rst b/typo3/sysext/core/Documentation/Changelog/master/Feature-72045-KeepTagsInHtmlParserWhenStrippingEmptyTags.rst
new file mode 100644
index 0000000000000000000000000000000000000000..77571fb69fc0703f9f9978e8a7ea6f0b0a3130b0
--- /dev/null
+++ b/typo3/sysext/core/Documentation/Changelog/master/Feature-72045-KeepTagsInHtmlParserWhenStrippingEmptyTags.rst
@@ -0,0 +1,28 @@
+====================================================
+Feature: #72045 - HTMLparser.stripEmptyTags.keepTags
+====================================================
+
+Description
+===========
+
+A new option for the HTMLparser.stripEmptyTags configuration is added.
+It allows keeping configured tags. Before this change only a list of tags
+could be provided that should be removed.
+
+The following example will strip all empty tags **except** ``tr`` and ``td`` tags.
+
+::
+
+ HTMLparser.stripEmptyTags = 1
+ HTMLparser.stripEmptyTags.keepTags = tr,td
+
+
+**Important!** If this setting is used the stripEmptyTags.tags configuration will
+have no effect any more. You can only use one option at a time.
+
+
+Impact
+======
+
+Unless the configuration of the HTMLparser is changed, the stripEmptyTags
+feature will work as before.
\ No newline at end of file
diff --git a/typo3/sysext/core/Tests/Unit/Html/HtmlParserTest.php b/typo3/sysext/core/Tests/Unit/Html/HtmlParserTest.php
index eaa7eb7f188dd1d3bdd1a9527a6bf04acfbe484d..01711fc020aa0a1fd6558e3fa2e48bdca8f3e3c4 100644
--- a/typo3/sysext/core/Tests/Unit/Html/HtmlParserTest.php
+++ b/typo3/sysext/core/Tests/Unit/Html/HtmlParserTest.php
@@ -294,23 +294,38 @@ class HtmlParserTest extends \TYPO3\CMS\Core\Tests\UnitTestCase
public function emptyTagsDataProvider()
{
return array(
- array(0 , null, false, '<h1></h1>', '<h1></h1>'),
- array(1 , null, false, '<h1></h1>', ''),
- array(1 , null, false, '<h1>hallo</h1>', '<h1>hallo</h1>'),
- array(1 , null, false, '<h1 class="something"></h1>', ''),
- array(1 , null, false, '<h1 class="something"></h1><h2></h2>', ''),
- array(1 , 'h2', false, '<h1 class="something"></h1><h2></h2>', '<h1 class="something"></h1>'),
- array(1 , 'h2, h1', false, '<h1 class="something"></h1><h2></h2>', ''),
- array(1 , null, false, '<div><p></p></div>', ''),
- array(1 , null, false, '<div><p> </p></div>', '<div><p> </p></div>'),
- array(1 , null, true, '<div><p> </p></div>', ''),
- array(1 , null, true, '<div> <p></p></div>', ''),
- array(1 , null, false, '<div>Some content<p></p></div>', '<div>Some content</div>'),
- array(1 , null, true, '<div>Some content<p></p></div>', '<div>Some content</div>'),
- array(1 , null, false, '<div>Some content</div>', '<div>Some content</div>'),
- array(1 , null, true, '<div>Some content</div>', '<div>Some content</div>'),
- array(1 , null, false, '<a href="#skiplinks">Skiplinks </a><b></b>', '<a href="#skiplinks">Skiplinks </a>'),
- array(1 , null, true, '<a href="#skiplinks">Skiplinks </a><b></b>', '<a href="#skiplinks">Skiplinks </a>'),
+ array(0, null, false, '<h1></h1>', '<h1></h1>'),
+ array(1, null, false, '<h1></h1>', ''),
+ array(1, null, false, '<h1>hallo</h1>', '<h1>hallo</h1>'),
+ array(1, null, false, '<h1 class="something"></h1>', ''),
+ array(1, null, false, '<h1 class="something"></h1><h2></h2>', ''),
+ array(1, 'h2', false, '<h1 class="something"></h1><h2></h2>', '<h1 class="something"></h1>'),
+ array(1, 'h2, h1', false, '<h1 class="something"></h1><h2></h2>', ''),
+ array(1, null, false, '<div><p></p></div>', ''),
+ array(1, null, false, '<div><p> </p></div>', '<div><p> </p></div>'),
+ array(1, null, true, '<div><p> </p></div>', ''),
+ array(1, null, true, '<div> <p></p></div>', ''),
+ array(1, null, false, '<div>Some content<p></p></div>', '<div>Some content</div>'),
+ array(1, null, true, '<div>Some content<p></p></div>', '<div>Some content</div>'),
+ array(1, null, false, '<div>Some content</div>', '<div>Some content</div>'),
+ array(1, null, true, '<div>Some content</div>', '<div>Some content</div>'),
+ array(1, null, false, '<a href="#skiplinks">Skiplinks </a><b></b>', '<a href="#skiplinks">Skiplinks </a>'),
+ array(1, null, true, '<a href="#skiplinks">Skiplinks </a><b></b>', '<a href="#skiplinks">Skiplinks </a>'),
+ array(0, '', false, '<h1></h1>', '<h1></h1>'),
+ array(1, '', false, '<h1></h1>', ''),
+ array(1, '', false, '<h1>hallo</h1>', '<h1>hallo</h1>'),
+ array(1, '', false, '<h1 class="something"></h1>', ''),
+ array(1, '', false, '<h1 class="something"></h1><h2></h2>', ''),
+ array(1, '', false, '<div><p></p></div>', ''),
+ array(1, '', false, '<div><p> </p></div>', '<div><p> </p></div>'),
+ array(1, '', true, '<div><p> </p></div>', ''),
+ array(1, '', true, '<div> <p></p></div>', ''),
+ array(1, '', false, '<div>Some content<p></p></div>', '<div>Some content</div>'),
+ array(1, '', true, '<div>Some content<p></p></div>', '<div>Some content</div>'),
+ array(1, '', false, '<div>Some content</div>', '<div>Some content</div>'),
+ array(1, '', true, '<div>Some content</div>', '<div>Some content</div>'),
+ array(1, '', false, '<a href="#skiplinks">Skiplinks </a><b></b>', '<a href="#skiplinks">Skiplinks </a>'),
+ array(1, '', true, '<a href="#skiplinks">Skiplinks </a><b></b>', '<a href="#skiplinks">Skiplinks </a>'),
);
}
@@ -338,6 +353,48 @@ class HtmlParserTest extends \TYPO3\CMS\Core\Tests\UnitTestCase
$this->assertEquals($expectedResult, $result);
}
+ /**
+ * @return array
+ */
+ public function stripEmptyTagsKeepsConfiguredTagsDataProvider() {
+ return [
+ array(
+ 'tr,td',
+ false,
+ '<div><p><tr><td></td></tr></p></div><div class="test"></div><tr></tr><p></p><td></td><i></i>',
+ '<div><p><tr><td></td></tr></p></div><tr></tr><td></td>'
+ ),
+ array(
+ 'tr,td',
+ true,
+ '<div><p><tr><td></td></tr></p></div><p class="test"> </p><tr></tr><p></p><td></td><i></i>',
+ '<div><p><tr><td></td></tr></p></div><tr></tr><td></td>'
+ ),
+ ];
+ }
+
+ /**
+ * @test
+ * @dataProvider stripEmptyTagsKeepsConfiguredTagsDataProvider
+ * @param string $tagList List of tags that should be kept, event if they are empty.
+ * @param bool $treatNonBreakingSpaceAsEmpty If true will be considered empty.
+ * @param string $content The HTML content that should be parsed.
+ * @param string $expectedResult The expected HTML code result.
+ */
+ public function stripEmptyTagsKeepsConfiguredTags($tagList, $treatNonBreakingSpaceAsEmpty, $content, $expectedResult) {
+ $tsConfig = array(
+ 'keepNonMatchedTags' => 1,
+ 'stripEmptyTags' => 1,
+ 'stripEmptyTags.' => array(
+ 'keepTags' => $tagList,
+ 'treatNonBreakingSpaceAsEmpty' => $treatNonBreakingSpaceAsEmpty
+ ),
+ );
+
+ $result = $this->parseConfigAndCleanHtml($tsConfig, $content);
+ $this->assertEquals($expectedResult, $result);
+ }
+
/**
* Calls HTMLparserConfig() and passes the generated config to the HTMLcleaner() method on the current subject.
*