/**
* Capture all attributes in given string
*
* @param Tag $tag Target tag
* @param string $elName Name of the HTML element
* @param string $str String containing the attribute declarations
* @return void
*/
protected function captureAttributes(Tag $tag, $elName, $str)
{
preg_match_all('/[a-z][-a-z0-9]*(?>\\s*=\\s*(?>"[^"]*"|\'[^\']*\'|[^\\s"\'=<>`]+))?/i', $str, $attrMatches);
foreach ($attrMatches[0] as $attrMatch) {
$pos = strpos($attrMatch, '=');
/**
* If there's no equal sign, it's a boolean attribute and we generate a value equal
* to the attribute's name, lowercased
*
* @link http://www.w3.org/html/wg/drafts/html/master/single-page.html#boolean-attributes
*/
if ($pos === false) {
$pos = strlen($attrMatch);
$attrMatch .= '=' . strtolower($attrMatch);
}
// Normalize the attribute name, remove the whitespace around its value to account
// for cases like <b title = "foo"/>
$attrName = strtolower(trim(substr($attrMatch, 0, $pos)));
$attrValue = trim(substr($attrMatch, 1 + $pos));
// Use the attribute's alias if applicable
if (isset($this->config['aliases'][$elName][$attrName])) {
$attrName = $this->config['aliases'][$elName][$attrName];
}
// Remove quotes around the value
if ($attrValue[0] === '"' || $attrValue[0] === "'") {
$attrValue = substr($attrValue, 1, -1);
}
$tag->setAttribute($attrName, html_entity_decode($attrValue, ENT_QUOTES, 'UTF-8'));
}
}
/**
* @testdox The [MEDIA] tag transfers its priority to the tag it creates
*/
public function testTagPriority()
{
$newTag = $this->getMockBuilder('s9e\\TextFormatter\\Parser\\Tag')->disableOriginalConstructor()->setMethods(['setAttributes', 'setSortPriority'])->getMock();
$newTag->expects($this->once())->method('setSortPriority')->with(123);
$tagStack = $this->getMockBuilder('s9e\\TextFormatter\\Parser')->disableOriginalConstructor()->setMethods(['addSelfClosingTag'])->getMock();
$tagStack->expects($this->once())->method('addSelfClosingTag')->will($this->returnValue($newTag));
$tag = new Tag(Tag::START_TAG, 'MEDIA', 0, 0);
$tag->setAttribute('media', 'foo');
$tag->setSortPriority(123);
Parser::filterTag($tag, $tagStack, ['foo.invalid' => 'foo']);
}
protected function captureAttributes(Tag $tag, $elName, $str)
{
\preg_match_all('/[a-z][-a-z0-9]*(?>\\s*=\\s*(?>"[^"]*"|\'[^\']*\'|[^\\s"\'=<>`]+))?/i', $str, $attrMatches);
foreach ($attrMatches[0] as $attrMatch) {
$pos = \strpos($attrMatch, '=');
if ($pos === \false) {
$pos = \strlen($attrMatch);
$attrMatch .= '=' . \strtolower($attrMatch);
}
$attrName = \strtolower(\trim(\substr($attrMatch, 0, $pos)));
$attrValue = \trim(\substr($attrMatch, 1 + $pos));
if (isset($this->config['aliases'][$elName][$attrName])) {
$attrName = $this->config['aliases'][$elName][$attrName];
}
if ($attrValue[0] === '"' || $attrValue[0] === "'") {
$attrValue = \substr($attrValue, 1, -1);
}
$tag->setAttribute($attrName, \html_entity_decode($attrValue, \ENT_QUOTES, 'UTF-8'));
}
}
protected function setLinkAttributes(Tag $tag, $linkInfo, $attrName)
{
$url = $linkInfo;
$title = '';
$pos = \strpos($linkInfo, ' ');
if ($pos !== \false) {
$url = \substr($linkInfo, 0, $pos);
$title = \substr(\trim(\substr($linkInfo, $pos)), 1, -1);
}
$tag->setAttribute($attrName, $this->decode($url));
if ($title > '') {
$tag->setAttribute('title', $this->decode($title));
}
}
/**
* Truncate the replacement text set in a LINK_TEXT tag
*
* @param \s9e\TextFormatter\Parser\Tag $tag LINK_TEXT tag
* @return bool Always true to indicate that the tag is valid
*/
public function truncate_text(\s9e\TextFormatter\Parser\Tag $tag)
{
$text = $tag->getAttribute('text');
if (utf8_strlen($text) > 55) {
$text = utf8_substr($text, 0, 39) . ' ... ' . utf8_substr($text, -10);
}
$tag->setAttribute('text', $text);
return true;
}
protected static function tagIsMissingAnyAttribute(Tag $tag, array $attrNames)
{
foreach ($attrNames as $attrName) {
if (!$tag->hasAttribute($attrName)) {
return \true;
}
}
return \false;
}
/**
* @testdox filterAttributes() calls the logger's setAttribute() and unsetAttribute() methods for each attribute with a filterChain
*/
public function testFilterAttributesCallsLoggerSetAttribute()
{
$logger = $this->getMock('s9e\\TextFormatter\\Parser\\Logger', ['setAttribute', 'unsetAttribute']);
$logger->expects($this->at(0))->method('setAttribute')->with('foo');
$logger->expects($this->at(2))->method('setAttribute')->with('bar');
$logger->expects($this->exactly(2))->method('unsetAttribute');
$tagConfig = new TagConfig();
$tagConfig->attributes->add('foo')->filterChain->append(function () {
});
$tagConfig->attributes->add('bar')->filterChain->append(function () {
});
$tag = new Tag(Tag::SELF_CLOSING_TAG, 'X', 0, 0);
$tag->setAttribute('foo', 'foo');
$tag->setAttribute('bar', 'bar');
Parser::filterAttributes($tag, $tagConfig->asConfig(), [], $logger);
}
/**
* Set a URL or IMG tag's attributes
*
* @param Tag $tag URL or IMG tag
* @param array $m Regexp captures
* @param string[] $attrNames List of attribute names
* @return void
*/
protected function setLinkAttributes(Tag $tag, array $m, array $attrNames)
{
if (isset($m[3])) {
$attrValues = $this->getInlineLinkAttributes($m);
} else {
$label = isset($m[2]) ? $m[2][0] : $m[1][0];
$attrValues = $this->getReferenceLinkAttributes($label);
}
foreach ($attrValues as $k => $attrValue) {
$tag->setAttribute($attrNames[$k], $attrValue);
}
}
/**
* @testdox gc() removes a paired tag's reference
*/
public function testGc()
{
$startTag = new Tag(Tag::START_TAG, 'X', 0, 0);
$endTag = new Tag(Tag::END_TAG, 'X', 1, 0);
$startTag->pairWith($endTag);
$this->assertSame($endTag, $startTag->getEndTag());
$this->assertSame($startTag, $endTag->getStartTag());
$startTag->gc();
$this->assertNull($startTag->getEndTag());
$endTag->gc();
$this->assertNull($endTag->getStartTag());
}
protected static function compareTags(Tag $a, Tag $b)
{
$aPos = $a->getPos();
$bPos = $b->getPos();
if ($aPos !== $bPos) {
return $bPos - $aPos;
}
if ($a->getSortPriority() !== $b->getSortPriority()) {
return $b->getSortPriority() - $a->getSortPriority();
}
$aLen = $a->getLen();
$bLen = $b->getLen();
if (!$aLen || !$bLen) {
if (!$aLen && !$bLen) {
$order = array(Tag::END_TAG => 0, Tag::SELF_CLOSING_TAG => 1, Tag::START_TAG => 2);
return $order[$b->getType()] - $order[$a->getType()];
}
return $aLen ? -1 : 1;
}
return $aLen - $bLen;
}
public function getData()
{
return [['foo bar', '<r><X><Y>foo</Y> <Z>bar</Z></X></r>', function ($configurator) {
$configurator->tags->add('X');
$configurator->tags->add('Y');
$configurator->tags->add('Z');
}, function ($parser) {
$parser->addStartTag('X', 0, 0);
$parser->addSelfClosingTag('Y', 0, 3);
$parser->addSelfClosingTag('Z', 4, 3);
$parser->addEndTag('X', 7, 0);
}], ['foo bar', '<r>foo <X>bar</X></r>', function ($configurator) {
$configurator->tags->add('X');
}, function ($parser) {
$parser->addStartTag('X', 4, 0);
}], ['foo bar', '<t>foo bar</t>', null, function ($parser) {
$parser->addStartTag('X', 4, 0);
}], ['foo bar', '<r><X>fo<Y>o</Y></X> bar</r>', function ($configurator) {
$configurator->tags->add('X');
$configurator->tags->add('Y');
}, function ($parser) {
$parser->addStartTag('X', 0, 0);
$parser->addEndTag('X', 3, 0);
$parser->addStartTag('Y', 2, 0);
$parser->addEndTag('Y', 4, 1);
}], ['foo bar', '<r><X>fo<Y>o</Y></X><Y> b</Y>ar</r>', function ($configurator) {
$configurator->tags->add('X');
$configurator->tags->add('Y')->rules->autoReopen();
}, function ($parser) {
$parser->addStartTag('X', 0, 0);
$parser->addEndTag('X', 3, 0);
$parser->addStartTag('Y', 2, 0);
$parser->addEndTag('Y', 5, 0);
}], ['foo bar', '<r><X>fo<Y attr="foo">o</Y></X><Y attr="foo"> b</Y>ar</r>', function ($configurator) {
$configurator->tags->add('X');
$tag = $configurator->tags->add('Y');
$tag->attributes->add('attr')->required = false;
$tag->rules->autoReopen();
}, function ($parser) {
$parser->addStartTag('X', 0, 0);
$parser->addEndTag('X', 3, 0);
$parser->addStartTag('Y', 2, 0)->setAttribute('attr', 'foo');
$parser->addEndTag('Y', 5, 0);
}], ['x [b][i]...[/b][/i] y', '<r>x <B><s>[b]</s><I><s>[i]</s>...</I><e>[/b]</e></B><i>[/i]</i> y</r>', function ($configurator) {
$configurator->tags->add('B');
$configurator->tags->add('I')->rules->autoReopen();
}, function ($parser) {
$parser->addStartTag('B', 2, 3);
$parser->addStartTag('I', 5, 3);
$parser->addEndTag('B', 11, 4);
$parser->addEndTag('I', 15, 4);
}], ['x [b][i]...[/b]![/i] y', '<r>x <B><s>[b]</s><I><s>[i]</s>...</I><e>[/b]</e></B><I>!<e>[/i]</e></I> y</r>', function ($configurator) {
$configurator->tags->add('B');
$configurator->tags->add('I')->rules->autoReopen();
}, function ($parser) {
$parser->addStartTag('B', 2, 3);
$parser->addStartTag('I', 5, 3);
$parser->addEndTag('B', 11, 4);
$parser->addEndTag('I', 16, 4);
}], ['x [b][i][u]...[/b][/u][/i] y', '<r>x <B><s>[b]</s><I><s>[i]</s><U><s>[u]</s>...</U></I><e>[/b]</e></B><i>[/u][/i]</i> y</r>', function ($configurator) {
$configurator->tags->add('B');
$configurator->tags->add('I')->rules->autoReopen();
$configurator->tags->add('U')->rules->autoReopen();
}, function ($parser) {
$parser->addStartTag('B', 2, 3);
$parser->addStartTag('I', 5, 3);
$parser->addStartTag('U', 8, 3);
$parser->addEndTag('B', 14, 4);
$parser->addEndTag('U', 18, 4);
$parser->addEndTag('I', 22, 4);
}], ['x [b][i][u]...[/b][/u][/i] y', '<r>x <B><s>[b]</s><I><s>[i]</s><U><s>[u]</s>...</U></I><e>[/b]</e></B><i>[/u][/i]</i> y</r>', function ($configurator) {
$configurator->tags->add('B');
$configurator->tags->add('I');
$configurator->tags->add('U');
}, function ($parser) {
$parser->addStartTag('B', 2, 3);
$parser->addStartTag('I', 5, 3);
$parser->addStartTag('U', 8, 3);
$parser->addEndTag('B', 14, 4);
$parser->addEndTag('U', 18, 4);
$parser->addEndTag('I', 22, 4);
}], ['x [b][i][u]...[/b][/i][/u] y', '<r>x <B><s>[b]</s><I><s>[i]</s><U><s>[u]</s>...</U></I><e>[/b]</e></B><i>[/i][/u]</i> y</r>', function ($configurator) {
$configurator->tags->add('B');
$configurator->tags->add('I')->rules->autoReopen();
$configurator->tags->add('U')->rules->autoReopen();
}, function ($parser) {
$parser->addStartTag('B', 2, 3);
$parser->addStartTag('I', 5, 3);
$parser->addStartTag('U', 8, 3);
$parser->addEndTag('B', 14, 4);
$parser->addEndTag('I', 18, 4);
$parser->addEndTag('U', 22, 4);
}], ['x [b][i][u]...[/b][/i][/u] y', '<r>x <B><s>[b]</s><I><s>[i]</s><U><s>[u]</s>...</U></I><e>[/b]</e></B>[/i][/u] y</r>', function ($configurator) {
$configurator->tags->add('B');
$configurator->tags->add('I')->rules->autoReopen();
$configurator->tags->add('U')->rules->autoReopen();
}, function ($parser) {
// Set maxFixingCost to 2 so that it allows [u] and [i] to be closed, without
// spending any efforts on reopening them
$parser->maxFixingCost = 2;
$parser->addStartTag('B', 2, 3);
$parser->addStartTag('I', 5, 3);
$parser->addStartTag('U', 8, 3);
$parser->addEndTag('B', 14, 4);
$parser->addEndTag('I', 18, 4);
$parser->addEndTag('U', 22, 4);
}], ['x [b][i][u]...[/b][/i]u[/u] y', '<r>x <B><s>[b]</s><I><s>[i]</s><U><s>[u]</s>...</U></I><e>[/b]</e></B><i>[/i]</i><U>u<e>[/u]</e></U> y</r>', function ($configurator) {
$configurator->tags->add('B');
$configurator->tags->add('I')->rules->autoReopen();
$configurator->tags->add('U')->rules->autoReopen();
}, function ($parser) {
$parser->addStartTag('B', 2, 3);
$parser->addStartTag('I', 5, 3);
$parser->addStartTag('U', 8, 3);
$parser->addEndTag('B', 14, 4);
$parser->addEndTag('I', 18, 4);
$parser->addEndTag('U', 23, 4);
}], ['x [i][b][u]...[/b][/i][/u] y', '<r>x <I><s>[i]</s><B><s>[b]</s><U><s>[u]</s>...</U><e>[/b]</e></B><e>[/i]</e></I><i>[/u]</i> y</r>', function ($configurator) {
$configurator->tags->add('B')->rules->autoReopen();
$configurator->tags->add('I')->rules->autoReopen();
$configurator->tags->add('U')->rules->autoReopen();
}, function ($parser) {
$parser->addStartTag('I', 2, 3);
$parser->addStartTag('B', 5, 3);
$parser->addStartTag('U', 8, 3);
$parser->addEndTag('B', 14, 4);
$parser->addEndTag('I', 18, 4);
$parser->addEndTag('U', 22, 4);
}], ['foo bar', '<r>foo <X>bar</X></r>', function ($configurator) {
$configurator->tags->add('X');
$configurator->tags->add('Y');
}, function ($parser) {
$parser->addStartTag('X', 4, 0);
$parser->addEndTag('Y', 5, 0);
}], ['foo bar', '<r><X>foo</X> bar</r>', function ($configurator) {
$configurator->tags->add('X');
}, function ($parser) {
$parser->addSelfClosingTag('X', 0, 3);
$parser->addSelfClosingTag('X', 0, 3);
$parser->addSelfClosingTag('X', 1, 1);
}], ['fooo bar', '<r><X>f<X>oo</X>o</X> bar</r>', function ($configurator) {
$configurator->tags->add('X')->nestingLimit = 2;
}, function ($parser) {
$parser->addStartTag('X', 0, 0);
$parser->addStartTag('X', 1, 0);
$parser->addSelfClosingTag('X', 2, 1);
$parser->addEndTag('X', 3, 0);
$parser->addEndTag('X', 4, 0);
}, [['err', 'Nesting limit exceeded', ['tag' => $this->runClosure(function () {
$tag = new Tag(Tag::SELF_CLOSING_TAG, 'X', 2, 1);
$tag->invalidate();
return $tag;
}), 'tagName' => 'X', 'nestingLimit' => 2]]]], ['foo bar', '<r><X>f</X><X>o</X>o bar</r>', function ($configurator) {
$configurator->tags->add('X')->tagLimit = 2;
}, function ($parser) {
$parser->addSelfClosingTag('X', 0, 1);
$parser->addSelfClosingTag('X', 1, 1);
$parser->addSelfClosingTag('X', 2, 1);
}, [['err', 'Tag limit exceeded', ['tag' => $this->runClosure(function () {
$tag = new Tag(Tag::SELF_CLOSING_TAG, 'X', 2, 1);
$tag->invalidate();
return $tag;
}), 'tagName' => 'X', 'tagLimit' => 2]]]], ['foo bar', '<r><X>foo</X> bar</r>', function ($configurator) {
$configurator->tags->add('X');
$configurator->tags->add('Y');
}, function ($parser) {
$parser->addSelfClosingTag('X', 0, 3);
$parser->addSelfClosingTag('Y', 1, 1)->cascadeInvalidationTo($parser->addSelfClosingTag('Y', 5, 1));
}], ["[pre]foo[b]x\ny[/b]bar[/pre]a\nb", "<r><PRE><s>[pre]</s>foo<B><s>[b]</s>x<br/>\ny<e>[/b]</e></B>bar<e>[/pre]</e></PRE>a<br/>\nb</r>", function ($configurator) {
$configurator->rootRules->enableAutoLineBreaks();
$configurator->tags->add('PRE')->rules->suspendAutoLineBreaks();
$configurator->tags->add('B');
}, function ($parser) {
$parser->addStartTag('PRE', 0, 5);
$parser->addEndTag('PRE', 21, 6);
$parser->addStartTag('B', 8, 3);
$parser->addEndTag('B', 14, 4);
}], ["[pre]foo[b]x\ny[/b]bar[/pre]a\nb", "<r><PRE><s>[pre]</s>foo<B><s>[b]</s>x\ny<e>[/b]</e></B>bar<e>[/pre]</e></PRE>a<br/>\nb</r>", function ($configurator) {
$configurator->rootRules->enableAutoLineBreaks();
$configurator->tags->add('PRE')->rules->disableAutoLineBreaks();
$configurator->tags->add('B');
}, function ($parser) {
$parser->addStartTag('PRE', 0, 5);
$parser->addEndTag('PRE', 21, 6);
$parser->addStartTag('B', 8, 3);
$parser->addEndTag('B', 14, 4);
}], ['foo bar', '<r><X>foo</X> bar</r>', function ($configurator) {
$configurator->tags->add('X');
}, function ($parser) {
$parser->addSelfClosingTag('X', 0, 3);
$parser->addIgnoreTag(2, 1);
}], ['foo bar', '<r><X>foo</X><i> b</i>ar</r>', function ($configurator) {
$configurator->tags->add('X');
}, function ($parser) {
$parser->addSelfClosingTag('X', 0, 3);
$parser->addIgnoreTag(2, 3);
}], ['foo bar', '<r>foo <X>bar</X></r>', function ($configurator) {
$configurator->tags->add('X');
}, function ($parser) {
$parser->addSelfClosingTag('X', 4, 3);
}], ['foo bar', '<t>foo bar</t>', function ($configurator) {
$configurator->tags->add('X');
}, function ($parser) {
$parser->addSelfClosingTag('X', 4, 4);
}], ['foo bar', '<r><X>foo bar</X></r>', function ($configurator) {
$configurator->tags->add('X');
}, function ($parser) {
$parser->addSelfClosingTag('X', 0, 7);
}], ['foo bar', '<t>foo bar</t>', function ($configurator) {
$configurator->tags->add('X');
}, function ($parser) {
$parser->addSelfClosingTag('X', 0, 8);
}], ['*foo* bar', '<r><X><s>*</s>foo<e>*</e></X> bar</r>', function ($configurator) {
$configurator->tags->add('X');
}, function ($parser) {
$parser->addStartTag('X', 0, 1)->pairWith($parser->addEndTag('X', 4, 1));
}], ['*foo* bar', '<r><X><s>*</s>foo* bar</X></r>', function ($configurator) {
$configurator->tags->add('X');
}, function ($parser) {
$parser->addStartTag('X', 0, 1)->pairWith($parser->addEndTag('X', 99, 1));
}], ['*foo* bar', '<r><X><s>*</s>foo* bar</X></r>', function ($configurator) {
$configurator->tags->add('X');
}, function ($parser) {
$parser->addStartTag('X', 0, 1)->pairWith($parser->addEndTag('X', 99, 1));
$parser->addEndTag('X', 4, 1);
}], ['*_foo* bar_', '<r><X><s>*</s><Y><s>_</s>foo</Y><e>*</e></X><Y> bar<e>_</e></Y></r>', function ($configurator) {
$configurator->tags->add('X');
$configurator->tags->add('Y')->rules->autoReopen();
}, function ($parser) {
$parser->addStartTag('X', 0, 1)->pairWith($parser->addEndTag('X', 5, 1));
$parser->addStartTag('Y', 1, 1)->pairWith($parser->addEndTag('Y', 10, 1));
$parser->addEndTag('Y', 6, 1);
}], ['**x**x***', '<r><X><s>**</s>x<Y><s>**</s>x<e>**</e></Y><e>*</e></X></r>', function ($configurator) {
$configurator->tags->add('X');
$configurator->tags->add('Y');
}, function ($parser) {
$parser->addStartTag('X', 0, 2)->pairWith($parser->addEndTag('X', 7, 2));
$parser->addStartTag('Y', 3, 2)->pairWith($parser->addEndTag('Y', 6, 2));
}], ['**x[**]x', '<r><X><s>**</s>x<Y>[**]</Y></X>x</r>', function ($configurator) {
$configurator->tags->add('X');
$configurator->tags->add('Y');
}, function ($parser) {
$parser->addStartTag('X', 0, 2)->pairWith($parser->addEndTag('X', 4, 2));
$parser->addSelfClosingTag('Y', 3, 4);
}], ['xy', '<t>x<br/>y</t>', null, function ($parser) {
$parser->addBrTag(1);
}], ['xy', '<t>xy</t>', function ($configurator) {
$configurator->rootRules->preventLineBreaks();
}, function ($parser) {
$parser->addBrTag(1);
}], ['xx', '<r><X>x</X><X>x</X></r>', function ($configurator) {
$configurator->tags->add('X')->rules->closeParent('X');
}, function ($parser) {
$parser->addStartTag('X', 0, 0);
$parser->addStartTag('X', 1, 0);
}], ['xx [hr] yy', '<r>xx <HR>[hr]</HR> yy</r>', function ($configurator) {
$configurator->tags->add('HR')->rules->autoClose();
}, function ($parser) {
$parser->addStartTag('HR', 3, 4);
}], ['xx [hr][/hr] yy', '<r>xx <HR><s>[hr]</s><e>[/hr]</e></HR> yy</r>', function ($configurator) {
$configurator->tags->add('HR')->rules->autoClose();
}, function ($parser) {
$parser->addStartTag('HR', 3, 4);
$parser->addEndTag('HR', 7, 5);
}], ['xx [img=foo.png] yy', '<r>xx <IMG src="foo.png">[img=foo.png]</IMG> yy</r>', function ($configurator) {
$tag = $configurator->tags->add('IMG');
$tag->attributes->add('src');
$tag->rules->autoClose();
}, function ($parser) {
$parser->addStartTag('IMG', 3, 13)->setAttribute('src', 'foo.png');
}], ['xx [img]foo.png[/img] yy', '<r>xx <IMG src="foo.png"><s>[img]</s>foo.png<e>[/img]</e></IMG> yy</r>', function ($configurator) {
$tag = $configurator->tags->add('IMG');
$tag->attributes->add('src');
$tag->rules->autoClose();
}, function ($parser) {
$tag = $parser->addStartTag('IMG', 3, 5);
$tag->setAttribute('src', 'foo.png');
$tag->pairWith($parser->addEndTag('IMG', 15, 6));
}], ['XYX', '<r><X><s>X</s><Y>Y</Y><e>X</e></X></r>', function ($configurator) {
$configurator->tags->add('X');
$configurator->tags->add('Y');
}, function ($parser) {
$parser->addStartTag('X', 0, 1);
$parser->addSelfClosingTag('Y', 1, 1);
$parser->addEndTag('X', 2, 1);
}], ['XYX', '<r><X><s>X</s>Y<e>X</e></X></r>', function ($configurator) {
$configurator->tags->add('X')->rules->denyChild('Y');
$configurator->tags->add('Y');
}, function ($parser) {
$parser->addStartTag('X', 0, 1);
$parser->addSelfClosingTag('Y', 1, 1);
$parser->addEndTag('X', 2, 1);
}], ['XYZYX', '<r><X><s>X</s><Y><s>Y</s><Z>Z</Z><e>Y</e></Y><e>X</e></X></r>', function ($configurator) {
$configurator->tags->add('X')->rules->denyChild('Z');
$configurator->tags->add('Y');
$configurator->tags->add('Z');
}, function ($parser) {
$parser->addStartTag('X', 0, 1);
$parser->addStartTag('Y', 1, 1);
$parser->addSelfClosingTag('Z', 2, 1);
$parser->addEndTag('Y', 3, 1);
$parser->addEndTag('X', 4, 1);
}], ['XYZYX', '<r><X><s>X</s><Y><s>Y</s>Z<e>Y</e></Y><e>X</e></X></r>', function ($configurator) {
$configurator->tags->add('X')->rules->denyChild('Z');
$configurator->tags->add('Y')->rules->isTransparent();
$configurator->tags->add('Z');
}, function ($parser) {
$parser->addStartTag('X', 0, 1);
$parser->addStartTag('Y', 1, 1);
$parser->addSelfClosingTag('Z', 2, 1);
$parser->addEndTag('Y', 3, 1);
$parser->addEndTag('X', 4, 1);
}], ['XYX', '<r><X><s>X</s>Y<e>X</e></X></r>', function ($configurator) {
$configurator->tags->add('X')->rules->denyDescendant('Y');
$configurator->tags->add('Y');
}, function ($parser) {
$parser->addStartTag('X', 0, 1);
$parser->addSelfClosingTag('Y', 1, 1);
$parser->addEndTag('X', 2, 1);
}], ['XYZYX', '<r><X><s>X</s><Y><s>Y</s>Z<e>Y</e></Y><e>X</e></X></r>', function ($configurator) {
$configurator->tags->add('X')->rules->denyDescendant('Z');
$configurator->tags->add('Y');
$configurator->tags->add('Z');
}, function ($parser) {
$parser->addStartTag('X', 0, 1);
$parser->addStartTag('Y', 1, 1);
$parser->addSelfClosingTag('Z', 2, 1);
$parser->addEndTag('Y', 3, 1);
$parser->addEndTag('X', 4, 1);
}], ['XYX', '<r><X><s>X</s>Y<e>X</e></X></r>', function ($configurator) {
$rules = $configurator->tags->add('X')->rules;
$rules->isTransparent();
$rules->denyChild('Y');
$configurator->tags->add('Y');
}, function ($parser) {
$parser->addStartTag('X', 0, 1);
$parser->addSelfClosingTag('Y', 1, 1);
$parser->addEndTag('X', 2, 1);
}], ['XYYYX', '<r><X><s>X</s><Y><s>Y</s><Y><s>Y</s><Y><s>Y</s></Y></Y></Y><e>X</e></X></r>', function ($configurator) {
$configurator->tags->add('X');
$configurator->tags->add('Y');
}, function ($parser) {
$parser->maxFixingCost = 0;
$parser->addStartTag('X', 0, 1);
$parser->addStartTag('Y', 1, 1);
$parser->addStartTag('Y', 2, 1);
$parser->addStartTag('Y', 3, 1);
$parser->addEndTag('X', 4, 1);
}, [['warn', 'Fixing cost limit exceeded']]], ['XYYYYX', '<r><X><s>X</s><Y><s>Y</s><Y>Y</Y><Y>Y</Y><e>Y</e></Y><e>X</e></X></r>', function ($configurator) {
$configurator->tags->add('X');
$configurator->tags->add('Y');
}, function ($parser) {
// Ensuring that we can still parse well-formed text if we disallow any fixing
$parser->maxFixingCost = 0;
$parser->addStartTag('X', 0, 1);
$parser->addStartTag('Y', 1, 1);
$parser->addSelfClosingTag('Y', 2, 1);
$parser->addSelfClosingTag('Y', 3, 1);
$parser->addEndTag('Y', 4, 1);
$parser->addEndTag('X', 5, 1);
}], ['..', '<r><X>.</X><X>.</X></r>', function ($configurator) {
$configurator->tags->add('X');
}, function ($parser) {
// NOTE: the tags are added in order of position, but the stack still needs to
// be sorted following the right tiebreakers. This is what we're testing
$parser->addEndTag('X', 2, 0);
$parser->addEndTag('X', 1, 0);
$parser->addStartTag('X', 1, 0);
$parser->addStartTag('X', 0, 0);
}], ['...', '<r><X>.</X><X>.</X><X>.</X></r>', function ($configurator) {
$configurator->tags->add('X')->filterChain->append(function ($tag, $parser) {
if ($tag->getPos() === 0) {
$parser->addSelfClosingTag('X', 2, 1);
}
return true;
})->addParameterByName('parser');
}, function ($parser) {
$parser->addSelfClosingTag('X', 0, 1);
$parser->addSelfClosingTag('X', 1, 1);
}], ['...', '<r><X>.</X><X>.</X><X>.</X></r>', function ($configurator) {
$configurator->tags->add('X')->filterChain->append(function ($tag, $parser) {
if ($tag->getPos() === 0) {
$parser->addSelfClosingTag('X', 1, 1);
}
return true;
})->addParameterByName('parser');
}, function ($parser) {
$parser->addSelfClosingTag('X', 0, 1);
$parser->addSelfClosingTag('X', 2, 1);
}], ['[UL]
[*] foo
[*] bar
[/UL]', '<r><UL><s>[UL]</s>
<LI><s>[*]</s> foo</LI>
<LI><s>[*]</s> bar</LI>
<e>[/UL]</e></UL></r>', function ($configurator) {
$configurator->tags->add('UL');
$rules = $configurator->tags->add('LI')->rules;
$rules->closeAncestor('LI');
$rules->ignoreSurroundingWhitespace();
}, function ($parser) {
$parser->addStartTag('UL', 0, 4);
$parser->addStartTag('LI', 10, 3);
$parser->addStartTag('LI', 23, 3);
$parser->addEndTag('UL', 35, 5);
}], ['[UL]
[*] foo
[*] bar
[/UL]', '<r><UL><s>[UL]</s>
<LI><s>[*]</s> foo</LI>
<LI><s>[*]</s> bar</LI>
<e>[/UL]</e></UL></r>', function ($configurator) {
$configurator->tags->add('UL');
$rules = $configurator->tags->add('LI')->rules;
$rules->closeParent('LI');
$rules->ignoreSurroundingWhitespace();
}, function ($parser) {
$parser->addStartTag('UL', 0, 4);
$parser->addStartTag('LI', 10, 3);
$parser->addStartTag('LI', 23, 3);
$parser->addEndTag('UL', 35, 5);
}], ['XX', '<r>X<X>X</X></r>', function ($configurator) {
$configurator->tags->add('X')->filterChain->append(function ($tag) {
return (bool) $tag->getPos();
});
}, function ($parser) {
$parser->addSelfClosingTag('X', 0, 1);
$parser->addSelfClosingTag('X', 1, 1);
}], ['XYX', '<r><X><s>X</s>Y<e>X</e></X></r>', function ($configurator) {
$configurator->tags->add('X')->rules->denyChild('Y');
$configurator->tags->add('Y');
}, function ($parser) {
$parser->addStartTag('X', 0, 1);
$parser->addSelfClosingTag('Y', 1, 1);
$parser->addEndTag('X', 2, 1);
}, [['warn', 'Tag is not allowed in this context', ['tagName' => 'Y', 'tag' => $this->runClosure(function () {
$tag = new Tag(Tag::SELF_CLOSING_TAG, 'Y', 1, 1);
$tag->invalidate();
return $tag;
})]]]], ['XX', '<r><X><s>X</s><e>X</e></X></r>', function ($configurator) {
$configurator->tags->add('X')->rules->denyChild('Y');
$configurator->tags->add('Y');
}, function ($parser) {
$parser->addStartTag('X', 0, 1);
$parser->addSelfClosingTag('Y', 1, 0);
$parser->addEndTag('X', 1, 1);
}, [['debug', 'Tag is not allowed in this context', ['tagName' => 'Y', 'tag' => $this->runClosure(function () {
$tag = new Tag(Tag::SELF_CLOSING_TAG, 'Y', 1, 0);
$tag->invalidate();
return $tag;
})]]]], ['X', '<t>X</t>', function ($configurator) {
$configurator->tags->add('X')->rules->requireAncestor('Y');
$configurator->tags->add('Y');
}, function ($parser) {
$parser->addSelfClosingTag('X', 0, 1);
}], ['.X.', '<r><NOPARSE><s>.</s>X<e>.</e></NOPARSE></r>', function ($configurator) {
$configurator->tags->add('NOPARSE')->rules->ignoreTags();
$configurator->tags->add('X');
}, function ($parser) {
$parser->addTagPair('NOPARSE', 0, 1, 2, 1);
$parser->addSelfClosingTag('X', 1, 1);
}], ['.X.', '<r><NOPARSE><s>.</s>X<e>.</e></NOPARSE></r>', function ($configurator) {
$configurator->tags->add('NOPARSE')->rules->ignoreTags();
$configurator->tags->add('X')->rules->closeParent('NOPARSE');
}, function ($parser) {
$parser->addTagPair('NOPARSE', 0, 1, 2, 1);
$parser->addSelfClosingTag('X', 1, 1);
}], ['X.X.X', '<r><X><s>X</s><NOPARSE><s>.</s>X<e>.</e></NOPARSE><e>X</e></X></r>', function ($configurator) {
$configurator->tags->add('NOPARSE')->rules->ignoreTags();
$configurator->tags->add('X');
}, function ($parser) {
$parser->addTagPair('NOPARSE', 1, 1, 3, 1);
$parser->addStartTag('X', 0, 1);
$parser->addEndTag('X', 2, 1);
$parser->addEndTag('X', 4, 1);
}], ['.X.', '<r><NOPARSE>.<i>X</i>.</NOPARSE></r>', function ($configurator) {
$configurator->tags->add('NOPARSE')->rules->ignoreTags();
}, function ($parser) {
$parser->addTagPair('NOPARSE', 0, 0, 3, 0);
$parser->addIgnoreTag(1, 1);
}], ['.XX.', '<r><NOPARSE><p>.X</p><p>X.</p></NOPARSE></r>', function ($configurator) {
$tag = $configurator->tags->add('NOPARSE');
$tag->rules->createParagraphs();
$tag->rules->ignoreTags();
}, function ($parser) {
$parser->addTagPair('NOPARSE', 0, 0, 4, 0);
$parser->addParagraphBreak(2);
}], ['.X.', '<r><NOPARSE>.<br/>X.</NOPARSE></r>', function ($configurator) {
$configurator->tags->add('NOPARSE')->rules->ignoreTags();
}, function ($parser) {
$parser->addTagPair('NOPARSE', 0, 0, 3, 0);
$parser->addBrTag(1);
}], ['foobar', '<t><p>foo</p><p>bar</p></t>', function ($configurator) {
$configurator->rootRules->createParagraphs();
}, function ($parser) {
$parser->addParagraphBreak(3);
}], ['foo|bar', '<r><p>foo</p><i>|</i><p>bar</p></r>', function ($configurator) {
$configurator->rootRules->createParagraphs();
}, function ($parser) {
$parser->addParagraphBreak(3);
$parser->addIgnoreTag(3, 1);
}], ["\n\n", "<r><X><br/>\n</X><X>\n</X></r>", function ($configurator) {
$configurator->rootRules->enableAutoLineBreaks();
$configurator->tags->add('X');
}, function ($parser) {
$parser->addTagPair('X', 0, 0, 1, 0);
$parser->addTagPair('X', 1, 0, 2, 0)->setFlags(Parser::RULE_SUSPEND_AUTO_BR);
}], ['xxx', "<r><T8>x<T8>x</T8>x</T8></r>", function ($configurator) {
// Create 9 tags with different rules so that they don't occupy the same bit in
// the allowed tags bitfields
for ($i = 0; $i <= 8; ++$i) {
$j = ($i + 1) % 9;
$configurator->tags->add('T' . $i)->rules->denyChild('T' . $j);
}
}, function ($parser) {
$parser->addTagPair('T8', 0, 0, 3, 0);
$parser->addSelfClosingTag('T8', 1, 1);
}], ['[x][y]..[/y][/x]', '<r><X><s>[x]</s></X><Y><s>[y]</s><X>..</X><e>[/y]</e></Y><i>[/x]</i></r>', function ($configurator) {
$configurator->tags->add('X')->rules->autoReopen();
$configurator->tags->add('Y')->rules->closeParent('X')->fosterParent('X');
}, function ($parser) {
$parser->addStartTag('X', 0, 3);
$parser->addStartTag('Y', 3, 3);
$parser->addEndTag('Y', 8, 4);
$parser->addEndTag('X', 12, 4);
}], ["foo\nbar", "<t>foo\nbar</t>"], ["foo\nbar", "<t>foo<br/>\nbar</t>", function ($configurator) {
$configurator->rootRules->enableAutoLineBreaks();
}], ["foo\nbar", "<t>foo\nbar</t>", function ($configurator) {
$configurator->rootRules->disableAutoLineBreaks();
$configurator->rootRules->enableAutoLineBreaks();
}], ["[Y]\n[N]\n[Y]\n[N]\n[/N]\n[/Y]\n[/N]\n[/Y]", "<r><Y><s>[Y]</s><br/>\n<N><s>[N]</s>\n<Y><s>[Y]</s><br/>\n<N><s>[N]</s>\n<e>[/N]</e></N><br/>\n<e>[/Y]</e></Y>\n<e>[/N]</e></N><br/>\n<e>[/Y]</e></Y></r>", function ($configurator) {
$configurator->tags->add('Y')->rules->enableAutoLineBreaks();
$configurator->tags->add('N')->rules->disableAutoLineBreaks();
}, function ($parser) {
$parser->addStartTag('Y', 0, 3);
$parser->addStartTag('N', 4, 3);
$parser->addStartTag('Y', 8, 3);
$parser->addStartTag('N', 12, 3);
$parser->addEndTag('N', 16, 4);
$parser->addEndTag('Y', 21, 4);
$parser->addEndTag('N', 26, 4);
$parser->addEndTag('Y', 31, 4);
}], ["[Y]\n[S]\n[X]\n[/X]\n[/S]\n[/Y]", "<r><Y><s>[Y]</s><br/>\n<S><s>[S]</s>\n<X><s>[X]</s><br/>\n<e>[/X]</e></X>\n<e>[/S]</e></S><br/>\n<e>[/Y]</e></Y></r>", function ($configurator) {
$configurator->tags->add('Y')->rules->enableAutoLineBreaks();
$configurator->tags->add('S')->rules->suspendAutoLineBreaks();
$configurator->tags->add('X');
}, function ($parser) {
$parser->addStartTag('Y', 0, 3);
$parser->addStartTag('S', 4, 3);
$parser->addStartTag('X', 8, 3);
$parser->addEndTag('X', 12, 4);
$parser->addEndTag('S', 17, 4);
$parser->addEndTag('Y', 22, 4);
}], ['...', '<t>...</t>', function ($configurator) {
$configurator->tags->add('X');
}, function ($parser) {
$parser->addSelfClosingTag('X', 1, 1);
$parser->addVerbatim(0, 3);
}], [".\n.\n", "<t>.\n.<br/>\n</t>", function ($configurator) {
$configurator->rootRules->enableAutoLineBreaks();
}, function ($parser) {
$parser->addVerbatim(1, 1)->setFlags(0);
}], [".\n.\n", "<t>.<br/>\n.\n</t>", function ($configurator) {
$configurator->rootRules->disableAutoLineBreaks();
}, function ($parser) {
$parser->addVerbatim(1, 1)->setFlags(Parser::RULE_ENABLE_AUTO_BR);
}], ['xxx', '<r><i>xxx</i></r>', null, function ($parser) {
$parser->addIgnoreTag(0, 99);
}]];
}
/**
* sortTags() callback
*
* Tags are stored as a stack, in LIFO order. We sort tags by position _descending_ so that they
* are processed in the order they appear in the text.
*
* @param Tag $a First tag to compare
* @param Tag $b Second tag to compare
* @return integer
*/
protected static function compareTags(Tag $a, Tag $b)
{
$aPos = $a->getPos();
$bPos = $b->getPos();
// First we order by pos descending
if ($aPos !== $bPos) {
return $bPos - $aPos;
}
// If the tags start at the same position, we'll use their sortPriority if applicable. Tags
// with a lower value get sorted last, which means they'll be processed first. IOW, -10 is
// processed before 10
if ($a->getSortPriority() !== $b->getSortPriority()) {
return $b->getSortPriority() - $a->getSortPriority();
}
// If the tags start at the same position and have the same priority, we'll sort them
// according to their length, with special considerations for zero-width tags
$aLen = $a->getLen();
$bLen = $b->getLen();
if (!$aLen || !$bLen) {
// Zero-width end tags are ordered after zero-width start tags so that a pair that ends
// with a zero-width tag has the opportunity to be closed before another pair starts
// with a zero-width tag. For example, the pairs that would enclose each of the letters
// in the string "XY". Self-closing tags are ordered between end tags and start tags in
// an attempt to keep them out of tag pairs
if (!$aLen && !$bLen) {
$order = [Tag::END_TAG => 0, Tag::SELF_CLOSING_TAG => 1, Tag::START_TAG => 2];
return $order[$b->getType()] - $order[$a->getType()];
}
// Here, we know that only one of $a or $b is a zero-width tags. Zero-width tags are
// ordered after wider tags so that they have a chance to be processed before the next
// character is consumed, which would force them to be skipped
return $aLen ? -1 : 1;
}
// Here we know that both tags start at the same position and have a length greater than 0.
// We sort tags by length ascending, so that the longest matches are processed first. If
// their length is identical, the order is undefined as PHP's sort isn't stable
return $aLen - $bLen;
}
/**
* @testdox addTagPair() returns the newly-created start tag
*/
public function testAddTagPairReturn()
{
$dummyStack = new DummyStack();
$startTag = new Tag(Tag::START_TAG, 'FOO', 1, 2);
$endTag = new Tag(Tag::END_TAG, 'FOO', 3, 4);
$startTag->pairWith($endTag);
$this->assertEquals($startTag, $dummyStack->addTagPair('FOO', 1, 2, 3, 4));
}
