MensBeam
/
HTML-Parser
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
Browse Source

Started fixing whitespace collapsing per specification

pretty-print
Dustin Wilson 3 years ago
parent
6da73a3dab
commit
b145e1ebc9
2 changed files with 228 additions and 27 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 79
      lib/Parser/Serializer.php
  2. 176
      tests/cases/TestSerializer.php

79
lib/Parser/Serializer.php
View File

@ -133,7 +133,7 @@ abstract class Serializer {
if ($htmlElement) {
// If the element's parent is to be treated as block then we need to modify
// whitespace.
if (!$first && static::treatAsBlock($node->parentNode)) {
if (!$first && self::treatAsBlock($node->parentNode)) {
$modify = true;
}
}
@ -150,7 +150,7 @@ abstract class Serializer {
$modify = true;
$foreignAsBlock = true;
} elseif (($node->parentNode->namespaceURI ?? Parser::HTML_NAMESPACE) === Parser::HTML_NAMESPACE) {
if (static::treatAsBlock($node->parentNode)) {
if (self::treatAsBlock($node->parentNode)) {
$modify = true;
$foreignAsBlock = true;
}
@ -178,8 +178,8 @@ abstract class Serializer {
// additional newline. This causes like elements to be grouped together.
$n = $node;
while ($n = $n->previousSibling) {
if (!$n instanceof \DOMText && !$n instanceof \DOMDocumentType) {
if (!$n instanceof \DOMElement || ($n->tagName !== $tagName && count(array_intersect([ $n->tagName, $tagName ], self::H_ELEMENTS)) !== 2)) {
if (!$n instanceof \DOMText) {
if ((!$n instanceof \DOMElement && !$n instanceof \DOMDocumentType) || ($n instanceof \DOMElement && $n->tagName !== $tagName && count(array_intersect([ $n->tagName, $tagName ], self::H_ELEMENTS)) !== 2)) {
$s .= "\n";
}
break;
@ -330,7 +330,7 @@ abstract class Serializer {
} while ($n = $n->nextSibling);
}
if ($firstElementChild !== null && ($foreignAsBlock || ($htmlElement && static::treatAsBlock($node)))) {
if ($firstElementChild !== null && ($foreignAsBlock || ($htmlElement && self::treatAsBlock($node)))) {
$s .= "\n" . str_repeat($indentChar, $indentionLevel * $indentStep);
}
}
@ -357,12 +357,37 @@ abstract class Serializer {
if ($serializerState['reformatWhitespace']) {
$preformattedContent = $serializerState['preformattedContent'] ?: static::isPreformattedContent($node);
if (!$preformattedContent && ($serializerState['foreignAsBlock'] || static::treatAsBlock($node->parentNode)) && strspn($data, Data::WHITESPACE) === strlen($data)) {
return $s;
}
if (!$preformattedContent) {
$treatAsBlock = self::treatAsBlock($node);
$modify = false;
if (($serializerState['foreignAsBlock'] || $treatAsBlock || (self::treatAsBlock($node->parentNode) && count($node->parentNode->childNodes) === 1)) && strspn($data, Data::WHITESPACE) === strlen($data)) {
return $s;
}
// Condense spaces and tabs into a single space.
$data = preg_replace('/ +/', ' ', str_replace("\t", ' ', $data));
if ($treatAsBlock) {
// Block formatting context -- remove all whitespace
$data = preg_replace(Data::WHITESPACE_REGEX, '', $data);
} else {
// Inline formatting context
$data = preg_replace([
// 1. Remove all whitespace before and after a newline
'/[\t\n\x0c\x0D ]*\n[\t\n\x0c\x0D ]*/',
// 2. Convert all tabs to a single space
'/\t/',
// 3. Convert all line breaks to a single space
'/\n/'
], [
"\n",
' ',
' '
], $data);
// 4. Convert multiple spaces to a single space even across inline elements.
// 5. Spaces at the beginning and ending of a line (beginning and ending of
// inline content) are removed.
}
}
}
$s .= self::escapeString($data);
@ -372,7 +397,7 @@ abstract class Serializer {
elseif ($node instanceof \DOMComment) {
if ($serializerState['reformatWhitespace'] && !$serializerState['first']) {
$preformattedContent = $serializerState['preformattedContent'] ?: static::isPreformattedContent($node);
if (!$preformattedContent && ($serializerState['foreignAsBlock'] || static::treatAsBlock($node->parentNode))) {
if (!$preformattedContent && ($serializerState['foreignAsBlock'] || self::treatAsBlock($node->parentNode))) {
$n = $node;
while ($n = $n->previousSibling) {
if (!$n instanceof \DOMText) {
@ -398,7 +423,7 @@ abstract class Serializer {
elseif ($node instanceof \DOMProcessingInstruction) {
if ($serializerState['reformatWhitespace'] && !$serializerState['first']) {
$preformattedContent = $serializerState['preformattedContent'] ?: static::isPreformattedContent($node);
if (!$preformattedContent && ($serializerState['foreignAsBlock'] || static::treatAsBlock($node->parentNode))) {
if (!$preformattedContent && ($serializerState['foreignAsBlock'] || self::treatAsBlock($node->parentNode))) {
$n = $node;
while ($n = $n->previousSibling) {
if (!$n instanceof \DOMText) {
@ -497,16 +522,32 @@ abstract class Serializer {
}
protected static function treatAsBlock(\DOMNode $node): bool {
// NOTE: This method is used only when pretty printing. Implementors of userland
// PHP DOM solutions with template contents will need to extend this method to
// check for any templates and look within their content fragments for "block"
// content.
if ($node instanceof \DOMDocument || $node instanceof \DOMDocumentFragment) {
if ($node instanceof \DOMDocument || $node instanceof \DOMDocumentFragment || ($node instanceof \DOMElement && in_array($node->tagName, self::BLOCK_ELEMENTS))) {
return true;
}
if (!$node instanceof \DOMElement) {
$node = $node->parentNode;
if ($node === null) {
return false;
}
}
$xpath = new \DOMXPath($node->ownerDocument);
return ($xpath->evaluate(self::BLOCK_QUERY, $node) > 0);
if ($xpath->evaluate(self::BLOCK_QUERY, $node) === 0) {
return static::treatAsBlockWithTemplates($node);
}
return true;
}
protected static function treatAsBlockWithTemplates(\DOMNode $node): bool {
// NOTE: This method is used only when pretty printing. Implementors of userland
// PHP DOM solutions with template contents will need to extend this method to
// check for any templates and look within their content fragments for "block"
// content.
return $result;
}
protected static function treatForeignRootAsBlock(\DOMNode $node): bool {
@ -518,7 +559,7 @@ abstract class Serializer {
if ($n instanceof \DOMDocument || $n instanceof \DOMDocumentFragment || ($n instanceof \DOMElement && $n->parentNode === null)) {
return true;
} elseif (($n->parentNode->namespaceURI ?? Parser::HTML_NAMESPACE) === Parser::HTML_NAMESPACE) {
if (static::treatAsBlock($n->parentNode)) {
if (self::treatAsBlock($n->parentNode)) {
return true;
}
break;

176
tests/cases/TestSerializer.php
View File

@ -122,22 +122,182 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
}
/** @dataProvider provideCustomSerializations */
public function testSerializeWithSimpleOptions(string $in, bool $boolAttr, bool $foreignVoid, string $exp): void {
public function testSerializeWithOptions(int $indentStep, bool $indentWithSpaces, bool $processingInstructions, bool $reformatWhitespace, bool $boolAttr, bool $foreignVoid, string $in, string $exp): void {
$config = new Config;
$config->indentStep = $indentStep;
$config->indentWithSpaces = $indentWithSpaces;
$config->processingInstructions = $processingInstructions;
$config->reformatWhitespace = $reformatWhitespace;
$config->serializeBooleanAttributeValues = $boolAttr;
$config->serializeForeignVoidEndTags = $foreignVoid;
$body = Parser::parse($in, "UTF-8")->document->getElementsByTagName("body")[0];
$act = Parser::serializeInner($body, $config);
$d = Parser::parse($in, "UTF-8")->document;
$act = Parser::serialize($d, $config);
$this->assertSame($exp, $act);
}
public function provideCustomSerializations(): iterable {
return [
['<a hidden="hidden"></a><b hidden=""></b><c hidden="HIDDEN"></c><d hidden="true"></d>', true, true, '<a hidden="hidden"></a><b hidden=""></b><c hidden="HIDDEN"></c><d hidden="true"></d>'],
['<a hidden="hidden"></a><b hidden=""></b><c hidden="HIDDEN"></c><d hidden="true"></d>', false, true, '<a hidden></a><b hidden></b><c hidden></c><d hidden="true"></d>'],
['<br><svg/><svg>blah</svg><math/><math>blah</math><input>', true, true, '<br><svg></svg><svg>blah</svg><math></math><math>blah</math><input>'],
['<br><svg/><svg>blah</svg><math/><math>blah</math><input>', true, false, '<br><svg/><svg>blah</svg><math/><math>blah</math><input>'],
['<audio loop hidden></audio><svg/>', false, false, '<audio loop hidden></audio><svg/>'],
// Boolean attribute values serialized
[0, false, false, false, true, true,
<<<HTML
<a hidden="hidden"></a><b hidden=""></b><c hidden="HIDDEN"></c><d hidden="true"></d>
HTML,
<<<HTML
<html><head></head><body><a hidden="hidden"></a><b hidden=""></b><c hidden="HIDDEN"></c><d hidden="true"></d></body></html>
HTML
],
// Boolean attribute values not serialized
[0, false, false, false, false, true,
<<<HTML
<a hidden="hidden"></a><b hidden=""></b><c hidden="HIDDEN"></c><d hidden="true"></d>
HTML,
<<<HTML
<html><head></head><body><a hidden></a><b hidden></b><c hidden></c><d hidden="true"></d></body></html>
HTML
],
// Boolean attribute values serialized, foreign void end tags serialized
[0, false, false, false, true, true,
<<<HTML
<br><svg/><svg>blah</svg><math/><math>blah</math><input>
HTML,
<<<HTML
<html><head></head><body><br><svg></svg><svg>blah</svg><math></math><math>blah</math><input></body></html>
HTML
],
// Boolean attribute values serialized, foreign void end tags not serialized
[0, false, false, false, true, false,
<<<HTML
<br><svg/><svg>blah</svg><math/><math>blah</math><input>
HTML,
<<<HTML
<html><head></head><body><br><svg/><svg>blah</svg><math/><math>blah</math><input></body></html>
HTML
],
// Neither attribute values nor foreign void end tags serialized
[0, false, false, false, false, false,
<<<HTML
<audio loop hidden></audio><svg/>
HTML,
<<<HTML
<html><head></head><body><audio loop hidden></audio><svg/></body></html>
HTML
],
// Reformat whitespace, empty document
[1, true, false, true, false, false,
<<<HTML
<html></html>
HTML,
<<<HTML
<html>
<head></head>
<body></body>
</html>
HTML
],
// Reformat whitespace, comment before doctype
[1, true, false, true, false, false,
<<<HTML
<!--data-->
<!DOCTYPE html>
<html></html>
HTML,
<<<HTML
<!--data-->
<!DOCTYPE html>
<html>
<head></head>
<body></body>
</html>
HTML
],
// Reformat whitespace, preformatted element
[1, true, false, true, false, false,
<<<HTML
<pre><code></code></pre>
HTML,
<<<HTML
<html>
<head></head>
<body>
<pre><code></code></pre>
</body>
</html>
HTML
],
// Reformat whitespace, element grouping, foreign "block" content, & foreign
// void end tags not serialized
[1, true, false, true, false, false,
<<<HTML
<div></div><svg><g id="ook"></g></svg>
HTML,
<<<HTML
<html>
<head></head>
<body>
<div></div>
<svg>
<g id="ook"/>
</svg>
</body>
</html>
HTML
],
// Reformat whitespace, whitespace collapsing, custom indentions
[4, true, false, true, false, false,
<<<HTML
<!DOCTYPE html>
<html>
<head>
</head>
<body>
ook
<div/>
</body>
</html>
HTML,
<<<HTML
<!DOCTYPE html>
<html>
<head></head>
<body>
ook
<div></div>
</body>
</html>
HTML
],
];
}

Write Preview
Loading…
Cancel
Save