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

100% coverage on pretty printing

pretty-print
Dustin Wilson 2 years ago
parent
8361ea0d88
commit
08ba468214
2 changed files with 374 additions and 59 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. 104
      lib/Parser/Serializer.php
  2. 329
      tests/cases/TestSerializer.php

104
lib/Parser/Serializer.php
View File

@ -11,9 +11,6 @@ use MensBeam\HTML\Parser;
abstract class Serializer {
use NameCoercion;
protected const BLOCK_ELEMENTS = [ 'address', 'article', 'aside', 'blockquote', 'canvas', 'dd', 'div', 'dl', 'dt', 'fieldset', 'figcaption', 'figure', 'footer', 'form', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'header', 'hr', 'li', 'main', 'nav', 'noscript', 'ol', 'p', 'pre', 'section', 'table', 'tfoot', 'ul', 'video' ];
// Elements treated as block elements when reformatting whitespace
protected const PRINTING_BLOCK_ELEMENTS = [ 'address', 'article', 'aside', 'blockquote', 'base', 'body', 'canvas', 'details', 'dialog', 'dd', 'div', 'dl', 'dt', 'fieldset', 'figcaption', 'figure', 'footer', 'form', 'frame', 'frameset', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'head', 'header', 'hr', 'html', 'isindex', 'li', 'link', 'main', 'meta', 'nav', 'noscript', 'ol', 'p', 'picture', 'pre', 'section', 'script', 'source', 'style', 'table', 'td', 'tfoot', 'th', 'thead', 'title', 'tr', 'ul', 'video' ];
// List of h-elements which are used to determine element grouping for the
// purposes of reformatting whitespace
protected const H_ELEMENTS = [ 'h1', 'h2', 'h3', 'h4', 'h5', 'h6' ];
@ -321,6 +318,7 @@ abstract class Serializer {
$firstElementChild = null;
if (property_exists($node, 'firstElementChild')) {
$firstElementChild = $node->firstElementChild;
// @codeCoverageIgnoreStart
} else {
$n = $node->firstChild;
do {
@ -330,6 +328,7 @@ abstract class Serializer {
}
} while ($n = $n->nextSibling);
}
// @codeCoverageIgnoreEnd
if ($firstElementChild !== null && ($foreignAsBlock || ($htmlElement && self::treatAsBlock($node)))) {
$s .= "\n" . str_repeat($indentChar, $indentionLevel * $indentStep);
@ -355,19 +354,21 @@ abstract class Serializer {
# Otherwise, append the value of current node's data IDL attribute, escaped as described below.
else {
$data = $node->data;
if ($serializerState['reformatWhitespace']) {
// The serializer should disable 'reformatWhitespace' on children of a
// preformatted element, but just in case check for it here.
$preformattedContent = $serializerState['preformattedContent'] ?: static::isPreformattedContent($node);
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($data)) {
if (($serializerState['foreignAsBlock'] || $treatAsBlock || ($node->parentNode !== null && self::treatAsBlock($node->parentNode) && count($node->parentNode->childNodes) === 1)) && strspn($data, Data::WHITESPACE) === strlen($data)) {
return $s;
}
if ($treatAsBlock) {
// Block formatting context -- remove all whitespace
$data = preg_replace(Data::WHITESPACE_REGEX, '', $data);
// Block formatting context -- trim data and convert all whitespace to a single
// space
$data = preg_replace('/[\t\n\x0c\x0D ]+/', ' ', trim($data));
if ($data === '') {
return $s;
}
@ -386,34 +387,55 @@ abstract class Serializer {
' '
], $data);
// 4. Convert multiple spaces to a single space even across inline elements.
//
// This will be accomplished by looking backwards through siblings, checking
// if the previous text node had whitespace at the end and then lobbing off
// whitespace at the beginning of the current text node. This has the added
// benefit of doing part of the work of #5 as well -- if it matches.
// Moonwalk and find the closest block element (actual block element, not
// elements treated as block for the purposes of serializing) then grab all
// descendant text nodes that aren't descendants of templates.
$xpath = new \DOMXPath($node->ownerDocument);
$previousTextNode = $xpath->query('./preceding-sibling::text()[1] | ./preceding-sibling::*/descendant::text()[1]', $node);
$ltrimmed = false;
if ($previousTextNode->length > 0) {
$data2 = $previousTextNode->item(0)->data;
if (preg_match('/[\t\n\x0c\x0D ]+$/', $data2)) {
$data = ltrim($data);
$ltrimmed = true;
$textNodes = $xpath->query('./ancestor::*[namespace-uri()="" or namespace-uri()="http://www.w3.org/1999/xhtml"][name()="address" or name()="article" or name()="aside" or name()="blockquote" or name="body" or name()="canvas" or name()="dd" or name()="div" or name()="dl" or name()="dt" or name()="fieldset" or name()="figcaption" or name()="figure" or name()="footer" or name()="form" or name()="h1" or name()="h2" or name()="h3" or name()="h4" or name()="h5" or name()="h6" or name()="head" or name()="header" or name()="hr" or name()="html" or name()="li" or name()="main" or name()="nav" or name()="ol" or name()="p" or name()="section" or name()="table" or name()="tfoot" or name()="ul" or name()="video"][1]/descendant::text()[not(ancestor::template[namespace-uri()="" or namespace-uri()="http://www.w3.org/1999/xhtml"])]', $node);
// If nothing was matched then the text node is either disconnected from its
// document and being serialized alone or an inline descendant of a document
// fragment.
if ($textNodes->length > 0) {
$firstOfLine = ($node === $textNodes->item(0));
$lastOfLine = ($node === $textNodes->item($textNodes->length - 1));
} else {
// If the text node is either disconnected from its document then firstOfLine
// and lastOfLine is true.
if ($node->parentNode === null) {
$firstOfLine = $lastOfLine = true;
}
// Otherwise, it's an inline descendant of a document fragment. Find its root
// node and then grab all text node descendants of that fragment.
else {
$n = $node;
while ($n = $n->parentNode) {
$root = $n;
}
$textNodes = $xpath->query('.//text()[not(ancestor::template[namespace-uri()="" or namespace-uri()="http://www.w3.org/1999/xhtml"])]', $root);
$firstOfLine = ($node === $textNodes->item(0));
$lastOfLine = ($node === $textNodes->item($textNodes->length - 1));
}
}
// 5. Spaces at the beginning and ending of a line (beginning and ending of
// inline content) are removed.
if (!$ltrimmed) {
$firstOfLine = $xpath->query('./ancestor::*[namespace-uri()="" or namespace-uri()="http://www.w3.org/1999/xhtml"][name()="address" or name()="article" or name()="aside" or name()="blockquote" or name="body" or name()="canvas" or name()="dd" or name()="div" or name()="dl" or name()="dt" or name()="fieldset" or name()="figcaption" or name()="figure" or name()="footer" or name()="form" or name()="h1" or name()="h2" or name()="h3" or name()="h4" or name()="h5" or name()="h6" or name="head" or name="header" or name="hr" or name="html" or name="li" or name="main" or name="nav" or name="ol" or name="p" or name="section" or name="table" or name="tfoot" or name="ul" or name="video"][1]/descendant::text()[1]', $node);
if ($firstOfLine->length > 0 && $node === $firstOfLine->item(0)) {
$data = ltrim($data);
// 4. Convert multiple spaces to a single space even across inline elements.
$data = preg_replace('/ +/', ' ', $data);
if (!$firstOfLine) {
foreach ($textNodes as $key => $t) {
if ($t === $node && preg_match('/[\t\n\x0c\x0D ]+$/', $textNodes[$key - 1]->data)) {
$data = ltrim($data);
break;
}
}
}
$lastOfLine = $xpath->query('./ancestor::*[namespace-uri()="" or namespace-uri()="http://www.w3.org/1999/xhtml"][name()="address" or name()="article" or name()="aside" or name()="blockquote" or name="body" or name()="canvas" or name()="dd" or name()="div" or name()="dl" or name()="dt" or name()="fieldset" or name()="figcaption" or name()="figure" or name()="footer" or name()="form" or name()="h1" or name()="h2" or name()="h3" or name()="h4" or name()="h5" or name()="h6" or name="head" or name="header" or name="hr" or name="html" or name="li" or name="main" or name="nav" or name="ol" or name="p" or name="section" or name="table" or name="tfoot" or name="ul" or name="video"][1]/descendant::text()[last()]', $node);
if ($lastOfLine->length > 0 && $node === $lastOfLine->item(0)) {
// 5. Spaces at the beginning and ending of a line (beginning and ending of
// inline content) are removed.
if ($firstOfLine) {
$data = ltrim($data);
}
if ($lastOfLine) {
$data = rtrim($data);
}
}
@ -552,7 +574,7 @@ abstract class Serializer {
}
protected static function treatAsBlock(\DOMNode $node): bool {
if ($node instanceof \DOMDocument || $node instanceof \DOMDocumentFragment || ($node instanceof \DOMElement && in_array($node->tagName, self::PRINTING_BLOCK_ELEMENTS))) {
if ($node instanceof \DOMDocument || $node instanceof \DOMDocumentFragment) {
return true;
}
@ -565,11 +587,12 @@ abstract class Serializer {
}
$xpath = new \DOMXPath($node->ownerDocument);
if ($xpath->evaluate(self::BLOCK_QUERY, $node) === 0) {
$result = ($xpath->evaluate(self::BLOCK_QUERY, $node) > 0);
if (!$result) {
return static::treatAsBlockWithTemplates($node);
}
return true;
return $result;
}
protected static function treatAsBlockWithTemplates(\DOMNode $node): bool {
@ -577,7 +600,7 @@ abstract class Serializer {
// 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;
return false;
}
protected static function treatForeignRootAsBlock(\DOMNode $node): bool {
@ -585,16 +608,17 @@ abstract class Serializer {
// PHP DOM solutions with template contents will need to extend this method to
// be able to moonwalk through document fragment hosts.
$n = $node;
while ($n = $n->parentNode) {
if ($n instanceof \DOMDocument || $n instanceof \DOMDocumentFragment || ($n instanceof \DOMElement && $n->parentNode === null)) {
do {
if ($n->parentNode !== null && ($n->parentNode->namespaceURI ?? Parser::HTML_NAMESPACE) !== Parser::HTML_NAMESPACE) {
continue;
}
if (self::treatAsBlock($n->parentNode)) {
return true;
} elseif (($n->parentNode->namespaceURI ?? Parser::HTML_NAMESPACE) === Parser::HTML_NAMESPACE) {
if (self::treatAsBlock($n->parentNode)) {
return true;
}
break;
}
}
break;
} while ($n = $n->parentNode);
return false;
}

329
tests/cases/TestSerializer.php
View File

@ -122,7 +122,7 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
}
/** @dataProvider provideCustomSerializations */
public function testSerializeWithOptions(int $indentStep, bool $indentWithSpaces, bool $processingInstructions, bool $reformatWhitespace, bool $boolAttr, bool $foreignVoid, string $in, string $exp): void {
public function testSerializeWithOptions(bool $fragment, ?string $fragmentContext, 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;
@ -130,15 +130,22 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
$config->reformatWhitespace = $reformatWhitespace;
$config->serializeBooleanAttributeValues = $boolAttr;
$config->serializeForeignVoidEndTags = $foreignVoid;
$d = Parser::parse($in, "UTF-8")->document;
$act = Parser::serialize($d, $config);
if (!$fragment) {
$d = Parser::parse($in, "UTF-8", $config)->document;
$act = Parser::serialize($d, $config);
} else {
$d = new \DOMDocument();
$act = Parser::serialize(Parser::parseFragment($d->createElement($fragmentContext), 0, $in, 'UTF-8', $config), $config);
}
$this->assertSame($exp, $act);
}
public function provideCustomSerializations(): iterable {
return [
// Boolean attribute values serialized
[0, false, false, false, true, true,
[false, null, 0, false, false, false, true, true,
<<<HTML
<a hidden="hidden"></a><b hidden=""></b><c hidden="HIDDEN"></c><d hidden="true"></d>
HTML,
@ -149,7 +156,7 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
],
// Boolean attribute values not serialized
[0, false, false, false, false, true,
[false, null, 0, false, false, false, false, true,
<<<HTML
<a hidden="hidden"></a><b hidden=""></b><c hidden="HIDDEN"></c><d hidden="true"></d>
HTML,
@ -160,7 +167,7 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
],
// Boolean attribute values serialized, foreign void end tags serialized
[0, false, false, false, true, true,
[false, null, 0, false, false, false, true, true,
<<<HTML
<br><svg/><svg>blah</svg><math/><math>blah</math><input>
HTML,
@ -171,7 +178,7 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
],
// Boolean attribute values serialized, foreign void end tags not serialized
[0, false, false, false, true, false,
[false, null, 0, false, false, false, true, false,
<<<HTML
<br><svg/><svg>blah</svg><math/><math>blah</math><input>
HTML,
@ -182,7 +189,7 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
],
// Neither attribute values nor foreign void end tags serialized
[0, false, false, false, false, false,
[false, null, 0, false, false, false, false, false,
<<<HTML
<audio loop hidden></audio><svg/>
HTML,
@ -193,7 +200,7 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
],
// Reformat whitespace, empty document
[1, true, false, true, false, false,
[false, null, 1, true, false, true, false, false,
<<<HTML
<html></html>
HTML,
@ -208,7 +215,7 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
],
// Reformat whitespace, comment before doctype
[1, true, false, true, false, false,
[false, null, 1, true, false, true, false, false,
<<<HTML
<!--data-->
<!DOCTYPE html>
@ -227,7 +234,7 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
],
// Reformat whitespace, preformatted element
[1, true, false, true, false, false,
[false, null, 1, true, false, true, false, false,
<<<HTML
<pre><code></code></pre>
HTML,
@ -245,7 +252,7 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
// Reformat whitespace, element grouping, foreign "block" content, & foreign
// void end tags not serialized
[1, true, false, true, false, false,
[false, null, 1, true, false, true, false, false,
<<<HTML
<div></div><svg><g id="ook"></g></svg>
HTML,
@ -265,8 +272,60 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
HTML
],
// Inline serialized comments and processing instructions, parsing of processing instructions off
[false, null, 1, true, false, true, false, false,
<<<HTML
<html>
<head></head>
<body>
<!--ook-->
<?ook eeeeek ?>
</body>
</html>
HTML,
<<<HTML
<html>
<head></head>
<body><!--ook--><!--?ook eeeeek ?--></body>
</html>
HTML
],
// Block serialized comments and processing instructions, parsing of processing instructions on
[false, null, 1, true, true, true, false, false,
<<<HTML
<html>
<head></head>
<body>
<div></div>
<!--ook-->
<?ook eeeeek ?>
<div></div>
</body>
</html>
HTML,
<<<HTML
<html>
<head></head>
<body>
<div></div>
<!--ook-->
<?ook eeeeek ?>
<div></div>
</body>
</html>
HTML
],
// Reformat whitespace, whitespace collapsing, custom indentions
[4, true, false, true, false, false,
[false, null, 4, true, false, true, false, false,
<<<HTML
<!DOCTYPE html>
<html>
@ -277,9 +336,15 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
</head>
<body>
ook
<div/>
ook eek
<pre>
This should be ignored
also this
</pre>
<div></div>
<p> Ook
<span> Eek!</span> </p>
</body>
</html>
HTML,
@ -289,15 +354,241 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
<html>
<head></head>
<body>
ook
<body>ook eek
<pre> This should be ignored
also this
</pre>
<div></div>
<p>Ook <span>Eek!</span></p>
</body>
</html>
HTML
],
// Fragment, html elements
[true, 'div', 1, true, false, true, false, false,
<<<HTML
<span> <span> Ook!</span></span>
HTML,
<<<HTML
<span><span>Ook!</span></span>
HTML
],
// Fragment, foreign elements
[true, 'div', 1, true, false, true, false, false,
<<<HTML
<svg> <g><path d=""/></g></svg>
HTML,
<<<HTML
<svg>
<g>
<path d=""/>
</g>
</svg>
HTML
],
// Fragment, foreign elements
[true, 'div', 1, true, false, true, false, false,
<<<HTML
<svg> <g><path d=""/></g></svg>
HTML,
<<<HTML
<svg>
<g>
<path d=""/>
</g>
</svg>
HTML
],
];
}
/** @dataProvider provideCustomSerializationsForNodes */
public function testSerializeNodesWithOptions(int $indentStep, bool $indentWithSpaces, bool $processingInstructions, bool $reformatWhitespace, bool $boolAttr, bool $foreignVoid, \Closure $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;
$act = $in($config);
$this->assertSame($exp, $act);
}
public function provideCustomSerializationsForNodes(): iterable {
return [
// Solo html element with context
[1, true, false, true, false, false,
function (Config $config): string {
$html = <<<HTML
<!DOCTYPE html>
<html>
<body>
<p> Ook! </p>
</body>
</html>
HTML;
$d = Parser::parse($html, "UTF-8")->document;
return Parser::serialize($d->getElementsByTagName('p')->item(0), $config);
},
<<<HTML
<p>Ook!</p>
HTML
],
// Solo html element without context
[1, true, false, true, false, false,
function (Config $config): string {
$html = <<<HTML
<!DOCTYPE html>
<html>
<body>
<p> Ook! </p>
</body>
</html>
HTML;
$d = Parser::parse($html, "UTF-8")->document;
$p = $d->getElementsByTagName('p')->item(0);
$p->parentNode->removeChild($p);
return Parser::serialize($p, $config);
},
<<<HTML
<p>Ook!</p>
HTML
],
// Solo svg element serializing as inline with context
[1, true, false, true, false, true,
function (Config $config): string {
$html = <<<HTML
<!DOCTYPE html>
<html>
<body>
<svg role="img" viewBox="0 0 26 26"><title>Ook</title>
<rect id="eek--a" width="5" height="5"/></svg>
</body>
</html>
HTML;
$d = Parser::parse($html, "UTF-8")->document;
$svg = $d->getElementsByTagName('svg')->item(0);
return Parser::serialize($svg, $config);
},
<<<HTML
<svg role="img" viewBox="0 0 26 26"><title>Ook</title> <rect id="eek--a" width="5" height="5"></rect></svg>
HTML
],
// Solo svg element serializing as block with context
[1, true, false, true, false, false,
function (Config $config): string {
$html = <<<HTML
<!DOCTYPE html>
<html>
<body>
<svg><g><g><rect id="eek--a" width="5" height="5"/></g></g></svg>
<div></div>
</body>
</html>
HTML;
$d = Parser::parse($html, "UTF-8")->document;
$svg = $d->getElementsByTagName('svg')->item(0);
$g = $svg->firstChild->firstChild;
return Parser::serialize($g, $config);
},
<<<HTML
<g>
<rect id="eek--a" width="5" height="5"/>
</g>
HTML
],
// Solo svg element without context
[1, true, false, true, false, true,
function (Config $config): string {
$html = <<<HTML
<!DOCTYPE html>
<html>
<body>
<svg role="img" viewBox="0 0 26 26"><title>Ook</title>
<rect id="eek--a" width="5" height="5"/></svg>
</body>
</html>
HTML;
$d = Parser::parse($html, "UTF-8")->document;
$svg = $d->getElementsByTagName('svg')->item(0);
$svg->parentNode->removeChild($svg);
return Parser::serialize($svg, $config);
},
<<<HTML
<svg role="img" viewBox="0 0 26 26">
<title>Ook</title>
<rect id="eek--a" width="5" height="5"></rect>
</svg>
HTML
],
/*
// Fragment, html elements
[true, 'div', 1, true, false, true, false, false,
<<<HTML
<span> <span> Ook!</span></span>
HTML,
<<<HTML
<span><span>Ook!</span></span>
HTML
],
*/
// Solo text node without context
[1, true, false, true, false, true,
function (Config $config): string {
$html = <<<HTML
<!DOCTYPE html>
<html>
<body>
OOK eeek ooooooook ook
</body>
</html>
HTML;
$d = Parser::parse($html, "UTF-8")->document;
$text = $d->getElementsByTagName('body')->item(0)->firstChild;
$text->parentNode->removeChild($text);
return Parser::serialize($text, $config);
},
<<<HTML
OOK eeek ooooooook ook
HTML
],
];
}

Write Preview
Loading…
Cancel
Save