|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Vikpe; |
|
4
|
|
|
|
|
5
|
|
|
class HtmlHeadingNormalizer |
|
6
|
|
|
{ |
|
7
|
|
|
public static function normalize($html, $base_level = 1) |
|
8
|
|
|
{ |
|
9
|
|
|
if (!self::htmlContainsHeadings($html)) { |
|
10
|
|
|
return $html; |
|
11
|
|
|
} |
|
12
|
|
|
|
|
13
|
|
|
$domDocument = new \DOMDocument(); |
|
14
|
|
|
$domDocument->loadHTML($html); |
|
15
|
|
|
|
|
16
|
|
|
$headingTagNames = ['h1', 'h2', 'h3', 'h4', 'h6']; |
|
17
|
|
|
|
|
18
|
|
|
$originalHeadingDomElements = []; |
|
19
|
|
|
$normalizedHeadingDomElements = []; |
|
20
|
|
|
|
|
21
|
|
|
foreach ($headingTagNames as $headingTagName) { |
|
22
|
|
|
$headingDomElements = $domDocument->getElementsByTagName($headingTagName); |
|
23
|
|
|
|
|
24
|
|
|
foreach ($headingDomElements as $headingDomElement) { |
|
25
|
|
|
$currentHeadingLevel = self::headingTagNameToNumber($headingDomElement->tagName); |
|
26
|
|
|
$newHeadingLevel = self::numberToHeadingLevel($base_level + $currentHeadingLevel - 1); |
|
27
|
|
|
|
|
28
|
|
|
if ($newHeadingLevel !== $currentHeadingLevel) { |
|
29
|
|
|
$originalHeadingDomElements[] = $headingDomElement; |
|
30
|
|
|
$normalizedHeadingDomElements[] = self::cloneDomElementWithNewTagName($headingDomElement, $newHeadingLevel); |
|
31
|
|
|
} |
|
32
|
|
|
} |
|
33
|
|
|
} |
|
34
|
|
|
|
|
35
|
|
|
foreach ($originalHeadingDomElements as $i => $needle) { |
|
36
|
|
|
$needle->parentNode->replaceChild($normalizedHeadingDomElements[$i], $needle); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
return $domDocument->saveHTML(); |
|
40
|
|
|
} |
|
41
|
|
|
|
|
42
|
|
|
private static function htmlContainsHeadings($html) |
|
43
|
|
|
{ |
|
44
|
|
|
$headingNeedle = '<h'; |
|
45
|
|
|
$containsHeadings = (false !== stripos($html, $headingNeedle)); |
|
46
|
|
|
|
|
47
|
|
|
return $containsHeadings; |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
private static function headingTagNameToNumber($headingLevel) |
|
51
|
|
|
{ |
|
52
|
|
|
return substr($headingLevel, 1); |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
private static function numberToHeadingLevel($number) |
|
56
|
|
|
{ |
|
57
|
|
|
return 'h' . $number; |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
private static function cloneDomElementWithNewTagName(\DOMElement $sourceDomElement, $newTagName) |
|
61
|
|
|
{ |
|
62
|
|
|
$targetDomElement = $sourceDomElement->parentNode->ownerDocument->createElement($newTagName); |
|
63
|
|
|
self::copyAttributes($sourceDomElement, $targetDomElement); |
|
64
|
|
|
self::moveChildNodes($sourceDomElement, $targetDomElement); |
|
65
|
|
|
|
|
66
|
|
|
return $targetDomElement; |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
private static function copyAttributes(\DOMElement $source, \DOMElement $target) |
|
70
|
|
|
{ |
|
71
|
|
|
foreach ($source->attributes as $attribute) { |
|
72
|
|
|
$target->setAttribute($attribute->name, $attribute->value); |
|
73
|
|
|
} |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
private static function moveChildNodes(\DOMElement $source, \DOMElement $target) |
|
77
|
|
|
{ |
|
78
|
|
|
while ($source->hasChildNodes()) { |
|
79
|
|
|
// appendChild() actually moves the childNode |
|
80
|
|
|
$target->appendChild($source->childNodes->item(0)); |
|
81
|
|
|
} |
|
82
|
|
|
} |
|
83
|
|
|
} |
|
84
|
|
|
|