1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Vikpe; |
4
|
|
|
|
5
|
|
|
class HtmlHeadingNormalizer |
6
|
|
|
{ |
7
|
|
|
public static function demote($html, $numberOfLevels) |
8
|
|
|
{ |
9
|
|
|
return self::normalize($html, $numberOfLevels); |
10
|
|
|
} |
11
|
|
|
|
12
|
|
|
public static function promote($html, $numberOfLevels) |
13
|
|
|
{ |
14
|
|
|
return self::normalize($html, -$numberOfLevels); |
15
|
|
|
} |
16
|
|
|
|
17
|
|
|
public static function min($html, $minLevel) |
18
|
|
|
{ |
19
|
|
|
if (!self::containsHeadings($html)) { |
20
|
|
|
return $html; |
21
|
|
|
} |
22
|
|
|
|
23
|
|
|
$currentMinLevel = min(self::headingLevels($html)); |
24
|
|
|
$levelDiff = $minLevel - $currentMinLevel; |
25
|
|
|
|
26
|
|
|
return self::normalize($html, $levelDiff); |
27
|
|
|
} |
28
|
|
|
|
29
|
|
|
private static function normalize($html, $numberOfLevels) |
30
|
|
|
{ |
31
|
|
|
$normalizationIsRequired = ((abs($numberOfLevels) > 0) && self::containsHeadings($html)); |
32
|
|
|
|
33
|
|
|
if (!$normalizationIsRequired) { |
34
|
|
|
return $html; |
35
|
|
|
} |
36
|
|
|
|
37
|
|
|
$domDocument = new \DOMDocument(); |
38
|
|
|
$domDocument->loadHTML($html); |
39
|
|
|
|
40
|
|
|
$originalHeadings = self::getHeadings($domDocument); |
41
|
|
|
$normalizedHeadings = self::normalizeHeadings($originalHeadings, $numberOfLevels); |
42
|
|
|
self::replaceHeadings($originalHeadings, $normalizedHeadings); |
43
|
|
|
|
44
|
|
|
return self::formatResult($domDocument, $html); |
45
|
|
|
} |
46
|
|
|
|
47
|
|
|
private static function getHeadings(\DOMDocument $domDocument) |
48
|
|
|
{ |
49
|
|
|
$tagNames = array('h1', 'h2', 'h3', 'h4', 'h5', 'h6'); |
50
|
|
|
$headings = array(); |
51
|
|
|
|
52
|
|
|
foreach ($tagNames as $tagName) { |
53
|
|
|
foreach ($domDocument->getElementsByTagName($tagName) as $heading) { |
54
|
|
|
$headings[] = $heading; |
55
|
|
|
} |
56
|
|
|
} |
57
|
|
|
|
58
|
|
|
return $headings; |
59
|
|
|
} |
60
|
|
|
|
61
|
|
|
private static function normalizeHeadings(array $originalHeadings, $numberOfLevels) |
62
|
|
|
{ |
63
|
|
|
$normalizedHeadings = array(); |
64
|
|
|
|
65
|
|
|
foreach ($originalHeadings as $heading) { |
66
|
|
|
$currentLevel = self::tagNameToLevel($heading->tagName); |
67
|
|
|
$newLevel = $currentLevel + $numberOfLevels; |
68
|
|
|
|
69
|
|
|
$normalizedHeadings[] = self::cloneHeading($heading, $newLevel); |
70
|
|
|
} |
71
|
|
|
|
72
|
|
|
return $normalizedHeadings; |
73
|
|
|
} |
74
|
|
|
|
75
|
|
|
private static function replaceHeadings(array $needles, array $replacements) |
76
|
|
|
{ |
77
|
|
|
foreach ($needles as $i => $needle) { |
78
|
|
|
$needle->parentNode->replaceChild($replacements[$i], $needle); |
79
|
|
|
} |
80
|
|
|
} |
81
|
|
|
|
82
|
|
|
private static function containsHeadings($html) |
83
|
|
|
{ |
84
|
|
|
$headingNeedle = '<h'; |
85
|
|
|
$containsHeadings = (false !== stripos($html, $headingNeedle)); |
86
|
|
|
|
87
|
|
|
return $containsHeadings; |
88
|
|
|
} |
89
|
|
|
|
90
|
|
|
private static function tagNameToLevel($tagName) |
91
|
|
|
{ |
92
|
|
|
return substr($tagName, 1); |
93
|
|
|
} |
94
|
|
|
|
95
|
|
|
private static function levelToTagName($level) |
96
|
|
|
{ |
97
|
|
|
return 'h'.$level; |
98
|
|
|
} |
99
|
|
|
|
100
|
|
|
private static function cloneHeading(\DOMElement $sourceHeading, $newLevel) |
101
|
|
|
{ |
102
|
|
|
$tagName = self::levelToTagName($newLevel); |
103
|
|
|
|
104
|
|
|
$targetHeading = $sourceHeading->parentNode->ownerDocument->createElement($tagName); |
105
|
|
|
self::copyAttributes($sourceHeading, $targetHeading); |
106
|
|
|
self::moveChildNodes($sourceHeading, $targetHeading); |
107
|
|
|
|
108
|
|
|
return $targetHeading; |
109
|
|
|
} |
110
|
|
|
|
111
|
|
|
private static function copyAttributes(\DOMElement $source, \DOMElement $target) |
112
|
|
|
{ |
113
|
|
|
foreach ($source->attributes as $attribute) { |
114
|
|
|
$target->setAttribute($attribute->name, $attribute->value); |
115
|
|
|
} |
116
|
|
|
} |
117
|
|
|
|
118
|
|
|
private static function moveChildNodes(\DOMElement $source, \DOMElement $target) |
119
|
|
|
{ |
120
|
|
|
while ($source->hasChildNodes()) { |
121
|
|
|
// appendChild() actually moves the childNode |
122
|
|
|
$target->appendChild($source->childNodes->item(0)); |
123
|
|
|
} |
124
|
|
|
} |
125
|
|
|
|
126
|
|
|
private static function formatResult(\DOMDocument $domDocument, $originalHtml) |
127
|
|
|
{ |
128
|
|
|
if (!self::containsDocType($originalHtml)) { |
129
|
|
|
$domDocument->removeChild($domDocument->doctype); |
130
|
|
|
} |
131
|
|
|
|
132
|
|
|
if (self::containsHtmlTag($originalHtml)) { |
133
|
|
|
return $domDocument->saveHTML(); |
134
|
|
|
} else { |
135
|
|
|
$bodyDomElement = $domDocument->getElementsByTagName('body') |
136
|
|
|
->item(0); |
137
|
|
|
|
138
|
|
|
$html = $domDocument->saveHTML($bodyDomElement); |
139
|
|
|
|
140
|
|
|
return str_replace(array('<body>', '</body>'), '', $html); |
141
|
|
|
} |
142
|
|
|
} |
143
|
|
|
|
144
|
|
|
private static function containsDocType($html) |
145
|
|
|
{ |
146
|
|
|
return self::stringContains($html, '<!DOCTYPE'); |
147
|
|
|
} |
148
|
|
|
|
149
|
|
|
private static function stringContains($string, $needle) |
150
|
|
|
{ |
151
|
|
|
return false !== strpos($string, $needle); |
152
|
|
|
} |
153
|
|
|
|
154
|
|
|
private static function containsHtmlTag($html) |
155
|
|
|
{ |
156
|
|
|
return self::stringContains($html, '<html'); |
157
|
|
|
} |
158
|
|
|
|
159
|
|
|
private static function headingLevels($html) |
160
|
|
|
{ |
161
|
|
|
$domDocument = new \DOMDocument(); |
162
|
|
|
$domDocument->loadHTML($html); |
163
|
|
|
$headings = self::getHeadings($domDocument); |
164
|
|
|
$headingLevels = array(); |
165
|
|
|
|
166
|
|
|
foreach ($headings as $heading) { |
167
|
|
|
$headingLevels[] = self::tagNameToLevel($heading->tagName); |
168
|
|
|
} |
169
|
|
|
|
170
|
|
|
return $headingLevels; |
171
|
|
|
} |
172
|
|
|
} |
173
|
|
|
|