|
1
|
|
|
<?php |
|
2
|
|
|
namespace Bookdown\Bookdown\Process\Headings; |
|
3
|
|
|
|
|
4
|
|
|
use Psr\Log\LoggerInterface; |
|
5
|
|
|
use Bookdown\Bookdown\Content\Page; |
|
6
|
|
|
use Bookdown\Bookdown\Content\HeadingFactory; |
|
7
|
|
|
use Bookdown\Bookdown\Fsio; |
|
8
|
|
|
use Bookdown\Bookdown\Process\ProcessInterface; |
|
9
|
|
|
use DomDocument; |
|
10
|
|
|
use DomNode; |
|
11
|
|
|
use DomNodeList; |
|
12
|
|
|
use DomText; |
|
13
|
|
|
use DomXpath; |
|
14
|
|
|
|
|
15
|
|
|
class HeadingsProcess implements ProcessInterface |
|
16
|
|
|
{ |
|
17
|
|
|
protected $page; |
|
18
|
|
|
|
|
19
|
|
|
protected $html; |
|
20
|
|
|
|
|
21
|
|
|
protected $doc; |
|
22
|
|
|
|
|
23
|
|
|
protected $counts = array(); |
|
24
|
|
|
|
|
25
|
|
|
protected $headings = array(); |
|
26
|
|
|
|
|
27
|
|
|
protected $headingFactory; |
|
28
|
|
|
|
|
29
|
|
|
protected $fsio; |
|
30
|
|
|
|
|
31
|
|
|
protected $logger; |
|
32
|
|
|
|
|
33
|
14 |
|
public function __construct( |
|
34
|
|
|
LoggerInterface $logger, |
|
35
|
|
|
Fsio $fsio, |
|
36
|
|
|
HeadingFactory $headingFactory |
|
37
|
|
|
) { |
|
38
|
14 |
|
$this->logger = $logger; |
|
39
|
14 |
|
$this->fsio = $fsio; |
|
40
|
14 |
|
$this->headingFactory = $headingFactory; |
|
41
|
14 |
|
} |
|
42
|
|
|
|
|
43
|
13 |
|
public function __invoke(Page $page) |
|
44
|
|
|
{ |
|
45
|
13 |
|
$this->logger->info(" Processing headings for {$page->getTarget()}"); |
|
46
|
|
|
|
|
47
|
13 |
|
$this->reset($page); |
|
48
|
|
|
|
|
49
|
13 |
|
$this->loadHtml(); |
|
50
|
13 |
|
if ($this->html) { |
|
51
|
12 |
|
$this->loadDomDocument(); |
|
52
|
12 |
|
$this->processHeadingNodes(); |
|
53
|
12 |
|
$this->saveHtml(); |
|
54
|
12 |
|
} |
|
55
|
|
|
|
|
56
|
13 |
|
$page->setHeadings($this->headings); |
|
57
|
13 |
|
} |
|
58
|
|
|
|
|
59
|
13 |
|
protected function reset(Page $page) |
|
60
|
|
|
{ |
|
61
|
13 |
|
$this->page = $page; |
|
62
|
13 |
|
$this->html = null; |
|
63
|
13 |
|
$this->doc = null; |
|
64
|
13 |
|
$this->counts = array( |
|
65
|
13 |
|
'h2' => 0, |
|
66
|
13 |
|
'h3' => 0, |
|
67
|
13 |
|
'h4' => 0, |
|
68
|
13 |
|
'h5' => 0, |
|
69
|
13 |
|
'h6' => 0, |
|
70
|
|
|
); |
|
71
|
13 |
|
$this->headings = array(); |
|
72
|
|
|
|
|
73
|
13 |
|
if ($this->page->isIndex()) { |
|
74
|
12 |
|
$this->headings[] = $this->headingFactory->newInstance( |
|
75
|
12 |
|
$this->page->getNumber(), |
|
76
|
12 |
|
$this->page->getTitle(), |
|
77
|
12 |
|
$this->page->getHref() |
|
78
|
12 |
|
); |
|
79
|
12 |
|
} |
|
80
|
13 |
|
} |
|
81
|
|
|
|
|
82
|
13 |
|
protected function loadHtml() |
|
83
|
|
|
{ |
|
84
|
13 |
|
$this->html = $this->fsio->get($this->page->getTarget()); |
|
85
|
13 |
|
} |
|
86
|
|
|
|
|
87
|
12 |
|
protected function saveHtml() |
|
88
|
|
|
{ |
|
89
|
12 |
|
$this->fsio->put($this->page->getTarget(), $this->html); |
|
90
|
12 |
|
} |
|
91
|
|
|
|
|
92
|
12 |
|
protected function loadDomDocument() |
|
93
|
|
|
{ |
|
94
|
12 |
|
$this->doc = new DomDocument(); |
|
95
|
12 |
|
$this->doc->formatOutput = true; |
|
96
|
12 |
|
$this->doc->loadHtml(mb_convert_encoding($this->html, 'HTML-ENTITIES', 'UTF-8'), LIBXML_HTML_NODEFDTD); |
|
97
|
12 |
|
} |
|
98
|
|
|
|
|
99
|
12 |
|
protected function processHeadingNodes() |
|
100
|
|
|
{ |
|
101
|
12 |
|
$nodes = $this->getHeadingNodes(); |
|
102
|
12 |
|
$this->setPageTitle($nodes); |
|
103
|
12 |
|
$this->addHeadings($nodes); |
|
104
|
12 |
|
$this->setHtmlFromDomDocument(); |
|
105
|
12 |
|
} |
|
106
|
|
|
|
|
107
|
12 |
|
protected function getHeadingNodes() |
|
108
|
|
|
{ |
|
109
|
12 |
|
$xpath = new DomXpath($this->doc); |
|
110
|
|
|
$query = '/html/body/*[self::h1 or self::h2 or self::h3 or self::h4 ' |
|
111
|
12 |
|
. ' or self::h5 or self::h6]'; |
|
112
|
12 |
|
return $xpath->query($query); |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
12 |
|
protected function setPageTitle(DomNodeList $nodes) |
|
116
|
|
|
{ |
|
117
|
12 |
|
$node = $nodes->item(0); |
|
118
|
12 |
|
if ($node) { |
|
119
|
12 |
|
$this->page->setTitle($node->nodeValue); |
|
120
|
12 |
|
} |
|
121
|
12 |
|
} |
|
122
|
|
|
|
|
123
|
12 |
|
protected function addHeadings($nodes) |
|
124
|
|
|
{ |
|
125
|
12 |
|
foreach ($nodes as $node) { |
|
126
|
12 |
|
$this->addHeading($node); |
|
127
|
12 |
|
} |
|
128
|
12 |
|
} |
|
129
|
|
|
|
|
130
|
12 |
|
protected function addHeading(DomNode $node) |
|
131
|
|
|
{ |
|
132
|
12 |
|
$heading = $this->newHeading($node); |
|
133
|
12 |
|
$this->headings[] = $heading; |
|
134
|
|
|
|
|
135
|
12 |
|
$number = new DOMText(); |
|
136
|
12 |
|
$number->nodeValue = $heading->getNumber() . ' '; |
|
137
|
12 |
|
$node->insertBefore($number, $node->firstChild); |
|
138
|
|
|
|
|
139
|
12 |
|
$node->setAttribute('id', $heading->getAnchor()); |
|
140
|
12 |
|
} |
|
141
|
|
|
|
|
142
|
12 |
|
protected function newHeading(DomNode $node) |
|
143
|
|
|
{ |
|
144
|
|
|
// the full heading number |
|
145
|
12 |
|
$number = $this->getHeadingNumber($node); |
|
146
|
|
|
|
|
147
|
|
|
// strip the leading <hN> and the closing </hN> |
|
148
|
|
|
// this assumes the <hN> tag has no attributes |
|
149
|
12 |
|
$title = substr($node->C14N(), 4, -5); |
|
150
|
|
|
|
|
151
|
|
|
// lose the trailing dot for the ID |
|
152
|
12 |
|
$id = substr($number, 0, -1); |
|
153
|
|
|
|
|
154
|
12 |
|
return $this->headingFactory->newInstance( |
|
155
|
12 |
|
$number, |
|
156
|
12 |
|
$title, |
|
157
|
12 |
|
$this->page->getHref(), |
|
158
|
|
|
$id |
|
159
|
12 |
|
); |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
12 |
|
protected function getHeadingNumber(DomNode $node) |
|
163
|
|
|
{ |
|
164
|
12 |
|
$this->setCounts($node); |
|
165
|
12 |
|
$string = ''; |
|
166
|
12 |
|
foreach ($this->counts as $count) { |
|
167
|
12 |
|
if (! $count) { |
|
168
|
12 |
|
break; |
|
169
|
|
|
} |
|
170
|
12 |
|
$string .= "{$count}."; |
|
171
|
12 |
|
} |
|
172
|
12 |
|
return $this->page->getNumber() . $string; |
|
173
|
|
|
} |
|
174
|
|
|
|
|
175
|
12 |
|
protected function setCounts(DomNode $node) |
|
176
|
|
|
{ |
|
177
|
12 |
|
foreach ($this->counts as $level => $count) { |
|
178
|
12 |
|
if ($level == $node->nodeName) { |
|
179
|
12 |
|
$this->counts[$level] ++; |
|
180
|
12 |
|
} |
|
181
|
12 |
|
if ($level > $node->nodeName) { |
|
182
|
12 |
|
$this->counts[$level] = 0; |
|
183
|
12 |
|
} |
|
184
|
12 |
|
} |
|
185
|
12 |
|
} |
|
186
|
|
|
|
|
187
|
12 |
|
protected function setHtmlFromDomDocument() |
|
188
|
|
|
{ |
|
189
|
|
|
// retain the modified html |
|
190
|
12 |
|
$this->html = trim($this->doc->saveHtml($this->doc->documentElement)); |
|
191
|
|
|
|
|
192
|
|
|
// strip the html and body tags added by DomDocument |
|
193
|
12 |
|
$this->html = substr( |
|
194
|
12 |
|
$this->html, |
|
195
|
12 |
|
strlen('<html><body>'), |
|
196
|
12 |
|
-1 * strlen('</body></html>') |
|
197
|
12 |
|
); |
|
198
|
|
|
|
|
199
|
|
|
// still may be whitespace all about |
|
200
|
12 |
|
$this->html = trim($this->html) . PHP_EOL; |
|
201
|
12 |
|
} |
|
202
|
|
|
} |
|
203
|
|
|
|
bookdown /
Bookdown.Bookdown
