|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace App\Docsets; |
|
4
|
|
|
|
|
5
|
|
|
use Illuminate\Support\Str; |
|
6
|
|
|
use Illuminate\Support\Collection; |
|
7
|
|
|
use Wa72\HtmlPageDom\HtmlPageCrawler; |
|
8
|
|
|
use Illuminate\Support\Facades\Storage; |
|
9
|
|
|
|
|
10
|
|
|
class Tiki extends BaseDocset |
|
11
|
|
|
{ |
|
12
|
|
|
public const CODE = 'tiki'; |
|
13
|
|
|
public const NAME = 'Tiki'; |
|
14
|
|
|
public const URL = 'doc.tiki.org'; |
|
15
|
|
|
public const INDEX = 'All-the-Documentation.html'; |
|
16
|
|
|
public const PLAYGROUND = ''; |
|
17
|
|
|
public const ICON_16 = '../../icons/icon.png'; |
|
18
|
|
|
public const ICON_32 = '../../icons/[email protected]'; |
|
19
|
|
|
public const EXTERNAL_DOMAINS = [ |
|
20
|
|
|
'themes.tiki.org', |
|
21
|
|
|
]; |
|
22
|
|
|
|
|
23
|
|
|
|
|
24
|
|
|
public function grab(): bool |
|
25
|
|
|
{ |
|
26
|
|
|
system( |
|
27
|
|
|
"wget doc.tiki.org/All-the-Documentation \ |
|
28
|
|
|
--mirror \ |
|
29
|
|
|
-e robots=off \ |
|
30
|
|
|
--header 'Cookie: javascript_enabled_detect=true' \ |
|
31
|
|
|
--reject-regex='/Plugins-|Plugins\.html|fullscreen=|PDF\.js|tikiversion=|comzone=|structure=|wp_files_sort_mode[0-9]=|offset=|\?refresh|\?session_filters|\?sort_mode' \ |
|
32
|
|
|
--accept-regex='/Plugin|/LIST|Tiki_org_family|\.css|\.js|\.jpg|\.png|\.gif|\.svg|\.ico|\.webmanifest' \ |
|
33
|
|
|
--page-requisites \ |
|
34
|
|
|
--adjust-extension \ |
|
35
|
|
|
--convert-links \ |
|
36
|
|
|
--span-hosts \ |
|
37
|
|
|
--domains={$this->externalDomains()} \ |
|
38
|
|
|
--directory-prefix=storage/{$this->downloadedDirectory()}", |
|
39
|
|
|
$result |
|
40
|
|
|
); |
|
41
|
|
|
|
|
42
|
|
|
return $result === 0; |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
1 |
|
public function entries(string $file): Collection |
|
46
|
|
|
{ |
|
47
|
1 |
|
$crawler = HtmlPageCrawler::create(Storage::get($file)); |
|
48
|
|
|
|
|
49
|
1 |
|
$entries = collect(); |
|
50
|
1 |
|
$entries = $entries->merge($this->pluginEntries($crawler, $file)); |
|
51
|
|
|
// $entries = $entries->merge($this->sectionEntries($crawler, $file)); |
|
52
|
|
|
|
|
53
|
1 |
|
return $entries; |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
1 |
|
protected function pluginEntries(HtmlPageCrawler $crawler, string $file) |
|
57
|
|
|
{ |
|
58
|
1 |
|
$entries = collect(); |
|
59
|
|
|
|
|
60
|
1 |
|
if (! in_array(basename($file), ['All-the-Documentation.html', 'site.webmanifest.html'])) { |
|
61
|
1 |
|
$path = $crawler->filter('link[rel=canonical]')->attr('href'); |
|
62
|
|
|
|
|
63
|
|
|
$crawler->filter('#top h1:first-of-type')->each(function (HtmlPageCrawler $node) use ($entries, $file, $path) { |
|
64
|
1 |
|
$entries->push([ |
|
65
|
1 |
|
'name' => $node->text(), |
|
66
|
1 |
|
'type' => 'Plugin', |
|
67
|
1 |
|
'path' => Str::after($file . '#' . Str::slug($path), $this->innerDirectory()), |
|
68
|
|
|
]); |
|
69
|
1 |
|
}); |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
1 |
|
return $entries; |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
protected function sectionEntries(HtmlPageCrawler $crawler, string $file) |
|
76
|
|
|
{ |
|
77
|
|
|
$entries = collect(); |
|
78
|
|
|
|
|
79
|
|
|
$crawler->filter('h2')->each(function (HtmlPageCrawler $node) use ($entries, $file) { |
|
80
|
|
|
if (! in_array(basename($file), ['All-the-Documentation.html', 'site.webmanifest.html'])) { |
|
81
|
|
|
$entries->push([ |
|
82
|
|
|
'name' => trim($node->text()), |
|
83
|
|
|
'type' => 'Section', |
|
84
|
|
|
'path' => Str::after($file . '#' . Str::slug($node->text()), $this->innerDirectory()), |
|
85
|
|
|
]); |
|
86
|
|
|
} |
|
87
|
|
|
}); |
|
88
|
|
|
|
|
89
|
|
|
return $entries; |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
1 |
|
public function format(string $html): string |
|
93
|
|
|
{ |
|
94
|
1 |
|
$crawler = HtmlPageCrawler::create($html); |
|
95
|
|
|
|
|
96
|
1 |
|
$this->removeNavbar($crawler); |
|
97
|
1 |
|
$this->removeLeftSidebarButton($crawler); |
|
98
|
1 |
|
$this->removeFullscreenButton($crawler); |
|
99
|
1 |
|
$this->removePageTopModules($crawler); |
|
100
|
1 |
|
$this->removeWikiActionsWrapper($crawler); |
|
101
|
1 |
|
$this->removeBreadcrumb($crawler); |
|
102
|
1 |
|
$this->removeTopbar($crawler); |
|
103
|
1 |
|
$this->removeLeftSidebar($crawler); |
|
104
|
1 |
|
$this->removeRightSidebar($crawler); |
|
105
|
1 |
|
$this->removePagebar($crawler); |
|
106
|
1 |
|
$this->removeFooter($crawler); |
|
107
|
1 |
|
$this->removeUnwantedJavaScript($crawler); |
|
108
|
|
|
|
|
109
|
1 |
|
$this->updateCss($crawler); |
|
110
|
|
|
|
|
111
|
1 |
|
$this->insertDashTableOfContents($crawler); |
|
112
|
|
|
|
|
113
|
1 |
|
return $crawler->saveHTML(); |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
1 |
|
protected function removeNavbar(HtmlPageCrawler $crawler) |
|
117
|
|
|
{ |
|
118
|
1 |
|
$crawler->filter('nav.navbar')->remove(); |
|
119
|
1 |
|
} |
|
120
|
|
|
|
|
121
|
1 |
|
protected function removeLeftSidebarButton(HtmlPageCrawler $crawler) |
|
122
|
|
|
{ |
|
123
|
1 |
|
$crawler->filter('#row-middle > div.side-col-toggle-container')->remove(); |
|
124
|
1 |
|
} |
|
125
|
|
|
|
|
126
|
1 |
|
protected function removeFullscreenButton(HtmlPageCrawler $crawler) |
|
127
|
|
|
{ |
|
128
|
1 |
|
$crawler->filter('#fullscreenbutton')->remove(); |
|
129
|
1 |
|
} |
|
130
|
|
|
|
|
131
|
1 |
|
protected function removePageTopModules(HtmlPageCrawler $crawler) |
|
132
|
|
|
{ |
|
133
|
1 |
|
$crawler->filter('#pagetop_modules')->remove(); |
|
134
|
1 |
|
} |
|
135
|
|
|
|
|
136
|
1 |
|
protected function removeWikiActionsWrapper(HtmlPageCrawler $crawler) |
|
137
|
|
|
{ |
|
138
|
1 |
|
$crawler->filter('#col1 > div.wikiactions_wrapper')->remove(); |
|
139
|
1 |
|
} |
|
140
|
|
|
|
|
141
|
1 |
|
protected function removeBreadcrumb(HtmlPageCrawler $crawler) |
|
142
|
|
|
{ |
|
143
|
1 |
|
$crawler->filter('nav.nav-breadcrumb')->remove(); |
|
144
|
1 |
|
} |
|
145
|
|
|
|
|
146
|
1 |
|
protected function removeTopbar(HtmlPageCrawler $crawler) |
|
147
|
|
|
{ |
|
148
|
1 |
|
$crawler->filter('#topbar')->remove(); |
|
149
|
1 |
|
} |
|
150
|
|
|
|
|
151
|
1 |
|
protected function removeLeftSidebar(HtmlPageCrawler $crawler) |
|
152
|
|
|
{ |
|
153
|
1 |
|
$crawler->filter('#col2')->remove(); |
|
154
|
1 |
|
} |
|
155
|
|
|
|
|
156
|
1 |
|
protected function removeRightSidebar(HtmlPageCrawler $crawler) |
|
157
|
|
|
{ |
|
158
|
1 |
|
$crawler->filter('script[src*="autoToc.js"]')->remove(); |
|
159
|
1 |
|
} |
|
160
|
|
|
|
|
161
|
1 |
|
protected function removePagebar(HtmlPageCrawler $crawler) |
|
162
|
|
|
{ |
|
163
|
1 |
|
$crawler->filter('#page-bar')->remove(); |
|
164
|
1 |
|
} |
|
165
|
|
|
|
|
166
|
1 |
|
protected function removeFooter(HtmlPageCrawler $crawler) |
|
167
|
|
|
{ |
|
168
|
1 |
|
$crawler->filter('#footer')->remove(); |
|
169
|
1 |
|
} |
|
170
|
|
|
|
|
171
|
1 |
|
protected function removeUnwantedJavaScript(HtmlPageCrawler $crawler) |
|
172
|
|
|
{ |
|
173
|
1 |
|
$crawler->filter('script[src*=autosave]')->remove(); |
|
174
|
1 |
|
$crawler->filter('script[src*=gtag]')->remove(); |
|
175
|
1 |
|
$crawler->filter('noscript')->remove(); |
|
176
|
1 |
|
$crawler->filterXPath("//script[text()[contains(.,'piwik.tiki.org')]]")->remove(); |
|
177
|
1 |
|
$crawler->filterXPath("//script[text()[contains(.,'gtag')]]")->remove(); |
|
178
|
1 |
|
} |
|
179
|
|
|
|
|
180
|
1 |
|
protected function updateCSS(HtmlPageCrawler $crawler) |
|
181
|
|
|
{ |
|
182
|
1 |
|
$this->updateTopPadding($crawler); |
|
183
|
1 |
|
$this->updateArticlePadding($crawler); |
|
184
|
1 |
|
} |
|
185
|
|
|
|
|
186
|
1 |
|
protected function updateTopPadding(HtmlPageCrawler $crawler) |
|
187
|
|
|
{ |
|
188
|
1 |
|
$crawler->filter('body') |
|
189
|
1 |
|
->removeClass('navbar-padding') |
|
190
|
1 |
|
->addClass('hide_zone_left') |
|
191
|
1 |
|
->css('padding-top', '0') |
|
192
|
|
|
; |
|
193
|
1 |
|
} |
|
194
|
|
|
|
|
195
|
1 |
|
protected function updateArticlePadding(HtmlPageCrawler $crawler) |
|
196
|
|
|
{ |
|
197
|
1 |
|
$crawler->filter('article#top') |
|
198
|
1 |
|
->css('padding-top', '44px') |
|
199
|
|
|
; |
|
200
|
1 |
|
} |
|
201
|
|
|
|
|
202
|
1 |
|
protected function insertDashTableOfContents(HtmlPageCrawler $crawler) |
|
203
|
|
|
{ |
|
204
|
1 |
|
$crawler->filter('h1') |
|
205
|
1 |
|
->before('<a name="//apple_ref/cpp/Section/Top" class="dashAnchor"></a>'); |
|
206
|
|
|
|
|
207
|
|
|
$crawler->filter('h2')->each(static function (HtmlPageCrawler $node) { |
|
208
|
1 |
|
$node->before( |
|
209
|
1 |
|
'<a id="' . Str::slug($node->text()) . '" name="//apple_ref/cpp/Section/' . rawurlencode($node->text()) . '" class="dashAnchor"></a>' |
|
210
|
|
|
); |
|
211
|
1 |
|
}); |
|
212
|
1 |
|
} |
|
213
|
|
|
} |
|
214
|
|
|
|
godbout /
dash-docset-builder
