|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace App\Docsets; |
|
4
|
|
|
|
|
5
|
|
|
use Godbout\DashDocsetBuilder\Docsets\BaseDocset; |
|
6
|
|
|
use Illuminate\Support\Collection; |
|
7
|
|
|
use Illuminate\Support\Facades\Storage; |
|
8
|
|
|
use Illuminate\Support\Str; |
|
9
|
|
|
use Wa72\HtmlPageDom\HtmlPageCrawler; |
|
10
|
|
|
|
|
11
|
|
|
class Tiki extends BaseDocset |
|
12
|
|
|
{ |
|
13
|
|
|
public const CODE = 'tiki'; |
|
14
|
|
|
public const NAME = 'Tiki'; |
|
15
|
|
|
public const URL = 'doc.tiki.org'; |
|
16
|
|
|
public const INDEX = 'All-the-Documentation.html'; |
|
17
|
|
|
public const PLAYGROUND = 'https://tiki.org/Demo'; |
|
18
|
|
|
public const ICON_16 = '../../icons/icon.png'; |
|
19
|
|
|
public const ICON_32 = '../../icons/[email protected]'; |
|
20
|
|
|
public const EXTERNAL_DOMAINS = [ |
|
21
|
|
|
'themes.tiki.org', |
|
22
|
|
|
]; |
|
23
|
|
|
|
|
24
|
|
|
|
|
25
|
|
|
public function grab(): bool |
|
26
|
|
|
{ |
|
27
|
|
|
$toIgnore = implode('|', [ |
|
28
|
|
|
'\?refresh', |
|
29
|
|
|
'\?session_filters', |
|
30
|
|
|
'\?sort_mode', |
|
31
|
|
|
'/Plugins-', |
|
32
|
|
|
'comzone=', |
|
33
|
|
|
'cookietab=', |
|
34
|
|
|
'fullscreen=', |
|
35
|
|
|
'offset=', |
|
36
|
|
|
'todate=', |
|
37
|
|
|
'viewmode=', |
|
38
|
|
|
'PDF\.js', |
|
39
|
|
|
'Plugins(\?structure=.*)?$', |
|
40
|
|
|
'tikiversion=', |
|
41
|
|
|
'wp_files_sort_mode[0-9]=', |
|
42
|
|
|
]); |
|
43
|
|
|
|
|
44
|
|
|
$toGet = implode('|', [ |
|
45
|
|
|
'\.css', |
|
46
|
|
|
'\.gif', |
|
47
|
|
|
'\.ico', |
|
48
|
|
|
'\.jpg', |
|
49
|
|
|
'\.js', |
|
50
|
|
|
'\.png', |
|
51
|
|
|
'\.svg', |
|
52
|
|
|
'\.webmanifest', |
|
53
|
|
|
'/display', |
|
54
|
|
|
'/LIST', |
|
55
|
|
|
'/Module-', |
|
56
|
|
|
'/Plugin[^-]', |
|
57
|
|
|
'-Field(s)?(\?structure=HomePage\+UserGuide)?$', |
|
58
|
|
|
'Tiki_org_family', |
|
59
|
|
|
'[^:=]Wiki-Syntax' |
|
60
|
|
|
]); |
|
61
|
|
|
|
|
62
|
|
|
system( |
|
63
|
|
|
"echo; wget doc.tiki.org/All-the-Documentation \ |
|
64
|
|
|
--mirror \ |
|
65
|
|
|
--trust-server-names \ |
|
66
|
|
|
--header 'Cookie: javascript_enabled_detect=true' \ |
|
67
|
|
|
--reject-regex='{$toIgnore}' \ |
|
68
|
|
|
--accept-regex='{$toGet}' \ |
|
69
|
|
|
--ignore-case \ |
|
70
|
|
|
--page-requisites \ |
|
71
|
|
|
--adjust-extension \ |
|
72
|
|
|
--convert-links \ |
|
73
|
|
|
--span-hosts \ |
|
74
|
|
|
--domains={$this->externalDomains()} \ |
|
75
|
|
|
--directory-prefix=storage/{$this->downloadedDirectory()} \ |
|
76
|
|
|
-e robots=off \ |
|
77
|
|
|
--quiet \ |
|
78
|
|
|
--show-progress", |
|
79
|
|
|
$result |
|
80
|
|
|
); |
|
81
|
|
|
|
|
82
|
|
|
return $result === 0; |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
12 |
|
public function entries(string $file): Collection |
|
86
|
|
|
{ |
|
87
|
12 |
|
$crawler = HtmlPageCrawler::create(Storage::get($file)); |
|
88
|
|
|
|
|
89
|
12 |
|
$entries = collect(); |
|
90
|
12 |
|
$entries = $entries->merge($this->pluginEntries($crawler, $file)); |
|
91
|
12 |
|
$entries = $entries->merge($this->moduleEntries($crawler, $file)); |
|
92
|
12 |
|
$entries = $entries->merge($this->fieldEntries($crawler, $file)); |
|
93
|
12 |
|
$entries = $entries->merge($this->styleEntries($crawler, $file)); |
|
94
|
|
|
|
|
95
|
12 |
|
return $entries; |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
12 |
|
protected function pluginEntries(HtmlPageCrawler $crawler, string $file) |
|
99
|
|
|
{ |
|
100
|
12 |
|
$entries = collect(); |
|
101
|
|
|
|
|
102
|
12 |
|
if (preg_match('/Plugin/i', $file)) { |
|
103
|
12 |
|
$path = $crawler->filter('link[rel=canonical]')->attr('href'); |
|
104
|
|
|
|
|
105
|
|
|
$crawler->filter('#page-data > h1:first-of-type')->each(function (HtmlPageCrawler $node) use ($entries, $file, $path) { |
|
106
|
12 |
|
$entries->push([ |
|
107
|
12 |
|
'name' => $node->text(), |
|
108
|
12 |
|
'type' => 'Plugin', |
|
109
|
12 |
|
'path' => Str::after($file . '#' . Str::slug($path), $this->innerDirectory()), |
|
110
|
|
|
]); |
|
111
|
12 |
|
}); |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
12 |
|
return $entries; |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
12 |
|
protected function moduleEntries(HtmlPageCrawler $crawler, string $file) |
|
118
|
|
|
{ |
|
119
|
12 |
|
$entries = collect(); |
|
120
|
|
|
|
|
121
|
12 |
|
if (preg_match('/Module/i', $file)) { |
|
122
|
6 |
|
$path = $crawler->filter('link[rel=canonical]')->attr('href'); |
|
123
|
|
|
|
|
124
|
|
|
$crawler->filter('#page-data > h1:first-of-type')->each(function (HtmlPageCrawler $node) use ($entries, $file, $path) { |
|
125
|
6 |
|
$entries->push([ |
|
126
|
6 |
|
'name' => $node->text(), |
|
127
|
6 |
|
'type' => 'Module', |
|
128
|
6 |
|
'path' => Str::after($file . '#' . Str::slug($path), $this->innerDirectory()), |
|
129
|
|
|
]); |
|
130
|
6 |
|
}); |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
12 |
|
return $entries; |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
12 |
|
protected function fieldEntries(HtmlPageCrawler $crawler, string $file) |
|
137
|
|
|
{ |
|
138
|
12 |
|
$entries = collect(); |
|
139
|
|
|
|
|
140
|
12 |
|
if (preg_match('/Tracker-Field/i', $file)) { |
|
141
|
6 |
|
$path = $crawler->filter('link[rel=canonical]')->attr('href'); |
|
142
|
|
|
|
|
143
|
|
|
$crawler->filter('#page-data > h1:first-of-type')->each(function (HtmlPageCrawler $node) use ($entries, $file, $path) { |
|
144
|
6 |
|
$entries->push([ |
|
145
|
6 |
|
'name' => $node->text(), |
|
146
|
6 |
|
'type' => 'Field', |
|
147
|
6 |
|
'path' => Str::after($file . '#' . Str::slug($path), $this->innerDirectory()), |
|
148
|
|
|
]); |
|
149
|
6 |
|
}); |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
12 |
|
return $entries; |
|
153
|
|
|
} |
|
154
|
|
|
|
|
155
|
12 |
|
protected function styleEntries(HtmlPageCrawler $crawler, string $file) |
|
156
|
|
|
{ |
|
157
|
12 |
|
$entries = collect(); |
|
158
|
|
|
|
|
159
|
12 |
|
if (preg_match('/Wiki-Syntax/i', $file)) { |
|
160
|
6 |
|
$path = $crawler->filter('link[rel=canonical]')->attr('href'); |
|
161
|
|
|
|
|
162
|
|
|
$crawler->filter('#page-data > h1:first-of-type')->each(function (HtmlPageCrawler $node) use ($entries, $file, $path) { |
|
163
|
6 |
|
$entries->push([ |
|
164
|
6 |
|
'name' => $node->text(), |
|
165
|
6 |
|
'type' => 'Style', |
|
166
|
6 |
|
'path' => Str::after($file . '#' . Str::slug($path), $this->innerDirectory()), |
|
167
|
|
|
]); |
|
168
|
6 |
|
}); |
|
169
|
|
|
} |
|
170
|
|
|
|
|
171
|
12 |
|
return $entries; |
|
172
|
|
|
} |
|
173
|
|
|
|
|
174
|
12 |
|
public function format(string $file): string |
|
175
|
|
|
{ |
|
176
|
12 |
|
$crawler = HtmlPageCrawler::create(Storage::get($file)); |
|
177
|
|
|
|
|
178
|
12 |
|
$this->removeNavbar($crawler); |
|
179
|
12 |
|
$this->removeLeftSidebarButton($crawler); |
|
180
|
12 |
|
$this->removeFullscreenButton($crawler); |
|
181
|
12 |
|
$this->removePageTopModules($crawler); |
|
182
|
12 |
|
$this->removeWikiActionsWrapper($crawler); |
|
183
|
12 |
|
$this->removeBreadcrumb($crawler); |
|
184
|
12 |
|
$this->removeTopbar($crawler); |
|
185
|
12 |
|
$this->removeLeftSidebar($crawler); |
|
186
|
12 |
|
$this->removeRightSidebar($crawler); |
|
187
|
12 |
|
$this->removePagebar($crawler); |
|
188
|
12 |
|
$this->removeFooter($crawler); |
|
189
|
12 |
|
$this->removeUnwantedJavaScript($crawler); |
|
190
|
|
|
|
|
191
|
12 |
|
$this->updateCss($crawler); |
|
192
|
|
|
|
|
193
|
12 |
|
$this->insertOnlineRedirection($crawler); |
|
194
|
12 |
|
$this->insertDashTableOfContents($crawler); |
|
195
|
|
|
|
|
196
|
12 |
|
return $crawler->saveHTML(); |
|
197
|
|
|
} |
|
198
|
|
|
|
|
199
|
12 |
|
protected function removeNavbar(HtmlPageCrawler $crawler) |
|
200
|
|
|
{ |
|
201
|
12 |
|
$crawler->filter('nav.navbar')->remove(); |
|
202
|
12 |
|
} |
|
203
|
|
|
|
|
204
|
12 |
|
protected function removeLeftSidebarButton(HtmlPageCrawler $crawler) |
|
205
|
|
|
{ |
|
206
|
12 |
|
$crawler->filter('#row-middle > div.side-col-toggle-container')->remove(); |
|
207
|
12 |
|
} |
|
208
|
|
|
|
|
209
|
12 |
|
protected function removeFullscreenButton(HtmlPageCrawler $crawler) |
|
210
|
|
|
{ |
|
211
|
12 |
|
$crawler->filter('#fullscreenbutton')->remove(); |
|
212
|
12 |
|
} |
|
213
|
|
|
|
|
214
|
12 |
|
protected function removePageTopModules(HtmlPageCrawler $crawler) |
|
215
|
|
|
{ |
|
216
|
12 |
|
$crawler->filter('#pagetop_modules')->remove(); |
|
217
|
12 |
|
} |
|
218
|
|
|
|
|
219
|
12 |
|
protected function removeWikiActionsWrapper(HtmlPageCrawler $crawler) |
|
220
|
|
|
{ |
|
221
|
12 |
|
$crawler->filter('#col1 > div.wikiactions_wrapper')->remove(); |
|
222
|
12 |
|
} |
|
223
|
|
|
|
|
224
|
12 |
|
protected function removeBreadcrumb(HtmlPageCrawler $crawler) |
|
225
|
|
|
{ |
|
226
|
12 |
|
$crawler->filter('nav.nav-breadcrumb')->remove(); |
|
227
|
12 |
|
} |
|
228
|
|
|
|
|
229
|
12 |
|
protected function removeTopbar(HtmlPageCrawler $crawler) |
|
230
|
|
|
{ |
|
231
|
12 |
|
$crawler->filter('#topbar')->remove(); |
|
232
|
12 |
|
} |
|
233
|
|
|
|
|
234
|
12 |
|
protected function removeLeftSidebar(HtmlPageCrawler $crawler) |
|
235
|
|
|
{ |
|
236
|
12 |
|
$crawler->filter('#col2')->remove(); |
|
237
|
12 |
|
} |
|
238
|
|
|
|
|
239
|
12 |
|
protected function removeRightSidebar(HtmlPageCrawler $crawler) |
|
240
|
|
|
{ |
|
241
|
12 |
|
$crawler->filter('script[src*="autoToc.js"]')->remove(); |
|
242
|
12 |
|
} |
|
243
|
|
|
|
|
244
|
12 |
|
protected function removePagebar(HtmlPageCrawler $crawler) |
|
245
|
|
|
{ |
|
246
|
12 |
|
$crawler->filter('#page-bar')->remove(); |
|
247
|
12 |
|
} |
|
248
|
|
|
|
|
249
|
12 |
|
protected function removeFooter(HtmlPageCrawler $crawler) |
|
250
|
|
|
{ |
|
251
|
12 |
|
$crawler->filter('#footer')->remove(); |
|
252
|
12 |
|
} |
|
253
|
|
|
|
|
254
|
12 |
|
protected function removeUnwantedJavaScript(HtmlPageCrawler $crawler) |
|
255
|
|
|
{ |
|
256
|
12 |
|
$crawler->filter('script[src*=autosave]')->remove(); |
|
257
|
12 |
|
$crawler->filter('script[src*=gtag]')->remove(); |
|
258
|
12 |
|
$crawler->filter('noscript')->remove(); |
|
259
|
12 |
|
$crawler->filterXPath("//script[text()[contains(.,'piwik.tiki.org')]]")->remove(); |
|
260
|
12 |
|
$crawler->filterXPath("//script[text()[contains(.,'gtag')]]")->remove(); |
|
261
|
12 |
|
} |
|
262
|
|
|
|
|
263
|
12 |
|
protected function updateCSS(HtmlPageCrawler $crawler) |
|
264
|
|
|
{ |
|
265
|
12 |
|
$this->updateTopPadding($crawler); |
|
266
|
12 |
|
$this->updateArticlePadding($crawler); |
|
267
|
12 |
|
} |
|
268
|
|
|
|
|
269
|
12 |
|
protected function updateTopPadding(HtmlPageCrawler $crawler) |
|
270
|
|
|
{ |
|
271
|
12 |
|
$crawler->filter('body') |
|
272
|
12 |
|
->removeClass('navbar-padding') |
|
273
|
12 |
|
->addClass('hide_zone_left') |
|
274
|
12 |
|
->css('padding-top', '0') |
|
275
|
|
|
; |
|
276
|
12 |
|
} |
|
277
|
|
|
|
|
278
|
12 |
|
protected function updateArticlePadding(HtmlPageCrawler $crawler) |
|
279
|
|
|
{ |
|
280
|
12 |
|
$crawler->filter('article#top') |
|
281
|
12 |
|
->css('padding-top', '44px') |
|
282
|
|
|
; |
|
283
|
12 |
|
} |
|
284
|
|
|
|
|
285
|
12 |
|
protected function insertOnlineRedirection(HtmlPageCrawler $crawler) |
|
286
|
|
|
{ |
|
287
|
12 |
|
$onlineUrl = ''; |
|
288
|
12 |
|
$meta = $crawler->filter('meta[property="og:url"]'); |
|
289
|
|
|
|
|
290
|
12 |
|
if ($meta->getDOMDocument()) { |
|
291
|
12 |
|
$onlineUrl = $meta->attr('content'); |
|
292
|
|
|
} |
|
293
|
|
|
|
|
294
|
12 |
|
$crawler->filter('html')->prepend("<!-- Online page at $onlineUrl -->"); |
|
295
|
12 |
|
} |
|
296
|
|
|
|
|
297
|
12 |
|
protected function insertDashTableOfContents(HtmlPageCrawler $crawler) |
|
298
|
|
|
{ |
|
299
|
12 |
|
$crawler->filter('#page-data > h1:first-of-type') |
|
300
|
12 |
|
->before('<a name="//apple_ref/cpp/Section/Top" class="dashAnchor"></a>'); |
|
301
|
|
|
|
|
302
|
|
|
$crawler->filter('h2')->each(static function (HtmlPageCrawler $node) { |
|
303
|
12 |
|
$node->before( |
|
304
|
12 |
|
'<a id="' . Str::slug($node->text()) . '" name="//apple_ref/cpp/Section/' . rawurlencode($node->text()) . '" class="dashAnchor"></a>' |
|
305
|
|
|
); |
|
306
|
12 |
|
}); |
|
307
|
12 |
|
} |
|
308
|
|
|
} |
|
309
|
|
|
|