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 = 'https://tiki.org/Demo'; |
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
|
|
|
$toIgnore = implode('|', [ |
27
|
|
|
'\?refresh', |
28
|
|
|
'\?session_filters', |
29
|
|
|
'\?sort_mode', |
30
|
|
|
'/Plugins-', |
31
|
|
|
'comzone=', |
32
|
|
|
'cookietab=', |
33
|
|
|
'fullscreen=', |
34
|
|
|
'offset=', |
35
|
|
|
'todate=', |
36
|
|
|
'viewmode=', |
37
|
|
|
'PDF\.js', |
38
|
|
|
'Plugins$', |
39
|
|
|
'structure=', |
40
|
|
|
'tikiversion=', |
41
|
|
|
'wp_files_sort_mode[0-9]=', |
42
|
|
|
]); |
43
|
|
|
|
44
|
|
|
$toGet = implode('|', [ |
45
|
|
|
'\.css', |
46
|
|
|
'\.gif', |
47
|
|
|
'\.ico', |
48
|
|
|
'\.jpg', |
49
|
|
|
'\.js', |
50
|
|
|
'\.png', |
51
|
|
|
'\.svg', |
52
|
|
|
'\.webmanifest', |
53
|
|
|
'/display', |
54
|
|
|
'/LIST', |
55
|
|
|
'/Module-', |
56
|
|
|
'/Plugin[^-]', |
57
|
|
|
'-Field$', |
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
|
2 |
|
public function entries(string $file): Collection |
86
|
|
|
{ |
87
|
2 |
|
$crawler = HtmlPageCrawler::create(Storage::get($file)); |
88
|
|
|
|
89
|
2 |
|
$entries = collect(); |
90
|
2 |
|
$entries = $entries->merge($this->pluginEntries($crawler, $file)); |
91
|
2 |
|
$entries = $entries->merge($this->moduleEntries($crawler, $file)); |
92
|
2 |
|
$entries = $entries->merge($this->fieldEntries($crawler, $file)); |
93
|
2 |
|
$entries = $entries->merge($this->styleEntries($crawler, $file)); |
94
|
|
|
|
95
|
2 |
|
return $entries; |
96
|
|
|
} |
97
|
|
|
|
98
|
2 |
|
protected function pluginEntries(HtmlPageCrawler $crawler, string $file) |
99
|
|
|
{ |
100
|
2 |
|
$entries = collect(); |
101
|
|
|
|
102
|
2 |
|
if (preg_match('/Plugin/i', $file)) { |
103
|
2 |
|
$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
|
2 |
|
$entries->push([ |
107
|
2 |
|
'name' => $node->text(), |
108
|
2 |
|
'type' => 'Plugin', |
109
|
2 |
|
'path' => Str::after($file . '#' . Str::slug($path), $this->innerDirectory()), |
110
|
|
|
]); |
111
|
2 |
|
}); |
112
|
|
|
} |
113
|
|
|
|
114
|
2 |
|
return $entries; |
115
|
|
|
} |
116
|
|
|
|
117
|
2 |
|
protected function moduleEntries(HtmlPageCrawler $crawler, string $file) |
118
|
|
|
{ |
119
|
2 |
|
$entries = collect(); |
120
|
|
|
|
121
|
2 |
|
if (preg_match('/Module/i', $file)) { |
122
|
1 |
|
$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
|
1 |
|
$entries->push([ |
126
|
1 |
|
'name' => $node->text(), |
127
|
1 |
|
'type' => 'Module', |
128
|
1 |
|
'path' => Str::after($file . '#' . Str::slug($path), $this->innerDirectory()), |
129
|
|
|
]); |
130
|
1 |
|
}); |
131
|
|
|
} |
132
|
|
|
|
133
|
2 |
|
return $entries; |
134
|
|
|
} |
135
|
|
|
|
136
|
2 |
|
protected function fieldEntries(HtmlPageCrawler $crawler, string $file) |
137
|
|
|
{ |
138
|
2 |
|
$entries = collect(); |
139
|
|
|
|
140
|
2 |
|
if (preg_match('/Tracker-Field/i', $file)) { |
141
|
1 |
|
$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
|
1 |
|
$entries->push([ |
145
|
1 |
|
'name' => $node->text(), |
146
|
1 |
|
'type' => 'Field', |
147
|
1 |
|
'path' => Str::after($file . '#' . Str::slug($path), $this->innerDirectory()), |
148
|
|
|
]); |
149
|
1 |
|
}); |
150
|
|
|
} |
151
|
|
|
|
152
|
2 |
|
return $entries; |
153
|
|
|
} |
154
|
|
|
|
155
|
2 |
|
protected function styleEntries(HtmlPageCrawler $crawler, string $file) |
156
|
|
|
{ |
157
|
2 |
|
$entries = collect(); |
158
|
|
|
|
159
|
2 |
|
if (preg_match('/Wiki-Syntax/i', $file)) { |
160
|
1 |
|
$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
|
1 |
|
$entries->push([ |
164
|
1 |
|
'name' => $node->text(), |
165
|
1 |
|
'type' => 'Style', |
166
|
1 |
|
'path' => Str::after($file . '#' . Str::slug($path), $this->innerDirectory()), |
167
|
|
|
]); |
168
|
1 |
|
}); |
169
|
|
|
} |
170
|
|
|
|
171
|
2 |
|
return $entries; |
172
|
|
|
} |
173
|
|
|
|
174
|
2 |
|
public function format(string $html): string |
175
|
|
|
{ |
176
|
2 |
|
$crawler = HtmlPageCrawler::create($html); |
177
|
|
|
|
178
|
2 |
|
$this->removeNavbar($crawler); |
179
|
2 |
|
$this->removeLeftSidebarButton($crawler); |
180
|
2 |
|
$this->removeFullscreenButton($crawler); |
181
|
2 |
|
$this->removePageTopModules($crawler); |
182
|
2 |
|
$this->removeWikiActionsWrapper($crawler); |
183
|
2 |
|
$this->removeBreadcrumb($crawler); |
184
|
2 |
|
$this->removeTopbar($crawler); |
185
|
2 |
|
$this->removeLeftSidebar($crawler); |
186
|
2 |
|
$this->removeRightSidebar($crawler); |
187
|
2 |
|
$this->removePagebar($crawler); |
188
|
2 |
|
$this->removeFooter($crawler); |
189
|
2 |
|
$this->removeUnwantedJavaScript($crawler); |
190
|
|
|
|
191
|
2 |
|
$this->updateCss($crawler); |
192
|
|
|
|
193
|
2 |
|
$this->insertOnlineRedirection($crawler); |
194
|
2 |
|
$this->insertDashTableOfContents($crawler); |
195
|
|
|
|
196
|
2 |
|
return $crawler->saveHTML(); |
197
|
|
|
} |
198
|
|
|
|
199
|
2 |
|
protected function removeNavbar(HtmlPageCrawler $crawler) |
200
|
|
|
{ |
201
|
2 |
|
$crawler->filter('nav.navbar')->remove(); |
202
|
2 |
|
} |
203
|
|
|
|
204
|
2 |
|
protected function removeLeftSidebarButton(HtmlPageCrawler $crawler) |
205
|
|
|
{ |
206
|
2 |
|
$crawler->filter('#row-middle > div.side-col-toggle-container')->remove(); |
207
|
2 |
|
} |
208
|
|
|
|
209
|
2 |
|
protected function removeFullscreenButton(HtmlPageCrawler $crawler) |
210
|
|
|
{ |
211
|
2 |
|
$crawler->filter('#fullscreenbutton')->remove(); |
212
|
2 |
|
} |
213
|
|
|
|
214
|
2 |
|
protected function removePageTopModules(HtmlPageCrawler $crawler) |
215
|
|
|
{ |
216
|
2 |
|
$crawler->filter('#pagetop_modules')->remove(); |
217
|
2 |
|
} |
218
|
|
|
|
219
|
2 |
|
protected function removeWikiActionsWrapper(HtmlPageCrawler $crawler) |
220
|
|
|
{ |
221
|
2 |
|
$crawler->filter('#col1 > div.wikiactions_wrapper')->remove(); |
222
|
2 |
|
} |
223
|
|
|
|
224
|
2 |
|
protected function removeBreadcrumb(HtmlPageCrawler $crawler) |
225
|
|
|
{ |
226
|
2 |
|
$crawler->filter('nav.nav-breadcrumb')->remove(); |
227
|
2 |
|
} |
228
|
|
|
|
229
|
2 |
|
protected function removeTopbar(HtmlPageCrawler $crawler) |
230
|
|
|
{ |
231
|
2 |
|
$crawler->filter('#topbar')->remove(); |
232
|
2 |
|
} |
233
|
|
|
|
234
|
2 |
|
protected function removeLeftSidebar(HtmlPageCrawler $crawler) |
235
|
|
|
{ |
236
|
2 |
|
$crawler->filter('#col2')->remove(); |
237
|
2 |
|
} |
238
|
|
|
|
239
|
2 |
|
protected function removeRightSidebar(HtmlPageCrawler $crawler) |
240
|
|
|
{ |
241
|
2 |
|
$crawler->filter('script[src*="autoToc.js"]')->remove(); |
242
|
2 |
|
} |
243
|
|
|
|
244
|
2 |
|
protected function removePagebar(HtmlPageCrawler $crawler) |
245
|
|
|
{ |
246
|
2 |
|
$crawler->filter('#page-bar')->remove(); |
247
|
2 |
|
} |
248
|
|
|
|
249
|
2 |
|
protected function removeFooter(HtmlPageCrawler $crawler) |
250
|
|
|
{ |
251
|
2 |
|
$crawler->filter('#footer')->remove(); |
252
|
2 |
|
} |
253
|
|
|
|
254
|
2 |
|
protected function removeUnwantedJavaScript(HtmlPageCrawler $crawler) |
255
|
|
|
{ |
256
|
2 |
|
$crawler->filter('script[src*=autosave]')->remove(); |
257
|
2 |
|
$crawler->filter('script[src*=gtag]')->remove(); |
258
|
2 |
|
$crawler->filter('noscript')->remove(); |
259
|
2 |
|
$crawler->filterXPath("//script[text()[contains(.,'piwik.tiki.org')]]")->remove(); |
260
|
2 |
|
$crawler->filterXPath("//script[text()[contains(.,'gtag')]]")->remove(); |
261
|
2 |
|
} |
262
|
|
|
|
263
|
2 |
|
protected function updateCSS(HtmlPageCrawler $crawler) |
264
|
|
|
{ |
265
|
2 |
|
$this->updateTopPadding($crawler); |
266
|
2 |
|
$this->updateArticlePadding($crawler); |
267
|
2 |
|
} |
268
|
|
|
|
269
|
2 |
|
protected function updateTopPadding(HtmlPageCrawler $crawler) |
270
|
|
|
{ |
271
|
2 |
|
$crawler->filter('body') |
272
|
2 |
|
->removeClass('navbar-padding') |
273
|
2 |
|
->addClass('hide_zone_left') |
274
|
2 |
|
->css('padding-top', '0') |
275
|
|
|
; |
276
|
2 |
|
} |
277
|
|
|
|
278
|
2 |
|
protected function updateArticlePadding(HtmlPageCrawler $crawler) |
279
|
|
|
{ |
280
|
2 |
|
$crawler->filter('article#top') |
281
|
2 |
|
->css('padding-top', '44px') |
282
|
|
|
; |
283
|
2 |
|
} |
284
|
|
|
|
285
|
2 |
|
protected function insertOnlineRedirection(HtmlPageCrawler $crawler) |
286
|
|
|
{ |
287
|
2 |
|
$onlineUrl = ''; |
288
|
2 |
|
$meta = $crawler->filter('meta[property="og:url"]'); |
289
|
|
|
|
290
|
2 |
|
if ($meta->getDOMDocument()) { |
291
|
2 |
|
$onlineUrl = $meta->attr('content'); |
292
|
|
|
} |
293
|
|
|
|
294
|
2 |
|
$crawler->filter('html')->prepend("<!-- Online page at $onlineUrl -->"); |
295
|
2 |
|
} |
296
|
|
|
|
297
|
2 |
|
protected function insertDashTableOfContents(HtmlPageCrawler $crawler) |
298
|
|
|
{ |
299
|
2 |
|
$crawler->filter('#page-data > h1:first-of-type') |
300
|
2 |
|
->before('<a name="//apple_ref/cpp/Section/Top" class="dashAnchor"></a>'); |
301
|
|
|
|
302
|
|
|
$crawler->filter('h2')->each(static function (HtmlPageCrawler $node) { |
303
|
2 |
|
$node->before( |
304
|
2 |
|
'<a id="' . Str::slug($node->text()) . '" name="//apple_ref/cpp/Section/' . rawurlencode($node->text()) . '" class="dashAnchor"></a>' |
305
|
|
|
); |
306
|
2 |
|
}); |
307
|
2 |
|
} |
308
|
|
|
} |
309
|
|
|
|