|
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 Ploi extends BaseDocset |
|
11
|
|
|
{ |
|
12
|
|
|
public const CODE = 'ploi-api'; |
|
13
|
|
|
public const NAME = 'Ploi API'; |
|
14
|
|
|
public const URL = 'developers.ploi.io'; |
|
15
|
|
|
public const INDEX = 'index.html'; |
|
16
|
|
|
public const PLAYGROUND = ''; |
|
17
|
|
|
public const ICON_16 = '../documentator.s3.eu-west-3.amazonaws.com/11/conversions/favicon-favicon-16.png'; |
|
18
|
|
|
public const ICON_32 = '../documentator.s3.eu-west-3.amazonaws.com/11/conversions/favicon-favicon-32.png'; |
|
19
|
|
|
public const EXTERNAL_DOMAINS = [ |
|
20
|
|
|
'documentator.s3.eu-west-3.amazonaws.com' |
|
21
|
|
|
]; |
|
22
|
|
|
|
|
23
|
|
|
|
|
24
|
|
|
public function grab(): bool |
|
25
|
|
|
{ |
|
26
|
|
|
system( |
|
27
|
|
|
"echo; wget {$this->url()} \ |
|
28
|
|
|
--mirror \ |
|
29
|
|
|
--trust-server-names \ |
|
30
|
|
|
--page-requisites \ |
|
31
|
|
|
--adjust-extension \ |
|
32
|
|
|
--convert-links \ |
|
33
|
|
|
--span-hosts \ |
|
34
|
|
|
--domains={$this->externalDomains()} \ |
|
35
|
|
|
--directory-prefix=storage/{$this->downloadedDirectory()} \ |
|
36
|
|
|
-e robots=off \ |
|
37
|
|
|
--quiet \ |
|
38
|
|
|
--show-progress", |
|
39
|
|
|
$result |
|
40
|
|
|
); |
|
41
|
|
|
|
|
42
|
|
|
return $result === 0; |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
2 |
|
public function entries(string $file): Collection |
|
46
|
|
|
{ |
|
47
|
2 |
|
$crawler = HtmlPageCrawler::create(Storage::get($file)); |
|
48
|
|
|
|
|
49
|
2 |
|
$entries = collect(); |
|
50
|
2 |
|
$entries = $entries->merge($this->guideEntries($crawler, $file)); |
|
51
|
|
|
|
|
52
|
2 |
|
return $entries; |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
2 |
|
protected function guideEntries(HtmlPageCrawler $crawler, string $file) |
|
56
|
|
|
{ |
|
57
|
2 |
|
$entries = collect(); |
|
58
|
|
|
|
|
59
|
2 |
|
if (Str::contains($file, "{$this->url()}/index.html")) { |
|
60
|
|
|
$crawler->filter('aside a.ml-2')->each(function (HtmlPageCrawler $node) use ($entries, $file) { |
|
|
|
|
|
|
61
|
2 |
|
$entries->push([ |
|
62
|
2 |
|
'name' => trim($node->text()), |
|
63
|
2 |
|
'type' => 'Guide', |
|
64
|
2 |
|
'path' => $this->url() . '/' . $node->attr('href'), |
|
65
|
|
|
]); |
|
66
|
2 |
|
}); |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
2 |
|
return $entries; |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
2 |
|
public function format(string $html): string |
|
73
|
|
|
{ |
|
74
|
2 |
|
$crawler = HtmlPageCrawler::create($html); |
|
75
|
|
|
|
|
76
|
2 |
|
$this->removeHeader($crawler); |
|
77
|
2 |
|
$this->removeSidebar($crawler); |
|
78
|
2 |
|
$this->removePreviousAndNextNavigation($crawler); |
|
79
|
2 |
|
$this->updateTopPadding($crawler); |
|
80
|
2 |
|
$this->removeUnwantedJavaScript($crawler); |
|
81
|
2 |
|
$this->insertDashTableOfContents($crawler); |
|
82
|
|
|
|
|
83
|
2 |
|
return $crawler->saveHTML(); |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
2 |
|
protected function removeHeader(HtmlPageCrawler $crawler) |
|
87
|
|
|
{ |
|
88
|
2 |
|
$crawler->filter('header')->remove(); |
|
89
|
2 |
|
} |
|
90
|
|
|
|
|
91
|
2 |
|
protected function removeSidebar(HtmlPageCrawler $crawler) |
|
92
|
|
|
{ |
|
93
|
2 |
|
$crawler->filter('aside')->remove(); |
|
94
|
2 |
|
} |
|
95
|
|
|
|
|
96
|
2 |
|
protected function removePreviousAndNextNavigation(HtmlPageCrawler $crawler) |
|
97
|
|
|
{ |
|
98
|
2 |
|
$crawler->filter('#previous-and-next')->remove(); |
|
99
|
2 |
|
} |
|
100
|
|
|
|
|
101
|
2 |
|
protected function updateTopPadding(HtmlPageCrawler $crawler) |
|
102
|
|
|
{ |
|
103
|
2 |
|
$crawler->filter('main') |
|
104
|
2 |
|
->addClass('md:pt-12') |
|
105
|
|
|
; |
|
106
|
2 |
|
} |
|
107
|
|
|
|
|
108
|
2 |
|
protected function removeUnwantedJavaScript(HtmlPageCrawler $crawler) |
|
109
|
|
|
{ |
|
110
|
2 |
|
$crawler->filter('script[src*=gtag]')->remove(); |
|
111
|
2 |
|
$crawler->filterXPath("//script[text()[contains(.,'gtag')]]")->remove(); |
|
112
|
2 |
|
} |
|
113
|
|
|
|
|
114
|
2 |
|
protected function insertDashTableOfContents(HtmlPageCrawler $crawler) |
|
115
|
|
|
{ |
|
116
|
2 |
|
$crawler->filter('h1') |
|
117
|
2 |
|
->before('<a name="//apple_ref/cpp/Section/Top" class="dashAnchor"></a>'); |
|
118
|
|
|
|
|
119
|
|
|
$crawler->filter('h2, h3, h4')->each(static function (HtmlPageCrawler $node) { |
|
120
|
2 |
|
$node->before( |
|
121
|
2 |
|
'<a id="' . Str::slug($node->text()) . '" name="//apple_ref/cpp/Section/' . rawurlencode($node->text()) . '" class="dashAnchor"></a>' |
|
122
|
|
|
); |
|
123
|
2 |
|
}); |
|
124
|
|
|
|
|
125
|
|
|
$crawler->filterXPath('//p[not(descendant::code)]')->each(static function (HtmlPageCrawler $node) { |
|
126
|
2 |
|
$node->before( |
|
127
|
2 |
|
'<a id="' . Str::slug($node->text()) . '" name="//apple_ref/cpp/Section/' . rawurlencode($node->text()) . '" class="dashAnchor"></a>' |
|
128
|
|
|
); |
|
129
|
2 |
|
}); |
|
130
|
2 |
|
} |
|
131
|
|
|
} |
|
132
|
|
|
|
godbout /
dash-docset-builder
This check looks for imports that have been defined, but are not used in the scope.