1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Godbout\DashDocsetBuilder\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 RickAstley extends BaseDocset |
11
|
|
|
{ |
12
|
|
|
public const CODE = 'rick-astley'; |
13
|
|
|
public const NAME = 'Rick Astley'; |
14
|
|
|
public const URL = 'rickastley.co.uk'; |
15
|
|
|
public const INDEX = 'index.html'; |
16
|
|
|
public const PLAYGROUND = ''; |
17
|
|
|
public const ICON_16 = 'icons/favicon-16x16.png'; |
18
|
|
|
public const ICON_32 = 'icons/favicon-32x32.png'; |
19
|
|
|
public const EXTERNAL_DOMAINS = [ |
20
|
|
|
'fonts.googleapis.com', |
21
|
|
|
'widget.songkick.com', |
22
|
|
|
'cdn-images.mailchimp.com', |
23
|
|
|
's3.amazonaws.com', |
24
|
|
|
'ajax.googleapis.com' |
25
|
|
|
]; |
26
|
|
|
|
27
|
|
|
|
28
|
|
|
public function entries(string $file): Collection |
29
|
|
|
{ |
30
|
|
|
$entries = collect(); |
31
|
|
|
|
32
|
|
|
$crawler = HtmlPageCrawler::create(Storage::get($file)); |
33
|
|
|
|
34
|
|
|
if (Str::contains($file, "{$this->url()}/index.html")) { |
35
|
|
|
$crawler->filter('#main_menu li:not(:first-child) a')->each(function () use ($entries) { |
36
|
|
|
$entries->push([ |
37
|
|
|
'name' => 'Rick Astley - Official Website', |
38
|
|
|
'type' => 'Guide', |
39
|
|
|
'path' => $this->url() . '/index.html' |
40
|
|
|
]); |
41
|
|
|
}); |
42
|
|
|
} |
43
|
|
|
|
44
|
|
|
$crawler->filter('#main_menu li:not(:first-child) a')->each(function (HtmlPageCrawler $node) use ($entries) { |
45
|
|
|
$entries->push([ |
46
|
|
|
'name' => $node->text(), |
47
|
|
|
'type' => 'Section', |
48
|
|
|
'path' => $this->url() . '/' . $node->attr('href') |
49
|
|
|
]); |
50
|
|
|
}); |
51
|
|
|
|
52
|
|
|
return $entries; |
53
|
|
|
} |
54
|
|
|
|
55
|
|
|
public function format(string $html): string |
56
|
|
|
{ |
57
|
|
|
$crawler = HtmlPageCrawler::create($html); |
58
|
|
|
|
59
|
|
|
$this->removeHeader($crawler); |
60
|
|
|
$this->removeFooter($crawler); |
61
|
|
|
|
62
|
|
|
$this->removeUnwantedHTML($crawler); |
63
|
|
|
$this->removeUnwantedJavaScript($crawler); |
64
|
|
|
|
65
|
|
|
$this->insertDashTableOfContents($crawler); |
66
|
|
|
|
67
|
|
|
return $crawler->saveHTML(); |
68
|
|
|
} |
69
|
|
|
|
70
|
|
|
protected function removeHeader(HtmlPageCrawler $crawler) |
71
|
|
|
{ |
72
|
|
|
$crawler->filter('#header')->remove(); |
73
|
|
|
} |
74
|
|
|
|
75
|
|
|
protected function removeFooter(HtmlPageCrawler $crawler) |
76
|
|
|
{ |
77
|
|
|
$crawler->filter('#footer')->remove(); |
78
|
|
|
} |
79
|
|
|
|
80
|
|
|
protected function removeUnwantedHTML(HtmlPageCrawler $crawler) |
81
|
|
|
{ |
82
|
|
|
$crawler->filterXPath("//img[@src[contains(.,'secure.adnxs.com')]]")->remove(); |
83
|
|
|
} |
84
|
|
|
|
85
|
|
|
protected function removeUnwantedJavaScript(HtmlPageCrawler $crawler) |
86
|
|
|
{ |
87
|
|
|
$crawler->filter('noscript')->remove(); |
88
|
|
|
$crawler->filterXPath("//script[@src[contains(.,'platform.twitter.com')]]")->remove(); |
89
|
|
|
$crawler->filterXPath("//script[@src[contains(.,'googletagmanager')]]")->remove(); |
90
|
|
|
$crawler->filterXPath("//script[@src[contains(.,'googleadservices')]]")->remove(); |
91
|
|
|
$crawler->filterXPath("//script[text()[contains(.,'googletagmanager')]]")->remove(); |
92
|
|
|
$crawler->filterXPath("//script[text()[contains(.,'gtag')]]")->remove(); |
93
|
|
|
$crawler->filterXPath("//script[text()[contains(.,'connect.facebook.net')]]")->remove(); |
94
|
|
|
$crawler->filterXPath("//script[text()[contains(.,'google_conversion_id')]]")->remove(); |
95
|
|
|
$crawler->filterXPath("//script[text()[contains(.,'googleadservices')]]")->remove(); |
96
|
|
|
$crawler->filterXPath("//script[text()[contains(.,'platform.twitter.com')]]")->remove(); |
97
|
|
|
$crawler->filterXPath("//script[text()[contains(.,'twttr.conversion')]]")->remove(); |
98
|
|
|
} |
99
|
|
|
|
100
|
|
|
protected function insertDashTableOfContents(HtmlPageCrawler $crawler) |
101
|
|
|
{ |
102
|
|
|
$crawler->filter('head') |
103
|
|
|
->before('<a name="//apple_ref/cpp/Section/Top" class="dashAnchor"></a>'); |
104
|
|
|
|
105
|
|
|
$crawler->filter('div.page_title, div.product_title')->each(static function (HtmlPageCrawler $node) { |
106
|
|
|
$node->before( |
107
|
|
|
'<a id="' . Str::slug($node->text()) . '" name="//apple_ref/cpp/Section/' . rawurlencode($node->text()) . '" class="dashAnchor"></a>' |
108
|
|
|
); |
109
|
|
|
}); |
110
|
|
|
} |
111
|
|
|
} |
112
|
|
|
|