|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace ReliqArts\Scavenger\Service; |
|
6
|
|
|
|
|
7
|
|
|
class Scanner |
|
8
|
|
|
{ |
|
9
|
|
|
/** |
|
10
|
|
|
* List of words (regex) we don't want in our scraps.'. |
|
11
|
|
|
* |
|
12
|
|
|
* @var array |
|
13
|
|
|
*/ |
|
14
|
|
|
protected array $badWords = []; |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* Scanner constructor. |
|
18
|
|
|
*/ |
|
19
|
|
|
public function __construct(array $badWords = []) |
|
20
|
|
|
{ |
|
21
|
|
|
$this->badWords = $badWords; |
|
22
|
|
|
} |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* Determine whether a scrap data has bad words and therefore is unwanted. |
|
26
|
|
|
* |
|
27
|
|
|
* @param array $badWords List of words (regex) we don't want in our scraps.'. |
|
28
|
|
|
*/ |
|
29
|
|
|
public function hasBadWords(array $scrapData, array $badWords = []) |
|
30
|
|
|
{ |
|
31
|
|
|
$invalid = false; |
|
32
|
|
|
$badWords = array_merge($this->badWords, $badWords); |
|
33
|
|
|
|
|
34
|
|
|
if (count($badWords)) { |
|
35
|
|
|
$badWordsRegex = sprintf('/(%s)/i', implode(')|(', $badWords)); |
|
36
|
|
|
|
|
37
|
|
|
// check for bad words |
|
38
|
|
|
foreach ($scrapData as $attr => $value) { |
|
39
|
|
|
if (!ConfigProvider::isSpecialKey($attr) && preg_match($badWordsRegex, $value)) { |
|
40
|
|
|
$invalid = true; |
|
41
|
|
|
|
|
42
|
|
|
break; |
|
43
|
|
|
} |
|
44
|
|
|
} |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
return $invalid; |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* Scour a string and pluck details. |
|
52
|
|
|
* |
|
53
|
|
|
* @param string $string the string to be scoured |
|
54
|
|
|
* @param array $map map to use for detail scouring |
|
55
|
|
|
* @param bool $retain whether to leave match in source string |
|
56
|
|
|
*/ |
|
57
|
|
|
public static function pluckDetails(string &$string, array $map = [], bool $retain = false): array |
|
58
|
|
|
{ |
|
59
|
|
|
$details = []; |
|
60
|
|
|
|
|
61
|
|
|
// Pluck mapped details from string |
|
62
|
|
|
foreach ($map as $attr => $regex) { |
|
63
|
|
|
// match and replace details in string |
|
64
|
|
|
$string = preg_replace_callback($regex, static function ($m) use ($attr, &$details, $retain) { |
|
65
|
|
|
// grab match |
|
66
|
|
|
$match = trim($m[0]); |
|
67
|
|
|
$details[$attr] = $match; |
|
68
|
|
|
// return match if it should be left in string |
|
69
|
|
|
if ($retain) { |
|
70
|
|
|
return $match; |
|
71
|
|
|
} |
|
72
|
|
|
// @noinspection PhpInconsistentReturnPointsInspection |
|
73
|
|
|
}, $string); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
return $details; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
/** |
|
80
|
|
|
* Searches array for needles. The first one found is returned. |
|
81
|
|
|
* If needles aren't supplied the first non-empty item in array is returned. |
|
82
|
|
|
* |
|
83
|
|
|
* @param array $haystack array to search |
|
84
|
|
|
* @param array $needles optional list of items to check for |
|
85
|
|
|
*/ |
|
86
|
|
|
public static function firstNonEmpty(array &$haystack, array $needles = []) |
|
87
|
|
|
{ |
|
88
|
|
|
if (!empty($needles)) { |
|
89
|
|
|
foreach ($needles as $value) { |
|
90
|
|
|
if (!empty($haystack[$value])) { |
|
91
|
|
|
return $haystack[$value]; |
|
92
|
|
|
} |
|
93
|
|
|
} |
|
94
|
|
|
} else { |
|
95
|
|
|
foreach ($haystack as $value) { |
|
96
|
|
|
if (!empty($value)) { |
|
97
|
|
|
return $value; |
|
98
|
|
|
} |
|
99
|
|
|
} |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
return null; |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
/** |
|
106
|
|
|
* Convert <br/> to newlines. |
|
107
|
|
|
*/ |
|
108
|
|
|
public static function br2nl(string $text): string |
|
109
|
|
|
{ |
|
110
|
|
|
return preg_replace('/<br[\\/]?>/', "\n", $text); |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
|
|
public static function cleanText(string $text): string |
|
114
|
|
|
{ |
|
115
|
|
|
return self::removeReturnsAndTabs(strip_tags($text)); |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
/** |
|
119
|
|
|
* Remove tabs and newlines from text. |
|
120
|
|
|
*/ |
|
121
|
|
|
private static function removeReturnsAndTabs(string $text): string |
|
122
|
|
|
{ |
|
123
|
|
|
$text = preg_replace('/\\s{2,}/', ' ', preg_replace("/[\r\n\t]+/", ' ', $text)); |
|
124
|
|
|
|
|
125
|
|
|
return str_replace(' / ', null, $text); |
|
126
|
|
|
} |
|
127
|
|
|
} |
|
128
|
|
|
|
reliqarts /
laravel-scavenger
