|
1
|
|
|
<?php namespace Tamtamchik\NameCase; |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* Class Formatter. |
|
5
|
|
|
*/ |
|
6
|
|
|
class Formatter |
|
7
|
|
|
{ |
|
8
|
|
|
// Default options. |
|
9
|
|
|
private static $options = [ |
|
10
|
|
|
'lazy' => true, |
|
11
|
|
|
'irish' => true, |
|
12
|
|
|
'spanish' => true, |
|
13
|
|
|
]; |
|
14
|
|
|
|
|
15
|
|
|
// Irish exceptions. |
|
16
|
|
|
private static $exceptions = [ |
|
17
|
|
|
'\bMacEdo' => 'Macedo', |
|
18
|
|
|
'\bMacEvicius' => 'Macevicius', |
|
19
|
|
|
'\bMacHado' => 'Machado', |
|
20
|
|
|
'\bMacHar' => 'Machar', |
|
21
|
|
|
'\bMacHin' => 'Machin', |
|
22
|
|
|
'\bMacHlin' => 'Machlin', |
|
23
|
|
|
'\bMacIas' => 'Macias', |
|
24
|
|
|
'\bMacIulis' => 'Maciulis', |
|
25
|
|
|
'\bMacKie' => 'Mackie', |
|
26
|
|
|
'\bMacKle' => 'Mackle', |
|
27
|
|
|
'\bMacKlin' => 'Macklin', |
|
28
|
|
|
'\bMacKmin' => 'Mackmin', |
|
29
|
|
|
'\bMacQuarie' => 'Macquarie', |
|
30
|
|
|
]; |
|
31
|
|
|
|
|
32
|
|
|
// General replacements. |
|
33
|
|
|
private static $replacements = [ |
|
34
|
|
|
'\bAl(?=\s+\w)' => 'al', // al Arabic or forename Al. |
|
35
|
|
|
'\b(Bin|Binti|Binte)\b' => 'bin', // bin, binti, binte Arabic |
|
36
|
|
|
'\bAp\b' => 'ap', // ap Welsh. |
|
37
|
|
|
'\bBen(?=\s+\w)' => 'ben', // ben Hebrew or forename Ben. |
|
38
|
|
|
'\bDell([ae])\b' => 'dell\1', // della and delle Italian. |
|
39
|
|
|
'\bD([aeiou])\b' => 'd\1', // da, de, di Italian; du French; do Brasil |
|
40
|
|
|
'\bD([ao]s)\b' => 'd\1', // das, dos Brasileiros |
|
41
|
|
|
'\bDe([lr])\b' => 'de\1', // del Italian; der Dutch/Flemish. |
|
42
|
|
|
'\bEl\b' => 'el', // el Greek or El Spanish. |
|
43
|
|
|
'\bLa\b' => 'la', // la French or La Spanish. |
|
44
|
|
|
'\bL([eo])\b' => 'l\1', // lo Italian; le French. |
|
45
|
|
|
'\bVan(?=\s+\w)' => 'van', // van German or forename Van. |
|
46
|
|
|
'\bVon\b' => 'von', // von Dutch/Flemish |
|
47
|
|
|
]; |
|
48
|
|
|
|
|
49
|
|
|
// Spanish conjunctions. |
|
50
|
|
|
private static $conjunctions = ["Y", "E", "I"]; |
|
51
|
|
|
|
|
52
|
|
|
// Roman letters regexp. |
|
53
|
|
|
private static $romanRegex = '\b((?:[Xx]{1,3}|[Xx][Ll]|[Ll][Xx]{0,3})?(?:[Ii]{1,3}|[Ii][VvXx]|[Vv][Ii]{0,3})?)\b'; |
|
54
|
|
|
|
|
55
|
|
|
/** |
|
56
|
|
|
* Main function for NameCase. |
|
57
|
|
|
* |
|
58
|
|
|
* @param string $string |
|
59
|
|
|
* @param array $options |
|
60
|
|
|
* |
|
61
|
|
|
* @return string |
|
62
|
|
|
*/ |
|
63
|
6 |
|
public static function nc($string, array $options = []) |
|
64
|
|
|
{ |
|
65
|
6 |
|
self::$options = array_merge(self::$options, $options); |
|
66
|
|
|
|
|
67
|
|
|
// Do not do anything if string is mixed and lazy option is true. |
|
68
|
6 |
|
if ($options['lazy'] && self::skipMixed($string)) return $string; |
|
69
|
|
|
|
|
70
|
|
|
// Capitalize |
|
71
|
|
|
$string = self::capitalize($string); |
|
72
|
|
|
$string = self::updateIrish($string); |
|
73
|
|
|
|
|
74
|
|
|
// Fixes for "son (daughter) of" etc |
|
75
|
|
|
foreach (self::$replacements as $pattern => $replacement) { |
|
76
|
|
|
$string = mb_ereg_replace($pattern, $replacement, $string); |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
$string = self::updateRoman($string); |
|
80
|
|
|
$string = self::fixConjunction($string); |
|
81
|
|
|
|
|
82
|
|
|
return $string; |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
/** |
|
86
|
|
|
* Capitalize first letters. |
|
87
|
|
|
* |
|
88
|
|
|
* @param string $string |
|
89
|
|
|
* |
|
90
|
|
|
* @return string |
|
91
|
|
|
*/ |
|
92
|
|
|
private static function capitalize($string) |
|
93
|
|
|
{ |
|
94
|
|
|
$string = mb_strtolower($string); |
|
95
|
|
|
|
|
96
|
|
|
$string = mb_ereg_replace_callback('\b\w', function ($matches) { |
|
97
|
|
|
return mb_strtoupper($matches[0]); |
|
98
|
|
|
}, $string); |
|
99
|
|
|
|
|
100
|
|
|
// Lowercase 's |
|
101
|
|
|
$string = mb_ereg_replace_callback('\'\w\b', function ($matches) { |
|
102
|
|
|
return mb_strtolower($matches[0]); |
|
103
|
|
|
}, $string); |
|
104
|
|
|
|
|
105
|
|
|
return $string; |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
/** |
|
109
|
|
|
* Skip if string is mixed case. |
|
110
|
|
|
* |
|
111
|
|
|
* @param string $string |
|
112
|
|
|
* |
|
113
|
|
|
* @return bool |
|
114
|
|
|
*/ |
|
115
|
|
|
private static function skipMixed($string) |
|
116
|
|
|
{ |
|
117
|
|
|
$firstLetterLower = $string[0] == mb_strtolower($string[0]); |
|
118
|
|
|
$allLowerOrUpper = (mb_strtolower($string) == $string || mb_strtoupper($string) == $string); |
|
119
|
|
|
|
|
120
|
|
|
return ! ($firstLetterLower || $allLowerOrUpper); |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
/** |
|
124
|
|
|
* Update for Irish names. |
|
125
|
|
|
* |
|
126
|
|
|
* @param string $string |
|
127
|
|
|
* |
|
128
|
|
|
* @return string |
|
129
|
|
|
*/ |
|
130
|
|
|
private static function updateIrish($string) |
|
131
|
|
|
{ |
|
132
|
|
|
if ( ! self::$options['irish']) return $string; |
|
133
|
|
|
|
|
134
|
|
|
if (mb_ereg_match('.*?\bMac[A-Za-z]{2,}[^aciozj]\b', $string) || mb_ereg_match('.*?\bMc', $string)) { |
|
135
|
|
|
|
|
136
|
|
|
$string = mb_ereg_replace_callback('\b(Ma?c)([A-Za-z]+)', function ($matches) { |
|
137
|
|
|
return $matches[1] . mb_strtoupper(mb_substr($matches[2], 0, 1)) . mb_substr($matches[2], 1); |
|
138
|
|
|
}, $string); |
|
139
|
|
|
|
|
140
|
|
|
// Now fix "Mac" exceptions |
|
141
|
|
|
foreach (self::$exceptions as $pattern => $replacement) { |
|
142
|
|
|
$string = mb_ereg_replace($pattern, $replacement, $string); |
|
143
|
|
|
} |
|
144
|
|
|
} |
|
145
|
|
|
|
|
146
|
|
|
return mb_ereg_replace('Macmurdo', 'MacMurdo', $string); |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
/** |
|
150
|
|
|
* Fix Spanish conjunctions. |
|
151
|
|
|
* |
|
152
|
|
|
* @param string $string |
|
153
|
|
|
* |
|
154
|
|
|
* @return string |
|
155
|
|
|
*/ |
|
156
|
|
|
private static function fixConjunction($string) |
|
157
|
|
|
{ |
|
158
|
|
|
if ( ! self::$options['spanish']) return $string; |
|
159
|
|
|
|
|
160
|
|
|
foreach (self::$conjunctions as $conjunction) { |
|
161
|
|
|
$string = mb_ereg_replace('\b' . $conjunction . '\b', mb_strtolower($conjunction), $string); |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
|
|
return $string; |
|
165
|
|
|
} |
|
166
|
|
|
|
|
167
|
|
|
/** |
|
168
|
|
|
* Fix roman numeral names. |
|
169
|
|
|
* |
|
170
|
|
|
* @param $string |
|
171
|
|
|
* |
|
172
|
|
|
* @return string |
|
173
|
|
|
*/ |
|
174
|
|
|
private static function updateRoman($string) |
|
175
|
|
|
{ |
|
176
|
|
|
return mb_ereg_replace_callback(self::$romanRegex, function ($matches) { |
|
177
|
|
|
return mb_strtoupper($matches[0]); |
|
178
|
|
|
}, $string); |
|
179
|
|
|
} |
|
180
|
|
|
} |
|
181
|
|
|
|
tamtamchik /
namecase
