|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace kouinkouin\StreetParser; |
|
4
|
|
|
|
|
5
|
|
|
class StreetParser |
|
6
|
|
|
{ |
|
7
|
|
|
/** |
|
8
|
|
|
* @param string $fullStreet |
|
9
|
|
|
* |
|
10
|
|
|
* @return array |
|
11
|
|
|
*/ |
|
12
|
10 |
|
public function getStreetDataFromFullStreet($fullStreet) |
|
13
|
|
|
{ |
|
14
|
10 |
|
$bestResult = ['score' => -1]; |
|
15
|
|
|
|
|
16
|
10 |
|
foreach ($this->getRegexLines() as $regexLine) { |
|
17
|
10 |
|
$result = $this->getLineResult($regexLine, $fullStreet); |
|
18
|
|
|
|
|
19
|
10 |
|
if ($result['score'] > $bestResult['score']) { |
|
20
|
10 |
|
$bestResult = $result; |
|
21
|
|
|
} |
|
22
|
|
|
} |
|
23
|
|
|
|
|
24
|
10 |
|
return $bestResult; |
|
25
|
|
|
} |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* @return array |
|
29
|
|
|
*/ |
|
30
|
10 |
|
private function getRegexLines() |
|
31
|
|
|
{ |
|
32
|
|
|
// Use https://www.regex101.com to debug regex :) |
|
33
|
|
|
return [ |
|
34
|
|
|
[ |
|
35
|
10 |
|
'regex' => '/(.*)/', // basic |
|
36
|
|
|
'street' => 1, |
|
37
|
|
|
'number' => -1, |
|
38
|
|
|
'box' => -1, |
|
39
|
|
|
], |
|
40
|
|
|
// Belgian format. "," mandatory to separate the street and the number+box |
|
41
|
|
|
[ |
|
42
|
|
|
'regex' => '/^(.*\w)\ *\,\ *(\d+\w*)[\/\ ]*(\d*\w*)$/', |
|
43
|
|
|
'street' => 1, |
|
44
|
|
|
'number' => 2, |
|
45
|
|
|
'box' => 3, |
|
46
|
|
|
], |
|
47
|
|
|
// Belgian format. Without "," to separate the street and the number. Don't detect the box |
|
48
|
|
|
[ |
|
49
|
|
|
'regex' => '/^(.*[a-zA-Z])\ *(\d+\w*)([\/\ ]*)(\d*\w*)$/', |
|
50
|
|
|
'street' => 1, |
|
51
|
|
|
'number' => 2, |
|
52
|
|
|
'box' => -1, |
|
53
|
|
|
], |
|
54
|
|
|
// Belgian format. Without "," to separate the street and the number. Don't detect the box |
|
55
|
|
|
[ |
|
56
|
|
|
'regex' => '/^(.*[a-zA-Z])\ *(\d+\w*)[\/\ ]+(\d+\w*)$/', |
|
57
|
|
|
'street' => 1, |
|
58
|
|
|
'number' => 2, |
|
59
|
|
|
'box' => 3, |
|
60
|
|
|
], |
|
61
|
|
|
]; |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
/** |
|
65
|
|
|
* @param array $regexLine |
|
66
|
|
|
* @param string $fullStreet |
|
67
|
|
|
* |
|
68
|
|
|
* @return array |
|
69
|
|
|
*/ |
|
70
|
10 |
|
private function getLineResult($regexLine, $fullStreet) |
|
71
|
|
|
{ |
|
72
|
10 |
|
if (preg_match($regexLine['regex'], $fullStreet, $matches)) { |
|
73
|
10 |
|
return $this->getMatchResult($matches, $regexLine); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
10 |
|
return ['score' => -1]; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
/** |
|
80
|
|
|
* @param array $matches |
|
81
|
|
|
* @param array $regexLine |
|
82
|
|
|
* |
|
83
|
|
|
* @return array |
|
84
|
|
|
*/ |
|
85
|
10 |
|
private function getMatchResult(array $matches, array $regexLine) |
|
86
|
|
|
{ |
|
87
|
10 |
|
$result = []; |
|
88
|
10 |
|
$score = 0; |
|
89
|
|
|
|
|
90
|
10 |
|
foreach (['street', 'number', 'box'] as $streetItem) { |
|
91
|
10 |
|
if (isset($matches[$regexLine[$streetItem]])) { |
|
92
|
10 |
|
$score++; |
|
93
|
10 |
|
$result[$streetItem] = $matches[$regexLine[$streetItem]]; |
|
94
|
|
|
} else { |
|
95
|
10 |
|
$result[$streetItem] = ''; |
|
96
|
|
|
} |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
10 |
|
$result['score'] = $score; |
|
100
|
|
|
|
|
101
|
10 |
|
return $result; |
|
102
|
|
|
} |
|
103
|
|
|
} |
|
104
|
|
|
|
kouinkouin /
street-parser
