|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Ngtfkx\Laradeck\AddressGenerator\Commands; |
|
4
|
|
|
|
|
5
|
|
|
use Illuminate\Console\Command; |
|
6
|
|
|
use Illuminate\Support\Collection; |
|
7
|
|
|
use Illuminate\Support\Facades\Storage; |
|
8
|
|
|
use Symfony\Component\DomCrawler\Crawler; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* Парсер улиц и домов с сайта www.city-address.ru |
|
12
|
|
|
* |
|
13
|
|
|
* Class ParseCityAddressRu |
|
14
|
|
|
* @package Ngtfkx\Laradeck\AddressGenerator\Commands |
|
15
|
|
|
*/ |
|
16
|
|
|
class ParseCityAddressRu extends Command |
|
17
|
|
|
{ |
|
18
|
|
|
/** |
|
19
|
|
|
* The name and signature of the console command. |
|
20
|
|
|
* |
|
21
|
|
|
* @var string |
|
22
|
|
|
*/ |
|
23
|
|
|
protected $signature = 'address:city-address-ru {city} {url} {--limit=0}'; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* The console command description. |
|
27
|
|
|
* |
|
28
|
|
|
* @var string |
|
29
|
|
|
*/ |
|
30
|
|
|
protected $description = 'Parse streets and building numbers from www.city-address.ru'; |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* @var Collection |
|
34
|
|
|
*/ |
|
35
|
|
|
protected $streets; |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* @var int |
|
39
|
|
|
*/ |
|
40
|
|
|
protected $cityId; |
|
41
|
|
|
|
|
42
|
|
|
/** |
|
43
|
|
|
* @var int |
|
44
|
|
|
*/ |
|
45
|
|
|
protected $limit; |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* @var string |
|
49
|
|
|
*/ |
|
50
|
|
|
protected $url; |
|
51
|
|
|
|
|
52
|
|
|
/** |
|
53
|
|
|
* @var string |
|
54
|
|
|
*/ |
|
55
|
|
|
protected $name; |
|
56
|
|
|
|
|
57
|
|
|
public function __construct() |
|
58
|
|
|
{ |
|
59
|
|
|
$this->streets = new Collection(); |
|
60
|
|
|
|
|
61
|
|
|
parent::__construct(); |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
public function handle() |
|
65
|
|
|
{ |
|
66
|
|
|
$this->cityId = $this->argument('city'); |
|
67
|
|
|
|
|
68
|
|
|
$this->url = $this->argument('url'); |
|
69
|
|
|
|
|
70
|
|
|
$this->limit = (int)$this->option('limit'); |
|
71
|
|
|
|
|
72
|
|
|
$this->parse(); |
|
73
|
|
|
|
|
74
|
|
|
$this->generateFile(); |
|
75
|
|
|
|
|
76
|
|
|
$this->info('Data was saved'); |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
/** |
|
80
|
|
|
* Парсинг данных |
|
81
|
|
|
*/ |
|
82
|
|
|
protected function parse() |
|
83
|
|
|
{ |
|
84
|
|
|
$firstPage = $this->download($this->url); |
|
85
|
|
|
|
|
86
|
|
|
$allPages = (int)(new Crawler($firstPage))->filter('.uk-pagination > li')->last()->text(); |
|
87
|
|
|
|
|
88
|
|
|
$name = (new Crawler($firstPage))->filterXpath('//meta[@name="geo.placename"]')->attr('content'); |
|
89
|
|
|
|
|
90
|
|
|
$this->name = explode(', ', $name)[0]; |
|
91
|
|
|
|
|
92
|
|
|
$k = 0; |
|
93
|
|
|
|
|
94
|
|
|
for ($i = 1; $i <= $allPages; $i++) { |
|
95
|
|
|
$page = $this->download($this->url . '/page-' . $i . '/'); |
|
96
|
|
|
|
|
97
|
|
|
$this->line('Downloading streets and building numbers from page #' . $i . ' from ' . $allPages); |
|
98
|
|
|
|
|
99
|
|
|
$streets = (new Crawler($page))->filter('.c4 > a'); |
|
100
|
|
|
|
|
101
|
|
|
$streets->each(function ($node) { |
|
102
|
|
|
$url = 'http://www.city-address.ru' . $node->attr('href'); |
|
103
|
|
|
|
|
104
|
|
|
$page = $this->download($url); |
|
105
|
|
|
|
|
106
|
|
|
$numbers = collect((new Crawler($page))->filter('.c6 > div > a')->extract('_text')) |
|
107
|
|
|
->reject(function ($item) { |
|
108
|
|
|
return strpos($item, 'дом ') !== 0; |
|
109
|
|
|
})->map(function ($item) { |
|
110
|
|
|
return str_replace('дом № ', '', $item); |
|
111
|
|
|
})->toArray(); |
|
112
|
|
|
|
|
113
|
|
|
$this->streets->put($node->text(), $numbers); |
|
114
|
|
|
}); |
|
115
|
|
|
|
|
116
|
|
|
if (++$k == $this->limit) { |
|
117
|
|
|
$this->warn('Interrupted by the limit'); |
|
118
|
|
|
break; |
|
119
|
|
|
} |
|
120
|
|
|
} |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
/** |
|
124
|
|
|
* Скачивание страница |
|
125
|
|
|
* |
|
126
|
|
|
* @param string $url |
|
127
|
|
|
* @return string |
|
128
|
|
|
*/ |
|
129
|
|
|
protected function download(string $url): string |
|
130
|
|
|
{ |
|
131
|
|
|
$content = file_get_contents($url); |
|
132
|
|
|
|
|
133
|
|
|
return $content; |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
/** |
|
137
|
|
|
* Запись данных в файл |
|
138
|
|
|
*/ |
|
139
|
|
|
protected function generateFile() |
|
140
|
|
|
{ |
|
141
|
|
|
$output = '<?php' . PHP_EOL . PHP_EOL; |
|
142
|
|
|
$output .= '/**' . PHP_EOL; |
|
143
|
|
|
$output .= ' * ' . $this->name . ', Россия' . PHP_EOL; |
|
144
|
|
|
$output .= ' */' . PHP_EOL . PHP_EOL; |
|
145
|
|
|
$output .= 'return [' . PHP_EOL; |
|
146
|
|
|
foreach ($this->streets as $street => $numbers) { |
|
147
|
|
|
$glueNumbers = implode('", "', $numbers); |
|
148
|
|
|
if (!empty($glueNumbers)) { |
|
149
|
|
|
$output .= ' "' . trim($street, ";") . '" => ["' . $glueNumbers . '"],' . PHP_EOL; |
|
150
|
|
|
} |
|
151
|
|
|
} |
|
152
|
|
|
$output .= '];' . PHP_EOL; |
|
153
|
|
|
|
|
154
|
|
|
Storage::put($this->cityId . '.php', $output); |
|
155
|
|
|
} |
|
156
|
|
|
} |
|
157
|
|
|
|
ngtfkx /
laradeck-address-generator
