|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace SimaLand\API\Parser; |
|
4
|
|
|
|
|
5
|
|
|
use SimaLand\API\AbstractList; |
|
6
|
|
|
use SimaLand\API\BaseObject; |
|
7
|
|
|
use SimaLand\API\Record; |
|
8
|
|
|
|
|
9
|
|
|
/** |
|
10
|
|
|
* Загрузка и сохранение всех записей сущностей. |
|
11
|
|
|
* |
|
12
|
|
|
* ```php |
|
13
|
|
|
* |
|
14
|
|
|
* $client = new \SimaLand\API\Rest\Client([ |
|
15
|
|
|
* 'login' => 'login', |
|
16
|
|
|
* 'password' => 'password' |
|
17
|
|
|
* ]); |
|
18
|
|
|
* $itemList = new \SimaLand\API\Entities\ItemList($client); |
|
19
|
|
|
* $itemStorage = new Json(['filename' => 'path/to/item.txt']); |
|
20
|
|
|
* $categoryList = new \SimaLand\API\Entities\CategoryList($client); |
|
21
|
|
|
* $categoryStorage = new Json(['filename' => 'path/to/category.txt']); |
|
22
|
|
|
* $parser = new Parser(['metaFilename' => 'path/to/file']); |
|
23
|
|
|
* $parser->addEntity($itemList, $itemStorage); |
|
24
|
|
|
* $parser->addEntity($categoryList, $categoryStorage); |
|
25
|
|
|
* $parser->run(); |
|
26
|
|
|
* |
|
27
|
|
|
* ``` |
|
28
|
|
|
*/ |
|
29
|
|
|
class Parser extends BaseObject |
|
30
|
|
|
{ |
|
31
|
|
|
/** |
|
32
|
|
|
* Кол-во итераций после которых сохраняются мета данные. |
|
33
|
|
|
* |
|
34
|
|
|
* @var int |
|
35
|
|
|
*/ |
|
36
|
|
|
public $iterationCount = 1000; |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* @var array |
|
40
|
|
|
*/ |
|
41
|
|
|
protected $list = []; |
|
42
|
|
|
|
|
43
|
|
|
/** |
|
44
|
|
|
* Путь до файла с мета данными. |
|
45
|
|
|
* |
|
46
|
|
|
* @var string |
|
47
|
|
|
*/ |
|
48
|
|
|
protected $metaFilename; |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* Мета данные. |
|
52
|
|
|
* |
|
53
|
|
|
* @var array |
|
54
|
|
|
*/ |
|
55
|
|
|
protected $metaData = []; |
|
56
|
|
|
|
|
57
|
|
|
/** |
|
58
|
|
|
* @inheritdoc |
|
59
|
|
|
*/ |
|
60
|
15 |
|
public function __construct(array $options = []) |
|
61
|
|
|
{ |
|
62
|
15 |
|
if (empty($options['metaFilename'])) { |
|
63
|
3 |
|
throw new \Exception('Param "metaFilename" can`t be empty'); |
|
64
|
|
|
} |
|
65
|
12 |
|
$this->metaFilename = $options['metaFilename']; |
|
66
|
12 |
|
unset($options['metaFilename']); |
|
67
|
12 |
|
parent::__construct($options); |
|
68
|
12 |
|
} |
|
69
|
|
|
|
|
70
|
|
|
/** |
|
71
|
|
|
* @param AbstractList $entity |
|
72
|
|
|
* @param StorageInterface $storage |
|
73
|
|
|
* @return Parser |
|
74
|
|
|
*/ |
|
75
|
12 |
|
public function addEntity(AbstractList $entity, StorageInterface $storage) |
|
76
|
|
|
{ |
|
77
|
12 |
|
$this->list[] = [ |
|
78
|
12 |
|
'entity' => $entity, |
|
79
|
8 |
|
'storage' => $storage |
|
80
|
4 |
|
]; |
|
81
|
12 |
|
return $this; |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
/** |
|
85
|
|
|
* Сбросить мета данные. |
|
86
|
|
|
* |
|
87
|
|
|
* @return Parser |
|
88
|
|
|
*/ |
|
89
|
9 |
|
public function reset() |
|
90
|
|
|
{ |
|
91
|
9 |
|
if (file_exists($this->metaFilename)) { |
|
92
|
9 |
|
unlink($this->metaFilename); |
|
93
|
3 |
|
} |
|
94
|
9 |
|
return $this; |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
/** |
|
98
|
|
|
* Запустить парсер. |
|
99
|
|
|
* |
|
100
|
|
|
* @param bool|false $continue Продолжить парсить с место обрыва. |
|
101
|
|
|
*/ |
|
102
|
12 |
|
public function run($continue = true) |
|
103
|
|
|
{ |
|
104
|
12 |
|
$logger = $this->getLogger(); |
|
105
|
12 |
|
$this->loadMetaData(); |
|
106
|
12 |
|
foreach ($this->list as $el) { |
|
107
|
|
|
/** @var AbstractList $entity */ |
|
108
|
12 |
|
$entity = $el['entity']; |
|
109
|
12 |
|
$entityName = $entity->getEntity(); |
|
110
|
12 |
|
if ($continue && isset($this->metaData[$entityName])) { |
|
111
|
6 |
|
if (isset($this->metaData[$entityName]['finish']) && $this->metaData[$entityName]['finish']) { |
|
112
|
3 |
|
continue; |
|
113
|
|
|
} |
|
114
|
3 |
|
$entity->addGetParams($this->metaData[$entityName]); |
|
115
|
1 |
|
} |
|
116
|
|
|
/** @var StorageInterface $storage */ |
|
117
|
12 |
|
$storage = $el['storage']; |
|
118
|
12 |
|
$logger->info("Parse \"{$entityName}\""); |
|
119
|
12 |
|
foreach ($entity as $key => $record) { |
|
120
|
9 |
|
if ($continue) { |
|
121
|
9 |
|
$this->fillAndSaveMetaData($entity, $record, $key); |
|
122
|
3 |
|
} |
|
123
|
9 |
|
$storage->save($record); |
|
124
|
4 |
|
} |
|
125
|
12 |
|
$logger->info("Finish parse \"{$entityName}\""); |
|
126
|
12 |
|
if ($continue) { |
|
127
|
12 |
|
$this->finishParseEntity($entity); |
|
128
|
12 |
|
$this->saveMetaData(); |
|
129
|
4 |
|
} |
|
130
|
4 |
|
} |
|
131
|
12 |
|
} |
|
132
|
|
|
|
|
133
|
|
|
/** |
|
134
|
|
|
* Загрузить мета данные. |
|
135
|
|
|
*/ |
|
136
|
12 |
|
protected function loadMetaData() |
|
137
|
|
|
{ |
|
138
|
12 |
|
if (!file_exists($this->metaFilename)) { |
|
139
|
12 |
|
return; |
|
140
|
|
|
} |
|
141
|
6 |
|
$data = file_get_contents($this->metaFilename); |
|
142
|
6 |
|
$this->metaData = (array)json_decode($data, true); |
|
143
|
6 |
|
} |
|
144
|
|
|
|
|
145
|
|
|
/** |
|
146
|
|
|
* Заполнить мета данные. |
|
147
|
|
|
* |
|
148
|
|
|
* @param AbstractList $entity |
|
149
|
|
|
* @param Record $record |
|
150
|
|
|
* @param int $i |
|
151
|
|
|
*/ |
|
152
|
9 |
|
protected function fillAndSaveMetaData(AbstractList $entity, Record $record, $i) |
|
153
|
|
|
{ |
|
154
|
9 |
|
$entityName = $entity->getEntity(); |
|
155
|
9 |
|
$pageKey = $record->meta ? $entity->keyThreads : $entity->keyAlternativePagination; |
|
156
|
9 |
|
if (!isset($this->metaData[$entityName])) { |
|
157
|
9 |
|
$this->metaData[$entityName] = [ |
|
158
|
9 |
|
$pageKey => null, |
|
159
|
9 |
|
'perPage' => $record->meta['perPage'], |
|
160
|
|
|
]; |
|
161
|
9 |
|
return; |
|
162
|
|
|
} |
|
163
|
9 |
|
if ($record->meta) { |
|
164
|
6 |
|
if ($this->metaData[$entityName][$pageKey] == $record->meta['currentPage']) { |
|
165
|
6 |
|
return; |
|
166
|
|
|
} |
|
167
|
6 |
|
$this->metaData[$entityName][$pageKey] = $record->meta['currentPage']; |
|
168
|
2 |
|
} else { |
|
169
|
6 |
|
$id = $record->data['id']; |
|
170
|
6 |
|
if ($i % $this->iterationCount != 0) { |
|
171
|
6 |
|
return; |
|
172
|
|
|
} |
|
173
|
3 |
|
$this->metaData[$entityName][$pageKey] = $id; |
|
174
|
|
|
} |
|
175
|
9 |
|
$this->saveMetaData(); |
|
176
|
9 |
|
} |
|
177
|
|
|
|
|
178
|
|
|
/** |
|
179
|
|
|
* Записать в мета данные об успешном сохранение сущности. |
|
180
|
|
|
* |
|
181
|
|
|
* @param AbstractList $entity |
|
182
|
|
|
*/ |
|
183
|
12 |
|
protected function finishParseEntity(AbstractList $entity) |
|
184
|
|
|
{ |
|
185
|
12 |
|
$entityName = $entity->getEntity(); |
|
186
|
12 |
|
if (!isset($this->metaData[$entityName])) { |
|
187
|
3 |
|
$this->metaData[$entityName] = []; |
|
188
|
1 |
|
} |
|
189
|
12 |
|
$this->metaData[$entityName]['finish'] = true; |
|
190
|
12 |
|
} |
|
191
|
|
|
|
|
192
|
|
|
/** |
|
193
|
|
|
* Сохранить мета данные. |
|
194
|
|
|
*/ |
|
195
|
12 |
|
protected function saveMetaData() |
|
196
|
|
|
{ |
|
197
|
12 |
|
$data = json_encode($this->metaData); |
|
198
|
12 |
|
file_put_contents($this->metaFilename, $data); |
|
199
|
12 |
|
} |
|
200
|
|
|
} |
|
201
|
|
|
|
sima-land /
api-php-client
