1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Sheerockoff\BitrixElastic; |
4
|
|
|
|
5
|
|
|
use Elasticsearch\Client; |
6
|
|
|
use Exception; |
7
|
|
|
use InvalidArgumentException; |
8
|
|
|
use stdClass; |
9
|
|
|
|
10
|
|
|
class Finder |
11
|
|
|
{ |
12
|
|
|
private const LOGIC_AND = 'AND'; |
13
|
|
|
private const LOGIC_OR = 'OR'; |
14
|
|
|
|
15
|
|
|
/** @var Client */ |
16
|
|
|
private $elastic; |
17
|
|
|
|
18
|
|
|
/** @var Mapper */ |
19
|
|
|
private $mapper; |
20
|
|
|
|
21
|
|
|
/** @var bool */ |
22
|
|
|
private $strictMode; |
23
|
|
|
|
24
|
28 |
|
public function __construct(Client $elastic, Mapper $mapper, $strictMode = true) |
25
|
|
|
{ |
26
|
28 |
|
$this->elastic = $elastic; |
27
|
28 |
|
$this->mapper = $mapper; |
28
|
28 |
|
$this->strictMode = $strictMode; |
29
|
|
|
} |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* @throws Exception |
33
|
|
|
*/ |
34
|
22 |
|
public function search(string $index, array $filter, array $sort = ['SORT' => 'ASC', 'ID' => 'DESC'], array $parameters = []): array |
35
|
|
|
{ |
36
|
22 |
|
$params = $this->prefabElasticSearchParams($index, $filter, $sort, $parameters); |
37
|
12 |
|
return $this->elastic->search($params); |
38
|
|
|
} |
39
|
|
|
|
40
|
|
|
/** |
41
|
|
|
* @throws Exception |
42
|
|
|
*/ |
43
|
23 |
|
public function prefabElasticSearchParams(string $index, array $filter, array $sort = ['SORT' => 'ASC', 'ID' => 'DESC'], array $parameters = []): array |
44
|
|
|
{ |
45
|
23 |
|
$mapping = $this->mapper->getCachedMapping($index); |
46
|
23 |
|
$filter = $this->prefabFilter($filter); |
47
|
23 |
|
$filter = $this->normalizeFilter($mapping, $filter); |
48
|
20 |
|
$query = $this->prepareFilterQuery($filter); |
49
|
|
|
|
50
|
15 |
|
$params = ['index' => $index, 'body' => ['query' => $query]]; |
51
|
15 |
|
$params['body'] = array_merge($params['body'], $this->normalizeSort($mapping, $sort)); |
52
|
|
|
|
53
|
13 |
|
return array_merge($params, $parameters); |
54
|
|
|
} |
55
|
|
|
|
56
|
23 |
|
private function prefabFilter(array $filter): array |
57
|
|
|
{ |
58
|
23 |
|
$changedFilter = $filter; |
59
|
23 |
|
$includeSubsections = false; |
60
|
23 |
|
foreach ($filter as $key => $rawValue) { |
61
|
18 |
|
if (!preg_match('/^(?<operator>\W*)(?<property>\w+)$/ui', $key, $matches)) { |
62
|
2 |
|
continue; |
63
|
|
|
} |
64
|
|
|
|
65
|
16 |
|
if ($matches['property'] !== 'INCLUDE_SUBSECTIONS') { |
66
|
16 |
|
continue; |
67
|
|
|
} |
68
|
|
|
|
69
|
1 |
|
$includeSubsections = $rawValue && $rawValue !== 'N'; |
70
|
1 |
|
if (array_key_exists($key, $changedFilter)) { |
71
|
1 |
|
unset($changedFilter[$key]); |
72
|
|
|
} |
73
|
|
|
|
74
|
1 |
|
break; |
75
|
|
|
} |
76
|
|
|
|
77
|
23 |
|
foreach ($filter as $key => $rawValue) { |
78
|
18 |
|
if (!preg_match('/^(?<operator>\W*)(?<property>\w+)$/ui', $key, $matches)) { |
79
|
2 |
|
continue; |
80
|
|
|
} |
81
|
|
|
|
82
|
16 |
|
if ($matches['property'] === 'IBLOCK_SECTION_ID' || $matches['property'] === 'SECTION_ID') { |
83
|
1 |
|
if (array_key_exists($key, $changedFilter)) { |
84
|
1 |
|
unset($changedFilter[$key]); |
85
|
|
|
} |
86
|
|
|
|
87
|
1 |
|
if ($includeSubsections) { |
88
|
1 |
|
$changedFilter['NAV_CHAIN_IDS'] = $rawValue; |
89
|
|
|
} else { |
90
|
1 |
|
$changedFilter['GROUP_IDS'] = $rawValue; |
91
|
|
|
} |
92
|
16 |
|
} elseif ($matches['property'] === 'SECTION_CODE') { |
93
|
1 |
|
if (array_key_exists($key, $changedFilter)) { |
94
|
1 |
|
unset($changedFilter[$key]); |
95
|
|
|
} |
96
|
|
|
|
97
|
1 |
|
if ($includeSubsections) { |
98
|
1 |
|
$changedFilter['NAV_CHAIN_CODES'] = $rawValue; |
99
|
|
|
} else { |
100
|
1 |
|
$changedFilter['GROUP_CODES'] = $rawValue; |
101
|
|
|
} |
102
|
|
|
} |
103
|
|
|
} |
104
|
|
|
|
105
|
23 |
|
return $changedFilter; |
106
|
|
|
} |
107
|
|
|
|
108
|
|
|
/** |
109
|
|
|
* @throws InvalidArgumentException |
110
|
|
|
*/ |
111
|
27 |
|
private function normalizeFilter(IndexMapping $mapping, array $filter): array |
112
|
|
|
{ |
113
|
27 |
|
$normalizedFilter = []; |
114
|
27 |
|
foreach ($filter as $k => $v) { |
115
|
22 |
|
if (is_array($v) && array_key_exists('LOGIC', $v)) { |
116
|
8 |
|
$subFilter = $v; |
117
|
8 |
|
unset($subFilter['LOGIC']); |
118
|
8 |
|
$subFilter = $this->normalizeFilter($mapping, $subFilter); |
119
|
7 |
|
$normalizedFilter[$k] = array_merge(['LOGIC' => $v['LOGIC']], $subFilter); |
120
|
7 |
|
continue; |
121
|
|
|
} |
122
|
|
|
|
123
|
22 |
|
if (!preg_match('/^(?<operator>\W*)(?<property>\w+)$/ui', $k, $matches)) { |
124
|
4 |
|
if ($this->strictMode) throw new InvalidArgumentException("Неверный ключ фильтра ($k)."); |
125
|
2 |
|
continue; |
126
|
|
|
} |
127
|
|
|
|
128
|
18 |
|
$property = $matches['property']; |
129
|
|
|
|
130
|
18 |
|
if (!$mapping->getProperties()->offsetExists($property)) { |
131
|
2 |
|
if ($this->strictMode) throw new InvalidArgumentException("$property не найден в карте индекса."); |
132
|
1 |
|
continue; |
133
|
|
|
} |
134
|
|
|
|
135
|
16 |
|
$isAlias = $mapping->getProperty($property)->get('type') === 'alias'; |
136
|
16 |
|
if ($isAlias) { |
137
|
5 |
|
$aliasPath = $mapping->getProperty($property)->get('path'); |
138
|
5 |
|
if (!$aliasPath) { |
139
|
2 |
|
if ($this->strictMode) throw new InvalidArgumentException("В $property типа alias не указан path."); |
140
|
1 |
|
continue; |
141
|
|
|
} |
142
|
|
|
|
143
|
3 |
|
if (!$mapping->getProperties()->offsetExists($aliasPath)) { |
144
|
2 |
|
if ($this->strictMode) { |
145
|
1 |
|
throw new InvalidArgumentException( |
146
|
1 |
|
"$property типа alias указывает на $aliasPath, который не найден в карте индекса." |
147
|
1 |
|
); |
148
|
|
|
} |
149
|
|
|
|
150
|
1 |
|
continue; |
151
|
|
|
} |
152
|
|
|
|
153
|
1 |
|
$property = $aliasPath; |
154
|
|
|
} |
155
|
|
|
|
156
|
12 |
|
if (is_array($v)) { |
157
|
5 |
|
$value = array_map(function ($v) use ($mapping, $property) { |
158
|
5 |
|
return $mapping->getProperty($property)->normalizeValue($v); |
159
|
5 |
|
}, $v); |
160
|
|
|
} else { |
161
|
8 |
|
$value = $mapping->getProperty($property)->normalizeValue($v); |
162
|
|
|
} |
163
|
|
|
|
164
|
12 |
|
$normalizedFilter[$k] = $value; |
165
|
|
|
} |
166
|
|
|
|
167
|
22 |
|
return $normalizedFilter; |
168
|
|
|
} |
169
|
|
|
|
170
|
|
|
/** |
171
|
|
|
* @throws InvalidArgumentException |
172
|
|
|
*/ |
173
|
15 |
|
private function normalizeSort(IndexMapping $mapping, array $sort): array |
174
|
|
|
{ |
175
|
15 |
|
$elasticSorts = []; |
176
|
15 |
|
foreach ($sort as $property => $term) { |
177
|
15 |
|
if (!$mapping->getProperties()->offsetExists($property)) { |
178
|
2 |
|
if ($this->strictMode) throw new InvalidArgumentException("$property не найден в карте индекса для сортировки."); |
179
|
1 |
|
continue; |
180
|
|
|
} |
181
|
|
|
|
182
|
13 |
|
$emptySort = null; |
183
|
13 |
|
if (preg_match('/(?<first>nulls\s*,\s*)?(?<dir>asc|desc)(?<last>\s*,\s*nulls)?/ui', $term, $matches)) { |
184
|
11 |
|
$sortOrder = strtolower($matches['dir']); |
185
|
11 |
|
if (!empty($matches['first'])) { |
186
|
1 |
|
$emptySort = 'asc'; |
187
|
11 |
|
} elseif (!empty($matches['last'])) { |
188
|
11 |
|
$emptySort = 'desc'; |
189
|
|
|
} |
190
|
|
|
} else { |
191
|
2 |
|
if ($this->strictMode) throw new InvalidArgumentException("Неверный формат сортировки ($term)."); |
192
|
1 |
|
continue; |
193
|
|
|
} |
194
|
|
|
|
195
|
11 |
|
$propMap = $mapping->getProperty($property); |
196
|
11 |
|
if (isset($propMap->getData()['fields']['enum'])) { |
197
|
1 |
|
$sortField = $property . '_VALUE'; |
198
|
|
|
} else { |
199
|
11 |
|
$sortField = $property; |
200
|
|
|
} |
201
|
|
|
|
202
|
11 |
|
if ($emptySort) { |
203
|
1 |
|
$emptyValuesForTypes = [ |
204
|
1 |
|
'integer' => '0', |
205
|
1 |
|
'long' => '0', |
206
|
1 |
|
'float' => '0.0', |
207
|
1 |
|
'double' => '0.0', |
208
|
1 |
|
'date' => 'null', |
209
|
1 |
|
'boolean' => 'false', |
210
|
1 |
|
'string' => '', |
211
|
1 |
|
]; |
212
|
|
|
|
213
|
1 |
|
$elasticSorts[] = [ |
214
|
1 |
|
'_script' => [ |
215
|
1 |
|
'type' => 'number', |
216
|
1 |
|
'script' => [ |
217
|
1 |
|
'lang' => 'painless', |
218
|
1 |
|
'source' => sprintf( |
219
|
1 |
|
"if (doc['%s'].empty || doc['%s'].value == %s) { return 0; } else { return 1; }", |
220
|
1 |
|
$sortField, |
221
|
1 |
|
$sortField, |
222
|
1 |
|
$emptyValuesForTypes[$propMap->get('type')] ?? 'null' |
223
|
1 |
|
) |
224
|
1 |
|
], |
225
|
1 |
|
'order' => $emptySort |
226
|
1 |
|
] |
227
|
1 |
|
]; |
228
|
|
|
} |
229
|
|
|
|
230
|
11 |
|
$elasticSorts[] = [$sortField => ['order' => $sortOrder]]; |
231
|
|
|
} |
232
|
|
|
|
233
|
13 |
|
return $elasticSorts ? ['sort' => $elasticSorts] : []; |
234
|
|
|
} |
235
|
|
|
|
236
|
|
|
/** |
237
|
|
|
* @throws InvalidArgumentException |
238
|
|
|
*/ |
239
|
22 |
|
private function prepareFilterQuery(array $filter, string $logic = self::LOGIC_AND): array |
240
|
|
|
{ |
241
|
22 |
|
if (!in_array($logic, [self::LOGIC_AND, self::LOGIC_OR], true)) { |
242
|
1 |
|
throw new InvalidArgumentException("Неверный логический оператор ($logic)."); |
243
|
|
|
} |
244
|
|
|
|
245
|
22 |
|
$terms = []; |
246
|
22 |
|
foreach ($filter as $k => $value) { |
247
|
15 |
|
if (is_array($value) && array_key_exists('LOGIC', $value)) { |
248
|
7 |
|
$subFilter = $value; |
249
|
7 |
|
unset($subFilter['LOGIC']); |
250
|
7 |
|
$subQuery = $this->prepareFilterQuery($subFilter, strtoupper($value['LOGIC'])); |
251
|
4 |
|
$terms = array_merge_recursive($terms, [($logic === self::LOGIC_OR ? 'should' : 'must') => [$subQuery]]); |
252
|
|
|
} else { |
253
|
13 |
|
if (!preg_match('/^(?<operator>\W*)(?<field>\w+)$/ui', $k, $matches)) { |
254
|
2 |
|
if ($this->strictMode) throw new InvalidArgumentException("Неверный ключ фильтра ($k)."); |
255
|
1 |
|
continue; |
256
|
|
|
} |
257
|
|
|
|
258
|
11 |
|
$operator = $matches['operator'] ?: '='; |
259
|
11 |
|
$field = $matches['field']; |
260
|
|
|
|
261
|
|
|
try { |
262
|
11 |
|
if ($logic === self::LOGIC_OR) { |
263
|
5 |
|
$entry = $this->or($field, $operator, $value); |
264
|
|
|
} else { |
265
|
7 |
|
$entry = $this->and($field, $operator, $value); |
266
|
|
|
} |
267
|
8 |
|
} catch (Exception $exception) { |
268
|
8 |
|
if ($this->strictMode) throw $exception; |
269
|
4 |
|
continue; |
270
|
|
|
} |
271
|
|
|
|
272
|
3 |
|
$terms = array_merge_recursive($terms, $entry); |
273
|
|
|
} |
274
|
|
|
} |
275
|
|
|
|
276
|
16 |
|
return $terms ? ['bool' => $terms] : ['match_all' => new stdClass()]; |
277
|
|
|
} |
278
|
|
|
|
279
|
|
|
/** |
280
|
|
|
* @throws InvalidArgumentException |
281
|
|
|
*/ |
282
|
7 |
|
private function and(string $field, string $operator, $value) |
283
|
|
|
{ |
284
|
7 |
|
$handlers = [ |
285
|
7 |
|
'=' => function ($field, $value) { |
286
|
2 |
|
if ($value === false || $value === null) { |
287
|
1 |
|
return ['must_not' => [['exists' => ['field' => $field]]]]; |
288
|
|
|
} else { |
289
|
2 |
|
return ['must' => [[(is_array($value) ? 'terms' : 'term') => [$field => $value]]]]; |
290
|
|
|
} |
291
|
7 |
|
}, |
292
|
7 |
|
'!' => function ($field, $value) { |
293
|
1 |
|
if ($value === false || $value === null) { |
294
|
1 |
|
return ['must' => [['exists' => ['field' => $field]]]]; |
295
|
|
|
} else { |
296
|
1 |
|
return ['must_not' => [[(is_array($value) ? 'terms' : 'term') => [$field => $value]]]]; |
297
|
|
|
} |
298
|
7 |
|
}, |
299
|
7 |
|
'%' => function ($field, $value) { |
300
|
1 |
|
return ['must' => [['wildcard' => [$field => "*$value*"]]]]; |
301
|
7 |
|
}, |
302
|
7 |
|
'>' => function ($field, $value) { |
303
|
2 |
|
return ['must' => [['range' => [$field => ['gt' => $value]]]]]; |
304
|
7 |
|
}, |
305
|
7 |
|
'>=' => function ($field, $value) { |
306
|
1 |
|
return ['must' => [['range' => [$field => ['gte' => $value]]]]]; |
307
|
7 |
|
}, |
308
|
7 |
|
'<' => function ($field, $value) { |
309
|
1 |
|
return ['must' => [['range' => [$field => ['lt' => $value]]]]]; |
310
|
7 |
|
}, |
311
|
7 |
|
'<=' => function ($field, $value) { |
312
|
1 |
|
return ['must' => [['range' => [$field => ['lte' => $value]]]]]; |
313
|
7 |
|
}, |
314
|
7 |
|
'><' => function ($field, $value) { |
315
|
3 |
|
if (!isset($value[0]) || !isset($value[1])) { |
316
|
2 |
|
throw new InvalidArgumentException("Для фильтра ><$field должен быть указан массив значений."); |
317
|
|
|
} |
318
|
|
|
|
319
|
1 |
|
return ['must' => [['range' => [$field => ['gte' => min($value), 'lte' => max($value)]]]]]; |
320
|
7 |
|
} |
321
|
7 |
|
]; |
322
|
|
|
|
323
|
7 |
|
if (!array_key_exists($operator, $handlers)) { |
324
|
2 |
|
throw new InvalidArgumentException("Невозможно отфильтровать $field по оператору $operator."); |
325
|
|
|
} |
326
|
|
|
|
327
|
5 |
|
return $handlers[$operator]($field, $value); |
328
|
|
|
} |
329
|
|
|
|
330
|
|
|
/** |
331
|
|
|
* @throws InvalidArgumentException |
332
|
|
|
*/ |
333
|
5 |
|
private function or($field, $operator, $value) |
334
|
|
|
{ |
335
|
5 |
|
$handlers = [ |
336
|
5 |
|
'=' => function ($field, $value) { |
337
|
1 |
|
if ($value === false || $value === null) { |
338
|
1 |
|
return ['should' => [['bool' => ['must_not' => [['exists' => ['field' => $field]]]]]]]; |
339
|
|
|
} else { |
340
|
1 |
|
return ['should' => [[(is_array($value) ? 'terms' : 'term') => [$field => $value]]]]; |
341
|
|
|
} |
342
|
5 |
|
}, |
343
|
5 |
|
'!' => function ($field, $value) { |
344
|
1 |
|
if ($value === false || $value === null) { |
345
|
1 |
|
return ['should' => [['exists' => ['field' => $field]]]]; |
346
|
|
|
} else { |
347
|
1 |
|
return ['should' => [['bool' => ['must_not' => [[(is_array($value) ? 'terms' : 'term') => [$field => $value]]]]]]]; |
348
|
|
|
} |
349
|
5 |
|
}, |
350
|
5 |
|
'%' => function ($field, $value) { |
351
|
1 |
|
return ['should' => [['wildcard' => [$field => "*$value*"]]]]; |
352
|
5 |
|
}, |
353
|
5 |
|
'>' => function ($field, $value) { |
354
|
1 |
|
return ['should' => [['range' => [$field => ['gt' => $value]]]]]; |
355
|
5 |
|
}, |
356
|
5 |
|
'>=' => function ($field, $value) { |
357
|
1 |
|
return ['should' => [['range' => [$field => ['gte' => $value]]]]]; |
358
|
5 |
|
}, |
359
|
5 |
|
'<' => function ($field, $value) { |
360
|
1 |
|
return ['should' => [['range' => [$field => ['lt' => $value]]]]]; |
361
|
5 |
|
}, |
362
|
5 |
|
'<=' => function ($field, $value) { |
363
|
1 |
|
return ['should' => [['range' => [$field => ['lte' => $value]]]]]; |
364
|
5 |
|
}, |
365
|
5 |
|
'><' => function ($field, $value) { |
366
|
3 |
|
if (!isset($value[0]) || !isset($value[1])) { |
367
|
2 |
|
throw new InvalidArgumentException("Для фильтра ><$field должен быть указан массив значений."); |
368
|
|
|
} |
369
|
|
|
|
370
|
1 |
|
return ['should' => [['range' => [$field => ['gte' => min($value), 'lte' => max($value)]]]]]; |
371
|
5 |
|
} |
372
|
5 |
|
]; |
373
|
|
|
|
374
|
5 |
|
if (!array_key_exists($operator, $handlers)) { |
375
|
2 |
|
throw new InvalidArgumentException("Невозможно отфильтровать $field по оператору $operator."); |
376
|
|
|
} |
377
|
|
|
|
378
|
3 |
|
return $handlers[$operator]($field, $value); |
379
|
|
|
} |
380
|
|
|
} |