|
1
|
|
|
<?php |
|
2
|
|
|
declare(strict_types=1); |
|
3
|
|
|
|
|
4
|
|
|
namespace Level23\Druid\Tasks; |
|
5
|
|
|
|
|
6
|
|
|
use Level23\Druid\Context\TaskContext; |
|
7
|
|
|
use Level23\Druid\Transforms\TransformSpec; |
|
8
|
|
|
use Level23\Druid\Dimensions\TimestampSpec; |
|
9
|
|
|
use Level23\Druid\TuningConfig\TuningConfig; |
|
10
|
|
|
use Level23\Druid\Collections\AggregationCollection; |
|
11
|
|
|
use Level23\Druid\InputSources\InputSourceInterface; |
|
12
|
|
|
use Level23\Druid\InputFormats\InputFormatInterface; |
|
13
|
|
|
use Level23\Druid\Granularities\GranularityInterface; |
|
14
|
|
|
use Level23\Druid\Collections\SpatialDimensionCollection; |
|
15
|
|
|
|
|
16
|
|
|
class IndexTask implements TaskInterface |
|
17
|
|
|
{ |
|
18
|
|
|
protected ?TuningConfig $tuningConfig; |
|
19
|
|
|
|
|
20
|
|
|
protected ?TaskContext $context; |
|
21
|
|
|
|
|
22
|
|
|
protected bool $appendToExisting = false; |
|
23
|
|
|
|
|
24
|
|
|
protected InputSourceInterface $inputSource; |
|
25
|
|
|
|
|
26
|
|
|
protected string $dateSource; |
|
27
|
|
|
|
|
28
|
|
|
protected GranularityInterface $granularity; |
|
29
|
|
|
|
|
30
|
|
|
protected ?TransformSpec $transformSpec; |
|
31
|
|
|
|
|
32
|
|
|
protected ?AggregationCollection $aggregations; |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* @var array<array<string,string|bool>> |
|
36
|
|
|
*/ |
|
37
|
|
|
protected array $dimensions = []; |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* Whether this task should be executed parallel. |
|
41
|
|
|
* |
|
42
|
|
|
* @var bool |
|
43
|
|
|
*/ |
|
44
|
|
|
protected bool $parallel = false; |
|
45
|
|
|
|
|
46
|
|
|
protected ?string $taskId; |
|
47
|
|
|
|
|
48
|
|
|
protected ?InputFormatInterface $inputFormat; |
|
49
|
|
|
|
|
50
|
|
|
protected TimestampSpec $timestampSpec; |
|
51
|
|
|
|
|
52
|
|
|
protected ?SpatialDimensionCollection $spatialDimensions = null; |
|
53
|
|
|
|
|
54
|
|
|
/** |
|
55
|
|
|
* IndexTask constructor. |
|
56
|
|
|
* |
|
57
|
|
|
* @param string $dateSource |
|
58
|
|
|
* @param \Level23\Druid\InputSources\InputSourceInterface $inputSource |
|
59
|
|
|
* @param \Level23\Druid\Granularities\GranularityInterface $granularity |
|
60
|
|
|
* @param \Level23\Druid\Dimensions\TimestampSpec $timestampSpec |
|
61
|
|
|
* @param \Level23\Druid\Transforms\TransformSpec|null $transformSpec |
|
62
|
|
|
* @param \Level23\Druid\TuningConfig\TuningConfig|null $tuningConfig |
|
63
|
|
|
* @param \Level23\Druid\Context\TaskContext|null $context |
|
64
|
|
|
* @param \Level23\Druid\Collections\AggregationCollection|null $aggregations |
|
65
|
|
|
* @param array<array<string,string|bool>> $dimensions |
|
66
|
|
|
* @param string|null $taskId |
|
67
|
|
|
* @param \Level23\Druid\InputFormats\InputFormatInterface|null $inputFormat |
|
68
|
|
|
* @param \Level23\Druid\Collections\SpatialDimensionCollection|null $spatialDimensions |
|
69
|
|
|
*/ |
|
70
|
3 |
|
public function __construct( |
|
71
|
|
|
string $dateSource, |
|
72
|
|
|
InputSourceInterface $inputSource, |
|
73
|
|
|
GranularityInterface $granularity, |
|
74
|
|
|
TimestampSpec $timestampSpec, |
|
75
|
|
|
?TransformSpec $transformSpec = null, |
|
76
|
|
|
?TuningConfig $tuningConfig = null, |
|
77
|
|
|
?TaskContext $context = null, |
|
78
|
|
|
?AggregationCollection $aggregations = null, |
|
79
|
|
|
array $dimensions = [], |
|
80
|
|
|
?string $taskId = null, |
|
81
|
|
|
?InputFormatInterface $inputFormat = null, |
|
82
|
|
|
?SpatialDimensionCollection $spatialDimensions = null |
|
83
|
|
|
) { |
|
84
|
3 |
|
$this->tuningConfig = $tuningConfig; |
|
85
|
3 |
|
$this->context = $context; |
|
86
|
3 |
|
$this->inputSource = $inputSource; |
|
87
|
3 |
|
$this->dateSource = $dateSource; |
|
88
|
3 |
|
$this->granularity = $granularity; |
|
89
|
3 |
|
$this->transformSpec = $transformSpec; |
|
90
|
3 |
|
$this->aggregations = $aggregations; |
|
91
|
3 |
|
$this->dimensions = $dimensions; |
|
92
|
3 |
|
$this->taskId = $taskId; |
|
93
|
3 |
|
$this->inputFormat = $inputFormat; |
|
94
|
3 |
|
$this->timestampSpec = $timestampSpec; |
|
95
|
3 |
|
$this->spatialDimensions = $spatialDimensions; |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* Return the task in a format so that we can send it to druid. |
|
100
|
|
|
* |
|
101
|
|
|
* @return array<string,array<string,array<string,array<int|string,array<mixed>|bool|int|string>|bool|int|string|null>|bool|int|string>|string> |
|
102
|
|
|
*/ |
|
103
|
2 |
|
public function toArray(): array |
|
104
|
|
|
{ |
|
105
|
2 |
|
$result = [ |
|
106
|
2 |
|
'type' => $this->parallel ? 'index_parallel' : 'index', |
|
107
|
2 |
|
'spec' => [ |
|
108
|
2 |
|
'dataSchema' => [ |
|
109
|
2 |
|
'dataSource' => $this->dateSource, |
|
110
|
2 |
|
'timestampSpec' => $this->timestampSpec->toArray(), |
|
111
|
2 |
|
'dimensionsSpec' => [ |
|
112
|
2 |
|
'dimensions' => $this->dimensions, |
|
113
|
2 |
|
], |
|
114
|
2 |
|
'metricsSpec' => ($this->aggregations?->toArray()), |
|
115
|
2 |
|
'granularitySpec' => $this->granularity->toArray(), |
|
116
|
2 |
|
'transformSpec' => ($this->transformSpec?->toArray()), |
|
117
|
2 |
|
], |
|
118
|
2 |
|
'ioConfig' => [ |
|
119
|
2 |
|
'type' => $this->parallel ? 'index_parallel' : 'index', |
|
120
|
2 |
|
'inputSource' => $this->inputSource->toArray(), |
|
121
|
2 |
|
'inputFormat' => $this->inputFormat?->toArray(), |
|
122
|
2 |
|
'appendToExisting' => $this->appendToExisting, |
|
123
|
2 |
|
], |
|
124
|
2 |
|
], |
|
125
|
2 |
|
]; |
|
126
|
|
|
|
|
127
|
|
|
// Add our spatial dimensions if supplied. |
|
128
|
2 |
|
if (!empty($this->spatialDimensions)) { |
|
129
|
1 |
|
$result['spec']['dataSchema']['dimensionsSpec']['spatialDimensions'] = $this->spatialDimensions->toArray(); |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
2 |
|
$context = $this->context ? $this->context->toArray() : []; |
|
133
|
2 |
|
if (count($context) > 0) { |
|
134
|
1 |
|
$result['context'] = $context; |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
2 |
|
if ($this->tuningConfig) { |
|
138
|
1 |
|
$this->tuningConfig->setType($this->parallel ? 'index_parallel' : 'index'); |
|
139
|
1 |
|
$result['spec']['tuningConfig'] = $this->tuningConfig->toArray(); |
|
140
|
|
|
} |
|
141
|
|
|
|
|
142
|
2 |
|
if ($this->taskId) { |
|
143
|
1 |
|
$result['id'] = $this->taskId; |
|
144
|
|
|
} |
|
145
|
|
|
|
|
146
|
2 |
|
return $result; |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
/** |
|
150
|
|
|
* @param bool $appendToExisting |
|
151
|
|
|
*/ |
|
152
|
2 |
|
public function setAppendToExisting(bool $appendToExisting): void |
|
153
|
|
|
{ |
|
154
|
2 |
|
$this->appendToExisting = $appendToExisting; |
|
155
|
|
|
} |
|
156
|
|
|
|
|
157
|
|
|
/** |
|
158
|
|
|
* Whether this task should be executed parallel. |
|
159
|
|
|
* |
|
160
|
|
|
* @param bool $parallel |
|
161
|
|
|
*/ |
|
162
|
2 |
|
public function setParallel(bool $parallel): void |
|
163
|
|
|
{ |
|
164
|
2 |
|
$this->parallel = $parallel; |
|
165
|
|
|
} |
|
166
|
|
|
|
|
167
|
|
|
/** |
|
168
|
|
|
* @param InputFormatInterface $inputFormat |
|
169
|
|
|
* |
|
170
|
|
|
* @return IndexTask |
|
171
|
|
|
*/ |
|
172
|
2 |
|
public function setInputFormat(InputFormatInterface $inputFormat): IndexTask |
|
173
|
|
|
{ |
|
174
|
2 |
|
$this->inputFormat = $inputFormat; |
|
175
|
|
|
|
|
176
|
2 |
|
return $this; |
|
177
|
|
|
} |
|
178
|
|
|
} |
level23 /
druid-client
