|
1
|
|
|
<?php |
|
2
|
|
|
namespace frictionlessdata\tableschema\Fields; |
|
3
|
|
|
|
|
4
|
|
|
use frictionlessdata\tableschema\Exceptions\FieldValidationException; |
|
5
|
|
|
|
|
6
|
|
|
class FieldsInferrer |
|
7
|
|
|
{ |
|
8
|
|
|
/** |
|
9
|
|
|
* @param null|array $rows optional initial rows to infer by, each row is an array of field name => field value |
|
10
|
|
|
*/ |
|
11
|
|
|
public function __construct($rows=null, $lenient=false) |
|
12
|
|
|
{ |
|
13
|
|
|
$this->lenient = $lenient; |
|
14
|
|
|
if (!empty($rows)) { |
|
15
|
|
|
$this->addRows($rows); |
|
16
|
|
|
} |
|
17
|
|
|
} |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* add rows and updates the fieldsPopularity array - to make the inferred fields more accurate |
|
21
|
|
|
* @param $rows |
|
22
|
|
|
* @throws FieldValidationException |
|
23
|
|
|
*/ |
|
24
|
|
|
public function addRows($rows) |
|
25
|
|
|
{ |
|
26
|
|
|
foreach ($rows as $row) { |
|
27
|
|
|
$this->inputRows[] = $row; |
|
28
|
|
|
$inferredRow = $this->inferRow($row); |
|
29
|
|
|
foreach ($this->getFieldNames() as $fieldName) { |
|
30
|
|
|
/** @var BaseField $inferredField */ |
|
31
|
|
|
$inferredField = $inferredRow[$fieldName]; |
|
32
|
|
|
$inferredFieldType = $inferredField->getInferIdentifier($this->lenient); |
|
33
|
|
|
if (!array_key_exists($fieldName, $this->fieldsPopularity)) { |
|
34
|
|
|
$this->fieldsPopularity[$fieldName] = []; |
|
35
|
|
|
$this->fieldsPopularityObjects[$fieldName] = []; |
|
36
|
|
|
} |
|
37
|
|
|
if (!array_key_exists($inferredFieldType, $this->fieldsPopularity[$fieldName])) { |
|
38
|
|
|
$this->fieldsPopularity[$fieldName][$inferredFieldType] = 0; |
|
39
|
|
|
$this->fieldsPopularityObjects[$fieldName][$inferredFieldType] = $inferredField; |
|
40
|
|
|
} |
|
41
|
|
|
$this->fieldsPopularity[$fieldName][$inferredFieldType]++; |
|
42
|
|
|
arsort($this->fieldsPopularity[$fieldName]); |
|
43
|
|
|
} |
|
44
|
|
|
} |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* return the best inferred fields along with the best value casting according to the rows received so far |
|
49
|
|
|
* @return array field name => inferred field object |
|
50
|
|
|
* @throws FieldValidationException |
|
51
|
|
|
*/ |
|
52
|
|
|
public function infer() |
|
53
|
|
|
{ |
|
54
|
|
|
$bestInferredFields = []; |
|
55
|
|
|
foreach ($this->fieldsPopularity as $fieldName => $fieldTypesPopularity) { |
|
56
|
|
|
$bestInferredFields[$fieldName] = $this->inferField($fieldName, $fieldTypesPopularity); |
|
57
|
|
|
} |
|
58
|
|
|
return $bestInferredFields; |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
/** |
|
62
|
|
|
* returns all the input rows got so far with the best cast value for each field |
|
63
|
|
|
* @return array of arrays of field name => best cast value |
|
64
|
|
|
*/ |
|
65
|
|
|
public function castRows() |
|
66
|
|
|
{ |
|
67
|
|
|
return $this->castRows; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
protected $inputRows = []; |
|
71
|
|
|
protected $castRows = []; |
|
72
|
|
|
protected $fieldsPopularity = []; |
|
73
|
|
|
protected $fieldsPopularityObjects = []; |
|
74
|
|
|
protected $lenient; |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* infer field objects for the given row |
|
78
|
|
|
* raises exception if fails to infer a field |
|
79
|
|
|
* @param $row array field name => value to infer by |
|
80
|
|
|
* @return array field name => inferred field object |
|
81
|
|
|
* @throws FieldValidationException |
|
82
|
|
|
*/ |
|
83
|
|
|
protected function inferRow($row) |
|
84
|
|
|
{ |
|
85
|
|
|
$rowFields = []; |
|
86
|
|
|
foreach ($row as $k => $v) { |
|
87
|
|
|
$rowFields[$k] = FieldsFactory::infer($v, (object)["name" => $k], $this->lenient); |
|
88
|
|
|
} |
|
89
|
|
|
return $rowFields; |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
/** |
|
93
|
|
|
* @return array |
|
94
|
|
|
*/ |
|
95
|
|
|
protected function getFieldNames() |
|
96
|
|
|
{ |
|
97
|
|
|
// we assume csv file where all rows have the same column positions |
|
98
|
|
|
// so we can use the first row to get the field names |
|
99
|
|
|
return array_keys($this->inputRows[0]); |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
/** |
|
103
|
|
|
* finds the best inferred fields for the given field name according to the popularity |
|
104
|
|
|
* also updates the castRows array with the latest cast values |
|
105
|
|
|
* @param $fieldName |
|
106
|
|
|
* @param $fieldTypesPopularity |
|
107
|
|
|
* @return BaseField|null |
|
108
|
|
|
*/ |
|
109
|
|
|
protected function inferField($fieldName, $fieldTypesPopularity) |
|
110
|
|
|
{ |
|
111
|
|
|
// the $fieldTypesPopularity array is already sorted with most popular fields first |
|
112
|
|
|
$inferredField = null; |
|
113
|
|
|
foreach (array_keys($fieldTypesPopularity) as $inferredFieldType) { |
|
114
|
|
|
/** @var BaseField $inferredField */ |
|
115
|
|
|
$inferredField = $this->fieldsPopularityObjects[$fieldName][$inferredFieldType]; |
|
116
|
|
|
try { |
|
117
|
|
|
$rowNum = 0; |
|
118
|
|
|
foreach ($this->inputRows as $inputRow) { |
|
119
|
|
|
if (!array_key_exists($rowNum, $this->castRows)) $this->castRows[$rowNum] = []; |
|
120
|
|
|
$this->castRows[$rowNum][$fieldName] = $inferredField->castValue($inputRow[$fieldName]); |
|
121
|
|
|
$rowNum++; |
|
122
|
|
|
} |
|
123
|
|
|
break; |
|
124
|
|
|
} catch (FieldValidationException $e) { |
|
125
|
|
|
// a row failed validation for this field type, will continue to the next one according to popularity |
|
126
|
|
|
continue; |
|
127
|
|
|
} |
|
128
|
|
|
} |
|
129
|
|
|
return $inferredField; |
|
130
|
|
|
} |
|
131
|
|
|
} |
OriHoch /
tableschema-php
