1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace frictionlessdata\tableschema\DataSources; |
4
|
|
|
|
5
|
|
|
use frictionlessdata\tableschema\Exceptions\DataSourceException; |
6
|
|
|
|
7
|
|
|
/** |
8
|
|
|
* handles reading data from a csv source |
9
|
|
|
* responsible for finding the header row based on options |
10
|
|
|
* support skipping rows from the csv. |
11
|
|
|
*/ |
12
|
|
|
class CsvDataSource extends BaseDataSource |
13
|
|
|
{ |
14
|
|
|
/** |
15
|
|
|
* @throws DataSourceException |
16
|
|
|
*/ |
17
|
|
|
public function open() |
18
|
|
|
{ |
19
|
|
|
$this->curRowNum = 0; |
20
|
|
|
try { |
21
|
|
|
$this->resource = fopen($this->dataSource, 'r'); |
22
|
|
|
} catch (\Exception $e) { |
23
|
|
|
throw new DataSourceException($e->getMessage()); |
24
|
|
|
} |
25
|
|
|
$this->headerRow = $this->getOption('headerRow'); |
26
|
|
|
if ($this->headerRow) { |
27
|
|
|
$headerRowNum = 0; |
28
|
|
|
$defaultSkipRows = 0; |
29
|
|
|
} else { |
30
|
|
|
$defaultSkipRows = $headerRowNum = $this->getOption('headerRowNum', 1); |
31
|
|
|
} |
32
|
|
|
$skipRows = $this->getOption('skipRows', $defaultSkipRows); |
33
|
|
|
if ($skipRows > 0) { |
34
|
|
|
foreach (range(1, $skipRows) as $i) { |
35
|
|
|
$row = $this->getRow(); |
36
|
|
|
$this->skippedRows[] = $row; |
37
|
|
|
if ($i == $headerRowNum) { |
38
|
|
|
$this->headerRow = $row; |
39
|
|
|
} |
40
|
|
|
} |
41
|
|
|
} |
42
|
|
|
if (!$this->headerRow) { |
43
|
|
|
throw new DataSourceException('Failed to get header row'); |
44
|
|
|
} |
45
|
|
|
} |
46
|
|
|
|
47
|
|
|
/** |
48
|
|
|
* @return array |
49
|
|
|
*/ |
50
|
|
|
public function getSkippedRows() |
51
|
|
|
{ |
52
|
|
|
return $this->skippedRows; |
53
|
|
|
} |
54
|
|
|
|
55
|
|
|
/** |
56
|
|
|
* @return array |
57
|
|
|
* |
58
|
|
|
* @throws DataSourceException |
59
|
|
|
*/ |
60
|
|
|
public function getNextLine() |
61
|
|
|
{ |
62
|
|
|
$row = $this->getRow(); |
63
|
|
|
$colNum = 0; |
64
|
|
|
$obj = []; |
65
|
|
|
foreach ($this->headerRow as $fieldName) { |
66
|
|
|
$obj[$fieldName] = $row[$colNum++]; |
67
|
|
|
} |
68
|
|
|
|
69
|
|
|
return $obj; |
70
|
|
|
} |
71
|
|
|
|
72
|
|
|
/** |
73
|
|
|
* @return bool |
74
|
|
|
* |
75
|
|
|
* @throws DataSourceException |
76
|
|
|
*/ |
77
|
|
|
public function isEof() |
78
|
|
|
{ |
79
|
|
|
try { |
80
|
|
|
return feof($this->resource); |
81
|
|
|
} catch (\Exception $e) { |
82
|
|
|
throw new DataSourceException($e->getMessage(), $this->curRowNum); |
83
|
|
|
} |
84
|
|
|
} |
85
|
|
|
|
86
|
|
|
/** |
87
|
|
|
* @throws DataSourceException |
88
|
|
|
*/ |
89
|
|
|
public function close() |
90
|
|
|
{ |
91
|
|
|
try { |
92
|
|
|
fclose($this->resource); |
93
|
|
|
} catch (\Exception $e) { |
94
|
|
|
throw new DataSourceException($e->getMessage(), $this->curRowNum); |
95
|
|
|
} |
96
|
|
|
} |
97
|
|
|
|
98
|
|
|
public function save($outputDataSource) { |
99
|
|
|
$file = fopen($outputDataSource, 'w'); |
100
|
|
|
fputcsv($file, $this->headerRow); |
101
|
|
|
while (!$this->isEof()) { |
102
|
|
|
fputcsv($file, array_values($this->getNextLine())); |
103
|
|
|
} |
104
|
|
|
fclose($file); |
105
|
|
|
} |
106
|
|
|
|
107
|
|
|
protected $resource; |
108
|
|
|
protected $headerRow; |
109
|
|
|
protected $skippedRows; |
110
|
|
|
protected $curRowNum; |
111
|
|
|
|
112
|
|
|
/** |
113
|
|
|
* @return array |
114
|
|
|
* |
115
|
|
|
* @throws DataSourceException |
116
|
|
|
*/ |
117
|
|
|
protected function getRow() |
118
|
|
|
{ |
119
|
|
|
++$this->curRowNum; |
120
|
|
|
try { |
121
|
|
|
return fgetcsv($this->resource); |
122
|
|
|
} catch (\Exception $e) { |
123
|
|
|
throw new DataSourceException($e->getMessage(), $this->curRowNum); |
124
|
|
|
} |
125
|
|
|
} |
126
|
|
|
} |
127
|
|
|
|