|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace ByJG\AnyDataset\Dataset; |
|
4
|
|
|
|
|
5
|
|
|
use ByJG\AnyDataset\Exception\DatasetException; |
|
6
|
|
|
use ByJG\AnyDataset\Exception\NotFoundException; |
|
7
|
|
|
use Exception; |
|
8
|
|
|
use InvalidArgumentException; |
|
9
|
|
|
|
|
10
|
|
|
class TextFileDataset |
|
11
|
|
|
{ |
|
12
|
|
|
|
|
13
|
|
|
const CSVFILE = '/[|,;](?=(?:[^"]*"[^"]*")*(?![^"]*"))/'; |
|
14
|
|
|
const CSVFILE_SEMICOLON = '/[;](?=(?:[^"]*"[^"]*")*(?![^"]*"))/'; |
|
15
|
|
|
const CSVFILE_COMMA = '/[,](?=(?:[^"]*"[^"]*")*(?![^"]*"))/'; |
|
16
|
|
|
|
|
17
|
|
|
protected $source; |
|
18
|
|
|
protected $fields; |
|
19
|
|
|
protected $fieldexpression; |
|
20
|
|
|
protected $sourceType; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* Text File Data Set |
|
24
|
|
|
* |
|
25
|
|
|
* @param string $source |
|
26
|
|
|
* @param array $fields |
|
27
|
|
|
* @param string $fieldexpression |
|
28
|
|
|
* @throws NotFoundException |
|
29
|
|
|
*/ |
|
30
|
15 |
|
public function __construct($source, $fields, $fieldexpression = null) |
|
31
|
|
|
{ |
|
32
|
15 |
|
if (is_null($fieldexpression)) { |
|
33
|
|
|
$fieldexpression = TextFileDataset::CSVFILE; |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
15 |
|
if (!is_array($fields)) { |
|
37
|
|
|
throw new InvalidArgumentException("You must define an array of fields."); |
|
38
|
|
|
} |
|
39
|
15 |
|
if (!preg_match('~(http|https|ftp)://~', $source)) { |
|
40
|
9 |
|
$this->source = $source; |
|
41
|
|
|
|
|
42
|
9 |
|
if (!file_exists($this->source)) { |
|
43
|
1 |
|
throw new NotFoundException("The specified file " . $this->source . " does not exists"); |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
8 |
|
$this->sourceType = "FILE"; |
|
47
|
|
|
} else { |
|
48
|
6 |
|
$this->source = $source; |
|
49
|
6 |
|
$this->sourceType = "HTTP"; |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
|
|
53
|
14 |
|
$this->fields = $fields; |
|
54
|
|
|
|
|
55
|
14 |
|
if ($fieldexpression == 'CSVFILE') { |
|
56
|
|
|
$this->fieldexpression = TextFileDataset::CSVFILE; |
|
57
|
|
|
} else { |
|
58
|
14 |
|
$this->fieldexpression = $fieldexpression; |
|
59
|
|
|
} |
|
60
|
14 |
|
} |
|
61
|
|
|
|
|
62
|
|
|
/** |
|
63
|
|
|
* @access public |
|
64
|
|
|
* @return GenericIterator |
|
65
|
|
|
* @throws DatasetException |
|
66
|
|
|
* @throws Exception |
|
67
|
|
|
*/ |
|
68
|
14 |
|
public function getIterator() |
|
69
|
|
|
{ |
|
70
|
14 |
|
$old = ini_set('auto_detect_line_endings', true); |
|
71
|
14 |
|
$handle = @fopen($this->source, "r"); |
|
72
|
14 |
|
ini_set('auto_detect_line_endings', $old); |
|
73
|
14 |
|
if (!$handle) { |
|
74
|
2 |
|
throw new DatasetException("TextFileDataset failed to open resource"); |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
try { |
|
78
|
12 |
|
$iterator = new TextFileIterator($handle, $this->fields, $this->fieldexpression); |
|
79
|
12 |
|
return $iterator; |
|
80
|
|
|
} catch (Exception $ex) { |
|
81
|
|
|
fclose($handle); |
|
82
|
|
|
throw $ex; |
|
83
|
|
|
} |
|
84
|
|
|
} |
|
85
|
|
|
} |
|
86
|
|
|
|
byjg /
anydataset
