|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/* |
|
4
|
|
|
* This file is part of YaEtl. |
|
5
|
|
|
* (c) Fabrice de Stefanis / https://github.com/fab2s/YaEtl |
|
6
|
|
|
* This source file is licensed under the MIT license which you will |
|
7
|
|
|
* find in the LICENSE file or at https://opensource.org/licenses/MIT |
|
8
|
|
|
*/ |
|
9
|
|
|
|
|
10
|
|
|
namespace fab2s\YaEtl\Extractors\File; |
|
11
|
|
|
|
|
12
|
|
|
use fab2s\NodalFlow\NodalFlowException; |
|
13
|
|
|
use fab2s\NodalFlow\YaEtlException; |
|
14
|
|
|
use fab2s\OpinHelpers\Bom; |
|
15
|
|
|
use fab2s\YaEtl\Extractors\ExtractorAbstract; |
|
16
|
|
|
use fab2s\YaEtl\Traits\FileHandlerTrait; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* Class FileExtractorAbstract |
|
20
|
|
|
*/ |
|
21
|
|
|
abstract class FileExtractorAbstract extends ExtractorAbstract |
|
22
|
|
|
{ |
|
23
|
|
|
use FileHandlerTrait; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* @param resource|string $input |
|
27
|
|
|
* |
|
28
|
|
|
* @throws YaEtlException |
|
29
|
|
|
* @throws NodalFlowException |
|
30
|
|
|
*/ |
|
31
|
|
|
public function __construct($input) |
|
32
|
|
|
{ |
|
33
|
|
|
$this->initHandle($input, 'rb'); |
|
34
|
|
|
|
|
35
|
|
|
parent::__construct(); |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* @param mixed $param |
|
40
|
|
|
* |
|
41
|
|
|
* @return bool |
|
42
|
|
|
*/ |
|
43
|
|
|
public function extract($param = null) |
|
44
|
|
|
{ |
|
45
|
|
|
$this->getCarrier()->getFlowMap()->incrementNode($this->getId(), 'num_extract'); |
|
46
|
|
|
|
|
47
|
|
|
return !feof($this->handle); |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* @return bool |
|
52
|
|
|
*/ |
|
53
|
|
|
protected function readBom() |
|
54
|
|
|
{ |
|
55
|
|
|
if (false === ($buffer = $this->getNextNonEmptyChars())) { |
|
56
|
|
|
return false; |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
|
|
/* @var string $buffer */ |
|
60
|
|
|
$firstCharPos = ftell($this->handle); |
|
61
|
|
|
if (false === ($chars = fread($this->handle, 3))) { |
|
62
|
|
|
return false; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
/* @var string $chars */ |
|
66
|
|
|
$buffer .= $chars; |
|
67
|
|
|
if ($bom = Bom::extract($buffer)) { |
|
68
|
|
|
$this->encoding = Bom::getBomEncoding($bom); |
|
69
|
|
|
|
|
70
|
|
|
return !fseek($this->handle, $firstCharPos + strlen($bom) - 1); |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
return !fseek($this->handle, $firstCharPos - 1); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* @return string|false |
|
78
|
|
|
*/ |
|
79
|
|
|
protected function getNextNonEmptyLine() |
|
80
|
|
|
{ |
|
81
|
|
|
while (false !== ($line = fgets($this->handle))) { |
|
82
|
|
|
if ('' === ($line = trim($line))) { |
|
83
|
|
|
continue; |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
return $line; |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
return false; |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
/** |
|
93
|
|
|
* @return string|false |
|
94
|
|
|
*/ |
|
95
|
|
|
protected function getNextNonEmptyChars() |
|
96
|
|
|
{ |
|
97
|
|
|
do { |
|
98
|
|
|
if (false === ($char = fread($this->handle, 1))) { |
|
99
|
|
|
return false; |
|
100
|
|
|
} |
|
101
|
|
|
} while (trim($char) === ''); |
|
102
|
|
|
|
|
103
|
|
|
return $char; |
|
104
|
|
|
} |
|
105
|
|
|
} |
|
106
|
|
|
|