|
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\Loaders\File; |
|
11
|
|
|
|
|
12
|
|
|
use fab2s\NodalFlow\NodalFlowException; |
|
13
|
|
|
use fab2s\NodalFlow\YaEtlException; |
|
14
|
|
|
use fab2s\YaEtl\Extractors\File\FileExtractorAbstract; |
|
15
|
|
|
use fab2s\YaEtl\Traits\CsvHandlerTrait; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* Class CsvExtractor |
|
19
|
|
|
*/ |
|
20
|
|
|
class CsvExtractor extends FileExtractorAbstract |
|
21
|
|
|
{ |
|
22
|
|
|
use CsvHandlerTrait; |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* CsvExtractor constructor |
|
26
|
|
|
* |
|
27
|
|
|
* @param resource|string $input |
|
28
|
|
|
* @param string $delimiter |
|
29
|
|
|
* @param string $enclosure |
|
30
|
|
|
* @param string $escape |
|
31
|
|
|
* |
|
32
|
|
|
* @throws NodalFlowException |
|
33
|
|
|
* @throws YaEtlException |
|
34
|
|
|
*/ |
|
35
|
|
|
public function __construct($input, $delimiter = ',', $enclosure = '"', $escape = '"') |
|
36
|
|
|
{ |
|
37
|
|
|
parent::__construct($input); |
|
38
|
|
|
$this->delimiter = $delimiter; |
|
39
|
|
|
$this->enclosure = $enclosure; |
|
40
|
|
|
$this->escape = $escape; |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
/** |
|
44
|
|
|
* @param mixed $param |
|
45
|
|
|
* |
|
46
|
|
|
* @return \Generator |
|
47
|
|
|
*/ |
|
48
|
|
|
public function getTraversable($param = null) |
|
49
|
|
|
{ |
|
50
|
|
|
while ($this->extract($param)) { |
|
51
|
|
|
if (!$this->readBom() || !$this->readSep() || false === ($firstRecord = $this->readHeader())) { |
|
52
|
|
|
return; |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
/* @var array $firstRecord */ |
|
56
|
|
|
yield $this->bakeRecord($firstRecord); |
|
57
|
|
|
while (false !== ($record = fgetcsv($this->handle, 0, $this->delimiter, $this->enclosure, $this->escape))) { |
|
58
|
|
|
/* @var array $record */ |
|
59
|
|
|
yield $this->bakeRecord($record); |
|
60
|
|
|
} |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
$this->releaseHandle(); |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
/** |
|
67
|
|
|
* @param array $record |
|
68
|
|
|
* |
|
69
|
|
|
* @return array |
|
70
|
|
|
*/ |
|
71
|
|
|
protected function bakeRecord($record) |
|
72
|
|
|
{ |
|
73
|
|
|
return isset($this->header) ? array_combine($this->header, $record) : $record; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* @return bool |
|
78
|
|
|
*/ |
|
79
|
|
|
protected function readHeader() |
|
80
|
|
|
{ |
|
81
|
|
|
if (false === ($firstRecord = $this->getNextNonEmptyRecord())) { |
|
82
|
|
|
return false; |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
if ($this->useHeader && !isset($this->header)) { |
|
86
|
|
|
$this->header = array_map('trim', $firstRecord); |
|
87
|
|
|
|
|
88
|
|
|
return $this->getNextNonEmptyRecord(); |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
return $firstRecord; |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
/** |
|
95
|
|
|
* @return bool |
|
96
|
|
|
*/ |
|
97
|
|
|
protected function readSep() |
|
98
|
|
|
{ |
|
99
|
|
|
if (false === ($firstChar = $this->getNextNonEmptyChars())) { |
|
100
|
|
|
return false; |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
$firstCharPos = ftell($this->handle); |
|
104
|
|
|
/* @var string $firstChar */ |
|
105
|
|
|
if ($firstChar === 's') { |
|
106
|
|
|
if (false === ($chars = fread($this->handle, 4))) { |
|
107
|
|
|
return false; |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
/* @var string $chars */ |
|
111
|
|
|
$line = $firstChar . $chars; |
|
112
|
|
|
|
|
113
|
|
|
if (strpos($line, 'sep=') === 0) { |
|
114
|
|
|
$this->useSep = true; |
|
115
|
|
|
$this->delimiter = $line[4]; |
|
116
|
|
|
|
|
117
|
|
|
return !fseek($this->handle, $firstCharPos + 5); |
|
118
|
|
|
} |
|
119
|
|
|
} |
|
120
|
|
|
|
|
121
|
|
|
return !fseek($this->handle, $firstCharPos - 1); |
|
122
|
|
|
} |
|
123
|
|
|
|
|
124
|
|
|
/** |
|
125
|
|
|
* @return array|false |
|
126
|
|
|
*/ |
|
127
|
|
|
protected function getNextNonEmptyRecord() |
|
128
|
|
|
{ |
|
129
|
|
|
$lastPos = ftell($this->handle); |
|
130
|
|
|
do { |
|
131
|
|
|
$record = fgetcsv($this->handle, 0, $this->delimiter, $this->enclosure, $this->escape); |
|
132
|
|
|
// since it is unclear if all PHP versions do return false, empty array or array(null) |
|
133
|
|
|
// for blank lines, we won't rely on false for EOF |
|
134
|
|
|
if (empty($record) || $record === [null]) { |
|
135
|
|
|
// at least make sure we go forward |
|
136
|
|
|
$pos = ftell($this->handle); |
|
137
|
|
|
if (feof($this->handle) || $pos <= $lastPos) { |
|
138
|
|
|
return false; |
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
|
|
$lastPos = $pos; |
|
142
|
|
|
continue; |
|
143
|
|
|
} |
|
144
|
|
|
|
|
145
|
|
|
return $record; |
|
146
|
|
|
} while (!feof($this->handle)); |
|
147
|
|
|
} |
|
148
|
|
|
} |
|
149
|
|
|
|
fab2s /
YaEtl
