|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* CSVelte: Slender, elegant CSV for PHP |
|
4
|
|
|
* |
|
5
|
|
|
* Inspired by Python's CSV module and Frictionless Data and the W3C's CSV |
|
6
|
|
|
* standardization efforts, CSVelte was written in an effort to take all the |
|
7
|
|
|
* suck out of working with CSV. |
|
8
|
|
|
* |
|
9
|
|
|
* @copyright Copyright (c) 2018 Luke Visinoni |
|
10
|
|
|
* @author Luke Visinoni <[email protected]> |
|
11
|
|
|
* @license See LICENSE file (MIT license) |
|
12
|
|
|
*/ |
|
13
|
|
|
namespace CSVelte\Sniffer; |
|
14
|
|
|
|
|
15
|
|
|
abstract class AbstractSniffer |
|
16
|
|
|
{ |
|
17
|
|
|
/** |
|
18
|
|
|
* Placeholder strings -- hold the place of newlines and delimiters contained |
|
19
|
|
|
* within quoted text so that the explode method doesn't split incorrectly. |
|
20
|
|
|
*/ |
|
21
|
|
|
const PLACEHOLDER_NEWLINE = '[__NEWLINE__]'; |
|
22
|
|
|
const PLACEHOLDER_DELIM = '[__DELIMIT__]'; |
|
23
|
|
|
|
|
24
|
|
|
protected $options = []; |
|
25
|
|
|
|
|
26
|
7 |
|
public function __construct(array $options = []) |
|
27
|
|
|
{ |
|
28
|
7 |
|
$this->setOptions($options); |
|
29
|
7 |
|
} |
|
30
|
|
|
|
|
31
|
7 |
|
protected function setOptions(array $options) |
|
32
|
|
|
{ |
|
33
|
7 |
|
$this->options = array_merge($this->options, $options); |
|
34
|
7 |
|
return $this; |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
protected function setOption($option, $value) |
|
38
|
|
|
{ |
|
39
|
|
|
if (array_key_exists($option, $this->options)) { |
|
40
|
|
|
$this->options[$option] = $value; |
|
41
|
|
|
} |
|
42
|
|
|
return $this; |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
6 |
|
protected function getOption($option) |
|
46
|
|
|
{ |
|
47
|
6 |
|
if (array_key_exists($option, $this->options)) { |
|
48
|
6 |
|
return $this->options[$option]; |
|
49
|
|
|
}; |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
/** |
|
53
|
|
|
* Replace all instances of newlines and whatever character you specify (as |
|
54
|
|
|
* the delimiter) that are contained within quoted text. The replacements are |
|
55
|
|
|
* simply a special placeholder string. |
|
56
|
|
|
* |
|
57
|
|
|
* @param string $data The string to do the replacements on |
|
58
|
|
|
* @param string $delim The delimiter character to replace |
|
59
|
|
|
* |
|
60
|
|
|
* @return string The data with replacements performed |
|
61
|
|
|
*/ |
|
62
|
3 |
|
protected function replaceQuotedSpecialChars($data, $delim = null, $eol = null) |
|
63
|
|
|
{ |
|
64
|
3 |
|
if (is_null($eol)) { |
|
65
|
2 |
|
$eol = "\r\n|\r|\n"; |
|
66
|
2 |
|
} |
|
67
|
3 |
|
return preg_replace_callback('/([\'"])(.*)\1/imsU', function ($matches) use ($delim, $eol) { |
|
68
|
3 |
|
$ret = preg_replace("/({$eol})/", static::PLACEHOLDER_NEWLINE, $matches[0]); |
|
69
|
3 |
|
if (!is_null($delim)) { |
|
70
|
3 |
|
$ret = str_replace($delim, static::PLACEHOLDER_DELIM, $ret); |
|
71
|
3 |
|
} |
|
72
|
3 |
|
return $ret; |
|
73
|
3 |
|
}, $data); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* Replaces all quoted columns with a blank string. I was using this method |
|
78
|
|
|
* to prevent explode() from incorrectly splitting at delimiters and newlines |
|
79
|
|
|
* within quotes when parsing a file. But this was before I wrote the |
|
80
|
|
|
* replaceQuotedSpecialChars method which (at least to me) makes more sense. |
|
81
|
|
|
* |
|
82
|
|
|
* @param string $data The string to replace quoted strings within |
|
83
|
|
|
* |
|
84
|
|
|
* @return string The input string with quoted strings removed |
|
85
|
|
|
*/ |
|
86
|
3 |
|
protected function removeQuotedStrings($data) |
|
87
|
|
|
{ |
|
88
|
3 |
|
return preg_replace($pattern = '/(["\'])(?:(?=(\\\\?))\2.)*?\1/sm', $replace = '', $data); |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
2 |
|
protected function unQuote($string) |
|
92
|
|
|
{ |
|
93
|
2 |
|
return preg_replace('/^(["\'])(.*)\1$/', '\2', (string) $string); |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
/** |
|
97
|
|
|
* Analyze data (sniff) |
|
98
|
|
|
* |
|
99
|
|
|
* @param string $data The data to analyze (sniff) |
|
100
|
|
|
* |
|
101
|
|
|
* @return string|string[] |
|
102
|
|
|
*/ |
|
103
|
|
|
abstract public function sniff($data); |
|
104
|
|
|
} |
nozavroni /
csvelte
