|
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
|
|
|
use CSVelte\Dialect; |
|
16
|
|
|
|
|
17
|
|
|
use function Noz\collect; |
|
18
|
|
|
use function Stringy\create as s; |
|
19
|
|
|
|
|
20
|
|
|
class SniffQuoteStyle extends AbstractSniffer |
|
21
|
|
|
{ |
|
22
|
|
|
/** |
|
23
|
|
|
* Guess quoting style |
|
24
|
|
|
* |
|
25
|
|
|
* The quoting style refers to which types of columns are quoted within a csv dataset. The dialect class defines |
|
26
|
|
|
* four possible quoting styles; all, none, minimal, or non-numeric. This class attempts to determine which of those |
|
27
|
|
|
* four it is by analyzing the content within each quoted value. |
|
28
|
|
|
* |
|
29
|
|
|
* @param string $data The data to analyze |
|
30
|
|
|
* |
|
31
|
|
|
* @return int |
|
32
|
|
|
*/ |
|
33
|
1 |
|
public function sniff($data) |
|
34
|
|
|
{ |
|
35
|
1 |
|
$styles = collect([ |
|
36
|
1 |
|
Dialect::QUOTE_NONE => true, |
|
37
|
1 |
|
Dialect::QUOTE_ALL => true, |
|
38
|
1 |
|
Dialect::QUOTE_MINIMAL => true, |
|
39
|
1 |
|
Dialect::QUOTE_NONNUMERIC => true |
|
40
|
1 |
|
]); |
|
41
|
|
|
|
|
42
|
1 |
|
$delimiter = $this->getOption('delimiter'); |
|
43
|
1 |
|
$lineTerminator = $this->getOption('lineTerminator') ?: "\n"; |
|
44
|
1 |
|
$quoted = collect(); |
|
45
|
1 |
|
collect(explode($lineTerminator, $this->replaceQuotedSpecialChars($data, $delimiter, $lineTerminator))) |
|
46
|
|
|
->each(function($line, $line_no) use (&$styles, $quoted, $delimiter) { |
|
|
|
|
|
|
47
|
1 |
|
$values = explode($delimiter, $line); |
|
48
|
1 |
|
foreach ($values as $value) { |
|
49
|
1 |
|
if ($this->isQuoted($value)) { |
|
50
|
|
|
// remove surrounding quotes |
|
51
|
1 |
|
$value = $this->unQuote($value); |
|
52
|
1 |
|
$styles[Dialect::QUOTE_NONE] = false; |
|
53
|
1 |
|
if (s($value)->containsAny([static::PLACEHOLDER_DELIM, static::PLACEHOLDER_NEWLINE, '"', "'"])) { |
|
54
|
1 |
|
$quoted->add(Dialect::QUOTE_MINIMAL); |
|
55
|
1 |
|
} elseif (!is_numeric((string) $value)) { |
|
56
|
1 |
|
$quoted->add(Dialect::QUOTE_NONNUMERIC); |
|
57
|
1 |
|
} else { |
|
58
|
1 |
|
$quoted->add(Dialect::QUOTE_ALL); |
|
59
|
|
|
} |
|
60
|
1 |
|
} else { |
|
61
|
1 |
|
$styles[Dialect::QUOTE_ALL] = false; |
|
62
|
|
|
} |
|
63
|
1 |
|
} |
|
64
|
1 |
|
}); |
|
65
|
|
|
|
|
66
|
|
|
// @todo the following can almost certainly be cleaned up considerably |
|
67
|
1 |
|
if ($styles[Dialect::QUOTE_ALL]) { |
|
68
|
1 |
|
return Dialect::QUOTE_ALL; |
|
69
|
1 |
|
} elseif ($styles[Dialect::QUOTE_NONE]) { |
|
70
|
1 |
|
return Dialect::QUOTE_NONE; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
1 |
|
$types = $quoted->distinct(); |
|
74
|
|
|
|
|
75
|
1 |
|
if ($types->contains(Dialect::QUOTE_NONNUMERIC) && $types->contains(Dialect::QUOTE_MINIMAL)) { |
|
76
|
|
|
// if both non-numeric and minimal then it isn't minimal |
|
77
|
|
|
$types = $types->filter(function($type) { |
|
78
|
|
|
return $type !== Dialect::QUOTE_MINIMAL; |
|
79
|
|
|
}); |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
1 |
|
if ($types->count() == 1) { |
|
83
|
1 |
|
return $types->getValueAt(1); |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
return Dialect::QUOTE_MINIMAL; |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
|
|
/** |
|
90
|
|
|
* Determine whether a particular string of data has quotes around it. |
|
91
|
|
|
* |
|
92
|
|
|
* @param string $data The data to check |
|
93
|
|
|
* |
|
94
|
|
|
* @return bool |
|
95
|
|
|
*/ |
|
96
|
1 |
|
protected function isQuoted($data) |
|
97
|
|
|
{ |
|
98
|
1 |
|
return preg_match('/^([\'"])[^\1]*\1$/', $data); |
|
99
|
|
|
} |
|
100
|
|
|
} |
nozavroni /
csvelte
This check looks from parameters that have been defined for a function or method, but which are not used in the method body.