|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* @package s9e\TextFormatter |
|
5
|
|
|
* @copyright Copyright (c) 2010-2018 The s9e Authors |
|
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
|
7
|
|
|
*/ |
|
8
|
|
|
namespace s9e\TextFormatter\Configurator\Helpers; |
|
9
|
|
|
|
|
10
|
|
|
use RuntimeException; |
|
11
|
|
|
use s9e\TextFormatter\Utils\XPath; |
|
12
|
|
|
|
|
13
|
|
|
abstract class XPathHelper |
|
14
|
|
|
{ |
|
15
|
|
|
/** |
|
16
|
|
|
* Return the list of variables used in a given XPath expression |
|
17
|
|
|
* |
|
18
|
|
|
* @param string $expr XPath expression |
|
19
|
|
|
* @return array Alphabetically sorted list of unique variable names |
|
20
|
|
|
*/ |
|
21
|
|
|
public static function getVariables($expr) |
|
22
|
|
|
{ |
|
23
|
|
|
// First, remove strings' contents to prevent false-positives |
|
24
|
|
|
$expr = preg_replace('/(["\']).*?\\1/s', '$1$1', $expr); |
|
25
|
|
|
|
|
26
|
|
|
// Capture all the variable names |
|
27
|
|
|
preg_match_all('/\\$(\\w+)/', $expr, $matches); |
|
28
|
|
|
|
|
29
|
|
|
// Dedupe and sort names |
|
30
|
|
|
$varNames = array_unique($matches[1]); |
|
31
|
|
|
sort($varNames); |
|
32
|
|
|
|
|
33
|
|
|
return $varNames; |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* Determine whether given XPath expression definitely evaluates to a number |
|
38
|
|
|
* |
|
39
|
|
|
* @param string $expr XPath expression |
|
40
|
|
|
* @return bool Whether given XPath expression definitely evaluates to a number |
|
41
|
|
|
*/ |
|
42
|
|
|
public static function isExpressionNumeric($expr) |
|
43
|
|
|
{ |
|
44
|
|
|
// Trim the expression and remove parentheses that are not part of a function call. PCRE |
|
45
|
|
|
// does not support lookbehind assertions of variable length so we have to flip the string. |
|
46
|
|
|
// We exclude the XPath operator "div" (flipped into "vid") to avoid false positives |
|
47
|
|
|
$expr = strrev(preg_replace('(\\((?!\\s*(?!vid(?!\\w))\\w))', ' ', strrev($expr))); |
|
48
|
|
|
$expr = str_replace(')', ' ', $expr); |
|
49
|
|
|
if (preg_match('(^\\s*([$@][-\\w]++|-?\\.\\d++|-?\\d++(?:\\.\\d++)?)(?>\\s*(?>[-+*]|div)\\s*(?1))++\\s*$)', $expr)) |
|
50
|
|
|
{ |
|
51
|
|
|
return true; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
return false; |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
/** |
|
58
|
|
|
* Remove extraneous space in a given XPath expression |
|
59
|
|
|
* |
|
60
|
|
|
* @param string $expr Original XPath expression |
|
61
|
|
|
* @return string Minified XPath expression |
|
62
|
|
|
*/ |
|
63
|
|
|
public static function minify($expr) |
|
64
|
|
|
{ |
|
65
|
|
|
$old = $expr; |
|
66
|
|
|
$strings = []; |
|
67
|
|
|
|
|
68
|
|
|
// Trim the surrounding whitespace then temporarily remove literal strings |
|
69
|
|
|
$expr = preg_replace_callback( |
|
70
|
|
|
'/"[^"]*"|\'[^\']*\'/', |
|
71
|
|
|
function ($m) use (&$strings) |
|
72
|
|
|
{ |
|
73
|
|
|
$uniqid = '(' . sha1(uniqid()) . ')'; |
|
74
|
|
|
$strings[$uniqid] = $m[0]; |
|
75
|
|
|
|
|
76
|
|
|
return $uniqid; |
|
77
|
|
|
}, |
|
78
|
|
|
trim($expr) |
|
79
|
|
|
); |
|
80
|
|
|
|
|
81
|
|
|
if (preg_match('/[\'"]/', $expr)) |
|
82
|
|
|
{ |
|
83
|
|
|
throw new RuntimeException("Cannot parse XPath expression '" . $old . "'"); |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
// Normalize whitespace to a single space |
|
87
|
|
|
$expr = preg_replace('/\\s+/', ' ', $expr); |
|
88
|
|
|
|
|
89
|
|
|
// Remove the space between a non-word character and a word character |
|
90
|
|
|
$expr = preg_replace('/([-a-z_0-9]) ([^-a-z_0-9])/i', '$1$2', $expr); |
|
91
|
|
|
$expr = preg_replace('/([^-a-z_0-9]) ([-a-z_0-9])/i', '$1$2', $expr); |
|
92
|
|
|
|
|
93
|
|
|
// Remove the space between two non-word characters as long as they're not two - |
|
94
|
|
|
$expr = preg_replace('/(?!- -)([^-a-z_0-9]) ([^-a-z_0-9])/i', '$1$2', $expr); |
|
95
|
|
|
|
|
96
|
|
|
// Remove the space between a - and a word character, as long as there's a space before - |
|
97
|
|
|
$expr = preg_replace('/ - ([a-z_0-9])/i', ' -$1', $expr); |
|
98
|
|
|
|
|
99
|
|
|
// Remove the spaces between a number and the div operator and the next token |
|
100
|
|
|
$expr = preg_replace('/((?:^|[ \\(])\\d+) div ?/', '$1div', $expr); |
|
101
|
|
|
|
|
102
|
|
|
// Remove the space between the div operator the next token |
|
103
|
|
|
$expr = preg_replace('/([^-a-z_0-9]div) (?=[$0-9@])/', '$1', $expr); |
|
104
|
|
|
|
|
105
|
|
|
// Restore the literals |
|
106
|
|
|
$expr = strtr($expr, $strings); |
|
107
|
|
|
|
|
108
|
|
|
return $expr; |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
/** |
|
112
|
|
|
* Parse an XPath expression that is composed entirely of equality tests between a variable part |
|
113
|
|
|
* and a constant part |
|
114
|
|
|
* |
|
115
|
|
|
* @param string $expr |
|
116
|
|
|
* @return array|false |
|
117
|
|
|
*/ |
|
118
|
|
|
public static function parseEqualityExpr($expr) |
|
119
|
|
|
{ |
|
120
|
|
|
// Match an equality between a variable and a literal or the concatenation of strings |
|
121
|
|
|
$eq = '(?<equality>' |
|
122
|
|
|
. '(?<key>@[-\\w]+|\\$\\w+|\\.)' |
|
123
|
|
|
. '(?<operator>\\s*=\\s*)' |
|
124
|
|
|
. '(?:' |
|
125
|
|
|
. '(?<literal>(?<string>"[^"]*"|\'[^\']*\')|0|[1-9][0-9]*)' |
|
126
|
|
|
. '|' |
|
127
|
|
|
. '(?<concat>concat\\(\\s*(?&string)\\s*(?:,\\s*(?&string)\\s*)+\\))' |
|
128
|
|
|
. ')' |
|
129
|
|
|
. '|' |
|
130
|
|
|
. '(?:(?<literal>(?&literal))|(?<concat>(?&concat)))(?&operator)(?<key>(?&key))' |
|
131
|
|
|
. ')'; |
|
132
|
|
|
|
|
133
|
|
|
// Match a string that is entirely composed of equality checks separated with "or" |
|
134
|
|
|
$regexp = '(^(?J)\\s*' . $eq . '\\s*(?:or\\s*(?&equality)\\s*)*$)'; |
|
135
|
|
|
|
|
136
|
|
|
if (!preg_match($regexp, $expr)) |
|
137
|
|
|
{ |
|
138
|
|
|
return false; |
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
|
|
preg_match_all("((?J)$eq)", $expr, $matches, PREG_SET_ORDER); |
|
142
|
|
|
|
|
143
|
|
|
$map = []; |
|
144
|
|
|
foreach ($matches as $m) |
|
145
|
|
|
{ |
|
146
|
|
|
$key = $m['key']; |
|
147
|
|
|
if (!empty($m['concat'])) |
|
148
|
|
|
{ |
|
149
|
|
|
preg_match_all('(\'[^\']*\'|"[^"]*")', $m['concat'], $strings); |
|
150
|
|
|
|
|
151
|
|
|
$value = ''; |
|
152
|
|
|
foreach ($strings[0] as $string) |
|
153
|
|
|
{ |
|
154
|
|
|
$value .= substr($string, 1, -1); |
|
155
|
|
|
} |
|
156
|
|
|
} |
|
157
|
|
|
else |
|
158
|
|
|
{ |
|
159
|
|
|
$value = $m['literal']; |
|
160
|
|
|
if ($value[0] === "'" || $value[0] === '"') |
|
161
|
|
|
{ |
|
162
|
|
|
$value = substr($value, 1, -1); |
|
163
|
|
|
} |
|
164
|
|
|
} |
|
165
|
|
|
|
|
166
|
|
|
$map[$key][] = $value; |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
return $map; |
|
170
|
|
|
} |
|
171
|
|
|
} |
s9e /
TextFormatter
