1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace FluidXml; |
4
|
|
|
|
5
|
|
|
class CssTranslator |
6
|
|
|
{ |
7
|
|
|
const TOKEN = '/{([[:alpha:]]+)(\d+)}/i'; |
8
|
|
|
const MAP = [ |
9
|
|
|
// Empty part of #id and .class |
10
|
|
|
[ '(?<=^|\s) # The begining or an empty space. |
11
|
|
|
(?=[.#\[]) # . | # | [', |
12
|
|
|
'*', |
13
|
|
|
'TAG', |
14
|
|
|
false ], |
15
|
|
|
// #id |
16
|
|
|
[ '\# |
17
|
|
|
([\w\-]+)', |
18
|
|
|
'[@id="\1"]', |
19
|
|
|
'ID', |
20
|
|
|
false ], |
21
|
|
|
// .class |
22
|
|
|
[ '\. |
23
|
|
|
([\w\-]+)', |
24
|
|
|
'[ contains( concat(" ", normalize-space(@class), " "), concat(" ", "\1", " ") ) ]', |
25
|
|
|
'CLASS', |
26
|
|
|
false ], |
27
|
|
|
// [attr] |
28
|
|
|
[ '\[ |
29
|
|
|
([\w\-]+) |
30
|
|
|
\]', |
31
|
|
|
'[@\1]', |
32
|
|
|
'ATTR', |
33
|
|
|
false ], |
34
|
|
|
// [attr="val"] |
35
|
|
|
[ '\[ |
36
|
|
|
([\w\-]+) |
37
|
|
|
=["\'] |
38
|
|
|
(.*) |
39
|
|
|
["\'] |
40
|
|
|
\]', |
41
|
|
|
'[@\1="\2"]', |
42
|
|
|
'ATTR', |
43
|
|
|
false ], |
44
|
|
|
// ns|A |
45
|
|
|
[ '([\w\-]+) |
46
|
|
|
\| |
47
|
|
|
(?=\w) # A namespace must be followed at least by a character.', |
48
|
|
|
'\1:', |
49
|
|
|
'NS', |
50
|
|
|
false ], |
51
|
|
|
// :root |
52
|
|
|
[ ':root\b', |
53
|
|
|
'/*', |
54
|
|
|
'TAG', |
55
|
|
|
false ], |
56
|
|
|
// A |
57
|
|
|
[ '(?<=^|\s|\}) |
58
|
|
|
( [\w\-]+ | \* )', |
59
|
|
|
'\1', |
60
|
|
|
'TAG', |
61
|
|
|
false ], |
62
|
|
|
// Aggregates the components of a tag in an expression. |
63
|
|
|
[ '({NS\d+})? |
64
|
|
|
({TAG\d+}) |
65
|
|
|
((?:{ATTR\d+})*|) |
66
|
|
|
((?:{ID\d+})*|) |
67
|
|
|
((?:{CLASS\d+})*|)', |
68
|
|
|
'\1\2\3\4\5', |
69
|
|
|
'EXP', |
70
|
|
|
false ], |
71
|
|
|
[ '({EXP\d+}) |
72
|
|
|
:first-child', |
73
|
|
|
'*[1]/self::\1', |
74
|
|
|
'EXP', |
75
|
|
|
false ], |
76
|
|
|
// {} + {} |
77
|
|
|
[ '({EXP\d+}) |
78
|
|
|
\s* \+ \s* |
79
|
|
|
({EXP\d+})', |
80
|
|
|
'\1/following-sibling::*[1]/self::\2', |
81
|
|
|
'EXP', |
82
|
|
|
true ], |
83
|
|
|
// {} ~ {} |
84
|
|
|
[ '({EXP\d+}) |
85
|
|
|
\s* \~ \s* |
86
|
|
|
({EXP\d+})', |
87
|
|
|
'\1/following-sibling::*/self::\2', |
88
|
|
|
'EXP', |
89
|
|
|
true ], |
90
|
|
|
// {} > {} |
91
|
|
|
[ '({EXP\d+}) |
92
|
|
|
\s* > \s* |
93
|
|
|
({EXP\d+})', |
94
|
|
|
'\1/\2', |
95
|
|
|
'EXP', |
96
|
|
|
true ], |
97
|
|
|
// {} {} |
98
|
|
|
[ '({EXP\d+}) |
99
|
|
|
\s+ |
100
|
|
|
({EXP\d+})', |
101
|
|
|
'\1//\2', |
102
|
|
|
'EXP', |
103
|
|
|
true ], |
104
|
|
|
// {}, {} |
105
|
|
|
[ '({EXP\d+}) |
106
|
|
|
\s* , \s* |
107
|
|
|
({EXP\d+})', |
108
|
|
|
'\1|\2', |
109
|
|
|
'EXP', |
110
|
|
|
true ] |
111
|
|
|
]; |
112
|
|
|
|
113
|
|
|
public static function xpath($css) |
114
|
|
|
{ |
115
|
|
|
$xpath = $css; |
116
|
|
|
|
117
|
|
|
$stack = []; |
118
|
|
|
$index = 0; |
119
|
|
|
|
120
|
|
|
foreach (self::MAP as $o) { |
121
|
|
|
// The regexes have a common wrapper. |
122
|
|
|
list($search, $replace, $id, $repeat) = $o; |
123
|
|
|
$search = "/{$search}/xi"; |
124
|
|
|
|
125
|
|
|
do { |
126
|
|
|
$prev_xpath = $xpath; |
127
|
|
|
self::tokenize($search, $replace, $id, $xpath, $stack, $index); |
128
|
|
|
} while ($repeat && $xpath !== $prev_xpath); |
129
|
|
|
} |
130
|
|
|
|
131
|
|
|
self::translateStack($stack, $xpath); |
132
|
|
|
|
133
|
|
|
$xpath = \trim($xpath); |
134
|
|
|
$xpath = ".//$xpath"; |
135
|
|
|
$xpath = \str_replace('|', '|.//', $xpath); |
136
|
|
|
$xpath = \str_replace('.///', '/', $xpath); |
137
|
|
|
|
138
|
|
|
return $xpath; |
139
|
|
|
} |
140
|
|
|
|
141
|
|
|
protected static function tokenize($search, $replace, $id, &$xpath, &$stack, &$index) |
142
|
|
|
{ |
143
|
|
|
// The search can return 0, 1 or more fund patterns. |
144
|
|
|
$matches_count = \preg_match_all($search, $xpath, $matches, \PREG_SET_ORDER | \PREG_OFFSET_CAPTURE); |
145
|
|
|
|
146
|
|
|
// \PREG_OFFSET_CAPTURE calculates offsets starting from the begining of the string $xpath. |
147
|
|
|
// Rewriting $xpath from left (first matches) creates a mismatch between the calculated offsets |
148
|
|
|
// and the actual ones. This problem can be avoided rewriting $xpath from right (last matches). |
149
|
|
|
for ($i = $matches_count - 1; $i >= 0; --$i) { |
150
|
|
|
$match = $matches[$i]; |
151
|
|
|
|
152
|
|
|
// The count of the groups must exclude the entire recognized string entry. |
153
|
|
|
$groups_count = \count($match) - 1; |
154
|
|
|
|
155
|
|
|
// $match[0] is the entire recognized string. |
156
|
|
|
// $match[>0] are the captured groups. |
157
|
|
|
// $match[n][0] is the actual string. |
158
|
|
|
// $match[n][1] is the position of the string. |
159
|
|
|
list($pattern, $pattern_pos) = $match[0]; |
160
|
|
|
|
161
|
|
|
$xpath = \substr_replace($xpath, "{{$id}{$index}}", $pattern_pos, \strlen($pattern)); |
162
|
|
|
|
163
|
|
|
$groups_values = []; |
164
|
|
|
for ($ii = 1; $ii <= $groups_count; ++$ii) { |
165
|
|
|
// 0 is the group value, 1 is the group position. |
166
|
|
|
$groups_values[$ii] = $match[$ii][0]; |
167
|
|
|
} |
168
|
|
|
|
169
|
|
|
$stack[$index] = [$replace, $groups_values, $pattern]; |
170
|
|
|
++$index; |
171
|
|
|
} |
172
|
|
|
} |
173
|
|
|
|
174
|
|
|
protected static function translateStack(&$stack, &$xpath) |
175
|
|
|
{ |
176
|
|
|
do { |
177
|
|
|
$matches_count = \preg_match_all(self::TOKEN, $xpath, $matches, \PREG_SET_ORDER); |
178
|
|
|
|
179
|
|
|
for ($i = 0; $i < $matches_count; ++$i) { |
180
|
|
|
list($pattern, $type, $id) = $matches[$i]; |
|
|
|
|
181
|
|
|
|
182
|
|
|
$id = \intval($id); |
183
|
|
|
|
184
|
|
|
list($replace, $groups) = $stack[$id]; |
185
|
|
|
|
186
|
|
|
foreach ($groups as $k => $v) { |
187
|
|
|
$replace = \str_replace("\\$k", $v, $replace); |
188
|
|
|
} |
189
|
|
|
|
190
|
|
|
$xpath = \str_replace("{{$type}{$id}}", $replace, $xpath); |
191
|
|
|
} |
192
|
|
|
} while ($matches_count > 0); |
193
|
|
|
} |
194
|
|
|
} |
195
|
|
|
|
This checks looks for assignemnts to variables using the
list(...)
function, where not all assigned variables are subsequently used.Consider the following code example.
Only the variables
$a
and$c
are used. There was no need to assign$b
.Instead, the list call could have been.