oga 0.1.3 → 0.2.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/README.md +57 -0
- data/doc/changelog.md +128 -0
- data/doc/css/common.css +5 -4
- data/doc/css_selectors.md +935 -0
- data/doc/manually_creating_documents.md +67 -0
- data/doc/xml_namespaces.md +63 -0
- data/ext/c/lexer.c +745 -628
- data/ext/c/lexer.h +8 -0
- data/ext/c/lexer.rl +44 -7
- data/ext/java/org/liboga/xml/Lexer.java +351 -232
- data/ext/java/org/liboga/xml/Lexer.rl +29 -8
- data/ext/ragel/base_lexer.rl +68 -18
- data/lib/oga.rb +4 -1
- data/lib/oga/css/lexer.rb +743 -0
- data/lib/oga/css/parser.rb +828 -0
- data/lib/oga/version.rb +1 -1
- data/lib/oga/xml/attribute.rb +3 -1
- data/lib/oga/xml/element.rb +15 -1
- data/lib/oga/xml/entities.rb +60 -0
- data/lib/oga/xml/html_void_elements.rb +2 -0
- data/lib/oga/xml/lexer.rb +36 -28
- data/lib/oga/xml/node_set.rb +22 -0
- data/lib/oga/xml/parser.rb +149 -128
- data/lib/oga/xml/querying.rb +24 -0
- data/lib/oga/xml/sax_parser.rb +55 -1
- data/lib/oga/xml/text.rb +6 -1
- data/lib/oga/xpath/evaluator.rb +138 -101
- data/lib/oga/xpath/lexer.rb +1205 -1294
- data/lib/oga/xpath/parser.rb +228 -204
- metadata +9 -4
- data/lib/oga/xpath/node.rb +0 -10
@@ -40,6 +40,9 @@ public class Lexer extends RubyObject
|
|
40
40
|
/* Used by Ragel to keep track of the current state. */
|
41
41
|
int act;
|
42
42
|
int cs;
|
43
|
+
int top;
|
44
|
+
int lines;
|
45
|
+
int[] stack;
|
43
46
|
|
44
47
|
/**
|
45
48
|
* Sets up the current class in the Ruby runtime.
|
@@ -90,15 +93,18 @@ public class Lexer extends RubyObject
|
|
90
93
|
|
91
94
|
byte[] data = rb_str.getBytes();
|
92
95
|
|
93
|
-
int ts
|
94
|
-
int te
|
95
|
-
int p
|
96
|
-
int mark
|
97
|
-
int
|
98
|
-
int
|
96
|
+
int ts = 0;
|
97
|
+
int te = 0;
|
98
|
+
int p = 0;
|
99
|
+
int mark = 0;
|
100
|
+
int lines = this.lines;
|
101
|
+
int pe = data.length;
|
102
|
+
int eof = data.length;
|
99
103
|
|
100
104
|
%% write exec;
|
101
105
|
|
106
|
+
this.lines = lines;
|
107
|
+
|
102
108
|
return context.nil;
|
103
109
|
}
|
104
110
|
|
@@ -108,8 +114,10 @@ public class Lexer extends RubyObject
|
|
108
114
|
@JRubyMethod
|
109
115
|
public IRubyObject reset_native(ThreadContext context)
|
110
116
|
{
|
111
|
-
this.act
|
112
|
-
this.
|
117
|
+
this.act = 0;
|
118
|
+
this.top = 0;
|
119
|
+
this.stack = new int[4];
|
120
|
+
this.cs = java_lexer_start;
|
113
121
|
|
114
122
|
return context.nil;
|
115
123
|
}
|
@@ -141,11 +149,24 @@ public class Lexer extends RubyObject
|
|
141
149
|
|
142
150
|
this.callMethod(context, name);
|
143
151
|
}
|
152
|
+
|
153
|
+
/**
|
154
|
+
* Advances the line number by `amount` lines.
|
155
|
+
*/
|
156
|
+
public void advance_line(int amount)
|
157
|
+
{
|
158
|
+
ThreadContext context = this.runtime.getCurrentContext();
|
159
|
+
RubyFixnum lines = this.runtime.newFixnum(amount);
|
160
|
+
|
161
|
+
this.callMethod(context, "advance_line", lines);
|
162
|
+
}
|
144
163
|
}
|
145
164
|
|
146
165
|
%%{
|
147
166
|
variable act this.act;
|
148
167
|
variable cs this.cs;
|
168
|
+
variable stack this.stack;
|
169
|
+
variable top this.top;
|
149
170
|
|
150
171
|
include base_lexer "base_lexer.rl";
|
151
172
|
}%%
|
data/ext/ragel/base_lexer.rl
CHANGED
@@ -35,7 +35,12 @@
|
|
35
35
|
# stack.
|
36
36
|
#
|
37
37
|
|
38
|
-
newline
|
38
|
+
newline = '\n' | '\r\n';
|
39
|
+
|
40
|
+
action count_newlines {
|
41
|
+
if ( fc == '\n' ) lines++;
|
42
|
+
}
|
43
|
+
|
39
44
|
whitespace = [ \t];
|
40
45
|
ident_char = [a-zA-Z0-9\-_];
|
41
46
|
identifier = ident_char+;
|
@@ -101,6 +106,8 @@
|
|
101
106
|
callback("on_text", data, encoding, mark, ts);
|
102
107
|
callback_simple("on_proc_ins_end");
|
103
108
|
|
109
|
+
mark = 0;
|
110
|
+
|
104
111
|
fnext main;
|
105
112
|
};
|
106
113
|
|
@@ -116,15 +123,49 @@
|
|
116
123
|
dquote = '"';
|
117
124
|
squote = "'";
|
118
125
|
|
119
|
-
|
120
|
-
|
126
|
+
action emit_string {
|
127
|
+
callback("on_string_body", data, encoding, ts, te);
|
121
128
|
|
122
|
-
|
129
|
+
if ( lines > 0 )
|
130
|
+
{
|
131
|
+
advance_line(lines);
|
123
132
|
|
124
|
-
|
125
|
-
|
133
|
+
lines = 0;
|
134
|
+
}
|
126
135
|
}
|
127
136
|
|
137
|
+
action start_string_squote {
|
138
|
+
callback_simple("on_string_squote");
|
139
|
+
|
140
|
+
fcall string_squote;
|
141
|
+
}
|
142
|
+
|
143
|
+
action start_string_dquote {
|
144
|
+
callback_simple("on_string_dquote");
|
145
|
+
|
146
|
+
fcall string_dquote;
|
147
|
+
}
|
148
|
+
|
149
|
+
string_squote := |*
|
150
|
+
^squote* $count_newlines => emit_string;
|
151
|
+
|
152
|
+
squote => {
|
153
|
+
callback_simple("on_string_squote");
|
154
|
+
|
155
|
+
fret;
|
156
|
+
};
|
157
|
+
*|;
|
158
|
+
|
159
|
+
string_dquote := |*
|
160
|
+
^dquote* $count_newlines => emit_string;
|
161
|
+
|
162
|
+
dquote => {
|
163
|
+
callback_simple("on_string_dquote");
|
164
|
+
|
165
|
+
fret;
|
166
|
+
};
|
167
|
+
*|;
|
168
|
+
|
128
169
|
# DOCTYPES
|
129
170
|
#
|
130
171
|
# http://www.w3.org/TR/html-markup/syntax.html#doctype-syntax
|
@@ -156,7 +197,8 @@
|
|
156
197
|
};
|
157
198
|
|
158
199
|
# Lex the public/system IDs as regular strings.
|
159
|
-
|
200
|
+
squote => start_string_squote;
|
201
|
+
dquote => start_string_dquote;
|
160
202
|
|
161
203
|
# Whitespace inside doctypes is ignored since there's no point in
|
162
204
|
# including it.
|
@@ -196,7 +238,8 @@
|
|
196
238
|
callback("on_attribute", data, encoding, ts, te);
|
197
239
|
};
|
198
240
|
|
199
|
-
|
241
|
+
squote => start_string_squote;
|
242
|
+
dquote => start_string_dquote;
|
200
243
|
|
201
244
|
any;
|
202
245
|
*|;
|
@@ -254,7 +297,8 @@
|
|
254
297
|
};
|
255
298
|
|
256
299
|
# Attribute values.
|
257
|
-
|
300
|
+
squote => start_string_squote;
|
301
|
+
dquote => start_string_dquote;
|
258
302
|
|
259
303
|
# We're done with the open tag of the element.
|
260
304
|
'>' => {
|
@@ -289,14 +333,19 @@
|
|
289
333
|
# long. Because of this "<!" is used instead of "<!--".
|
290
334
|
|
291
335
|
terminate_text = '</' | '<!' | '<?' | element_start;
|
292
|
-
allowed_text = any* -- terminate_text;
|
336
|
+
allowed_text = (any* -- terminate_text) $count_newlines;
|
293
337
|
|
294
338
|
text := |*
|
295
|
-
|
296
|
-
# rules below, but only if those don't match.
|
297
|
-
terminate_text => {
|
339
|
+
terminate_text | allowed_text => {
|
298
340
|
callback("on_text", data, encoding, ts, te);
|
299
341
|
|
342
|
+
if ( lines > 0 )
|
343
|
+
{
|
344
|
+
advance_line(lines);
|
345
|
+
|
346
|
+
lines = 0;
|
347
|
+
}
|
348
|
+
|
300
349
|
fnext main;
|
301
350
|
};
|
302
351
|
|
@@ -307,12 +356,13 @@
|
|
307
356
|
p = mark - 1;
|
308
357
|
mark = 0;
|
309
358
|
|
310
|
-
|
311
|
-
|
359
|
+
if ( lines > 0 )
|
360
|
+
{
|
361
|
+
advance_line(lines);
|
362
|
+
|
363
|
+
lines = 0;
|
364
|
+
}
|
312
365
|
|
313
|
-
# Just regular text.
|
314
|
-
allowed_text => {
|
315
|
-
callback("on_text", data, encoding, ts, te);
|
316
366
|
fnext main;
|
317
367
|
};
|
318
368
|
*|;
|
data/lib/oga.rb
CHANGED
@@ -21,6 +21,7 @@ end
|
|
21
21
|
#:nocov:
|
22
22
|
|
23
23
|
require_relative 'oga/xml/html_void_elements'
|
24
|
+
require_relative 'oga/xml/entities'
|
24
25
|
require_relative 'oga/xml/querying'
|
25
26
|
require_relative 'oga/xml/traversal'
|
26
27
|
require_relative 'oga/xml/node'
|
@@ -43,7 +44,9 @@ require_relative 'oga/xml/pull_parser'
|
|
43
44
|
require_relative 'oga/html/parser'
|
44
45
|
require_relative 'oga/html/sax_parser'
|
45
46
|
|
46
|
-
require_relative 'oga/xpath/node'
|
47
47
|
require_relative 'oga/xpath/lexer'
|
48
48
|
require_relative 'oga/xpath/parser'
|
49
49
|
require_relative 'oga/xpath/evaluator'
|
50
|
+
|
51
|
+
require_relative 'oga/css/lexer'
|
52
|
+
require_relative 'oga/css/parser'
|
@@ -0,0 +1,743 @@
|
|
1
|
+
|
2
|
+
# line 1 "lib/oga/css/lexer.rl"
|
3
|
+
|
4
|
+
# line 3 "lib/oga/css/lexer.rl"
|
5
|
+
module Oga
|
6
|
+
module CSS
|
7
|
+
##
|
8
|
+
# Lexer for turning CSS expressions into a sequence of tokens. Tokens are
|
9
|
+
# returned as arrays with every array having two values:
|
10
|
+
#
|
11
|
+
# 1. The token type as a Symbol
|
12
|
+
# 2. The token value, or nil if there is no value.
|
13
|
+
#
|
14
|
+
# ## Thread Safety
|
15
|
+
#
|
16
|
+
# Similar to the XPath lexer this lexer keeps track of an internal state. As
|
17
|
+
# a result it's not safe to share the same instance of this lexer between
|
18
|
+
# multiple threads. However, no global state is used so you can use separate
|
19
|
+
# instances in threads just fine.
|
20
|
+
#
|
21
|
+
class Lexer
|
22
|
+
|
23
|
+
# line 24 "lib/oga/css/lexer.rb"
|
24
|
+
class << self
|
25
|
+
attr_accessor :_css_lexer_trans_keys
|
26
|
+
private :_css_lexer_trans_keys, :_css_lexer_trans_keys=
|
27
|
+
end
|
28
|
+
self._css_lexer_trans_keys = [
|
29
|
+
0, 0, 43, 57, 118, 118,
|
30
|
+
101, 101, 110, 110, 100,
|
31
|
+
100, 100, 100, 34, 34,
|
32
|
+
61, 61, 39, 39, 61, 61,
|
33
|
+
61, 61, 61, 61, 9,
|
34
|
+
126, 9, 44, 9, 32,
|
35
|
+
0, 0, 9, 32, 0, 0,
|
36
|
+
0, 0, 9, 32, 45,
|
37
|
+
122, 9, 32, 9, 122,
|
38
|
+
9, 32, 48, 57, 43, 57,
|
39
|
+
45, 122, 9, 126, 9,
|
40
|
+
32, 61, 61, 45, 122,
|
41
|
+
0
|
42
|
+
]
|
43
|
+
|
44
|
+
class << self
|
45
|
+
attr_accessor :_css_lexer_key_spans
|
46
|
+
private :_css_lexer_key_spans, :_css_lexer_key_spans=
|
47
|
+
end
|
48
|
+
self._css_lexer_key_spans = [
|
49
|
+
0, 15, 1, 1, 1, 1, 1, 1,
|
50
|
+
1, 1, 1, 1, 1, 118, 36, 24,
|
51
|
+
0, 24, 0, 0, 24, 78, 24, 114,
|
52
|
+
24, 10, 15, 78, 118, 24, 1, 78
|
53
|
+
]
|
54
|
+
|
55
|
+
class << self
|
56
|
+
attr_accessor :_css_lexer_index_offsets
|
57
|
+
private :_css_lexer_index_offsets, :_css_lexer_index_offsets=
|
58
|
+
end
|
59
|
+
self._css_lexer_index_offsets = [
|
60
|
+
0, 0, 16, 18, 20, 22, 24, 26,
|
61
|
+
28, 30, 32, 34, 36, 38, 157, 194,
|
62
|
+
219, 220, 245, 246, 247, 272, 351, 376,
|
63
|
+
491, 516, 527, 543, 622, 741, 766, 768
|
64
|
+
]
|
65
|
+
|
66
|
+
class << self
|
67
|
+
attr_accessor :_css_lexer_indicies
|
68
|
+
private :_css_lexer_indicies, :_css_lexer_indicies=
|
69
|
+
end
|
70
|
+
self._css_lexer_indicies = [
|
71
|
+
1, 0, 1, 0, 0, 2, 2, 2,
|
72
|
+
2, 2, 2, 2, 2, 2, 2, 0,
|
73
|
+
3, 4, 5, 4, 6, 4, 7, 4,
|
74
|
+
8, 4, 10, 9, 11, 4, 10, 12,
|
75
|
+
13, 4, 14, 4, 15, 4, 17, 16,
|
76
|
+
16, 16, 16, 16, 16, 16, 16, 16,
|
77
|
+
16, 16, 16, 16, 16, 16, 16, 16,
|
78
|
+
16, 16, 16, 16, 16, 17, 16, 16,
|
79
|
+
18, 16, 16, 16, 16, 19, 16, 20,
|
80
|
+
21, 22, 16, 23, 16, 16, 16, 16,
|
81
|
+
16, 16, 16, 16, 16, 16, 16, 24,
|
82
|
+
16, 16, 16, 25, 16, 16, 26, 26,
|
83
|
+
26, 26, 26, 26, 26, 26, 26, 26,
|
84
|
+
26, 26, 26, 26, 26, 26, 26, 26,
|
85
|
+
26, 26, 26, 26, 26, 26, 26, 26,
|
86
|
+
27, 16, 16, 16, 26, 16, 26, 26,
|
87
|
+
26, 26, 26, 26, 26, 26, 26, 26,
|
88
|
+
26, 26, 26, 26, 26, 26, 26, 26,
|
89
|
+
26, 26, 26, 26, 26, 26, 26, 26,
|
90
|
+
16, 28, 16, 29, 16, 17, 30, 30,
|
91
|
+
30, 30, 30, 30, 30, 30, 30, 30,
|
92
|
+
30, 30, 30, 30, 30, 30, 30, 30,
|
93
|
+
30, 30, 30, 30, 17, 30, 30, 30,
|
94
|
+
30, 30, 30, 30, 30, 30, 30, 30,
|
95
|
+
22, 30, 22, 31, 31, 31, 31, 31,
|
96
|
+
31, 31, 31, 31, 31, 31, 31, 31,
|
97
|
+
31, 31, 31, 31, 31, 31, 31, 31,
|
98
|
+
31, 22, 31, 32, 21, 33, 33, 33,
|
99
|
+
33, 33, 33, 33, 33, 33, 33, 33,
|
100
|
+
33, 33, 33, 33, 33, 33, 33, 33,
|
101
|
+
33, 33, 33, 21, 33, 34, 35, 25,
|
102
|
+
36, 36, 36, 36, 36, 36, 36, 36,
|
103
|
+
36, 36, 36, 36, 36, 36, 36, 36,
|
104
|
+
36, 36, 36, 36, 36, 36, 25, 36,
|
105
|
+
26, 37, 37, 26, 26, 26, 26, 26,
|
106
|
+
26, 26, 26, 26, 26, 37, 37, 37,
|
107
|
+
37, 37, 37, 37, 26, 26, 26, 26,
|
108
|
+
26, 26, 26, 26, 26, 26, 26, 26,
|
109
|
+
26, 26, 26, 26, 26, 26, 26, 26,
|
110
|
+
26, 26, 26, 26, 26, 26, 37, 37,
|
111
|
+
37, 37, 26, 37, 26, 26, 26, 26,
|
112
|
+
26, 26, 26, 26, 26, 26, 26, 26,
|
113
|
+
26, 26, 26, 26, 26, 26, 26, 26,
|
114
|
+
26, 26, 26, 26, 26, 26, 37, 29,
|
115
|
+
38, 38, 38, 38, 38, 38, 38, 38,
|
116
|
+
38, 38, 38, 38, 38, 38, 38, 38,
|
117
|
+
38, 38, 38, 38, 38, 38, 29, 38,
|
118
|
+
39, 4, 4, 4, 4, 4, 4, 4,
|
119
|
+
4, 4, 4, 4, 4, 4, 4, 4,
|
120
|
+
4, 4, 4, 4, 4, 4, 4, 39,
|
121
|
+
4, 4, 4, 4, 4, 4, 4, 4,
|
122
|
+
40, 41, 1, 4, 42, 4, 4, 2,
|
123
|
+
2, 2, 2, 2, 2, 2, 2, 2,
|
124
|
+
2, 4, 4, 4, 4, 4, 4, 4,
|
125
|
+
43, 43, 43, 43, 43, 43, 43, 43,
|
126
|
+
43, 43, 43, 43, 43, 43, 43, 43,
|
127
|
+
43, 43, 43, 43, 43, 43, 43, 43,
|
128
|
+
43, 43, 4, 4, 4, 4, 43, 4,
|
129
|
+
43, 43, 43, 43, 44, 43, 43, 43,
|
130
|
+
43, 43, 43, 43, 43, 45, 46, 43,
|
131
|
+
43, 43, 43, 43, 43, 43, 43, 43,
|
132
|
+
43, 43, 4, 39, 47, 47, 47, 47,
|
133
|
+
47, 47, 47, 47, 47, 47, 47, 47,
|
134
|
+
47, 47, 47, 47, 47, 47, 47, 47,
|
135
|
+
47, 47, 39, 47, 2, 2, 2, 2,
|
136
|
+
2, 2, 2, 2, 2, 2, 48, 1,
|
137
|
+
49, 1, 49, 49, 2, 2, 2, 2,
|
138
|
+
2, 2, 2, 2, 2, 2, 49, 43,
|
139
|
+
50, 50, 43, 43, 43, 43, 43, 43,
|
140
|
+
43, 43, 43, 43, 50, 50, 50, 50,
|
141
|
+
50, 50, 50, 43, 43, 43, 43, 43,
|
142
|
+
43, 43, 43, 43, 43, 43, 43, 43,
|
143
|
+
43, 43, 43, 43, 43, 43, 43, 43,
|
144
|
+
43, 43, 43, 43, 43, 50, 50, 50,
|
145
|
+
50, 43, 50, 43, 43, 43, 43, 43,
|
146
|
+
43, 43, 43, 43, 43, 43, 43, 43,
|
147
|
+
43, 43, 43, 43, 43, 43, 43, 43,
|
148
|
+
43, 43, 43, 43, 43, 50, 51, 4,
|
149
|
+
4, 4, 4, 4, 4, 4, 4, 4,
|
150
|
+
4, 4, 4, 4, 4, 4, 4, 4,
|
151
|
+
4, 4, 4, 4, 4, 51, 4, 9,
|
152
|
+
4, 52, 4, 4, 12, 4, 4, 53,
|
153
|
+
4, 4, 4, 4, 4, 4, 4, 4,
|
154
|
+
4, 4, 4, 4, 4, 4, 4, 4,
|
155
|
+
4, 4, 54, 4, 4, 4, 55, 55,
|
156
|
+
55, 55, 55, 55, 55, 55, 55, 55,
|
157
|
+
55, 55, 55, 55, 55, 55, 55, 55,
|
158
|
+
55, 55, 55, 55, 55, 55, 55, 55,
|
159
|
+
4, 4, 56, 57, 55, 4, 55, 55,
|
160
|
+
55, 55, 55, 55, 55, 55, 55, 55,
|
161
|
+
55, 55, 55, 55, 55, 55, 55, 55,
|
162
|
+
55, 55, 55, 55, 55, 55, 55, 55,
|
163
|
+
4, 58, 4, 59, 4, 51, 60, 60,
|
164
|
+
60, 60, 60, 60, 60, 60, 60, 60,
|
165
|
+
60, 60, 60, 60, 60, 60, 60, 60,
|
166
|
+
60, 60, 60, 60, 51, 60, 62, 61,
|
167
|
+
55, 61, 61, 55, 55, 55, 55, 55,
|
168
|
+
55, 55, 55, 55, 55, 61, 61, 61,
|
169
|
+
61, 61, 61, 61, 55, 55, 55, 55,
|
170
|
+
55, 55, 55, 55, 55, 55, 55, 55,
|
171
|
+
55, 55, 55, 55, 55, 55, 55, 55,
|
172
|
+
55, 55, 55, 55, 55, 55, 61, 61,
|
173
|
+
61, 61, 55, 61, 55, 55, 55, 55,
|
174
|
+
55, 55, 55, 55, 55, 55, 55, 55,
|
175
|
+
55, 55, 55, 55, 55, 55, 55, 55,
|
176
|
+
55, 55, 55, 55, 55, 55, 61, 0
|
177
|
+
]
|
178
|
+
|
179
|
+
class << self
|
180
|
+
attr_accessor :_css_lexer_trans_targs
|
181
|
+
private :_css_lexer_trans_targs, :_css_lexer_trans_targs=
|
182
|
+
end
|
183
|
+
self._css_lexer_trans_targs = [
|
184
|
+
23, 1, 25, 3, 0, 4, 23, 6,
|
185
|
+
23, 7, 28, 28, 9, 28, 28, 28,
|
186
|
+
13, 14, 16, 13, 13, 17, 15, 18,
|
187
|
+
19, 20, 21, 13, 13, 22, 13, 13,
|
188
|
+
13, 13, 13, 13, 13, 13, 13, 24,
|
189
|
+
23, 23, 26, 27, 2, 23, 5, 23,
|
190
|
+
23, 23, 23, 29, 8, 30, 28, 31,
|
191
|
+
28, 10, 11, 12, 28, 28, 28
|
192
|
+
]
|
193
|
+
|
194
|
+
class << self
|
195
|
+
attr_accessor :_css_lexer_trans_actions
|
196
|
+
private :_css_lexer_trans_actions, :_css_lexer_trans_actions=
|
197
|
+
end
|
198
|
+
self._css_lexer_trans_actions = [
|
199
|
+
1, 0, 0, 0, 0, 0, 2, 0,
|
200
|
+
3, 0, 4, 5, 0, 6, 7, 8,
|
201
|
+
11, 0, 0, 12, 13, 0, 0, 0,
|
202
|
+
0, 0, 0, 14, 15, 0, 16, 17,
|
203
|
+
18, 19, 20, 21, 22, 23, 24, 0,
|
204
|
+
26, 27, 28, 0, 0, 29, 0, 30,
|
205
|
+
31, 32, 33, 0, 0, 0, 34, 0,
|
206
|
+
35, 0, 0, 0, 36, 37, 38
|
207
|
+
]
|
208
|
+
|
209
|
+
class << self
|
210
|
+
attr_accessor :_css_lexer_to_state_actions
|
211
|
+
private :_css_lexer_to_state_actions, :_css_lexer_to_state_actions=
|
212
|
+
end
|
213
|
+
self._css_lexer_to_state_actions = [
|
214
|
+
0, 0, 0, 0, 0, 0, 0, 0,
|
215
|
+
0, 0, 0, 0, 0, 9, 0, 0,
|
216
|
+
0, 0, 0, 0, 0, 0, 0, 25,
|
217
|
+
0, 0, 0, 0, 9, 0, 0, 0
|
218
|
+
]
|
219
|
+
|
220
|
+
class << self
|
221
|
+
attr_accessor :_css_lexer_from_state_actions
|
222
|
+
private :_css_lexer_from_state_actions, :_css_lexer_from_state_actions=
|
223
|
+
end
|
224
|
+
self._css_lexer_from_state_actions = [
|
225
|
+
0, 0, 0, 0, 0, 0, 0, 0,
|
226
|
+
0, 0, 0, 0, 0, 10, 0, 0,
|
227
|
+
0, 0, 0, 0, 0, 0, 0, 10,
|
228
|
+
0, 0, 0, 0, 10, 0, 0, 0
|
229
|
+
]
|
230
|
+
|
231
|
+
class << self
|
232
|
+
attr_accessor :_css_lexer_eof_trans
|
233
|
+
private :_css_lexer_eof_trans, :_css_lexer_eof_trans=
|
234
|
+
end
|
235
|
+
self._css_lexer_eof_trans = [
|
236
|
+
0, 1, 0, 0, 0, 0, 0, 0,
|
237
|
+
0, 0, 0, 0, 0, 0, 31, 32,
|
238
|
+
33, 34, 35, 36, 37, 38, 39, 0,
|
239
|
+
48, 49, 50, 51, 0, 61, 62, 62
|
240
|
+
]
|
241
|
+
|
242
|
+
class << self
|
243
|
+
attr_accessor :css_lexer_start
|
244
|
+
end
|
245
|
+
self.css_lexer_start = 13;
|
246
|
+
class << self
|
247
|
+
attr_accessor :css_lexer_first_final
|
248
|
+
end
|
249
|
+
self.css_lexer_first_final = 13;
|
250
|
+
class << self
|
251
|
+
attr_accessor :css_lexer_error
|
252
|
+
end
|
253
|
+
self.css_lexer_error = 0;
|
254
|
+
|
255
|
+
class << self
|
256
|
+
attr_accessor :css_lexer_en_pseudo_args
|
257
|
+
end
|
258
|
+
self.css_lexer_en_pseudo_args = 23;
|
259
|
+
class << self
|
260
|
+
attr_accessor :css_lexer_en_predicate
|
261
|
+
end
|
262
|
+
self.css_lexer_en_predicate = 28;
|
263
|
+
class << self
|
264
|
+
attr_accessor :css_lexer_en_main
|
265
|
+
end
|
266
|
+
self.css_lexer_en_main = 13;
|
267
|
+
|
268
|
+
|
269
|
+
# line 21 "lib/oga/css/lexer.rl"
|
270
|
+
|
271
|
+
# % fix highlight
|
272
|
+
|
273
|
+
##
|
274
|
+
# @param [String] data The data to lex.
|
275
|
+
#
|
276
|
+
def initialize(data)
|
277
|
+
@data = data
|
278
|
+
end
|
279
|
+
|
280
|
+
##
|
281
|
+
# Gathers all the tokens for the input and returns them as an Array.
|
282
|
+
#
|
283
|
+
# @see [#advance]
|
284
|
+
# @return [Array]
|
285
|
+
#
|
286
|
+
def lex
|
287
|
+
tokens = []
|
288
|
+
|
289
|
+
advance do |type, value|
|
290
|
+
tokens << [type, value]
|
291
|
+
end
|
292
|
+
|
293
|
+
return tokens
|
294
|
+
end
|
295
|
+
|
296
|
+
##
|
297
|
+
# Advances through the input and generates the corresponding tokens. Each
|
298
|
+
# token is yielded to the supplied block.
|
299
|
+
#
|
300
|
+
# This method stores the supplied block in `@block` and resets it after
|
301
|
+
# the lexer loop has finished.
|
302
|
+
#
|
303
|
+
# @see [#add_token]
|
304
|
+
#
|
305
|
+
def advance(&block)
|
306
|
+
@block = block
|
307
|
+
|
308
|
+
data = @data # saves ivar lookups while lexing.
|
309
|
+
ts = nil
|
310
|
+
te = nil
|
311
|
+
stack = []
|
312
|
+
top = 0
|
313
|
+
cs = self.class.css_lexer_start
|
314
|
+
act = 0
|
315
|
+
eof = @data.bytesize
|
316
|
+
p = 0
|
317
|
+
pe = eof
|
318
|
+
|
319
|
+
_css_lexer_eof_trans = self.class.send(:_css_lexer_eof_trans)
|
320
|
+
_css_lexer_from_state_actions = self.class.send(:_css_lexer_from_state_actions)
|
321
|
+
_css_lexer_index_offsets = self.class.send(:_css_lexer_index_offsets)
|
322
|
+
_css_lexer_indicies = self.class.send(:_css_lexer_indicies)
|
323
|
+
_css_lexer_key_spans = self.class.send(:_css_lexer_key_spans)
|
324
|
+
_css_lexer_to_state_actions = self.class.send(:_css_lexer_to_state_actions)
|
325
|
+
_css_lexer_trans_actions = self.class.send(:_css_lexer_trans_actions)
|
326
|
+
_css_lexer_trans_keys = self.class.send(:_css_lexer_trans_keys)
|
327
|
+
_css_lexer_trans_targs = self.class.send(:_css_lexer_trans_targs)
|
328
|
+
|
329
|
+
|
330
|
+
# line 331 "lib/oga/css/lexer.rb"
|
331
|
+
begin
|
332
|
+
testEof = false
|
333
|
+
_slen, _trans, _keys, _inds, _acts, _nacts = nil
|
334
|
+
_goto_level = 0
|
335
|
+
_resume = 10
|
336
|
+
_eof_trans = 15
|
337
|
+
_again = 20
|
338
|
+
_test_eof = 30
|
339
|
+
_out = 40
|
340
|
+
while true
|
341
|
+
if _goto_level <= 0
|
342
|
+
if p == pe
|
343
|
+
_goto_level = _test_eof
|
344
|
+
next
|
345
|
+
end
|
346
|
+
if cs == 0
|
347
|
+
_goto_level = _out
|
348
|
+
next
|
349
|
+
end
|
350
|
+
end
|
351
|
+
if _goto_level <= _resume
|
352
|
+
case _css_lexer_from_state_actions[cs]
|
353
|
+
when 10 then
|
354
|
+
# line 1 "NONE"
|
355
|
+
begin
|
356
|
+
ts = p
|
357
|
+
end
|
358
|
+
# line 359 "lib/oga/css/lexer.rb"
|
359
|
+
end
|
360
|
+
_keys = cs << 1
|
361
|
+
_inds = _css_lexer_index_offsets[cs]
|
362
|
+
_slen = _css_lexer_key_spans[cs]
|
363
|
+
_wide = ( _wide)
|
364
|
+
_wide = (data.getbyte(p) || 0)
|
365
|
+
_trans = if ( _slen > 0 &&
|
366
|
+
_css_lexer_trans_keys[_keys] <= _wide &&
|
367
|
+
_wide <= _css_lexer_trans_keys[_keys + 1]
|
368
|
+
) then
|
369
|
+
_css_lexer_indicies[ _inds + _wide - _css_lexer_trans_keys[_keys] ]
|
370
|
+
else
|
371
|
+
_css_lexer_indicies[ _inds + _slen ]
|
372
|
+
end
|
373
|
+
end
|
374
|
+
if _goto_level <= _eof_trans
|
375
|
+
cs = _css_lexer_trans_targs[_trans]
|
376
|
+
if _css_lexer_trans_actions[_trans] != 0
|
377
|
+
case _css_lexer_trans_actions[_trans]
|
378
|
+
when 29 then
|
379
|
+
# line 249 "lib/oga/css/lexer.rl"
|
380
|
+
begin
|
381
|
+
te = p+1
|
382
|
+
begin add_token(:T_NTH) end
|
383
|
+
end
|
384
|
+
when 3 then
|
385
|
+
# line 251 "lib/oga/css/lexer.rl"
|
386
|
+
begin
|
387
|
+
te = p+1
|
388
|
+
begin add_token(:T_ODD) end
|
389
|
+
end
|
390
|
+
when 2 then
|
391
|
+
# line 252 "lib/oga/css/lexer.rl"
|
392
|
+
begin
|
393
|
+
te = p+1
|
394
|
+
begin add_token(:T_EVEN) end
|
395
|
+
end
|
396
|
+
when 26 then
|
397
|
+
# line 233 "lib/oga/css/lexer.rl"
|
398
|
+
begin
|
399
|
+
te = p+1
|
400
|
+
begin
|
401
|
+
add_token(:T_RPAREN)
|
402
|
+
|
403
|
+
cs = 13;
|
404
|
+
end
|
405
|
+
end
|
406
|
+
when 27 then
|
407
|
+
# line 160 "lib/oga/css/lexer.rl"
|
408
|
+
begin
|
409
|
+
te = p+1
|
410
|
+
begin
|
411
|
+
emit(:T_IDENT, ts, te)
|
412
|
+
end
|
413
|
+
end
|
414
|
+
when 30 then
|
415
|
+
# line 244 "lib/oga/css/lexer.rl"
|
416
|
+
begin
|
417
|
+
te = p
|
418
|
+
p = p - 1; end
|
419
|
+
when 32 then
|
420
|
+
# line 250 "lib/oga/css/lexer.rl"
|
421
|
+
begin
|
422
|
+
te = p
|
423
|
+
p = p - 1; begin add_token(:T_MINUS) end
|
424
|
+
end
|
425
|
+
when 31 then
|
426
|
+
# line 193 "lib/oga/css/lexer.rl"
|
427
|
+
begin
|
428
|
+
te = p
|
429
|
+
p = p - 1; begin
|
430
|
+
value = slice_input(ts, te).to_i
|
431
|
+
|
432
|
+
add_token(:T_INT, value)
|
433
|
+
end
|
434
|
+
end
|
435
|
+
when 33 then
|
436
|
+
# line 160 "lib/oga/css/lexer.rl"
|
437
|
+
begin
|
438
|
+
te = p
|
439
|
+
p = p - 1; begin
|
440
|
+
emit(:T_IDENT, ts, te)
|
441
|
+
end
|
442
|
+
end
|
443
|
+
when 1 then
|
444
|
+
# line 1 "NONE"
|
445
|
+
begin
|
446
|
+
case act
|
447
|
+
when 0 then
|
448
|
+
begin begin
|
449
|
+
cs = 0
|
450
|
+
_goto_level = _again
|
451
|
+
next
|
452
|
+
end
|
453
|
+
end
|
454
|
+
when 3 then
|
455
|
+
begin begin p = ((te))-1; end
|
456
|
+
add_token(:T_MINUS) end
|
457
|
+
end
|
458
|
+
end
|
459
|
+
when 34 then
|
460
|
+
# line 290 "lib/oga/css/lexer.rl"
|
461
|
+
begin
|
462
|
+
te = p+1
|
463
|
+
begin add_token(:T_EQ) end
|
464
|
+
end
|
465
|
+
when 8 then
|
466
|
+
# line 291 "lib/oga/css/lexer.rl"
|
467
|
+
begin
|
468
|
+
te = p+1
|
469
|
+
begin add_token(:T_SPACE_IN) end
|
470
|
+
end
|
471
|
+
when 6 then
|
472
|
+
# line 292 "lib/oga/css/lexer.rl"
|
473
|
+
begin
|
474
|
+
te = p+1
|
475
|
+
begin add_token(:T_STARTS_WITH) end
|
476
|
+
end
|
477
|
+
when 5 then
|
478
|
+
# line 293 "lib/oga/css/lexer.rl"
|
479
|
+
begin
|
480
|
+
te = p+1
|
481
|
+
begin add_token(:T_ENDS_WITH) end
|
482
|
+
end
|
483
|
+
when 38 then
|
484
|
+
# line 294 "lib/oga/css/lexer.rl"
|
485
|
+
begin
|
486
|
+
te = p+1
|
487
|
+
begin add_token(:T_IN) end
|
488
|
+
end
|
489
|
+
when 7 then
|
490
|
+
# line 295 "lib/oga/css/lexer.rl"
|
491
|
+
begin
|
492
|
+
te = p+1
|
493
|
+
begin add_token(:T_HYPHEN_IN) end
|
494
|
+
end
|
495
|
+
when 35 then
|
496
|
+
# line 272 "lib/oga/css/lexer.rl"
|
497
|
+
begin
|
498
|
+
te = p+1
|
499
|
+
begin
|
500
|
+
add_token(:T_RBRACK)
|
501
|
+
|
502
|
+
cs = 13;
|
503
|
+
end
|
504
|
+
end
|
505
|
+
when 4 then
|
506
|
+
# line 212 "lib/oga/css/lexer.rl"
|
507
|
+
begin
|
508
|
+
te = p+1
|
509
|
+
begin
|
510
|
+
emit(:T_STRING, ts + 1, te - 1)
|
511
|
+
end
|
512
|
+
end
|
513
|
+
when 36 then
|
514
|
+
# line 280 "lib/oga/css/lexer.rl"
|
515
|
+
begin
|
516
|
+
te = p
|
517
|
+
p = p - 1; end
|
518
|
+
when 37 then
|
519
|
+
# line 160 "lib/oga/css/lexer.rl"
|
520
|
+
begin
|
521
|
+
te = p
|
522
|
+
p = p - 1; begin
|
523
|
+
emit(:T_IDENT, ts, te)
|
524
|
+
end
|
525
|
+
end
|
526
|
+
when 14 then
|
527
|
+
# line 266 "lib/oga/css/lexer.rl"
|
528
|
+
begin
|
529
|
+
te = p+1
|
530
|
+
begin
|
531
|
+
add_token(:T_LBRACK)
|
532
|
+
|
533
|
+
cs = 28;
|
534
|
+
end
|
535
|
+
end
|
536
|
+
when 15 then
|
537
|
+
# line 145 "lib/oga/css/lexer.rl"
|
538
|
+
begin
|
539
|
+
te = p+1
|
540
|
+
begin
|
541
|
+
add_token(:T_PIPE)
|
542
|
+
end
|
543
|
+
end
|
544
|
+
when 12 then
|
545
|
+
# line 227 "lib/oga/css/lexer.rl"
|
546
|
+
begin
|
547
|
+
te = p+1
|
548
|
+
begin
|
549
|
+
add_token(:T_LPAREN)
|
550
|
+
|
551
|
+
cs = 23;
|
552
|
+
end
|
553
|
+
end
|
554
|
+
when 13 then
|
555
|
+
# line 160 "lib/oga/css/lexer.rl"
|
556
|
+
begin
|
557
|
+
te = p+1
|
558
|
+
begin
|
559
|
+
emit(:T_IDENT, ts, te)
|
560
|
+
end
|
561
|
+
end
|
562
|
+
when 11 then
|
563
|
+
# line 315 "lib/oga/css/lexer.rl"
|
564
|
+
begin
|
565
|
+
te = p+1
|
566
|
+
end
|
567
|
+
when 22 then
|
568
|
+
# line 304 "lib/oga/css/lexer.rl"
|
569
|
+
begin
|
570
|
+
te = p
|
571
|
+
p = p - 1; begin add_token(:T_GREATER) end
|
572
|
+
end
|
573
|
+
when 19 then
|
574
|
+
# line 305 "lib/oga/css/lexer.rl"
|
575
|
+
begin
|
576
|
+
te = p
|
577
|
+
p = p - 1; begin add_token(:T_PLUS) end
|
578
|
+
end
|
579
|
+
when 24 then
|
580
|
+
# line 306 "lib/oga/css/lexer.rl"
|
581
|
+
begin
|
582
|
+
te = p
|
583
|
+
p = p - 1; begin add_token(:T_TILDE) end
|
584
|
+
end
|
585
|
+
when 17 then
|
586
|
+
# line 149 "lib/oga/css/lexer.rl"
|
587
|
+
begin
|
588
|
+
te = p
|
589
|
+
p = p - 1; begin
|
590
|
+
add_token(:T_COMMA)
|
591
|
+
end
|
592
|
+
end
|
593
|
+
when 16 then
|
594
|
+
# line 135 "lib/oga/css/lexer.rl"
|
595
|
+
begin
|
596
|
+
te = p
|
597
|
+
p = p - 1; begin
|
598
|
+
add_token(:T_SPACE)
|
599
|
+
end
|
600
|
+
end
|
601
|
+
when 23 then
|
602
|
+
# line 160 "lib/oga/css/lexer.rl"
|
603
|
+
begin
|
604
|
+
te = p
|
605
|
+
p = p - 1; begin
|
606
|
+
emit(:T_IDENT, ts, te)
|
607
|
+
end
|
608
|
+
end
|
609
|
+
when 18 then
|
610
|
+
# line 139 "lib/oga/css/lexer.rl"
|
611
|
+
begin
|
612
|
+
add_token(:T_HASH) end
|
613
|
+
# line 302 "lib/oga/css/lexer.rl"
|
614
|
+
begin
|
615
|
+
te = p
|
616
|
+
p = p - 1; end
|
617
|
+
when 20 then
|
618
|
+
# line 140 "lib/oga/css/lexer.rl"
|
619
|
+
begin
|
620
|
+
add_token(:T_DOT) end
|
621
|
+
# line 302 "lib/oga/css/lexer.rl"
|
622
|
+
begin
|
623
|
+
te = p
|
624
|
+
p = p - 1; end
|
625
|
+
when 21 then
|
626
|
+
# line 141 "lib/oga/css/lexer.rl"
|
627
|
+
begin
|
628
|
+
add_token(:T_COLON) end
|
629
|
+
# line 302 "lib/oga/css/lexer.rl"
|
630
|
+
begin
|
631
|
+
te = p
|
632
|
+
p = p - 1; end
|
633
|
+
when 28 then
|
634
|
+
# line 1 "NONE"
|
635
|
+
begin
|
636
|
+
te = p+1
|
637
|
+
end
|
638
|
+
# line 250 "lib/oga/css/lexer.rl"
|
639
|
+
begin
|
640
|
+
act = 3; end
|
641
|
+
# line 641 "lib/oga/css/lexer.rb"
|
642
|
+
end
|
643
|
+
end
|
644
|
+
end
|
645
|
+
if _goto_level <= _again
|
646
|
+
case _css_lexer_to_state_actions[cs]
|
647
|
+
when 9 then
|
648
|
+
# line 1 "NONE"
|
649
|
+
begin
|
650
|
+
ts = nil; end
|
651
|
+
when 25 then
|
652
|
+
# line 1 "NONE"
|
653
|
+
begin
|
654
|
+
ts = nil; end
|
655
|
+
# line 1 "NONE"
|
656
|
+
begin
|
657
|
+
act = 0
|
658
|
+
end
|
659
|
+
# line 659 "lib/oga/css/lexer.rb"
|
660
|
+
end
|
661
|
+
|
662
|
+
if cs == 0
|
663
|
+
_goto_level = _out
|
664
|
+
next
|
665
|
+
end
|
666
|
+
p += 1
|
667
|
+
if p != pe
|
668
|
+
_goto_level = _resume
|
669
|
+
next
|
670
|
+
end
|
671
|
+
end
|
672
|
+
if _goto_level <= _test_eof
|
673
|
+
if p == eof
|
674
|
+
if _css_lexer_eof_trans[cs] > 0
|
675
|
+
_trans = _css_lexer_eof_trans[cs] - 1;
|
676
|
+
_goto_level = _eof_trans
|
677
|
+
next;
|
678
|
+
end
|
679
|
+
end
|
680
|
+
|
681
|
+
end
|
682
|
+
if _goto_level <= _out
|
683
|
+
break
|
684
|
+
end
|
685
|
+
end
|
686
|
+
end
|
687
|
+
|
688
|
+
# line 81 "lib/oga/css/lexer.rl"
|
689
|
+
|
690
|
+
# % fix highlight
|
691
|
+
ensure
|
692
|
+
@block = nil
|
693
|
+
end
|
694
|
+
|
695
|
+
private
|
696
|
+
|
697
|
+
##
|
698
|
+
# Emits a token of which the value is based on the supplied start/stop
|
699
|
+
# position.
|
700
|
+
#
|
701
|
+
# @param [Symbol] type The token type.
|
702
|
+
# @param [Fixnum] start
|
703
|
+
# @param [Fixnum] stop
|
704
|
+
#
|
705
|
+
# @see [#text]
|
706
|
+
# @see [#add_token]
|
707
|
+
#
|
708
|
+
def emit(type, start, stop)
|
709
|
+
value = slice_input(start, stop)
|
710
|
+
|
711
|
+
add_token(type, value)
|
712
|
+
end
|
713
|
+
|
714
|
+
##
|
715
|
+
# Returns the text between the specified start and stop position.
|
716
|
+
#
|
717
|
+
# @param [Fixnum] start
|
718
|
+
# @param [Fixnum] stop
|
719
|
+
# @return [String]
|
720
|
+
#
|
721
|
+
def slice_input(start, stop)
|
722
|
+
return @data.byteslice(start, stop - start)
|
723
|
+
end
|
724
|
+
|
725
|
+
##
|
726
|
+
# Yields a new token to the supplied block.
|
727
|
+
#
|
728
|
+
# @param [Symbol] type The token type.
|
729
|
+
# @param [String] value The token value.
|
730
|
+
#
|
731
|
+
# @yieldparam [Symbol] type
|
732
|
+
# @yieldparam [String|NilClass] value
|
733
|
+
#
|
734
|
+
def add_token(type, value = nil)
|
735
|
+
@block.call(type, value)
|
736
|
+
end
|
737
|
+
|
738
|
+
|
739
|
+
# line 317 "lib/oga/css/lexer.rl"
|
740
|
+
|
741
|
+
end # Lexer
|
742
|
+
end # CSS
|
743
|
+
end # Oga
|