oga 1.2.3 → 1.3.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/doc/css_selectors.md +1 -1
- data/lib/oga.rb +6 -1
- data/lib/oga/blacklist.rb +0 -10
- data/lib/oga/css/lexer.rb +530 -255
- data/lib/oga/css/parser.rb +232 -230
- data/lib/oga/entity_decoder.rb +0 -4
- data/lib/oga/html/entities.rb +0 -4
- data/lib/oga/html/parser.rb +0 -4
- data/lib/oga/html/sax_parser.rb +0 -4
- data/lib/oga/lru.rb +0 -26
- data/lib/oga/oga.rb +0 -8
- data/lib/oga/ruby/generator.rb +225 -0
- data/lib/oga/ruby/node.rb +189 -0
- data/lib/oga/version.rb +1 -1
- data/lib/oga/whitelist.rb +0 -6
- data/lib/oga/xml/attribute.rb +13 -20
- data/lib/oga/xml/cdata.rb +0 -4
- data/lib/oga/xml/character_node.rb +0 -8
- data/lib/oga/xml/comment.rb +0 -4
- data/lib/oga/xml/default_namespace.rb +0 -2
- data/lib/oga/xml/doctype.rb +0 -8
- data/lib/oga/xml/document.rb +10 -14
- data/lib/oga/xml/element.rb +1 -52
- data/lib/oga/xml/entities.rb +0 -26
- data/lib/oga/xml/expanded_name.rb +12 -0
- data/lib/oga/xml/html_void_elements.rb +0 -2
- data/lib/oga/xml/lexer.rb +0 -86
- data/lib/oga/xml/namespace.rb +0 -10
- data/lib/oga/xml/node.rb +18 -34
- data/lib/oga/xml/node_set.rb +0 -50
- data/lib/oga/xml/parser.rb +13 -50
- data/lib/oga/xml/processing_instruction.rb +0 -8
- data/lib/oga/xml/pull_parser.rb +0 -18
- data/lib/oga/xml/querying.rb +58 -19
- data/lib/oga/xml/sax_parser.rb +0 -18
- data/lib/oga/xml/text.rb +0 -12
- data/lib/oga/xml/traversal.rb +0 -4
- data/lib/oga/xml/xml_declaration.rb +0 -8
- data/lib/oga/xpath/compiler.rb +1568 -0
- data/lib/oga/xpath/conversion.rb +102 -0
- data/lib/oga/xpath/lexer.rb +1844 -1238
- data/lib/oga/xpath/parser.rb +182 -153
- metadata +7 -3
- data/lib/oga/xpath/evaluator.rb +0 -1800
@@ -0,0 +1,12 @@
|
|
1
|
+
module Oga
|
2
|
+
module XML
|
3
|
+
module ExpandedName
|
4
|
+
# Returns the expanded name of the current Element or Attribute.
|
5
|
+
#
|
6
|
+
# @return [String]
|
7
|
+
def expanded_name
|
8
|
+
namespace_name ? "#{namespace_name}:#{name}" : name
|
9
|
+
end
|
10
|
+
end # ExpandedName
|
11
|
+
end # XML
|
12
|
+
end # Oga
|
@@ -1,12 +1,10 @@
|
|
1
1
|
module Oga
|
2
2
|
module XML
|
3
|
-
##
|
4
3
|
# Names of the HTML void elements that should be handled when HTML lexing
|
5
4
|
# is enabled.
|
6
5
|
#
|
7
6
|
# @api private
|
8
7
|
# @return [Oga::Whitelist]
|
9
|
-
#
|
10
8
|
HTML_VOID_ELEMENTS = Whitelist.new(%w{
|
11
9
|
area base br col command embed hr img input keygen link meta param source
|
12
10
|
track wbr
|
data/lib/oga/xml/lexer.rb
CHANGED
@@ -1,6 +1,5 @@
|
|
1
1
|
module Oga
|
2
2
|
module XML
|
3
|
-
##
|
4
3
|
# Low level lexer that supports both XML and HTML (using an extra option).
|
5
4
|
# To lex HTML input set the `:html` option to `true` when creating an
|
6
5
|
# instance of the lexer:
|
@@ -46,7 +45,6 @@ module Oga
|
|
46
45
|
# Strict mode only applies to XML documents.
|
47
46
|
#
|
48
47
|
# @private
|
49
|
-
#
|
50
48
|
class Lexer
|
51
49
|
# These are all constant/frozen to remove the need for String allocations
|
52
50
|
# every time they are referenced in the lexer.
|
@@ -96,12 +94,9 @@ module Oga
|
|
96
94
|
HTML_CLOSE_SELF[key.upcase] = HTML_CLOSE_SELF[key]
|
97
95
|
end
|
98
96
|
|
99
|
-
##
|
100
97
|
# Names of HTML tags of which the content should be lexed as-is.
|
101
|
-
#
|
102
98
|
LITERAL_HTML_ELEMENTS = Whitelist.new([HTML_SCRIPT, HTML_STYLE])
|
103
99
|
|
104
|
-
##
|
105
100
|
# @param [String|IO] data The data to lex. This can either be a String or
|
106
101
|
# an IO instance.
|
107
102
|
#
|
@@ -113,7 +108,6 @@ module Oga
|
|
113
108
|
#
|
114
109
|
# @option options [TrueClass|FalseClass] :strict Enables/disables strict
|
115
110
|
# parsing of XML documents, disabled by default.
|
116
|
-
#
|
117
111
|
def initialize(data, options = {})
|
118
112
|
@data = data
|
119
113
|
@html = options[:html]
|
@@ -122,11 +116,9 @@ module Oga
|
|
122
116
|
reset
|
123
117
|
end
|
124
118
|
|
125
|
-
##
|
126
119
|
# Resets the internal state of the lexer. Typically you don't need to
|
127
120
|
# call this method yourself as its called by #lex after lexing a given
|
128
121
|
# String.
|
129
|
-
#
|
130
122
|
def reset
|
131
123
|
@line = 1
|
132
124
|
@elements = []
|
@@ -136,12 +128,10 @@ module Oga
|
|
136
128
|
reset_native
|
137
129
|
end
|
138
130
|
|
139
|
-
##
|
140
131
|
# Yields the data to lex to the supplied block.
|
141
132
|
#
|
142
133
|
# @return [String]
|
143
134
|
# @yieldparam [String]
|
144
|
-
#
|
145
135
|
def read_data
|
146
136
|
if @data.is_a?(String)
|
147
137
|
yield @data
|
@@ -157,7 +147,6 @@ module Oga
|
|
157
147
|
end
|
158
148
|
end
|
159
149
|
|
160
|
-
##
|
161
150
|
# Gathers all the tokens for the input and returns them as an Array.
|
162
151
|
#
|
163
152
|
# This method resets the internal state of the lexer after consuming the
|
@@ -165,7 +154,6 @@ module Oga
|
|
165
154
|
#
|
166
155
|
# @see #advance
|
167
156
|
# @return [Array]
|
168
|
-
#
|
169
157
|
def lex
|
170
158
|
tokens = []
|
171
159
|
|
@@ -178,7 +166,6 @@ module Oga
|
|
178
166
|
tokens
|
179
167
|
end
|
180
168
|
|
181
|
-
##
|
182
169
|
# Advances through the input and generates the corresponding tokens. Each
|
183
170
|
# token is yielded to the supplied block.
|
184
171
|
#
|
@@ -196,7 +183,6 @@ module Oga
|
|
196
183
|
# @yieldparam [Symbol] type
|
197
184
|
# @yieldparam [String] value
|
198
185
|
# @yieldparam [Fixnum] line
|
199
|
-
#
|
200
186
|
def advance(&block)
|
201
187
|
@block = block
|
202
188
|
|
@@ -212,44 +198,33 @@ module Oga
|
|
212
198
|
@block = nil
|
213
199
|
end
|
214
200
|
|
215
|
-
##
|
216
201
|
# @return [TrueClass|FalseClass]
|
217
|
-
#
|
218
202
|
def html?
|
219
203
|
@html == true
|
220
204
|
end
|
221
205
|
|
222
|
-
##
|
223
206
|
# @return [TrueClass|FalseClass]
|
224
|
-
#
|
225
207
|
def strict?
|
226
208
|
@strict
|
227
209
|
end
|
228
210
|
|
229
|
-
##
|
230
211
|
# @return [TrueClass|FalseClass]
|
231
|
-
#
|
232
212
|
def html_script?
|
233
213
|
html? && current_element == HTML_SCRIPT
|
234
214
|
end
|
235
215
|
|
236
|
-
##
|
237
216
|
# @return [TrueClass|FalseClass]
|
238
|
-
#
|
239
217
|
def html_style?
|
240
218
|
html? && current_element == HTML_STYLE
|
241
219
|
end
|
242
220
|
|
243
221
|
private
|
244
222
|
|
245
|
-
##
|
246
223
|
# @param [Fixnum] amount The amount of lines to advance.
|
247
|
-
#
|
248
224
|
def advance_line(amount = 1)
|
249
225
|
@line += amount
|
250
226
|
end
|
251
227
|
|
252
|
-
##
|
253
228
|
# Calls the supplied block with the information of the current token.
|
254
229
|
#
|
255
230
|
# @param [Symbol] type The token type.
|
@@ -258,192 +233,145 @@ module Oga
|
|
258
233
|
# @yieldparam [String] type
|
259
234
|
# @yieldparam [String] value
|
260
235
|
# @yieldparam [Fixnum] line
|
261
|
-
#
|
262
236
|
def add_token(type, value = nil)
|
263
237
|
@block.call(type, value, @line)
|
264
238
|
end
|
265
239
|
|
266
|
-
##
|
267
240
|
# Returns the name of the element we're currently in.
|
268
241
|
#
|
269
242
|
# @return [String]
|
270
|
-
#
|
271
243
|
def current_element
|
272
244
|
@elements.last
|
273
245
|
end
|
274
246
|
|
275
|
-
##
|
276
247
|
# Called when processing a single quote.
|
277
|
-
#
|
278
248
|
def on_string_squote
|
279
249
|
add_token(:T_STRING_SQUOTE)
|
280
250
|
end
|
281
251
|
|
282
|
-
##
|
283
252
|
# Called when processing a double quote.
|
284
|
-
#
|
285
253
|
def on_string_dquote
|
286
254
|
add_token(:T_STRING_DQUOTE)
|
287
255
|
end
|
288
256
|
|
289
|
-
##
|
290
257
|
# Called when processing the body of a string.
|
291
258
|
#
|
292
259
|
# @param [String] value The data between the quotes.
|
293
|
-
#
|
294
260
|
def on_string_body(value)
|
295
261
|
add_token(:T_STRING_BODY, value)
|
296
262
|
end
|
297
263
|
|
298
|
-
##
|
299
264
|
# Called when a doctype starts.
|
300
|
-
#
|
301
265
|
def on_doctype_start
|
302
266
|
add_token(:T_DOCTYPE_START)
|
303
267
|
end
|
304
268
|
|
305
|
-
##
|
306
269
|
# Called on the identifier specifying the type of the doctype.
|
307
270
|
#
|
308
271
|
# @param [String] value
|
309
|
-
#
|
310
272
|
def on_doctype_type(value)
|
311
273
|
add_token(:T_DOCTYPE_TYPE, value)
|
312
274
|
end
|
313
275
|
|
314
|
-
##
|
315
276
|
# Called on the identifier specifying the name of the doctype.
|
316
277
|
#
|
317
278
|
# @param [String] value
|
318
|
-
#
|
319
279
|
def on_doctype_name(value)
|
320
280
|
add_token(:T_DOCTYPE_NAME, value)
|
321
281
|
end
|
322
282
|
|
323
|
-
##
|
324
283
|
# Called on the end of a doctype.
|
325
|
-
#
|
326
284
|
def on_doctype_end
|
327
285
|
add_token(:T_DOCTYPE_END)
|
328
286
|
end
|
329
287
|
|
330
|
-
##
|
331
288
|
# Called on an inline doctype block.
|
332
289
|
#
|
333
290
|
# @param [String] value
|
334
|
-
#
|
335
291
|
def on_doctype_inline(value)
|
336
292
|
add_token(:T_DOCTYPE_INLINE, value)
|
337
293
|
end
|
338
294
|
|
339
|
-
##
|
340
295
|
# Called on the open CDATA tag.
|
341
|
-
#
|
342
296
|
def on_cdata_start
|
343
297
|
add_token(:T_CDATA_START)
|
344
298
|
end
|
345
299
|
|
346
|
-
##
|
347
300
|
# Called on the closing CDATA tag.
|
348
|
-
#
|
349
301
|
def on_cdata_end
|
350
302
|
add_token(:T_CDATA_END)
|
351
303
|
end
|
352
304
|
|
353
|
-
##
|
354
305
|
# Called for the body of a CDATA tag.
|
355
306
|
#
|
356
307
|
# @param [String] value
|
357
|
-
#
|
358
308
|
def on_cdata_body(value)
|
359
309
|
add_token(:T_CDATA_BODY, value)
|
360
310
|
end
|
361
311
|
|
362
|
-
##
|
363
312
|
# Called on the open comment tag.
|
364
|
-
#
|
365
313
|
def on_comment_start
|
366
314
|
add_token(:T_COMMENT_START)
|
367
315
|
end
|
368
316
|
|
369
|
-
##
|
370
317
|
# Called on the closing comment tag.
|
371
|
-
#
|
372
318
|
def on_comment_end
|
373
319
|
add_token(:T_COMMENT_END)
|
374
320
|
end
|
375
321
|
|
376
|
-
##
|
377
322
|
# Called on a comment.
|
378
323
|
#
|
379
324
|
# @param [String] value
|
380
|
-
#
|
381
325
|
def on_comment_body(value)
|
382
326
|
add_token(:T_COMMENT_BODY, value)
|
383
327
|
end
|
384
328
|
|
385
|
-
##
|
386
329
|
# Called on the start of an XML declaration tag.
|
387
|
-
#
|
388
330
|
def on_xml_decl_start
|
389
331
|
add_token(:T_XML_DECL_START)
|
390
332
|
end
|
391
333
|
|
392
|
-
##
|
393
334
|
# Called on the end of an XML declaration tag.
|
394
|
-
#
|
395
335
|
def on_xml_decl_end
|
396
336
|
add_token(:T_XML_DECL_END)
|
397
337
|
end
|
398
338
|
|
399
|
-
##
|
400
339
|
# Called on the start of a processing instruction.
|
401
|
-
#
|
402
340
|
def on_proc_ins_start
|
403
341
|
add_token(:T_PROC_INS_START)
|
404
342
|
end
|
405
343
|
|
406
|
-
##
|
407
344
|
# Called on a processing instruction name.
|
408
345
|
#
|
409
346
|
# @param [String] value
|
410
|
-
#
|
411
347
|
def on_proc_ins_name(value)
|
412
348
|
add_token(:T_PROC_INS_NAME, value)
|
413
349
|
end
|
414
350
|
|
415
|
-
##
|
416
351
|
# Called on the body of a processing instruction.
|
417
352
|
#
|
418
353
|
# @param [String] value
|
419
|
-
#
|
420
354
|
def on_proc_ins_body(value)
|
421
355
|
add_token(:T_PROC_INS_BODY, value)
|
422
356
|
end
|
423
357
|
|
424
|
-
##
|
425
358
|
# Called on the end of a processing instruction.
|
426
|
-
#
|
427
359
|
def on_proc_ins_end
|
428
360
|
add_token(:T_PROC_INS_END)
|
429
361
|
end
|
430
362
|
|
431
|
-
##
|
432
363
|
# Called on the name of an element.
|
433
364
|
#
|
434
365
|
# @param [String] name The name of the element, including namespace.
|
435
|
-
#
|
436
366
|
def on_element_name(name)
|
437
367
|
before_html_element_name(name) if html?
|
438
368
|
|
439
369
|
add_element(name)
|
440
370
|
end
|
441
371
|
|
442
|
-
##
|
443
372
|
# Handles inserting of any missing tags whenever a new HTML tag is opened.
|
444
373
|
#
|
445
374
|
# @param [String] name
|
446
|
-
#
|
447
375
|
def before_html_element_name(name)
|
448
376
|
close_current = HTML_CLOSE_SELF[current_element]
|
449
377
|
|
@@ -463,27 +391,21 @@ module Oga
|
|
463
391
|
end
|
464
392
|
end
|
465
393
|
|
466
|
-
##
|
467
394
|
# @param [String] name
|
468
|
-
#
|
469
395
|
def add_element(name)
|
470
396
|
@elements << name
|
471
397
|
|
472
398
|
add_token(:T_ELEM_NAME, name)
|
473
399
|
end
|
474
400
|
|
475
|
-
##
|
476
401
|
# Called on the element namespace.
|
477
402
|
#
|
478
403
|
# @param [String] namespace
|
479
|
-
#
|
480
404
|
def on_element_ns(namespace)
|
481
405
|
add_token(:T_ELEM_NS, namespace)
|
482
406
|
end
|
483
407
|
|
484
|
-
##
|
485
408
|
# Called on the closing `>` of the open tag of an element.
|
486
|
-
#
|
487
409
|
def on_element_open_end
|
488
410
|
return unless html?
|
489
411
|
|
@@ -496,12 +418,10 @@ module Oga
|
|
496
418
|
end
|
497
419
|
end
|
498
420
|
|
499
|
-
##
|
500
421
|
# Called on the closing tag of an element.
|
501
422
|
#
|
502
423
|
# @param [String] name The name of the element (minus namespace
|
503
424
|
# prefix). This is not set for self closing tags.
|
504
|
-
#
|
505
425
|
def on_element_end(name = nil)
|
506
426
|
return if @elements.empty?
|
507
427
|
|
@@ -520,31 +440,25 @@ module Oga
|
|
520
440
|
@elements.pop
|
521
441
|
end
|
522
442
|
|
523
|
-
##
|
524
443
|
# Called on regular text values.
|
525
444
|
#
|
526
445
|
# @param [String] value
|
527
|
-
#
|
528
446
|
def on_text(value)
|
529
447
|
return if value.empty?
|
530
448
|
|
531
449
|
add_token(:T_TEXT, value)
|
532
450
|
end
|
533
451
|
|
534
|
-
##
|
535
452
|
# Called on attribute namespaces.
|
536
453
|
#
|
537
454
|
# @param [String] value
|
538
|
-
#
|
539
455
|
def on_attribute_ns(value)
|
540
456
|
add_token(:T_ATTR_NS, value)
|
541
457
|
end
|
542
458
|
|
543
|
-
##
|
544
459
|
# Called on tag attributes.
|
545
460
|
#
|
546
461
|
# @param [String] value
|
547
|
-
#
|
548
462
|
def on_attribute(value)
|
549
463
|
add_token(:T_ATTR, value)
|
550
464
|
end
|
data/lib/oga/xml/namespace.rb
CHANGED
@@ -1,9 +1,7 @@
|
|
1
1
|
module Oga
|
2
2
|
module XML
|
3
|
-
##
|
4
3
|
# The Namespace class contains information about XML namespaces such as the
|
5
4
|
# name and URI.
|
6
|
-
#
|
7
5
|
class Namespace
|
8
6
|
# @return [String]
|
9
7
|
attr_accessor :name
|
@@ -11,35 +9,27 @@ module Oga
|
|
11
9
|
# @return [String]
|
12
10
|
attr_accessor :uri
|
13
11
|
|
14
|
-
##
|
15
12
|
# @param [Hash] options
|
16
13
|
#
|
17
14
|
# @option options [String] :name
|
18
15
|
# @option options [String] :uri
|
19
|
-
#
|
20
16
|
def initialize(options = {})
|
21
17
|
@name = options[:name]
|
22
18
|
@uri = options[:uri]
|
23
19
|
end
|
24
20
|
|
25
|
-
##
|
26
21
|
# @return [String]
|
27
|
-
#
|
28
22
|
def to_s
|
29
23
|
name.to_s
|
30
24
|
end
|
31
25
|
|
32
|
-
##
|
33
26
|
# @return [String]
|
34
|
-
#
|
35
27
|
def inspect
|
36
28
|
"Namespace(name: #{name.inspect} uri: #{uri.inspect})"
|
37
29
|
end
|
38
30
|
|
39
|
-
##
|
40
31
|
# @param [Oga::XML::Namespace] other
|
41
32
|
# @return [TrueClass|FalseClass]
|
42
|
-
#
|
43
33
|
def ==(other)
|
44
34
|
other.is_a?(self.class) && name == other.name && uri == other.uri
|
45
35
|
end
|