oga 1.2.3-java → 1.3.0-java

Sign up to get free protection for your applications and to get access to all the features.
Files changed (46) hide show
  1. checksums.yaml +4 -4
  2. data/doc/css_selectors.md +1 -1
  3. data/lib/liboga.jar +0 -0
  4. data/lib/oga.rb +6 -1
  5. data/lib/oga/blacklist.rb +0 -10
  6. data/lib/oga/css/lexer.rb +530 -255
  7. data/lib/oga/css/parser.rb +232 -230
  8. data/lib/oga/entity_decoder.rb +0 -4
  9. data/lib/oga/html/entities.rb +0 -4
  10. data/lib/oga/html/parser.rb +0 -4
  11. data/lib/oga/html/sax_parser.rb +0 -4
  12. data/lib/oga/lru.rb +0 -26
  13. data/lib/oga/oga.rb +0 -8
  14. data/lib/oga/ruby/generator.rb +225 -0
  15. data/lib/oga/ruby/node.rb +189 -0
  16. data/lib/oga/version.rb +1 -1
  17. data/lib/oga/whitelist.rb +0 -6
  18. data/lib/oga/xml/attribute.rb +13 -20
  19. data/lib/oga/xml/cdata.rb +0 -4
  20. data/lib/oga/xml/character_node.rb +0 -8
  21. data/lib/oga/xml/comment.rb +0 -4
  22. data/lib/oga/xml/default_namespace.rb +0 -2
  23. data/lib/oga/xml/doctype.rb +0 -8
  24. data/lib/oga/xml/document.rb +10 -14
  25. data/lib/oga/xml/element.rb +1 -52
  26. data/lib/oga/xml/entities.rb +0 -26
  27. data/lib/oga/xml/expanded_name.rb +12 -0
  28. data/lib/oga/xml/html_void_elements.rb +0 -2
  29. data/lib/oga/xml/lexer.rb +0 -86
  30. data/lib/oga/xml/namespace.rb +0 -10
  31. data/lib/oga/xml/node.rb +18 -34
  32. data/lib/oga/xml/node_set.rb +0 -50
  33. data/lib/oga/xml/parser.rb +13 -50
  34. data/lib/oga/xml/processing_instruction.rb +0 -8
  35. data/lib/oga/xml/pull_parser.rb +0 -18
  36. data/lib/oga/xml/querying.rb +58 -19
  37. data/lib/oga/xml/sax_parser.rb +0 -18
  38. data/lib/oga/xml/text.rb +0 -12
  39. data/lib/oga/xml/traversal.rb +0 -4
  40. data/lib/oga/xml/xml_declaration.rb +0 -8
  41. data/lib/oga/xpath/compiler.rb +1568 -0
  42. data/lib/oga/xpath/conversion.rb +102 -0
  43. data/lib/oga/xpath/lexer.rb +1844 -1238
  44. data/lib/oga/xpath/parser.rb +182 -153
  45. metadata +7 -3
  46. data/lib/oga/xpath/evaluator.rb +0 -1800
@@ -1,6 +1,5 @@
1
1
  module Oga
2
2
  module XML
3
- ##
4
3
  # The SaxParser class provides the basic interface for writing custom SAX
5
4
  # parsers. All callback methods defined in {Oga::XML::Parser} are delegated
6
5
  # to a dedicated handler class.
@@ -66,12 +65,9 @@ module Oga
66
65
  # attribute names (optionally prefixed by their namespace) and their values.
67
66
  # You can overwrite `on_attribute` to control individual attributes and
68
67
  # `on_attributes` to control the final set.
69
- #
70
68
  class SaxParser < Parser
71
- ##
72
69
  # @param [Object] handler The SAX handler to delegate callbacks to.
73
70
  # @see [Oga::XML::Parser#initialize]
74
- #
75
71
  def initialize(handler, *args)
76
72
  @handler = handler
77
73
 
@@ -89,38 +85,32 @@ module Oga
89
85
  EOF
90
86
  end
91
87
 
92
- ##
93
88
  # Manually overwrite `on_element` so we can ensure that `after_element`
94
89
  # always receives the namespace and name.
95
90
  #
96
91
  # @see [Oga::XML::Parser#on_element]
97
92
  # @return [Array]
98
- #
99
93
  def on_element(namespace, name, attrs = [])
100
94
  run_callback(:on_element, namespace, name, attrs)
101
95
 
102
96
  [namespace, name]
103
97
  end
104
98
 
105
- ##
106
99
  # Manually overwrite `after_element` so it can take a namespace and name.
107
100
  # This differs a bit from the regular `after_element` which only takes an
108
101
  # {Oga::XML::Element} instance.
109
102
  #
110
103
  # @param [Array] namespace_with_name
111
- #
112
104
  def after_element(namespace_with_name)
113
105
  run_callback(:after_element, *namespace_with_name)
114
106
 
115
107
  return
116
108
  end
117
109
 
118
- ##
119
110
  # Manually overwrite this method since for this one we _do_ want the
120
111
  # return value so it can be passed to `on_element`.
121
112
  #
122
113
  # @see [Oga::XML::Parser#on_attribute]
123
- #
124
114
  def on_attribute(name, ns = nil, value = nil)
125
115
  if @handler.respond_to?(:on_attribute)
126
116
  return run_callback(:on_attribute, name, ns, value)
@@ -135,12 +125,10 @@ module Oga
135
125
  {key => value}
136
126
  end
137
127
 
138
- ##
139
128
  # Merges the attributes together into a Hash.
140
129
  #
141
130
  # @param [Array] attrs
142
131
  # @return [Hash]
143
- #
144
132
  def on_attributes(attrs)
145
133
  if @handler.respond_to?(:on_attributes)
146
134
  return run_callback(:on_attributes, attrs)
@@ -156,9 +144,7 @@ module Oga
156
144
  merged
157
145
  end
158
146
 
159
- ##
160
147
  # @param [String] text
161
- #
162
148
  def on_text(text)
163
149
  if @handler.respond_to?(:on_text)
164
150
  unless inside_literal_html?
@@ -173,17 +159,13 @@ module Oga
173
159
 
174
160
  private
175
161
 
176
- ##
177
162
  # @return [TrueClass|FalseClass]
178
- #
179
163
  def inside_literal_html?
180
164
  @lexer.html_script? || @lexer.html_style?
181
165
  end
182
166
 
183
- ##
184
167
  # @param [Symbol] method
185
168
  # @param [Array] args
186
- #
187
169
  def run_callback(method, *args)
188
170
  @handler.send(method, *args) if @handler.respond_to?(method)
189
171
  end
@@ -1,9 +1,7 @@
1
1
  module Oga
2
2
  module XML
3
- ##
4
3
  # Class containing information about a single text node. Text nodes don't
5
4
  # have any children, attributes and the likes; just text.
6
- #
7
5
  class Text < CharacterNode
8
6
  def initialize(*args)
9
7
  super
@@ -11,20 +9,16 @@ module Oga
11
9
  @decoded = false
12
10
  end
13
11
 
14
- ##
15
12
  # @param [String] value
16
- #
17
13
  def text=(value)
18
14
  @decoded = false
19
15
  @text = value
20
16
  end
21
17
 
22
- ##
23
18
  # Returns the text as a String. Upon the first call any XML/HTML entities
24
19
  # are decoded.
25
20
  #
26
21
  # @return [String]
27
- #
28
22
  def text
29
23
  if decode_entities?
30
24
  @text = EntityDecoder.try_decode(@text, html?)
@@ -34,9 +28,7 @@ module Oga
34
28
  @text
35
29
  end
36
30
 
37
- ##
38
31
  # @see [Oga::XML::CharacterNode#to_xml]
39
- #
40
32
  def to_xml
41
33
  return super if inside_literal_html?
42
34
 
@@ -45,16 +37,12 @@ module Oga
45
37
 
46
38
  private
47
39
 
48
- ##
49
40
  # @return [TrueClass|FalseClass]
50
- #
51
41
  def decode_entities?
52
42
  !@decoded && !inside_literal_html?
53
43
  end
54
44
 
55
- ##
56
45
  # @return [TrueClass|FalseClass]
57
- #
58
46
  def inside_literal_html?
59
47
  node = parent
60
48
 
@@ -1,10 +1,7 @@
1
1
  module Oga
2
2
  module XML
3
- ##
4
3
  # Module that provides methods to traverse DOM trees.
5
- #
6
4
  module Traversal
7
- ##
8
5
  # Traverses through the node and yields every child node to the supplied
9
6
  # block.
10
7
  #
@@ -29,7 +26,6 @@ module Oga
29
26
  # end
30
27
  #
31
28
  # @yieldparam [Oga::XML::Node] The current node.
32
- #
33
29
  def each_node
34
30
  visit = children.to_a.reverse
35
31
 
@@ -1,8 +1,6 @@
1
1
  module Oga
2
2
  module XML
3
- ##
4
3
  # Class containing information about an XML declaration tag.
5
- #
6
4
  class XmlDeclaration
7
5
  # @return [String]
8
6
  attr_accessor :version
@@ -14,24 +12,20 @@ module Oga
14
12
  # @return [String]
15
13
  attr_accessor :standalone
16
14
 
17
- ##
18
15
  # @param [Hash] options
19
16
  #
20
17
  # @option options [String] :version
21
18
  # @option options [String] :encoding
22
19
  # @option options [String] :standalone
23
- #
24
20
  def initialize(options = {})
25
21
  @version = options[:version] || '1.0'
26
22
  @encoding = options[:encoding] || 'UTF-8'
27
23
  @standalone = options[:standalone]
28
24
  end
29
25
 
30
- ##
31
26
  # Converts the declaration tag to XML.
32
27
  #
33
28
  # @return [String]
34
- #
35
29
  def to_xml
36
30
  pairs = []
37
31
 
@@ -44,9 +38,7 @@ module Oga
44
38
  "<?xml #{pairs.join(' ')} ?>"
45
39
  end
46
40
 
47
- ##
48
41
  # @return [String]
49
- #
50
42
  def inspect
51
43
  segments = []
52
44
 
@@ -0,0 +1,1568 @@
1
+ module Oga
2
+ module XPath
3
+ # Compiling of XPath ASTs into Ruby code.
4
+ #
5
+ # The Compiler class can be used to turn an XPath AST into Ruby source code
6
+ # that can be executed to match XML nodes in a given input document/element.
7
+ # Compiled source code is cached per expression, removing the need for
8
+ # recompiling the same expression over and over again.
9
+ #
10
+ # @private
11
+ class Compiler
12
+ # @return [Oga::LRU]
13
+ CACHE = LRU.new
14
+
15
+ # Wildcard for node names/namespace prefixes.
16
+ STAR = '*'
17
+
18
+ # Node types that require a NodeSet to push nodes into.
19
+ RETURN_NODESET = [:path, :absolute_path, :axis, :predicate]
20
+
21
+ # Hash containing all operator callbacks, the conversion methods and the
22
+ # Ruby methods to use.
23
+ OPERATORS = {
24
+ :on_add => [:to_float, :+],
25
+ :on_sub => [:to_float, :-],
26
+ :on_div => [:to_float, :/],
27
+ :on_gt => [:to_float, :>],
28
+ :on_gte => [:to_float, :>=],
29
+ :on_lt => [:to_float, :<],
30
+ :on_lte => [:to_float, :<=],
31
+ :on_mul => [:to_float, :*],
32
+ :on_mod => [:to_float, :%],
33
+ :on_and => [:to_boolean, :and],
34
+ :on_or => [:to_boolean, :or]
35
+ }
36
+
37
+ # Compiles and caches an AST.
38
+ #
39
+ # @see [#compile]
40
+ def self.compile_with_cache(ast)
41
+ CACHE.get_or_set(ast) { new.compile(ast) }
42
+ end
43
+
44
+ def initialize
45
+ reset
46
+ end
47
+
48
+ # Resets the internal state.
49
+ def reset
50
+ @literal_id = 0
51
+
52
+ @predicate_nodesets = []
53
+ @predicate_indexes = []
54
+ end
55
+
56
+ # Compiles an XPath AST into a Ruby Proc.
57
+ #
58
+ # @param [AST::Node] ast
59
+ # @return [Proc]
60
+ def compile(ast)
61
+ document = literal(:node)
62
+ matched = matched_literal
63
+
64
+ if return_nodeset?(ast)
65
+ ruby_ast = process(ast, document) { |node| matched.push(node) }
66
+ else
67
+ ruby_ast = process(ast, document)
68
+ end
69
+
70
+ vars = variables_literal.assign(self.nil)
71
+
72
+ proc_ast = literal(:lambda).add_block(document, vars) do
73
+ input_assign = original_input_literal.assign(document)
74
+
75
+ if return_nodeset?(ast)
76
+ body = matched.assign(literal(XML::NodeSet).new)
77
+ .followed_by(ruby_ast)
78
+ .followed_by(matched)
79
+ else
80
+ body = ruby_ast
81
+ end
82
+
83
+ input_assign.followed_by(body)
84
+ end
85
+
86
+ generator = Ruby::Generator.new
87
+ source = generator.process(proc_ast)
88
+
89
+ eval(source)
90
+ ensure
91
+ reset
92
+ end
93
+
94
+ # Processes a single XPath AST node.
95
+ #
96
+ # @param [AST::Node] ast
97
+ # @param [Oga::Ruby::Node] input
98
+ # @return [Oga::Ruby::Node]
99
+ def process(ast, input, &block)
100
+ send("on_#{ast.type}", ast, input, &block)
101
+ end
102
+
103
+ # @param [AST::Node] ast
104
+ # @param [Oga::Ruby::Node] input
105
+ # @return [Oga::Ruby::Node]
106
+ def on_absolute_path(ast, input, &block)
107
+ if ast.children.empty?
108
+ matched_literal.push(input.root_node)
109
+ else
110
+ process(ast.children[0], input.root_node, &block)
111
+ end
112
+ end
113
+
114
+ # Dispatches the processing of axes to dedicated methods. This works
115
+ # similar to {#process} except the handler names are "on_axis_X" with "X"
116
+ # being the axis name.
117
+ #
118
+ # @param [AST::Node] ast
119
+ # @param [Oga::Ruby::Node] input
120
+ # @return [Oga::Ruby::Node]
121
+ def on_axis(ast, input, &block)
122
+ name, test, following = *ast
123
+
124
+ handler = name.gsub('-', '_')
125
+
126
+ send(:"on_axis_#{handler}", test, input) do |matched|
127
+ process_following_or_yield(following, matched, &block)
128
+ end
129
+ end
130
+
131
+ # @param [AST::Node] ast
132
+ # @param [Oga::Ruby::Node] input
133
+ # @return [Oga::Ruby::Node]
134
+ def on_axis_child(ast, input)
135
+ child = unique_literal(:child)
136
+
137
+ document_or_node(input).if_true do
138
+ input.children.each.add_block(child) do
139
+ process(ast, child).if_true { yield child }
140
+ end
141
+ end
142
+ end
143
+
144
+ # @param [AST::Node] ast
145
+ # @param [Oga::Ruby::Node] input
146
+ # @return [Oga::Ruby::Node]
147
+ def on_axis_attribute(ast, input)
148
+ input.is_a?(XML::Element).if_true do
149
+ attribute = unique_literal(:attribute)
150
+
151
+ input.attributes.each.add_block(attribute) do
152
+ name_match = match_name_and_namespace(ast, attribute)
153
+
154
+ if name_match
155
+ name_match.if_true { yield attribute }
156
+ else
157
+ yield attribute
158
+ end
159
+ end
160
+ end
161
+ end
162
+
163
+ # @param [AST::Node] ast
164
+ # @param [Oga::Ruby::Node] input
165
+ # @return [Oga::Ruby::Node]
166
+ def on_axis_ancestor_or_self(ast, input)
167
+ parent = unique_literal(:parent)
168
+
169
+ process(ast, input).and(input.is_a?(XML::Node))
170
+ .if_true { yield input }
171
+ .followed_by do
172
+ attribute_or_node(input).if_true do
173
+ input.each_ancestor.add_block(parent) do
174
+ process(ast, parent).if_true { yield parent }
175
+ end
176
+ end
177
+ end
178
+ end
179
+
180
+ # @param [AST::Node] ast
181
+ # @param [Oga::Ruby::Node] input
182
+ # @return [Oga::Ruby::Node]
183
+ def on_axis_ancestor(ast, input)
184
+ parent = unique_literal(:parent)
185
+
186
+ attribute_or_node(input).if_true do
187
+ input.each_ancestor.add_block(parent) do
188
+ process(ast, parent).if_true { yield parent }
189
+ end
190
+ end
191
+ end
192
+
193
+ # @param [AST::Node] ast
194
+ # @param [Oga::Ruby::Node] input
195
+ # @return [Oga::Ruby::Node]
196
+ def on_axis_descendant_or_self(ast, input)
197
+ node = unique_literal(:descendant)
198
+
199
+ document_or_node(input).if_true do
200
+ process(ast, input)
201
+ .if_true { yield input }
202
+ .followed_by do
203
+ input.each_node.add_block(node) do
204
+ process(ast, node).if_true { yield node }
205
+ end
206
+ end
207
+ end
208
+ end
209
+
210
+ # @param [AST::Node] ast
211
+ # @param [Oga::Ruby::Node] input
212
+ # @return [Oga::Ruby::Node]
213
+ def on_axis_descendant(ast, input)
214
+ node = unique_literal(:descendant)
215
+
216
+ document_or_node(input).if_true do
217
+ input.each_node.add_block(node) do
218
+ process(ast, node).if_true { yield node }
219
+ end
220
+ end
221
+ end
222
+
223
+ # @param [AST::Node] ast
224
+ # @param [Oga::Ruby::Node] input
225
+ # @return [Oga::Ruby::Node]
226
+ def on_axis_parent(ast, input)
227
+ parent = unique_literal(:parent)
228
+
229
+ attribute_or_node(input).if_true do
230
+ parent.assign(input.parent).followed_by do
231
+ process(ast, parent).if_true { yield parent }
232
+ end
233
+ end
234
+ end
235
+
236
+ # @param [AST::Node] ast
237
+ # @param [Oga::Ruby::Node] input
238
+ # @return [Oga::Ruby::Node]
239
+ def on_axis_self(ast, input)
240
+ process(ast, input).if_true { yield input }
241
+ end
242
+
243
+ # @param [AST::Node] ast
244
+ # @param [Oga::Ruby::Node] input
245
+ # @return [Oga::Ruby::Node]
246
+ def on_axis_following_sibling(ast, input)
247
+ orig_input = original_input_literal
248
+ doc_node = literal(:doc_node)
249
+ check = literal(:check)
250
+ parent = literal(:parent)
251
+ root = literal(:root)
252
+
253
+ orig_input.is_a?(XML::Node)
254
+ .if_true { root.assign(orig_input.parent) }
255
+ .else { root.assign(orig_input) }
256
+ .followed_by do
257
+ input.is_a?(XML::Node).and(input.parent)
258
+ .if_true { parent.assign(input.parent) }
259
+ .else { parent.assign(self.nil) }
260
+ end
261
+ .followed_by(check.assign(self.false))
262
+ .followed_by do
263
+ document_or_node(root).if_true do
264
+ root.each_node.add_block(doc_node) do
265
+ doc_node.eq(input)
266
+ .if_true do
267
+ check.assign(self.true)
268
+ .followed_by(throw_message(:skip_children))
269
+ end
270
+ .followed_by do
271
+ check.not.or(parent != doc_node.parent).if_true do
272
+ send_message(:next)
273
+ end
274
+ end
275
+ .followed_by do
276
+ process(ast, doc_node).if_true { yield doc_node }
277
+ end
278
+ end
279
+ end
280
+ end
281
+ end
282
+
283
+ # @param [AST::Node] ast
284
+ # @param [Oga::Ruby::Node] input
285
+ # @return [Oga::Ruby::Node]
286
+ def on_axis_following(ast, input)
287
+ orig_input = original_input_literal
288
+ doc_node = literal(:doc_node)
289
+ check = literal(:check)
290
+ root = literal(:root)
291
+
292
+ orig_input.is_a?(XML::Node)
293
+ .if_true { root.assign(orig_input.root_node) }
294
+ .else { root.assign(orig_input) }
295
+ .followed_by(check.assign(self.false))
296
+ .followed_by do
297
+ document_or_node(root).if_true do
298
+ root.each_node.add_block(doc_node) do
299
+ doc_node.eq(input)
300
+ .if_true do
301
+ check.assign(self.true)
302
+ .followed_by(throw_message(:skip_children))
303
+ end
304
+ .followed_by do
305
+ check.if_false { send_message(:next) }
306
+ end
307
+ .followed_by do
308
+ process(ast, doc_node).if_true { yield doc_node }
309
+ end
310
+ end
311
+ end
312
+ end
313
+ end
314
+
315
+ # @param [AST::Node] ast
316
+ # @param [Oga::Ruby::Node] input
317
+ # @return [Oga::Ruby::Node]
318
+ def on_axis_namespace(ast, input)
319
+ underscore = literal(:_)
320
+ node = unique_literal(:namespace)
321
+
322
+ name = string(ast.children[1])
323
+ star = string(STAR)
324
+
325
+ input.is_a?(XML::Element).if_true do
326
+ input.available_namespaces.each.add_block(underscore, node) do
327
+ node.name.eq(name).or(name.eq(star)).if_true { yield node }
328
+ end
329
+ end
330
+ end
331
+
332
+ # @param [AST::Node] ast
333
+ # @param [Oga::Ruby::Node] input
334
+ # @return [Oga::Ruby::Node]
335
+ def on_axis_preceding(ast, input)
336
+ root = literal(:root)
337
+ doc_node = literal(:doc_node)
338
+
339
+ input.is_a?(XML::Node).if_true do
340
+ root.assign(input.root_node)
341
+ .followed_by do
342
+ document_or_node(root).if_true do
343
+ root.each_node.add_block(doc_node) do
344
+ doc_node.eq(input)
345
+ .if_true { self.break }
346
+ .followed_by do
347
+ process(ast, doc_node).if_true { yield doc_node }
348
+ end
349
+ end
350
+ end
351
+ end
352
+ end
353
+ end
354
+
355
+ # @param [AST::Node] ast
356
+ # @param [Oga::Ruby::Node] input
357
+ # @return [Oga::Ruby::Node]
358
+ def on_axis_preceding_sibling(ast, input)
359
+ orig_input = original_input_literal
360
+ check = literal(:check)
361
+ root = literal(:root)
362
+ parent = literal(:parent)
363
+ doc_node = literal(:doc_node)
364
+
365
+ orig_input.is_a?(XML::Node)
366
+ .if_true { root.assign(orig_input.parent) }
367
+ .else { root.assign(orig_input) }
368
+ .followed_by(check.assign(self.false))
369
+ .followed_by do
370
+ input.is_a?(XML::Node).and(input.parent)
371
+ .if_true { parent.assign(input.parent) }
372
+ .else { parent.assign(self.nil) }
373
+ end
374
+ .followed_by do
375
+ document_or_node(root).if_true do
376
+ root.each_node.add_block(doc_node) do
377
+ doc_node.eq(input)
378
+ .if_true { self.break }
379
+ .followed_by do
380
+ doc_node.parent.eq(parent).if_true do
381
+ process(ast, doc_node).if_true { yield doc_node }
382
+ end
383
+ end
384
+ end
385
+ end
386
+ end
387
+ end
388
+
389
+ # @param [AST::Node] ast
390
+ # @param [Oga::Ruby::Node] input
391
+ # @return [Oga::Ruby::Node]
392
+ def on_predicate(ast, input, &block)
393
+ test, predicate, following = *ast
394
+
395
+ index_var = unique_literal(:index)
396
+
397
+ if number?(predicate)
398
+ method = :on_predicate_index
399
+ elsif has_call_node?(predicate, 'last')
400
+ method = :on_predicate_temporary
401
+ else
402
+ method = :on_predicate_direct
403
+ end
404
+
405
+ @predicate_indexes << index_var
406
+
407
+ ast = index_var.assign(literal(1)).followed_by do
408
+ send(method, input, test, predicate) do |matched|
409
+ process_following_or_yield(following, matched, &block)
410
+ end
411
+ end
412
+
413
+ @predicate_indexes.pop
414
+
415
+ ast
416
+ end
417
+
418
+ # Processes a predicate that requires a temporary NodeSet.
419
+ #
420
+ # @param [Oga::Ruby::Node] input
421
+ # @param [AST::Node] test
422
+ # @param [AST::Node] predicate
423
+ # @return [Oga::Ruby::Node]
424
+ def on_predicate_temporary(input, test, predicate)
425
+ temp_set = unique_literal(:temp_set)
426
+ pred_node = unique_literal(:pred_node)
427
+ pred_var = unique_literal(:pred_var)
428
+ conversion = literal(Conversion)
429
+
430
+ index_var = predicate_index
431
+ index_step = literal(1)
432
+
433
+ @predicate_nodesets << temp_set
434
+
435
+ ast = temp_set.assign(literal(XML::NodeSet).new)
436
+ .followed_by do
437
+ process(test, input) { |node| temp_set << node }
438
+ end
439
+ .followed_by do
440
+ temp_set.each.add_block(pred_node) do
441
+ pred_ast = process(predicate, pred_node)
442
+
443
+ pred_var.assign(pred_ast)
444
+ .followed_by do
445
+ pred_var.is_a?(Numeric).if_true do
446
+ pred_var.assign(pred_var.to_i.eq(index_var))
447
+ end
448
+ end
449
+ .followed_by do
450
+ conversion.to_boolean(pred_var).if_true { yield pred_node }
451
+ end
452
+ .followed_by do
453
+ index_var.assign(index_var + index_step)
454
+ end
455
+ end
456
+ end
457
+
458
+ @predicate_nodesets.pop
459
+
460
+ ast
461
+ end
462
+
463
+ # Processes a predicate that doesn't require temporary NodeSet.
464
+ #
465
+ # @param [Oga::Ruby::Node] input
466
+ # @param [AST::Node] test
467
+ # @param [AST::Node] predicate
468
+ # @return [Oga::Ruby::Node]
469
+ def on_predicate_direct(input, test, predicate)
470
+ pred_var = unique_literal(:pred_var)
471
+ index_var = predicate_index
472
+ index_step = literal(1)
473
+ conversion = literal(Conversion)
474
+
475
+ process(test, input) do |matched_test_node|
476
+ if return_nodeset?(predicate)
477
+ pred_ast = catch_message(:predicate_matched) do
478
+ process(predicate, matched_test_node) do
479
+ throw_message(:predicate_matched, self.true)
480
+ end
481
+ end
482
+ else
483
+ pred_ast = process(predicate, matched_test_node)
484
+ end
485
+
486
+ pred_var.assign(pred_ast)
487
+ .followed_by do
488
+ pred_var.is_a?(Numeric).if_true do
489
+ pred_var.assign(pred_var.to_i.eq(index_var))
490
+ end
491
+ end
492
+ .followed_by do
493
+ conversion.to_boolean(pred_var).if_true do
494
+ yield matched_test_node
495
+ end
496
+ end
497
+ .followed_by do
498
+ index_var.assign(index_var + index_step)
499
+ end
500
+ end
501
+ end
502
+
503
+ # Processes a predicate that uses a literal index.
504
+ #
505
+ # @param [Oga::Ruby::Node] input
506
+ # @param [AST::Node] test
507
+ # @param [AST::Node] predicate
508
+ # @return [Oga::Ruby::Node]
509
+ def on_predicate_index(input, test, predicate)
510
+ index_var = predicate_index
511
+ index_step = literal(1)
512
+
513
+ index = process(predicate, input).to_i
514
+
515
+ process(test, input) do |matched_test_node|
516
+ index_var.eq(index)
517
+ .if_true do
518
+ yield matched_test_node
519
+ end
520
+ .followed_by do
521
+ index_var.assign(index_var + index_step)
522
+ end
523
+ end
524
+ end
525
+
526
+ # @param [AST::Node] ast
527
+ # @param [Oga::Ruby::Node] input
528
+ # @return [Oga::Ruby::Node]
529
+ def on_test(ast, input)
530
+ condition = element_or_attribute(input)
531
+ name_match = match_name_and_namespace(ast, input)
532
+
533
+ name_match ? condition.and(name_match) : condition
534
+ end
535
+
536
+ # Processes the `=` operator.
537
+ #
538
+ # @see [#operator]
539
+ def on_eq(ast, input, &block)
540
+ conv = literal(Conversion)
541
+
542
+ operator(ast, input) do |left, right|
543
+ mass_assign([left, right], conv.to_compatible_types(left, right))
544
+ .followed_by do
545
+ operation = left.eq(right)
546
+
547
+ block ? operation.if_true(&block) : operation
548
+ end
549
+ end
550
+ end
551
+
552
+ # Processes the `!=` operator.
553
+ #
554
+ # @see [#operator]
555
+ def on_neq(ast, input, &block)
556
+ conv = literal(Conversion)
557
+
558
+ operator(ast, input) do |left, right|
559
+ mass_assign([left, right], conv.to_compatible_types(left, right))
560
+ .followed_by do
561
+ operation = left != right
562
+
563
+ block ? operation.if_true(&block) : operation
564
+ end
565
+ end
566
+ end
567
+
568
+ OPERATORS.each do |callback, (conv_method, ruby_method)|
569
+ define_method(callback) do |ast, input, &block|
570
+ conversion = literal(XPath::Conversion)
571
+
572
+ operator(ast, input) do |left, right|
573
+ lval = conversion.__send__(conv_method, left)
574
+ rval = conversion.__send__(conv_method, right)
575
+ operation = lval.__send__(ruby_method, rval)
576
+
577
+ block ? conversion.to_boolean(operation).if_true(&block) : operation
578
+ end
579
+ end
580
+ end
581
+
582
+ # Processes the `|` operator.
583
+ #
584
+ # @see [#operator]
585
+ def on_pipe(ast, input, &block)
586
+ left, right = *ast
587
+
588
+ union = unique_literal(:union)
589
+ conversion = literal(Conversion)
590
+
591
+ union.assign(literal(XML::NodeSet).new)
592
+ .followed_by(process(left, input) { |node| union << node })
593
+ .followed_by(process(right, input) { |node| union << node })
594
+ .followed_by do
595
+ # block present means we're in a predicate
596
+ block ? conversion.to_boolean(union).if_true(&block) : union
597
+ end
598
+ end
599
+
600
+ # @param [AST::Node] ast
601
+ # @return [Oga::Ruby::Node]
602
+ def on_string(ast, *)
603
+ string(ast.children[0])
604
+ end
605
+
606
+ # @param [AST::Node] ast
607
+ # @return [Oga::Ruby::Node]
608
+ def on_int(ast, *)
609
+ literal(ast.children[0].to_f.to_s)
610
+ end
611
+
612
+ # @param [AST::Node] ast
613
+ # @return [Oga::Ruby::Node]
614
+ def on_float(ast, *)
615
+ literal(ast.children[0].to_s)
616
+ end
617
+
618
+ # @param [AST::Node] ast
619
+ # @return [Oga::Ruby::Node]
620
+ def on_var(ast, *)
621
+ name = ast.children[0]
622
+
623
+ variables_literal.and(variables_literal[string(name)])
624
+ .or(send_message(:raise, string("Undefined XPath variable: #{name}")))
625
+ end
626
+
627
+ # Delegates function calls to specific handlers.
628
+ #
629
+ # @param [AST::Node] ast
630
+ # @param [Oga::Ruby::Node] input
631
+ # @return [Oga::Ruby::Node]
632
+ def on_call(ast, input, &block)
633
+ name, *args = *ast
634
+
635
+ handler = name.gsub('-', '_')
636
+
637
+ send(:"on_call_#{handler}", input, *args, &block)
638
+ end
639
+
640
+ # @return [Oga::Ruby::Node]
641
+ def on_call_true(*)
642
+ block_given? ? yield : self.true
643
+ end
644
+
645
+ # @return [Oga::Ruby::Node]
646
+ def on_call_false(*)
647
+ self.false
648
+ end
649
+
650
+ # @param [Oga::Ruby::Node] input
651
+ # @param [AST::Node] arg
652
+ # @return [Oga::Ruby::Node]
653
+ def on_call_boolean(input, arg)
654
+ arg_ast = try_match_first_node(arg, input)
655
+ call_arg = unique_literal(:call_arg)
656
+ conversion = literal(Conversion)
657
+
658
+ call_arg.assign(arg_ast).followed_by do
659
+ converted = conversion.to_boolean(call_arg)
660
+
661
+ block_given? ? converted.if_true { yield } : converted
662
+ end
663
+ end
664
+
665
+ # @param [Oga::Ruby::Node] input
666
+ # @param [AST::Node] arg
667
+ # @return [Oga::Ruby::Node]
668
+ def on_call_ceiling(input, arg)
669
+ arg_ast = try_match_first_node(arg, input)
670
+ call_arg = unique_literal(:call_arg)
671
+ conversion = literal(Conversion)
672
+
673
+ call_arg.assign(arg_ast)
674
+ .followed_by do
675
+ call_arg.assign(conversion.to_float(call_arg))
676
+ end
677
+ .followed_by do
678
+ call_arg.nan?
679
+ .if_true { call_arg }
680
+ .else { block_given? ? yield : call_arg.ceil.to_f }
681
+ end
682
+ end
683
+
684
+ # @param [Oga::Ruby::Node] input
685
+ # @param [AST::Node] arg
686
+ # @return [Oga::Ruby::Node]
687
+ def on_call_floor(input, arg)
688
+ arg_ast = try_match_first_node(arg, input)
689
+ call_arg = unique_literal(:call_arg)
690
+ conversion = literal(Conversion)
691
+
692
+ call_arg.assign(arg_ast)
693
+ .followed_by do
694
+ call_arg.assign(conversion.to_float(call_arg))
695
+ end
696
+ .followed_by do
697
+ call_arg.nan?
698
+ .if_true { call_arg }
699
+ .else { block_given? ? yield : call_arg.floor.to_f }
700
+ end
701
+ end
702
+
703
+ # @param [Oga::Ruby::Node] input
704
+ # @param [AST::Node] arg
705
+ # @return [Oga::Ruby::Node]
706
+ def on_call_round(input, arg)
707
+ arg_ast = try_match_first_node(arg, input)
708
+ call_arg = unique_literal(:call_arg)
709
+ conversion = literal(Conversion)
710
+
711
+ call_arg.assign(arg_ast)
712
+ .followed_by do
713
+ call_arg.assign(conversion.to_float(call_arg))
714
+ end
715
+ .followed_by do
716
+ call_arg.nan?
717
+ .if_true { call_arg }
718
+ .else { block_given? ? yield : call_arg.round.to_f }
719
+ end
720
+ end
721
+
722
+ # @param [Oga::Ruby::Node] input
723
+ # @param [Array<AST::Node>] args
724
+ # @return [Oga::Ruby::Node]
725
+ def on_call_concat(input, *args)
726
+ conversion = literal(Conversion)
727
+ assigns = []
728
+ conversions = []
729
+
730
+ args.each do |arg|
731
+ arg_var = unique_literal(:concat_arg)
732
+ arg_ast = try_match_first_node(arg, input)
733
+
734
+ assigns << arg_var.assign(arg_ast)
735
+ conversions << conversion.to_string(arg_var)
736
+ end
737
+
738
+ concatted = assigns.inject(:followed_by)
739
+ .followed_by(conversions.inject(:+))
740
+
741
+ block_given? ? concatted.empty?.if_false { yield } : concatted
742
+ end
743
+
744
+ # @param [Oga::Ruby::Node] input
745
+ # @param [AST::Node] haystack
746
+ # @param [AST::Node] needle
747
+ # @return [Oga::Ruby::Node]
748
+ def on_call_contains(input, haystack, needle)
749
+ haystack_lit = unique_literal(:haystack)
750
+ needle_lit = unique_literal(:needle)
751
+ conversion = literal(Conversion)
752
+
753
+ haystack_lit.assign(try_match_first_node(haystack, input))
754
+ .followed_by do
755
+ needle_lit.assign(try_match_first_node(needle, input))
756
+ end
757
+ .followed_by do
758
+ converted = conversion.to_string(haystack_lit)
759
+ .include?(conversion.to_string(needle_lit))
760
+
761
+ block_given? ? converted.if_true { yield } : converted
762
+ end
763
+ end
764
+
765
+ # @param [Oga::Ruby::Node] input
766
+ # @param [AST::Node] arg
767
+ # @return [Oga::Ruby::Node]
768
+ def on_call_count(input, arg)
769
+ count = unique_literal(:count)
770
+
771
+ unless return_nodeset?(arg)
772
+ raise TypeError, 'count() can only operate on NodeSet instances'
773
+ end
774
+
775
+ count.assign(literal(0.0))
776
+ .followed_by do
777
+ process(arg, input) { count.assign(count + literal(1)) }
778
+ end
779
+ .followed_by do
780
+ block_given? ? count.zero?.if_false { yield } : count
781
+ end
782
+ end
783
+
784
+ # Processes the `id()` function call.
785
+ #
786
+ # The XPath specification states that this function's behaviour should be
787
+ # controlled by a DTD. If a DTD were to specify that the ID attribute for
788
+ # a certain element would be "foo" then this function should use said
789
+ # attribute.
790
+ #
791
+ # Oga does not support DTD parsing/evaluation and as such always uses the
792
+ # "id" attribute.
793
+ #
794
+ # This function searches the entire document for a matching node,
795
+ # regardless of the current position.
796
+ #
797
+ # @param [Oga::Ruby::Node] input
798
+ # @param [AST::Node] arg
799
+ # @return [Oga::Ruby::Node]
800
+ def on_call_id(input, arg)
801
+ orig_input = original_input_literal
802
+ node = unique_literal(:node)
803
+ ids_var = unique_literal('ids')
804
+ matched = unique_literal('id_matched')
805
+ id_str_var = unique_literal('id_string')
806
+ attr_var = unique_literal('attr')
807
+
808
+ matched.assign(literal(XML::NodeSet).new)
809
+ .followed_by do
810
+ # When using some sort of path we'll want the text of all matched
811
+ # nodes.
812
+ if return_nodeset?(arg)
813
+ ids_var.assign(literal(:[])).followed_by do
814
+ process(arg, input) { |element| ids_var << element.text }
815
+ end
816
+
817
+ # For everything else we'll cast the value to a string and split it
818
+ # on every space.
819
+ else
820
+ conversion = literal(Conversion).to_string(ids_var)
821
+ .split(string(' '))
822
+
823
+ ids_var.assign(process(arg, input))
824
+ .followed_by(ids_var.assign(conversion))
825
+ end
826
+ end
827
+ .followed_by do
828
+ id_str_var.assign(string('id'))
829
+ end
830
+ .followed_by do
831
+ orig_input.each_node.add_block(node) do
832
+ node.is_a?(XML::Element).if_true do
833
+ attr_var.assign(node.attribute(id_str_var)).followed_by do
834
+ attr_var.and(ids_var.include?(attr_var.value))
835
+ .if_true { block_given? ? yield : matched << node }
836
+ end
837
+ end
838
+ end
839
+ end
840
+ .followed_by(matched)
841
+ end
842
+
843
+ # @param [Oga::Ruby::Node] input
844
+ # @param [AST::Node] arg
845
+ # @return [Oga::Ruby::Node]
846
+ def on_call_lang(input, arg)
847
+ lang_var = unique_literal('lang')
848
+ node = unique_literal('node')
849
+ found = unique_literal('found')
850
+ xml_lang = unique_literal('xml_lang')
851
+ matched = unique_literal('matched')
852
+
853
+ conversion = literal(Conversion)
854
+
855
+ ast = lang_var.assign(try_match_first_node(arg, input))
856
+ .followed_by do
857
+ lang_var.assign(conversion.to_string(lang_var))
858
+ end
859
+ .followed_by do
860
+ matched.assign(self.false)
861
+ end
862
+ .followed_by do
863
+ node.assign(input)
864
+ end
865
+ .followed_by do
866
+ xml_lang.assign(string('xml:lang'))
867
+ end
868
+ .followed_by do
869
+ node.respond_to?(symbol(:attribute)).while_true do
870
+ found.assign(node.get(xml_lang))
871
+ .followed_by do
872
+ found.if_true do
873
+ found.eq(lang_var)
874
+ .if_true do
875
+ if block_given?
876
+ yield
877
+ else
878
+ matched.assign(self.true).followed_by(self.break)
879
+ end
880
+ end
881
+ .else { self.break }
882
+ end
883
+ end
884
+ .followed_by(node.assign(node.parent))
885
+ end
886
+ end
887
+
888
+ block_given? ? ast : ast.followed_by(matched)
889
+ end
890
+
891
+ # @param [Oga::Ruby::Node] input
892
+ # @param [AST::Node] arg
893
+ # @return [Oga::Ruby::Node]
894
+ def on_call_local_name(input, arg = nil)
895
+ argument_or_first_node(input, arg) do |arg_var|
896
+ arg_var
897
+ .if_true do
898
+ ensure_element_or_attribute(arg_var)
899
+ .followed_by { block_given? ? yield : arg_var.name }
900
+ end
901
+ .else { string('') }
902
+ end
903
+ end
904
+
905
+ # @param [Oga::Ruby::Node] input
906
+ # @param [AST::Node] arg
907
+ # @return [Oga::Ruby::Node]
908
+ def on_call_name(input, arg = nil)
909
+ argument_or_first_node(input, arg) do |arg_var|
910
+ arg_var
911
+ .if_true do
912
+ ensure_element_or_attribute(arg_var)
913
+ .followed_by { block_given? ? yield : arg_var.expanded_name }
914
+ end
915
+ .else { string('') }
916
+ end
917
+ end
918
+
919
+ # @param [Oga::Ruby::Node] input
920
+ # @param [AST::Node] arg
921
+ # @return [Oga::Ruby::Node]
922
+ def on_call_namespace_uri(input, arg = nil)
923
+ default = string('')
924
+
925
+ argument_or_first_node(input, arg) do |arg_var|
926
+ arg_var
927
+ .if_true do
928
+ ensure_element_or_attribute(arg_var).followed_by do
929
+ arg_var.namespace
930
+ .if_true { block_given? ? yield : arg_var.namespace.uri }
931
+ .else { default } # no yield so predicates aren't matched
932
+ end
933
+ end
934
+ .else { default }
935
+ end
936
+ end
937
+
938
+ # @param [Oga::Ruby::Node] input
939
+ # @param [AST::Node] arg
940
+ # @return [Oga::Ruby::Node]
941
+ def on_call_normalize_space(input, arg = nil)
942
+ conversion = literal(Conversion)
943
+ norm_var = unique_literal(:normalized)
944
+
945
+ find = literal('/\s+/')
946
+ replace = string(' ')
947
+
948
+ argument_or_first_node(input, arg) do |arg_var|
949
+ norm_var
950
+ .assign(conversion.to_string(arg_var).strip.gsub(find, replace))
951
+ .followed_by do
952
+ norm_var.empty?
953
+ .if_true { string('') }
954
+ .else { block_given? ? yield : norm_var }
955
+ end
956
+ end
957
+ end
958
+
959
+ # @param [Oga::Ruby::Node] input
960
+ # @param [AST::Node] arg
961
+ # @return [Oga::Ruby::Node]
962
+ def on_call_not(input, arg)
963
+ arg_ast = try_match_first_node(arg, input)
964
+ call_arg = unique_literal(:call_arg)
965
+ conversion = literal(Conversion)
966
+
967
+ call_arg.assign(arg_ast).followed_by do
968
+ converted = conversion.to_boolean(call_arg).not
969
+
970
+ block_given? ? converted.if_true { yield } : converted
971
+ end
972
+ end
973
+
974
+ # @param [Oga::Ruby::Node] input
975
+ # @param [AST::Node] arg
976
+ # @return [Oga::Ruby::Node]
977
+ def on_call_number(input, arg = nil)
978
+ convert_var = unique_literal(:convert)
979
+ conversion = literal(Conversion)
980
+
981
+ argument_or_first_node(input, arg) do |arg_var|
982
+ convert_var.assign(conversion.to_float(arg_var)).followed_by do
983
+ if block_given?
984
+ convert_var.zero?.if_false { yield }
985
+ else
986
+ convert_var
987
+ end
988
+ end
989
+ end
990
+ end
991
+
992
+ # @param [Oga::Ruby::Node] input
993
+ # @param [AST::Node] haystack
994
+ # @param [AST::Node] needle
995
+ # @return [Oga::Ruby::Node]
996
+ def on_call_starts_with(input, haystack, needle)
997
+ haystack_var = unique_literal(:haystack)
998
+ needle_var = unique_literal(:needle)
999
+ conversion = literal(Conversion)
1000
+
1001
+ haystack_var.assign(try_match_first_node(haystack, input))
1002
+ .followed_by do
1003
+ needle_var.assign(try_match_first_node(needle, input))
1004
+ end
1005
+ .followed_by do
1006
+ haystack_var.assign(conversion.to_string(haystack_var))
1007
+ .followed_by do
1008
+ needle_var.assign(conversion.to_string(needle_var))
1009
+ end
1010
+ .followed_by do
1011
+ equal = needle_var.empty?
1012
+ .or(haystack_var.start_with?(needle_var))
1013
+
1014
+ block_given? ? equal.if_true { yield } : equal
1015
+ end
1016
+ end
1017
+ end
1018
+
1019
+ # @param [Oga::Ruby::Node] input
1020
+ # @param [AST::Node] arg
1021
+ # @return [Oga::Ruby::Node]
1022
+ def on_call_string_length(input, arg = nil)
1023
+ convert_var = unique_literal(:convert)
1024
+ conversion = literal(Conversion)
1025
+
1026
+ argument_or_first_node(input, arg) do |arg_var|
1027
+ convert_var.assign(conversion.to_string(arg_var).length)
1028
+ .followed_by do
1029
+ if block_given?
1030
+ convert_var.zero?.if_false { yield }
1031
+ else
1032
+ convert_var.to_f
1033
+ end
1034
+ end
1035
+ end
1036
+ end
1037
+
1038
+ # @param [Oga::Ruby::Node] input
1039
+ # @param [AST::Node] arg
1040
+ # @return [Oga::Ruby::Node]
1041
+ def on_call_string(input, arg = nil)
1042
+ convert_var = unique_literal(:convert)
1043
+ conversion = literal(Conversion)
1044
+
1045
+ argument_or_first_node(input, arg) do |arg_var|
1046
+ convert_var.assign(conversion.to_string(arg_var))
1047
+ .followed_by do
1048
+ if block_given?
1049
+ convert_var.empty?.if_false { yield }
1050
+ else
1051
+ convert_var
1052
+ end
1053
+ end
1054
+ end
1055
+ end
1056
+
1057
+ # @param [Oga::Ruby::Node] input
1058
+ # @param [AST::Node] haystack
1059
+ # @param [AST::Node] needle
1060
+ # @return [Oga::Ruby::Node]
1061
+ def on_call_substring_before(input, haystack, needle)
1062
+ haystack_var = unique_literal(:haystack)
1063
+ needle_var = unique_literal(:needle)
1064
+ conversion = literal(Conversion)
1065
+
1066
+ before = unique_literal(:before)
1067
+ sep = unique_literal(:sep)
1068
+ after = unique_literal(:after)
1069
+
1070
+ haystack_var.assign(try_match_first_node(haystack, input))
1071
+ .followed_by do
1072
+ needle_var.assign(try_match_first_node(needle, input))
1073
+ end
1074
+ .followed_by do
1075
+ converted = conversion.to_string(haystack_var)
1076
+ .partition(conversion.to_string(needle_var))
1077
+
1078
+ mass_assign([before, sep, after], converted).followed_by do
1079
+ sep.empty?
1080
+ .if_true { sep }
1081
+ .else { block_given? ? yield : before }
1082
+ end
1083
+ end
1084
+ end
1085
+
1086
+ # @param [Oga::Ruby::Node] input
1087
+ # @param [AST::Node] haystack
1088
+ # @param [AST::Node] needle
1089
+ # @return [Oga::Ruby::Node]
1090
+ def on_call_substring_after(input, haystack, needle)
1091
+ haystack_var = unique_literal(:haystack)
1092
+ needle_var = unique_literal(:needle)
1093
+ conversion = literal(Conversion)
1094
+
1095
+ before = unique_literal(:before)
1096
+ sep = unique_literal(:sep)
1097
+ after = unique_literal(:after)
1098
+
1099
+ haystack_var.assign(try_match_first_node(haystack, input))
1100
+ .followed_by do
1101
+ needle_var.assign(try_match_first_node(needle, input))
1102
+ end
1103
+ .followed_by do
1104
+ converted = conversion.to_string(haystack_var)
1105
+ .partition(conversion.to_string(needle_var))
1106
+
1107
+ mass_assign([before, sep, after], converted).followed_by do
1108
+ sep.empty?
1109
+ .if_true { sep }
1110
+ .else { block_given? ? yield : after }
1111
+ end
1112
+ end
1113
+ end
1114
+
1115
+ # @param [Oga::Ruby::Node] input
1116
+ # @param [AST::Node] haystack
1117
+ # @param [AST::Node] start
1118
+ # @param [AST::Node] length
1119
+ # @return [Oga::Ruby::Node]
1120
+ def on_call_substring(input, haystack, start, length = nil)
1121
+ haystack_var = unique_literal(:haystack)
1122
+ start_var = unique_literal(:start)
1123
+ stop_var = unique_literal(:stop)
1124
+ length_var = unique_literal(:length)
1125
+ conversion = literal(Conversion)
1126
+
1127
+ haystack_var.assign(try_match_first_node(haystack, input))
1128
+ .followed_by do
1129
+ start_var.assign(try_match_first_node(start, input))
1130
+ .followed_by do
1131
+ start_var.assign(start_var - literal(1))
1132
+ end
1133
+ end
1134
+ .followed_by do
1135
+ if length
1136
+ length_var.assign(try_match_first_node(length, input))
1137
+ .followed_by do
1138
+ length_int = conversion.to_float(length_var)
1139
+ .to_i - literal(1)
1140
+
1141
+ stop_var.assign(start_var + length_int)
1142
+ end
1143
+ else
1144
+ stop_var.assign(literal(-1))
1145
+ end
1146
+ end
1147
+ .followed_by do
1148
+ substring = conversion
1149
+ .to_string(haystack_var)[range(start_var, stop_var)]
1150
+
1151
+ block_given? ? substring.empty?.if_false { yield } : substring
1152
+ end
1153
+ end
1154
+
1155
+ # @param [Oga::Ruby::Node] input
1156
+ # @param [AST::Node] arg
1157
+ # @return [Oga::Ruby::Node]
1158
+ def on_call_sum(input, arg)
1159
+ unless return_nodeset?(arg)
1160
+ raise TypeError, 'sum() can only operate on a path, axis or predicate'
1161
+ end
1162
+
1163
+ sum_var = unique_literal(:sum)
1164
+ conversion = literal(Conversion)
1165
+
1166
+ sum_var.assign(literal(0.0))
1167
+ .followed_by do
1168
+ process(arg, input) do |matched_node|
1169
+ sum_var.assign(sum_var + conversion.to_float(matched_node.text))
1170
+ end
1171
+ end
1172
+ .followed_by do
1173
+ block_given? ? sum_var.zero?.if_false { yield } : sum_var
1174
+ end
1175
+ end
1176
+
1177
+ # @param [Oga::Ruby::Node] input
1178
+ # @param [AST::Node] source
1179
+ # @param [AST::Node] find
1180
+ # @param [AST::Node] replace
1181
+ # @return [Oga::Ruby::Node]
1182
+ def on_call_translate(input, source, find, replace)
1183
+ source_var = unique_literal(:source)
1184
+ find_var = unique_literal(:find)
1185
+ replace_var = unique_literal(:replace)
1186
+ replaced_var = unique_literal(:replaced)
1187
+ conversion = literal(Conversion)
1188
+
1189
+ char = unique_literal(:char)
1190
+ index = unique_literal(:index)
1191
+
1192
+ source_var.assign(try_match_first_node(source, input))
1193
+ .followed_by do
1194
+ replaced_var.assign(conversion.to_string(source_var))
1195
+ end
1196
+ .followed_by do
1197
+ find_var.assign(try_match_first_node(find, input))
1198
+ end
1199
+ .followed_by do
1200
+ find_var.assign(conversion.to_string(find_var).chars.to_array)
1201
+ end
1202
+ .followed_by do
1203
+ replace_var.assign(try_match_first_node(replace, input))
1204
+ end
1205
+ .followed_by do
1206
+ replace_var.assign(conversion.to_string(replace_var).chars.to_array)
1207
+ end
1208
+ .followed_by do
1209
+ find_var.each_with_index.add_block(char, index) do
1210
+ replace_with = replace_var[index]
1211
+ .if_true { replace_var[index] }
1212
+ .else { string('') }
1213
+
1214
+ replaced_var.assign(replaced_var.gsub(char, replace_with))
1215
+ end
1216
+ end
1217
+ .followed_by do
1218
+ replaced_var
1219
+ end
1220
+ end
1221
+
1222
+ # @return [Oga::Ruby::Node]
1223
+ def on_call_last(*)
1224
+ set = predicate_nodeset
1225
+
1226
+ unless set
1227
+ raise 'last() can only be used in a predicate'
1228
+ end
1229
+
1230
+ set.length.to_f
1231
+ end
1232
+
1233
+ # @return [Oga::Ruby::Node]
1234
+ def on_call_position(*)
1235
+ index = predicate_index
1236
+
1237
+ unless index
1238
+ raise 'position() can only be used in a predicate'
1239
+ end
1240
+
1241
+ index.to_f
1242
+ end
1243
+
1244
+ # Delegates type tests to specific handlers.
1245
+ #
1246
+ # @param [AST::Node] ast
1247
+ # @param [Oga::Ruby::Node] input
1248
+ # @return [Oga::Ruby::Node]
1249
+ def on_type_test(ast, input, &block)
1250
+ name, following = *ast
1251
+
1252
+ handler = name.gsub('-', '_')
1253
+
1254
+ send(:"on_type_test_#{handler}", input) do |matched|
1255
+ process_following_or_yield(following, matched, &block)
1256
+ end
1257
+ end
1258
+
1259
+ # @param [Oga::Ruby::Node] input
1260
+ # @return [Oga::Ruby::Node]
1261
+ def on_type_test_comment(input)
1262
+ input.is_a?(XML::Comment)
1263
+ end
1264
+
1265
+ # @param [Oga::Ruby::Node] input
1266
+ # @return [Oga::Ruby::Node]
1267
+ def on_type_test_text(input)
1268
+ input.is_a?(XML::Text)
1269
+ end
1270
+
1271
+ # @param [Oga::Ruby::Node] input
1272
+ # @return [Oga::Ruby::Node]
1273
+ def on_type_test_processing_instruction(input)
1274
+ input.is_a?(XML::ProcessingInstruction)
1275
+ end
1276
+
1277
+ # @param [Oga::Ruby::Node] input
1278
+ # @return [Oga::Ruby::Node]
1279
+ def on_type_test_node(input)
1280
+ document_or_node(input).or(input.is_a?(XML::Attribute))
1281
+ end
1282
+
1283
+ # @param [#to_s] value
1284
+ # @return [Oga::Ruby::Node]
1285
+ def literal(value)
1286
+ Ruby::Node.new(:lit, [value.to_s])
1287
+ end
1288
+
1289
+ # @param [Oga::Ruby::Node] start
1290
+ # @param [Oga::Ruby::Node] stop
1291
+ # @return [Oga::Ruby::Node]
1292
+ def range(start, stop)
1293
+ Ruby::Node.new(:range, [start, stop])
1294
+ end
1295
+
1296
+ # @param [String] name
1297
+ # @return [Oga::Ruby::Node]
1298
+ def unique_literal(name)
1299
+ new_id = @literal_id += 1
1300
+
1301
+ literal("#{name}#{new_id}")
1302
+ end
1303
+
1304
+ # @param [#to_s] value
1305
+ # @return [Oga::Ruby::Node]
1306
+ def string(value)
1307
+ Ruby::Node.new(:string, [value.to_s])
1308
+ end
1309
+
1310
+ # @param [String] value
1311
+ # @return [Oga::Ruby::Node]
1312
+ def symbol(value)
1313
+ Ruby::Node.new(:symbol, [value.to_sym])
1314
+ end
1315
+
1316
+ # @param [String] name
1317
+ # @param [Array] args
1318
+ # @return [Oga::Ruby::Node]
1319
+ def send_message(name, *args)
1320
+ Ruby::Node.new(:send, [nil, name.to_s, *args])
1321
+ end
1322
+
1323
+ # @param [Class] klass
1324
+ # @param [String] message
1325
+ # @return [Oga::Ruby::Node]
1326
+ def raise_message(klass, message)
1327
+ send_message(:raise, literal(klass), string(message))
1328
+ end
1329
+
1330
+ # @return [Oga::Ruby::Node]
1331
+ def nil
1332
+ @nil ||= literal(:nil)
1333
+ end
1334
+
1335
+ # @return [Oga::Ruby::Node]
1336
+ def true
1337
+ @true ||= literal(:true)
1338
+ end
1339
+
1340
+ # @return [Oga::Ruby::Node]
1341
+ def false
1342
+ @false ||= literal(:false)
1343
+ end
1344
+
1345
+ # @param [Oga::Ruby::Node] node
1346
+ # @return [Oga::Ruby::Node]
1347
+ def element_or_attribute(node)
1348
+ node.is_a?(XML::Element).or(node.is_a?(XML::Attribute))
1349
+ end
1350
+
1351
+ # @param [Oga::Ruby::Node] node
1352
+ # @return [Oga::Ruby::Node]
1353
+ def attribute_or_node(node)
1354
+ node.is_a?(XML::Attribute).or(node.is_a?(XML::Node))
1355
+ end
1356
+
1357
+ # @param [Oga::Ruby::Node] node
1358
+ # @return [Oga::Ruby::Node]
1359
+ def document_or_node(node)
1360
+ node.is_a?(XML::Document).or(node.is_a?(XML::Node))
1361
+ end
1362
+
1363
+ # @param [AST::Node] ast
1364
+ # @param [Oga::Ruby::Node] input
1365
+ # @return [Oga::Ruby::Node]
1366
+ def match_name_and_namespace(ast, input)
1367
+ ns, name = *ast
1368
+
1369
+ condition = nil
1370
+ name_str = string(name)
1371
+ zero = literal(0)
1372
+
1373
+ if name != STAR
1374
+ condition = input.name.eq(name_str)
1375
+ .or(input.name.casecmp(name_str).eq(zero))
1376
+ end
1377
+
1378
+ if ns and ns != STAR
1379
+ ns_match = input.namespace_name.eq(string(ns))
1380
+ condition = condition ? condition.and(ns_match) : ns_match
1381
+ end
1382
+
1383
+ condition
1384
+ end
1385
+
1386
+ # Returns an AST matching the first node of a node set.
1387
+ #
1388
+ # @param [Oga::Ruby::Node] ast
1389
+ # @param [Oga::Ruby::Node] input
1390
+ # @return [Oga::Ruby::Node]
1391
+ def match_first_node(ast, input)
1392
+ catch_message(:value) do
1393
+ process(ast, input) do |node|
1394
+ throw_message(:value, node)
1395
+ end
1396
+ end
1397
+ end
1398
+
1399
+ # Tries to match the first node in a set, otherwise processes it as usual.
1400
+ #
1401
+ # @see [#match_first_node]
1402
+ def try_match_first_node(ast, input, optimize_first = true)
1403
+ if return_nodeset?(ast) and optimize_first
1404
+ match_first_node(ast, input)
1405
+ else
1406
+ process(ast, input)
1407
+ end
1408
+ end
1409
+
1410
+ # @param [Oga::Ruby::Node] input
1411
+ # @return [Oga::Ruby::Node]
1412
+ def ensure_element_or_attribute(input)
1413
+ element_or_attribute(input).if_false do
1414
+ raise_message(TypeError, 'argument is not an Element or Attribute')
1415
+ end
1416
+ end
1417
+
1418
+ # @param [Oga::Ruby::Node] input
1419
+ # @param [AST::Ruby] arg
1420
+ # @return [Oga::Ruby::Node]
1421
+ def argument_or_first_node(input, arg = nil)
1422
+ arg_ast = arg ? try_match_first_node(arg, input) : input
1423
+ arg_var = unique_literal(:argument_or_first_node)
1424
+
1425
+ arg_var.assign(arg_ast).followed_by { yield arg_var }
1426
+ end
1427
+
1428
+ # Generates the code for an operator.
1429
+ #
1430
+ # The generated code is optimized so that expressions such as `a/b = c`
1431
+ # only match the first node in both arms instead of matching all available
1432
+ # nodes first. Because numeric operators only ever operates on the first
1433
+ # node in a set we can simply ditch the rest, possibly speeding things up
1434
+ # quite a bit. This only works if one of the arms is:
1435
+ #
1436
+ # * a path
1437
+ # * an axis
1438
+ # * a predicate
1439
+ #
1440
+ # Everything else is processed the usual (and possibly slower) way.
1441
+ #
1442
+ # @param [AST::Node] ast
1443
+ # @param [Oga::Ruby::Node] input
1444
+ # @param [TrueClass|FalseClass] optimize_first
1445
+ # @return [Oga::Ruby::Node]
1446
+ def operator(ast, input, optimize_first = true)
1447
+ left, right = *ast
1448
+
1449
+ left_var = unique_literal(:op_left)
1450
+ right_var = unique_literal(:op_right)
1451
+
1452
+ left_ast = try_match_first_node(left, input, optimize_first)
1453
+ right_ast = try_match_first_node(right, input, optimize_first)
1454
+
1455
+ left_var.assign(left_ast)
1456
+ .followed_by(right_var.assign(right_ast))
1457
+ .followed_by { yield left_var, right_var }
1458
+ end
1459
+
1460
+ # @return [Oga::Ruby::Node]
1461
+ def matched_literal
1462
+ literal(:matched)
1463
+ end
1464
+
1465
+ # @return [Oga::Ruby::Node]
1466
+ def original_input_literal
1467
+ literal(:original_input)
1468
+ end
1469
+
1470
+ # @return [Oga::Ruby::Node]
1471
+ def variables_literal
1472
+ literal(:variables)
1473
+ end
1474
+
1475
+ # @param [AST::Node] ast
1476
+ # @return [Oga::Ruby::Node]
1477
+ def to_int(ast)
1478
+ literal(ast.children[0].to_i.to_s)
1479
+ end
1480
+
1481
+ # @param [Array] vars The variables to assign.
1482
+ # @param [Oga::Ruby::Node] value
1483
+ # @return [Oga::Ruby::Node]
1484
+ def mass_assign(vars, value)
1485
+ Ruby::Node.new(:massign, [vars, value])
1486
+ end
1487
+
1488
+ # @param [AST::Node] ast
1489
+ # @return [TrueClass|FalseClass]
1490
+ def number?(ast)
1491
+ ast.type == :int || ast.type == :float
1492
+ end
1493
+
1494
+ # @param [AST::Node] ast
1495
+ # @return [TrueClass|FalseClass]
1496
+ def string?(ast)
1497
+ ast.type == :string
1498
+ end
1499
+
1500
+ # @param [Symbol] name
1501
+ # @return [Oga::Ruby::Node]
1502
+ def catch_message(name)
1503
+ send_message(:catch, symbol(name)).add_block do
1504
+ # Ensure that the "catch" only returns a value when "throw" is
1505
+ # actually invoked.
1506
+ yield.followed_by(self.nil)
1507
+ end
1508
+ end
1509
+
1510
+ # @param [Symbol] name
1511
+ # @param [Array] args
1512
+ # @return [Oga::Ruby::Node]
1513
+ def throw_message(name, *args)
1514
+ send_message(:throw, symbol(name), *args)
1515
+ end
1516
+
1517
+ # @return [Oga::Ruby::Node]
1518
+ def break
1519
+ send_message(:break)
1520
+ end
1521
+
1522
+ # @param [AST::Node] ast
1523
+ # @return [TrueClass|FalseClass]
1524
+ def return_nodeset?(ast)
1525
+ RETURN_NODESET.include?(ast.type)
1526
+ end
1527
+
1528
+ # @param [AST::Node] ast
1529
+ # @return [TrueClass|FalseClass]
1530
+ def has_call_node?(ast, name)
1531
+ visit = [ast]
1532
+
1533
+ until visit.empty?
1534
+ current = visit.pop
1535
+
1536
+ return true if current.type == :call && current.children[0] == name
1537
+
1538
+ current.children.each do |child|
1539
+ visit << child if child.is_a?(AST::Node)
1540
+ end
1541
+ end
1542
+
1543
+ false
1544
+ end
1545
+
1546
+ # @return [Oga::Ruby::Node]
1547
+ def predicate_index
1548
+ @predicate_indexes.last
1549
+ end
1550
+
1551
+ # @return [Oga::Ruby::Node]
1552
+ def predicate_nodeset
1553
+ @predicate_nodesets.last
1554
+ end
1555
+
1556
+ # @param [AST::Node] following
1557
+ # @param [Oga::Ruby::Node] matched
1558
+ # @return [Oga::Ruby::Node]
1559
+ def process_following_or_yield(following, matched, &block)
1560
+ if following
1561
+ process(following, matched, &block)
1562
+ else
1563
+ yield matched
1564
+ end
1565
+ end
1566
+ end # Compiler
1567
+ end # XPath
1568
+ end # Oga