taurus 0.1.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +7 -0
- data/.rspec +3 -0
- data/.rubocop.yml +8 -0
- data/CHANGELOG.md +518 -0
- data/CLAUDE.md +104 -0
- data/LICENSE.md +33 -0
- data/README.adoc +1529 -0
- data/Rakefile +7 -0
- data/TODO.impl/01-architecture.md +217 -0
- data/TODO.impl/02-ffi-declarations.md +236 -0
- data/TODO.impl/03-document-node-element-nodeset.md +382 -0
- data/TODO.impl/04-sax-parser.md +203 -0
- data/TODO.impl/05-serialize-c14n-memory-specs-css.md +276 -0
- data/benchmark/README.md +168 -0
- data/benchmark/taurus_vs_nokogiri.rb +105 -0
- data/docs/ARCHITECTURE.adoc +559 -0
- data/docs/BUILD.md +395 -0
- data/docs/ERROR_MESSAGES.md +458 -0
- data/docs/FFI_ARCHITECTURE.md +439 -0
- data/docs/FUTURE_VISION.md +303 -0
- data/docs/GITHUB_ACTIONS.md +293 -0
- data/docs/OPTIMIZATIONS_IMPLEMENTED.adoc +459 -0
- data/docs/PERFORMANCE.adoc +668 -0
- data/docs/PERFORMANCE.md +448 -0
- data/docs/RELEASE_NOTES_v1.0.0.md +515 -0
- data/docs/XPATH_SPEC_COMPLIANCE.md +298 -0
- data/docs/completion/taurus.bash +86 -0
- data/docs/completion/taurus.zsh +74 -0
- data/docs/man/taurus-format.1 +227 -0
- data/docs/man/taurus-parse.1 +178 -0
- data/docs/man/taurus-xpath.1 +312 -0
- data/docs/man/taurus.1 +160 -0
- data/docs/v0.9.0_PERFORMANCE_IMPROVEMENTS.md +217 -0
- data/docs/v0.9.0_RELEASE_SUMMARY.md +281 -0
- data/docs/v1.0.0_CONTINUATION_PLAN.md +172 -0
- data/docs/v1.0.0_CONTINUATION_PROMPT.md +382 -0
- data/docs/v1.0.0_SESSION_6_CONTINUATION.md +434 -0
- data/docs/v1.0.0_SESSION_6_PROMPT.md +231 -0
- data/docs/v1.0.0_STATUS_TRACKER.md +224 -0
- data/docs/v1.1.0_CONTINUATION_PLAN.md +299 -0
- data/docs/v1.1.0_FINAL_CONTINUATION_PLAN.md +201 -0
- data/docs/v1.1.0_SESSION_3_PROMPT.md +223 -0
- data/docs/v1.1.0_STATUS_TRACKER.md +355 -0
- data/docs/xml-performance.adoc +115 -0
- data/docs/xpath-performance.adoc +379 -0
- data/lib/taurus/version.rb +5 -0
- data/lib/taurus/xml/attr.rb +43 -0
- data/lib/taurus/xml/c14n.rb +23 -0
- data/lib/taurus/xml/cdata.rb +16 -0
- data/lib/taurus/xml/comment.rb +16 -0
- data/lib/taurus/xml/css_to_xpath.rb +177 -0
- data/lib/taurus/xml/doc_type.rb +54 -0
- data/lib/taurus/xml/document.rb +202 -0
- data/lib/taurus/xml/document_fragment.rb +42 -0
- data/lib/taurus/xml/element.rb +278 -0
- data/lib/taurus/xml/ffi.rb +420 -0
- data/lib/taurus/xml/namespace.rb +43 -0
- data/lib/taurus/xml/node.rb +221 -0
- data/lib/taurus/xml/node_set.rb +143 -0
- data/lib/taurus/xml/parse_options.rb +19 -0
- data/lib/taurus/xml/processing_instruction.rb +26 -0
- data/lib/taurus/xml/sax/document.rb +45 -0
- data/lib/taurus/xml/sax/parser.rb +148 -0
- data/lib/taurus/xml/sax.rb +12 -0
- data/lib/taurus/xml/searchable.rb +93 -0
- data/lib/taurus/xml/text.rb +16 -0
- data/lib/taurus/xml.rb +29 -0
- data/lib/taurus.rb +7 -0
- data/taurus.gemspec +42 -0
- metadata +157 -0
|
@@ -0,0 +1,420 @@
|
|
|
1
|
+
# frozen_string_literal: true
|
|
2
|
+
|
|
3
|
+
require "ffi"
|
|
4
|
+
|
|
5
|
+
module Taurus
|
|
6
|
+
module XML
|
|
7
|
+
module FFI
|
|
8
|
+
extend ::FFI::Library
|
|
9
|
+
|
|
10
|
+
ffi_lib [
|
|
11
|
+
ENV["TAURUS_LIB_PATH"],
|
|
12
|
+
File.expand_path("../../libtaurus.dylib", __dir__),
|
|
13
|
+
File.expand_path("../../libtaurus.so", __dir__),
|
|
14
|
+
"/usr/local/lib/libtaurus.dylib",
|
|
15
|
+
"/usr/local/lib/libtaurus.so",
|
|
16
|
+
"taurus",
|
|
17
|
+
].compact
|
|
18
|
+
|
|
19
|
+
typedef :pointer, :taurus_document
|
|
20
|
+
typedef :pointer, :taurus_element
|
|
21
|
+
typedef :pointer, :taurus_node_ref
|
|
22
|
+
typedef :pointer, :taurus_attribute
|
|
23
|
+
typedef :pointer, :taurus_doctype
|
|
24
|
+
typedef :pointer, :taurus_xpath_result
|
|
25
|
+
typedef :pointer, :taurus_xpath_var_set
|
|
26
|
+
typedef :pointer, :taurus_sax_parser
|
|
27
|
+
typedef :int, :taurus_status
|
|
28
|
+
|
|
29
|
+
class SAXHandler < ::FFI::Struct
|
|
30
|
+
layout \
|
|
31
|
+
:start_document, :pointer,
|
|
32
|
+
:end_document, :pointer,
|
|
33
|
+
:start_element, :pointer,
|
|
34
|
+
:end_element, :pointer,
|
|
35
|
+
:characters, :pointer,
|
|
36
|
+
:comment, :pointer,
|
|
37
|
+
:cdata, :pointer,
|
|
38
|
+
:processing_instruction, :pointer,
|
|
39
|
+
:start_prefix_mapping, :pointer,
|
|
40
|
+
:end_prefix_mapping, :pointer,
|
|
41
|
+
:error, :pointer
|
|
42
|
+
end
|
|
43
|
+
|
|
44
|
+
class SerializeOptions < ::FFI::Struct
|
|
45
|
+
layout \
|
|
46
|
+
:indent, :int,
|
|
47
|
+
:xml_declaration, :int,
|
|
48
|
+
:encoding, :pointer
|
|
49
|
+
end
|
|
50
|
+
|
|
51
|
+
attach_function :taurus_version, [], :string
|
|
52
|
+
attach_function :taurus_version_components, [:pointer, :pointer, :pointer], :void
|
|
53
|
+
|
|
54
|
+
attach_function :taurus_parse_string,
|
|
55
|
+
[:string, :size_t, :pointer], :taurus_document
|
|
56
|
+
attach_function :taurus_parse_string_inplace,
|
|
57
|
+
[:pointer, :size_t, :pointer], :taurus_document
|
|
58
|
+
attach_function :taurus_parse_string_with_encoding,
|
|
59
|
+
[:string, :size_t, :pointer], :taurus_document
|
|
60
|
+
attach_function :taurus_parse_file,
|
|
61
|
+
[:string, :pointer], :taurus_document
|
|
62
|
+
attach_function :taurus_load_file,
|
|
63
|
+
[:string, :pointer], :pointer
|
|
64
|
+
attach_function :taurus_document_free,
|
|
65
|
+
[:taurus_document], :void
|
|
66
|
+
attach_function :taurus_document_root,
|
|
67
|
+
[:taurus_document], :taurus_element
|
|
68
|
+
attach_function :taurus_document_encoding,
|
|
69
|
+
[:taurus_document], :string
|
|
70
|
+
attach_function :taurus_document_finalize_strings,
|
|
71
|
+
[:taurus_document], :int
|
|
72
|
+
attach_function :taurus_document_adopt_child,
|
|
73
|
+
[:taurus_document, :taurus_document], :void
|
|
74
|
+
attach_function :taurus_document_set_strict,
|
|
75
|
+
[:taurus_document, :int], :taurus_status
|
|
76
|
+
attach_function :taurus_document_get_strict,
|
|
77
|
+
[:taurus_document], :int
|
|
78
|
+
attach_function :taurus_document_freeze,
|
|
79
|
+
[:taurus_document], :taurus_status
|
|
80
|
+
attach_function :taurus_document_is_frozen,
|
|
81
|
+
[:taurus_document], :int
|
|
82
|
+
attach_function :taurus_set_strict_mode, [:int], :void
|
|
83
|
+
attach_function :taurus_get_strict_mode, [], :int
|
|
84
|
+
attach_function :taurus_set_max_depth, [:int], :void
|
|
85
|
+
attach_function :taurus_get_max_depth, [], :int
|
|
86
|
+
attach_function :taurus_xinclude_process,
|
|
87
|
+
[:taurus_document, :string], :taurus_status
|
|
88
|
+
attach_function :taurus_xinclude_is_include_element,
|
|
89
|
+
[:taurus_element], :int
|
|
90
|
+
attach_function :taurus_xinclude_is_fallback_element,
|
|
91
|
+
[:taurus_element], :int
|
|
92
|
+
attach_function :taurus_xinclude_get_href, [:taurus_element], :string
|
|
93
|
+
attach_function :taurus_xinclude_get_parse, [:taurus_element], :string
|
|
94
|
+
attach_function :taurus_xinclude_get_xpointer, [:taurus_element], :string
|
|
95
|
+
# taurus_xinclude_get_encoding is declared in taurus.h but not exported
|
|
96
|
+
# from the v0.4.4 dylib (likely same visibility bug as the serialize
|
|
97
|
+
# functions; see upstream issue #166). Re-add once #173 / v0.4.5 ships.
|
|
98
|
+
|
|
99
|
+
attach_function :taurus_node_get_type,
|
|
100
|
+
[:taurus_node_ref], :int
|
|
101
|
+
attach_function :taurus_node_first_child,
|
|
102
|
+
[:taurus_node_ref], :taurus_node_ref
|
|
103
|
+
attach_function :taurus_node_last_child,
|
|
104
|
+
[:taurus_node_ref], :taurus_node_ref
|
|
105
|
+
attach_function :taurus_node_next_sibling,
|
|
106
|
+
[:taurus_node_ref], :taurus_node_ref
|
|
107
|
+
attach_function :taurus_node_previous_sibling,
|
|
108
|
+
[:taurus_node_ref], :taurus_node_ref
|
|
109
|
+
attach_function :taurus_node_child_count,
|
|
110
|
+
[:taurus_node_ref], :size_t
|
|
111
|
+
attach_function :taurus_node_as_element,
|
|
112
|
+
[:taurus_node_ref], :taurus_element
|
|
113
|
+
attach_function :taurus_element_as_node,
|
|
114
|
+
[:taurus_element], :taurus_node_ref
|
|
115
|
+
attach_function :taurus_node_get_binding_wrapper,
|
|
116
|
+
[:taurus_node_ref], :pointer
|
|
117
|
+
attach_function :taurus_node_set_binding_wrapper,
|
|
118
|
+
[:taurus_node_ref, :pointer], :void
|
|
119
|
+
attach_function :taurus_node_parent,
|
|
120
|
+
[:taurus_node_ref], :taurus_element
|
|
121
|
+
attach_function :taurus_node_unlink,
|
|
122
|
+
[:taurus_node_ref], :taurus_status
|
|
123
|
+
attach_function :taurus_node_line,
|
|
124
|
+
[:taurus_node_ref], :int
|
|
125
|
+
attach_function :taurus_node_compare,
|
|
126
|
+
[:taurus_node_ref, :taurus_node_ref], :int
|
|
127
|
+
attach_function :taurus_node_traverse,
|
|
128
|
+
[:taurus_node_ref, :int, :pointer, :pointer], :int
|
|
129
|
+
|
|
130
|
+
attach_function :taurus_text_node_get_content,
|
|
131
|
+
[:taurus_node_ref], :string
|
|
132
|
+
attach_function :taurus_comment_node_get_content,
|
|
133
|
+
[:taurus_node_ref], :string
|
|
134
|
+
attach_function :taurus_cdata_node_get_content,
|
|
135
|
+
[:taurus_node_ref], :string
|
|
136
|
+
attach_function :taurus_pi_node_get_target,
|
|
137
|
+
[:taurus_node_ref], :string
|
|
138
|
+
attach_function :taurus_pi_node_get_data,
|
|
139
|
+
[:taurus_node_ref], :string
|
|
140
|
+
attach_function :taurus_text_node_create,
|
|
141
|
+
[:taurus_document, :string], :taurus_node_ref
|
|
142
|
+
attach_function :taurus_comment_node_create,
|
|
143
|
+
[:taurus_document, :string], :taurus_node_ref
|
|
144
|
+
attach_function :taurus_cdata_node_create,
|
|
145
|
+
[:taurus_document, :string], :taurus_node_ref
|
|
146
|
+
attach_function :taurus_pi_node_create,
|
|
147
|
+
[:taurus_document, :string, :string], :taurus_node_ref
|
|
148
|
+
attach_function :taurus_text_node_set_content,
|
|
149
|
+
[:taurus_node_ref, :string], :taurus_status
|
|
150
|
+
attach_function :taurus_comment_node_set_content,
|
|
151
|
+
[:taurus_node_ref, :string], :taurus_status
|
|
152
|
+
attach_function :taurus_cdata_node_set_content,
|
|
153
|
+
[:taurus_node_ref, :string], :taurus_status
|
|
154
|
+
attach_function :taurus_pi_node_set_target,
|
|
155
|
+
[:taurus_node_ref, :string], :taurus_status
|
|
156
|
+
attach_function :taurus_pi_node_set_data,
|
|
157
|
+
[:taurus_node_ref, :string], :taurus_status
|
|
158
|
+
|
|
159
|
+
attach_function :taurus_element_name,
|
|
160
|
+
[:taurus_element], :string
|
|
161
|
+
attach_function :taurus_element_text,
|
|
162
|
+
[:taurus_element], :string
|
|
163
|
+
attach_function :taurus_element_text_int,
|
|
164
|
+
[:taurus_element, :int], :int
|
|
165
|
+
attach_function :taurus_element_text_uint,
|
|
166
|
+
[:taurus_element, :uint], :uint
|
|
167
|
+
attach_function :taurus_element_text_double,
|
|
168
|
+
[:taurus_element, :double], :double
|
|
169
|
+
attach_function :taurus_element_text_float,
|
|
170
|
+
[:taurus_element, :float], :float
|
|
171
|
+
attach_function :taurus_element_text_bool,
|
|
172
|
+
[:taurus_element, :int], :int
|
|
173
|
+
attach_function :taurus_element_attribute,
|
|
174
|
+
[:taurus_element, :string], :string
|
|
175
|
+
attach_function :taurus_element_attribute_string,
|
|
176
|
+
[:taurus_element, :string, :string], :string
|
|
177
|
+
attach_function :taurus_element_attribute_int,
|
|
178
|
+
[:taurus_element, :string, :int], :int
|
|
179
|
+
attach_function :taurus_element_attribute_uint,
|
|
180
|
+
[:taurus_element, :string, :uint], :uint
|
|
181
|
+
attach_function :taurus_element_attribute_double,
|
|
182
|
+
[:taurus_element, :string, :double], :double
|
|
183
|
+
attach_function :taurus_element_attribute_float,
|
|
184
|
+
[:taurus_element, :string, :float], :float
|
|
185
|
+
attach_function :taurus_element_attribute_bool,
|
|
186
|
+
[:taurus_element, :string, :int], :int
|
|
187
|
+
# taurus_element_has_attribute is declared in taurus.h but not exported
|
|
188
|
+
# from the v0.4.4 dylib (likely visibility bug; see upstream #166).
|
|
189
|
+
# Ruby-level Element#key? will emulate via attribute lookup.
|
|
190
|
+
attach_function :taurus_element_attribute_count,
|
|
191
|
+
[:taurus_element], :size_t
|
|
192
|
+
attach_function :taurus_element_attribute_name_at,
|
|
193
|
+
[:taurus_element, :size_t], :string
|
|
194
|
+
attach_function :taurus_element_attribute_value_at,
|
|
195
|
+
[:taurus_element, :size_t], :string
|
|
196
|
+
attach_function :taurus_element_remove_all_attributes,
|
|
197
|
+
[:taurus_element], :taurus_status
|
|
198
|
+
|
|
199
|
+
attach_function :taurus_element_child_count,
|
|
200
|
+
[:taurus_element], :size_t
|
|
201
|
+
attach_function :taurus_element_child,
|
|
202
|
+
[:taurus_element, :size_t], :taurus_element
|
|
203
|
+
attach_function :taurus_element_parent,
|
|
204
|
+
[:taurus_element], :taurus_element
|
|
205
|
+
attach_function :taurus_element_root,
|
|
206
|
+
[:taurus_element], :taurus_element
|
|
207
|
+
attach_function :taurus_element_child_value,
|
|
208
|
+
[:taurus_element], :string
|
|
209
|
+
attach_function :taurus_element_hash_value,
|
|
210
|
+
[:taurus_element], :size_t
|
|
211
|
+
attach_function :taurus_element_find_child,
|
|
212
|
+
[:taurus_element, :string], :taurus_element
|
|
213
|
+
attach_function :taurus_element_find_child_by_attr,
|
|
214
|
+
[:taurus_element, :string, :string, :string], :taurus_element
|
|
215
|
+
attach_function :taurus_element_next_sibling,
|
|
216
|
+
[:taurus_element, :string], :taurus_element
|
|
217
|
+
attach_function :taurus_element_previous_sibling,
|
|
218
|
+
[:taurus_element, :string], :taurus_element
|
|
219
|
+
attach_function :taurus_element_first_child,
|
|
220
|
+
[:taurus_element, :string], :taurus_element
|
|
221
|
+
attach_function :taurus_element_last_child,
|
|
222
|
+
[:taurus_element, :string], :taurus_element
|
|
223
|
+
attach_function :taurus_element_first_child_any,
|
|
224
|
+
[:taurus_element], :taurus_element
|
|
225
|
+
attach_function :taurus_element_last_child_any,
|
|
226
|
+
[:taurus_element], :taurus_element
|
|
227
|
+
attach_function :taurus_element_next_sibling_any,
|
|
228
|
+
[:taurus_element], :taurus_element
|
|
229
|
+
attach_function :taurus_element_previous_sibling_any,
|
|
230
|
+
[:taurus_element], :taurus_element
|
|
231
|
+
|
|
232
|
+
attach_function :taurus_element_create,
|
|
233
|
+
[:taurus_document, :string], :taurus_element
|
|
234
|
+
attach_function :taurus_element_set_name,
|
|
235
|
+
[:taurus_element, :string], :taurus_status
|
|
236
|
+
attach_function :taurus_element_set_text,
|
|
237
|
+
[:taurus_element, :string], :taurus_status
|
|
238
|
+
attach_function :taurus_element_set_attribute,
|
|
239
|
+
[:taurus_element, :string, :string], :taurus_status
|
|
240
|
+
attach_function :taurus_element_set_attribute_bool,
|
|
241
|
+
[:taurus_element, :string, :int], :taurus_status
|
|
242
|
+
attach_function :taurus_element_set_attribute_double,
|
|
243
|
+
[:taurus_element, :string, :double], :taurus_status
|
|
244
|
+
attach_function :taurus_element_set_attribute_float,
|
|
245
|
+
[:taurus_element, :string, :float], :taurus_status
|
|
246
|
+
attach_function :taurus_element_set_attribute_int,
|
|
247
|
+
[:taurus_element, :string, :int], :taurus_status
|
|
248
|
+
attach_function :taurus_element_set_attribute_uint,
|
|
249
|
+
[:taurus_element, :string, :uint], :taurus_status
|
|
250
|
+
attach_function :taurus_element_remove_attribute,
|
|
251
|
+
[:taurus_element, :string], :taurus_status
|
|
252
|
+
attach_function :taurus_element_append_child,
|
|
253
|
+
[:taurus_element, :taurus_element], :taurus_status
|
|
254
|
+
attach_function :taurus_element_prepend_child,
|
|
255
|
+
[:taurus_element, :taurus_element], :taurus_status
|
|
256
|
+
attach_function :taurus_element_insert_before,
|
|
257
|
+
[:taurus_element, :taurus_element], :taurus_status
|
|
258
|
+
attach_function :taurus_element_insert_after,
|
|
259
|
+
[:taurus_element, :taurus_element], :taurus_status
|
|
260
|
+
attach_function :taurus_element_remove_child,
|
|
261
|
+
[:taurus_element, :taurus_element], :taurus_status
|
|
262
|
+
attach_function :taurus_element_remove_children,
|
|
263
|
+
[:taurus_element], :taurus_status
|
|
264
|
+
attach_function :taurus_element_append_copy,
|
|
265
|
+
[:taurus_element, :taurus_element], :taurus_element
|
|
266
|
+
attach_function :taurus_element_prepend_copy,
|
|
267
|
+
[:taurus_element, :taurus_element], :taurus_element
|
|
268
|
+
attach_function :taurus_element_insert_copy_before,
|
|
269
|
+
[:taurus_element, :taurus_element], :taurus_element
|
|
270
|
+
attach_function :taurus_element_insert_copy_after,
|
|
271
|
+
[:taurus_element, :taurus_element], :taurus_element
|
|
272
|
+
|
|
273
|
+
attach_function :taurus_element_copy,
|
|
274
|
+
[:taurus_element, :taurus_document], :taurus_element
|
|
275
|
+
attach_function :taurus_document_copy,
|
|
276
|
+
[:taurus_document], :taurus_document
|
|
277
|
+
attach_function :taurus_node_get_xpath,
|
|
278
|
+
[:taurus_node_ref], :pointer
|
|
279
|
+
attach_function :taurus_parse_fragment,
|
|
280
|
+
[:string, :size_t, :taurus_document, :pointer], :taurus_element
|
|
281
|
+
|
|
282
|
+
attach_function :taurus_element_namespace,
|
|
283
|
+
[:taurus_element], :string
|
|
284
|
+
attach_function :taurus_element_namespace_for_prefix,
|
|
285
|
+
[:taurus_element, :string], :string
|
|
286
|
+
attach_function :taurus_element_namespace_count,
|
|
287
|
+
[:taurus_element], :size_t
|
|
288
|
+
attach_function :taurus_element_namespace_decl_prefix,
|
|
289
|
+
[:taurus_element, :size_t], :string
|
|
290
|
+
attach_function :taurus_element_namespace_decl_uri,
|
|
291
|
+
[:taurus_element, :size_t], :string
|
|
292
|
+
attach_function :taurus_element_add_namespace_definition,
|
|
293
|
+
[:taurus_element, :string, :string], :taurus_status
|
|
294
|
+
attach_function :taurus_element_set_default_namespace,
|
|
295
|
+
[:taurus_element, :string], :taurus_status
|
|
296
|
+
attach_function :taurus_element_remove_namespace_definition,
|
|
297
|
+
[:taurus_element, :string], :taurus_status
|
|
298
|
+
attach_function :taurus_namespace_uri, [:string], :string
|
|
299
|
+
attach_function :taurus_namespace_prefix, [:string], :string
|
|
300
|
+
|
|
301
|
+
attach_function :taurus_xpath_eval,
|
|
302
|
+
[:taurus_document, :taurus_element, :string], :taurus_xpath_result
|
|
303
|
+
attach_function :taurus_xpath_eval_with_vars,
|
|
304
|
+
[:taurus_document, :string, :taurus_xpath_var_set], :taurus_xpath_result
|
|
305
|
+
attach_function :taurus_xpath_eval_with_vars_context,
|
|
306
|
+
[:taurus_document, :taurus_element, :string, :taurus_xpath_var_set],
|
|
307
|
+
:taurus_xpath_result
|
|
308
|
+
attach_function :taurus_xpath_result_type,
|
|
309
|
+
[:taurus_xpath_result], :int
|
|
310
|
+
attach_function :taurus_xpath_result_count,
|
|
311
|
+
[:taurus_xpath_result], :size_t
|
|
312
|
+
attach_function :taurus_xpath_result_get,
|
|
313
|
+
[:taurus_xpath_result, :size_t], :taurus_element
|
|
314
|
+
attach_function :taurus_xpath_result_get_nodes,
|
|
315
|
+
[:taurus_xpath_result, :pointer, :size_t], :size_t
|
|
316
|
+
attach_function :taurus_xpath_result_boolean,
|
|
317
|
+
[:taurus_xpath_result], :int
|
|
318
|
+
attach_function :taurus_xpath_result_number,
|
|
319
|
+
[:taurus_xpath_result], :double
|
|
320
|
+
attach_function :taurus_xpath_result_string,
|
|
321
|
+
[:taurus_xpath_result], :pointer
|
|
322
|
+
attach_function :taurus_xpath_result_free,
|
|
323
|
+
[:taurus_xpath_result], :void
|
|
324
|
+
attach_function :taurus_xpath_function_supported, [:string], :int
|
|
325
|
+
attach_function :taurus_xpath_supported_functions, [], :pointer
|
|
326
|
+
|
|
327
|
+
attach_function :taurus_xpath_variable_set_new,
|
|
328
|
+
[], :taurus_xpath_var_set
|
|
329
|
+
attach_function :taurus_xpath_variable_set_free,
|
|
330
|
+
[:taurus_xpath_var_set], :void
|
|
331
|
+
attach_function :taurus_xpath_variable_set_boolean,
|
|
332
|
+
[:taurus_xpath_var_set, :string, :int], :taurus_status
|
|
333
|
+
attach_function :taurus_xpath_variable_set_number,
|
|
334
|
+
[:taurus_xpath_var_set, :string, :double], :taurus_status
|
|
335
|
+
attach_function :taurus_xpath_variable_set_string,
|
|
336
|
+
[:taurus_xpath_var_set, :string, :string], :taurus_status
|
|
337
|
+
|
|
338
|
+
attach_function :taurus_sax_parse,
|
|
339
|
+
[:string, :size_t, :pointer, :pointer], :int
|
|
340
|
+
attach_function :taurus_sax_parser_create,
|
|
341
|
+
[:pointer, :pointer], :taurus_sax_parser
|
|
342
|
+
attach_function :taurus_sax_parser_feed,
|
|
343
|
+
[:taurus_sax_parser, :string, :size_t, :int], :int
|
|
344
|
+
attach_function :taurus_sax_parser_free,
|
|
345
|
+
[:taurus_sax_parser], :void
|
|
346
|
+
attach_function :taurus_sax_parser_set_streaming,
|
|
347
|
+
[:taurus_sax_parser, :int], :int
|
|
348
|
+
|
|
349
|
+
attach_function :taurus_document_serialize,
|
|
350
|
+
[:taurus_document, :pointer], :pointer
|
|
351
|
+
attach_function :taurus_element_serialize,
|
|
352
|
+
[:taurus_element, :pointer], :pointer
|
|
353
|
+
attach_function :taurus_document_save_file,
|
|
354
|
+
[:taurus_document, :string, :pointer], :taurus_status
|
|
355
|
+
attach_function :taurus_c14n_canonicalize,
|
|
356
|
+
[:taurus_document, :int, :int], :pointer
|
|
357
|
+
attach_function :taurus_c14n_canonicalize_subtree,
|
|
358
|
+
[:taurus_element, :int, :int], :pointer
|
|
359
|
+
attach_function :taurus_c14n_canonicalize_ex,
|
|
360
|
+
[:taurus_document, :int, :int, :pointer, :int], :pointer
|
|
361
|
+
attach_function :taurus_c14n_canonicalize_subtree_ex,
|
|
362
|
+
[:taurus_element, :int, :int, :pointer, :int], :pointer
|
|
363
|
+
|
|
364
|
+
attach_function :taurus_document_internal_subset,
|
|
365
|
+
[:taurus_document], :taurus_doctype
|
|
366
|
+
attach_function :taurus_doctype_get_name,
|
|
367
|
+
[:taurus_doctype], :string
|
|
368
|
+
attach_function :taurus_doctype_get_root_name,
|
|
369
|
+
[:taurus_doctype], :string
|
|
370
|
+
attach_function :taurus_doctype_get_public_id,
|
|
371
|
+
[:taurus_doctype], :string
|
|
372
|
+
attach_function :taurus_doctype_get_system_id,
|
|
373
|
+
[:taurus_doctype], :string
|
|
374
|
+
attach_function :taurus_doctype_get_internal_subset,
|
|
375
|
+
[:taurus_doctype], :string
|
|
376
|
+
|
|
377
|
+
attach_function :taurus_free_string, [:pointer], :void
|
|
378
|
+
attach_function :taurus_explicit_cleanup, [], :void
|
|
379
|
+
attach_function :taurus_set_memory_management_functions,
|
|
380
|
+
[:pointer, :pointer], :void
|
|
381
|
+
attach_function :taurus_get_memory_allocation_function, [], :pointer
|
|
382
|
+
attach_function :taurus_get_memory_deallocation_function, [], :pointer
|
|
383
|
+
attach_function :taurus_document_set_allocators,
|
|
384
|
+
[:taurus_document, :pointer, :pointer], :taurus_status
|
|
385
|
+
attach_function :taurus_status_string, [:taurus_status], :string
|
|
386
|
+
|
|
387
|
+
TAURUS_OK = 0
|
|
388
|
+
TAURUS_ERROR_MEMORY = -1
|
|
389
|
+
TAURUS_ERROR_PARSE = -2
|
|
390
|
+
TAURUS_ERROR_XPATH = -3
|
|
391
|
+
TAURUS_ERROR_NULL_ARG = -4
|
|
392
|
+
TAURUS_ERROR_INVALID_ARG = -5
|
|
393
|
+
TAURUS_ERROR_NOT_FOUND = -6
|
|
394
|
+
TAURUS_ERROR_IO = -7
|
|
395
|
+
TAURUS_ERROR_NOT_IMPLEMENTED = -8
|
|
396
|
+
|
|
397
|
+
XPATH_NODESET = 0
|
|
398
|
+
XPATH_BOOLEAN = 1
|
|
399
|
+
XPATH_NUMBER = 2
|
|
400
|
+
XPATH_STRING = 3
|
|
401
|
+
|
|
402
|
+
NODE_ELEMENT = 0
|
|
403
|
+
NODE_TEXT = 1
|
|
404
|
+
NODE_COMMENT = 2
|
|
405
|
+
NODE_CDATA = 3
|
|
406
|
+
NODE_PI = 4
|
|
407
|
+
NODE_DOCTYPE = 5
|
|
408
|
+
NODE_ATTRIBUTE = 6
|
|
409
|
+
|
|
410
|
+
C14N_1_0 = 0
|
|
411
|
+
C14N_1_1 = 1
|
|
412
|
+
|
|
413
|
+
C14N_MODE_CANONICAL = 0
|
|
414
|
+
C14N_MODE_EXCLUSIVE = 1
|
|
415
|
+
|
|
416
|
+
TRAVERSE_PRE_ORDER = 0
|
|
417
|
+
TRAVERSE_POST_ORDER = 1
|
|
418
|
+
end
|
|
419
|
+
end
|
|
420
|
+
end
|
|
@@ -0,0 +1,43 @@
|
|
|
1
|
+
# frozen_string_literal: true
|
|
2
|
+
|
|
3
|
+
# libtaurus's TaurusNamespace is `const char*` (the URI directly). The Ruby
|
|
4
|
+
# Namespace wrapper holds (element, uri, optional prefix). When the prefix
|
|
5
|
+
# isn't supplied (e.g. via Element#namespace), it can be derived by walking
|
|
6
|
+
# the element's namespace declarations via taurus_element_namespace_decl_*.
|
|
7
|
+
|
|
8
|
+
class Taurus::XML::Namespace
|
|
9
|
+
attr_reader :element, :href, :prefix
|
|
10
|
+
|
|
11
|
+
def initialize(element, href, prefix: :derive)
|
|
12
|
+
@element = element
|
|
13
|
+
@href = href
|
|
14
|
+
@prefix = prefix == :derive ? derive_prefix : prefix
|
|
15
|
+
end
|
|
16
|
+
|
|
17
|
+
def document
|
|
18
|
+
@element&.document
|
|
19
|
+
end
|
|
20
|
+
|
|
21
|
+
def ==(other)
|
|
22
|
+
other.is_a?(Taurus::XML::Namespace) &&
|
|
23
|
+
@href == other.href && @prefix == other.prefix
|
|
24
|
+
end
|
|
25
|
+
|
|
26
|
+
def inspect
|
|
27
|
+
"#<#{self.class.name} prefix=#{@prefix.inspect} href=#{@href.inspect}>"
|
|
28
|
+
end
|
|
29
|
+
|
|
30
|
+
private
|
|
31
|
+
|
|
32
|
+
def derive_prefix
|
|
33
|
+
return nil if @element.nil?
|
|
34
|
+
count = Taurus::XML::FFI.taurus_element_namespace_count(@element.c_ptr)
|
|
35
|
+
count.times do |i|
|
|
36
|
+
uri = Taurus::XML::FFI.taurus_element_namespace_decl_uri(@element.c_ptr, i)
|
|
37
|
+
next unless uri == @href
|
|
38
|
+
prefix = Taurus::XML::FFI.taurus_element_namespace_decl_prefix(@element.c_ptr, i)
|
|
39
|
+
return prefix.nil? || prefix.empty? ? nil : prefix
|
|
40
|
+
end
|
|
41
|
+
nil
|
|
42
|
+
end
|
|
43
|
+
end
|
|
@@ -0,0 +1,221 @@
|
|
|
1
|
+
# frozen_string_literal: true
|
|
2
|
+
|
|
3
|
+
class Taurus::XML::Node
|
|
4
|
+
attr_reader :c_ptr, :document
|
|
5
|
+
|
|
6
|
+
def initialize(c_ptr, document, parent: nil)
|
|
7
|
+
@c_ptr = c_ptr
|
|
8
|
+
@document = document
|
|
9
|
+
@parent = parent
|
|
10
|
+
end
|
|
11
|
+
|
|
12
|
+
def self.wrap(c_ptr, document, parent: nil)
|
|
13
|
+
# Per-document weak-ref cache. Returns the existing wrapper when the
|
|
14
|
+
# same c_ptr is wrapped twice (common in children/sibling walks,
|
|
15
|
+
# repeated xpath queries, traverse-then-access patterns). The cache
|
|
16
|
+
# dies with the document so no stale entries.
|
|
17
|
+
if document && (cached = document.wrapper_cache[c_ptr.address])
|
|
18
|
+
return cached
|
|
19
|
+
end
|
|
20
|
+
|
|
21
|
+
node =
|
|
22
|
+
case Taurus::XML::FFI.taurus_node_get_type(c_ptr)
|
|
23
|
+
when Taurus::XML::FFI::NODE_ELEMENT
|
|
24
|
+
Taurus::XML::Element.new(c_ptr, document, parent: parent)
|
|
25
|
+
when Taurus::XML::FFI::NODE_TEXT
|
|
26
|
+
Taurus::XML::Text.new(c_ptr, document, parent: parent)
|
|
27
|
+
when Taurus::XML::FFI::NODE_COMMENT
|
|
28
|
+
Taurus::XML::Comment.new(c_ptr, document, parent: parent)
|
|
29
|
+
when Taurus::XML::FFI::NODE_CDATA
|
|
30
|
+
Taurus::XML::CDATA.new(c_ptr, document, parent: parent)
|
|
31
|
+
when Taurus::XML::FFI::NODE_PI
|
|
32
|
+
Taurus::XML::ProcessingInstruction.new(c_ptr, document, parent: parent)
|
|
33
|
+
else
|
|
34
|
+
new(c_ptr, document, parent: parent)
|
|
35
|
+
end
|
|
36
|
+
|
|
37
|
+
document.wrapper_cache[c_ptr.address] = node if document
|
|
38
|
+
node
|
|
39
|
+
end
|
|
40
|
+
|
|
41
|
+
def name
|
|
42
|
+
raise NotImplementedError, "#{self.class}#name not implemented"
|
|
43
|
+
end
|
|
44
|
+
|
|
45
|
+
def content
|
|
46
|
+
raise NotImplementedError, "#{self.class}#content not implemented"
|
|
47
|
+
end
|
|
48
|
+
alias_method :text, :content
|
|
49
|
+
alias_method :inner_text, :content
|
|
50
|
+
|
|
51
|
+
def type
|
|
52
|
+
Taurus::XML::FFI.taurus_node_get_type(@c_ptr)
|
|
53
|
+
end
|
|
54
|
+
alias_method :node_type, :type
|
|
55
|
+
|
|
56
|
+
def element?; type == Taurus::XML::FFI::NODE_ELEMENT; end
|
|
57
|
+
def text?; type == Taurus::XML::FFI::NODE_TEXT; end
|
|
58
|
+
def comment?; type == Taurus::XML::FFI::NODE_COMMENT; end
|
|
59
|
+
def cdata?; type == Taurus::XML::FFI::NODE_CDATA; end
|
|
60
|
+
def processing_instruction?
|
|
61
|
+
type == Taurus::XML::FFI::NODE_PI
|
|
62
|
+
end
|
|
63
|
+
alias_method :pi?, :processing_instruction?
|
|
64
|
+
|
|
65
|
+
def parent
|
|
66
|
+
return @parent if @parent
|
|
67
|
+
ptr = Taurus::XML::FFI.taurus_node_parent(@c_ptr)
|
|
68
|
+
return nil if ptr.null?
|
|
69
|
+
Taurus::XML::Element.new(ptr, @document)
|
|
70
|
+
end
|
|
71
|
+
|
|
72
|
+
def line
|
|
73
|
+
Taurus::XML::FFI.taurus_node_line(@c_ptr)
|
|
74
|
+
end
|
|
75
|
+
|
|
76
|
+
def <=>(other)
|
|
77
|
+
return nil unless other.is_a?(Taurus::XML::Node)
|
|
78
|
+
return nil unless @document == other.document
|
|
79
|
+
Taurus::XML::FFI.taurus_node_compare(@c_ptr, other.c_ptr)
|
|
80
|
+
end
|
|
81
|
+
|
|
82
|
+
def child
|
|
83
|
+
ptr = Taurus::XML::FFI.taurus_node_first_child(@c_ptr)
|
|
84
|
+
return nil if ptr.null?
|
|
85
|
+
Taurus::XML::Node.wrap(ptr, @document, parent: as_element_or_self)
|
|
86
|
+
end
|
|
87
|
+
|
|
88
|
+
def children
|
|
89
|
+
nodes = []
|
|
90
|
+
ptr = Taurus::XML::FFI.taurus_node_first_child(@c_ptr)
|
|
91
|
+
until ptr.nil? || ptr.null?
|
|
92
|
+
nodes << Taurus::XML::Node.wrap(ptr, @document, parent: as_element_or_self)
|
|
93
|
+
ptr = Taurus::XML::FFI.taurus_node_next_sibling(ptr)
|
|
94
|
+
end
|
|
95
|
+
Taurus::XML::NodeSet.new(@document, nodes)
|
|
96
|
+
end
|
|
97
|
+
|
|
98
|
+
def next_sibling
|
|
99
|
+
ptr = Taurus::XML::FFI.taurus_node_next_sibling(@c_ptr)
|
|
100
|
+
return nil if ptr.null?
|
|
101
|
+
Taurus::XML::Node.wrap(ptr, @document, parent: @parent)
|
|
102
|
+
end
|
|
103
|
+
alias_method :next, :next_sibling
|
|
104
|
+
|
|
105
|
+
def previous_sibling
|
|
106
|
+
ptr = Taurus::XML::FFI.taurus_node_previous_sibling(@c_ptr)
|
|
107
|
+
return nil if ptr.null?
|
|
108
|
+
Taurus::XML::Node.wrap(ptr, @document, parent: @parent)
|
|
109
|
+
end
|
|
110
|
+
alias_method :previous, :previous_sibling
|
|
111
|
+
|
|
112
|
+
def first_element_child
|
|
113
|
+
ptr = Taurus::XML::FFI.taurus_node_first_child(@c_ptr)
|
|
114
|
+
until ptr.nil? || ptr.null?
|
|
115
|
+
node = Taurus::XML::Node.wrap(ptr, @document, parent: as_element_or_self)
|
|
116
|
+
return node if node.element?
|
|
117
|
+
ptr = Taurus::XML::FFI.taurus_node_next_sibling(ptr)
|
|
118
|
+
end
|
|
119
|
+
nil
|
|
120
|
+
end
|
|
121
|
+
|
|
122
|
+
def last_element_child
|
|
123
|
+
children.reverse_each.find(&:element?)
|
|
124
|
+
end
|
|
125
|
+
|
|
126
|
+
def element_children
|
|
127
|
+
children.select(&:element?)
|
|
128
|
+
end
|
|
129
|
+
alias_method :elements, :element_children
|
|
130
|
+
|
|
131
|
+
def next_element
|
|
132
|
+
sibling = next_sibling
|
|
133
|
+
sibling = sibling.next_sibling until sibling.nil? || sibling.element?
|
|
134
|
+
sibling
|
|
135
|
+
end
|
|
136
|
+
|
|
137
|
+
def previous_element
|
|
138
|
+
sibling = previous_sibling
|
|
139
|
+
sibling = sibling.previous_sibling until sibling.nil? || sibling.element?
|
|
140
|
+
sibling
|
|
141
|
+
end
|
|
142
|
+
|
|
143
|
+
def unlink
|
|
144
|
+
status = Taurus::XML::FFI.taurus_node_unlink(@c_ptr)
|
|
145
|
+
raise Taurus::XML::Error,
|
|
146
|
+
Taurus::XML::FFI.taurus_status_string(status) unless status == Taurus::XML::FFI::TAURUS_OK
|
|
147
|
+
@parent = nil
|
|
148
|
+
self
|
|
149
|
+
end
|
|
150
|
+
alias_method :remove, :unlink
|
|
151
|
+
|
|
152
|
+
# Walks the subtree in post-order DFS (matches Nokogiri's semantics).
|
|
153
|
+
#
|
|
154
|
+
# Specialized hot path: skips the intermediate NodeSet allocation that
|
|
155
|
+
# Element#children would create, walking via raw FFI calls and wrapping
|
|
156
|
+
# nodes directly. Saves one Array + one NodeSet allocation per parent
|
|
157
|
+
# node. For a tree of N nodes that's ~N fewer allocations on a full
|
|
158
|
+
# traversal.
|
|
159
|
+
#
|
|
160
|
+
# Still pays ~2 FFI calls per visited node (first_child + next_sibling).
|
|
161
|
+
# Beating Nokogiri on this benchmark needs C-side traverse with a
|
|
162
|
+
# callback (libtaurus #273); the per-node FFI cost is the floor.
|
|
163
|
+
def traverse
|
|
164
|
+
return enum_for(:traverse) unless block_given?
|
|
165
|
+
callback = ::FFI::Function.new(:int, [:pointer, :pointer], blocking: true) do |node_ptr, _|
|
|
166
|
+
yield Taurus::XML::Node.wrap(node_ptr, @document)
|
|
167
|
+
0
|
|
168
|
+
end
|
|
169
|
+
Taurus::XML::FFI.taurus_node_traverse(
|
|
170
|
+
@c_ptr, Taurus::XML::FFI::TRAVERSE_POST_ORDER, callback, nil)
|
|
171
|
+
end
|
|
172
|
+
|
|
173
|
+
def path
|
|
174
|
+
str_ptr = Taurus::XML::FFI.taurus_node_get_xpath(@c_ptr)
|
|
175
|
+
return nil if str_ptr.null?
|
|
176
|
+
str_ptr.read_string.tap { Taurus::XML::FFI.taurus_free_string(str_ptr) }
|
|
177
|
+
end
|
|
178
|
+
|
|
179
|
+
def css_path
|
|
180
|
+
return nil if path.nil?
|
|
181
|
+
path.split("/").filter_map do |part|
|
|
182
|
+
next nil if part.empty?
|
|
183
|
+
part.gsub(/\[(\d+)\]/, ':nth-of-type(\1)')
|
|
184
|
+
end.join(" > ")
|
|
185
|
+
end
|
|
186
|
+
|
|
187
|
+
def dup
|
|
188
|
+
elem_ptr = Taurus::XML::FFI.taurus_node_as_element(@c_ptr)
|
|
189
|
+
raise Taurus::XML::Error, "dup is only supported for element nodes" if elem_ptr.null?
|
|
190
|
+
copy_ptr = Taurus::XML::FFI.taurus_element_copy(elem_ptr, @document.c_ptr)
|
|
191
|
+
raise Taurus::XML::Error, "taurus_element_copy failed" if copy_ptr.null?
|
|
192
|
+
Taurus::XML::Element.new(copy_ptr, @document)
|
|
193
|
+
end
|
|
194
|
+
alias_method :clone, :dup
|
|
195
|
+
|
|
196
|
+
def ==(other)
|
|
197
|
+
return false unless other.is_a?(Taurus::XML::Node)
|
|
198
|
+
@c_ptr == other.c_ptr
|
|
199
|
+
end
|
|
200
|
+
|
|
201
|
+
def inspect
|
|
202
|
+
"#<#{self.class.name} ptr=#{c_ptr}>"
|
|
203
|
+
end
|
|
204
|
+
|
|
205
|
+
protected
|
|
206
|
+
|
|
207
|
+
def as_element_or_self
|
|
208
|
+
is_a?(Taurus::XML::Element) ? self : nil
|
|
209
|
+
end
|
|
210
|
+
|
|
211
|
+
private
|
|
212
|
+
|
|
213
|
+
# Walks the subtree in post-order DFS (matches Nokogiri's semantics).
|
|
214
|
+
#
|
|
215
|
+
# Specialized hot path for #traverse: skips the intermediate NodeSet
|
|
216
|
+
# allocation that Element#children would create, walking via raw FFI
|
|
217
|
+
# calls and wrapping nodes directly. Saves one Array + one NodeSet
|
|
218
|
+
# allocation per parent node.
|
|
219
|
+
#
|
|
220
|
+
include Taurus::XML::Searchable
|
|
221
|
+
end
|