regex-treetop 1.4.8
Sign up to get free protection for your applications and to get access to all the features.
- data/LICENSE +19 -0
- data/README.md +164 -0
- data/Rakefile +19 -0
- data/bin/tt +112 -0
- data/doc/contributing_and_planned_features.markdown +103 -0
- data/doc/grammar_composition.markdown +65 -0
- data/doc/index.markdown +90 -0
- data/doc/pitfalls_and_advanced_techniques.markdown +51 -0
- data/doc/semantic_interpretation.markdown +189 -0
- data/doc/site.rb +112 -0
- data/doc/sitegen.rb +65 -0
- data/doc/syntactic_recognition.markdown +100 -0
- data/doc/using_in_ruby.markdown +21 -0
- data/examples/lambda_calculus/arithmetic.rb +551 -0
- data/examples/lambda_calculus/arithmetic.treetop +97 -0
- data/examples/lambda_calculus/arithmetic_node_classes.rb +7 -0
- data/examples/lambda_calculus/arithmetic_test.rb +54 -0
- data/examples/lambda_calculus/lambda_calculus +0 -0
- data/examples/lambda_calculus/lambda_calculus.rb +718 -0
- data/examples/lambda_calculus/lambda_calculus.treetop +132 -0
- data/examples/lambda_calculus/lambda_calculus_node_classes.rb +5 -0
- data/examples/lambda_calculus/lambda_calculus_test.rb +89 -0
- data/examples/lambda_calculus/test_helper.rb +18 -0
- data/lib/treetop.rb +16 -0
- data/lib/treetop/bootstrap_gen_1_metagrammar.rb +45 -0
- data/lib/treetop/compiler.rb +6 -0
- data/lib/treetop/compiler/grammar_compiler.rb +44 -0
- data/lib/treetop/compiler/lexical_address_space.rb +17 -0
- data/lib/treetop/compiler/metagrammar.rb +3392 -0
- data/lib/treetop/compiler/metagrammar.treetop +454 -0
- data/lib/treetop/compiler/node_classes.rb +21 -0
- data/lib/treetop/compiler/node_classes/anything_symbol.rb +18 -0
- data/lib/treetop/compiler/node_classes/atomic_expression.rb +14 -0
- data/lib/treetop/compiler/node_classes/character_class.rb +28 -0
- data/lib/treetop/compiler/node_classes/choice.rb +31 -0
- data/lib/treetop/compiler/node_classes/declaration_sequence.rb +24 -0
- data/lib/treetop/compiler/node_classes/grammar.rb +28 -0
- data/lib/treetop/compiler/node_classes/inline_module.rb +27 -0
- data/lib/treetop/compiler/node_classes/nonterminal.rb +13 -0
- data/lib/treetop/compiler/node_classes/optional.rb +19 -0
- data/lib/treetop/compiler/node_classes/parenthesized_expression.rb +9 -0
- data/lib/treetop/compiler/node_classes/parsing_expression.rb +146 -0
- data/lib/treetop/compiler/node_classes/parsing_rule.rb +55 -0
- data/lib/treetop/compiler/node_classes/predicate.rb +45 -0
- data/lib/treetop/compiler/node_classes/predicate_block.rb +16 -0
- data/lib/treetop/compiler/node_classes/regex.rb +23 -0
- data/lib/treetop/compiler/node_classes/repetition.rb +55 -0
- data/lib/treetop/compiler/node_classes/sequence.rb +71 -0
- data/lib/treetop/compiler/node_classes/terminal.rb +20 -0
- data/lib/treetop/compiler/node_classes/transient_prefix.rb +9 -0
- data/lib/treetop/compiler/node_classes/treetop_file.rb +9 -0
- data/lib/treetop/compiler/ruby_builder.rb +113 -0
- data/lib/treetop/ruby_extensions.rb +2 -0
- data/lib/treetop/ruby_extensions/string.rb +42 -0
- data/lib/treetop/runtime.rb +5 -0
- data/lib/treetop/runtime/compiled_parser.rb +118 -0
- data/lib/treetop/runtime/interval_skip_list.rb +4 -0
- data/lib/treetop/runtime/interval_skip_list/head_node.rb +15 -0
- data/lib/treetop/runtime/interval_skip_list/interval_skip_list.rb +200 -0
- data/lib/treetop/runtime/interval_skip_list/node.rb +164 -0
- data/lib/treetop/runtime/syntax_node.rb +114 -0
- data/lib/treetop/runtime/terminal_parse_failure.rb +16 -0
- data/lib/treetop/runtime/terminal_syntax_node.rb +17 -0
- data/lib/treetop/version.rb +9 -0
- metadata +138 -0
@@ -0,0 +1,454 @@
|
|
1
|
+
module Treetop
|
2
|
+
module Compiler
|
3
|
+
grammar Metagrammar
|
4
|
+
rule treetop_file
|
5
|
+
requires:(space? require_statement)* prefix:space? module_or_grammar:(module_declaration / grammar) suffix:space? {
|
6
|
+
def compile
|
7
|
+
requires.text_value + prefix.text_value + module_or_grammar.compile + suffix.text_value
|
8
|
+
end
|
9
|
+
}
|
10
|
+
end
|
11
|
+
|
12
|
+
rule require_statement
|
13
|
+
prefix:space? "require" [ \t]+ [^\n\r]+ [\n\r]
|
14
|
+
end
|
15
|
+
|
16
|
+
rule module_declaration
|
17
|
+
prefix:('module' space [A-Z] alphanumeric_char* space) module_contents:(module_declaration / grammar) suffix:(space 'end') {
|
18
|
+
def compile
|
19
|
+
prefix.text_value + module_contents.compile + suffix.text_value
|
20
|
+
end
|
21
|
+
}
|
22
|
+
end
|
23
|
+
|
24
|
+
rule grammar
|
25
|
+
'grammar' space grammar_name space ('do' space)? declaration_sequence space? 'end' <Grammar>
|
26
|
+
end
|
27
|
+
|
28
|
+
rule grammar_name
|
29
|
+
([A-Z] alphanumeric_char*)
|
30
|
+
end
|
31
|
+
|
32
|
+
rule declaration_sequence
|
33
|
+
head:declaration tail:(space declaration)* <DeclarationSequence> {
|
34
|
+
def declarations
|
35
|
+
[head] + tail
|
36
|
+
end
|
37
|
+
|
38
|
+
def tail
|
39
|
+
super.elements.map { |elt| elt.declaration }
|
40
|
+
end
|
41
|
+
}
|
42
|
+
/
|
43
|
+
'' {
|
44
|
+
def compile(builder)
|
45
|
+
end
|
46
|
+
}
|
47
|
+
end
|
48
|
+
|
49
|
+
rule declaration
|
50
|
+
parsing_rule / include_declaration
|
51
|
+
end
|
52
|
+
|
53
|
+
rule include_declaration
|
54
|
+
'include' space [A-Z] (alphanumeric_char / '::')* {
|
55
|
+
def compile(builder)
|
56
|
+
builder << text_value
|
57
|
+
end
|
58
|
+
}
|
59
|
+
end
|
60
|
+
|
61
|
+
rule parsing_rule
|
62
|
+
'rule' space nonterminal space ('do' space)? parsing_expression space 'end' <ParsingRule>
|
63
|
+
end
|
64
|
+
|
65
|
+
rule parsing_expression
|
66
|
+
choice / sequence / primary
|
67
|
+
end
|
68
|
+
|
69
|
+
rule choice
|
70
|
+
head:alternative tail:(space? '/' space? alternative)+ <Choice> {
|
71
|
+
def alternatives
|
72
|
+
[head] + tail
|
73
|
+
end
|
74
|
+
|
75
|
+
def tail
|
76
|
+
super.elements.map {|elt| elt.alternative}
|
77
|
+
end
|
78
|
+
|
79
|
+
def inline_modules
|
80
|
+
(alternatives.map {|alt| alt.inline_modules }).flatten
|
81
|
+
end
|
82
|
+
}
|
83
|
+
end
|
84
|
+
|
85
|
+
rule sequence
|
86
|
+
head:labeled_sequence_primary tail:(space labeled_sequence_primary)+ node_class_declarations <Sequence> {
|
87
|
+
def sequence_elements
|
88
|
+
[head] + tail
|
89
|
+
end
|
90
|
+
|
91
|
+
def tail
|
92
|
+
super.elements.map {|elt| elt.labeled_sequence_primary }
|
93
|
+
end
|
94
|
+
|
95
|
+
def inline_modules
|
96
|
+
(sequence_elements.map {|elt| elt.inline_modules}).flatten +
|
97
|
+
[sequence_element_accessor_module] +
|
98
|
+
node_class_declarations.inline_modules
|
99
|
+
end
|
100
|
+
|
101
|
+
def inline_module_name
|
102
|
+
node_class_declarations.inline_module_name
|
103
|
+
end
|
104
|
+
}
|
105
|
+
end
|
106
|
+
|
107
|
+
rule alternative
|
108
|
+
sequence / primary
|
109
|
+
end
|
110
|
+
|
111
|
+
rule primary
|
112
|
+
prefix atomic {
|
113
|
+
def compile(address, builder, parent_expression=nil)
|
114
|
+
prefix.compile(address, builder, self)
|
115
|
+
end
|
116
|
+
|
117
|
+
def prefixed_expression
|
118
|
+
atomic
|
119
|
+
end
|
120
|
+
|
121
|
+
def inline_modules
|
122
|
+
atomic.inline_modules
|
123
|
+
end
|
124
|
+
|
125
|
+
def inline_module_name
|
126
|
+
nil
|
127
|
+
end
|
128
|
+
}
|
129
|
+
/
|
130
|
+
prefix space? predicate_block {
|
131
|
+
def compile(address, builder, parent_expression=nil)
|
132
|
+
prefix.compile(address, builder, self)
|
133
|
+
end
|
134
|
+
def prefixed_expression
|
135
|
+
predicate_block
|
136
|
+
end
|
137
|
+
def inline_modules
|
138
|
+
[]
|
139
|
+
end
|
140
|
+
}
|
141
|
+
/
|
142
|
+
atomic suffix node_class_declarations {
|
143
|
+
def compile(address, builder, parent_expression=nil)
|
144
|
+
suffix.compile(address, builder, self)
|
145
|
+
end
|
146
|
+
|
147
|
+
def optional_expression
|
148
|
+
atomic
|
149
|
+
end
|
150
|
+
|
151
|
+
def node_class_name
|
152
|
+
node_class_declarations.node_class_name
|
153
|
+
end
|
154
|
+
|
155
|
+
def inline_modules
|
156
|
+
atomic.inline_modules + node_class_declarations.inline_modules
|
157
|
+
end
|
158
|
+
|
159
|
+
def inline_module_name
|
160
|
+
node_class_declarations.inline_module_name
|
161
|
+
end
|
162
|
+
}
|
163
|
+
/
|
164
|
+
atomic node_class_declarations {
|
165
|
+
def compile(address, builder, parent_expression=nil)
|
166
|
+
atomic.compile(address, builder, self)
|
167
|
+
end
|
168
|
+
|
169
|
+
def node_class_name
|
170
|
+
node_class_declarations.node_class_name
|
171
|
+
end
|
172
|
+
|
173
|
+
def inline_modules
|
174
|
+
atomic.inline_modules + node_class_declarations.inline_modules
|
175
|
+
end
|
176
|
+
|
177
|
+
def inline_module_name
|
178
|
+
node_class_declarations.inline_module_name
|
179
|
+
end
|
180
|
+
}
|
181
|
+
end
|
182
|
+
|
183
|
+
rule labeled_sequence_primary
|
184
|
+
label sequence_primary {
|
185
|
+
def compile(lexical_address, builder)
|
186
|
+
sequence_primary.compile(lexical_address, builder)
|
187
|
+
end
|
188
|
+
|
189
|
+
def inline_modules
|
190
|
+
sequence_primary.inline_modules
|
191
|
+
end
|
192
|
+
|
193
|
+
def label_name
|
194
|
+
if label.name
|
195
|
+
label.name
|
196
|
+
elsif sequence_primary.instance_of?(Nonterminal)
|
197
|
+
sequence_primary.text_value
|
198
|
+
else
|
199
|
+
nil
|
200
|
+
end
|
201
|
+
end
|
202
|
+
}
|
203
|
+
end
|
204
|
+
|
205
|
+
rule label
|
206
|
+
(alpha_char alphanumeric_char*) ':' {
|
207
|
+
def name
|
208
|
+
elements[0].text_value
|
209
|
+
end
|
210
|
+
}
|
211
|
+
/
|
212
|
+
'' {
|
213
|
+
def name
|
214
|
+
nil
|
215
|
+
end
|
216
|
+
}
|
217
|
+
end
|
218
|
+
|
219
|
+
rule sequence_primary
|
220
|
+
prefix atomic {
|
221
|
+
def compile(lexical_address, builder)
|
222
|
+
prefix.compile(lexical_address, builder, self)
|
223
|
+
end
|
224
|
+
|
225
|
+
def prefixed_expression
|
226
|
+
elements[1]
|
227
|
+
end
|
228
|
+
|
229
|
+
def inline_modules
|
230
|
+
atomic.inline_modules
|
231
|
+
end
|
232
|
+
|
233
|
+
def inline_module_name
|
234
|
+
nil
|
235
|
+
end
|
236
|
+
}
|
237
|
+
/
|
238
|
+
prefix space? predicate_block {
|
239
|
+
def compile(address, builder, parent_expression=nil)
|
240
|
+
prefix.compile(address, builder, self)
|
241
|
+
end
|
242
|
+
def prefixed_expression
|
243
|
+
predicate_block
|
244
|
+
end
|
245
|
+
def inline_modules
|
246
|
+
[]
|
247
|
+
end
|
248
|
+
}
|
249
|
+
/
|
250
|
+
atomic suffix {
|
251
|
+
def compile(lexical_address, builder)
|
252
|
+
suffix.compile(lexical_address, builder, self)
|
253
|
+
end
|
254
|
+
|
255
|
+
def node_class_name
|
256
|
+
nil
|
257
|
+
end
|
258
|
+
|
259
|
+
def inline_modules
|
260
|
+
atomic.inline_modules
|
261
|
+
end
|
262
|
+
|
263
|
+
def inline_module_name
|
264
|
+
nil
|
265
|
+
end
|
266
|
+
}
|
267
|
+
/
|
268
|
+
atomic
|
269
|
+
end
|
270
|
+
|
271
|
+
rule suffix
|
272
|
+
repetition_suffix / optional_suffix
|
273
|
+
end
|
274
|
+
|
275
|
+
rule optional_suffix
|
276
|
+
'?' <Optional>
|
277
|
+
end
|
278
|
+
|
279
|
+
rule node_class_declarations
|
280
|
+
node_class_expression trailing_inline_module {
|
281
|
+
def node_class_name
|
282
|
+
node_class_expression.node_class_name
|
283
|
+
end
|
284
|
+
|
285
|
+
def inline_modules
|
286
|
+
trailing_inline_module.inline_modules
|
287
|
+
end
|
288
|
+
|
289
|
+
def inline_module
|
290
|
+
trailing_inline_module.inline_module
|
291
|
+
end
|
292
|
+
|
293
|
+
def inline_module_name
|
294
|
+
inline_module.module_name if inline_module
|
295
|
+
end
|
296
|
+
}
|
297
|
+
end
|
298
|
+
|
299
|
+
rule repetition_suffix
|
300
|
+
'+' <OneOrMore> / '*' <ZeroOrMore>
|
301
|
+
end
|
302
|
+
|
303
|
+
rule prefix
|
304
|
+
'&' <AndPredicate> / '!' <NotPredicate> / '~' <TransientPrefix>
|
305
|
+
end
|
306
|
+
|
307
|
+
rule atomic
|
308
|
+
regex
|
309
|
+
/
|
310
|
+
terminal
|
311
|
+
/
|
312
|
+
nonterminal
|
313
|
+
/
|
314
|
+
parenthesized_expression
|
315
|
+
end
|
316
|
+
|
317
|
+
rule regex
|
318
|
+
'r(' regex_body ')' <Regex> {
|
319
|
+
def inline_modules
|
320
|
+
[]
|
321
|
+
end
|
322
|
+
}
|
323
|
+
end
|
324
|
+
|
325
|
+
rule regex_body
|
326
|
+
(nested_parens_in_regex / [^)])*
|
327
|
+
end
|
328
|
+
|
329
|
+
rule nested_parens_in_regex
|
330
|
+
'(' regex_body ')'
|
331
|
+
end
|
332
|
+
|
333
|
+
rule parenthesized_expression
|
334
|
+
'(' space? parsing_expression space? ')' <ParenthesizedExpression> {
|
335
|
+
def inline_modules
|
336
|
+
parsing_expression.inline_modules
|
337
|
+
end
|
338
|
+
}
|
339
|
+
end
|
340
|
+
|
341
|
+
rule nonterminal
|
342
|
+
!keyword_inside_grammar (alpha_char alphanumeric_char*) <Nonterminal>
|
343
|
+
end
|
344
|
+
|
345
|
+
rule terminal
|
346
|
+
quoted_string / character_class / anything_symbol
|
347
|
+
end
|
348
|
+
|
349
|
+
rule quoted_string
|
350
|
+
(single_quoted_string / double_quoted_string) {
|
351
|
+
def string
|
352
|
+
super.text_value
|
353
|
+
end
|
354
|
+
}
|
355
|
+
end
|
356
|
+
|
357
|
+
rule double_quoted_string
|
358
|
+
'"' string:(!'"' ("\\\\" / '\"' / .))* '"' <Terminal>
|
359
|
+
end
|
360
|
+
|
361
|
+
rule single_quoted_string
|
362
|
+
"'" string:(!"'" ("\\\\" / "\\'" / .))* "'" <Terminal>
|
363
|
+
end
|
364
|
+
|
365
|
+
rule character_class
|
366
|
+
'[' characters:(!']' ('\\' . /!'\\' .))+ ']' <CharacterClass> {
|
367
|
+
def characters
|
368
|
+
super.text_value
|
369
|
+
end
|
370
|
+
}
|
371
|
+
end
|
372
|
+
|
373
|
+
rule anything_symbol
|
374
|
+
'.' <AnythingSymbol>
|
375
|
+
end
|
376
|
+
|
377
|
+
rule node_class_expression
|
378
|
+
space '<' (!'>' .)+ '>' {
|
379
|
+
def node_class_name
|
380
|
+
elements[2].text_value
|
381
|
+
end
|
382
|
+
}
|
383
|
+
/
|
384
|
+
'' {
|
385
|
+
def node_class_name
|
386
|
+
nil
|
387
|
+
end
|
388
|
+
}
|
389
|
+
end
|
390
|
+
|
391
|
+
rule trailing_inline_module
|
392
|
+
space inline_module {
|
393
|
+
def inline_modules
|
394
|
+
[inline_module]
|
395
|
+
end
|
396
|
+
|
397
|
+
def inline_module_name
|
398
|
+
inline_module.module_name
|
399
|
+
end
|
400
|
+
}
|
401
|
+
/
|
402
|
+
'' {
|
403
|
+
def inline_modules
|
404
|
+
[]
|
405
|
+
end
|
406
|
+
|
407
|
+
def inline_module
|
408
|
+
nil
|
409
|
+
end
|
410
|
+
|
411
|
+
def inline_module_name
|
412
|
+
nil
|
413
|
+
end
|
414
|
+
}
|
415
|
+
end
|
416
|
+
|
417
|
+
rule predicate_block
|
418
|
+
'' inline_module <PredicateBlock>
|
419
|
+
end
|
420
|
+
|
421
|
+
rule inline_module
|
422
|
+
'{' (inline_module / ![{}] .)* '}' <InlineModule>
|
423
|
+
end
|
424
|
+
|
425
|
+
rule keyword_inside_grammar
|
426
|
+
('rule' / 'end') !non_space_char
|
427
|
+
end
|
428
|
+
|
429
|
+
rule non_space_char
|
430
|
+
!space .
|
431
|
+
end
|
432
|
+
|
433
|
+
rule alpha_char
|
434
|
+
[A-Za-z_]
|
435
|
+
end
|
436
|
+
|
437
|
+
rule alphanumeric_char
|
438
|
+
alpha_char / [0-9]
|
439
|
+
end
|
440
|
+
|
441
|
+
rule space
|
442
|
+
(white / comment_to_eol)+
|
443
|
+
end
|
444
|
+
|
445
|
+
rule comment_to_eol
|
446
|
+
'#' (!"\n" .)*
|
447
|
+
end
|
448
|
+
|
449
|
+
rule white
|
450
|
+
[ \t\n\r]
|
451
|
+
end
|
452
|
+
end
|
453
|
+
end
|
454
|
+
end
|