1 | n/a | #------------------------------------------------------------------------------- |
---|
2 | n/a | # Parser for ASDL [1] definition files. Reads in an ASDL description and parses |
---|
3 | n/a | # it into an AST that describes it. |
---|
4 | n/a | # |
---|
5 | n/a | # The EBNF we're parsing here: Figure 1 of the paper [1]. Extended to support |
---|
6 | n/a | # modules and attributes after a product. Words starting with Capital letters |
---|
7 | n/a | # are terminals. Literal tokens are in "double quotes". Others are |
---|
8 | n/a | # non-terminals. Id is either TokenId or ConstructorId. |
---|
9 | n/a | # |
---|
10 | n/a | # module ::= "module" Id "{" [definitions] "}" |
---|
11 | n/a | # definitions ::= { TypeId "=" type } |
---|
12 | n/a | # type ::= product | sum |
---|
13 | n/a | # product ::= fields ["attributes" fields] |
---|
14 | n/a | # fields ::= "(" { field, "," } field ")" |
---|
15 | n/a | # field ::= TypeId ["?" | "*"] [Id] |
---|
16 | n/a | # sum ::= constructor { "|" constructor } ["attributes" fields] |
---|
17 | n/a | # constructor ::= ConstructorId [fields] |
---|
18 | n/a | # |
---|
19 | n/a | # [1] "The Zephyr Abstract Syntax Description Language" by Wang, et. al. See |
---|
20 | n/a | # http://asdl.sourceforge.net/ |
---|
21 | n/a | #------------------------------------------------------------------------------- |
---|
22 | n/a | from collections import namedtuple |
---|
23 | n/a | import re |
---|
24 | n/a | |
---|
25 | n/a | __all__ = [ |
---|
26 | n/a | 'builtin_types', 'parse', 'AST', 'Module', 'Type', 'Constructor', |
---|
27 | n/a | 'Field', 'Sum', 'Product', 'VisitorBase', 'Check', 'check'] |
---|
28 | n/a | |
---|
29 | n/a | # The following classes define nodes into which the ASDL description is parsed. |
---|
30 | n/a | # Note: this is a "meta-AST". ASDL files (such as Python.asdl) describe the AST |
---|
31 | n/a | # structure used by a programming language. But ASDL files themselves need to be |
---|
32 | n/a | # parsed. This module parses ASDL files and uses a simple AST to represent them. |
---|
33 | n/a | # See the EBNF at the top of the file to understand the logical connection |
---|
34 | n/a | # between the various node types. |
---|
35 | n/a | |
---|
36 | n/a | builtin_types = {'identifier', 'string', 'bytes', 'int', 'object', 'singleton', |
---|
37 | n/a | 'constant'} |
---|
38 | n/a | |
---|
39 | n/a | class AST: |
---|
40 | n/a | def __repr__(self): |
---|
41 | n/a | raise NotImplementedError |
---|
42 | n/a | |
---|
43 | n/a | class Module(AST): |
---|
44 | n/a | def __init__(self, name, dfns): |
---|
45 | n/a | self.name = name |
---|
46 | n/a | self.dfns = dfns |
---|
47 | n/a | self.types = {type.name: type.value for type in dfns} |
---|
48 | n/a | |
---|
49 | n/a | def __repr__(self): |
---|
50 | n/a | return 'Module({0.name}, {0.dfns})'.format(self) |
---|
51 | n/a | |
---|
52 | n/a | class Type(AST): |
---|
53 | n/a | def __init__(self, name, value): |
---|
54 | n/a | self.name = name |
---|
55 | n/a | self.value = value |
---|
56 | n/a | |
---|
57 | n/a | def __repr__(self): |
---|
58 | n/a | return 'Type({0.name}, {0.value})'.format(self) |
---|
59 | n/a | |
---|
60 | n/a | class Constructor(AST): |
---|
61 | n/a | def __init__(self, name, fields=None): |
---|
62 | n/a | self.name = name |
---|
63 | n/a | self.fields = fields or [] |
---|
64 | n/a | |
---|
65 | n/a | def __repr__(self): |
---|
66 | n/a | return 'Constructor({0.name}, {0.fields})'.format(self) |
---|
67 | n/a | |
---|
68 | n/a | class Field(AST): |
---|
69 | n/a | def __init__(self, type, name=None, seq=False, opt=False): |
---|
70 | n/a | self.type = type |
---|
71 | n/a | self.name = name |
---|
72 | n/a | self.seq = seq |
---|
73 | n/a | self.opt = opt |
---|
74 | n/a | |
---|
75 | n/a | def __repr__(self): |
---|
76 | n/a | if self.seq: |
---|
77 | n/a | extra = ", seq=True" |
---|
78 | n/a | elif self.opt: |
---|
79 | n/a | extra = ", opt=True" |
---|
80 | n/a | else: |
---|
81 | n/a | extra = "" |
---|
82 | n/a | if self.name is None: |
---|
83 | n/a | return 'Field({0.type}{1})'.format(self, extra) |
---|
84 | n/a | else: |
---|
85 | n/a | return 'Field({0.type}, {0.name}{1})'.format(self, extra) |
---|
86 | n/a | |
---|
87 | n/a | class Sum(AST): |
---|
88 | n/a | def __init__(self, types, attributes=None): |
---|
89 | n/a | self.types = types |
---|
90 | n/a | self.attributes = attributes or [] |
---|
91 | n/a | |
---|
92 | n/a | def __repr__(self): |
---|
93 | n/a | if self.attributes: |
---|
94 | n/a | return 'Sum({0.types}, {0.attributes})'.format(self) |
---|
95 | n/a | else: |
---|
96 | n/a | return 'Sum({0.types})'.format(self) |
---|
97 | n/a | |
---|
98 | n/a | class Product(AST): |
---|
99 | n/a | def __init__(self, fields, attributes=None): |
---|
100 | n/a | self.fields = fields |
---|
101 | n/a | self.attributes = attributes or [] |
---|
102 | n/a | |
---|
103 | n/a | def __repr__(self): |
---|
104 | n/a | if self.attributes: |
---|
105 | n/a | return 'Product({0.fields}, {0.attributes})'.format(self) |
---|
106 | n/a | else: |
---|
107 | n/a | return 'Product({0.fields})'.format(self) |
---|
108 | n/a | |
---|
109 | n/a | # A generic visitor for the meta-AST that describes ASDL. This can be used by |
---|
110 | n/a | # emitters. Note that this visitor does not provide a generic visit method, so a |
---|
111 | n/a | # subclass needs to define visit methods from visitModule to as deep as the |
---|
112 | n/a | # interesting node. |
---|
113 | n/a | # We also define a Check visitor that makes sure the parsed ASDL is well-formed. |
---|
114 | n/a | |
---|
115 | n/a | class VisitorBase(object): |
---|
116 | n/a | """Generic tree visitor for ASTs.""" |
---|
117 | n/a | def __init__(self): |
---|
118 | n/a | self.cache = {} |
---|
119 | n/a | |
---|
120 | n/a | def visit(self, obj, *args): |
---|
121 | n/a | klass = obj.__class__ |
---|
122 | n/a | meth = self.cache.get(klass) |
---|
123 | n/a | if meth is None: |
---|
124 | n/a | methname = "visit" + klass.__name__ |
---|
125 | n/a | meth = getattr(self, methname, None) |
---|
126 | n/a | self.cache[klass] = meth |
---|
127 | n/a | if meth: |
---|
128 | n/a | try: |
---|
129 | n/a | meth(obj, *args) |
---|
130 | n/a | except Exception as e: |
---|
131 | n/a | print("Error visiting %r: %s" % (obj, e)) |
---|
132 | n/a | raise |
---|
133 | n/a | |
---|
134 | n/a | class Check(VisitorBase): |
---|
135 | n/a | """A visitor that checks a parsed ASDL tree for correctness. |
---|
136 | n/a | |
---|
137 | n/a | Errors are printed and accumulated. |
---|
138 | n/a | """ |
---|
139 | n/a | def __init__(self): |
---|
140 | n/a | super(Check, self).__init__() |
---|
141 | n/a | self.cons = {} |
---|
142 | n/a | self.errors = 0 |
---|
143 | n/a | self.types = {} |
---|
144 | n/a | |
---|
145 | n/a | def visitModule(self, mod): |
---|
146 | n/a | for dfn in mod.dfns: |
---|
147 | n/a | self.visit(dfn) |
---|
148 | n/a | |
---|
149 | n/a | def visitType(self, type): |
---|
150 | n/a | self.visit(type.value, str(type.name)) |
---|
151 | n/a | |
---|
152 | n/a | def visitSum(self, sum, name): |
---|
153 | n/a | for t in sum.types: |
---|
154 | n/a | self.visit(t, name) |
---|
155 | n/a | |
---|
156 | n/a | def visitConstructor(self, cons, name): |
---|
157 | n/a | key = str(cons.name) |
---|
158 | n/a | conflict = self.cons.get(key) |
---|
159 | n/a | if conflict is None: |
---|
160 | n/a | self.cons[key] = name |
---|
161 | n/a | else: |
---|
162 | n/a | print('Redefinition of constructor {}'.format(key)) |
---|
163 | n/a | print('Defined in {} and {}'.format(conflict, name)) |
---|
164 | n/a | self.errors += 1 |
---|
165 | n/a | for f in cons.fields: |
---|
166 | n/a | self.visit(f, key) |
---|
167 | n/a | |
---|
168 | n/a | def visitField(self, field, name): |
---|
169 | n/a | key = str(field.type) |
---|
170 | n/a | l = self.types.setdefault(key, []) |
---|
171 | n/a | l.append(name) |
---|
172 | n/a | |
---|
173 | n/a | def visitProduct(self, prod, name): |
---|
174 | n/a | for f in prod.fields: |
---|
175 | n/a | self.visit(f, name) |
---|
176 | n/a | |
---|
177 | n/a | def check(mod): |
---|
178 | n/a | """Check the parsed ASDL tree for correctness. |
---|
179 | n/a | |
---|
180 | n/a | Return True if success. For failure, the errors are printed out and False |
---|
181 | n/a | is returned. |
---|
182 | n/a | """ |
---|
183 | n/a | v = Check() |
---|
184 | n/a | v.visit(mod) |
---|
185 | n/a | |
---|
186 | n/a | for t in v.types: |
---|
187 | n/a | if t not in mod.types and not t in builtin_types: |
---|
188 | n/a | v.errors += 1 |
---|
189 | n/a | uses = ", ".join(v.types[t]) |
---|
190 | n/a | print('Undefined type {}, used in {}'.format(t, uses)) |
---|
191 | n/a | return not v.errors |
---|
192 | n/a | |
---|
193 | n/a | # The ASDL parser itself comes next. The only interesting external interface |
---|
194 | n/a | # here is the top-level parse function. |
---|
195 | n/a | |
---|
196 | n/a | def parse(filename): |
---|
197 | n/a | """Parse ASDL from the given file and return a Module node describing it.""" |
---|
198 | n/a | with open(filename) as f: |
---|
199 | n/a | parser = ASDLParser() |
---|
200 | n/a | return parser.parse(f.read()) |
---|
201 | n/a | |
---|
202 | n/a | # Types for describing tokens in an ASDL specification. |
---|
203 | n/a | class TokenKind: |
---|
204 | n/a | """TokenKind is provides a scope for enumerated token kinds.""" |
---|
205 | n/a | (ConstructorId, TypeId, Equals, Comma, Question, Pipe, Asterisk, |
---|
206 | n/a | LParen, RParen, LBrace, RBrace) = range(11) |
---|
207 | n/a | |
---|
208 | n/a | operator_table = { |
---|
209 | n/a | '=': Equals, ',': Comma, '?': Question, '|': Pipe, '(': LParen, |
---|
210 | n/a | ')': RParen, '*': Asterisk, '{': LBrace, '}': RBrace} |
---|
211 | n/a | |
---|
212 | n/a | Token = namedtuple('Token', 'kind value lineno') |
---|
213 | n/a | |
---|
214 | n/a | class ASDLSyntaxError(Exception): |
---|
215 | n/a | def __init__(self, msg, lineno=None): |
---|
216 | n/a | self.msg = msg |
---|
217 | n/a | self.lineno = lineno or '<unknown>' |
---|
218 | n/a | |
---|
219 | n/a | def __str__(self): |
---|
220 | n/a | return 'Syntax error on line {0.lineno}: {0.msg}'.format(self) |
---|
221 | n/a | |
---|
222 | n/a | def tokenize_asdl(buf): |
---|
223 | n/a | """Tokenize the given buffer. Yield Token objects.""" |
---|
224 | n/a | for lineno, line in enumerate(buf.splitlines(), 1): |
---|
225 | n/a | for m in re.finditer(r'\s*(\w+|--.*|.)', line.strip()): |
---|
226 | n/a | c = m.group(1) |
---|
227 | n/a | if c[0].isalpha(): |
---|
228 | n/a | # Some kind of identifier |
---|
229 | n/a | if c[0].isupper(): |
---|
230 | n/a | yield Token(TokenKind.ConstructorId, c, lineno) |
---|
231 | n/a | else: |
---|
232 | n/a | yield Token(TokenKind.TypeId, c, lineno) |
---|
233 | n/a | elif c[:2] == '--': |
---|
234 | n/a | # Comment |
---|
235 | n/a | break |
---|
236 | n/a | else: |
---|
237 | n/a | # Operators |
---|
238 | n/a | try: |
---|
239 | n/a | op_kind = TokenKind.operator_table[c] |
---|
240 | n/a | except KeyError: |
---|
241 | n/a | raise ASDLSyntaxError('Invalid operator %s' % c, lineno) |
---|
242 | n/a | yield Token(op_kind, c, lineno) |
---|
243 | n/a | |
---|
244 | n/a | class ASDLParser: |
---|
245 | n/a | """Parser for ASDL files. |
---|
246 | n/a | |
---|
247 | n/a | Create, then call the parse method on a buffer containing ASDL. |
---|
248 | n/a | This is a simple recursive descent parser that uses tokenize_asdl for the |
---|
249 | n/a | lexing. |
---|
250 | n/a | """ |
---|
251 | n/a | def __init__(self): |
---|
252 | n/a | self._tokenizer = None |
---|
253 | n/a | self.cur_token = None |
---|
254 | n/a | |
---|
255 | n/a | def parse(self, buf): |
---|
256 | n/a | """Parse the ASDL in the buffer and return an AST with a Module root. |
---|
257 | n/a | """ |
---|
258 | n/a | self._tokenizer = tokenize_asdl(buf) |
---|
259 | n/a | self._advance() |
---|
260 | n/a | return self._parse_module() |
---|
261 | n/a | |
---|
262 | n/a | def _parse_module(self): |
---|
263 | n/a | if self._at_keyword('module'): |
---|
264 | n/a | self._advance() |
---|
265 | n/a | else: |
---|
266 | n/a | raise ASDLSyntaxError( |
---|
267 | n/a | 'Expected "module" (found {})'.format(self.cur_token.value), |
---|
268 | n/a | self.cur_token.lineno) |
---|
269 | n/a | name = self._match(self._id_kinds) |
---|
270 | n/a | self._match(TokenKind.LBrace) |
---|
271 | n/a | defs = self._parse_definitions() |
---|
272 | n/a | self._match(TokenKind.RBrace) |
---|
273 | n/a | return Module(name, defs) |
---|
274 | n/a | |
---|
275 | n/a | def _parse_definitions(self): |
---|
276 | n/a | defs = [] |
---|
277 | n/a | while self.cur_token.kind == TokenKind.TypeId: |
---|
278 | n/a | typename = self._advance() |
---|
279 | n/a | self._match(TokenKind.Equals) |
---|
280 | n/a | type = self._parse_type() |
---|
281 | n/a | defs.append(Type(typename, type)) |
---|
282 | n/a | return defs |
---|
283 | n/a | |
---|
284 | n/a | def _parse_type(self): |
---|
285 | n/a | if self.cur_token.kind == TokenKind.LParen: |
---|
286 | n/a | # If we see a (, it's a product |
---|
287 | n/a | return self._parse_product() |
---|
288 | n/a | else: |
---|
289 | n/a | # Otherwise it's a sum. Look for ConstructorId |
---|
290 | n/a | sumlist = [Constructor(self._match(TokenKind.ConstructorId), |
---|
291 | n/a | self._parse_optional_fields())] |
---|
292 | n/a | while self.cur_token.kind == TokenKind.Pipe: |
---|
293 | n/a | # More constructors |
---|
294 | n/a | self._advance() |
---|
295 | n/a | sumlist.append(Constructor( |
---|
296 | n/a | self._match(TokenKind.ConstructorId), |
---|
297 | n/a | self._parse_optional_fields())) |
---|
298 | n/a | return Sum(sumlist, self._parse_optional_attributes()) |
---|
299 | n/a | |
---|
300 | n/a | def _parse_product(self): |
---|
301 | n/a | return Product(self._parse_fields(), self._parse_optional_attributes()) |
---|
302 | n/a | |
---|
303 | n/a | def _parse_fields(self): |
---|
304 | n/a | fields = [] |
---|
305 | n/a | self._match(TokenKind.LParen) |
---|
306 | n/a | while self.cur_token.kind == TokenKind.TypeId: |
---|
307 | n/a | typename = self._advance() |
---|
308 | n/a | is_seq, is_opt = self._parse_optional_field_quantifier() |
---|
309 | n/a | id = (self._advance() if self.cur_token.kind in self._id_kinds |
---|
310 | n/a | else None) |
---|
311 | n/a | fields.append(Field(typename, id, seq=is_seq, opt=is_opt)) |
---|
312 | n/a | if self.cur_token.kind == TokenKind.RParen: |
---|
313 | n/a | break |
---|
314 | n/a | elif self.cur_token.kind == TokenKind.Comma: |
---|
315 | n/a | self._advance() |
---|
316 | n/a | self._match(TokenKind.RParen) |
---|
317 | n/a | return fields |
---|
318 | n/a | |
---|
319 | n/a | def _parse_optional_fields(self): |
---|
320 | n/a | if self.cur_token.kind == TokenKind.LParen: |
---|
321 | n/a | return self._parse_fields() |
---|
322 | n/a | else: |
---|
323 | n/a | return None |
---|
324 | n/a | |
---|
325 | n/a | def _parse_optional_attributes(self): |
---|
326 | n/a | if self._at_keyword('attributes'): |
---|
327 | n/a | self._advance() |
---|
328 | n/a | return self._parse_fields() |
---|
329 | n/a | else: |
---|
330 | n/a | return None |
---|
331 | n/a | |
---|
332 | n/a | def _parse_optional_field_quantifier(self): |
---|
333 | n/a | is_seq, is_opt = False, False |
---|
334 | n/a | if self.cur_token.kind == TokenKind.Asterisk: |
---|
335 | n/a | is_seq = True |
---|
336 | n/a | self._advance() |
---|
337 | n/a | elif self.cur_token.kind == TokenKind.Question: |
---|
338 | n/a | is_opt = True |
---|
339 | n/a | self._advance() |
---|
340 | n/a | return is_seq, is_opt |
---|
341 | n/a | |
---|
342 | n/a | def _advance(self): |
---|
343 | n/a | """ Return the value of the current token and read the next one into |
---|
344 | n/a | self.cur_token. |
---|
345 | n/a | """ |
---|
346 | n/a | cur_val = None if self.cur_token is None else self.cur_token.value |
---|
347 | n/a | try: |
---|
348 | n/a | self.cur_token = next(self._tokenizer) |
---|
349 | n/a | except StopIteration: |
---|
350 | n/a | self.cur_token = None |
---|
351 | n/a | return cur_val |
---|
352 | n/a | |
---|
353 | n/a | _id_kinds = (TokenKind.ConstructorId, TokenKind.TypeId) |
---|
354 | n/a | |
---|
355 | n/a | def _match(self, kind): |
---|
356 | n/a | """The 'match' primitive of RD parsers. |
---|
357 | n/a | |
---|
358 | n/a | * Verifies that the current token is of the given kind (kind can |
---|
359 | n/a | be a tuple, in which the kind must match one of its members). |
---|
360 | n/a | * Returns the value of the current token |
---|
361 | n/a | * Reads in the next token |
---|
362 | n/a | """ |
---|
363 | n/a | if (isinstance(kind, tuple) and self.cur_token.kind in kind or |
---|
364 | n/a | self.cur_token.kind == kind |
---|
365 | n/a | ): |
---|
366 | n/a | value = self.cur_token.value |
---|
367 | n/a | self._advance() |
---|
368 | n/a | return value |
---|
369 | n/a | else: |
---|
370 | n/a | raise ASDLSyntaxError( |
---|
371 | n/a | 'Unmatched {} (found {})'.format(kind, self.cur_token.kind), |
---|
372 | n/a | self.cur_token.lineno) |
---|
373 | n/a | |
---|
374 | n/a | def _at_keyword(self, keyword): |
---|
375 | n/a | return (self.cur_token.kind == TokenKind.TypeId and |
---|
376 | n/a | self.cur_token.value == keyword) |
---|