| OLD | NEW |
| (Empty) |
| 1 # Copyright (C) 2013 Google Inc. All rights reserved. | |
| 2 # | |
| 3 # Redistribution and use in source and binary forms, with or without | |
| 4 # modification, are permitted provided that the following conditions are | |
| 5 # met: | |
| 6 # | |
| 7 # * Redistributions of source code must retain the above copyright | |
| 8 # notice, this list of conditions and the following disclaimer. | |
| 9 # * Redistributions in binary form must reproduce the above | |
| 10 # copyright notice, this list of conditions and the following disclaimer | |
| 11 # in the documentation and/or other materials provided with the | |
| 12 # distribution. | |
| 13 # * Neither the name of Google Inc. nor the names of its | |
| 14 # contributors may be used to endorse or promote products derived from | |
| 15 # this software without specific prior written permission. | |
| 16 # | |
| 17 # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS | |
| 18 # "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT | |
| 19 # LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR | |
| 20 # A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT | |
| 21 # OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, | |
| 22 # SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT | |
| 23 # LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, | |
| 24 # DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY | |
| 25 # THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT | |
| 26 # (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE | |
| 27 # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. | |
| 28 | |
| 29 """Parser for Blink IDL. | |
| 30 | |
| 31 The parser uses the PLY (Python Lex-Yacc) library to build a set of parsing | |
| 32 rules which understand the Blink dialect of Web IDL. | |
| 33 It derives from a standard Web IDL parser, overriding rules where Blink IDL | |
| 34 differs syntactically or semantically from the base parser, or where the base | |
| 35 parser diverges from the Web IDL standard. | |
| 36 | |
| 37 Web IDL: | |
| 38 http://www.w3.org/TR/WebIDL/ | |
| 39 Web IDL Grammar: | |
| 40 http://www.w3.org/TR/WebIDL/#idl-grammar | |
| 41 PLY: | |
| 42 http://www.dabeaz.com/ply/ | |
| 43 | |
| 44 Design doc: | |
| 45 http://www.chromium.org/developers/design-documents/idl-compiler#TOC-Front-end | |
| 46 """ | |
| 47 | |
| 48 # Disable check for line length and Member as Function due to how grammar rules | |
| 49 # are defined with PLY | |
| 50 # | |
| 51 # pylint: disable=R0201 | |
| 52 # pylint: disable=C0301 | |
| 53 # | |
| 54 # Disable attribute validation, as lint can't import parent class to check | |
| 55 # pylint: disable=E1101 | |
| 56 | |
| 57 import os.path | |
| 58 import sys | |
| 59 | |
| 60 # PLY is in Chromium src/third_party/ply | |
| 61 module_path, module_name = os.path.split(__file__) | |
| 62 third_party = os.path.join(module_path, os.pardir, os.pardir, os.pardir, os.pard
ir, 'third_party') | |
| 63 # Insert at front to override system libraries, and after path[0] == script dir | |
| 64 sys.path.insert(1, third_party) | |
| 65 from ply import yacc | |
| 66 | |
| 67 # Base parser is in Chromium src/tools/idl_parser | |
| 68 tools_dir = os.path.join(module_path, os.pardir, os.pardir, os.pardir, os.pardir
, 'tools') | |
| 69 sys.path.append(tools_dir) | |
| 70 from idl_parser.idl_parser import IDLParser, ListFromConcat | |
| 71 from idl_parser.idl_parser import ParseFile as parse_file | |
| 72 | |
| 73 from blink_idl_lexer import BlinkIDLLexer | |
| 74 import blink_idl_lexer | |
| 75 | |
| 76 | |
| 77 # Explicitly set starting symbol to rule defined only in base parser. | |
| 78 # BEWARE that the starting symbol should NOT be defined in both the base parser | |
| 79 # and the derived one, as otherwise which is used depends on which line number | |
| 80 # is lower, which is fragile. Instead, either use one in base parser or | |
| 81 # create a new symbol, so that this is unambiguous. | |
| 82 # FIXME: unfortunately, this doesn't work in PLY 3.4, so need to duplicate the | |
| 83 # rule below. | |
| 84 STARTING_SYMBOL = 'Definitions' | |
| 85 | |
| 86 # We ignore comments (and hence don't need 'Top') but base parser preserves them | |
| 87 # FIXME: Upstream: comments should be removed in base parser | |
| 88 REMOVED_RULES = ['Top', # [0] | |
| 89 'Comments', # [0.1] | |
| 90 'CommentsRest', # [0.2] | |
| 91 ] | |
| 92 | |
| 93 # Remove rules from base class | |
| 94 # FIXME: add a class method upstream: @classmethod IDLParser._RemoveRules | |
| 95 for rule in REMOVED_RULES: | |
| 96 production_name = 'p_' + rule | |
| 97 delattr(IDLParser, production_name) | |
| 98 | |
| 99 | |
| 100 class BlinkIDLParser(IDLParser): | |
| 101 # [1] | |
| 102 # FIXME: Need to duplicate rule for starting symbol here, with line number | |
| 103 # *lower* than in the base parser (idl_parser.py). | |
| 104 # This is a bug in PLY: it determines starting symbol by lowest line number. | |
| 105 # This can be overridden by the 'start' parameter, but as of PLY 3.4 this | |
| 106 # doesn't work correctly. | |
| 107 def p_Definitions(self, p): | |
| 108 """Definitions : ExtendedAttributeList Definition Definitions | |
| 109 | """ | |
| 110 if len(p) > 1: | |
| 111 p[2].AddChildren(p[1]) | |
| 112 p[0] = ListFromConcat(p[2], p[3]) | |
| 113 | |
| 114 # Below are grammar rules used by yacc, given by functions named p_<RULE>. | |
| 115 # * The docstring is the production rule in BNF (grammar). | |
| 116 # * The body is the yacc action (semantics). | |
| 117 # | |
| 118 # The PLY framework builds the actual low-level parser by introspecting this | |
| 119 # parser object, selecting all attributes named p_<RULE> as grammar rules. | |
| 120 # It extracts the docstrings and uses them as the production rules, building | |
| 121 # the table of a LALR parser, and uses the body of the functions as actions. | |
| 122 # | |
| 123 # Reference: | |
| 124 # http://www.dabeaz.com/ply/ply.html#ply_nn23 | |
| 125 # | |
| 126 # Review of yacc: | |
| 127 # Yacc parses a token stream, internally producing a Concrete Syntax Tree | |
| 128 # (CST), where each node corresponds to a production rule in the grammar. | |
| 129 # At each node, it runs an action, which is usually "produce a node in the | |
| 130 # Abstract Syntax Tree (AST)" or "ignore this node" (for nodes in the CST | |
| 131 # that aren't included in the AST, since only needed for parsing). | |
| 132 # | |
| 133 # The rules use pseudo-variables; in PLY syntax: | |
| 134 # p[0] is the left side: assign return value to p[0] instead of returning, | |
| 135 # p[1] ... p[n] are the right side: the values can be accessed, and they | |
| 136 # can be modified. | |
| 137 # (In yacc these are $$ and $1 ... $n.) | |
| 138 # | |
| 139 # The rules can look cryptic at first, but there are a few standard | |
| 140 # transforms from the CST to AST. With these in mind, the actions should | |
| 141 # be reasonably legible. | |
| 142 # | |
| 143 # * Ignore production | |
| 144 # Discard this branch. Primarily used when one alternative is empty. | |
| 145 # | |
| 146 # Sample code: | |
| 147 # if len(p) > 1: | |
| 148 # p[0] = ... | |
| 149 # # Note no assignment if len(p) == 1 | |
| 150 # | |
| 151 # * Eliminate singleton production | |
| 152 # Discard this node in the CST, pass the next level down up the tree. | |
| 153 # Used to ignore productions only necessary for parsing, but not needed | |
| 154 # in the AST. | |
| 155 # | |
| 156 # Sample code: | |
| 157 # p[0] = p[1] | |
| 158 # | |
| 159 # * Build node | |
| 160 # The key type of rule. In this parser, produces object of class IDLNode. | |
| 161 # There are several helper functions: | |
| 162 # * BuildProduction: actually builds an IDLNode, based on a production. | |
| 163 # * BuildAttribute: builds an IDLAttribute, which is a temporary | |
| 164 # object to hold a name-value pair, which is then | |
| 165 # set as a Property of the IDLNode when the IDLNode | |
| 166 # is built. | |
| 167 # * BuildNamed: Same as BuildProduction, and sets the 'NAME' property. | |
| 168 # * BuildTrue: BuildAttribute with value True, for flags. | |
| 169 # See base idl_parser.py for definitions and more examples of use. | |
| 170 # | |
| 171 # Sample code: | |
| 172 # # Build node of type NodeType, with value p[1], and children. | |
| 173 # p[0] = self.BuildProduction('NodeType', p, 1, children) | |
| 174 # | |
| 175 # # Build named node of type NodeType, with name and value p[1]. | |
| 176 # # (children optional) | |
| 177 # p[0] = self.BuildNamed('NodeType', p, 1) | |
| 178 # | |
| 179 # # Make a list | |
| 180 # # Used if one node has several children. | |
| 181 # children = ListFromConcat(p[2], p[3]) | |
| 182 # p[0] = self.BuildProduction('NodeType', p, 1, children) | |
| 183 # | |
| 184 # # Also used to collapse the right-associative tree | |
| 185 # # produced by parsing a list back into a single list. | |
| 186 # """Foos : Foo Foos | |
| 187 # |""" | |
| 188 # if len(p) > 1: | |
| 189 # p[0] = ListFromConcat(p[1], p[2]) | |
| 190 # | |
| 191 # # Add children. | |
| 192 # # Primarily used to add attributes, produced via BuildTrue. | |
| 193 # # p_StaticAttribute | |
| 194 # """StaticAttribute : STATIC Attribute""" | |
| 195 # p[2].AddChildren(self.BuildTrue('STATIC')) | |
| 196 # p[0] = p[2] | |
| 197 # | |
| 198 # Numbering scheme for the rules is: | |
| 199 # [1] for Web IDL spec (or additions in base parser) | |
| 200 # These should all be upstreamed to the base parser. | |
| 201 # [b1] for Blink IDL changes (overrides Web IDL) | |
| 202 # [b1.1] for Blink IDL additions, auxiliary rules for [b1] | |
| 203 # Numbers are as per Candidate Recommendation 19 April 2012: | |
| 204 # http://www.w3.org/TR/2012/CR-WebIDL-20120419/ | |
| 205 | |
| 206 # [3] Override action, since we distinguish callbacks | |
| 207 # FIXME: Upstream | |
| 208 def p_CallbackOrInterface(self, p): | |
| 209 """CallbackOrInterface : CALLBACK CallbackRestOrInterface | |
| 210 | Interface""" | |
| 211 if len(p) > 2: | |
| 212 p[2].AddChildren(self.BuildTrue('CALLBACK')) | |
| 213 p[0] = p[2] | |
| 214 else: | |
| 215 p[0] = p[1] | |
| 216 | |
| 217 # [b27] Add strings, more 'Literal' productions | |
| 218 # 'Literal's needed because integers and strings are both internally strings | |
| 219 def p_ConstValue(self, p): | |
| 220 """ConstValue : BooleanLiteral | |
| 221 | FloatLiteral | |
| 222 | IntegerLiteral | |
| 223 | StringLiteral | |
| 224 | null""" | |
| 225 # Standard is (no 'string', fewer 'Literal's): | |
| 226 # ConstValue : BooleanLiteral | |
| 227 # | FloatLiteral | |
| 228 # | integer | |
| 229 # | NULL | |
| 230 p[0] = p[1] | |
| 231 | |
| 232 # [b27.1] | |
| 233 def p_IntegerLiteral(self, p): | |
| 234 """IntegerLiteral : integer""" | |
| 235 p[0] = ListFromConcat(self.BuildAttribute('TYPE', 'integer'), | |
| 236 self.BuildAttribute('NAME', p[1])) | |
| 237 | |
| 238 # [b27.2] | |
| 239 def p_StringLiteral(self, p): | |
| 240 """StringLiteral : string""" | |
| 241 p[0] = ListFromConcat(self.BuildAttribute('TYPE', 'DOMString'), | |
| 242 self.BuildAttribute('NAME', p[1])) | |
| 243 | |
| 244 # [b47] | |
| 245 def p_ExceptionMember(self, p): | |
| 246 """ExceptionMember : Const | |
| 247 | ExceptionField | |
| 248 | Attribute | |
| 249 | ExceptionOperation""" | |
| 250 # Standard is (no Attribute, no ExceptionOperation): | |
| 251 # ExceptionMember : Const | |
| 252 # | ExceptionField | |
| 253 # FIXME: In DOMException.idl, Attributes should be changed to | |
| 254 # ExceptionFields, and Attribute removed from this rule. | |
| 255 p[0] = p[1] | |
| 256 | |
| 257 # [b47.1] FIXME: rename to ExceptionAttribute | |
| 258 def p_Attribute(self, p): | |
| 259 """Attribute : ReadOnly ATTRIBUTE Type identifier ';'""" | |
| 260 p[0] = self.BuildNamed('Attribute', p, 4, | |
| 261 ListFromConcat(p[1], p[3])) | |
| 262 | |
| 263 # [b47.2] | |
| 264 def p_ExceptionOperation(self, p): | |
| 265 """ExceptionOperation : Type identifier '(' ')' ';'""" | |
| 266 # Needed to handle one case in DOMException.idl: | |
| 267 # // Override in a Mozilla compatible format | |
| 268 # [NotEnumerable] DOMString toString(); | |
| 269 # Limited form of Operation to prevent others from being added. | |
| 270 # FIXME: Should be a stringifier instead. | |
| 271 p[0] = self.BuildNamed('ExceptionOperation', p, 2, p[1]) | |
| 272 | |
| 273 # Extended attributes | |
| 274 # [b49] Override base parser: remove comment field, since comments stripped | |
| 275 # FIXME: Upstream | |
| 276 def p_ExtendedAttributeList(self, p): | |
| 277 """ExtendedAttributeList : '[' ExtendedAttribute ExtendedAttributes ']' | |
| 278 | '[' ']' | |
| 279 | """ | |
| 280 if len(p) > 3: | |
| 281 items = ListFromConcat(p[2], p[3]) | |
| 282 p[0] = self.BuildProduction('ExtAttributes', p, 1, items) | |
| 283 | |
| 284 # Error handling for ExtendedAttributeList. | |
| 285 # We can't upstream this because we override ExtendedAttributeList. | |
| 286 def p_ExtendedAttributeListError(self, p): | |
| 287 """ExtendedAttributeList : '[' ExtendedAttribute ',' error""" | |
| 288 p[0] = self.BuildError(p, "ExtendedAttributeList") | |
| 289 | |
| 290 # [b50] Allow optional trailing comma | |
| 291 # Blink-only, marked as WONTFIX in Web IDL spec: | |
| 292 # https://www.w3.org/Bugs/Public/show_bug.cgi?id=22156 | |
| 293 def p_ExtendedAttributes(self, p): | |
| 294 """ExtendedAttributes : ',' ExtendedAttribute ExtendedAttributes | |
| 295 | ',' | |
| 296 |""" | |
| 297 if len(p) > 3: | |
| 298 p[0] = ListFromConcat(p[2], p[3]) | |
| 299 | |
| 300 # [b51] Add ExtendedAttributeStringLiteral and ExtendedAttributeStringLitera
lList | |
| 301 def p_ExtendedAttribute(self, p): | |
| 302 """ExtendedAttribute : ExtendedAttributeNoArgs | |
| 303 | ExtendedAttributeArgList | |
| 304 | ExtendedAttributeIdent | |
| 305 | ExtendedAttributeIdentList | |
| 306 | ExtendedAttributeNamedArgList | |
| 307 | ExtendedAttributeStringLiteral | |
| 308 | ExtendedAttributeStringLiteralList""" | |
| 309 p[0] = p[1] | |
| 310 | |
| 311 # [59] | |
| 312 # FIXME: Upstream UnionType | |
| 313 def p_UnionType(self, p): | |
| 314 """UnionType : '(' UnionMemberType OR UnionMemberType UnionMemberTypes '
)'""" | |
| 315 members = ListFromConcat(p[2], p[4], p[5]) | |
| 316 p[0] = self.BuildProduction('UnionType', p, 1, members) | |
| 317 | |
| 318 # [60] | |
| 319 def p_UnionMemberType(self, p): | |
| 320 """UnionMemberType : NonAnyType | |
| 321 | UnionType TypeSuffix | |
| 322 | ANY '[' ']' TypeSuffix""" | |
| 323 if len(p) == 2: | |
| 324 p[0] = self.BuildProduction('Type', p, 1, p[1]) | |
| 325 elif len(p) == 3: | |
| 326 p[0] = self.BuildProduction('Type', p, 1, ListFromConcat(p[1], p[2])
) | |
| 327 else: | |
| 328 any_node = ListFromConcat(self.BuildProduction('Any', p, 1), p[4]) | |
| 329 p[0] = self.BuildProduction('Type', p, 1, any_node) | |
| 330 | |
| 331 # [61] | |
| 332 def p_UnionMemberTypes(self, p): | |
| 333 """UnionMemberTypes : OR UnionMemberType UnionMemberTypes | |
| 334 |""" | |
| 335 if len(p) > 2: | |
| 336 p[0] = ListFromConcat(p[2], p[3]) | |
| 337 | |
| 338 # [70] Override base parser to remove non-standard sized array | |
| 339 # FIXME: Upstream | |
| 340 def p_TypeSuffix(self, p): | |
| 341 """TypeSuffix : '[' ']' TypeSuffix | |
| 342 | '?' TypeSuffixStartingWithArray | |
| 343 |""" | |
| 344 if len(p) == 4: | |
| 345 p[0] = self.BuildProduction('Array', p, 1, p[3]) | |
| 346 elif len(p) == 3: | |
| 347 p[0] = ListFromConcat(self.BuildTrue('NULLABLE'), p[2]) | |
| 348 | |
| 349 # Blink extension: Add support for string literal Extended Attribute values | |
| 350 def p_ExtendedAttributeStringLiteral(self, p): | |
| 351 """ExtendedAttributeStringLiteral : identifier '=' StringLiteral """ | |
| 352 def unwrap_string(ls): | |
| 353 """Reach in and grab the string literal's "NAME".""" | |
| 354 return ls[1].value | |
| 355 | |
| 356 value = self.BuildAttribute('VALUE', unwrap_string(p[3])) | |
| 357 p[0] = self.BuildNamed('ExtAttribute', p, 1, value) | |
| 358 | |
| 359 # Blink extension: Add support for compound Extended Attribute values over s
tring literals ("A","B") | |
| 360 def p_ExtendedAttributeStringLiteralList(self, p): | |
| 361 """ExtendedAttributeStringLiteralList : identifier '=' '(' StringLiteral
List ')' """ | |
| 362 value = self.BuildAttribute('VALUE', p[4]) | |
| 363 p[0] = self.BuildNamed('ExtAttribute', p, 1, value) | |
| 364 | |
| 365 # Blink extension: one or more string literals. The values aren't propagated
as literals, | |
| 366 # but their by their value only. | |
| 367 def p_StringLiteralList(self, p): | |
| 368 """StringLiteralList : StringLiteral ',' StringLiteralList | |
| 369 | StringLiteral""" | |
| 370 def unwrap_string(ls): | |
| 371 """Reach in and grab the string literal's "NAME".""" | |
| 372 return ls[1].value | |
| 373 | |
| 374 if len(p) > 3: | |
| 375 p[0] = ListFromConcat(unwrap_string(p[1]), p[3]) | |
| 376 else: | |
| 377 p[0] = ListFromConcat(unwrap_string(p[1])) | |
| 378 | |
| 379 def __init__(self, | |
| 380 # common parameters | |
| 381 debug=False, | |
| 382 # local parameters | |
| 383 rewrite_tables=False, | |
| 384 # idl_parser parameters | |
| 385 lexer=None, verbose=False, mute_error=False, | |
| 386 # yacc parameters | |
| 387 outputdir='', optimize=True, write_tables=False, | |
| 388 picklefile=None): | |
| 389 if debug: | |
| 390 # Turn off optimization and caching, and write out tables, | |
| 391 # to help debugging | |
| 392 optimize = False | |
| 393 outputdir = None | |
| 394 picklefile = None | |
| 395 write_tables = True | |
| 396 if outputdir: | |
| 397 picklefile = picklefile or os.path.join(outputdir, 'parsetab.pickle'
) | |
| 398 if rewrite_tables: | |
| 399 try: | |
| 400 os.unlink(picklefile) | |
| 401 except OSError: | |
| 402 pass | |
| 403 | |
| 404 lexer = lexer or BlinkIDLLexer(debug=debug, | |
| 405 outputdir=outputdir, | |
| 406 optimize=optimize) | |
| 407 self.lexer = lexer | |
| 408 self.tokens = lexer.KnownTokens() | |
| 409 # Using SLR (instead of LALR) generates the table faster, | |
| 410 # but produces the same output. This is ok b/c Web IDL (and Blink IDL) | |
| 411 # is an SLR grammar (as is often the case for simple LL(1) grammars). | |
| 412 # | |
| 413 # Optimized mode substantially decreases startup time (by disabling | |
| 414 # error checking), and also allows use of Python's optimized mode. | |
| 415 # See: Using Python's Optimized Mode | |
| 416 # http://www.dabeaz.com/ply/ply.html#ply_nn38 | |
| 417 # | |
| 418 # |picklefile| allows simpler importing than |tabmodule| (parsetab.py), | |
| 419 # as we don't need to modify sys.path; virtually identical speed. | |
| 420 # See: CHANGES, Version 3.2 | |
| 421 # http://ply.googlecode.com/svn/trunk/CHANGES | |
| 422 self.yaccobj = yacc.yacc(module=self, | |
| 423 start=STARTING_SYMBOL, | |
| 424 method='SLR', | |
| 425 debug=debug, | |
| 426 optimize=optimize, | |
| 427 write_tables=write_tables, | |
| 428 picklefile=picklefile) | |
| 429 self.parse_debug = debug | |
| 430 self.verbose = verbose | |
| 431 self.mute_error = mute_error | |
| 432 self._parse_errors = 0 | |
| 433 self._parse_warnings = 0 | |
| 434 self._last_error_msg = None | |
| 435 self._last_error_lineno = 0 | |
| 436 self._last_error_pos = 0 | |
| 437 | |
| 438 | |
| 439 ################################################################################ | |
| 440 | |
| 441 def main(argv): | |
| 442 # If file itself executed, cache lex/parse tables | |
| 443 try: | |
| 444 outputdir = argv[1] | |
| 445 except IndexError as err: | |
| 446 print 'Usage: %s OUTPUT_DIR' % argv[0] | |
| 447 return 1 | |
| 448 blink_idl_lexer.main(argv) | |
| 449 # Important: rewrite_tables=True causes the cache file to be deleted if it | |
| 450 # exists, thus making sure that PLY doesn't load it instead of regenerating | |
| 451 # the parse table. | |
| 452 parser = BlinkIDLParser(outputdir=outputdir, rewrite_tables=True) | |
| 453 | |
| 454 | |
| 455 if __name__ == '__main__': | |
| 456 sys.exit(main(sys.argv)) | |
| OLD | NEW |