ebnf.py 24.5 KB
Newer Older
1
2
#!/usr/bin/python3

3
"""ebnf.py - EBNF -> Python-Parser compilation for DHParser
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

Copyright 2016  by Eckhart Arnold (arnold@badw.de)
                Bavarian Academy of Sciences an Humanities (badw.de)

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
implied.  See the License for the specific language governing
permissions and limitations under the License.
"""

Eckhart Arnold's avatar
Eckhart Arnold committed
21
# import collections
22
23
import keyword
from functools import partial
24

25
26
27
28
29
try:
    import regex as re
except ImportError:
    import re

30
31
from .__init__ import __version__
from .toolkit import load_if_file, escape_re, md5, sane_parser_name
32
from .parsers import GrammarBase, mixin_comment, Forward, RE, NegativeLookahead, \
33
34
    Alternative, Sequence, Optional, Required, OneOrMore, ZeroOrMore, Token, CompilerBase, \
    Capture, Retrieve
35
from .syntaxtree import Node, traverse, remove_enclosing_delimiters, reduce_single_child, \
di68kap's avatar
di68kap committed
36
    replace_by_single_child, TOKEN_KEYWORD, remove_expendables, remove_tokens, flatten, \
37
    forbid, assert_content, WHITESPACE_KEYWORD
38
39


Eckhart Arnold's avatar
Eckhart Arnold committed
40
41
42
43
44
__all__ = ['EBNFGrammar',
           'EBNFTransTable',
           'EBNFCompilerError',
           # 'Scanner',
           'EBNFCompiler']
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65


class EBNFGrammar(GrammarBase):
    r"""Parser for an EBNF source file, with this grammar:

    # EBNF-Grammar in EBNF

    @ comment    =  /#.*(?:\n|$)/                    # comments start with '#' and eat all chars up to and including '\n'
    @ whitespace =  /\s*/                            # whitespace includes linefeed
    @ literalws  =  right                            # trailing whitespace of literals will be ignored tacitly

    syntax     =  [~//] { definition | directive } §EOF
    definition =  symbol §"=" expression
    directive  =  "@" §symbol §"=" ( regexp | literal | list_ )

    expression =  term { "|" term }
    term       =  { factor }+
    factor     =  [flowmarker] [retrieveop] symbol !"="   # negative lookahead to be sure it's not a definition
                | [flowmarker] literal
                | [flowmarker] regexp
                | [flowmarker] group
66
                | [flowmarker] regexchain
67
68
69
70
71
72
73
74
75
                | [flowmarker] oneormore
                | repetition
                | option

    flowmarker =  "!"  | "&"  | "§" |                # '!' negative lookahead, '&' positive lookahead, '§' required
                  "-!" | "-&"                        # '-' negative lookbehind, '-&' positive lookbehind
    retrieveop =  "::" | ":"                         # '::' pop, ':' retrieve

    group      =  "(" expression §")"
76
    regexchain =  "<" expression §">"                # compiles "expression" into a singular regular expression
77
78
    oneormore  =  "{" expression "}+"
    repetition =  "{" expression §"}"
79
80
81
    option     =  "[" expression §"]"

    link       = regexp | symbol | literal           # semantic restriction: symbol must evaluate to a regexp or chain
82
83
84
85
86
87
88

    symbol     =  /(?!\d)\w+/~                       # e.g. expression, factor, parameter_list
    literal    =  /"(?:[^"]|\\")*?"/~                # e.g. "(", '+', 'while'
                | /'(?:[^']|\\')*?'/~                # whitespace following literals will be ignored tacitly.
    regexp     =  /~?\/(?:[^\/]|(?<=\\)\/)*\/~?/~    # e.g. /\w+/, ~/#.*(?:\n|$)/~
                                                     # '~' is a whitespace-marker, if present leading or trailing
                                                     # whitespace of a regular expression will be ignored tacitly.
89
    list_      =  /\w+/~ { "," /\w+/~ }              # comma separated list of symbols, e.g. BEGIN_LIST, END_LIST,
90
91
92
93
                                                     # BEGIN_QUOTE, END_QUOTE ; see CommonMark/markdown.py for an exmaple
    EOF =  !/./
    """
    expression = Forward()
94
    source_hash__ = "a410e1727fb7575e98ff8451dbf8f3bd"
95
    parser_initialization__ = "upon instatiation"
96
97
    COMMENT__ = r'#.*(?:\n|$)'
    WSP__ = mixin_comment(whitespace=r'\s*', comment=r'#.*(?:\n|$)')
98
    wspL__ = ''
99
    wspR__ = WSP__
100
    EOF = NegativeLookahead(RE('.', wR=''))
101
    list_ = Sequence(RE('\\w+'), ZeroOrMore(Sequence(Token(","), RE('\\w+'))))
102
103
104
    regexp = RE('~?/(?:[^/]|(?<=\\\\)/)*/~?')
    literal = Alternative(RE('"(?:[^"]|\\\\")*?"'), RE("'(?:[^']|\\\\')*?'"))
    symbol = RE('(?!\\d)\\w+')
105
106
    link = Alternative(regexp, symbol, literal)
    option = Sequence(Token("["), expression, Required(Token("]")))
107
108
    repetition = Sequence(Token("{"), expression, Required(Token("}")))
    oneormore = Sequence(Token("{"), expression, Token("}+"))
109
    regexchain = Sequence(Token("<"), expression, Required(Token(">")))
110
111
112
113
114
    group = Sequence(Token("("), expression, Required(Token(")")))
    retrieveop = Alternative(Token("::"), Token(":"))
    flowmarker = Alternative(Token("!"), Token("&"), Token("§"), Token("-!"), Token("-&"))
    factor = Alternative(Sequence(Optional(flowmarker), Optional(retrieveop), symbol, NegativeLookahead(Token("="))),
                         Sequence(Optional(flowmarker), literal), Sequence(Optional(flowmarker), regexp),
115
116
                         Sequence(Optional(flowmarker), group), Sequence(Optional(flowmarker), regexchain),
                         Sequence(Optional(flowmarker), oneormore), repetition, option)
117
118
119
120
    term = OneOrMore(factor)
    expression.set(Sequence(term, ZeroOrMore(Sequence(Token("|"), term))))
    directive = Sequence(Token("@"), Required(symbol), Required(Token("=")), Alternative(regexp, literal, list_))
    definition = Sequence(symbol, Required(Token("=")), expression)
121
    syntax = Sequence(Optional(RE('', wR='', wL=WSP__)), ZeroOrMore(Alternative(definition, directive)), Required(EOF))
122
123
124
    root__ = syntax


125
126
#TODO: Add Capture and Retrieve Validation: A variable mustn't be captured twice before retrival?!?

127
EBNF_transformation_table = {
128
129
130
131
132
133
134
135
136
137
138
139
140
    # AST Transformations for EBNF-grammar
    "syntax":
        remove_expendables,
    "directive, definition":
        partial(remove_tokens, tokens={'@', '='}),
    "expression, chain":
        [replace_by_single_child, flatten,
         partial(remove_tokens, tokens={'|', '--'})],
    "term":
        [replace_by_single_child, flatten],  # supports both idioms:  "{ factor }+" and "factor { factor }"
    "factor, flowmarker, retrieveop":
        replace_by_single_child,
    "group":
141
142
143
        [remove_enclosing_delimiters, replace_by_single_child],
    "oneormore, repetition, option, regexchain":
        [reduce_single_child, remove_enclosing_delimiters],
144
    "symbol, literal, regexp":
145
146
147
        [remove_expendables, reduce_single_child],
    (TOKEN_KEYWORD, WHITESPACE_KEYWORD):
        [remove_expendables, reduce_single_child],
148
149
    "list_":
        [partial(remove_tokens, tokens={','})],
150
151
152
153
    "":
        [remove_expendables, replace_by_single_child]
}

154

155
EBNF_validation_table = {
156
157
158
159
160
    # Semantic validation on the AST
    "repetition, option, oneormore":
        [partial(forbid, child_tags=['repetition', 'option', 'oneormore']),
         partial(assert_content, regex=r'(?!§)')],
}
161

162

163
164
165
def EBNFTransform(syntax_tree):
    for processing_table in [EBNF_transformation_table, EBNF_validation_table]:
        traverse(syntax_tree, processing_table)
di68kap's avatar
di68kap committed
166
167


168
169
170
171
172
173
174
175
176
177
178
179
180
181
class EBNFCompilerError(Exception):
    """Error raised by `EBNFCompiler` class. (Not compilation errors
    in the strict sense, see `CompilationError` below)"""
    pass


class EBNFCompiler(CompilerBase):
    """Generates a Parser from an abstract syntax tree of a grammar specified
    in EBNF-Notation.
    """
    COMMENT_KEYWORD = "COMMENT__"
    RESERVED_SYMBOLS = {TOKEN_KEYWORD, WHITESPACE_KEYWORD, COMMENT_KEYWORD}
    AST_ERROR = "Badly structured syntax tree. " \
                "Potentially due to erroneuos AST transformation."
182
183
184
185
    PREFIX_TABLE = {'§': 'Required',
                    '&': 'Lookahead', '!': 'NegativeLookahead',
                    '-&': 'Lookbehind', '-!': 'NegativeLookbehind',
                    '::': 'Pop', ':': 'Retrieve'}
186
187
188
    WHITESPACE = {'horizontal': r'[\t ]*',  # default: horizontal
                  'linefeed': r'[ \t]*\n?(?!\s*\n)[ \t]*',
                  'vertical': r'\s*'}
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203

    def __init__(self, grammar_name="", source_text=""):
        super(EBNFCompiler, self).__init__()
        assert grammar_name == "" or re.match('\w+\Z', grammar_name)
        self.grammar_name = grammar_name
        self.source_text = load_if_file(source_text)
        self._reset()

    def _reset(self):
        self.rules = set()
        self.symbols = set()
        self.variables = set()
        self.definition_names = []
        self.recursive = set()
        self.root = ""
204
        self.directives = {'whitespace': self.WHITESPACE['horizontal'],
205
                           'comment': '',
206
                           'literalws': ['right'],
207
                           'tokens': set(),     # alt. 'scanner_tokens'
208
                           'counterpart': set()}     # alt. 'retrieve_counterpart'
209
210
211
212
213
214
215
216
217

    def gen_scanner_skeleton(self):
        name = self.grammar_name + "Scanner"
        return "def %s(text):\n    return text\n" % name

    def gen_AST_skeleton(self):
        if not self.definition_names:
            raise EBNFCompilerError('Compiler has not been run before calling '
                                    '"gen_AST_Skeleton()"!')
di68kap's avatar
di68kap committed
218
219
220
        tt_name = self.grammar_name + '_ASTTransform'
        pl_name = self.grammar_name + '_ASTPipeline'
        transtable = [tt_name + ' = {',
221
222
223
                      '    # AST Transformations for the ' +
                      self.grammar_name + '-grammar']
        for name in self.definition_names:
di68kap's avatar
di68kap committed
224
            transtable.append('    "' + name + '": no_operation,')
225
        transtable += ['    "": no_operation', '}', '',  pl_name + ' = [%s]' % tt_name, '']
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
        return '\n'.join(transtable)

    def gen_compiler_skeleton(self):
        if not self.definition_names:
            raise EBNFCompilerError('Compiler has not been run before calling '
                                    '"gen_Compiler_Skeleton()"!')
        compiler = ['class ' + self.grammar_name + 'Compiler(CompilerBase):',
                    '    """Compiler for the abstract-syntax-tree of a ' +
                    self.grammar_name + ' source file.',
                    '    """', '',
                    '    def __init__(self, grammar_name="' +
                    self.grammar_name + '"):',
                    '        super(' + self.grammar_name +
                    'Compiler, self).__init__()',
                    "        assert re.match('\w+\Z', grammar_name)", '']
        for name in self.definition_names:
            if name == self.root:
                compiler += ['    def ' + name + '(self, node):',
                             '        return node', '']
            else:
                compiler += ['    def ' + name + '(self, node):',
                             '        pass', '']
248
        return '\n'.join(compiler)
249
250
251
252
253
254

    def gen_parser(self, definitions):
        # fix capture of variables that have been defined before usage [sic!]
        if self.variables:
            for i in range(len(definitions)):
                if definitions[i][0] in self.variables:
255
                    definitions[i] = (definitions[i][0], 'Capture(%s)' % definitions[1])
256
257

        self.definition_names = [defn[0] for defn in definitions]
Eckhart Arnold's avatar
Eckhart Arnold committed
258
259
260
261
        definitions.append(('wspR__', WHITESPACE_KEYWORD
                            if 'right' in self.directives['literalws'] else "''"))
        definitions.append(('wspL__', WHITESPACE_KEYWORD
                            if 'left' in self.directives['literalws'] else "''"))
262
263
264
265
266
267
268
269
        definitions.append((WHITESPACE_KEYWORD,
                            ("mixin_comment(whitespace="
                             "r'{whitespace}', comment=r'{comment}')").
                            format(**self.directives)))
        definitions.append((self.COMMENT_KEYWORD, "r'{comment}'".format(**self.directives)))

        # prepare parser class header and docstring and
        # add EBNF grammar to the doc string of the parser class
270
        article = 'an ' if self.grammar_name[0:1] in "AaEeIiOoUu" else 'a '  # what about 'hour', 'universe' etc.?
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
        declarations = ['class ' + self.grammar_name +
                        'Grammar(GrammarBase):',
                        'r"""Parser for ' + article + self.grammar_name +
                        ' source file' +
                        (', with this grammar:' if self.source_text else '.')]
        definitions.append(('parser_initialization__', '"upon instatiation"'))
        if self.source_text:
            definitions.append(('source_hash__',
                                '"%s"' % md5(self.source_text, __version__)))
            declarations.append('')
            declarations += [line for line in self.source_text.split('\n')]
            while declarations[-1].strip() == '':
                declarations = declarations[:-1]
        declarations.append('"""')

286
287
288
289
290
291
292
        # add default functions for counterpart filters of pop or retrieve operators
        for symbol in self.directives['counterpart']:
            # declarations.append('def %s_counterpart(value): \n' % symbol +
            #                     '        return value.replace("(", ")").replace("[", "]")'
            #                     '.replace("{", "}").replace(">", "<")\n')
            declarations.append(symbol + '_counterpart = lambda value: value.replace("(", ")")'
                                '.replace("[", "]").replace("{", "}").replace(">", "<")')
293

294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
        # turn definitions into declarations in reverse order
        self.root = definitions[0][0] if definitions else ""
        definitions.reverse()
        declarations += [symbol + ' = Forward()'
                         for symbol in sorted(list(self.recursive))]
        for symbol, statement in definitions:
            if symbol in self.recursive:
                declarations += [symbol + '.set(' + statement + ')']
            else:
                declarations += [symbol + ' = ' + statement]
        for nd in self.symbols:
            if nd.result not in self.rules:
                nd.add_error("Missing production for symbol '%s'" % nd.result)
        if self.root and 'root__' not in self.symbols:
            declarations.append('root__ = ' + self.root)
        declarations.append('')
        return '\n    '.join(declarations)

    def syntax(self, node):
        self._reset()
        definitions = []

        # drop the wrapping sequence node
        if isinstance(node.parser, Sequence) and \
                isinstance(node.result[0].parser, ZeroOrMore):
            node = node.result[0]

        # compile definitions and directives and collect definitions
        for nd in node.result:
            if nd.parser.name == "definition":
                definitions.append(self.compile__(nd))
            else:
                assert nd.parser.name == "directive", nd.as_sexpr()
                self.compile__(nd)

        return self.gen_parser(definitions)

    def definition(self, node):
        rule = node.result[0].result
333
334
335
        if rule in self.rules:
            node.add_error('A rule with name "%s" has already been defined.' % rule)
        elif rule in EBNFCompiler.RESERVED_SYMBOLS:
336
337
338
339
            node.add_error('Symbol "%s" is a reserved symbol.' % rule)
        elif not sane_parser_name(rule):
            node.add_error('Illegal symbol "%s". Symbols must not start or '
                           ' end with a doube underscore "__".' % rule)
340
        elif rule in self.directives['tokens']:
341
342
343
344
345
346
347
348
349
            node.add_error('Symbol "%s" has already been defined as '
                           'a scanner token.' % rule)
        elif keyword.iskeyword(rule):
            node.add_error('Python keyword "%s" may not be used as a symbol. '
                           % rule + '(This may change in the furute.)')
        try:
            self.rules.add(rule)
            defn = self.compile__(node.result[1])
            if rule in self.variables:
350
                defn = 'Capture(%s)' % defn
351
352
353
354
355
                self.variables.remove(rule)
        except TypeError as error:
            errmsg = EBNFCompiler.AST_ERROR + " (" + str(error) + ")\n" + node.as_sexpr()
            node.add_error(errmsg)
            rule, defn = rule + ':error', '"' + errmsg + '"'
Eckhart Arnold's avatar
Eckhart Arnold committed
356
        return rule, defn
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373

    @staticmethod
    def _check_rx(node, rx):
        """Checks whether the string `rx` represents a valid regular
        expression. Makes sure that multiline regular expressions are
        prepended by the multiline-flag. Returns the regular expression string.
        """
        rx = rx if rx.find('\n') < 0 or rx[0:4] == '(?x)' else '(?x)' + rx
        try:
            re.compile(rx)
        except Exception as re_error:
            node.add_error("malformed regular expression %s: %s" %
                           (repr(rx), str(re_error)))
        return rx

    def directive(self, node):
        key = node.result[0].result.lower()
374
        assert key not in self.directives['tokens']
375
        if key in {'comment', 'whitespace'}:
376
377
            if node.result[1].parser.name == "list_":
                if len(node.result[1].result) != 1:
Eckhart Arnold's avatar
Eckhart Arnold committed
378
                    node.add_error('Directive "%s" must have one, but not %i values.' %
379
380
                                   (key, len(node.result[1])))
                value = self.compile__(node.result[1]).pop()
381
382
                if key == 'whitespace' and value in EBNFCompiler.WHITESPACE:
                    value = EBNFCompiler.WHITESPACE[value]  # replace whitespace-name by regex
383
                else:
384
                    node.add_error('Value "%s" not allowed for directive "%s".' % (value, key))
385
            else:
386
387
388
389
390
391
392
393
                value = node.result[1].result.strip("~")
                if value != node.result[1].result:
                    node.add_error("Whitespace marker '~' not allowed in definition of "
                                   "%s regular expression." % key)
                if value[0] + value[-1] in {'""', "''"}:
                    value = escape_re(value[1:-1])
                elif value[0] + value[-1] == '//':
                    value = self._check_rx(node, value[1:-1])
394
395
396
                if key == 'whitespace' and not re.match(value, ''):
                    node.add_error("Implicit whitespace should always match the empty string, "
                                   "/%s/ does not." % value)
397
            self.directives[key] = value
398

399
400
401
        elif key == 'literalws':
            value = {item.lower() for item in self.compile__(node.result[1])}
            if (len(value - {'left', 'right', 'both', 'none'}) > 0
Eckhart Arnold's avatar
Eckhart Arnold committed
402
                    or ('none' in value and len(value) > 1)):
403
404
405
406
407
408
409
                node.add_error('Directive "literalws" allows the values '
                               '`left`, `right`, `both` or `none`, '
                               'but not `%s`' % ", ".join(value))
            ws = {'left', 'right'} if 'both' in value \
                else {} if 'none' in value else value
            self.directives[key] = list(ws)

410
        elif key in {'tokens', 'scanner_tokens'}:
411
            self.directives['tokens'] |= self.compile__(node.result[1])
412

413
414
        elif key in {'counterpart', 'retrieve_counterpart'}:
            self.directives['counterpart'] |= self.compile__(node.result[1])
415

416
417
418
        else:
            node.add_error('Unknown directive %s ! (Known ones are %s .)' %
                           (key,
419
                            ', '.join(list(self.directives.keys()))))
420
421
        return ""

422
    def non_terminal(self, node, parser_class, custom_args=[]):
423
424
425
        """Compiles any non-terminal, where `parser_class` indicates the Parser class
        name for the particular non-terminal.
        """
426
        arguments = [self.compile__(r) for r in node.result] + custom_args
427
428
429
430
431
432
433
434
435
436
437
438
439
        return parser_class + '(' + ', '.join(arguments) + ')'

    def expression(self, node):
        return self.non_terminal(node, 'Alternative')

    def term(self, node):
        return self.non_terminal(node, 'Sequence')

    def factor(self, node):
        assert isinstance(node.parser, Sequence), node.as_sexpr()  # these assert statements can be removed
        assert node.children
        assert len(node.result) >= 2, node.as_sexpr()
        prefix = node.result[0].result
440
        custom_args = []
441
442
443

        if prefix in {'::', ':'}:
            assert len(node.result) == 2
444
445
            arg = node.result[-1]
            argstr = str(arg)
446
            if arg.parser.name != 'symbol':
Eckhart Arnold's avatar
Eckhart Arnold committed
447
                node.add_error(('Retrieve Operator "%s" requires a symbol, '
448
449
                                'and not a %s.') % (prefix, str(arg.parser)))
                return str(arg.result)
450
451
            if str(arg) in self.directives['counterpart']:
                custom_args = ['counterpart=%s_counterpart' % str(arg)]
452
453
            self.variables.add(arg.result)

454
        elif len(node.result) > 2:
455
456
            # shift = (Node(node.parser, node.result[1].result),)
            # node.result[1].result = shift + node.result[2:]
457
            node.result[1].result = (Node(node.result[1].parser, node.result[1].result),) \
458
459
460
461
462
                                    + node.result[2:]
            node.result[1].parser = node.parser
            node.result = (node.result[0], node.result[1])

        node.result = node.result[1:]
463
464
465
466
467
        try:
            parser_class = self.PREFIX_TABLE[prefix]
            return self.non_terminal(node, parser_class, custom_args)
        except KeyError:
            node.add_error('Unknown prefix "%s".' % prefix)
468
469
470
471
472
473
474
475
476
477

    def option(self, node):
        return self.non_terminal(node, 'Optional')

    def repetition(self, node):
        return self.non_terminal(node, 'ZeroOrMore')

    def oneormore(self, node):
        return self.non_terminal(node, 'OneOrMore')

478
479
480
    def regexchain(self, node):
        raise EBNFCompilerError("Not yet implemented!")

481
482
483
484
485
    def group(self, node):
        raise EBNFCompilerError("Group nodes should have been eliminated by "
                                "AST transformation!")

    def symbol(self, node):
486
        if node.result in self.directives['tokens']:
487
488
489
490
491
492
493
494
            return 'ScannerToken("' + node.result + '")'
        else:
            self.symbols.add(node)
            if node.result in self.rules:
                self.recursive.add(node.result)
            return node.result

    def literal(self, node):
495
        return 'Token(' + node.result.replace('\\', r'\\') + ')'  # return 'Token(' + ', '.join([node.result]) + ')' ?
496
497
498
499
500
501
502
503
504
505
506

    def regexp(self, node):
        rx = node.result
        name = []
        if rx[:2] == '~/':
            if not 'left' in self.directives['literalws']:
                name = ['wL=' + WHITESPACE_KEYWORD] + name
            rx = rx[1:]
        elif 'left' in self.directives['literalws']:
            name = ["wL=''"] + name
        if rx[-2:] == '/~':
Eckhart Arnold's avatar
Eckhart Arnold committed
507
            if 'right' not in self.directives['literalws']:
508
509
510
511
512
513
514
515
516
517
518
519
520
521
                name = ['wR=' + WHITESPACE_KEYWORD] + name
            rx = rx[:-1]
        elif 'right' in self.directives['literalws']:
            name = ["wR=''"] + name
        try:
            arg = repr(self._check_rx(node, rx[1:-1].replace(r'\/', '/')))
        except AttributeError as error:
            errmsg = EBNFCompiler.AST_ERROR + " (" + str(error) + ")\n" + \
                     node.as_sexpr()
            node.add_error(errmsg)
            return '"' + errmsg + '"'
        return 'RE(' + ', '.join([arg] + name) + ')'

    def list_(self, node):
522
523
        assert node.children
        return set(item.result.strip() for item in node.result)
524
525


526
527
528
def grammar_changed(grammar_class, grammar_source):
    """Returns ``True`` if ``grammar_class`` does not reflect the latest
    changes of ``grammar_source``
529
530
531
532

    Parameters:
        grammar_class:  the parser class representing the grammar
            or the file name of a compiler suite containing the grammar
533
534
535
        grammar_source:  File name or string representation of the
            EBNF code of the grammar
            
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
    Returns (bool):
        True, if the source text of the grammar is different from the
        source from which the grammar class was generated
    """
    grammar = load_if_file(grammar_source)
    chksum = md5(grammar, __version__)
    if isinstance(grammar_class, str):
        # grammar_class = load_compiler_suite(grammar_class)[1]
        with open(grammar_class, 'r', encoding='utf8') as f:
            pycode = f.read()
        m = re.search('class \w*\(GrammarBase\)', pycode)
        if m:
            m = re.search('    source_hash__ *= *"([a-z0-9]*)"',
                          pycode[m.span()[1]:])
            return not (m and m.groups() and m.groups()[-1] == chksum)
        else:
            return True
    else:
        return chksum != grammar_class.source_hash__