ebnf.py 27.5 KB
Newer Older
1
"""ebnf.py - EBNF -> Python-Parser compilation for DHParser
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19

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.
"""

from functools import partial
20
21
import keyword
import os
22
23
24
25
26
try:
    import regex as re
except ImportError:
    import re

27
from .toolkit import load_if_file, escape_re, md5, sane_parser_name
Eckhart Arnold's avatar
Eckhart Arnold committed
28
from .parsers import GrammarBase, mixin_comment, nil_scanner, Forward, RE, NegativeLookahead, \
29
30
    Alternative, Sequence, Optional, Required, OneOrMore, ZeroOrMore, Token, CompilerBase, \
    Capture, Retrieve
31
from .syntaxtree import Node, traverse, remove_enclosing_delimiters, reduce_single_child, \
di68kap's avatar
di68kap committed
32
    replace_by_single_child, TOKEN_KEYWORD, remove_expendables, remove_tokens, flatten, \
33
    forbid, assert_content, WHITESPACE_KEYWORD
Eckhart Arnold's avatar
Eckhart Arnold committed
34
from .versionnumber import __version__
35
36


37
38
39
40
41
__all__ = ['get_ebnf_scanner',
           'get_ebnf_grammar',
           'get_ebnf_transformer',
           'get_ebnf_compiler',
           'EBNFGrammar',
42
           'EBNFTransformer',
Eckhart Arnold's avatar
Eckhart Arnold committed
43
           'EBNFCompilerError',
44
45
           'EBNFCompiler',
           'grammar_changed']
46
47


Eckhart Arnold's avatar
Eckhart Arnold committed
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67


########################################################################
#
# EBNF scanning
#
########################################################################


def get_ebnf_scanner():
    return nil_scanner


########################################################################
#
# EBNF parsing
#
########################################################################


68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
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
87
                | [flowmarker] regexchain
88
89
90
91
92
93
94
95
96
                | [flowmarker] oneormore
                | repetition
                | option

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

    group      =  "(" expression §")"
97
    regexchain =  ">" expression §"<"                # compiles "expression" into a singular regular expression
98
99
    oneormore  =  "{" expression "}+"
    repetition =  "{" expression §"}"
100
101
    option     =  "[" expression §"]"

102
103
104
105
106
107
    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|$)/~
                                                     # '~' is a whitespace-marker, if present leading or trailing
                                                     # whitespace of a regular expression will be ignored tacitly.
108
    list_      =  /\w+/~ { "," /\w+/~ }              # comma separated list of symbols, e.g. BEGIN_LIST, END_LIST,
109
110
111
112
                                                     # BEGIN_QUOTE, END_QUOTE ; see CommonMark/markdown.py for an exmaple
    EOF =  !/./
    """
    expression = Forward()
113
    source_hash__ = "a410e1727fb7575e98ff8451dbf8f3bd"
114
    parser_initialization__ = "upon instatiation"
115
116
    COMMENT__ = r'#.*(?:\n|$)'
    WSP__ = mixin_comment(whitespace=r'\s*', comment=r'#.*(?:\n|$)')
117
    wspL__ = ''
118
    wspR__ = WSP__
119
    EOF = NegativeLookahead(RE('.', wR=''))
120
    list_ = Sequence(RE('\\w+'), ZeroOrMore(Sequence(Token(","), RE('\\w+'))))
121
122
123
    regexp = RE('~?/(?:[^/]|(?<=\\\\)/)*/~?')
    literal = Alternative(RE('"(?:[^"]|\\\\")*?"'), RE("'(?:[^']|\\\\')*?'"))
    symbol = RE('(?!\\d)\\w+')
124
    option = Sequence(Token("["), expression, Required(Token("]")))
125
126
    repetition = Sequence(Token("{"), expression, Required(Token("}")))
    oneormore = Sequence(Token("{"), expression, Token("}+"))
127
    regexchain = Sequence(Token("<"), expression, Required(Token(">")))
128
129
130
131
132
    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),
133
134
                         Sequence(Optional(flowmarker), group), Sequence(Optional(flowmarker), regexchain),
                         Sequence(Optional(flowmarker), oneormore), repetition, option)
135
136
137
138
    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)
139
    syntax = Sequence(Optional(RE('', wR='', wL=WSP__)), ZeroOrMore(Alternative(definition, directive)), Required(EOF))
140
141
142
    root__ = syntax


Eckhart Arnold's avatar
Eckhart Arnold committed
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
def grammar_changed(grammar_class, grammar_source):
    """Returns ``True`` if ``grammar_class`` does not reflect the latest
    changes of ``grammar_source``

    Parameters:
        grammar_class:  the parser class representing the grammar
            or the file name of a compiler suite containing the grammar
        grammar_source:  File name or string representation of the
            EBNF code of the grammar

    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__


def get_ebnf_grammar():
    global thread_local_ebnf_grammar_singleton
    try:
        grammar = thread_local_ebnf_grammar_singleton
        return grammar
    except NameError:
        thread_local_ebnf_grammar_singleton = EBNFGrammar()
        return thread_local_ebnf_grammar_singleton


########################################################################
#
# EBNF concrete to abstract syntax tree transformation and validation
#
########################################################################


191
192
#TODO: Add Capture and Retrieve Validation: A variable mustn't be captured twice before retrival?!?

193
EBNF_transformation_table = {
194
195
196
197
198
    # AST Transformations for EBNF-grammar
    "syntax":
        remove_expendables,
    "directive, definition":
        partial(remove_tokens, tokens={'@', '='}),
Eckhart Arnold's avatar
Eckhart Arnold committed
199
    "expression":
200
        [replace_by_single_child, flatten,
Eckhart Arnold's avatar
Eckhart Arnold committed
201
         partial(remove_tokens, tokens={'|'})],
202
203
204
205
206
    "term":
        [replace_by_single_child, flatten],  # supports both idioms:  "{ factor }+" and "factor { factor }"
    "factor, flowmarker, retrieveop":
        replace_by_single_child,
    "group":
207
208
209
        [remove_enclosing_delimiters, replace_by_single_child],
    "oneormore, repetition, option, regexchain":
        [reduce_single_child, remove_enclosing_delimiters],
210
    "symbol, literal, regexp":
211
212
213
        [remove_expendables, reduce_single_child],
    (TOKEN_KEYWORD, WHITESPACE_KEYWORD):
        [remove_expendables, reduce_single_child],
214
    "list_":
215
        [flatten, partial(remove_tokens, tokens={','})],
216
217
218
219
    "":
        [remove_expendables, replace_by_single_child]
}

220

221
EBNF_validation_table = {
222
223
224
225
226
    # Semantic validation on the AST
    "repetition, option, oneormore":
        [partial(forbid, child_tags=['repetition', 'option', 'oneormore']),
         partial(assert_content, regex=r'(?!§)')],
}
227

228

229
def EBNFTransformer(syntax_tree):
230
231
    for processing_table in [EBNF_transformation_table, EBNF_validation_table]:
        traverse(syntax_tree, processing_table)
di68kap's avatar
di68kap committed
232
233


Eckhart Arnold's avatar
Eckhart Arnold committed
234
def get_ebnf_transformer():
235
    return EBNFTransformer
Eckhart Arnold's avatar
Eckhart Arnold committed
236
237
238
239
240
241
242
243


########################################################################
#
# EBNF abstract syntax tree to Python parser compilation
#
########################################################################

244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
SCANNER_FACTORY = '''
def get_{NAME}_scanner():
    return {NAME}Scanner
'''


GRAMMAR_FACTORY = '''
def get_{NAME}_grammar():
    global thread_local_{NAME}_grammar_singleton
    try:
        grammar = thread_local_{NAME}_grammar_singleton
        return grammar
    except NameError:
        thread_local_{NAME}_grammar_singleton = {NAME}Grammar()
        return thread_local_{NAME}_grammar_singleton
'''


TRANSFORMER_FACTORY = '''
def get_{NAME}_transformer():
    return {NAME}Transform
'''


COMPILER_FACTORY = '''
def get_{NAME}_compiler(grammar_name="{NAME}", 
                        grammar_source=""):
    global thread_local_{NAME}_compiler_singleton
    try:
        compiler = thread_local_{NAME}_compiler_singleton
        compiler.set_grammar_name(grammar_name, grammar_source)
        return compiler
    except NameError:
        thread_local_{NAME}_compiler_singleton = \\
            {NAME}Compiler(grammar_name, grammar_source)
        return thread_local_{NAME}_compiler_singleton 
'''

Eckhart Arnold's avatar
Eckhart Arnold committed
282

283
284
285
286
287
288
289
290
291
292
293
294
295
296
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."
297
298
299
300
    PREFIX_TABLE = {'§': 'Required',
                    '&': 'Lookahead', '!': 'NegativeLookahead',
                    '-&': 'Lookbehind', '-!': 'NegativeLookbehind',
                    '::': 'Pop', ':': 'Retrieve'}
301
302
303
    WHITESPACE = {'horizontal': r'[\t ]*',  # default: horizontal
                  'linefeed': r'[ \t]*\n?(?!\s*\n)[ \t]*',
                  'vertical': r'\s*'}
304

305
    def __init__(self, grammar_name="", grammar_source=""):
Eckhart Arnold's avatar
Eckhart Arnold committed
306
        super(EBNFCompiler, self).__init__(grammar_name, grammar_source)
307
308
309
        self._reset()

    def _reset(self):
Eckhart Arnold's avatar
Eckhart Arnold committed
310
        self._result = None
311
312
        self.rules = set()
        self.variables = set()
313
        self.symbol_nodes = []
314
315
316
        self.definition_names = []
        self.recursive = set()
        self.root = ""
317
        self.directives = {'whitespace': self.WHITESPACE['horizontal'],
318
                           'comment': '',
319
                           'literalws': ['right'],
320
                           'tokens': set(),     # alt. 'scanner_tokens'
321
                           'counterpart': set()}     # alt. 'retrieve_counterpart'
322

Eckhart Arnold's avatar
Eckhart Arnold committed
323
324
325
326
    @property
    def result(self):
        return self._result

327
328
    def gen_scanner_skeleton(self):
        name = self.grammar_name + "Scanner"
329
330
        return "def %s(text):\n    return text\n" % name \
               + SCANNER_FACTORY.format(NAME=self.grammar_name)
331

332
    def gen_transformer_skeleton(self):
333
        if not self.definition_names:
Eckhart Arnold's avatar
Eckhart Arnold committed
334
335
            raise EBNFCompilerError('Compiler must be run before calling '
                                    '"gen_transformer_Skeleton()"!')
336
337
        tt_name = self.grammar_name + '_AST_transformation_table'
        tf_name = self.grammar_name + 'Transform'
di68kap's avatar
di68kap committed
338
        transtable = [tt_name + ' = {',
339
340
341
                      '    # AST Transformations for the ' +
                      self.grammar_name + '-grammar']
        for name in self.definition_names:
di68kap's avatar
di68kap committed
342
            transtable.append('    "' + name + '": no_operation,')
343
344
        transtable += ['    "": no_operation', '}', '',  tf_name +
                       ' = partial(traverse, processing_table=%s)' % tt_name, '']
345
        transtable += [TRANSFORMER_FACTORY.format(NAME=self.grammar_name)]
346
347
348
349
350
351
352
353
354
355
356
        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="' +
Eckhart Arnold's avatar
Eckhart Arnold committed
357
                    self.grammar_name + '", grammar_source=""):',
358
                    '        super(' + self.grammar_name +
Eckhart Arnold's avatar
Eckhart Arnold committed
359
                    'Compiler, self).__init__(grammar_name, grammar_source)',
360
361
                    "        assert re.match('\w+\Z', grammar_name)", '']
        for name in self.definition_names:
362
            method_name = CompilerBase.derive_method_name(name)
363
            if name == self.root:
364
                compiler += ['    def ' + method_name + '(self, node):',
365
366
                             '        return node', '']
            else:
367
                compiler += ['    def ' + method_name + '(self, node):',
368
                             '        pass', '']
369
        compiler += [COMPILER_FACTORY.format(NAME=self.grammar_name)]
370
        return '\n'.join(compiler)
371

Eckhart Arnold's avatar
Eckhart Arnold committed
372
    def assemble_parser(self, definitions):
373
        # fix capture of variables that have been defined before usage [sic!]
374

375
376
377
        if self.variables:
            for i in range(len(definitions)):
                if definitions[i][0] in self.variables:
378
                    definitions[i] = (definitions[i][0], 'Capture(%s)' % definitions[1])
379
380

        self.definition_names = [defn[0] for defn in definitions]
Eckhart Arnold's avatar
Eckhart Arnold committed
381
382
383
384
        definitions.append(('wspR__', WHITESPACE_KEYWORD
                            if 'right' in self.directives['literalws'] else "''"))
        definitions.append(('wspL__', WHITESPACE_KEYWORD
                            if 'left' in self.directives['literalws'] else "''"))
385
386
387
388
389
390
391
392
        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
393

394
        article = 'an ' if self.grammar_name[0:1] in "AaEeIiOoUu" else 'a '  # what about 'hour', 'universe' etc.?
395
396
397
398
        declarations = ['class ' + self.grammar_name +
                        'Grammar(GrammarBase):',
                        'r"""Parser for ' + article + self.grammar_name +
                        ' source file' +
399
                        (', with this grammar:' if self.grammar_source else '.')]
400
        definitions.append(('parser_initialization__', '"upon instatiation"'))
401
        if self.grammar_source:
402
            definitions.append(('source_hash__',
403
                                '"%s"' % md5(self.grammar_source, __version__)))
404
            declarations.append('')
405
            declarations += [line for line in self.grammar_source.split('\n')]
406
407
408
409
            while declarations[-1].strip() == '':
                declarations = declarations[:-1]
        declarations.append('"""')

410
        # add default functions for counterpart filters of pop or retrieve operators
411

412
413
414
415
416
417
        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(">", "<")')
418

419
        # turn definitions into declarations in reverse order
420

421
422
423
424
425
426
427
428
429
        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]
430
        for nd in self.symbol_nodes:
431
432
            if nd.result not in self.rules:
                nd.add_error("Missing production for symbol '%s'" % nd.result)
433
        if self.root and 'root__' not in self.rules:
434
435
            declarations.append('root__ = ' + self.root)
        declarations.append('')
Eckhart Arnold's avatar
Eckhart Arnold committed
436
437
438
        self._result = '\n    '.join(declarations) \
                       + GRAMMAR_FACTORY.format(NAME=self.grammar_name)
        return self._result
439

440
    def on_syntax(self, node):
441
442
443
444
        self._reset()
        definitions = []

        # drop the wrapping sequence node
445
        if len(node.children) == 1 and not node.result[0].parser.name:
446
447
448
449
450
            node = node.result[0]

        # compile definitions and directives and collect definitions
        for nd in node.result:
            if nd.parser.name == "definition":
451
                definitions.append(self._compile(nd))
452
453
            else:
                assert nd.parser.name == "directive", nd.as_sexpr()
454
                self._compile(nd)
455

Eckhart Arnold's avatar
Eckhart Arnold committed
456
        return self.assemble_parser(definitions)
457

458
    def on_definition(self, node):
459
        rule = node.result[0].result
460
461
462
        if rule in self.rules:
            node.add_error('A rule with name "%s" has already been defined.' % rule)
        elif rule in EBNFCompiler.RESERVED_SYMBOLS:
463
464
465
466
            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)
467
        elif rule in self.directives['tokens']:
468
469
470
471
472
473
474
            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)
475
            defn = self._compile(node.result[1])
476
            if rule in self.variables:
477
                defn = 'Capture(%s)' % defn
478
479
480
481
482
                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
483
        return rule, defn
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498

    @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

499
    def on_directive(self, node):
500
        key = node.result[0].result.lower()
501
        assert key not in self.directives['tokens']
502
        if key in {'comment', 'whitespace'}:
503
504
            if node.result[1].parser.name == "list_":
                if len(node.result[1].result) != 1:
Eckhart Arnold's avatar
Eckhart Arnold committed
505
                    node.add_error('Directive "%s" must have one, but not %i values.' %
506
                                   (key, len(node.result[1])))
507
                value = self._compile(node.result[1]).pop()
508
509
                if key == 'whitespace' and value in EBNFCompiler.WHITESPACE:
                    value = EBNFCompiler.WHITESPACE[value]  # replace whitespace-name by regex
510
                else:
511
                    node.add_error('Value "%s" not allowed for directive "%s".' % (value, key))
512
            else:
513
514
515
516
517
518
519
520
                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])
521
522
523
                if key == 'whitespace' and not re.match(value, ''):
                    node.add_error("Implicit whitespace should always match the empty string, "
                                   "/%s/ does not." % value)
524
            self.directives[key] = value
525

526
        elif key == 'literalws':
527
            value = {item.lower() for item in self._compile(node.result[1])}
528
            if (len(value - {'left', 'right', 'both', 'none'}) > 0
Eckhart Arnold's avatar
Eckhart Arnold committed
529
                    or ('none' in value and len(value) > 1)):
530
531
532
533
534
535
536
                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)

537
        elif key in {'tokens', 'scanner_tokens'}:
538
            self.directives['tokens'] |= self._compile(node.result[1])
539

540
        elif key in {'counterpart', 'retrieve_counterpart'}:
541
            self.directives['counterpart'] |= self._compile(node.result[1])
542

543
544
545
        else:
            node.add_error('Unknown directive %s ! (Known ones are %s .)' %
                           (key,
546
                            ', '.join(list(self.directives.keys()))))
547
548
        return ""

549
    def non_terminal(self, node, parser_class, custom_args=[]):
550
551
552
        """Compiles any non-terminal, where `parser_class` indicates the Parser class
        name for the particular non-terminal.
        """
553
        arguments = [self._compile(r) for r in node.result] + custom_args
554
555
        return parser_class + '(' + ', '.join(arguments) + ')'

556
    def on_expression(self, node):
557
558
        return self.non_terminal(node, 'Alternative')

559
    def on_term(self, node):
560
561
        return self.non_terminal(node, 'Sequence')

562
    def on_factor(self, node):
563
564
565
566
        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
567
        custom_args = []
568
569
570

        if prefix in {'::', ':'}:
            assert len(node.result) == 2
571
            arg = node.result[-1]
572
            if arg.parser.name != 'symbol':
Eckhart Arnold's avatar
Eckhart Arnold committed
573
                node.add_error(('Retrieve Operator "%s" requires a symbol, '
574
575
                                'and not a %s.') % (prefix, str(arg.parser)))
                return str(arg.result)
576
577
            if str(arg) in self.directives['counterpart']:
                custom_args = ['counterpart=%s_counterpart' % str(arg)]
578
579
            self.variables.add(arg.result)

580
        elif len(node.result) > 2:
581
582
            # shift = (Node(node.parser, node.result[1].result),)
            # node.result[1].result = shift + node.result[2:]
583
            node.result[1].result = (Node(node.result[1].parser, node.result[1].result),) \
584
585
586
587
588
                                    + node.result[2:]
            node.result[1].parser = node.parser
            node.result = (node.result[0], node.result[1])

        node.result = node.result[1:]
589
590
591
592
593
        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)
594

595
    def on_option(self, node):
596
597
        return self.non_terminal(node, 'Optional')

598
    def on_repetition(self, node):
599
600
        return self.non_terminal(node, 'ZeroOrMore')

601
    def on_oneormore(self, node):
602
603
        return self.non_terminal(node, 'OneOrMore')

604
    def on_regexchain(self, node):
605
606
        raise EBNFCompilerError("Not yet implemented!")

607
    def on_group(self, node):
608
609
610
        raise EBNFCompilerError("Group nodes should have been eliminated by "
                                "AST transformation!")

611
    def on_symbol(self, node):
612
        if node.result in self.directives['tokens']:
613
614
            return 'ScannerToken("' + node.result + '")'
        else:
615
            self.symbol_nodes.append(node)
616
617
618
619
            if node.result in self.rules:
                self.recursive.add(node.result)
            return node.result

620
    def on_literal(self, node):
621
        return 'Token(' + node.result.replace('\\', r'\\') + ')'  # return 'Token(' + ', '.join([node.result]) + ')' ?
622

623
    def on_regexp(self, node):
624
625
626
627
628
629
630
631
632
        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
633
            if 'right' not in self.directives['literalws']:
634
635
636
637
638
639
640
641
642
643
644
645
646
                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) + ')'

647
    def on_list_(self, node):
648
649
        assert node.children
        return set(item.result.strip() for item in node.result)
650
651


Eckhart Arnold's avatar
Eckhart Arnold committed
652
653
654
655
656
657
658
659
660
def get_ebnf_compiler(grammar_name="", grammar_source=""):
    global thread_local_ebnf_compiler_singleton
    try:
        compiler = thread_local_ebnf_compiler_singleton
        compiler.set_grammar_name(grammar_name, grammar_source)
        return compiler
    except NameError:
        thread_local_ebnf_compiler_singleton = EBNFCompiler(grammar_name, grammar_source)
        return thread_local_ebnf_compiler_singleton