def make_messy_parse_function(regexes, rules, eof=False, post_lexer=None,
extra_names=()):
"""
NOT_RPYTHON: This function is only called to process the grammarfile, which
occurs before translation, so it does not need to be RPython.
The parse function it returns parses Fundy code into a "messy" AST, which
can be cleaned up using the ToAST object obtained from parse_ebnf(grammar).
"""
names, regexes = zip(*regexes)
if "IGNORE" in names:
ignore = ["IGNORE"]
else:
ignore = []
check_for_missing_names(names + extra_names, regexes, rules)
lexer = Lexer(list(regexes), list(names), ignore=ignore)
parser = PackratParser(rules, rules[0].nonterminal)
def parse(s):
tokens = lexer.tokenize(s, eof=eof)
if post_lexer is not None:
tokens = post_lexer(tokens)
s = parser.parse(tokens)
return s
return parse
def _get_parse_tools():
grammar = _get_grammar_file()
try:
regexs, rules, transformer_class = parse_ebnf(grammar)
except ParseError as e:
print e.nice_error_message('grammar.ebnf', grammar)
raise e
names, regexs = zip(*regexs)
check_for_missing_names(names, regexs, rules)
lexer = PieLexer(list(regexs), list(names))
parser = PackratParser(rules, rules[0].nonterminal)
def parse(s):
tokens = lexer.tokenize(s)
s = parser.parse(tokens)
return s
return parse, transformer_class()