path: root/ebnf.py

import string

class EBNF:
    IDENT   = 0
    LITERAL = 2
    LPAREN  = 3
    LBRAK   = 4
    LBRACE  = 5
    BAR     = 6
    EQL     = 7
    RPAREN  = 8
    RBRAK   = 9
    RBRACE  = 10
    PERIOD  = 11
    OTHER   = 12

    __input__  = None
    __char__   = None
    __symbol__ = None
    __data__   = None

    def __init__(self, input):
        self.__input__ = input
        self.__get_char__()

    def __get_char__(self):
        self.__char__ = self.__input__.read(1)
        if self.__char__ == '':
            raise EOFError
        return self.__char__

    def __get_sym__(self):
        self.__data__ = None

        while self.__char__ in string.whitespace:
            self.__get_char__()

        if self.__char__ in string.letters + "_":
            self.__symbol__ = self.IDENT
            self.__data__ = self.__char__
            while self.__get_char__() in string.letters + "_":
                self.__data__ += self.__char__
            return

        elif self.__char__ == '"':
            self.__symbol__ = self.LITERAL
            self.__data__ = ""
            while self.__get_char__() != '"':
                self.__data__ += self.__char__

        elif self.__char__ == '=':
            self.__symbol__ = self.EQL
        elif self.__char__ == '(':
            self.__symbol__ = self.LPAREN
        elif self.__char__ == ')':
            self.__symbol__ = self.RPAREN
        elif self.__char__ == '[':
            self.__symbol__ = self.LBRAK
        elif self.__char__ == ']':
            self.__symbol__ = self.RBRAK
        elif self.__char__ == '{':
            self.__symbol__ = self.LBRACE
        elif self.__char__ == '}':
            self.__symbol__ = self.RBRACE
        elif self.__char__ == '|':
            self.__symbol__ = self.BAR
        elif self.__char__ == '.':
            self.__symbol__ = self.PERIOD
        else:
            self.__symbol__ = self.OTHER

        self.__get_char__()

    def __error__(self, num):
        pos = self.__input__.tell()
        if pos > self.__lastpos__ + 2:
            print "ERROR: pos=%d, err=%d, sym=%d" % (pos, num, self.__symbol__)
            self.__lastpos__ = pos

    def __expression__(self):
        self.__term__()
        while self.__symbol__ == self.BAR:
            self.__get_sym__()
            self.__term__()

    def __term__(self):
        self.__factor__()
        while self.__symbol__ < self.BAR:
            self.__factor__()

    def __factor__(self):
        if self.__symbol__ == self.IDENT:
            print "record(TO, '%s', 1)" % (self.__data__,)
            self.__get_sym__()
        elif self.__symbol__ == self.LITERAL:
            print "record(T1, '%s', 0)" % (self.__data__,)
            self.__get_sym__()
        elif self.__symbol__ == self.LPAREN:
            self.__get_sym__()
            self.__expression__()
            if self.__symbol__ == self.RPAREN:
                self.__get_sym__()
            else:
                self.__error__(2)
        elif self.__symbol__ == self.LBRAK:
            self.__get_sym__()
            self.__expression__()
            if self.__symbol__ == self.RBRAK:
                self.__get_sym__()
            else:
                self.__error__(3)
        elif self.__symbol__ == self.LBRACE:
            self.__get_sym__()
            self.__expression__()
            if self.__symbol__ == self.RBRACE:
                self.__get_sym__()
            else:
                self.__error__(4)
        else:
            self.__error__(5)

    def __production__(self):
        self.__get_sym__()

        if self.__symbol__ == self.EQL:
            self.__get_sym__()
        else:
            self.__error__(7)

        self.__expression__()

        if self.__symbol__ == self.PERIOD:
            self.__get_sym__()
        else:
            self.__error__(8)

    def __syntax__(self):
        while self.__symbol__ == self.IDENT:
            self.__production__()

    def compile(self):
        try:
            self.__lastpos__ = 0
            self.__get_sym__()
            self.__syntax__()
        except EOFError:
            return

if __name__ == '__main__':
    import sys
    ebnf = EBNF(sys.stdin)
    ebnf.compile()