import abstract_types

tokens_enum = ['global_res_word','context_res_word', 'res_symbol',
	'identifier', 'string_type', 'int_type', 'float_type']
tokens = abstract_types.enum(tokens_enum)
	
glob_res = ['VAR', 'IF', 'WHILE', 'FOR', 'PRINT']

cont_res = ['break', 'next', 'loop', 'else']
	
symb_res = ['+', '-', '*', '/', '(', ')', '^', '%', 'EOF']
	
token_type = abstract_types.record()

token_type.infile = file('d:/source.bpl', 'r')
token_type.cur_str = ""
token_type.cur_str_len = 0
token_type.cur_line = 0
token_type.cur_pos = 0
token_type.type_of_token = tokens
token_type.read_str = ""
token_type.cap_str = ""
token_type.int_val = 0
token_type.float_val = 0.0
token_type.glob_res_word = glob_res
token_type.cont_res_word = cont_res
token_type.res_sym = symb_res

def is_delim(char):
	if ' ,;'.find(char) <> -1:
		return 1
	return 0

def is_ident(char):
	if 'ABCDEFGHIJKLMNOPQRSTUVWXYZ'.find(char) <> -1:
		return 1
	return 0

def is_num(char):
	if '0123456789'.find(char) <> -1:
		return 1
	return 0

def is_res_symbol(char):
	if char in ['+', '-', '*', '/', '(', ')', '^', '%', 'EOF']:
		return 1
	return 0

def numeric(num):
	if num.find(".") <> -1 or num.find("E") <> -1:
		return 1
	return 0

def advance_token(token):
	glob_res = ['VAR', 'IF', 'WHILE', 'FOR', 'PRINT']
	cont_res = ['BREAK', 'NEXT', 'LOOP', 'ELSE']
	symb_res = ['+', '-', '*', '/', '(', ')', '^', '%', 'EOF']
	
	token.cur_str = ""
	token.read_str = ""
	token.cap_str = ""
	while (token.cur_str[token.cur_pos] == " "):
		if token.cur_pos > token.cur_str_len:
			token.cur_str = token.infile.readline()
			token.cur_str_len = len(token.cur_str)
			token.cur_pos = 0
			if not token.cur_str:
				token.type_of_token = 'RESERVED_SYMBOL'
				token.res_sym = 'EOF'
	read_str_idx = 0
	while (not is_delim(token.cur_str[token.cur_pos])):
		token.read_str[read_str_idx] = token.cur_str[token.cur_pos]
		token.cap_str[read_str_idx] = token.cur_str[token.cur_pos].ucase()
		read_str_idx += 1
	if is_ident(token.cap_str[0]):
		if token.cap_str in glob_res:
			token.type_of_token = 'GLOBAL_RES_WORD'
			token.glob_res_word = token.cap_str
			return
		else:
			token.glob_res_word = 'UNDEFINED'
		if token.type_of_token  in cont_res:
			token.type_of_token = 'CONTEXT_RES_WORD'
			token.cont_res_word = token.cap_str
			return
		else:
			token.cont_res_word = 'UNDEFINED'
		token.type_of_token = 'IDENTIFIER'
		return
	if is_num(token.cap_str[0]):
		num = numeric(token.cap_str)
		if num:
			token.type_of_token = 'FLOAT'
			token.float_val = float(token.cap_str)
			return
		else:
			token.type_of_token = 'INT'
			token.int_val = int(token.cap_str)
			return
	if is_res_symbol(token.cap_str[0]):
		token.res_sym = token.cap_str
		token.type_of_token = 'RESERVED'
		return
	
	token.type_of_token = 'UNKNOWN'
	return