编译技术-词法分析

对循环语句和条件判断语句编写词法分析编译程序，只能通过一遍扫描完成。
实验要求：
(1) 关键字：
for if then else while do until int input output
所有关键字都是小写。
(2）运算符和分隔符：
： = + - * / < > <= <> >= ; ( ) #
(2) 其他标识符（ID）和整型常数（NUM），通过以下模式定义：
ID=letter(letter | digit)*
NUM=digit digit*
(4)空格由空白、制表符和换行符组成。空格一般用来分隔ID、NUM、运算符、分隔符和关键字，词法分析阶段通常被忽略。

词法分析程序的功能
输入：源程序
输出：二元组（词法记号,属性值/其在符号表中的位置）构成的序列。
例如：对源程序
int x:=5; if (x>0) then x:=2*x+1/3; else x:=2/x; #
经词法分析后输出如下序列：
（30, int）（10,’x’）(18, :=) (11,5) (26, ；) (2, if ) (27,( )……

keyword = ["for","if","then","else","while","do","until","int","input","output"]
def getkey(temp):
    key = -1
    assert isinstance(temp,str)
    if temp == "for" : key = 1
    elif temp == "if" : key = 2
    elif temp == "then" : key = 3
    elif temp == "else" : key = 4
    elif temp == "while" : key = 5
    elif temp == "do" : key = 6
    elif temp == "until" : key = 29
    elif temp == "input" : key = 31
    elif temp == "int" : key = 30
    elif temp == "output" : key = 32
    elif temp == "alnum" : key = 10
    elif temp == "digit" : key = 11
    elif temp == "+" : key = 13
    elif temp == "-" : key = 14
    elif temp == "*" : key = 15
    elif temp == "/" : key = 16
    elif temp == ":" : key = 17
    elif temp == ":=" : key = 18
    elif temp == "<" : key = 20
    elif temp == "<>" : key = 21
    elif temp == "<=" : key = 22
    elif temp == ">" : key = 23
    elif temp == ">=" : key = 24
    elif temp == "=" : key = 25
    elif temp == ";" : key = 26
    elif temp == "(" : key = 27
    elif temp == ")" : key = 28
    elif temp == "#" : key = 0
    elif temp == "float" : key = 12
    elif temp.isdigit():key = 110
    elif temp.isalnum():key = 100
    else : key = -1
    return key

def iseasy(key):
    if key == 0 or key == 13 or key == 14 or key == 15 or key == 16 or key == 25  or key == 26 or key == 27 or key == 28:
        return True
    else:
        return False
l1 = []
l2 = []
def f(file_name):
    try:
        fp_read = open(file_name, 'r')
        while True:
            read = fp_read.readline()
            read = read.lower()
            if not read:
                break
            length = len(read)
            i = -1
            while i < length - 1:
                i += 1
                if getkey(read[i]) == -1:
                    continue
                elif iseasy(getkey(read[i])):
                    l1.append(read[i])
                    l2.append(getkey(read[i]))
                else:
                    if read[i] == ":":
                        i += 1
                        if read[i] == "=":
                            l1.append(":=")
                            l2.append(getkey(":="))
                        else:
                            i -= 1
                            l1.append(":")
                            l2.append(getkey(":"))
                    elif read[i] == "<":
                        i += 1
                        if read[i] == ">":
                            l1.append("<>")
                            l2.append(getkey("<>"))
                        elif read[i] == "=":
                            l1.append("<=")
                            l2.append(getkey("<="))
                        else:
                            i -= 1
                            l1.append("<")
                            l2.append(getkey("<"))
                    elif read[i] == ">":
                        i += 1
                        if read[i] == "=":
                            l1.append(">=")
                            l2.append(getkey(">="))
                        else:
                            i -= 1
                            l1.append(">")
                            l2.append(getkey(">"))
                    elif read[i].isdigit():
                        flag = False
                        floatflag = False
                        s = str(read[i])
                        while True:
                            i += 1
                            if read[i].isdigit():
                                s = str(s) + str(read[i])
                            elif read[i].isalpha() and floatflag == False:
                                s = str(s) + str(read[i])
                                flag = True
                            elif read[i] == "." and floatflag == False and flag == False:
                                s = str(s) + str(read[i])
                                floatflag = True
                            else:
                                break
                        i -= 1
                        if flag:
                            l1.append(s)
                            l2.append(getkey("alnum"))
                        elif floatflag:
                            l1.append(s)
                            l2.append(getkey("float"))
                        else:
                            l1.append(s)
                            l2.append(getkey("digit"))
                    elif read[i].isalpha():
                        s = str(read[i])
                        while True:
                            i += 1
                            if read[i].isalnum():
                                s = str(s) + str(read[i])
                            else:
                                break
                        i -= 1
                        if s in keyword:
                            l1.append(s)
                            l2.append(getkey(s))
                        else:
                            l1.append("'"+s+"'")
                            l2.append(getkey("alnum"))
    except Exception:
        print(file_name, ': This FileName Not Found!')
def main():
    f("C:\\Users\\Qzh_794\\PycharmProjects\\hello\\1.txt")
    l1.reverse()
    l2.reverse()
    while l1 and l2:
        print("("+str(l2.pop())+","+str(l1.pop())+")")
if __name__ == '__main__':
    main()