【编译原理】词法分析器

来源：互联网发布：淘宝节点考试在哪里编辑：程序博客网时间：2024/04/29 20:18
#include <stdio.h>#include <iostream>#include <string.h>using namespace std;// char *keyword[6] = {"begin", "if", "then", "while", "do", "end"};// char *opword[16] = {":", ":=", "+", "-", "*", "/", "<", "<=", "<>", ">", ">=", "=", ";", "(", ")", "#"};bool showline = true;int syn;char token[100];int num;int row;bool huiche = false;int t2, t1;int STATE;int inSTATE;bool inzhushi = false;void clearpos(){    memset(token, 0, sizeof(token));    t1 = 0;    t2 = 1;    num = 0;}bool isvar(char *s){    int l = strlen(s);    if (!isalpha(s[0]))        return false;    for (int i = 0; i < l; i++)    {        if (isdigit(s[i]) || isalpha(s[i]))            continue;        else            return false;    }    return true;}bool isint(char *s){    for (int i = 0; i < strlen(s); i++)    {        if (!isdigit(s[i]))            return false;    }    return true;}int getnum(char *s){    if (STATE == 2)        return 11;    if (strcmp(s, "begin") == 0)return 1;    else if (strcmp(s, "if") == 0)return 2;    else if (strcmp(s, "then") == 0)return 3;    else if (strcmp(s, "while") == 0)return 4;    else if (strcmp(s, "do") == 0)return 5;    else if (strcmp(s, "end") == 0)return 6;    else if (isvar(s))          return 10;    else if (strcmp(s, "+") == 0)return 13;    else if (strcmp(s, "-") == 0)return 14;    else if (strcmp(s, "*") == 0)return 15;    else if (strcmp(s, "/") == 0)return 16;    else if (strcmp(s, ":") == 0)return 17;    else if (strcmp(s, ":=") == 0)return 18;    else if (strcmp(s, "<") == 0)return 20;    else if (strcmp(s, "<>") == 0)return 21;    else if (strcmp(s, "<=") == 0)return 22;    else if (strcmp(s, ">") == 0)return 23;    else if (strcmp(s, ">=") == 0)return 24;    else if (strcmp(s, "=") == 0)return 25;    else if (strcmp(s, ";") == 0)return 26;    else if (strcmp(s, "(") == 0)return 27;    else if (strcmp(s, ")") == 0)return 28;    else if (strcmp(s, "#") == 0)return 0;    else if (s[0] == '/' && s[1] == '*')return -2;    else return -1;}void output(){    if (huiche && showline)    {        huiche = false;        printf("\n");        printf("%d: ", row);    }    syn = getnum(token);    if (syn == -1)    {        printf("(error!,%s)\n", token);        return;    }    else if (syn == -2)    {        inzhushi = true;        inSTATE = -2;        int l = strlen(token);        if (token[l - 1] == '/' && token[l - 2] == '*' && l >= 4)        {            inzhushi = false;            STATE = 0;        }        return;    }    else if (STATE == 2)    {        if (isint(token))            printf("(%d,%s)", 11, token);        else            printf("(error!,%s)\n", token);    }    else    {        printf("(%d,%s)", syn, token);    }    if (showline == false)        printf("\n");}int main(){    freopen("C:\\Users\\Sky\\Desktop\\1.in", "r", stdin);    char ch;    row = 1;    clearpos();    STATE = 0;    num = 0;    t1 = 0;    if (showline)        printf("%d: ", row);    while (scanf("%c", &ch),ch!='#')    {        if (inzhushi)        {            if (ch == '\n' && showline)            {                row++;                printf("\n");                printf("%d: ", row);            }            if (inSTATE == -2)//注释起始状态            {                clearpos();                if (ch == '*')                    inSTATE = 0;            }            else if (inSTATE == -1)//有*状态1            {                if (ch == '*')                    inSTATE = 0;                else                    inSTATE = -2;            }            else if (inSTATE == 0)//有*状态2            {                if (ch == '*')                    inSTATE = 0;                else if (ch == '/' )                {                    inzhushi = false;                    STATE = 0;                }                else                    inSTATE = -1;            }            continue;        }        if (STATE == 0)//空状态        {            if (isalpha(ch))            {                STATE = 1;                token[t1++] = ch;            }            else if (isdigit(ch))            {                STATE = 2;                // num = ch - '0';                token[t1++] = ch;            }            else if (ch == '\n' || ch == ' ')            {                STATE = 0;                if (ch == '\n')                {                    huiche = true;                    row++;                }            }            else            {                STATE = 3;                token[t1++] = ch;            }        }        else if (STATE == 1)//变量，关键字状态        {            if (isalpha(ch))            {                token[t1++] = ch;            }            else if (isdigit(ch))            {                token[t1++] = ch;            }            else if (ch == '\n' || ch == ' ')            {                output();                clearpos();                STATE = 0;                if (ch == '\n')                {                    huiche = true;                    row++;                }            }            else            {                output();                clearpos();                STATE = 3;                token[t1++] = ch;            }        }        else if (STATE == 2)//数字状态        {            if (isalpha(ch))            {                token[t1++] = ch;            }            else if (isdigit(ch))            {                // num *= 10;                // num += ch - '0';                token[t1++] = ch;            }            else if (ch == '\n' || ch == ' ')            {                output();                clearpos();                STATE = 0;                if (ch == '\n')                {                    huiche = true;                    row++;                }            }            else            {                output();                clearpos();                STATE = 3;                token[t1++] = ch;            }        }        else if (STATE == 3)//符号集状态        {            if (isalpha(ch))            {                output();                clearpos();                STATE = 1;                token[t1++] = ch;            }            else if (isdigit(ch))            {                output();                clearpos();                STATE = 2;                // num = ch - '0';                token[t1++] = ch;            }            else if (ch == '\n' || ch == ' ')            {                output();                clearpos();                STATE = 0;                if (ch == '\n')                {                    huiche = true;                    row++;                }            }            else            {                token[t1++] = ch;            }        }    }    return 0;}// begin// /*sdas/*d*/// x2231:=2;// if x>9// then x:=2*x+1/3;// end #
0 0