lex.cpp

一个小型C语言编译器

                        else
                            n = (n<<4) + d;
                    }
                    if ((char *)rcp - token <= 2)
                        //error("invalid hexadecimal constant `%S'\n", token, (char *)rcp-token);
                        printf("invalid hexadecimal constant `%S'\n", token, (char *)rcp-token);
                    cp = rcp;
                    //tsym = icon(n, overflow, 16);
                } else if (*token == '0') {
                    int err = 0, overflow = 0;
                    for ( ; map[*rcp]&DIGIT; rcp++) {
                        if (*rcp == '8' || *rcp == '9')
                            err = 1;
                        if (n&~((unsigned)-1 >> 3))
                            overflow = 1;
                        else
                            n = (n<<3) + (unsigned)(*rcp - '0');
                    }
                    if (*rcp == '.' || *rcp == 'e' || *rcp == 'E') {
                        cp   = rcp;
                        //tsym = fcon();
                        fcon();
                        return FCON;
                    }       
                    cp = rcp;
                    //tsym = icon(n, overflow, 8);
                    if (err)
                        //error("invalid octal constant `%S'\n", token, (char*)cp-token);
                        printf("invalid octal constant `%S'\n", token, (char*)cp-token);
                } else {
                    int overflow = 0;
                    for (n = *token - '0'; map[*rcp]&DIGIT; ) {
                        int d = *rcp++ - '0';
                        if (n > ((unsigned)UINT_MAX - d)/10)
                            overflow = 1;
                        else
                            n = 10*n + d;
                    }
                    if (*rcp == '.' || *rcp == 'e' || *rcp == 'E') {
                        cp = rcp;
                        //tsym = fcon();
                        fcon();
                        return FCON;
                    }       
                    cp = rcp;
                    //tsym = icon(n, overflow, 10);
                }
                return ICON;
            }
            case '.':
                if (rcp[0] == '.' && rcp[1] == '.') {
                    cp += 2;
                    return ELLIPSIS;
                }
                if ((map[*rcp]&DIGIT) == 0)
                    return '.';
                if (limit - rcp < MAXLINE) {
                    cp = rcp - 1;
                    fillbuf();
                    rcp = ++cp;
                }
                assert(cp == rcp);
                cp = rcp - 1;
                token = (char *)cp;
                //tsym = fcon();
                fcon();
                return FCON;
scon:   
            case '\'': case '"': 
            {
                static char cbuf[BUFSIZE+1];
                char *s = cbuf;
                int nbad = 0;
                *s++ = *--cp;
                do {
                    cp++;
                    while (*cp != cbuf[0]) {
                        int c;
                        if (map[*cp]&NEWLINE) {
                            if (cp < limit)
                                break;
                            cp++;
                            nextline();
                            if (cp == limit)
                                break;
                            continue;
                        }
                        c = *cp++;
                        if (c == '\\') {
                            if (map[*cp]&NEWLINE) {
                                if (cp < limit)
                                    break;
                                cp++;
                                nextline();
                            }
                            if (limit - cp < MAXTOKEN)
                                fillbuf();
                            c = backslash(cbuf[0]);
                        } else if (map[c] == 0)
                            nbad++;
                        if (s < &cbuf[sizeof cbuf] - 2)
                            *s++ = c;
                    }
                    if (*cp == cbuf[0])
                        cp++;
                    else
                      //error("missing %c\n", cbuf[0]);
                      printf("missing %c\n", cbuf[0]);
                } while (cbuf[0] == '"' && getchr() == '"');
                *s++ = 0;
                if (s >= &cbuf[sizeof cbuf])
                    //error("%s literal too long\n", cbuf[0] == '"' ? "string" : "character");
                    printf("%s literal too long\n", cbuf[0] == '"' ? "string" : "character");
                //if (Aflag >= 2 && cbuf[0] == '"' && s - cbuf - 1 > 509)
                //    warning("more than 509 characters in a string literal\n");
                //if (Aflag >= 2 && nbad)
                //    warning("%s literal contains non-portable characters\n", cbuf[0] == '"' ? "string" : "character");
                token = cbuf;
                //tsym = &tval;
                if (cbuf[0] == '"') {
                    //tval.type = array(chartype, s - cbuf - 1, 0);
                    //tval.u.c.v.p = cbuf + 1;
                    return SCON;
                } else {
                    if (s - cbuf > 3)
                        //warning("excess characters in multibyte character literal `%S' ignored\n", token, (char*)cp-token);
                        printf("excess characters in multibyte character literal `%S' ignored\n", token, (char*)cp-token);
                    else if (s - cbuf <= 2)
                        //error("missing '\n");
                        printf("missing '\n");
                    //tval.type = inttype;
                    //tval.u.c.v.i = cbuf[1];
                    return ICON;
                }
            }
            case 'a':
                if (rcp[0] == 'u'
                &&  rcp[1] == 't'
                &&  rcp[2] == 'o'
                && !(map[rcp[3]]&(DIGIT|LETTER))) {
                    cp = rcp + 3;
                    return AUTO;
                }
                goto id;
            case 'b':
                if (rcp[0] == 'r'
                &&  rcp[1] == 'e'
                &&  rcp[2] == 'a'
                &&  rcp[3] == 'k'
                && !(map[rcp[4]]&(DIGIT|LETTER))) {
                    cp = rcp + 4;
                    return BREAK;
                }
                goto id;
            case 'c':
                if (rcp[0] == 'a'
                &&  rcp[1] == 's'
                &&  rcp[2] == 'e'
                && !(map[rcp[3]]&(DIGIT|LETTER))) {
                    cp = rcp + 3;
                    return CASE;
                }
                if (rcp[0] == 'h'
                &&  rcp[1] == 'a'
                &&  rcp[2] == 'r'
                && !(map[rcp[3]]&(DIGIT|LETTER))) {
                    cp = rcp + 3;
                    //tsym = chartype->u.sym;
                    return CHAR;
                }
                if (rcp[0] == 'o'
                &&  rcp[1] == 'n'
                &&  rcp[2] == 's'
                &&  rcp[3] == 't'
                && !(map[rcp[4]]&(DIGIT|LETTER))) {
                    cp = rcp + 4;
                    return CONST;
                }
                if (rcp[0] == 'o'
                &&  rcp[1] == 'n'
                &&  rcp[2] == 't'
                &&  rcp[3] == 'i'
                &&  rcp[4] == 'n'
                &&  rcp[5] == 'u'
                &&  rcp[6] == 'e'
                && !(map[rcp[7]]&(DIGIT|LETTER))) {
                    cp = rcp + 7;
                    return CONTINUE;
                }
                goto id;
            case 'd':
                if (rcp[0] == 'e'
                &&  rcp[1] == 'f'
                &&  rcp[2] == 'a'
                &&  rcp[3] == 'u'
                &&  rcp[4] == 'l'
                &&  rcp[5] == 't'
                && !(map[rcp[6]]&(DIGIT|LETTER))) {
                    cp = rcp + 6;
                    return DEFAULT;
                }
                if (rcp[0] == 'o'
                &&  rcp[1] == 'u'
                &&  rcp[2] == 'b'
                &&  rcp[3] == 'l'
                &&  rcp[4] == 'e'
                && !(map[rcp[5]]&(DIGIT|LETTER))) {
                    cp = rcp + 5;
                    //tsym = doubletype->u.sym;
                    return DOUBLE;
                }
                if (rcp[0] == 'o'
                && !(map[rcp[1]]&(DIGIT|LETTER))) {
                    cp = rcp + 1;
                    return DO;
                }
                goto id;
            case 'e':
                if (rcp[0] == 'l'
                &&  rcp[1] == 's'
                &&  rcp[2] == 'e'
                && !(map[rcp[3]]&(DIGIT|LETTER))) {
                    cp = rcp + 3;
                    return ELSE;
                }
                if (rcp[0] == 'n'
                &&  rcp[1] == 'u'
                &&  rcp[2] == 'm'
                && !(map[rcp[3]]&(DIGIT|LETTER))) {
                    cp = rcp + 3;
                    return ENUM;
                }
                if (rcp[0] == 'x'
                &&  rcp[1] == 't'
                &&  rcp[2] == 'e'
                &&  rcp[3] == 'r'
                &&  rcp[4] == 'n'
                && !(map[rcp[5]]&(DIGIT|LETTER))) {
                    cp = rcp + 5;
                    return EXTERN;
                }
                goto id;
            case 'f':
                if (rcp[0] == 'l'
                &&  rcp[1] == 'o'
                &&  rcp[2] == 'a'
                &&  rcp[3] == 't'
                && !(map[rcp[4]]&(DIGIT|LETTER))) {
                    cp = rcp + 4;
                    //tsym = floattype->u.sym;
                    return FLOAT;
                }
                if (rcp[0] == 'o'
                &&  rcp[1] == 'r'
                && !(map[rcp[2]]&(DIGIT|LETTER))) {
                    cp = rcp + 2;
                    return FOR;
                }
                goto id;
            case 'g':
                if (rcp[0] == 'o'
                &&  rcp[1] == 't'
                &&  rcp[2] == 'o'
                && !(map[rcp[3]]&(DIGIT|LETTER))) {
                    cp = rcp + 3;
                    return GOTO;
                }
                goto id;
            case 'l':
                if (rcp[0] == 'o'
                &&  rcp[1] == 'n'
                &&  rcp[2] == 'g'
                && !(map[rcp[3]]&(DIGIT|LETTER))) {
                    cp = rcp + 3;
                    //tsym = longtype->u.sym;
                    return LONG;
                }
                goto id;
            case 'r':
                if (rcp[0] == 'e'
                &&  rcp[1] == 'g'
                &&  rcp[2] == 'i'
                &&  rcp[3] == 's'
                &&  rcp[4] == 't'
                &&  rcp[5] == 'e'
                &&  rcp[6] == 'r'
                && !(map[rcp[7]]&(DIGIT|LETTER))) {
                    cp = rcp + 7;
                    return REGISTER;
                }
                if (rcp[0] == 'e'
                &&  rcp[1] == 't'
                &&  rcp[2] == 'u'
                &&  rcp[3] == 'r'
                &&  rcp[4] == 'n'
                && !(map[rcp[5]]&(DIGIT|LETTER))) {
                    cp = rcp + 5;
                    return RETURN;
                }
                goto id;
            case 's':