toke.c

早期freebsd实现

	if (strEQ(d,"push")) {
	    yylval.ival = O_PUSH;
	    OPERATOR(PUSH);
	}
	if (strEQ(d,"pop"))
	    OPERATOR(POP);
	if (strEQ(d,"pack"))
	    FL2(O_PACK);
	if (strEQ(d,"package"))
	    OPERATOR(PACKAGE);
	if (strEQ(d,"pipe"))
	    FOP22(O_PIPE_OP);
	break;
    case 'q': case 'Q':
	SNARFWORD;
	if (strEQ(d,"q")) {
	    s = scanstr(s-1, SCAN_DEF);
	    TERM(RSTRING);
	}
	if (strEQ(d,"qq")) {
	    s = scanstr(s-2, SCAN_DEF);
	    TERM(RSTRING);
	}
	if (strEQ(d,"qx")) {
	    s = scanstr(s-2, SCAN_DEF);
	    TERM(RSTRING);
	}
	break;
    case 'r': case 'R':
	SNARFWORD;
	if (strEQ(d,"return"))
	    OLDLOP(O_RETURN);
	if (strEQ(d,"require")) {
	    allstabs = TRUE;		/* must initialize everything since */
	    UNI(O_REQUIRE);		/* we don't know what will be used */
	}
	if (strEQ(d,"reset"))
	    UNI(O_RESET);
	if (strEQ(d,"redo"))
	    LOOPX(O_REDO);
	if (strEQ(d,"rename"))
	    FUN2(O_RENAME);
	if (strEQ(d,"rand"))
	    UNI(O_RAND);
	if (strEQ(d,"rmdir"))
	    UNI(O_RMDIR);
	if (strEQ(d,"rindex"))
	    FUN2x(O_RINDEX);
	if (strEQ(d,"read"))
	    FOP3(O_READ);
	if (strEQ(d,"readdir"))
	    FOP(O_READDIR);
	if (strEQ(d,"rewinddir"))
	    FOP(O_REWINDDIR);
	if (strEQ(d,"recv"))
	    FOP4(O_RECV);
	if (strEQ(d,"reverse"))
	    LOP(O_REVERSE);
	if (strEQ(d,"readlink"))
	    UNI(O_READLINK);
	break;
    case 's': case 'S':
	if (s[1] == '\'') {
	    d = "s";
	    s++;
	}
	else {
	    SNARFWORD;
	}
	if (strEQ(d,"s")) {
	    s = scansubst(s);
	    if (yylval.arg)
		TERM(SUBST);
	    else
		RETURN(1);	/* force error */
	}
	switch (d[1]) {
	case 'a':
	case 'b':
	    break;
	case 'c':
	    if (strEQ(d,"scalar"))
		UNI(O_SCALAR);
	    break;
	case 'd':
	    break;
	case 'e':
	    if (strEQ(d,"select"))
		OPERATOR(SSELECT);
	    if (strEQ(d,"seek"))
		FOP3(O_SEEK);
	    if (strEQ(d,"semctl"))
		FUN4(O_SEMCTL);
	    if (strEQ(d,"semget"))
		FUN3(O_SEMGET);
	    if (strEQ(d,"semop"))
		FUN2(O_SEMOP);
	    if (strEQ(d,"send"))
		FOP3(O_SEND);
	    if (strEQ(d,"setpgrp"))
		FUN2(O_SETPGRP);
	    if (strEQ(d,"setpriority"))
		FUN3(O_SETPRIORITY);
	    if (strEQ(d,"sethostent"))
		FUN1(O_SHOSTENT);
	    if (strEQ(d,"setnetent"))
		FUN1(O_SNETENT);
	    if (strEQ(d,"setservent"))
		FUN1(O_SSERVENT);
	    if (strEQ(d,"setprotoent"))
		FUN1(O_SPROTOENT);
	    if (strEQ(d,"setpwent"))
		FUN0(O_SPWENT);
	    if (strEQ(d,"setgrent"))
		FUN0(O_SGRENT);
	    if (strEQ(d,"seekdir"))
		FOP2(O_SEEKDIR);
	    if (strEQ(d,"setsockopt"))
		FOP4(O_SSOCKOPT);
	    break;
	case 'f':
	case 'g':
	    break;
	case 'h':
	    if (strEQ(d,"shift"))
		TERM(SHIFT);
	    if (strEQ(d,"shmctl"))
		FUN3(O_SHMCTL);
	    if (strEQ(d,"shmget"))
		FUN3(O_SHMGET);
	    if (strEQ(d,"shmread"))
		FUN4(O_SHMREAD);
	    if (strEQ(d,"shmwrite"))
		FUN4(O_SHMWRITE);
	    if (strEQ(d,"shutdown"))
		FOP2(O_SHUTDOWN);
	    break;
	case 'i':
	    if (strEQ(d,"sin"))
		UNI(O_SIN);
	    break;
	case 'j':
	case 'k':
	    break;
	case 'l':
	    if (strEQ(d,"sleep"))
		UNI(O_SLEEP);
	    break;
	case 'm':
	case 'n':
	    break;
	case 'o':
	    if (strEQ(d,"socket"))
		FOP4(O_SOCKET);
	    if (strEQ(d,"socketpair"))
		FOP25(O_SOCKPAIR);
	    if (strEQ(d,"sort")) {
		checkcomma(s,d,"subroutine name");
		d = bufend;
		while (s < d && isSPACE(*s)) s++;
		if (*s == ';' || *s == ')')		/* probably a close */
		    fatal("sort is now a reserved word");
		if (isALPHA(*s) || *s == '_') {
		    /*SUPPRESS 530*/
		    for (d = s; isALNUM(*d); d++) ;
		    strncpy(tokenbuf,s,d-s);
		    tokenbuf[d-s] = '\0';
		    if (strNE(tokenbuf,"keys") &&
			strNE(tokenbuf,"values") &&
			strNE(tokenbuf,"split") &&
			strNE(tokenbuf,"grep") &&
			strNE(tokenbuf,"readdir") &&
			strNE(tokenbuf,"unpack") &&
			strNE(tokenbuf,"do") &&
			strNE(tokenbuf,"eval") &&
			(d >= bufend || isSPACE(*d)) )
			*(--s) = '\\';	/* force next ident to WORD */
		}
		LOP(O_SORT);
	    }
	    break;
	case 'p':
	    if (strEQ(d,"split"))
		TERM(SPLIT);
	    if (strEQ(d,"sprintf"))
		FL(O_SPRINTF);
	    if (strEQ(d,"splice")) {
		yylval.ival = O_SPLICE;
		OPERATOR(PUSH);
	    }
	    break;
	case 'q':
	    if (strEQ(d,"sqrt"))
		UNI(O_SQRT);
	    break;
	case 'r':
	    if (strEQ(d,"srand"))
		UNI(O_SRAND);
	    break;
	case 's':
	    break;
	case 't':
	    if (strEQ(d,"stat"))
		FOP(O_STAT);
	    if (strEQ(d,"study")) {
		sawstudy++;
		LFUN(O_STUDY);
	    }
	    break;
	case 'u':
	    if (strEQ(d,"substr"))
		FUN2x(O_SUBSTR);
	    if (strEQ(d,"sub")) {
		yylval.ival = savestack->ary_fill; /* restore stuff on reduce */
		savelong(&subline);
		saveitem(subname);

		subline = curcmd->c_line;
		d = bufend;
		while (s < d && isSPACE(*s))
		    s++;
		if (isALPHA(*s) || *s == '_' || *s == '\'') {
		    str_sset(subname,curstname);
		    str_ncat(subname,"'",1);
		    for (d = s+1; isALNUM(*d) || *d == '\''; d++)
			/*SUPPRESS 530*/
			;
		    if (d[-1] == '\'')
			d--;
		    str_ncat(subname,s,d-s);
		    *(--s) = '\\';	/* force next ident to WORD */
		}
		else
		    str_set(subname,"?");
		OPERATOR(SUB);
	    }
	    break;
	case 'v':
	case 'w':
	case 'x':
	    break;
	case 'y':
	    if (strEQ(d,"system")) {
		set_csh();
		LOP(O_SYSTEM);
	    }
	    if (strEQ(d,"symlink"))
		FUN2(O_SYMLINK);
	    if (strEQ(d,"syscall"))
		LOP(O_SYSCALL);
	    if (strEQ(d,"sysread"))
		FOP3(O_SYSREAD);
	    if (strEQ(d,"syswrite"))
		FOP3(O_SYSWRITE);
	    break;
	case 'z':
	    break;
	}
	break;
    case 't': case 'T':
	SNARFWORD;
	if (strEQ(d,"tr")) {
	    s = scantrans(s);
	    if (yylval.arg)
		TERM(TRANS);
	    else
		RETURN(1);	/* force error */
	}
	if (strEQ(d,"tell"))
	    FOP(O_TELL);
	if (strEQ(d,"telldir"))
	    FOP(O_TELLDIR);
	if (strEQ(d,"time"))
	    FUN0(O_TIME);
	if (strEQ(d,"times"))
	    FUN0(O_TMS);
	if (strEQ(d,"truncate"))
	    FOP2(O_TRUNCATE);
	break;
    case 'u': case 'U':
	SNARFWORD;
	if (strEQ(d,"using"))
	    OPERATOR(USING);
	if (strEQ(d,"until")) {
	    yylval.ival = curcmd->c_line;
	    OPERATOR(UNTIL);
	}
	if (strEQ(d,"unless")) {
	    yylval.ival = curcmd->c_line;
	    OPERATOR(UNLESS);
	}
	if (strEQ(d,"unlink"))
	    LOP(O_UNLINK);
	if (strEQ(d,"undef"))
	    LFUN(O_UNDEF);
	if (strEQ(d,"unpack"))
	    FUN2(O_UNPACK);
	if (strEQ(d,"utime"))
	    LOP(O_UTIME);
	if (strEQ(d,"umask"))
	    UNI(O_UMASK);
	if (strEQ(d,"unshift")) {
	    yylval.ival = O_UNSHIFT;
	    OPERATOR(PUSH);
	}
	break;
    case 'v': case 'V':
	SNARFWORD;
	if (strEQ(d,"values"))
	    HFUN(O_VALUES);
	if (strEQ(d,"vec")) {
	    sawvec = TRUE;
	    FUN3(O_VEC);
	}
	break;
    case 'w': case 'W':
	SNARFWORD;
	if (strEQ(d,"while")) {
	    yylval.ival = curcmd->c_line;
	    OPERATOR(WHILE);
	}
	if (strEQ(d,"warn"))
	    LOP(O_WARN);
	if (strEQ(d,"wait"))
	    FUN0(O_WAIT);
	if (strEQ(d,"waitpid"))
	    FUN2(O_WAITPID);
	if (strEQ(d,"wantarray")) {
	    yylval.arg = op_new(1);
	    yylval.arg->arg_type = O_ITEM;
	    yylval.arg[1].arg_type = A_WANTARRAY;
	    TERM(RSTRING);
	}
	if (strEQ(d,"write"))
	    FOP(O_WRITE);
	break;
    case 'x': case 'X':
	if (*s == 'x' && isDIGIT(s[1]) && !expectterm) {
	    s++;
	    MOP(O_REPEAT);
	}
	SNARFWORD;
	if (strEQ(d,"x")) {
	    if (!expectterm)
		MOP(O_REPEAT);
	    check_uni();
	}
	break;
    case 'y': case 'Y':
	if (s[1] == '\'') {
	    d = "y";
	    s++;
	}
	else {
	    SNARFWORD;
	}
	if (strEQ(d,"y")) {
	    s = scantrans(s);
	    TERM(TRANS);
	}
	break;
    case 'z': case 'Z':
	SNARFWORD;
	break;
    }
    yylval.cval = savestr(d);
    if (expectterm == 2) {		/* special case: start of statement */
	while (isSPACE(*s)) s++;
	if (*s == ':') {
	    s++;
	    CLINE;
	    OPERATOR(LABEL);
	}
	TERM(WORD);
    }
    expectterm = FALSE;
    if (oldoldbufptr && oldoldbufptr < bufptr) {
	while (isSPACE(*oldoldbufptr))
	    oldoldbufptr++;
	if (*oldoldbufptr == 'p' && strnEQ(oldoldbufptr,"print",5))
	    expectterm = TRUE;
	else if (*oldoldbufptr == 's' && strnEQ(oldoldbufptr,"sort",4))
	    expectterm = TRUE;
    }
    return (CLINE, bufptr = s, (int)WORD);
}

void
checkcomma(s,name,what)
register char *s;
char *name;
char *what;
{
    char *w;

    if (dowarn && *s == ' ' && s[1] == '(') {
	w = index(s,')');
	if (w)
	    for (w++; *w && isSPACE(*w); w++) ;
	if (!w || !*w || !index(";|}", *w))	/* an advisory hack only... */
	    warn("%s (...) interpreted as function",name);
    }
    while (s < bufend && isSPACE(*s))
	s++;
    if (*s == '(')
	s++;
    while (s < bufend && isSPACE(*s))
	s++;
    if (isALPHA(*s) || *s == '_') {
	w = s++;
	while (isALNUM(*s))
	    s++;
	while (s < bufend && isSPACE(*s))
	    s++;
	if (*s == ',') {
	    *s = '\0';
	    w = instr(
	      "tell eof times getlogin wait length shift umask getppid \
	      cos exp int log rand sin sqrt ord wantarray",
	      w);
	    *s = ',';
	    if (w)
		return;
	    fatal("No comma allowed after %s", what);
	}
    }
}

char *
scanident(s,send,dest)
register char *s;
register char *send;
char *dest;
{
    register char *d;
    int brackets = 0;

    reparse = Nullch;
    s++;
    d = dest;
    if (isDIGIT(*s)) {
	while (isDIGIT(*s))
	    *d++ = *s++;
    }
    else {
	while (isALNUM(*s) || *s == '\'')
	    *d++ = *s++;
    }
    while (d > dest+1 && d[-1] == '\'')
	d--,s--;
    *d = '\0';
    d = dest;
    if (!*d) {
	*d = *s++;
	if (*d == '{' /* } */ ) {
	    d = dest;
	    brackets++;
	    while (s < send && brackets) {
		if (!reparse && (d == dest || (*s && isALNUM(*s) ))) {
		    *d++ = *s++;
		    continue;
		}
		else if (!reparse)
		    reparse = s;
		switch (*s++) {
		/* { */
		case '}':
		    brackets--;
		    if (reparse && reparse == s - 1)
			reparse = Nullch;
		    break;
		case '{':   /* } */
		    brackets++;
		    break;
		}
	    }
	    *d = '\0';
	    d = dest;
	}
	else
	    d[1] = '\0';
    }
    if (*d == '^' && (isUPPER(*s) || index("[\\]^_?", *s))) {
#ifdef DEBUGGING
	if (*s == 'D')
	    debug |= 32768;
#endif
	*d = *s++ ^ 64;
    }
    return s;
}

void
scanconst(spat,string,len)
SPAT *spat;
char *string;
int len;
{
    register STR *tmpstr;
    register char *t;
    register char *d;
    register char *e;
    char *origstring = string;
    static char *vert = "|";

    if (ninstr(string, string+len, vert, vert+1))
	return;
    if (*string == '^')
	string++, len--;
    tmpstr = Str_new(86,len);
    str_nset(tmpstr,string,len);
    t = str_get(tmpstr);
    e = t + len;
    tmpstr->str_u.str_useful = 100;
    for (d=t; d < e; ) {
	switch (*d) {
	case '{':
	    if (isDIGIT(d[1]))
		e = d;
	    else
		goto defchar;
	    break;
	case '.': case '[': case '$': case '(': case ')': case '|': case '+':
	case '^':
	    e = d;
	    break;
	case '\\':
	    if (d[1] && index("wWbB0123456789sSdDlLuUExc",d[1])) {
		e = d;
		break;
	    }
	    Move(d+1,d,e-d,char);
	    e--;
	    switch(*d) {
	    case 'n':
		*d = '\n';
		break;
	    case 't':
		*d = '\t';
		break;
	    case 'f':
		*d = '\f';
		break;
	    case 'r':
		*d = '\r';
		break;
	    case 'e':
		*d = '\033';
		break;
	    case 'a':
		*d = '\007';
		break;
	    }