/* toke.c
*
- * Copyright (c) 1991-1994, Larry Wall
+ * Copyright (c) 1991-1997, Larry Wall
*
* You may distribute under the terms of either the GNU General Public
* License or the Artistic License, as specified in the README file.
static char *linestart; /* beg. of most recently read line */
+static char pending_ident; /* pending identifier lookup */
+
static struct {
I32 super_state; /* lexer state to save */
I32 sub_inwhat; /* "lex_inwhat" to use */
char *what;
char *s;
{
- char tmpbuf[128];
char *oldbp = bufptr;
bool is_first = (oldbufptr == linestart);
+ char *msg;
+
bufptr = s;
- sprintf(tmpbuf, "%s found where operator expected", what);
- yywarn(tmpbuf);
+ New(890, msg, strlen(what) + 40, char);
+ sprintf(msg, "%s found where operator expected", what);
+ yywarn(msg);
+ Safefree(msg);
if (is_first)
warn("\t(Missing semicolon on previous line?)\n");
else if (oldoldbufptr && isIDFIRST(*oldoldbufptr)) {
}
else if (multi_close < 32 || multi_close == 127) {
*tmpbuf = '^';
- tmpbuf[1] = multi_close ^ 64;
+ tmpbuf[1] = toCTRL(multi_close);
s = "\\n";
tmpbuf[2] = '\0';
s = tmpbuf;
void
lex_end()
{
+ doextract = FALSE;
}
static void
#define LOP(f,x) return lop(f,x,s)
static I32
-lop(f,x,s)
+lop
+#ifdef CAN_PROTOTYPE
+ (I32 f, expectation x, char *s)
+#else
+ (f,x,s)
I32 f;
expectation x;
char *s;
+#endif /* CAN_PROTOTYPE */
{
yylval.ival = f;
CLINE;
return THING;
}
if (op_type == OP_CONST || op_type == OP_READLINE) {
- yylval.opval = (OP*)newSVOP(op_type, 0, q(lex_stuff));
+ SV *sv = q(lex_stuff);
+ STRLEN len;
+ char *p = SvPV(sv, len);
+ yylval.opval = (OP*)newSVOP(op_type, 0, newSVpv(p, len));
+ SvREFCNT_dec(sv);
lex_stuff = Nullsv;
return THING;
}
continue;
case 'c':
s++;
- *d = *s++;
- if (isLOWER(*d))
- *d = toUPPER(*d);
- *d++ ^= 64;
+ len = *s++;
+ *d++ = toCTRL(len);
continue;
case 'b':
*d++ = '\b';
if (gv) {
if (GvIO(gv))
return 0;
- if (!GvCV(gv))
+ if (!GvCVu(gv))
gv = 0;
}
s = scan_word(s, tmpbuf, TRUE, &len);
}
if (!keyword(tmpbuf, len)) {
indirgv = gv_fetchpv(tmpbuf,FALSE, SVt_PVCV);
- if (indirgv && GvCV(indirgv))
+ if (indirgv && GvCVu(indirgv))
return 0;
/* filehandle or package name makes it a method */
if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
die("Can't upgrade filter_add data to SVt_PVIO");
IoDIRP(datasv) = (DIR*)funcp; /* stash funcp into spare field */
if (filter_debug)
- warn("filter_add func %lx (%s)", funcp, SvPV(datasv,na));
+ warn("filter_add func %p (%s)", funcp, SvPV(datasv,na));
av_unshift(rsfp_filters, 1);
av_store(rsfp_filters, 0, datasv) ;
return(datasv);
filter_t funcp;
{
if (filter_debug)
- warn("filter_del func %lx", funcp);
+ warn("filter_del func %p", funcp);
if (!rsfp_filters || AvFILL(rsfp_filters)<0)
return;
/* if filter is on top of stack (usual case) just pop it off */
/* Get function pointer hidden within datasv */
funcp = (filter_t)IoDIRP(datasv);
if (filter_debug)
- warn("filter_read %d: via function %lx (%s)\n",
+ warn("filter_read %d: via function %p (%s)\n",
idx, funcp, SvPV(datasv,na));
/* Call function. The function is expected to */
/* call "FILTER_READ(idx+1, buf_sv)" first. */
{ "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "TERMBLOCK" };
#endif
-extern int yychar; /* last token */
+EXT int yychar; /* last token */
int
yylex()
register I32 tmp;
STRLEN len;
+ if (pending_ident) {
+ char pit = pending_ident;
+ pending_ident = 0;
+
+ if (in_my) {
+ if (strchr(tokenbuf,':'))
+ croak(no_myglob,tokenbuf);
+ yylval.opval = newOP(OP_PADANY, 0);
+ yylval.opval->op_targ = pad_allocmy(tokenbuf);
+ return PRIVATEREF;
+ }
+
+ if (!strchr(tokenbuf,':') && (tmp = pad_findmy(tokenbuf))) {
+ if (last_lop_op == OP_SORT &&
+ tokenbuf[0] == '$' &&
+ (tokenbuf[1] == 'a' || tokenbuf[1] == 'b')
+ && !tokenbuf[2])
+ {
+ for (d = in_eval ? oldoldbufptr : linestart;
+ d < bufend && *d != '\n';
+ d++)
+ {
+ if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
+ croak("Can't use \"my %s\" in sort comparison",
+ tokenbuf);
+ }
+ }
+ }
+
+ yylval.opval = newOP(OP_PADANY, 0);
+ yylval.opval->op_targ = tmp;
+ return PRIVATEREF;
+ }
+
+ /* Force them to make up their mind on "@foo". */
+ if (pit == '@' && lex_state != LEX_NORMAL && !lex_brackets) {
+ GV *gv = gv_fetchpv(tokenbuf+1, FALSE, SVt_PVAV);
+ if (!gv || ((tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv))) {
+ char tmpbuf[1024];
+ sprintf(tmpbuf, "In string, %s now must be written as \\%s",
+ tokenbuf, tokenbuf);
+ yyerror(tmpbuf);
+ }
+ }
+
+ yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(tokenbuf+1, 0));
+ yylval.opval->op_private = OPpCONST_ENTERED;
+ gv_fetchpv(tokenbuf+1, in_eval ? GV_ADDMULTI : TRUE,
+ ((tokenbuf[0] == '$') ? SVt_PV
+ : (tokenbuf[0] == '@') ? SVt_PVAV
+ : SVt_PVHV));
+ return WORD;
+ }
+
switch (lex_state) {
#ifdef COMMENTARY
case LEX_NORMAL: /* Some compilers will produce faster */
s = bufptr;
Aop(OP_CONCAT);
}
- else
- return yylex();
- break;
+ return yylex();
case LEX_INTERPENDMAYBE:
if (intuit_more(bufptr)) {
retry:
switch (*s) {
default:
- warn("Unrecognized character \\%03o ignored", *s++ & 255);
- goto retry;
+ croak("Unrecognized character \\%03o", *s & 255);
case 4:
case 26:
goto fake_eof; /* emulate EOF on ^D or ^Z */
sv_catpv(linestr, "LINE: while (<>) {");
if (minus_l)
sv_catpv(linestr,"chomp;");
- if (minus_a){
- if (minus_F){
- char tmpbuf1[50];
- if ( splitstr[0] == '/' ||
- splitstr[0] == '\'' ||
- splitstr[0] == '"' )
- sprintf( tmpbuf1, "@F=split(%s);", splitstr );
- else
- sprintf( tmpbuf1, "@F=split('%s');", splitstr );
- sv_catpv(linestr,tmpbuf1);
+ if (minus_a) {
+ GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
+ if (gv)
+ GvIMPORTED_AV_on(gv);
+ if (minus_F) {
+ char *tmpbuf1;
+ New(201, tmpbuf1, strlen(splitstr) * 2 + 20, char);
+ if (strchr("/'\"", *splitstr)
+ && strchr(splitstr + 1, *splitstr))
+ sprintf(tmpbuf1, "@F=split(%s);", splitstr);
+ else {
+ char delim;
+ s = "'~#\200\1'"; /* surely one char is unused...*/
+ while (s[1] && strchr(splitstr, *s)) s++;
+ delim = *s;
+ sprintf(tmpbuf1, "@F=split(%s%c",
+ "q" + (delim == '\''), delim);
+ d = tmpbuf1 + strlen(tmpbuf1);
+ for (s = splitstr; *s; ) {
+ if (*s == '\\')
+ *d++ = '\\';
+ *d++ = *s++;
+ }
+ sprintf(d, "%c);", delim);
+ }
+ sv_catpv(linestr,tmpbuf1);
+ Safefree(tmpbuf1);
}
else
sv_catpv(linestr,"@F=split(' ');");
s++;
if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
s++;
- if (!in_eval && *s == '#' && s[1] == '!') {
+ d = Nullch;
+ if (!in_eval) {
+ if (*s == '#' && *(s+1) == '!')
+ d = s + 2;
+#ifdef ALTERNATE_SHEBANG
+ else {
+ static char as[] = ALTERNATE_SHEBANG;
+ if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
+ d = s + (sizeof(as) - 1);
+ }
+#endif /* ALTERNATE_SHEBANG */
+ }
+ if (d) {
+ char *ipath;
+ char *ipathend;
+
+ while (isSPACE(*d))
+ d++;
+ ipath = d;
+ while (*d && !isSPACE(*d))
+ d++;
+ ipathend = d;
+
+#ifdef ARG_ZERO_IS_SCRIPT
+ if (ipathend > ipath) {
+ /*
+ * HP-UX (at least) sets argv[0] to the script name,
+ * which makes $^X incorrect. And Digital UNIX and Linux,
+ * at least, set argv[0] to the basename of the Perl
+ * interpreter. So, having found "#!", we'll set it right.
+ */
+ SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
+ assert(SvPOK(x) || SvGMAGICAL(x));
+ if (sv_eq(x, GvSV(curcop->cop_filegv))) {
+ sv_setpvn(x, ipath, ipathend - ipath);
+ SvSETMAGIC(x);
+ }
+ TAINT_NOT; /* $^X is always tainted, but that's OK */
+ }
+#endif /* ARG_ZERO_IS_SCRIPT */
+
+ /*
+ * Look for options.
+ */
d = instr(s,"perl -");
if (!d)
d = instr(s,"perl");
+#ifdef ALTERNATE_SHEBANG
+ /*
+ * If the ALTERNATE_SHEBANG on this system starts with a
+ * character that can be part of a Perl expression, then if
+ * we see it but not "perl", we're probably looking at the
+ * start of Perl code, not a request to hand off to some
+ * other interpreter. Similarly, if "perl" is there, but
+ * not in the first 'word' of the line, we assume the line
+ * contains the start of the Perl program.
+ */
+ if (d && *s != '#') {
+ char *c = ipath;
+ while (*c && !strchr("; \t\r\n\f\v#", *c))
+ c++;
+ if (c < d)
+ d = Nullch; /* "perl" not in first word; ignore */
+ else
+ *s = '#'; /* Don't try to parse shebang line */
+ }
+#endif /* ALTERNATE_SHEBANG */
if (!d &&
+ *s == '#' &&
+ ipathend > ipath &&
!minus_c &&
!instr(s,"indir") &&
instr(origargv[0],"perl"))
{
char **newargv;
- char *cmd;
- s += 2;
- if (*s == ' ')
- s++;
- cmd = s;
- while (s < bufend && !isSPACE(*s))
- s++;
- *s++ = '\0';
+ *ipathend = '\0';
+ s = ipathend + 1;
while (s < bufend && isSPACE(*s))
s++;
if (s < bufend) {
}
else
newargv = origargv;
- newargv[0] = cmd;
- execv(cmd,newargv);
- croak("Can't exec %s", cmd);
+ newargv[0] = ipath;
+ execv(ipath, newargv);
+ croak("Can't exec %s", ipath);
}
if (d) {
int oldpdb = perldb;
while (*d == ' ' || *d == '\t') d++;
if (*d++ == '-') {
- while (d = moreswitches(d)) ;
+ do {
+ if (*d == 'M' || *d == 'm') {
+ char *m = d;
+ while (*d && !isSPACE(*d)) d++;
+ croak("Too late for \"-%.*s\" option",
+ (int)(d - m), m);
+ }
+ d = moreswitches(d);
+ } while (d);
if (perldb && !oldpdb ||
( minus_n || minus_p ) && !(oldn || oldp) )
/* if we have already added "LINE: while (<>) {",
return yylex();
}
goto retry;
- case ' ': case '\t': case '\f': case '\r': case 013:
+ case '\r':
+ warn("Illegal character \\%03o (carriage return)", '\r');
+ croak(
+ "(Maybe you didn't strip carriage returns after a network transfer?)\n");
+ case ' ': case '\t': case '\f': case 013:
s++;
goto retry;
case '#':
if (strnEQ(s,"=>",2)) {
if (dowarn)
warn("Ambiguous use of -%c => resolved to \"-%c\" =>",
- tmp, tmp);
+ (int)tmp, (int)tmp);
s = force_word(bufptr,WORD,FALSE,FALSE,FALSE);
OPERATOR('-'); /* unary minus */
}
case 'A': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTATIME);
case 'C': gv_fetchpv("\024",TRUE, SVt_PV); FTST(OP_FTCTIME);
default:
- croak("Unrecognized file test: -%c", tmp);
+ croak("Unrecognized file test: -%c", (int)tmp);
break;
}
}
Mop(OP_MULTIPLY);
case '%':
- if (expect != XOPERATOR) {
- s = scan_ident(s, bufend, tokenbuf + 1, TRUE);
- if (tokenbuf[1]) {
- expect = XOPERATOR;
- tokenbuf[0] = '%';
- if (in_my) {
- if (strchr(tokenbuf,':'))
- croak(no_myglob,tokenbuf);
- nextval[nexttoke].opval = newOP(OP_PADANY, 0);
- nextval[nexttoke].opval->op_targ = pad_allocmy(tokenbuf);
- force_next(PRIVATEREF);
- TERM('%');
- }
- if (!strchr(tokenbuf,':')) {
- if (tmp = pad_findmy(tokenbuf)) {
- nextval[nexttoke].opval = newOP(OP_PADANY, 0);
- nextval[nexttoke].opval->op_targ = tmp;
- force_next(PRIVATEREF);
- TERM('%');
- }
- }
- force_ident(tokenbuf + 1, *tokenbuf);
- }
- else
- PREREF('%');
- TERM('%');
+ if (expect == XOPERATOR) {
+ ++s;
+ Mop(OP_MODULO);
+ }
+ tokenbuf[0] = '%';
+ s = scan_ident(s, bufend, tokenbuf+1, TRUE);
+ if (!tokenbuf[1]) {
+ if (s == bufend)
+ yyerror("Final % should be \\% or %name");
+ PREREF('%');
}
- ++s;
- Mop(OP_MODULO);
+ pending_ident = '%';
+ TERM('%');
case '^':
s++;
else
lex_brackstack[lex_brackets++] = XOPERATOR;
OPERATOR(HASHBRACK);
- break;
case XOPERATOR:
while (s < bufend && (*s == ' ' || *s == '\t'))
s++;
- if (s < bufend && (isALPHA(*s) || *s == '_')) {
- d = scan_word(s, tokenbuf, FALSE, &len);
+ d = s;
+ tokenbuf[0] = '\0';
+ if (d < bufend && *d == '-') {
+ tokenbuf[0] = '-';
+ d++;
+ while (d < bufend && (*d == ' ' || *d == '\t'))
+ d++;
+ }
+ if (d < bufend && isIDFIRST(*d)) {
+ d = scan_word(d, tokenbuf + 1, FALSE, &len);
while (d < bufend && (*d == ' ' || *d == '\t'))
d++;
if (*d == '}') {
+ char minus = (tokenbuf[0] == '-');
if (dowarn &&
- (keyword(tokenbuf, len) ||
- perl_get_cv(tokenbuf, FALSE) ))
+ (keyword(tokenbuf + 1, len) ||
+ (minus && len == 1 && isALPHA(tokenbuf[1])) ||
+ perl_get_cv(tokenbuf + 1, FALSE) ))
warn("Ambiguous use of {%s} resolved to {\"%s\"}",
- tokenbuf, tokenbuf);
- s = force_word(s,WORD,FALSE,TRUE,FALSE);
+ tokenbuf + !minus, tokenbuf + !minus);
+ s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
+ if (minus)
+ force_next('-');
}
}
/* FALL THROUGH */
bufptr = s;
return yylex(); /* ignore fake brackets */
}
- if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
+ if (*s == '-' && s[1] == '>')
+ lex_state = LEX_INTERPENDMAYBE;
+ else if (*s != '[' && *s != '{')
lex_state = LEX_INTERPEND;
}
}
if (tmp == '~')
PMop(OP_MATCH);
if (dowarn && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
- warn("Reversed %c= operator",tmp);
+ warn("Reversed %c= operator",(int)tmp);
s--;
if (expect == XSTATE && isALPHA(tmp) &&
(s == linestart+1 || s[-2] == '\n') )
Rop(OP_GT);
case '$':
- if (s[1] == '#' && (isALPHA(s[2]) || strchr("_{$:", s[2]))) {
- s = scan_ident(s+1, bufend, tokenbuf+1, FALSE);
- if (expect == XOPERATOR) {
- if (lex_formbrack && lex_brackets == lex_formbrack) {
- expect = XTERM;
- depcom();
- return ','; /* grandfather non-comma-format format */
- }
- else
- no_op("Array length",s);
+ CLINE;
+
+ if (expect == XOPERATOR) {
+ if (lex_formbrack && lex_brackets == lex_formbrack) {
+ expect = XTERM;
+ depcom();
+ return ','; /* grandfather non-comma-format format */
}
- else if (!tokenbuf[1])
+ }
+
+ if (s[1] == '#' && (isALPHA(s[2]) || strchr("_{$:", s[2]))) {
+ if (expect == XOPERATOR)
+ no_op("Array length", bufptr);
+ tokenbuf[0] = '@';
+ s = scan_ident(s+1, bufend, tokenbuf+1, FALSE);
+ if (!tokenbuf[1])
PREREF(DOLSHARP);
- if (!strchr(tokenbuf+1,':')) {
- tokenbuf[0] = '@';
- if (tmp = pad_findmy(tokenbuf)) {
- nextval[nexttoke].opval = newOP(OP_PADANY, 0);
- nextval[nexttoke].opval->op_targ = tmp;
- expect = XOPERATOR;
- force_next(PRIVATEREF);
- TOKEN(DOLSHARP);
- }
- }
expect = XOPERATOR;
- force_ident(tokenbuf+1, *tokenbuf);
+ pending_ident = '#';
TOKEN(DOLSHARP);
}
+
+ if (expect == XOPERATOR)
+ no_op("Scalar", bufptr);
+ tokenbuf[0] = '$';
s = scan_ident(s, bufend, tokenbuf+1, FALSE);
- if (expect == XOPERATOR) {
- if (lex_formbrack && lex_brackets == lex_formbrack) {
- expect = XTERM;
- depcom();
- return ','; /* grandfather non-comma-format format */
- }
- else
- no_op("Scalar",s);
+ if (!tokenbuf[1]) {
+ if (s == bufend)
+ yyerror("Final $ should be \\$ or $name");
+ PREREF('$');
}
- if (tokenbuf[1]) {
- expectation oldexpect = expect;
- /* This kludge not intended to be bulletproof. */
- if (tokenbuf[1] == '[' && !tokenbuf[2]) {
- yylval.opval = newSVOP(OP_CONST, 0,
- newSViv((IV)compiling.cop_arybase));
- yylval.opval->op_private = OPpCONST_ARYBASE;
- TERM(THING);
- }
- tokenbuf[0] = '$';
- if (dowarn) {
- char *t;
- if (*s == '[' && oldexpect != XREF) {
- for (t = s+1; isSPACE(*t) || isALNUM(*t) || *t == '$'; t++) ;
+ /* This kludge not intended to be bulletproof. */
+ if (tokenbuf[1] == '[' && !tokenbuf[2]) {
+ yylval.opval = newSVOP(OP_CONST, 0,
+ newSViv((IV)compiling.cop_arybase));
+ yylval.opval->op_private = OPpCONST_ARYBASE;
+ TERM(THING);
+ }
+
+ d = s;
+ if (lex_state == LEX_NORMAL)
+ s = skipspace(s);
+
+ if ((expect != XREF || oldoldbufptr == last_lop) && intuit_more(s)) {
+ char *t;
+ if (*s == '[') {
+ tokenbuf[0] = '@';
+ if (dowarn) {
+ for(t = s + 1;
+ isSPACE(*t) || isALNUM(*t) || *t == '$';
+ t++) ;
if (*t++ == ',') {
bufptr = skipspace(bufptr);
- while (t < bufend && *t != ']') t++;
+ while (t < bufend && *t != ']')
+ t++;
warn("Multidimensional syntax %.*s not supported",
- t-bufptr+1, bufptr);
+ (t - bufptr) + 1, bufptr);
}
}
- if (*s == '{' && strEQ(tokenbuf, "$SIG") &&
- (t = strchr(s,'}')) && (t = strchr(t,'='))) {
+ }
+ else if (*s == '{') {
+ tokenbuf[0] = '%';
+ if (dowarn && strEQ(tokenbuf+1, "SIG") &&
+ (t = strchr(s, '}')) && (t = strchr(t, '=')))
+ {
char tmpbuf[1024];
STRLEN len;
for (t++; isSPACE(*t); t++) ;
}
}
}
- expect = XOPERATOR;
- if (lex_state == LEX_NORMAL && isSPACE(*s)) {
- bool islop = (last_lop == oldoldbufptr);
- s = skipspace(s);
- if (!islop || last_lop_op == OP_GREPSTART)
- expect = XOPERATOR;
- else if (strchr("$@\"'`q", *s))
- expect = XTERM; /* e.g. print $fh "foo" */
- else if (strchr("&*<%", *s) && isIDFIRST(s[1]))
- expect = XTERM; /* e.g. print $fh &sub */
- else if (isDIGIT(*s))
- expect = XTERM; /* e.g. print $fh 3 */
- else if (*s == '.' && isDIGIT(s[1]))
- expect = XTERM; /* e.g. print $fh .3 */
- else if (strchr("/?-+", *s) && !isSPACE(s[1]))
- expect = XTERM; /* e.g. print $fh -1 */
- else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]))
- expect = XTERM; /* print $fh <<"EOF" */
- }
- if (in_my) {
- if (strchr(tokenbuf,':'))
- croak(no_myglob,tokenbuf);
- nextval[nexttoke].opval = newOP(OP_PADANY, 0);
- nextval[nexttoke].opval->op_targ = pad_allocmy(tokenbuf);
- force_next(PRIVATEREF);
- }
- else if (!strchr(tokenbuf,':')) {
- if (oldexpect != XREF || oldoldbufptr == last_lop) {
- if (intuit_more(s)) {
- if (*s == '[')
- tokenbuf[0] = '@';
- else if (*s == '{')
- tokenbuf[0] = '%';
- }
- }
- if (tmp = pad_findmy(tokenbuf)) {
- if (last_lop_op == OP_SORT &&
- !tokenbuf[2] && *tokenbuf =='$' &&
- tokenbuf[1] <= 'b' && tokenbuf[1] >= 'a')
- {
- for (d = in_eval ? oldoldbufptr : linestart;
- d < bufend && *d != '\n';
- d++)
- {
- if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
- croak("Can't use \"my %s\" in sort comparison",
- tokenbuf);
- }
- }
- }
- nextval[nexttoke].opval = newOP(OP_PADANY, 0);
- nextval[nexttoke].opval->op_targ = tmp;
- force_next(PRIVATEREF);
+ }
+
+ expect = XOPERATOR;
+ if (lex_state == LEX_NORMAL && isSPACE(*d)) {
+ bool islop = (last_lop == oldoldbufptr);
+ if (!islop || last_lop_op == OP_GREPSTART)
+ expect = XOPERATOR;
+ else if (strchr("$@\"'`q", *s))
+ expect = XTERM; /* e.g. print $fh "foo" */
+ else if (strchr("&*<%", *s) && isIDFIRST(s[1]))
+ expect = XTERM; /* e.g. print $fh &sub */
+ else if (isIDFIRST(*s)) {
+ char tmpbuf[1024];
+ scan_word(s, tmpbuf, TRUE, &len);
+ if (keyword(tmpbuf, len))
+ expect = XTERM; /* e.g. print $fh length() */
+ else {
+ GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
+ if (gv && GvCVu(gv))
+ expect = XTERM; /* e.g. print $fh subr() */
}
- else
- force_ident(tokenbuf+1, *tokenbuf);
}
- else
- force_ident(tokenbuf+1, *tokenbuf);
- }
- else {
- if (s == bufend)
- yyerror("Final $ should be \\$ or $name");
- PREREF('$');
- }
+ else if (isDIGIT(*s))
+ expect = XTERM; /* e.g. print $fh 3 */
+ else if (*s == '.' && isDIGIT(s[1]))
+ expect = XTERM; /* e.g. print $fh .3 */
+ else if (strchr("/?-+", *s) && !isSPACE(s[1]))
+ expect = XTERM; /* e.g. print $fh -1 */
+ else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]))
+ expect = XTERM; /* print $fh <<"EOF" */
+ }
+ pending_ident = '$';
TOKEN('$');
case '@':
- s = scan_ident(s, bufend, tokenbuf+1, FALSE);
if (expect == XOPERATOR)
- no_op("Array",s);
- if (tokenbuf[1]) {
- GV* gv;
-
- tokenbuf[0] = '@';
- expect = XOPERATOR;
- if (in_my) {
- if (strchr(tokenbuf,':'))
- croak(no_myglob,tokenbuf);
- nextval[nexttoke].opval = newOP(OP_PADANY, 0);
- nextval[nexttoke].opval->op_targ = pad_allocmy(tokenbuf);
- force_next(PRIVATEREF);
- TERM('@');
- }
- else if (!strchr(tokenbuf,':')) {
- if (intuit_more(s)) {
- if (*s == '{')
- tokenbuf[0] = '%';
- }
- if (tmp = pad_findmy(tokenbuf)) {
- nextval[nexttoke].opval = newOP(OP_PADANY, 0);
- nextval[nexttoke].opval->op_targ = tmp;
- force_next(PRIVATEREF);
- TERM('@');
- }
- }
-
- /* Force them to make up their mind on "@foo". */
- if (lex_state != LEX_NORMAL && !lex_brackets &&
- ( !(gv = gv_fetchpv(tokenbuf+1, FALSE, SVt_PVAV)) ||
- (*tokenbuf == '@'
- ? !GvAV(gv)
- : !GvHV(gv) )))
- {
- char tmpbuf[1024];
- sprintf(tmpbuf, "Literal @%s now requires backslash",tokenbuf+1);
- yyerror(tmpbuf);
- }
+ no_op("Array", s);
+ tokenbuf[0] = '@';
+ s = scan_ident(s, bufend, tokenbuf+1, FALSE);
+ if (!tokenbuf[1]) {
+ if (s == bufend)
+ yyerror("Final @ should be \\@ or @name");
+ PREREF('@');
+ }
+ if (lex_state == LEX_NORMAL)
+ s = skipspace(s);
+ if ((expect != XREF || oldoldbufptr == last_lop) && intuit_more(s)) {
+ if (*s == '{')
+ tokenbuf[0] = '%';
/* Warn about @ where they meant $. */
if (dowarn) {
}
}
}
- force_ident(tokenbuf+1, *tokenbuf);
- }
- else {
- if (s == bufend)
- yyerror("Final @ should be \\@ or @name");
- PREREF('@');
}
+ pending_ident = '@';
TERM('@');
case '/': /* may either be division or pattern */
keylookup:
bufptr = s;
s = scan_word(s, tokenbuf, FALSE, &len);
-
- if (*s == ':' && s[1] == ':' && strNE(tokenbuf, "CORE"))
+
+ /* Some keywords can be followed by any delimiter, including ':' */
+ tmp = (len == 1 && strchr("msyq", tokenbuf[0]) ||
+ len == 2 && ((tokenbuf[0] == 't' && tokenbuf[1] == 'r') ||
+ (tokenbuf[0] == 'q' &&
+ strchr("qwx", tokenbuf[1]))));
+
+ /* x::* is just a word, unless x is "CORE" */
+ if (!tmp && *s == ':' && s[1] == ':' && strNE(tokenbuf, "CORE"))
goto just_a_word;
+ d = s;
+ while (d < bufend && isSPACE(*d))
+ d++; /* no comments skipped here, or s### is misparsed */
+
+ /* Is this a label? */
+ if (!tmp && expect == XSTATE
+ && d < bufend && *d == ':' && *(d + 1) != ':') {
+ s = d + 1;
+ yylval.pval = savepv(tokenbuf);
+ CLINE;
+ TOKEN(LABEL);
+ }
+
+ /* Check for keywords */
tmp = keyword(tokenbuf, len);
/* Is this a word before a => operator? */
- d = s;
- while (d < bufend && (*d == ' ' || *d == '\t'))
- d++; /* no comments skipped here, or s### is misparsed */
if (strnEQ(d,"=>",2)) {
CLINE;
if (dowarn && (tmp || perl_get_cv(tokenbuf, FALSE)))
croak("Bad name after %s::", tokenbuf);
}
- /* Do special processing at start of statement. */
-
- if (expect == XSTATE) {
- while (isSPACE(*s)) s++;
- if (*s == ':') { /* It's a label. */
- yylval.pval = savepv(tokenbuf);
- s++;
- CLINE;
- TOKEN(LABEL);
- }
- }
- else if (expect == XOPERATOR) {
+ if (expect == XOPERATOR) {
if (bufptr == linestart) {
curcop->cop_line--;
warn(warn_nosemi);
curcop->cop_line++;
}
else
- no_op("Bare word",s);
+ no_op("Bareword",s);
}
/* Look for a subroutine with this name in current package. */
/* (But it's an indir obj regardless for sort.) */
if ((last_lop_op == OP_SORT ||
- (!immediate_paren && (!gv || !GvCV(gv))) ) &&
+ (!immediate_paren && (!gv || !GvCVu(gv))) ) &&
(last_lop_op != OP_MAPSTART && last_lop_op != OP_GREPSTART)){
expect = (last_lop == oldoldbufptr) ? XTERM : XOPERATOR;
goto bareword;
/* If followed by var or block, call it a method (unless sub) */
- if ((*s == '$' || *s == '{') && (!gv || !GvCV(gv))) {
+ if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
last_lop = oldbufptr;
last_lop_op = OP_METHOD;
PREBLOCK(METHOD);
/* Not a method, so call it a subroutine (if defined) */
- if (gv && GvCV(gv)) {
+ if (gv && GvCVu(gv)) {
CV* cv = GvCV(gv);
if (*s == '(') {
nextval[nexttoke].opval = yylval.opval;
TOKEN(WORD);
}
+ case KEY___FILE__:
case KEY___LINE__:
- case KEY___FILE__: {
if (tokenbuf[2] == 'L')
(void)sprintf(tokenbuf,"%ld",(long)curcop->cop_line);
else
strcpy(tokenbuf, SvPVX(GvSV(curcop->cop_filegv)));
yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(tokenbuf,0));
TERM(THING);
- }
+
+ case KEY___PACKAGE__:
+ yylval.opval = (OP*)newSVOP(OP_CONST, 0,
+ (curstash
+ ? newSVsv(curstname)
+ : &sv_undef));
+ TERM(THING);
case KEY___DATA__:
case KEY___END__: {
FUN0(OP_GPWENT);
case KEY_getpwnam:
- FUN1(OP_GPWNAM);
+ UNI(OP_GPWNAM);
case KEY_getpwuid:
- FUN1(OP_GPWUID);
+ UNI(OP_GPWUID);
case KEY_getpeername:
UNI(OP_GETPEERNAME);
FUN0(OP_GGRENT);
case KEY_getgrnam:
- FUN1(OP_GGRNAM);
+ UNI(OP_GGRNAM);
case KEY_getgrgid:
- FUN1(OP_GGRGID);
+ UNI(OP_GGRGID);
case KEY_getlogin:
FUN0(OP_GETLOGIN);
LOP(OP_SETPRIORITY,XTERM);
case KEY_sethostent:
- FUN1(OP_SHOSTENT);
+ UNI(OP_SHOSTENT);
case KEY_setnetent:
- FUN1(OP_SNETENT);
+ UNI(OP_SNETENT);
case KEY_setservent:
- FUN1(OP_SSERVENT);
+ UNI(OP_SSERVENT);
case KEY_setprotoent:
- FUN1(OP_SPROTOENT);
+ UNI(OP_SPROTOENT);
case KEY_setpwent:
FUN0(OP_SPWENT);
/* Look for a prototype */
if (*s == '(') {
+ char *p;
+
s = scan_str(s);
if (!s) {
if (lex_stuff)
lex_stuff = Nullsv;
croak("Prototype not terminated");
}
+ /* strip spaces */
+ d = SvPVX(lex_stuff);
+ tmp = 0;
+ for (p = d; *p; ++p) {
+ if (!isSPACE(*p))
+ d[tmp++] = *p;
+ }
+ d[tmp] = '\0';
+ SvCUR(lex_stuff) = tmp;
+
nexttoke++;
nextval[1] = nextval[0];
nexttype[1] = nexttype[0];
switch (*d) {
case '_':
if (d[1] == '_') {
- if (strEQ(d,"__LINE__")) return -KEY___LINE__;
if (strEQ(d,"__FILE__")) return -KEY___FILE__;
+ if (strEQ(d,"__LINE__")) return -KEY___LINE__;
+ if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
if (strEQ(d,"__DATA__")) return KEY___DATA__;
if (strEQ(d,"__END__")) return KEY___END__;
}
return s;
}
if (*s == '$' && s[1] &&
- (isALPHA(s[1]) || strchr("$_{", s[1]) || strnEQ(s+1,"::",2)) )
- return s;
+ (isALNUM(s[1]) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
+ {
+ if (isDIGIT(s[1]) && lex_state == LEX_INTERPNORMAL)
+ deprecate("\"$$<digit>\" to mean \"${$}<digit>\"");
+ else
+ return s;
+ }
if (*s == '{') {
bracket = s;
s++;
*d = *s++;
d[1] = '\0';
if (*d == '^' && *s && (isUPPER(*s) || strchr("[\\]^_?", *s))) {
- *d = *s++ ^ 64;
+ *d = toCTRL(*s);
+ s++;
}
if (bracket) {
if (isSPACE(s[-1])) {
- while (s < send && (*s == ' ' || *s == '\t')) s++;
- *d = *s;
+ while (s < send) {
+ char ch = *s++;
+ if (ch != ' ' && ch != '\t') {
+ *d = ch;
+ break;
+ }
+ }
}
- if (isALPHA(*d) || *d == '_') {
+ if (isIDFIRST(*d)) {
d++;
while (isALNUM(*s) || *s == ':')
*d++ = *s++;
*d = '\0';
while (s < send && (*s == ' ' || *s == '\t')) s++;
- if ((*s == '[' || *s == '{')) {
+ if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
if (dowarn && keyword(dest, d - dest)) {
char *brack = *s == '[' ? "[...]" : "{...}";
warn("Ambiguous use of %c{%s%s} resolved to %c%s%s",
U16* pmfl;
int ch;
{
- if (ch == 'i') {
- sawi = TRUE;
+ if (ch == 'i')
*pmfl |= PMf_FOLD;
- }
else if (ch == 'g')
*pmfl |= PMf_GLOBAL;
else if (ch == 'o')
lex_stuff = Nullsv;
croak("Search pattern not terminated");
}
+
pm = (PMOP*)newPMOP(OP_MATCH, 0);
if (multi_open == '?')
pm->op_pmflags |= PMf_ONCE;
-
while (*s && strchr("iogmsx", *s))
pmflag(&pm->op_pmflags,*s++);
-
pm->op_pmpermflags = pm->op_pmflags;
+
lex_op = (OP*)pm;
yylval.ival = OP_MATCH;
return s;
{
register char *s;
register PMOP *pm;
+ I32 first_start;
I32 es = 0;
yylval.ival = OP_NULL;
if (s[-1] == multi_open)
s--;
+ first_start = multi_start;
s = scan_str(s);
if (!s) {
if (lex_stuff)
lex_repl = Nullsv;
croak("Substitution replacement not terminated");
}
+ multi_start = first_start; /* so whole substitution is taken together */
pm = (PMOP*)newPMOP(OP_SUBST, 0);
while (*s && strchr("iogmsex", *s)) {
) {
if (!(pm->op_pmregexp->reganch & ROPT_ANCH))
pm->op_pmflags |= PMf_SCANFIRST;
- else if (pm->op_pmflags & PMf_FOLD)
- return;
pm->op_pmshort = SvREFCNT_inc(pm->op_pmregexp->regstart);
pm->op_pmslen = SvCUR(pm->op_pmshort);
}
return;
}
}
- if (!pm->op_pmshort || /* promote the better string */
- ((pm->op_pmflags & PMf_SCANFIRST) &&
- (SvCUR(pm->op_pmshort) < SvCUR(pm->op_pmregexp->regmust)) )){
+ /* promote the better string */
+ if ((!pm->op_pmshort &&
+ !(pm->op_pmregexp->reganch & ROPT_ANCH_GPOS)) ||
+ ((pm->op_pmflags & PMf_SCANFIRST) &&
+ (SvCUR(pm->op_pmshort) < SvCUR(pm->op_pmregexp->regmust)))) {
SvREFCNT_dec(pm->op_pmshort); /* ok if null */
pm->op_pmshort = pm->op_pmregexp->regmust;
pm->op_pmslen = SvCUR(pm->op_pmshort);
char term;
register char *d;
char *peek;
+ int outer = (rsfp && !lex_inwhat);
s += 2;
d = tokenbuf;
- if (!rsfp)
+ if (!outer)
*d++ = '\n';
for (peek = s; *peek == ' ' || *peek == '\t'; peek++) ;
if (*peek && strchr("`'\"",*peek)) {
*d = '\0';
len = d - tokenbuf;
d = "\n";
- if (rsfp || !(d=ninstr(s,bufend,d,d+1)))
+ if (outer || !(d=ninstr(s,bufend,d,d+1)))
herewas = newSVpv(s,bufend-s);
else
s--, herewas = newSVpv(s,d-s);
multi_start = curcop->cop_line;
multi_open = multi_close = '<';
term = *tokenbuf;
- if (!rsfp) {
+ if (!outer) {
d = s;
while (s < bufend &&
- (*s != term || memcmp(s,tokenbuf,len) != 0) ) {
+ (*s != term || memNE(s,tokenbuf,len)) ) {
if (*s++ == '\n')
curcop->cop_line++;
}
else
sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
while (s >= bufend) { /* multiple line string? */
- if (!rsfp ||
+ if (!outer ||
!(oldoldbufptr = oldbufptr = s = linestart = filter_gets(linestr, rsfp, 0))) {
curcop->cop_line = multi_start;
missingterm(tokenbuf);
(I32)curcop->cop_line,sv);
}
bufend = SvPVX(linestr) + SvCUR(linestr);
- if (*s == term && memcmp(s,tokenbuf,len) == 0) {
+ if (*s == term && memEQ(s,tokenbuf,len)) {
s = bufend - 1;
*s = ' ';
sv_catsv(linestr,herewas);
{
register char *s = start;
register char *d;
- I32 tryi32;
+ I32 tryiv;
double value;
SV *sv;
I32 floatit;
}
*d = '\0';
sv = NEWSV(92,0);
+ SET_NUMERIC_STANDARD();
value = atof(tokenbuf);
- tryi32 = I_32(value);
- if (!floatit && (double)tryi32 == value)
- sv_setiv(sv,tryi32);
+ tryiv = I_V(value);
+ if (!floatit && (double)tryiv == value)
+ sv_setiv(sv, tryiv);
else
- sv_setnv(sv,value);
+ sv_setnv(sv, value);
break;
}
}
int
-start_subparse()
+start_subparse(is_format, flags)
+I32 is_format;
+U32 flags;
{
int oldsavestack_ix = savestack_ix;
CV* outsidecv = compcv;
AV* comppadlist;
-#ifndef __QNX__
if (compcv) {
assert(SvTYPE(compcv) == SVt_PVCV);
}
-#endif
save_I32(&subline);
save_item(subname);
SAVEI32(padix);
SAVEI32(pad_reset_pending);
compcv = (CV*)NEWSV(1104,0);
- sv_upgrade((SV *)compcv, SVt_PVCV);
+ sv_upgrade((SV *)compcv, is_format ? SVt_PVFM : SVt_PVCV);
+ CvFLAGS(compcv) |= flags;
comppad = newAV();
comppad_name = newAV();
yyerror(s)
char *s;
{
- char tmpbuf[258];
- char *tname = tmpbuf;
-
- if (bufptr > oldoldbufptr && bufptr - oldoldbufptr < 200 &&
+ char wbuf[40];
+ char *where = NULL;
+ char *context = NULL;
+ int contlen = -1;
+
+ if (!yychar || (yychar == ';' && !rsfp))
+ where = "at EOF";
+ else if (bufptr > oldoldbufptr && bufptr - oldoldbufptr < 200 &&
oldoldbufptr != oldbufptr && oldbufptr != bufptr) {
while (isSPACE(*oldoldbufptr))
oldoldbufptr++;
- sprintf(tname,"near \"%.*s\"",bufptr - oldoldbufptr, oldoldbufptr);
+ context = oldoldbufptr;
+ contlen = bufptr - oldoldbufptr;
}
else if (bufptr > oldbufptr && bufptr - oldbufptr < 200 &&
oldbufptr != bufptr) {
while (isSPACE(*oldbufptr))
oldbufptr++;
- sprintf(tname,"near \"%.*s\"",bufptr - oldbufptr, oldbufptr);
+ context = oldbufptr;
+ contlen = bufptr - oldbufptr;
}
else if (yychar > 255)
- tname = "next token ???";
- else if (!yychar || (yychar == ';' && !rsfp))
- (void)strcpy(tname,"at EOF");
+ where = "next token ???";
else if ((yychar & 127) == 127) {
if (lex_state == LEX_NORMAL ||
(lex_state == LEX_KNOWNEXT && lex_defer == LEX_NORMAL))
- (void)strcpy(tname,"at end of line");
+ where = "at end of line";
else if (lex_inpat)
- (void)strcpy(tname,"within pattern");
+ where = "within pattern";
else
- (void)strcpy(tname,"within string");
+ where = "within string";
}
else if (yychar < 32)
- (void)sprintf(tname,"next char ^%c",yychar+64);
+ (void)sprintf(where = wbuf, "next char ^%c", toCTRL(yychar));
+ else if (isPRINT_LC(yychar))
+ (void)sprintf(where = wbuf, "next char %c", yychar);
+ else
+ (void)sprintf(where = wbuf, "next char \\%03o", yychar & 255);
+ if (contlen == -1)
+ contlen = strlen(where);
+ (void)sprintf(buf, "%s at %s line %d, ",
+ s, SvPVX(GvSV(curcop->cop_filegv)), curcop->cop_line);
+ if (context)
+ (void)sprintf(buf+strlen(buf), "near \"%.*s\"\n", contlen, context);
else
- (void)sprintf(tname,"next char %c",yychar);
- (void)sprintf(buf, "%s at %s line %d, %s\n",
- s,SvPVX(GvSV(curcop->cop_filegv)),curcop->cop_line,tname);
- if (curcop->cop_line == multi_end && multi_start < multi_end) {
+ (void)sprintf(buf+strlen(buf), "%s\n", where);
+ if (multi_start < multi_end && (U32)(curcop->cop_line - multi_end) <= 1) {
sprintf(buf+strlen(buf),
- " (Might be a runaway multi-line %c%c string starting on line %ld)\n",
- multi_open,multi_close,(long)multi_start);
+ " (Might be a runaway multi-line %c%c string starting on line %ld)\n",
+ (int)multi_open,(int)multi_close,(long)multi_start);
multi_end = 0;
}
if (in_eval & 2)
projects / p5sagit/p5-mst-13.2.git / blobdiff