Gisle Aas writes:
: The included program will not compile. Perl5.005_52 says:
:
: Unrecognized character \xE5 at ident.t line 8.
:
: even though utf8.pm told me:
:
: Identifiers within the Perl script may contain Unicode alphanumeric
: characters, including ideographs. (You are currently on your own
: when it comes to using the canonical forms of characters--Perl
: doesn't (yet) attempt to canonicalize variable names for you.)
:
: (There might of other errors in this program, as I had a hard time
: reading it myself :-)
:
: Regards,
: Gisle
:
:
: begin 664 utf8/ident.t
: M<')I;G0@(C$N+C%<;B(["@HC(%1E<W0@:68(_at_)=71F(_dot_)"!C:&%R86-T97)S(&EN
: M(&ED96YT:69I97)S('-U8V@@87,@<')O8V5D=7)E(&YA;65S(&%N9`HC('9A
: M<FEA8FQE<R!W;W)K"@IU<V4(_at_)=71F(_dot_)#L*"G-U8B#ENZ`*>PH@("!P<FEN="`B
: M;F]T("(@=6YL97-S('-H:69T(#T](#,S.PH@("!P<FEN="`B;VL@,5QN(CL*
: M?0H*<W5B('-NP[AVPZ9R('L*("`@;7D@)'-NP[@@/2`S,SL*("`@;7D@),.E
: J(#T(_at_)-#4["B`@(.6[H"@D<V[#N"P@),.E*3L*?0H*<V[#N';#IG(H*3L*
: `
: end
: [Bulk--not acked]
Here's a patch for this (and some related problems).
[Sarathy, this is already in the repository.]
Larry
diff -c -r perl5.005_52/doop.c perl5.005_52pat/doop.c
*** perl5.005_52/doop.c Thu Sep 24 23:03:34 1998
--- perl5.005_52pat/doop.c Wed Oct 21 16:51:14 1998
***************
*** 215,224 ****
send = s + len;
while (s < send) {
! if ((uv = swash_fetch(rv, s)) < none) {
! s += UTF8SKIP(s);
matches++;
! }
}
return matches;
--- 215,223 ----
send = s + len;
while (s < send) {
! if ((uv = swash_fetch(rv, s)) < none)
matches++;
! s += UTF8SKIP(s);
}
return matches;
diff -c -r perl5.005_52/gv.c perl5.005_52pat/gv.c
*** perl5.005_52/gv.c Tue Sep 22 19:45:28 1998
--- perl5.005_52pat/gv.c Wed Oct 21 11:47:39 1998
***************
*** 499,505 ****
/* No stash in name, so see how we can default */
if (!stash) {
! if (isIDFIRST(*name)) {
bool global = FALSE;
if (isUPPER(*name)) {
--- 499,505 ----
/* No stash in name, so see how we can default */
if (!stash) {
! if (isIDFIRST(*name) || (IN_UTF8 && ((*name & 0xc0) == 0xc0) &&
isIDFIRST_utf8(name))) {
bool global = FALSE;
if (isUPPER(*name)) {
diff -c -r perl5.005_52/op.c perl5.005_52pat/op.c
*** perl5.005_52/op.c Thu Sep 24 03:16:53 1998
--- perl5.005_52pat/op.c Wed Oct 21 15:42:27 1998
***************
*** 118,124 ****
PADOFFSET off;
SV *sv;
! if (!(isALPHA(name[1]) || name[1] == '_' && (int)strlen(name) > 2)) {
if (!isPRINT(name[1])) {
name[3] = '\0';
name[2] = toCTRL(name[1]);
--- 118,128 ----
PADOFFSET off;
SV *sv;
! if (!(
! isALPHA(name[1]) ||
! (PL_hints & HINT_UTF8 && (name[1] & 0xc0) == 0xc0) ||
! name[1] == '_' && (int)strlen(name) > 2))
! {
if (!isPRINT(name[1])) {
name[3] = '\0';
name[2] = toCTRL(name[1]);
***************
*** 1662,1668 ****
dTHR;
if (ckWARN(WARN_PARENTHESIS) && PL_bufptr > PL_oldbufptr &&
PL_bufptr[-1] == ',') {
char *s;
! for (s = PL_bufptr; *s && (isALNUM(*s) || strchr("@$%, ",*s)); s++)
;
if (*s == ';' || *s == '=')
warner(WARN_PARENTHESIS, "Parens missing around \"%s\" list",
lex ? "my" : "local");
--- 1666,1672 ----
dTHR;
if (ckWARN(WARN_PARENTHESIS) && PL_bufptr > PL_oldbufptr &&
PL_bufptr[-1] == ',') {
char *s;
! for (s = PL_bufptr; *s && (isALNUM(*s) || (*s & 0x80) ||
strchr("@$%, ",*s)); s++) ;
if (*s == ';' || *s == '=')
warner(WARN_PARENTHESIS, "Parens missing around \"%s\" list",
lex ? "my" : "local");
diff -c -r perl5.005_52/pp.c perl5.005_52pat/pp.c
*** perl5.005_52/pp.c Wed Sep 23 03:29:38 1998
--- perl5.005_52pat/pp.c Wed Oct 21 11:55:55 1998
***************
*** 1316,1321 ****
--- 1316,1325 ----
sv_setsv(TARG, sv);
*SvPV_force(TARG, len) = *s == '-' ? '+' : '-';
}
+ else if (IN_UTF8 && *(U8*)s >= 0xc0 && isIDFIRST_utf8(s)) {
+ sv_setpvn(TARG, "-", 1);
+ sv_catsv(TARG, sv);
+ }
else
sv_setnv(TARG, -SvNV(sv));
SETTARG;
diff -c -r perl5.005_52/pp_hot.c perl5.005_52pat/pp_hot.c
*** perl5.005_52/pp_hot.c Sun Aug 9 23:36:13 1998
--- perl5.005_52pat/pp_hot.c Wed Oct 21 12:03:06 1998
***************
*** 2502,2511 ****
!(iogv = gv_fetchpv(packname, FALSE, SVt_PVIO)) ||
!(ob=(SV*)GvIO(iogv)))
{
! if (!packname || !isIDFIRST(*packname))
DIE("Can't call method \"%s\" %s", name,
SvOK(sv)? "without a package or object reference"
: "on an undefined value");
stash = gv_stashpvn(packname, packlen, TRUE);
goto fetch;
}
--- 2502,2517 ----
!(iogv = gv_fetchpv(packname, FALSE, SVt_PVIO)) ||
!(ob=(SV*)GvIO(iogv)))
{
! if (!packname ||
! ((*(U8*)packname >= 0xc0 && IN_UTF8)
! ? !isIDFIRST_utf8(packname)
! : !isIDFIRST(*packname)
! ))
! {
DIE("Can't call method \"%s\" %s", name,
SvOK(sv)? "without a package or object reference"
: "on an undefined value");
+ }
stash = gv_stashpvn(packname, packlen, TRUE);
goto fetch;
}
diff -c -r perl5.005_52/toke.c perl5.005_52pat/toke.c
*** perl5.005_52/toke.c Thu Sep 24 01:56:07 1998
--- perl5.005_52pat/toke.c Fri Oct 23 10:50:05 1998
***************
*** 61,66 ****
--- 61,78 ----
static char ident_too_long[] = "Identifier too long";
#define UTF (PL_hints & HINT_UTF8)
+ /*
+ * Note: we try to be careful never to call the isXXX_utf8() functions
+ * unless we're pretty sure we've seen the beginning of a UTF-8 character
+ * (that is, the two high bits are set). Otherwise we risk loading in the
+ * heavy-duty SWASHINIT and SWASHGET routines unnecessarily.
+ */
+ #define isIDFIRST_lazy(p) ((!UTF || (*((U8*)p) < 0xc0)) \
+ ? isIDFIRST(*(p)) \
+ : isIDFIRST_utf8((U8*)p))
+ #define isALNUM_lazy(p) ((!UTF || (*((U8*)p) < 0xc0)) \
+ ? isALNUM(*(p)) \
+ : isALNUM_utf8((U8*)p))
/* The following are arranged oddly so that the guard on the switch statement
* can get by with a single comparison (if the compiler is smart enough).
***************
*** 167,175 ****
yywarn(form("%s found where operator expected", what));
if (is_first)
warn("\t(Missing semicolon on previous line?)\n");
! else if (PL_oldoldbufptr && isIDFIRST(*PL_oldoldbufptr)) {
char *t;
! for (t = PL_oldoldbufptr; *t && (isALNUM(*t) || *t == ':'); t++) ;
if (t < PL_bufptr && isSPACE(*t))
warn("\t(Do you need to predeclare %.*s?)\n",
t - PL_oldoldbufptr, PL_oldoldbufptr);
--- 179,187 ----
yywarn(form("%s found where operator expected", what));
if (is_first)
warn("\t(Missing semicolon on previous line?)\n");
! else if (PL_oldoldbufptr && isIDFIRST_lazy(PL_oldoldbufptr)) {
char *t;
! for (t = PL_oldoldbufptr; *t && (isALNUM_lazy(t) || *t == ':'); t++) ;
if (t < PL_bufptr && isSPACE(*t))
warn("\t(Do you need to predeclare %.*s?)\n",
t - PL_oldoldbufptr, PL_oldoldbufptr);
***************
*** 476,482 ****
return;
while (isSPACE(*PL_last_uni))
PL_last_uni++;
! for (s = PL_last_uni; isALNUM(*s) || *s == '-'; s++) ;
if ((t = strchr(s, '(')) && t < PL_bufptr)
return;
ch = *s;
--- 488,494 ----
return;
while (isSPACE(*PL_last_uni))
PL_last_uni++;
! for (s = PL_last_uni; isALNUM_lazy(s) || *s == '-'; s++) ;
if ((t = strchr(s, '(')) && t < PL_bufptr)
return;
ch = *s;
***************
*** 552,558 ****
start = skipspace(start);
s = start;
! if (isIDFIRST(*s) ||
(allow_pack && *s == ':') ||
(allow_initial_tick && *s == '\'') )
{
--- 564,570 ----
start = skipspace(start);
s = start;
! if (isIDFIRST_lazy(s) ||
(allow_pack && *s == ':') ||
(allow_initial_tick && *s == '\'') )
{
***************
*** 993,999 ****
}
/* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
! else if (*s == '@' && s[1] && (isALNUM(s[1]) || strchr(":'{$", s[1])))
break;
/* check for embedded scalars. only stop if we're sure it's a
--- 1005,1011 ----
}
/* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
! else if (*s == '@' && s[1] && (isALNUM_lazy(s+1) || strchr(":'{$",
s[1])))
break;
/* check for embedded scalars. only stop if we're sure it's a
***************
*** 1247,1253 ****
case '&':
case '$':
weight -= seen[un_char] * 10;
! if (isALNUM(s[1])) {
scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE,
SVt_PV))
weight -= 100;
--- 1259,1265 ----
case '&':
case '$':
weight -= seen[un_char] * 10;
! if (isALNUM_lazy(s+1)) {
scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE,
SVt_PV))
weight -= 100;
***************
*** 1856,1871 ****
retry:
switch (*s) {
default:
! /*
! * Note: we try to be careful never to call the isXXX_utf8() functions
unless we're
! * pretty sure we've seen the beginning of a UTF-8 character (that is,
the two high
! * bits are set). Otherwise we risk loading in the heavy-duty
SWASHINIT and SWASHGET
! * routines unnecessarily. You will see this not just here but
throughout this file.
! */
! if (UTF && (*s & 0xc0) == 0x80) {
! if (isIDFIRST_utf8((U8*)s))
! goto keylookup;
! }
croak("Unrecognized character \\x%02X", *s & 255);
case 4:
case 26:
--- 1868,1875 ----
retry:
switch (*s) {
default:
! if (isIDFIRST_lazy(s))
! goto keylookup;
croak("Unrecognized character \\x%02X", *s & 255);
case 4:
case 26:
***************
*** 2216,2222 ****
else if (*s == '>') {
s++;
s = skipspace(s);
! if (isIDFIRST(*s)) {
s = force_word(s,METHOD,FALSE,TRUE,FALSE);
TOKEN(ARROW);
}
--- 2220,2226 ----
else if (*s == '>') {
s++;
s = skipspace(s);
! if (isIDFIRST_lazy(s)) {
s = force_word(s,METHOD,FALSE,TRUE,FALSE);
TOKEN(ARROW);
}
***************
*** 2361,2367 ****
while (d < PL_bufend && (*d == ' ' || *d == '\t'))
d++;
}
! if (d < PL_bufend && isIDFIRST(*d)) {
d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
FALSE, &len);
while (d < PL_bufend && (*d == ' ' || *d == '\t'))
--- 2365,2371 ----
while (d < PL_bufend && (*d == ' ' || *d == '\t'))
d++;
}
! if (d < PL_bufend && isIDFIRST_lazy(d)) {
d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
FALSE, &len);
while (d < PL_bufend && (*d == ' ' || *d == '\t'))
***************
*** 2449,2456 ****
}
t++;
}
! else if (isALPHA(*s)) {
! for (t++; t < PL_bufend && isALNUM(*t); t++) ;
}
while (t < PL_bufend && isSPACE(*t))
t++;
--- 2453,2460 ----
}
t++;
}
! else if (isIDFIRST_lazy(s)) {
! for (t++; t < PL_bufend && isALNUM_lazy(t); t++) ;
}
while (t < PL_bufend && isSPACE(*t))
t++;
***************
*** 2460,2466 ****
|| (*t == '=' && t[1] == '>')))
OPERATOR(HASHBRACK);
if (PL_expect == XREF)
! PL_expect = XTERM;
else {
PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
PL_expect = XSTATE;
--- 2464,2470 ----
|| (*t == '=' && t[1] == '>')))
OPERATOR(HASHBRACK);
if (PL_expect == XREF)
! PL_expect = XSTATE; /* was XTERM, trying XSTATE */
else {
PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
PL_expect = XSTATE;
***************
*** 2508,2514 ****
AOPERATOR(ANDAND);
s--;
if (PL_expect == XOPERATOR) {
! if (ckWARN(WARN_SEMICOLON) && isALPHA(*s) && PL_bufptr ==
PL_linestart) {
PL_curcop->cop_line--;
warner(WARN_SEMICOLON, warn_nosemi);
PL_curcop->cop_line++;
--- 2512,2518 ----
AOPERATOR(ANDAND);
s--;
if (PL_expect == XOPERATOR) {
! if (ckWARN(WARN_SEMICOLON) && isIDFIRST_lazy(s) && PL_bufptr ==
PL_linestart) {
PL_curcop->cop_line--;
warner(WARN_SEMICOLON, warn_nosemi);
PL_curcop->cop_line++;
***************
*** 2638,2644 ****
}
}
! if (s[1] == '#' && (isALPHA(s[2]) || strchr("_{$:+-", s[2]))) {
if (PL_expect == XOPERATOR)
no_op("Array length", PL_bufptr);
PL_tokenbuf[0] = '@';
--- 2642,2648 ----
}
}
! if (s[1] == '#' && (isIDFIRST_lazy(s+2) || strchr("{$:+-", s[2]))) {
if (PL_expect == XOPERATOR)
no_op("Array length", PL_bufptr);
PL_tokenbuf[0] = '@';
***************
*** 2679,2685 ****
PL_tokenbuf[0] = '@';
if (ckWARN(WARN_SYNTAX)) {
for(t = s + 1;
! isSPACE(*t) || isALNUM(*t) || *t == '$';
t++) ;
if (*t++ == ',') {
PL_bufptr = skipspace(PL_bufptr);
--- 2683,2689 ----
PL_tokenbuf[0] = '@';
if (ckWARN(WARN_SYNTAX)) {
for(t = s + 1;
! isSPACE(*t) || isALNUM_lazy(t) || *t == '$';
t++) ;
if (*t++ == ',') {
PL_bufptr = skipspace(PL_bufptr);
***************
*** 2699,2705 ****
char tmpbuf[sizeof PL_tokenbuf];
STRLEN len;
for (t++; isSPACE(*t); t++) ;
! if (isIDFIRST(*t)) {
t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
if (*t != '(' && perl_get_cv(tmpbuf, FALSE))
warner(WARN_SYNTAX,
--- 2703,2709 ----
char tmpbuf[sizeof PL_tokenbuf];
STRLEN len;
for (t++; isSPACE(*t); t++) ;
! if (isIDFIRST_lazy(t)) {
t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
if (*t != '(' && perl_get_cv(tmpbuf, FALSE))
warner(WARN_SYNTAX,
***************
*** 2716,2724 ****
PL_expect = XOPERATOR;
else if (strchr("$(_at_)\"'`q", *s))
PL_expect = XTERM; /* e.g. print $fh "foo" */
! else if (strchr("&*<%", *s) && isIDFIRST(s[1]))
PL_expect = XTERM; /* e.g. print $fh &sub */
! else if (isIDFIRST(*s)) {
char tmpbuf[sizeof PL_tokenbuf];
scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
if (tmp = keyword(tmpbuf, len)) {
--- 2720,2728 ----
PL_expect = XOPERATOR;
else if (strchr("$(_at_)\"'`q", *s))
PL_expect = XTERM; /* e.g. print $fh "foo" */
! else if (strchr("&*<%", *s) && isIDFIRST_lazy(s+1))
PL_expect = XTERM; /* e.g. print $fh &sub */
! else if (isIDFIRST_lazy(s)) {
char tmpbuf[sizeof PL_tokenbuf];
scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
if (tmp = keyword(tmpbuf, len)) {
***************
*** 2776,2782 ****
if (ckWARN(WARN_SYNTAX)) {
if (*s == '[' || *s == '{') {
char *t = s + 1;
! while (*t && (isALNUM(*t) || strchr(" \t$#+-'\"", *t)))
t++;
if (*t == '}' || *t == ']') {
t++;
--- 2780,2786 ----
if (ckWARN(WARN_SYNTAX)) {
if (*s == '[' || *s == '{') {
char *t = s + 1;
! while (*t && (isALNUM_lazy(t) || strchr(" \t$#+-'\"", *t)))
t++;
if (*t == '}' || *t == ']') {
t++;
***************
*** 2797,2803 ****
/* Disable warning on "study /blah/" */
if (PL_oldoldbufptr == PL_last_uni
&& (*PL_last_uni != 's' || s - PL_last_uni < 5
! || memNE(PL_last_uni, "study", 5) ||
isALNUM(PL_last_uni[5])))
check_uni();
s = scan_pat(s,OP_MATCH);
TERM(sublex_start());
--- 2801,2807 ----
/* Disable warning on "study /blah/" */
if (PL_oldoldbufptr == PL_last_uni
&& (*PL_last_uni != 's' || s - PL_last_uni < 5
! || memNE(PL_last_uni, "study", 5) ||
isALNUM_lazy(PL_last_uni+5)))
check_uni();
s = scan_pat(s,OP_MATCH);
TERM(sublex_start());
***************
*** 3110,3116 ****
/* Two barewords in a row may indicate method call. */
! if ((isALPHA(*s) || *s == '$') && (tmp=intuit_method(s,gv)))
return tmp;
/* If not a declared subroutine, it's an indirect object. */
--- 3114,3120 ----
/* Two barewords in a row may indicate method call. */
! if ((isIDFIRST_lazy(s) || *s == '$') &&
(tmp=intuit_method(s,gv)))
return tmp;
/* If not a declared subroutine, it's an indirect object. */
***************
*** 3154,3160 ****
/* If followed by a bareword, see if it looks like indir obj. */
! if ((isALPHA(*s) || *s == '$') && (tmp = intuit_method(s,gv)))
return tmp;
/* Not a method, so call it a subroutine (if defined) */
--- 3158,3164 ----
/* If followed by a bareword, see if it looks like indir obj. */
! if ((isIDFIRST_lazy(s) || *s == '$') && (tmp =
intuit_method(s,gv)))
return tmp;
/* Not a method, so call it a subroutine (if defined) */
***************
*** 3468,3480 ****
case KEY_foreach:
yylval.ival = PL_curcop->cop_line;
s = skipspace(s);
! if (PL_expect == XSTATE && isIDFIRST(*s)) {
char *p = s;
if ((PL_bufend - p) >= 3 &&
strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
p += 2;
p = skipspace(p);
! if (isIDFIRST(*p))
croak("Missing $ on loop variable");
}
OPERATOR(FOR);
--- 3472,3484 ----
case KEY_foreach:
yylval.ival = PL_curcop->cop_line;
s = skipspace(s);
! if (PL_expect == XSTATE && isIDFIRST_lazy(s)) {
char *p = s;
if ((PL_bufend - p) >= 3 &&
strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
p += 2;
p = skipspace(p);
! if (isIDFIRST_lazy(p))
croak("Missing $ on loop variable");
}
OPERATOR(FOR);
***************
*** 3662,3668 ****
TERM(sublex_start());
case KEY_map:
! LOP(OP_MAPSTART,XREF);
case KEY_mkdir:
LOP(OP_MKDIR,XTERM);
--- 3666,3672 ----
TERM(sublex_start());
case KEY_map:
! LOP(OP_MAPSTART, XREF);
case KEY_mkdir:
LOP(OP_MKDIR,XTERM);
***************
*** 3682,3688 ****
case KEY_my:
PL_in_my = TRUE;
s = skipspace(s);
! if (isIDFIRST(*s)) {
s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
if (!PL_in_my_stash) {
--- 3686,3692 ----
case KEY_my:
PL_in_my = TRUE;
s = skipspace(s);
! if (isIDFIRST_lazy(s)) {
s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
PL_in_my_stash = gv_stashpv(PL_tokenbuf, FALSE);
if (!PL_in_my_stash) {
***************
*** 3714,3722 ****
case KEY_open:
s = skipspace(s);
! if (isIDFIRST(*s)) {
char *t;
! for (d = s; isALNUM(*d); d++) ;
t = skipspace(d);
if (strchr("|&*+-=!?:.", *t))
warn("Precedence problem: open %.*s should be open(%.*s)",
--- 3718,3726 ----
case KEY_open:
s = skipspace(s);
! if (isIDFIRST_lazy(s)) {
char *t;
! for (d = s; isALNUM_lazy(d); d++) ;
t = skipspace(d);
if (strchr("|&*+-=!?:.", *t))
warn("Precedence problem: open %.*s should be open(%.*s)",
***************
*** 3839,3845 ****
case KEY_require:
*PL_tokenbuf = '\0';
s = force_word(s,WORD,TRUE,TRUE,FALSE);
! if (isIDFIRST(*PL_tokenbuf))
gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
else if (*s == '<')
yyerror("<> should be quotes");
--- 3843,3849 ----
case KEY_require:
*PL_tokenbuf = '\0';
s = force_word(s,WORD,TRUE,TRUE,FALSE);
! if (isIDFIRST_lazy(PL_tokenbuf))
gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
else if (*s == '<')
yyerror("<> should be quotes");
***************
*** 4023,4029 ****
really_sub:
s = skipspace(s);
! if (isIDFIRST(*s) || *s == '\'' || *s == ':') {
char tmpbuf[sizeof PL_tokenbuf];
PL_expect = XBLOCK;
d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
--- 4027,4033 ----
really_sub:
s = skipspace(s);
! if (isIDFIRST_lazy(s) || *s == '\'' || *s == ':') {
char tmpbuf[sizeof PL_tokenbuf];
PL_expect = XBLOCK;
d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
***************
*** 4895,4903 ****
s++;
while (s < PL_bufend && isSPACE(*s))
s++;
! if (isIDFIRST(*s)) {
w = s++;
! while (isALNUM(*s))
s++;
while (s < PL_bufend && isSPACE(*s))
s++;
--- 4899,4907 ----
s++;
while (s < PL_bufend && isSPACE(*s))
s++;
! if (isIDFIRST_lazy(s)) {
w = s++;
! while (isALNUM_lazy(s))
s++;
while (s < PL_bufend && isSPACE(*s))
s++;
***************
*** 4990,4998 ****
for (;;) {
if (d >= e)
croak(ident_too_long);
! if (isALNUM(*s))
*d++ = *s++;
! else if (*s == '\'' && allow_package && isIDFIRST(s[1])) {
*d++ = ':';
*d++ = ':';
s++;
--- 4994,5002 ----
for (;;) {
if (d >= e)
croak(ident_too_long);
! if (isALNUM(*s)) /* UTF handled below */
*d++ = *s++;
! else if (*s == '\'' && allow_package && isIDFIRST_lazy(s+1)) {
*d++ = ':';
*d++ = ':';
s++;
***************
*** 5001,5007 ****
*d++ = *s++;
*d++ = *s++;
}
! else if (UTF && (*s & 0xc0) == 0x80 && isALNUM_utf8((U8*)s)) {
char *t = s + UTF8SKIP(s);
while (*t & 0x80 && is_utf8_mark((U8*)t))
t += UTF8SKIP(t);
--- 5005,5011 ----
*d++ = *s++;
*d++ = *s++;
}
! else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
char *t = s + UTF8SKIP(s);
while (*t & 0x80 && is_utf8_mark((U8*)t))
t += UTF8SKIP(t);
***************
*** 5044,5052 ****
for (;;) {
if (d >= e)
croak(ident_too_long);
! if (isALNUM(*s))
*d++ = *s++;
! else if (*s == '\'' && isIDFIRST(s[1])) {
*d++ = ':';
*d++ = ':';
s++;
--- 5048,5056 ----
for (;;) {
if (d >= e)
croak(ident_too_long);
! if (isALNUM(*s)) /* UTF handled below */
*d++ = *s++;
! else if (*s == '\'' && isIDFIRST_lazy(s+1)) {
*d++ = ':';
*d++ = ':';
s++;
***************
*** 5055,5061 ****
*d++ = *s++;
*d++ = *s++;
}
! else if (UTF && (*s & 0xc0) == 0x80 && isALNUM_utf8((U8*)s)) {
char *t = s + UTF8SKIP(s);
while (*t & 0x80 && is_utf8_mark((U8*)t))
t += UTF8SKIP(t);
--- 5059,5065 ----
*d++ = *s++;
*d++ = *s++;
}
! else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
char *t = s + UTF8SKIP(s);
while (*t & 0x80 && is_utf8_mark((U8*)t))
t += UTF8SKIP(t);
***************
*** 5077,5083 ****
return s;
}
if (*s == '$' && s[1] &&
! (isALNUM(s[1]) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
{
return s;
}
--- 5081,5087 ----
return s;
}
if (*s == '$' && s[1] &&
! (isALNUM_lazy(s+1) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
{
return s;
}
***************
*** 5104,5114 ****
}
}
}
! if (isIDFIRST(*d) || (UTF && (*d & 0xc0) == 0x80 &&
isIDFIRST_utf8((U8*)d))) {
d++;
if (UTF) {
e = s;
! while (e < send && (isALNUM(*e) || ((*e & 0xc0) == 0x80 &&
isALNUM_utf8((U8*)e)) || *e == ':')) {
e += UTF8SKIP(e);
while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
e += UTF8SKIP(e);
--- 5108,5118 ----
}
}
}
! if (isIDFIRST_lazy(d)) {
d++;
if (UTF) {
e = s;
! while (e < send && isALNUM_lazy(e) || *e == ':') {
e += UTF8SKIP(e);
while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
e += UTF8SKIP(e);
***************
*** 5394,5402 ****
s++, term = '\'';
else
term = '"';
! if (!isALNUM(*s))
deprecate("bare << to mean <<\"\"");
! for (; isALNUM(*s); s++) {
if (d < e)
*d++ = *s;
}
--- 5398,5406 ----
s++, term = '\'';
else
term = '"';
! if (!isALNUM_lazy(s))
deprecate("bare << to mean <<\"\"");
! for (; isALNUM_lazy(s); s++) {
if (d < e)
*d++ = *s;
}
***************
*** 5577,5583 ****
if (*d == '$' && d[1]) d++;
/* allow <Pkg'VALUE> or <Pkg::VALUE> */
! while (*d && (isALNUM(*d) || *d == '\'' || *d == ':'))
d++;
/* If we've tried to read what we allow filehandles to look like, and
--- 5581,5587 ----
if (*d == '$' && d[1]) d++;
/* allow <Pkg'VALUE> or <Pkg::VALUE> */
! while (*d && (isALNUM_lazy(d) || *d == '\'' || *d == ':'))
d++;
/* If we've tried to read what we allow filehandles to look like, and