dist/Safe/t/safeutf8.t: Generalize to non-ASCII platform

[perl5.git] / toke.c

diff --git a/toke.c b/toke.c
index cb411d4..0eeafd4 100644 (file)
--- a/toke.c
+++ b/toke.c
@@ -1252,7 +1252,7 @@ buffer has reached the end of the input text.
 */
 
 #define LEX_FAKE_EOF 0x80000000
-#define LEX_NO_TERM  0x40000000
+#define LEX_NO_TERM  0x40000000 /* here-doc */
 
 bool
 Perl_lex_next_chunk(pTHX_ U32 flags)
@@ -1266,6 +1266,8 @@ Perl_lex_next_chunk(pTHX_ U32 flags)
     bool got_some;
     if (flags & ~(LEX_KEEP_PREVIOUS|LEX_FAKE_EOF|LEX_NO_TERM))
        Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_next_chunk");
+    if (!(flags & LEX_NO_TERM) && PL_lex_inwhat)
+       return FALSE;
     linestr = PL_parser->linestr;
     buf = SvPVX(linestr);
     if (!(flags & LEX_KEEP_PREVIOUS) &&
@@ -1520,6 +1522,8 @@ Perl_lex_read_space(pTHX_ U32 flags)
                incline(s);
                need_incline = 0;
            }
+       } else if (!c) {
+           s++;
        } else {
            break;
        }
@@ -1796,13 +1800,13 @@ S_skipspace_flags(pTHX_ char *s, U32 flags)
 {
     PERL_ARGS_ASSERT_SKIPSPACE_FLAGS;
     if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
-       while (s < PL_bufend && SPACE_OR_TAB(*s))
+       while (s < PL_bufend && (SPACE_OR_TAB(*s) || !*s))
            s++;
     } else {
        STRLEN bufptr_pos = PL_bufptr - SvPVX(PL_linestr);
        PL_bufptr = s;
        lex_read_space(flags | LEX_KEEP_PREVIOUS |
-               (PL_sublex_info.sub_inwhat || PL_lex_state == LEX_FORMLINE ?
+               (PL_lex_inwhat || PL_lex_state == LEX_FORMLINE ?
                    LEX_NO_NEXT_CHUNK : 0));
        s = PL_bufptr;
        PL_bufptr = SvPVX(PL_linestr) + bufptr_pos;
@@ -1905,6 +1909,7 @@ S_force_next(pTHX_ I32 type)
        tokereport(type, &NEXTVAL_NEXTTOKE);
     }
 #endif
+    assert(PL_nexttoke < C_ARRAY_LENGTH(PL_nexttype));
     PL_nexttype[PL_nexttoke] = type;
     PL_nexttoke++;
     if (PL_lex_state != LEX_KNOWNEXT) {
@@ -2376,6 +2381,7 @@ S_sublex_push(pTHX)
        popping.  We must not have a PL_lex_stuff value left dangling, as
        that breaks assumptions elsewhere.  See bug #123617.  */
     SAVEGENERICSV(PL_lex_stuff);
+    SAVEGENERICSV(PL_sublex_info.repl);
 
     PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
        = SvPVX(PL_linestr);
@@ -2481,7 +2487,6 @@ S_sublex_done(pTHX)
        PL_bufend = SvPVX(PL_linestr);
        PL_bufend += SvCUR(PL_linestr);
        PL_expect = XOPERATOR;
-       PL_sublex_info.sub_inwhat = 0;
        return ')';
     }
 }
@@ -4309,13 +4314,8 @@ Perl_yylex(pTHX)
        SvREFCNT_dec(tmp);
     } );
 
-    switch (PL_lex_state) {
-    case LEX_NORMAL:
-    case LEX_INTERPNORMAL:
-       break;
-
     /* when we've already built the next token, just pull it out of the queue */
-    case LEX_KNOWNEXT:
+    if (PL_nexttoke) {
        PL_nexttoke--;
        pl_yylval = PL_nextval[PL_nexttoke];
        if (!PL_nexttoke) {
@@ -4340,6 +4340,12 @@ Perl_yylex(pTHX)
            }
            return REPORT(next_type == 'p' ? pending_ident() : next_type);
        }
+    }
+
+    switch (PL_lex_state) {
+    case LEX_NORMAL:
+    case LEX_INTERPNORMAL:
+       break;
 
     /* interpolated case modifiers like \L \U, including \Q and \E.
        when we get here, PL_bufptr is at the \
@@ -4488,6 +4494,14 @@ Perl_yylex(pTHX)
        /* FALLTHROUGH */
 
     case LEX_INTERPEND:
+       /* Treat state as LEX_NORMAL if we have no inner lexing scope.
+          XXX This hack can be removed if we stop setting PL_lex_state to
+          LEX_KNOWNEXT, as can the hack under LEX_INTREPCONCAT below.  */
+       if (UNLIKELY(!PL_lex_inwhat)) {
+           PL_lex_state = LEX_NORMAL;
+           break;
+       }
+
        if (PL_lex_dojoin) {
            const U8 dojoin_was = PL_lex_dojoin;
            PL_lex_dojoin = FALSE;
@@ -4539,6 +4553,14 @@ Perl_yylex(pTHX)
            Perl_croak(aTHX_ "panic: INTERPCONCAT, lex_brackets=%ld",
                       (long) PL_lex_brackets);
 #endif
+       /* Treat state as LEX_NORMAL when not in an inner lexing scope.
+          XXX This hack can be removed if we stop setting PL_lex_state to
+          LEX_KNOWNEXT.  */
+       if (UNLIKELY(!PL_lex_inwhat)) {
+           PL_lex_state = LEX_NORMAL;
+           break;
+       }
+
        if (PL_bufptr == PL_bufend)
            return REPORT(sublex_done());
 
@@ -4619,7 +4641,8 @@ Perl_yylex(pTHX)
     case 26:
        goto fake_eof;                  /* emulate EOF on ^D or ^Z */
     case 0:
-       if (!PL_rsfp && (!PL_parser->filtered || s+1 < PL_bufend)) {
+       if ((!PL_rsfp || PL_lex_inwhat)
+        && (!PL_parser->filtered || s+1 < PL_bufend)) {
            PL_last_uni = 0;
            PL_last_lop = 0;
            if (PL_lex_brackets &&
@@ -4958,7 +4981,6 @@ Perl_yylex(pTHX)
        }
        if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
            PL_lex_state = LEX_FORMLINE;
-           NEXTVAL_NEXTTOKE.ival = 0;
            force_next(FORMRBRACK);
            TOKEN(';');
        }
@@ -5001,7 +5023,6 @@ Perl_yylex(pTHX)
                 incline(s);
            if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
                PL_lex_state = LEX_FORMLINE;
-               NEXTVAL_NEXTTOKE.ival = 0;
                force_next(FORMRBRACK);
                TOKEN(';');
            }
@@ -5750,13 +5771,12 @@ Perl_yylex(pTHX)
        PL_tokenbuf[0] = '&';
        s = scan_ident(s - 1, PL_tokenbuf + 1,
                       sizeof PL_tokenbuf - 1, TRUE);
+       pl_yylval.ival = (OPpENTERSUB_AMPER<<8);
        if (PL_tokenbuf[1]) {
-           PL_expect = XOPERATOR;
            force_ident_maybe_lex('&');
        }
        else
            PREREF('&');
-       pl_yylval.ival = (OPpENTERSUB_AMPER<<8);
        TERM('&');
 
     case '|':
@@ -8593,7 +8613,7 @@ S_scan_ident(pTHX_ char *s, char *dest, STRLEN destlen, I32 ck_uni)
 
     PERL_ARGS_ASSERT_SCAN_IDENT;
 
-    if (isSPACE(*s))
+    if (isSPACE(*s) || !*s)
        s = skipspace(s);
     if (isDIGIT(*s)) {
        while (isDIGIT(*s)) {