Change files which are mysteriously different to mainline to be
[p5sagit/p5-mst-13.2.git] / toke.c
diff --git a/toke.c b/toke.c
index e5f737b..274e506 100644 (file)
--- a/toke.c
+++ b/toke.c
@@ -80,15 +80,19 @@ static I32 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
 #endif
 
 #ifdef USE_PURE_BISON
-YYSTYPE* yylval_pointer = NULL;
-int* yychar_pointer = NULL;
+#  ifndef YYMAXLEVEL
+#    define YYMAXLEVEL 100
+#  endif
+YYSTYPE* yylval_pointer[YYMAXLEVEL];
+int* yychar_pointer[YYMAXLEVEL];
+int yyactlevel = 0;
 #  undef yylval
 #  undef yychar
-#  define yylval (*yylval_pointer)
-#  define yychar (*yychar_pointer)
-#  define PERL_YYLEX_PARAM yylval_pointer,yychar_pointer
-#  undef yylex
-#  define yylex()      Perl_yylex(aTHX_ yylval_pointer, yychar_pointer)
+#  define yylval (*yylval_pointer[yyactlevel])
+#  define yychar (*yychar_pointer[yyactlevel])
+#  define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
+#  undef yylex 
+#  define yylex()      Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
 #endif
 
 #include "keywords.h"
@@ -357,7 +361,6 @@ Perl_lex_start(pTHX_ SV *line)
            SAVEVPTR(PL_nextval[toke]);
        }
        SAVEI32(PL_nexttoke);
-       PL_nexttoke = 0;
     }
     SAVECOPLINE(PL_curcop);
     SAVEPPTR(PL_bufptr);
@@ -391,6 +394,7 @@ Perl_lex_start(pTHX_ SV *line)
     PL_lex_stuff = Nullsv;
     PL_lex_repl = Nullsv;
     PL_lex_inpat = 0;
+    PL_nexttoke = 0;
     PL_lex_inwhat = 0;
     PL_sublex_info.sub_inwhat = 0;
     PL_linestr = line;
@@ -809,10 +813,10 @@ Perl_str_to_version(pTHX_ SV *sv)
     bool utf = SvUTF8(sv) ? TRUE : FALSE;
     char *end = start + len;
     while (start < end) {
-       I32 skip;
+       STRLEN skip;
        UV n;
        if (utf)
-           n = utf8_to_uv((U8*)start, &skip);
+           n = utf8_to_uv((U8*)start, len, &skip, 0);
        else {
            n = *(U8*)start;
            skip = 1;
@@ -844,7 +848,7 @@ S_force_version(pTHX_ char *s)
         for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
         if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
            SV *ver;
-            s = scan_num(s);
+            s = scan_num(s, &yylval);
             version = yylval.opval;
            ver = cSVOPx(version)->op_sv;
            if (SvPOK(ver) && !SvNIOK(ver)) {
@@ -1183,14 +1187,13 @@ S_scan_const(pTHX_ char *start)
     register char *d = SvPVX(sv);              /* destination for copies */
     bool dorange = FALSE;                      /* are we in a translit range? */
     bool didrange = FALSE;                     /* did we just finish a range? */
-    bool has_utf = FALSE;                      /* embedded \x{} */
-    I32 len;                                   /* ? */
+    bool has_utf8 = FALSE;                     /* embedded \x{} */
     UV uv;
 
     I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
        ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
        : UTF;
-    I32 thisutf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
+    I32 this_utf8 = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
        ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
                                                OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
        : UTF;
@@ -1219,7 +1222,7 @@ S_scan_const(pTHX_ char *start)
                 if (min > max) {
                    Perl_croak(aTHX_
                               "Invalid [] range \"%c-%c\" in transliteration operator",
-                              min, max);
+                              (char)min, (char)max);
                 }
 
 #ifndef ASCIIish
@@ -1305,9 +1308,11 @@ S_scan_const(pTHX_ char *start)
                *d++ = *s++;
        }
 
-       /* check for embedded arrays (@foo, @:foo, @'foo, @{foo}, @$foo) */
+       /* check for embedded arrays
+          (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
+          */
        else if (*s == '@' && s[1]
-                && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$", s[1])))
+                && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
            break;
 
        /* check for embedded scalars.  only stop if we're sure it's a
@@ -1322,21 +1327,24 @@ S_scan_const(pTHX_ char *start)
 
        /* (now in tr/// code again) */
 
-       if (*s & 0x80 && thisutf) {
-          (void)utf8_to_uv((U8*)s, &len);
-          if (len == 1) {
-              /* illegal UTF8, make it valid */
-              char *old_pvx = SvPVX(sv);
-              /* need space for one extra char (NOTE: SvCUR() not set here) */
-              d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
-              d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
-          }
-          else {
-              while (len--)
-                  *d++ = *s++;
-          }
-          has_utf = TRUE;
-          continue;
+       if (*s & 0x80 && this_utf8) {
+           STRLEN len;
+           UV uv;
+
+           uv = utf8_to_uv((U8*)s, send - s, &len, UTF8_CHECK_ONLY);
+           if (len == 1) {
+               /* Illegal UTF8 (a high-bit byte), make it valid. */
+               char *old_pvx = SvPVX(sv);
+               /* need space for one extra char (NOTE: SvCUR() not set here) */
+               d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
+               d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
+           }
+           else {
+               while (len--)
+                   *d++ = *s++;
+           }
+           has_utf8 = TRUE;
+           continue;
        }
 
        /* backslashes */
@@ -1392,9 +1400,11 @@ S_scan_const(pTHX_ char *start)
            /* \132 indicates an octal constant */
            case '0': case '1': case '2': case '3':
            case '4': case '5': case '6': case '7':
-               len = 0;        /* disallow underscores */
-               uv = (UV)scan_oct(s, 3, &len);
-               s += len;
+               {
+                   STRLEN len = 0;     /* disallow underscores */
+                   uv = (UV)scan_oct(s, 3, &len);
+                   s += len;
+               }
                goto NUM_ESCAPE_INSERT;
 
            /* \x24 indicates a hex constant */
@@ -1406,14 +1416,19 @@ S_scan_const(pTHX_ char *start)
                        yyerror("Missing right brace on \\x{}");
                        e = s;
                    }
-                   len = 1;            /* allow underscores */
-                    uv = (UV)scan_hex(s + 1, e - s - 1, &len);
-                    s = e + 1;
+                   else {
+                       STRLEN len = 1;         /* allow underscores */
+                       uv = (UV)scan_hex(s + 1, e - s - 1, &len);
+                       has_utf8 = TRUE;
+                   }
+                   s = e + 1;
                }
                else {
-                   len = 0;            /* disallow underscores */
-                   uv = (UV)scan_hex(s, 2, &len);
-                   s += len;
+                   {
+                       STRLEN len = 0;         /* disallow underscores */
+                       uv = (UV)scan_hex(s, 2, &len);
+                       s += len;
+                   }
                }
 
              NUM_ESCAPE_INSERT:
@@ -1421,8 +1436,8 @@ S_scan_const(pTHX_ char *start)
                 * There will always enough room in sv since such escapes will
                 * be longer than any utf8 sequence they can end up as
                 */
-               if (uv > 127) {
-                   if (!thisutf && !has_utf && uv > 255) {
+               if (uv > 127 || has_utf8) {
+                   if (!this_utf8 && !has_utf8 && uv > 255) {
                        /* might need to recode whatever we have accumulated so far
                         * if it contains any hibit chars
                         */
@@ -1454,9 +1469,9 @@ S_scan_const(pTHX_ char *start)
                         }
                     }
 
-                    if (thisutf || uv > 255) {
+                    if (has_utf8 || uv > 255) {
                        d = (char*)uv_to_utf8((U8*)d, uv);
-                       has_utf = TRUE;
+                       this_utf8 = TRUE;
                     }
                    else {
                        *d++ = (char)uv;
@@ -1485,13 +1500,16 @@ S_scan_const(pTHX_ char *start)
                    res = new_constant( Nullch, 0, "charnames", 
                                        res, Nullsv, "\\N{...}" );
                    str = SvPV(res,len);
-                   if (!has_utf && SvUTF8(res)) {
+                   if (!has_utf8 && SvUTF8(res)) {
                        char *ostart = SvPVX(sv);
                        SvCUR_set(sv, d - ostart);
                        SvPOK_on(sv);
+                       *d = '\0';
                        sv_utf8_upgrade(sv);
+                       /* this just broke our allocation above... */
+                       SvGROW(sv, send - start);
                        d = SvPVX(sv) + SvCUR(sv);
-                       has_utf = TRUE;
+                       has_utf8 = TRUE;
                    }
                    if (len > e - s + 4) {
                        char *odest = SvPVX(sv);
@@ -1519,8 +1537,10 @@ S_scan_const(pTHX_ char *start)
                *d = toCTRL(*d); 
                d++;
 #else
-               len = *s++;
-               *d++ = toCTRL(len);
+               {
+                   U8 c = *s++;
+                   *d++ = toCTRL(c);
+               }
 #endif
                continue;
 
@@ -1568,7 +1588,7 @@ S_scan_const(pTHX_ char *start)
     *d = '\0';
     SvCUR_set(sv, d - SvPVX(sv));
     SvPOK_on(sv);
-    if (has_utf)
+    if (has_utf8)
        SvUTF8_on(sv);
 
     /* shrink the sv if we allocated more than we used */
@@ -2052,9 +2072,34 @@ S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
       if we already built the token before, use it.
 */
 
+#ifdef USE_PURE_BISON
+#ifdef __SC__
+#pragma segment Perl_yylex_r
+#endif
+int
+Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
+{
+    dTHR;
+    int r;
+
+    yylval_pointer[yyactlevel] = lvalp;
+    yychar_pointer[yyactlevel] = lcharp;
+    yyactlevel++;
+    if (yyactlevel >= YYMAXLEVEL)
+       Perl_croak(aTHX_ "panic: YYMAXLEVEL");
+
+    r = Perl_yylex(aTHX);
+
+    yyactlevel--;
+
+    return r;
+}
+#endif
+
 #ifdef __SC__
 #pragma segment Perl_yylex
 #endif
+
 int
 #ifdef USE_PURE_BISON
 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
@@ -2070,11 +2115,6 @@ Perl_yylex(pTHX)
     GV *gv = Nullgv;
     GV **gvp = 0;
 
-#ifdef USE_PURE_BISON
-    yylval_pointer = lvalp;
-    yychar_pointer = lcharp;
-#endif
-
     /* check if there's an identifier for us to look at */
     if (PL_pending_ident) {
         /* pit holds the identifier we read and pending_ident is reset */
@@ -3517,7 +3557,7 @@ Perl_yylex(pTHX)
        /* FALL THROUGH */
     case '0': case '1': case '2': case '3': case '4':
     case '5': case '6': case '7': case '8': case '9':
-       s = scan_num(s);
+       s = scan_num(s, &yylval);
        if (PL_expect == XOPERATOR)
            no_op("Number",s);
        TERM(THING);
@@ -3587,7 +3627,7 @@ Perl_yylex(pTHX)
            while (isDIGIT(*start) || *start == '_')
                start++;
            if (*start == '.' && isDIGIT(start[1])) {
-               s = scan_num(s);
+               s = scan_num(s, &yylval);
                TERM(THING);
            }
            /* avoid v123abc() or $h{v1}, allow C<print v10;> */
@@ -3598,7 +3638,7 @@ Perl_yylex(pTHX)
                gv = gv_fetchpv(s, FALSE, SVt_PVCV);
                *start = c;
                if (!gv) {
-                   s = scan_num(s);
+                   s = scan_num(s, &yylval);
                    TERM(THING);
                }
            }
@@ -6514,7 +6554,7 @@ S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
     register char term;                        /* terminating character */
     register char *to;                 /* current position in the sv's data */
     I32 brackets = 1;                  /* bracket nesting level */
-    bool has_utf = FALSE;              /* is there any utf8 content? */
+    bool has_utf8 = FALSE;             /* is there any utf8 content? */
 
     /* skip space before the delimiter */
     if (isSPACE(*s))
@@ -6526,7 +6566,7 @@ S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
     /* after skipping whitespace, the next character is the terminator */
     term = *s;
     if ((term & 0x80) && UTF)
-       has_utf = TRUE;
+       has_utf8 = TRUE;
 
     /* mark where we are */
     PL_multi_start = CopLINE(PL_curcop);
@@ -6572,8 +6612,8 @@ S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
                   have found the terminator */
                else if (*s == term)
                    break;
-               else if (!has_utf && (*s & 0x80) && UTF)
-                   has_utf = TRUE;
+               else if (!has_utf8 && (*s & 0x80) && UTF)
+                   has_utf8 = TRUE;
                *to = *s;
            }
        }
@@ -6601,8 +6641,8 @@ S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
                    break;
                else if (*s == PL_multi_open)
                    brackets++;
-               else if (!has_utf && (*s & 0x80) && UTF)
-                   has_utf = TRUE;
+               else if (!has_utf8 && (*s & 0x80) && UTF)
+                   has_utf8 = TRUE;
                *to = *s;
            }
        }
@@ -6662,7 +6702,7 @@ S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
 
     if (keep_delims)
        sv_catpvn(sv, s, 1);
-    if (has_utf)
+    if (has_utf8)
        SvUTF8_on(sv);
     PL_multi_end = CopLINE(PL_curcop);
     s++;
@@ -6707,7 +6747,7 @@ S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
 */
   
 char *
-Perl_scan_num(pTHX_ char *start)
+Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
 {
     register char *s = start;          /* current position in buffer */
     register char *d;                  /* destination in temp buffer */
@@ -7120,9 +7160,9 @@ vstring:
     /* make the op for the constant and return */
 
     if (sv)
-       yylval.opval = newSVOP(OP_CONST, 0, sv);
+       lvalp->opval = newSVOP(OP_CONST, 0, sv);
     else
-       yylval.opval = Nullop;
+       lvalp->opval = Nullop;
 
     return s;
 }
@@ -7354,7 +7394,7 @@ Perl_yyerror(pTHX_ char *s)
        qerror(msg);
     if (PL_error_count >= 10) {
        if (PL_in_eval && SvCUR(ERRSV))
-           Perl_croak(aTHX_ "%_%s has too many errors.\n",
+           Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
                       ERRSV, CopFILE(PL_curcop));
        else
            Perl_croak(aTHX_ "%s has too many errors.\n",