Re: [ID 20010831.001] SEGV from ($a, b) = (1, 2)

[perl5.git] / regexec.c

diff --git a/regexec.c b/regexec.c
index e95bd16..6adb9ca 100644 (file)
--- a/regexec.c
+++ b/regexec.c
@@ -80,12 +80,6 @@
 #define PERL_IN_REGEXEC_C
 #include "perl.h"
 
-#ifdef PERL_IN_XSUB_RE
-#  if defined(PERL_CAPI) || defined(PERL_OBJECT)
-#    include "XSUB.h"
-#  endif
-#endif
-
 #include "regcomp.h"
 
 #define RF_tainted     1               /* tainted information used? */
@@ -107,17 +101,17 @@
  */
 
 #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
-#define CHR_DIST(a,b) (DO_UTF8(PL_reg_sv) ? utf8_distance(a,b) : a - b)
+#define CHR_DIST(a,b) (PL_reg_sv_utf8 ? utf8_distance(a,b) : a - b)
 
 #define reghop_c(pos,off) ((char*)reghop((U8*)pos, off))
 #define reghopmaybe_c(pos,off) ((char*)reghopmaybe((U8*)pos, off))
-#define HOP(pos,off) (DO_UTF8(PL_reg_sv) ? reghop((U8*)pos, off) : (U8*)(pos + off))
-#define HOPMAYBE(pos,off) (DO_UTF8(PL_reg_sv) ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off))
+#define HOP(pos,off) (PL_reg_sv_utf8 ? reghop((U8*)pos, off) : (U8*)(pos + off))
+#define HOPMAYBE(pos,off) (PL_reg_sv_utf8 ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off))
 #define HOPc(pos,off) ((char*)HOP(pos,off))
 #define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off))
 
 #define HOPBACK(pos, off) (            \
-    (UTF && DO_UTF8(PL_reg_sv))                \
+    (UTF && PL_reg_sv_utf8)            \
        ? reghopmaybe((U8*)pos, -off)   \
     : (pos - off >= PL_bostr)          \
        ? (U8*)(pos - off)              \
@@ -127,14 +121,14 @@
 
 #define reghop3_c(pos,off,lim) ((char*)reghop3((U8*)pos, off, (U8*)lim))
 #define reghopmaybe3_c(pos,off,lim) ((char*)reghopmaybe3((U8*)pos, off, (U8*)lim))
-#define HOP3(pos,off,lim) (DO_UTF8(PL_reg_sv) ? reghop3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off))
-#define HOPMAYBE3(pos,off,lim) (DO_UTF8(PL_reg_sv) ? reghopmaybe3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off))
+#define HOP3(pos,off,lim) (PL_reg_sv_utf8 ? reghop3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off))
+#define HOPMAYBE3(pos,off,lim) (PL_reg_sv_utf8 ? reghopmaybe3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off))
 #define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim))
 #define HOPMAYBE3c(pos,off,lim) ((char*)HOPMAYBE3(pos,off,lim))
 
 #define LOAD_UTF8_CHARCLASS(a,b) STMT_START { if (!CAT2(PL_utf8_,a)) (void)CAT2(is_utf8_, a)((U8*)b); } STMT_END
 
-static void restore_pos(pTHXo_ void *arg);
+static void restore_pos(pTHX_ void *arg);
 
 STATIC CHECKPOINT
 S_regcppush(pTHX_ I32 parenfloor)
@@ -878,7 +872,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
        unsigned int c2;
        char *e;
        register I32 tmp = 1;   /* Scratch variable? */
-       register bool do_utf8 = DO_UTF8(PL_reg_sv);
+       register bool do_utf8 = PL_reg_sv_utf8;
 
        /* We know what class it must start with. */
        switch (OP(c)) {
@@ -1775,7 +1769,7 @@ got_it:
            sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is
                                                  restored, the value remains
                                                  the same. */
-       restore_pos(aTHXo_ 0);
+       restore_pos(aTHX_ 0);
     }
 
     /* make sure $`, $&, $', and $digit will work later */
@@ -1804,7 +1798,7 @@ phooey:
     DEBUG_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
                          PL_colors[4],PL_colors[5]));
     if (PL_reg_eval_set)
-       restore_pos(aTHXo_ 0);
+       restore_pos(aTHX_ 0);
     return 0;
 }
 
@@ -1841,7 +1835,7 @@ S_regtry(pTHX_ regexp *prog, char *startpos)
        if (PL_reg_sv) {
            /* Make $_ available to executed code. */
            if (PL_reg_sv != DEFSV) {
-               /* SAVE_DEFSV does *not* suffice here for USE_THREADS */
+               /* SAVE_DEFSV does *not* suffice here for USE_5005THREADS */
                SAVESPTR(DEFSV);
                DEFSV = PL_reg_sv;
            }
@@ -2009,7 +2003,7 @@ S_regmatch(pTHX_ regnode *prog)
 #if 0
     I32 firstcp = PL_savestack_ix;
 #endif
-    register bool do_utf8 = DO_UTF8(PL_reg_sv);
+    register bool do_utf8 = PL_reg_sv_utf8;
 
 #ifdef DEBUGGING
     PL_regindent++;
@@ -3033,12 +3027,15 @@ S_regmatch(pTHX_ regnode *prog)
                minmod = 0;
                if (ln && regrepeat_hard(scan, ln, &l) < ln)
                    sayNO;
-               if (ln && l == 0 && n >= ln
-                   /* In fact, this is tricky.  If paren, then the
-                      fact that we did/didnot match may influence
-                      future execution. */
-                   && !(paren && ln == 0))
-                   ln = n;
+               /* if we matched something zero-length we don't need to
+                  backtrack - capturing parens are already defined, so
+                  the caveat in the maximal case doesn't apply
+
+                  XXXX if ln == 0, we can redo this check first time
+                  through the following loop
+               */
+               if (ln && l == 0)
+                   n = ln;     /* don't backtrack */
                locinput = PL_reginput;
                if (PL_regkind[(U8)OP(next)] == EXACT) {
                    c1 = (U8)*STRING(next);
@@ -3060,7 +3057,7 @@ S_regmatch(pTHX_ regnode *prog)
                        UCHARAT(PL_reginput) == c2)
                    {
                        if (paren) {
-                           if (n) {
+                           if (ln) {
                                PL_regstartp[paren] =
                                    HOPc(PL_reginput, -l) - PL_bostr;
                                PL_regendp[paren] = PL_reginput - PL_bostr;
@@ -3084,12 +3081,13 @@ S_regmatch(pTHX_ regnode *prog)
            }
            else {
                n = regrepeat_hard(scan, n, &l);
-               if (n != 0 && l == 0
-                   /* In fact, this is tricky.  If paren, then the
-                      fact that we did/didnot match may influence
-                      future execution. */
-                   && !(paren && ln == 0))
-                   ln = n;
+               /* if we matched something zero-length we don't need to
+                  backtrack, unless the minimum count is zero and we
+                  are capturing the result - in that case the capture
+                  being defined or not may affect later execution
+               */
+               if (n != 0 && l == 0 && !(paren && ln == 0))
+                   ln = n;     /* don't backtrack */
                locinput = PL_reginput;
                DEBUG_r(
                    PerlIO_printf(Perl_debug_log,
@@ -3586,7 +3584,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max)
     register I32 c;
     register char *loceol = PL_regeol;
     register I32 hardcount = 0;
-    register bool do_utf8 = DO_UTF8(PL_reg_sv);
+    register bool do_utf8 = PL_reg_sv_utf8;
 
     scan = PL_reginput;
     if (max != REG_INFTY && max < loceol - scan)
@@ -3825,7 +3823,7 @@ S_regrepeat_hard(pTHX_ regnode *p, I32 max, I32 *lp)
        return 0;
 
     start = PL_reginput;
-    if (DO_UTF8(PL_reg_sv)) {
+    if (PL_reg_sv_utf8) {
        while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) {
            if (!count++) {
                l = 0;
@@ -4061,12 +4059,8 @@ S_reghopmaybe3(pTHX_ U8* s, I32 off, U8* lim)
     return s;
 }
 
-#ifdef PERL_OBJECT
-#include "XSUB.h"
-#endif
-
 static void
-restore_pos(pTHXo_ void *arg)
+restore_pos(pTHX_ void *arg)
 {
     if (PL_reg_eval_set) {
        if (PL_reg_oldsaved) {