*/
#ifdef PERL_EXT_RE_BUILD
-/* need to replace pregcomp et al, so enable that */
-# ifndef PERL_IN_XSUB_RE
-# define PERL_IN_XSUB_RE
-# endif
-/* need access to debugger hooks */
-# if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING)
-# define DEBUGGING
-# endif
-#endif
-
-#ifdef PERL_IN_XSUB_RE
-/* We *really* need to overwrite these symbols: */
-# define Perl_regexec_flags my_regexec
-# define Perl_regdump my_regdump
-# define Perl_regprop my_regprop
-# define Perl_re_intuit_start my_re_intuit_start
-/* *These* symbols are masked to allow static link. */
-# define Perl_pregexec my_pregexec
-# define Perl_reginitcolors my_reginitcolors
-# define Perl_regclass_swash my_regclass_swash
-
-# define PERL_NO_GET_CONTEXT
+#include "re_top.h"
#endif
/*
#define PERL_IN_REGEXEC_C
#include "perl.h"
-#include "regcomp.h"
+#ifdef PERL_IN_XSUB_RE
+# include "re_comp.h"
+#else
+# include "regcomp.h"
+#endif
#define RF_tainted 1 /* tainted information used? */
#define RF_warned 2 /* warned about big count? */
* pregexec and friends
*/
+#ifndef PERL_IN_XSUB_RE
/*
- pregexec - match a regexp against a string
*/
regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
nosave ? 0 : REXEC_COPY_STR);
}
-
+#endif
/*
* Need to implement the following flags for reg_anch:
/* Make sure there is a test for this +1 options in re_tests */
#define TRIE_INITAL_ACCEPT_BUFFLEN 4;
+/* this value indiciates that the c1/c2 "next char" test should be skipped */
+#define CHRTEST_VOID -1000
+
#define SLAB_FIRST(s) (&(s)->states[0])
#define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1])
STATIC regmatch_state *
S_push_slab(pTHX)
{
+#if PERL_VERSION < 9
+ dMY_CXT;
+#endif
regmatch_slab *s = PL_regmatch_slab->next;
if (!s) {
Newx(s, 1, regmatch_slab);
STATIC I32 /* 0 failure, 1 success */
S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
{
+#if PERL_VERSION < 9
+ dMY_CXT;
+#endif
dVAR;
register const bool do_utf8 = PL_reg_match_utf8;
const U32 uniflags = UTF8_ALLOW_DEFAULT;
scan += NEXT_OFF(scan); /* Skip former OPEN. */
PL_reginput = locinput;
st->u.curlym.maxwanted = st->minmod ? st->ln : n;
- if (st->u.curlym.maxwanted) {
- while (PL_reginput < PL_regeol && st->u.curlym.matches < st->u.curlym.maxwanted) {
- /* resume to current state on success */
- st->u.yes.prev_yes_state = yes_state;
- yes_state = st;
- REGMATCH(scan, CURLYM1);
- yes_state = st->u.yes.prev_yes_state;
- /*** all unsaved local vars undefined at this point */
- if (!result)
- break;
- /* on first match, determine length, u.curlym.l */
- if (!st->u.curlym.matches++) {
- if (PL_reg_match_utf8) {
- char *s = locinput;
- while (s < PL_reginput) {
- st->u.curlym.l++;
- s += UTF8SKIP(s);
- }
- }
- else {
- st->u.curlym.l = PL_reginput - locinput;
- }
- if (st->u.curlym.l == 0) {
- st->u.curlym.matches = st->u.curlym.maxwanted;
- break;
+ while (PL_reginput < PL_regeol && st->u.curlym.matches < st->u.curlym.maxwanted) {
+ /* resume to current state on success */
+ st->u.yes.prev_yes_state = yes_state;
+ yes_state = st;
+ REGMATCH(scan, CURLYM1);
+ yes_state = st->u.yes.prev_yes_state;
+ /*** all unsaved local vars undefined at this point */
+ if (!result)
+ break;
+ /* on first match, determine length, u.curlym.l */
+ if (!st->u.curlym.matches++) {
+ if (PL_reg_match_utf8) {
+ char *s = locinput;
+ while (s < PL_reginput) {
+ st->u.curlym.l++;
+ s += UTF8SKIP(s);
}
}
- locinput = PL_reginput;
+ else {
+ st->u.curlym.l = PL_reginput - locinput;
+ }
+ if (st->u.curlym.l == 0) {
+ st->u.curlym.matches = st->u.curlym.maxwanted;
+ break;
+ }
}
+ locinput = PL_reginput;
}
PL_reginput = locinput;
-
- if (st->minmod) {
+ if (st->u.curlym.matches < st->ln) {
st->minmod = 0;
- if (st->ln && st->u.curlym.matches < st->ln)
- sayNO;
- if (HAS_TEXT(next) || JUMPABLE(next)) {
- regnode *text_node = next;
+ sayNO;
+ }
- if (! HAS_TEXT(text_node)) FIND_NEXT_IMPT(text_node);
+ DEBUG_EXECUTE_r(
+ PerlIO_printf(Perl_debug_log,
+ "%*s matched %"IVdf" times, len=%"IVdf"...\n",
+ (int)(REPORT_CODE_OFF+PL_regindent*2), "",
+ (IV) st->u.curlym.matches, (IV)st->u.curlym.l)
+ );
+
+ /* calculate c1 and c1 for possible match of 1st char
+ * following curly */
+ st->u.curlym.c1 = st->u.curlym.c2 = CHRTEST_VOID;
+ if (HAS_TEXT(next) || JUMPABLE(next)) {
+ regnode *text_node = next;
+ if (! HAS_TEXT(text_node)) FIND_NEXT_IMPT(text_node);
+ if (HAS_TEXT(text_node)
+ && PL_regkind[(U8)OP(text_node)] != REF)
+ {
+ st->u.curlym.c1 = (U8)*STRING(text_node);
+ st->u.curlym.c2 =
+ (OP(text_node) == EXACTF || OP(text_node) == REFF)
+ ? PL_fold[st->u.curlym.c1]
+ : (OP(text_node) == EXACTFL || OP(text_node) == REFFL)
+ ? PL_fold_locale[st->u.curlym.c1]
+ : st->u.curlym.c1;
+ }
+ }
- if (! HAS_TEXT(text_node)) st->u.curlym.c1 = st->u.curlym.c2 = -1000;
- else {
- if (PL_regkind[(U8)OP(text_node)] == REF) {
- st->u.curlym.c1 = st->u.curlym.c2 = -1000;
- goto assume_ok_MM;
+ REGCP_SET(st->u.curlym.lastcp);
+
+ st->u.curlym.minmod = st->minmod;
+ st->minmod = 0;
+ while (st->u.curlym.matches >= st->ln
+ && (st->u.curlym.matches <= n
+ /* for REG_INFTY, ln could overflow to negative */
+ || (n == REG_INFTY && st->u.curlym.matches >= 0)))
+ {
+ /* If it could work, try it. */
+ if (st->u.curlym.c1 == CHRTEST_VOID ||
+ UCHARAT(PL_reginput) == st->u.curlym.c1 ||
+ UCHARAT(PL_reginput) == st->u.curlym.c2)
+ {
+ DEBUG_EXECUTE_r(
+ PerlIO_printf(Perl_debug_log,
+ "%*s trying tail with matches=%"IVdf"...\n",
+ (int)(REPORT_CODE_OFF+PL_regindent*2),
+ "", (IV)st->u.curlym.matches)
+ );
+ if (st->u.curlym.paren) {
+ if (st->u.curlym.matches) {
+ PL_regstartp[st->u.curlym.paren]
+ = HOPc(PL_reginput, -st->u.curlym.l) - PL_bostr;
+ PL_regendp[st->u.curlym.paren] = PL_reginput - PL_bostr;
}
- else { st->u.curlym.c1 = (U8)*STRING(text_node); }
- if (OP(text_node) == EXACTF || OP(text_node) == REFF)
- st->u.curlym.c2 = PL_fold[st->u.curlym.c1];
- else if (OP(text_node) == EXACTFL || OP(text_node) == REFFL)
- st->u.curlym.c2 = PL_fold_locale[st->u.curlym.c1];
else
- st->u.curlym.c2 = st->u.curlym.c1;
+ PL_regendp[st->u.curlym.paren] = -1;
}
+ /* resume to current state on success */
+ st->u.yes.prev_yes_state = yes_state;
+ yes_state = st;
+ REGMATCH(next, CURLYM2);
+ yes_state = st->u.yes.prev_yes_state;
+ /*** all unsaved local vars undefined at this point */
+ if (result)
+ /* XXX tmp sayYES; */
+ sayYES_FINAL;
+ REGCP_UNWIND(st->u.curlym.lastcp);
}
- else
- st->u.curlym.c1 = st->u.curlym.c2 = -1000;
- assume_ok_MM:
- REGCP_SET(st->u.curlym.lastcp);
- while (n >= st->ln || (n == REG_INFTY && st->ln > 0)) { /* ln overflow ? */
- /* If it could work, try it. */
- if (st->u.curlym.c1 == -1000 ||
- UCHARAT(PL_reginput) == st->u.curlym.c1 ||
- UCHARAT(PL_reginput) == st->u.curlym.c2)
- {
- if (st->u.curlym.paren) {
- if (st->ln) {
- PL_regstartp[st->u.curlym.paren] =
- HOPc(PL_reginput, -st->u.curlym.l) - PL_bostr;
- PL_regendp[st->u.curlym.paren] = PL_reginput - PL_bostr;
- }
- else
- PL_regendp[st->u.curlym.paren] = -1;
- }
- /* resume to current state on success */
- st->u.yes.prev_yes_state = yes_state;
- yes_state = st;
- REGMATCH(next, CURLYM2);
- yes_state = st->u.yes.prev_yes_state;
- /*** all unsaved local vars undefined at this point */
- if (result)
- /* XXX tmp sayYES; */
- sayYES_FINAL;
- REGCP_UNWIND(st->u.curlym.lastcp);
- }
- /* Couldn't or didn't -- move forward. */
+ /* Couldn't or didn't -- move forward/backward. */
+ if (st->u.curlym.minmod) {
PL_reginput = locinput;
/* resume to current state on success */
st->u.yes.prev_yes_state = yes_state;
yes_state = st->u.yes.prev_yes_state;
/*** all unsaved local vars undefined at this point */
if (result) {
- st->ln++;
+ st->u.curlym.matches++;
locinput = PL_reginput;
}
else
sayNO;
}
- }
- else {
- DEBUG_EXECUTE_r(
- PerlIO_printf(Perl_debug_log,
- "%*s matched %"IVdf" times, len=%"IVdf"...\n",
- (int)(REPORT_CODE_OFF+PL_regindent*2), "",
- (IV) st->u.curlym.matches, (IV)st->u.curlym.l)
- );
- if (st->u.curlym.matches >= st->ln) {
- if (HAS_TEXT(next) || JUMPABLE(next)) {
- regnode *text_node = next;
-
- if (! HAS_TEXT(text_node)) FIND_NEXT_IMPT(text_node);
-
- if (! HAS_TEXT(text_node)) st->u.curlym.c1 = st->u.curlym.c2 = -1000;
- else {
- if (PL_regkind[(U8)OP(text_node)] == REF) {
- st->u.curlym.c1 = st->u.curlym.c2 = -1000;
- goto assume_ok_REG;
- }
- else { st->u.curlym.c1 = (U8)*STRING(text_node); }
-
- if (OP(text_node) == EXACTF || OP(text_node) == REFF)
- st->u.curlym.c2 = PL_fold[st->u.curlym.c1];
- else if (OP(text_node) == EXACTFL || OP(text_node) == REFFL)
- st->u.curlym.c2 = PL_fold_locale[st->u.curlym.c1];
- else
- st->u.curlym.c2 = st->u.curlym.c1;
- }
- }
- else
- st->u.curlym.c1 = st->u.curlym.c2 = -1000;
- }
- assume_ok_REG:
- REGCP_SET(st->u.curlym.lastcp);
- while (st->u.curlym.matches >= st->ln) {
- /* If it could work, try it. */
- if (st->u.curlym.c1 == -1000 ||
- UCHARAT(PL_reginput) == st->u.curlym.c1 ||
- UCHARAT(PL_reginput) == st->u.curlym.c2)
- {
- DEBUG_EXECUTE_r(
- PerlIO_printf(Perl_debug_log,
- "%*s trying tail with matches=%"IVdf"...\n",
- (int)(REPORT_CODE_OFF+PL_regindent*2),
- "", (IV)st->u.curlym.matches)
- );
- if (st->u.curlym.paren) {
- if (st->u.curlym.matches) {
- PL_regstartp[st->u.curlym.paren]
- = HOPc(PL_reginput, -st->u.curlym.l) - PL_bostr;
- PL_regendp[st->u.curlym.paren] = PL_reginput - PL_bostr;
- }
- else
- PL_regendp[st->u.curlym.paren] = -1;
- }
- /* resume to current state on success */
- st->u.yes.prev_yes_state = yes_state;
- yes_state = st;
- REGMATCH(next, CURLYM4);
- yes_state = st->u.yes.prev_yes_state;
- /*** all unsaved local vars undefined at this point */
- if (result)
- /* XXX tmp sayYES; */
- sayYES_FINAL;
- REGCP_UNWIND(st->u.curlym.lastcp);
- }
- /* Couldn't or didn't -- back up. */
+ else {
st->u.curlym.matches--;
locinput = HOPc(locinput, -st->u.curlym.l);
PL_reginput = locinput;
if (! HAS_TEXT(text_node)) FIND_NEXT_IMPT(text_node);
- if (! HAS_TEXT(text_node)) st->u.plus.c1 = st->u.plus.c2 = -1000;
+ if (! HAS_TEXT(text_node))
+ st->u.plus.c1 = st->u.plus.c2 = CHRTEST_VOID;
else {
if (PL_regkind[(U8)OP(text_node)] == REF) {
- st->u.plus.c1 = st->u.plus.c2 = -1000;
+ st->u.plus.c1 = st->u.plus.c2 = CHRTEST_VOID;
goto assume_ok_easy;
}
else { s = (U8*)STRING(text_node); }
}
}
else
- st->u.plus.c1 = st->u.plus.c2 = -1000;
+ st->u.plus.c1 = st->u.plus.c2 = CHRTEST_VOID;
assume_ok_easy:
PL_reginput = locinput;
if (st->minmod) {
sayNO;
locinput = PL_reginput;
REGCP_SET(st->u.plus.lastcp);
- if (st->u.plus.c1 != -1000) {
+ if (st->u.plus.c1 != CHRTEST_VOID) {
st->u.plus.old = locinput;
st->u.plus.count = 0;
st->u.plus.count++;
}
} else {
- STRLEN len;
/* count initialised to
* utf8_distance(old, locinput) */
while (locinput <= st->u.plus.e) {
- UV c = utf8n_to_uvchr((U8*)locinput,
+ STRLEN len;
+ const UV c = utf8n_to_uvchr((U8*)locinput,
UTF8_MAXBYTES, &len,
uniflags);
if (c == (UV)st->u.plus.c1 || c == (UV)st->u.plus.c2)
else
while (n >= st->ln || (n == REG_INFTY && st->ln > 0)) { /* ln overflow ? */
UV c;
- if (st->u.plus.c1 != -1000) {
+ if (st->u.plus.c1 != CHRTEST_VOID) {
if (do_utf8)
c = utf8n_to_uvchr((U8*)PL_reginput,
UTF8_MAXBYTES, 0,
}
}
/* If it could work, try it. */
- else if (st->u.plus.c1 == -1000)
+ else if (st->u.plus.c1 == CHRTEST_VOID)
{
TRYPAREN(st->u.plus.paren, st->ln, PL_reginput, PLUS3);
/*** all unsaved local vars undefined at this point */
{
UV c = 0;
while (n >= st->ln) {
- if (st->u.plus.c1 != -1000) {
+ if (st->u.plus.c1 != CHRTEST_VOID) {
if (do_utf8)
c = utf8n_to_uvchr((U8*)PL_reginput,
UTF8_MAXBYTES, 0,
c = UCHARAT(PL_reginput);
}
/* If it could work, try it. */
- if (st->u.plus.c1 == -1000 || c == (UV)st->u.plus.c1 || c == (UV)st->u.plus.c2)
+ if (st->u.plus.c1 == CHRTEST_VOID || c == (UV)st->u.plus.c1 || c == (UV)st->u.plus.c2)
{
TRYPAREN(st->u.plus.paren, n, PL_reginput, PLUS4);
/*** all unsaved local vars undefined at this point */
* pop to the state marked by yes_state and continue from there */
/*XXX tmp for CURLYM*/
- regmatch_slab *oslab = PL_regmatch_slab;
- regmatch_state *ost = st, *oys=yes_state;
+ regmatch_slab * const oslab = PL_regmatch_slab;
+ regmatch_state * const ost = st;
+ regmatch_state * const oys = yes_state;
int odepth = depth;
assert(st != yes_state);
/* Restore parens of the caller without popping the
* savestack */
{
- I32 tmp = PL_savestack_ix;
+ const I32 tmp = PL_savestack_ix;
PL_savestack_ix = st->u.eval.lastcp;
regcppop(rex);
PL_savestack_ix = tmp;
case resume_CURLYM1:
case resume_CURLYM2:
case resume_CURLYM3:
- case resume_CURLYM4:
PL_regmatch_slab =oslab;
st = ost;
PL_regmatch_state = st;
goto resume_point_CURLYM2;
case resume_CURLYM3:
goto resume_point_CURLYM3;
- case resume_CURLYM4:
- goto resume_point_CURLYM4;
case resume_PLUS1:
goto resume_point_PLUS1;
case resume_PLUS2:
goto resume_point_CURLYM2;
case resume_CURLYM3:
goto resume_point_CURLYM3;
- case resume_CURLYM4:
- goto resume_point_CURLYM4;
case resume_IFMATCH:
yes_state = st->u.yes.prev_yes_state;
if (st->logical) {
/* free all slabs above current one */
if (orig_slab->next) {
- regmatch_slab *osl, *sl = orig_slab->next;
+ regmatch_slab *sl = orig_slab->next;
orig_slab->next = NULL;
while (sl) {
- osl = sl;
+ regmatch_slab * const osl = sl;
sl = sl->next;
Safefree(osl);
}
}
+#ifndef PERL_IN_XSUB_RE
/*
- regclass_swash - prepare the utf8 swash
*/
return sw;
}
+#endif
/*
- reginclass - determine if a character falls into a character class