Add Daniel Grisinger <dgris@dimensional.com>.

[p5sagit/p5-mst-13.2.git] / regcomp.c

diff --git a/regcomp.c b/regcomp.c
index fb2993b..f78388b 100644 (file)
--- a/regcomp.c
+++ b/regcomp.c
@@ -146,6 +146,7 @@ static void regtail _((regnode *, regnode *));
 static char* regwhite _((char *, char *));
 static char* nextchar _((void));
 static void re_croak2 _((const char* pat1,const char* pat2,...)) __attribute__((noreturn));
+static char* regpposixcc _((I32 value));
 #endif
 
 /* Length of a variant. */
@@ -776,7 +777,7 @@ reginitcolors(void)
                PL_colors[i] = ++s;
            }
            else
-               PL_colors[i] = "";
+               PL_colors[i] = s = "";
        }
     } else {
        while (i < 6) 
@@ -824,10 +825,8 @@ pregcomp(char *exp, char *xend, PMOP *pm)
        PL_reg_flags = 0;
 
     PL_regprecomp = savepvn(exp, xend - exp);
-    DEBUG_r(
-       if (!PL_colorset)
-           reginitcolors();
-       PerlIO_printf(Perl_debug_log, "%sCompiling%s RE `%s%*s%s'\n",
+    DEBUG_r(if (!PL_colorset) reginitcolors());
+    DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling%s RE `%s%*s%s'\n",
                      PL_colors[4],PL_colors[5],PL_colors[0],
                      xend - exp, PL_regprecomp, PL_colors[1]));
     PL_regflags = pm->op_pmflags;
@@ -1880,6 +1879,8 @@ tryagain:
                FAIL("trailing \\ in regexp");
            /* FALL THROUGH */
        default:
+           /* Do not generate `unrecognized' warnings here, we fall
+              back into the quick-grab loop below */
            goto defchar;
        }
        break;
@@ -2010,6 +2011,11 @@ tryagain:
                            FAIL("trailing \\ in regexp");
                        /* FALL THROUGH */
                    default:
+                       if (!SIZE_ONLY && ckWARN(WARN_UNSAFE) && isALPHA(*p))
+                           warner(WARN_UNSAFE, 
+                                  "/%.127s/: Unrecognized escape \\%c passed through",
+                                  PL_regprecomp,
+                                  *p);
                        goto normal_default;
                    }
                    break;
@@ -2095,6 +2101,45 @@ regwhite(char *p, char *e)
     return p;
 }
 
+/* parse POSIX character classes like [[:foo:]] */
+STATIC char*
+regpposixcc(I32 value)
+{
+    dTHR;
+    char *posixcc = 0;
+
+    if (value == '[' && PL_regcomp_parse + 1 < PL_regxend &&
+       /* I smell either [: or [= or [. -- POSIX has been here, right? */
+       (*PL_regcomp_parse == ':' ||
+        *PL_regcomp_parse == '=' ||
+        *PL_regcomp_parse == '.')) {
+       char  c = *PL_regcomp_parse;
+       char* s = PL_regcomp_parse++;
+           
+       while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != c)
+           PL_regcomp_parse++;
+       if (PL_regcomp_parse == PL_regxend)
+           /* Grandfather lone [:, [=, [. */
+           PL_regcomp_parse = s;
+       else {
+           PL_regcomp_parse++; /* skip over the c */
+           if (*PL_regcomp_parse == ']') {
+               /* Not Implemented Yet.
+                * (POSIX Extended Character Classes, that is)
+                * The text between e.g. [: and :] would start
+                * at s + 1 and stop at regcomp_parse - 2. */
+               if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY)
+                   warner(WARN_UNSAFE,
+                          "Character class syntax [%c %c] is reserved for future extensions", c, c);
+               PL_regcomp_parse++; /* skip over the ending ] */
+               posixcc = s + 1;
+           }
+       }
+    }
+
+    return posixcc;
+}
+
 STATIC regnode *
 regclass(void)
 {
@@ -2132,32 +2177,9 @@ regclass(void)
     while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') {
        skipcond:
        value = UCHARAT(PL_regcomp_parse++);
-       if (value == '[' && PL_regcomp_parse + 1 < PL_regxend &&
-           /* I smell either [: or [= or [. -- POSIX has been here, right? */
-           (*PL_regcomp_parse == ':' || *PL_regcomp_parse == '=' || *PL_regcomp_parse == '.')) {
-           char  posixccc = *PL_regcomp_parse;
-           char* posixccs = PL_regcomp_parse++;
-           
-           while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != posixccc)
-               PL_regcomp_parse++;
-           if (PL_regcomp_parse == PL_regxend)
-               /* Grandfather lone [:, [=, [. */
-               PL_regcomp_parse = posixccs;
-           else {
-               PL_regcomp_parse++; /* skip over the posixccc */
-               if (*PL_regcomp_parse == ']') {
-                   /* Not Implemented Yet.
-                    * (POSIX Extended Character Classes, that is)
-                    * The text between e.g. [: and :] would start
-                    * at posixccs + 1 and stop at regcomp_parse - 2. */
-                   if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY)
-                       warner(WARN_UNSAFE,
-                           "Character class syntax [%c %c] is reserved for future extensions", posixccc, posixccc);
-                   PL_regcomp_parse++; /* skip over the ending ] */
-               }
-           }
-       }
-       if (value == '\\') {
+       if (value == '[')
+           (void)regpposixcc(value); /* ignore the return value for now */
+       else if (value == '\\') {
            value = UCHARAT(PL_regcomp_parse++);
            switch (value) {
            case 'w':
@@ -2352,33 +2374,9 @@ regclassutf8(void)
        value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen);
        PL_regcomp_parse += numlen;
 
-       if (value == '[' && PL_regcomp_parse + 1 < PL_regxend &&
-           /* I smell either [: or [= or [. -- POSIX has been here, right? */
-           (*PL_regcomp_parse == ':' || *PL_regcomp_parse == '=' || *PL_regcomp_parse == '.')) {
-           char  posixccc = *PL_regcomp_parse;
-           char* posixccs = PL_regcomp_parse++;
-           
-           while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != posixccc)
-               PL_regcomp_parse++;
-           if (PL_regcomp_parse == PL_regxend)
-               /* Grandfather lone [:, [=, [. */
-               PL_regcomp_parse = posixccs;
-           else {
-               PL_regcomp_parse++; /* skip over the posixccc */
-               if (*PL_regcomp_parse == ']') {
-                   /* Not Implemented Yet.
-                    * (POSIX Extended Character Classes, that is)
-                    * The text between e.g. [: and :] would start
-                    * at posixccs + 1 and stop at regcomp_parse - 2. */
-                   if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY)
-                       warner(WARN_UNSAFE,
-                           "Character class syntax [%c %c] is reserved for future extensions", posixccc, posixccc);
-                   PL_regcomp_parse++; /* skip over the ending ] */
-               }
-           }
-       }
-
-       if (value == '\\') {
+       if (value == '[')
+           (void)regpposixcc(value); /* ignore the return value for now */
+       else if (value == '\\') {
            value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen);
            PL_regcomp_parse += numlen;
            switch (value) {
@@ -3222,4 +3220,15 @@ save_re_context(void)
     SAVESPTR(PL_regcode);              /* Code-emit pointer; &regdummy = don't */
     SAVEPPTR(PL_regxend);              /* End of input for compile */
     SAVEPPTR(PL_regcomp_parse);                /* Input-scan pointer. */
+    SAVESPTR(PL_reg_call_cc);          /* from regexec.c */
+    SAVESPTR(PL_reg_re);               /* from regexec.c */
+    SAVEPPTR(PL_reg_ganch);            /* from regexec.c */
+    SAVESPTR(PL_reg_sv);               /* from regexec.c */
+    SAVESPTR(PL_reg_magic);            /* from regexec.c */
+    SAVEI32(PL_reg_oldpos);                    /* from regexec.c */
+    SAVESPTR(PL_reg_oldcurpm);         /* from regexec.c */
+    SAVESPTR(PL_reg_curpm);            /* from regexec.c */
+#ifdef DEBUGGING
+    SAVEPPTR(PL_reg_starttry);         /* from regexec.c */    
+#endif
 }