This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regex: Use BOUNDU regnodes
authorKarl Williamson <public@khwilliamson.com>
Tue, 28 Dec 2010 23:13:49 +0000 (16:13 -0700)
committerKarl Williamson <public@khwilliamson.com>
Mon, 17 Jan 2011 02:13:25 +0000 (19:13 -0700)
This refactors one area in regexec.c to use BOUNDU, NBOUNDU for
efficiciency, and easier adding of the future BOUNDA.

regcomp.c
regcomp.sym
regexec.c

index cbcabdf..1eded27 100644 (file)
--- a/regcomp.c
+++ b/regcomp.c
@@ -7399,22 +7399,40 @@ tryagain:
        case 'b':
            RExC_seen_zerolen++;
            RExC_seen |= REG_SEEN_LOOKBEHIND;
-            if (LOC) {
-                ret = reg_node(pRExC_state, (U8)(BOUNDL));
-            } else {
-                ret = reg_node(pRExC_state, (U8)(BOUND));
+           switch (get_regex_charset(RExC_flags)) {
+               case REGEX_LOCALE_CHARSET:
+                   op = BOUNDL;
+                   break;
+               case REGEX_UNICODE_CHARSET:
+                   op = BOUNDU;
+                   break;
+               case REGEX_DEPENDS_CHARSET:
+                   op = BOUND;
+                   break;
+               default:
+                   goto bad_charset;
             }
+           ret = reg_node(pRExC_state, op);
            FLAGS(ret) = get_regex_charset(RExC_flags);
            *flagp |= SIMPLE;
            goto finish_meta_pat;
        case 'B':
            RExC_seen_zerolen++;
            RExC_seen |= REG_SEEN_LOOKBEHIND;
-            if (LOC) {
-                ret = reg_node(pRExC_state, (U8)(NBOUNDL));
-            } else {
-                ret = reg_node(pRExC_state, (U8)(NBOUND));
+           switch (get_regex_charset(RExC_flags)) {
+               case REGEX_LOCALE_CHARSET:
+                   op = NBOUNDL;
+                   break;
+               case REGEX_UNICODE_CHARSET:
+                   op = NBOUNDU;
+                   break;
+               case REGEX_DEPENDS_CHARSET:
+                   op = NBOUND;
+                   break;
+               default:
+                   goto bad_charset;
             }
+           ret = reg_node(pRExC_state, op);
            FLAGS(ret) = get_regex_charset(RExC_flags);
            *flagp |= SIMPLE;
            goto finish_meta_pat;
index 36ebe6a..2fb1785 100644 (file)
@@ -33,6 +33,7 @@ SEOL        EOL,        no        ; Same, assuming singleline.
 BOUND       BOUND,      no        ; Match "" at any word boundary using native charset semantics for non-utf8
 BOUNDL      BOUND,      no        ; Match "" at any locale word boundary
 BOUNDU      BOUND,      no        ; Match "" at any word boundary using Unicode semantics
+# All NBOUND nodes are required by a line regexec.c to be greater than all BOUND ones
 NBOUND      NBOUND,     no        ; Match "" at any word non-boundary using native charset semantics for non-utf8
 NBOUNDL     NBOUND,     no        ; Match "" at any locale word non-boundary
 NBOUNDU     NBOUND,     no        ; Match "" at any word non-boundary using Unicode semantics
index 4416d69..c360cb2 100644 (file)
--- a/regexec.c
+++ b/regexec.c
@@ -1332,6 +1332,63 @@ if ((!reginfo || regtry(reginfo, &s))) \
 #define DUMP_EXEC_POS(li,s,doutf8) \
     dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8)
 
+/* The only difference between the BOUND and NBOUND cases is that
+ * REXEC_FBC_TRYIT is called when matched in BOUND, and when non-matched in
+ * NBOUND.  This is accomplished by passing it in either the if or else clause,
+ * with the other one being empty */
+#define FBC_BOUND(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8) \
+    FBC_BOUND_COMMON(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8, REXEC_FBC_TRYIT, )
+
+#define FBC_NBOUND(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8) \
+    FBC_BOUND_COMMON(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8, , REXEC_FBC_TRYIT)
+
+/* Common to the BOUND and NBOUND cases.  Unfortunately the UTF8 tests need to
+ * be passed in completely with the variable name being tested, which isn't
+ * such a clean interface, but this is easier to read than it was before.  We
+ * are looking for the boundary (or non-boundary between a word and non-word
+ * character.  The utf8 and non-utf8 cases have the same logic, but the details
+ * must be different.  Find the "wordness" of the character just prior to this
+ * one, and compare it with the wordness of this one.  If they differ, we have
+ * a boundary.  At the beginning of the string, pretend that the previous
+ * character was a new-line */
+#define FBC_BOUND_COMMON(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8,                \
+                                                         IF_SUCCESS, IF_FAIL) \
+    if (utf8_target) {                                                         \
+       if (s == PL_bostr) {                                                   \
+           tmp = '\n';                                                        \
+       }                                                                      \
+       else {                                                                 \
+           U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);                 \
+           tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);       \
+       }                                                                      \
+       tmp = TEST1_UTF8;                                                      \
+       LOAD_UTF8_CHARCLASS_ALNUM();                                           \
+       REXEC_FBC_UTF8_SCAN(                                                   \
+           if (tmp == ! (TEST2_UTF8)) {                                       \
+               tmp = !tmp;                                                    \
+               IF_SUCCESS;                                                    \
+           }                                                                  \
+           else {                                                             \
+               IF_FAIL;                                                       \
+           }                                                                  \
+       );                                                                     \
+    }                                                                          \
+    else {  /* Not utf8 */                                                     \
+       tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';                         \
+       tmp = TEST_NON_UTF8(tmp);                                              \
+       REXEC_FBC_SCAN(                                                        \
+           if (tmp == ! TEST_NON_UTF8((U8) *s)) {                             \
+               tmp = !tmp;                                                    \
+               IF_SUCCESS;                                                    \
+           }                                                                  \
+           else {                                                             \
+               IF_FAIL;                                                       \
+           }                                                                  \
+       );                                                                     \
+    }                                                                          \
+    if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s)))           \
+       goto got_it;
+
 /* We know what class REx starts with.  Try to find this position... */
 /* if reginfo is NULL, its a dryrun */
 /* annoyingly all the vars in this routine have different names from their counterparts
@@ -1524,91 +1581,35 @@ S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s,
            break;
        case BOUNDL:
            PL_reg_flags |= RF_tainted;
-           /* FALL THROUGH */
-       case BOUND:
-           if (utf8_target) {
-               if (s == PL_bostr)
-                   tmp = '\n';
-               else {
-                   U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
-                   tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
-               }
-               tmp = (FLAGS(c) != REGEX_LOCALE_CHARSET ?
-                       isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp)));
-               LOAD_UTF8_CHARCLASS_ALNUM();
-               REXEC_FBC_UTF8_SCAN(
-                   if (tmp == !(FLAGS(c) != REGEX_LOCALE_CHARSET ?
-                                cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)) :
-                                isALNUM_LC_utf8((U8*)s)))
-                   {
-                       tmp = !tmp;
-                       REXEC_FBC_TRYIT;
-               }
-               );
-           }
-            else {  /* Not utf8 */
-               tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
-                tmp = ((FLAGS(c) == REGEX_LOCALE_CHARSET)
-                            ? isALNUM_LC(tmp)
-                            : (isWORDCHAR_L1(tmp)
-                               && (isASCII(tmp) || (FLAGS(c) == REGEX_UNICODE_CHARSET))));
-               REXEC_FBC_SCAN(
-                   if (tmp ==
-                        !((FLAGS(c) == REGEX_LOCALE_CHARSET)
-                          ? isALNUM_LC(*s)
-                          : (isWORDCHAR_L1((U8) *s)
-                             && (isASCII((U8) *s) || (FLAGS(c) == REGEX_UNICODE_CHARSET)))))
-                   {
-                       tmp = !tmp;
-                       REXEC_FBC_TRYIT;
-               }
-               );
-           }
-           if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s)))
-               goto got_it;
+           FBC_BOUND(isALNUM_LC,
+                     isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp)),
+                     isALNUM_LC_utf8((U8*)s));
            break;
        case NBOUNDL:
            PL_reg_flags |= RF_tainted;
-           /* FALL THROUGH */
+           FBC_NBOUND(isALNUM_LC,
+                      isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp)),
+                      isALNUM_LC_utf8((U8*)s));
+           break;
+       case BOUND:
+           FBC_BOUND(isWORDCHAR,
+                     isALNUM_uni(tmp),
+                     cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
+           break;
        case NBOUND:
-           if (utf8_target) {
-               if (s == PL_bostr)
-                   tmp = '\n';
-               else {
-                   U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
-                   tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
-               }
-               tmp = (FLAGS(c) != REGEX_LOCALE_CHARSET ?
-                       isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp)));
-               LOAD_UTF8_CHARCLASS_ALNUM();
-               REXEC_FBC_UTF8_SCAN(
-                   if (tmp == !(FLAGS(c) != REGEX_LOCALE_CHARSET ?
-                                cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)) :
-                                isALNUM_LC_utf8((U8*)s)))
-                       tmp = !tmp;
-                   else REXEC_FBC_TRYIT;
-               );
-           }
-           else {
-               tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
-                tmp = ((FLAGS(c) == REGEX_LOCALE_CHARSET)
-                            ? isALNUM_LC(tmp)
-                            : (isWORDCHAR_L1(tmp)
-                               && (isASCII(tmp) || (FLAGS(c) == REGEX_UNICODE_CHARSET))));
-               REXEC_FBC_SCAN(
-                   if (tmp == ! (
-                            (FLAGS(c) == REGEX_LOCALE_CHARSET)
-                            ? isALNUM_LC(*s)
-                            : (isWORDCHAR_L1((U8) *s)
-                               && (isASCII((U8) *s) || (FLAGS(c) == REGEX_UNICODE_CHARSET)))))
-                    {
-                       tmp = !tmp;
-                    }
-                   else REXEC_FBC_TRYIT;
-               );
-           }
-           if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, &s)))
-               goto got_it;
+           FBC_NBOUND(isWORDCHAR,
+                      isALNUM_uni(tmp),
+                      cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
+           break;
+       case BOUNDU:
+           FBC_BOUND(isWORDCHAR_L1,
+                     isALNUM_uni(tmp),
+                     cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
+           break;
+       case NBOUNDU:
+           FBC_NBOUND(isWORDCHAR_L1,
+                      isALNUM_uni(tmp),
+                      cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
            break;
        case ALNUML:
            REXEC_FBC_CSCAN_TAINT(
@@ -3641,12 +3642,18 @@ S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog)
            nextchr = UCHARAT(locinput);
            break;
        }
+
+       /* XXX Could improve efficiency by separating these all out using a
+        * macro or in-line function.  At that point regcomp.c would no longer
+        * have to set the FLAGS fields of these */
        case BOUNDL:
        case NBOUNDL:
            PL_reg_flags |= RF_tainted;
            /* FALL THROUGH */
        case BOUND:
+       case BOUNDU:
        case NBOUND:
+       case NBOUNDU:
            /* was last char in word? */
            if (utf8_target) {
                if (locinput == PL_bostr)
@@ -3656,7 +3663,7 @@ S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog)
 
                    ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags);
                }
-               if (OP(scan) == BOUND || OP(scan) == NBOUND) {
+               if (FLAGS(scan) != REGEX_LOCALE_CHARSET) {
                    ln = isALNUM_uni(ln);
                    LOAD_UTF8_CHARCLASS_ALNUM();
                    n = swash_fetch(PL_utf8_alnum, (U8*)locinput, utf8_target);
@@ -3669,24 +3676,27 @@ S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog)
            else {
                ln = (locinput != PL_bostr) ?
                    UCHARAT(locinput - 1) : '\n';
-               if (FLAGS(scan) == REGEX_UNICODE_CHARSET) {
-
-                    /* Here, can't be BOUNDL or NBOUNDL because they never set
-                     * the flags to REGEX_UNICODE_CHARSET */
-                    ln = isWORDCHAR_L1(ln);
-                    n = isWORDCHAR_L1(nextchr);
-                }
-                else if (OP(scan) == BOUND || OP(scan) == NBOUND) {
-                   ln = isALNUM(ln);
-                   n = isALNUM(nextchr);
-               }
-               else {
-                   ln = isALNUM_LC(ln);
-                   n = isALNUM_LC(nextchr);
+               switch (FLAGS(scan)) {
+                   case REGEX_UNICODE_CHARSET:
+                       ln = isWORDCHAR_L1(ln);
+                       n = isWORDCHAR_L1(nextchr);
+                       break;
+                   case REGEX_LOCALE_CHARSET:
+                       ln = isALNUM_LC(ln);
+                       n = isALNUM_LC(nextchr);
+                       break;
+                   case REGEX_DEPENDS_CHARSET:
+                       ln = isALNUM(ln);
+                       n = isALNUM(nextchr);
+                       break;
+                   default:
+                       Perl_croak(aTHX_ "panic: Unexpected FLAGS %u in op %u", FLAGS(scan), OP(scan));
+                       break;
                }
            }
-           if (((!ln) == (!n)) == (OP(scan) == BOUND ||
-                                   OP(scan) == BOUNDL))
+           /* Note requires that all BOUNDs be lower than all NBOUNDs in
+            * regcomp.sym */
+           if (((!ln) == (!n)) == (OP(scan) < NBOUND))
                    sayNO;
            break;
        case ANYOFV: