regcomp.c: Generate EXACTFU_SS only for non-UTF8

It turns out that now, the regular methods for handling multi-character
folds work for the ones involving LATIN SMALL LETTER SHARP S when the
pattern is in UTF-8.  So the special code for handling this case can be
removed, and a regular EXACTFU node is generated.  This has the
advantage of being trie-able, and requiring fewer operations at run
time, as the pattern is pre-folded at compile time, and doesn't have to
be re-folded during each backtracking at run-time.

This means that the EXACTFU_SS node type will only be generated for
non-UTF-8 patterns, and the handling of it is unchanged in these cases.

diff --git a/pod/perldebguts.pod b/pod/perldebguts.pod
index b076bcf..f670bb2 100644 (file)
--- a/pod/perldebguts.pod
+++ b/pod/perldebguts.pod
@@ -667,8 +667,8 @@ will be lost.
                              folded length <= unfolded).
 
  EXACTFU_SS       str        Match this string using /iu rules (w/len);
                              folded length <= unfolded).
 
  EXACTFU_SS       str        Match this string using /iu rules (w/len);

-                             (string folded iff in UTF-8; non-UTF8
-                             folded length > unfolded).
+                             (string not UTF-8, only portions guaranteed
+                             to be folded; folded length > unfolded).

  EXACTFLU8        str        Like EXACTFU, but use /il, UTF-8, folded,
                              and everything in it is above 255.
  EXACTFAA_NO_TRIE str        Match this string using /iaa rules (w/len)
  EXACTFLU8        str        Like EXACTFU, but use /il, UTF-8, folded,
                              and everything in it is above 255.
  EXACTFAA_NO_TRIE str        Match this string using /iaa rules (w/len)

diff --git a/regcomp.c b/regcomp.c
index 6f02661..11a155d 100644 (file)
--- a/regcomp.c
+++ b/regcomp.c
@@ -3856,16 +3856,18 @@ S_construct_ahocorasick_from_trie(pTHX_ RExC_state_t *pRExC_state, regnode *sour
  *      so that the optimizer doesn't reject these possibilities based on size
  *      constraints.
  * 2)   For the sequence involving the Sharp s (\xDF), the node type EXACTFU_SS
  *      so that the optimizer doesn't reject these possibilities based on size
  *      constraints.
  * 2)   For the sequence involving the Sharp s (\xDF), the node type EXACTFU_SS

- *      is used for an EXACTFU node that contains at least one "ss" sequence in
- *      it.  For non-UTF-8 patterns and strings, this is the only case where
- *      there is a possible fold length change.  That means that a regular
- *      EXACTFU node without UTF-8 involvement doesn't have to concern itself
- *      with length changes, and so can be processed faster.  regexec.c takes
- *      advantage of this.  Generally, an EXACTFish node that is in UTF-8 is
- *      pre-folded by regcomp.c (except EXACTFL, some of whose folds aren't
- *      known until runtime).  This saves effort in regex matching.  However,
- *      the pre-folding isn't done for non-UTF8 patterns because the fold of
- *      the MICRO SIGN requires UTF-8, and we don't want to slow things down by
+ *      is used in non-UTF-8 patterns for an EXACTFU node that contains at
+ *      least one "ss" sequence in it.  For UTF-8 patterns, the procedures in
+ *      step 1) above are sufficient to handle these, but for non-UTF-8
+ *      patterns and strings, this is the only case where there is a possible
+ *      fold length change.  That means that a regular EXACTFU node without
+ *      UTF-8 involvement doesn't have to concern itself with length changes,
+ *      and so can be processed faster.  regexec.c takes advantage of this.
+ *      Generally, an EXACTFish node that is in UTF-8 is pre-folded by
+ *      regcomp.c (except EXACTFL, some of whose folds aren't known until
+ *      runtime).  This saves effort in regex matching.  However, the
+ *      pre-folding isn't done for non-UTF8 patterns because the fold of the
+ *      MICRO SIGN requires UTF-8, and we don't want to slow things down by

  *      forcing the pattern into UTF8 unless necessary.  Also what EXACTF (and,
  *      again, EXACTFL) nodes fold to isn't known until runtime.  The fold
  *      possibilities for the non-UTF8 patterns are quite simple, except for
  *      forcing the pattern into UTF8 unless necessary.  Also what EXACTF (and,
  *      again, EXACTFL) nodes fold to isn't known until runtime.  The fold
  *      possibilities for the non-UTF8 patterns are quite simple, except for


@@ -4232,19 +4234,7 @@ S_join_exact(pTHX_ RExC_state_t *pRExC_state, regnode *scan,
                     continue;
                 }
 
                     continue;
                 }
 

-                /* Nodes with 'ss' require special handling, except for
-                 * EXACTFAA-ish for which there is no multi-char fold to this */
-                if (len == 2 && *s == 's' && *(s+1) == 's'
-                    && OP(scan) != EXACTFAA
-                    && OP(scan) != EXACTFAA_NO_TRIE)
-                {
-                    count = 2;
-                    if (OP(scan) != EXACTFL) {
-                        OP(scan) = EXACTFU_SS;
-                    }
-                    s += 2;
-                }
-                else { /* Here is a generic multi-char fold. */
+                { /* Here is a generic multi-char fold. */

                     U8* multi_end  = s + len;
 
                     /* Count how many characters are in it.  In the case of
                     U8* multi_end  = s + len;
 
                     /* Count how many characters are in it.  In the case of

diff --git a/regcomp.sym b/regcomp.sym
index ddf5ba8..235305d 100644 (file)
--- a/regcomp.sym
+++ b/regcomp.sym
@@ -107,7 +107,7 @@ EXACTFAA    EXACT,      str   ; Match this string using /iaa rules (w/len) (stri
 
 # End of important relative ordering.
 
 
 # End of important relative ordering.
 

-EXACTFU_SS  EXACT,      str      ; Match this string using /iu rules (w/len); (string folded iff in UTF-8; non-UTF8 folded length > unfolded).
+EXACTFU_SS  EXACT,      str      ; Match this string using /iu rules (w/len); (string not UTF-8, only portions guaranteed to be folded; folded length > unfolded).

 EXACTFLU8   EXACT,      str      ; Like EXACTFU, but use /il, UTF-8, folded, and everything in it is above 255.
 EXACTFAA_NO_TRIE  EXACT, str     ; Match this string using /iaa rules (w/len) (string not UTF-8, not guaranteed to be folded, not currently trie-able).
 
 EXACTFLU8   EXACT,      str      ; Like EXACTFU, but use /il, UTF-8, folded, and everything in it is above 255.
 EXACTFAA_NO_TRIE  EXACT, str     ; Match this string using /iaa rules (w/len) (string not UTF-8, not guaranteed to be folded, not currently trie-able).
 

diff --git a/regexec.c b/regexec.c
index f7da372..f025965 100644 (file)
--- a/regexec.c
+++ b/regexec.c
@@ -2278,9 +2278,7 @@ S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s,
         goto do_exactf_non_utf8;
 
     case EXACTFU_SS:
         goto do_exactf_non_utf8;
 
     case EXACTFU_SS:

-        if (is_utf8_pat) {
-            utf8_fold_flags = FOLDEQ_S2_ALREADY_FOLDED;
-        }
+        assert(! is_utf8_pat);

         goto do_exactf_utf8;
 
     case EXACTFLU8:
         goto do_exactf_utf8;
 
     case EXACTFLU8:


@@ -6378,6 +6376,8 @@ S_regmatch(pTHX_ regmatch_info *reginfo, char *startpos, regnode *prog)
            goto do_exactf;
 
        case EXACTFU_SS:         /*  /\x{df}/iu   */
            goto do_exactf;
 
        case EXACTFU_SS:         /*  /\x{df}/iu   */

+            assert(! is_utf8_pat);
+            /* FALLTHROUGH */

        case EXACTFU:            /*  /abc/iu      */
            folder = foldEQ_latin1;
            fold_array = PL_fold_latin1;
        case EXACTFU:            /*  /abc/iu      */
            folder = foldEQ_latin1;
            fold_array = PL_fold_latin1;


@@ -9131,7 +9131,7 @@ S_regrepeat(pTHX_ regexp *prog, char **startposp, const regnode *p,
     I32 hardcount = 0;  /* How many matches so far */
     bool utf8_target = reginfo->is_utf8_target;
     unsigned int to_complement = 0;  /* Invert the result? */
     I32 hardcount = 0;  /* How many matches so far */
     bool utf8_target = reginfo->is_utf8_target;
     unsigned int to_complement = 0;  /* Invert the result? */

-    UV utf8_flags;
+    UV utf8_flags = 0;

     _char_class_number classnum;
 
     PERL_ARGS_ASSERT_REGREPEAT;
     _char_class_number classnum;
 
     PERL_ARGS_ASSERT_REGREPEAT;


@@ -9277,7 +9277,6 @@ S_regrepeat(pTHX_ regexp *prog, char **startposp, const regnode *p,
 
     case EXACTF:   /* This node only generated for non-utf8 patterns */
         assert(! reginfo->is_utf8_pat);
 
     case EXACTF:   /* This node only generated for non-utf8 patterns */
         assert(! reginfo->is_utf8_pat);

-        utf8_flags = 0;

         goto do_exactf;
 
     case EXACTFLU8:
         goto do_exactf;
 
     case EXACTFLU8:


@@ -9296,9 +9295,13 @@ S_regrepeat(pTHX_ regexp *prog, char **startposp, const regnode *p,
        utf8_flags = FOLDEQ_S2_ALREADY_FOLDED;
         goto do_exactf;
 
        utf8_flags = FOLDEQ_S2_ALREADY_FOLDED;
         goto do_exactf;
 

-    case EXACTFU_SS:

     case EXACTFU:
     case EXACTFU:

-       utf8_flags = reginfo->is_utf8_pat ? FOLDEQ_S2_ALREADY_FOLDED : 0;
+       if (reginfo->is_utf8_pat) {
+            utf8_flags = FOLDEQ_S2_ALREADY_FOLDED;
+        }
+        /* FALLTHROUGH */
+
+    case EXACTFU_SS:

 
       do_exactf: {
         int c1, c2;
 
       do_exactf: {
         int c1, c2;

diff --git a/regnodes.h b/regnodes.h
index 94b4443..9fd01a7 100644 (file)
--- a/regnodes.h
+++ b/regnodes.h
@@ -53,7 +53,7 @@
 #define        EXACTFL                 39      /* 0x27 Match this string using /il rules (w/len); (string not guaranteed to be folded). */
 #define        EXACTFU                 40      /* 0x28 Match this string using /iu rules (w/len); (string folded iff in UTF-8; non-UTF8 folded length <= unfolded). */
 #define        EXACTFAA                41      /* 0x29 Match this string using /iaa rules (w/len) (string folded iff in UTF-8; non-UTF8 folded length <= unfolded). */
 #define        EXACTFL                 39      /* 0x27 Match this string using /il rules (w/len); (string not guaranteed to be folded). */
 #define        EXACTFU                 40      /* 0x28 Match this string using /iu rules (w/len); (string folded iff in UTF-8; non-UTF8 folded length <= unfolded). */
 #define        EXACTFAA                41      /* 0x29 Match this string using /iaa rules (w/len) (string folded iff in UTF-8; non-UTF8 folded length <= unfolded). */

-#define        EXACTFU_SS              42      /* 0x2a Match this string using /iu rules (w/len); (string folded iff in UTF-8; non-UTF8 folded length > unfolded). */
+#define        EXACTFU_SS              42      /* 0x2a Match this string using /iu rules (w/len); (string not UTF-8, only portions guaranteed to be folded; folded length > unfolded). */

 #define        EXACTFLU8               43      /* 0x2b Like EXACTFU, but use /il, UTF-8, folded, and everything in it is above 255. */
 #define        EXACTFAA_NO_TRIE        44      /* 0x2c Match this string using /iaa rules (w/len) (string not UTF-8, not guaranteed to be folded, not currently trie-able). */
 #define        EXACT_ONLY8             45      /* 0x2d Like EXACT, but only UTF-8 encoded targets can match */
 #define        EXACTFLU8               43      /* 0x2b Like EXACTFU, but use /il, UTF-8, folded, and everything in it is above 255. */
 #define        EXACTFAA_NO_TRIE        44      /* 0x2c Match this string using /iaa rules (w/len) (string not UTF-8, not guaranteed to be folded, not currently trie-able). */
 #define        EXACT_ONLY8             45      /* 0x2d Like EXACT, but only UTF-8 encoded targets can match */