X-Git-Url: https://perl5.git.perl.org/perl5.git/blobdiff_plain/5eb114cff44a7920ae194ed22b1af045e54f96d4..2b29d4bc27bf0f6ecfae06ec423a6643b3a6fe03:/inline_invlist.c

diff --git a/inline_invlist.c b/inline_invlist.c
index 936a298..1589f95 100644
--- a/inline_invlist.c
+++ b/inline_invlist.c
@@ -8,59 +8,44 @@
 
 #if defined(PERL_IN_UTF8_C) || defined(PERL_IN_REGCOMP_C) || defined(PERL_IN_REGEXEC_C)
 
-#define INVLIST_LEN_OFFSET 0	/* Number of elements in the inversion list */
-
-/* This is a combination of a version and data structure type, so that one
- * being passed in can be validated to be an inversion list of the correct
- * vintage.  When the structure of the header is changed, a new random number
- * in the range 2**31-1 should be generated.  Then, if an auxiliary program
- * doesn't change correspondingly, it will be discovered immediately */
-#define INVLIST_VERSION_ID_OFFSET 1
-#define INVLIST_VERSION_ID 1511554547
-
-#define INVLIST_OFFSET_OFFSET 2	/* 0 or 1 */
-/* The UV at this position contains either 0 or 1.  If 0, the inversion list
- * contains the code point U+00000, and begins at element [0] in the array,
- * which always contains 0.  If 1, the inversion list doesn't contain U+0000,
- * and it begins at element [1].  Inverting an inversion list consists of
- * adding or removing the 0 at the beginning of it.  By reserving a space for
- * that 0, inversion can be made very fast: we just flip this UV */
-
-/* For safety, when adding new elements, remember to #undef them at the end of
- * the inversion list code section */
-
-#define HEADER_LENGTH (INVLIST_OFFSET_OFFSET + 1) /* includes 1 for the constant
-                                                   0 element */
-
 /* An element is in an inversion list iff its index is even numbered: 0, 2, 4,
  * etc */
 #define ELEMENT_RANGE_MATCHES_INVLIST(i) (! ((i) & 1))
 #define PREV_RANGE_MATCHES_INVLIST(i) (! ELEMENT_RANGE_MATCHES_INVLIST(i))
 
-PERL_STATIC_INLINE STRLEN*
-S__get_invlist_len_addr(pTHX_ SV* invlist)
+/* This converts to/from our UVs to what the SV code is expecting: bytes. */
+#define TO_INTERNAL_SIZE(x) ((x) * sizeof(UV))
+#define FROM_INTERNAL_SIZE(x) ((x)/ sizeof(UV))
+
+PERL_STATIC_INLINE bool*
+S_get_invlist_offset_addr(SV* invlist)
 {
-    /* Return the address of the UV that contains the current number
-     * of used elements in the inversion list */
+    /* Return the address of the field that says whether the inversion list is
+     * offset (it contains 1) or not (contains 0) */
+    PERL_ARGS_ASSERT_GET_INVLIST_OFFSET_ADDR;
 
-    PERL_ARGS_ASSERT__GET_INVLIST_LEN_ADDR;
+    assert(SvTYPE(invlist) == SVt_INVLIST);
 
-    return &(LvTARGLEN(invlist));
+    return &(((XINVLIST*) SvANY(invlist))->is_offset);
 }
 
 PERL_STATIC_INLINE UV
-S__invlist_len(pTHX_ SV* const invlist)
+S__invlist_len(SV* const invlist)
 {
     /* Returns the current number of elements stored in the inversion list's
      * array */
 
     PERL_ARGS_ASSERT__INVLIST_LEN;
 
-    return *_get_invlist_len_addr(invlist);
+    assert(SvTYPE(invlist) == SVt_INVLIST);
+
+    return (SvCUR(invlist) == 0)
+           ? 0
+           : FROM_INTERNAL_SIZE(SvCUR(invlist)) - *get_invlist_offset_addr(invlist);
 }
 
 PERL_STATIC_INLINE bool
-S__invlist_contains_cp(pTHX_ SV* const invlist, const UV cp)
+S__invlist_contains_cp(SV* const invlist, const UV cp)
 {
     /* Does <invlist> contain code point <cp> as part of the set? */
 
@@ -71,4 +56,32 @@ S__invlist_contains_cp(pTHX_ SV* const invlist, const UV cp)
     return index >= 0 && ELEMENT_RANGE_MATCHES_INVLIST(index);
 }
 
+PERL_STATIC_INLINE UV*
+S_invlist_array(SV* const invlist)
+{
+    /* Returns the pointer to the inversion list's array.  Every time the
+     * length changes, this needs to be called in case malloc or realloc moved
+     * it */
+
+    PERL_ARGS_ASSERT_INVLIST_ARRAY;
+
+    /* Must not be empty.  If these fail, you probably didn't check for <len>
+     * being non-zero before trying to get the array */
+    assert(_invlist_len(invlist));
+
+    /* The very first element always contains zero, The array begins either
+     * there, or if the inversion list is offset, at the element after it.
+     * The offset header field determines which; it contains 0 or 1 to indicate
+     * how much additionally to add */
+    assert(0 == *(SvPVX(invlist)));
+    return ((UV *) SvPVX(invlist) + *get_invlist_offset_addr(invlist));
+}
+
+#   if defined(PERL_IN_UTF8_C) || defined(PERL_IN_REGEXEC_C)
+
+/* These symbols are only needed later in regcomp.c */
+#       undef TO_INTERNAL_SIZE
+#       undef FROM_INTERNAL_SIZE
+#   endif
+
 #endif