Try and fix my XCLASS misunderstandings

PCRE2Project · Nov 29, 2024 · 7056cdf · 7056cdf
1 parent 81b1169
commit 7056cdf
Show file tree

Hide file tree

Showing 4 changed files with 12 additions and 16 deletions.
diff --git a/src/pcre2_compile.c b/src/pcre2_compile.c
@@ -6466,7 +6466,7 @@ for (;; pptr++)
     /* Now emit the OP_CLASS/OP_NCLASS/OP_XCLASS/OP_ALLANY opcode. */
 
     pptr = PRIV(compile_class_not_nested)(options, xoptions, pptr + 1,
-                                          &code, meta == META_CLASS_NOT,
+                                          &code, meta == META_CLASS_NOT, FALSE,
                                           errorcodeptr, cb, lengthptr);
     if (pptr == NULL) return 0;
     PCRE2_ASSERT(*pptr == META_CLASS_END);

diff --git a/src/pcre2_compile.h b/src/pcre2_compile.h
@@ -263,7 +263,7 @@ void PRIV(update_classbits)(uint32_t ptype, uint32_t pdata, BOOL negated,
 OP_CLASS, OP_NCLASS, OP_XCLASS, or OP_ALLANY into pcode. */
 
 uint32_t *PRIV(compile_class_not_nested)(uint32_t options, uint32_t xoptions,
-  uint32_t *start_ptr, PCRE2_UCHAR **pcode, BOOL negate_class,
+  uint32_t *start_ptr, PCRE2_UCHAR **pcode, BOOL negate_class, BOOL always_map,
   int *errorcodeptr, compile_block *cb, PCRE2_SIZE *lengthptr);
 
 /* Compile the META codes in pptr into opcodes written to pcode. The pptr must

diff --git a/src/pcre2_compile_class.c b/src/pcre2_compile_class.c
@@ -1007,7 +1007,7 @@ become a single OP_CLASS OP_NCLASS, OP_XCLASS, or OP_ALLANY. */
 
 uint32_t *
 PRIV(compile_class_not_nested)(uint32_t options, uint32_t xoptions,
-  uint32_t *start_ptr, PCRE2_UCHAR **pcode, BOOL negate_class,
+  uint32_t *start_ptr, PCRE2_UCHAR **pcode, BOOL negate_class, BOOL always_map,
   int *errorcodeptr, compile_block *cb, PCRE2_SIZE *lengthptr)
 {
 uint32_t *pptr = start_ptr;
@@ -1637,7 +1637,8 @@ if ((xclass_props & XCLASS_REQUIRED) != 0)
   /* If the map is required, move up the extra data to make room for it;
   otherwise just move the code pointer to the end of the extra data. */
 
-  if ((xclass_props & XCLASS_HAS_8BIT_CHARS) != 0)
+  if ((xclass_props & XCLASS_HAS_8BIT_CHARS) != 0 ||
+      always_map)
     {
     *code++ |= XCL_MAP;
     (void)memmove(code + (32 / sizeof(PCRE2_UCHAR)), code,
@@ -2079,7 +2080,7 @@ switch (meta)
   prev_ptr = ptr;
   ptr = PRIV(compile_class_not_nested)(options, xoptions, ptr, &code,
                                        (meta != META_CLASS_NOT) == negated,
-                                       errorcodeptr, cb, lengthptr);
+                                       TRUE, errorcodeptr, cb, lengthptr);
   if (ptr == NULL) return FALSE;
 
   /* We must have a 100% guarantee that ptr increases when

diff --git a/src/pcre2_xclass.c b/src/pcre2_xclass.c
@@ -79,20 +79,15 @@ const uint8_t *next_char;
 utf = TRUE;
 #endif
 
-/* Code points < 256 are matched against a bitmap, if one is present. If no
-bitmap is present, then the XCLASS does not match any code points < 256. */
-
-if (c < 256)
-  {
-  if ((*data & XCL_MAP) != 0)
-    return (((const uint8_t *)data+1)[c/8] & (1u << (c&7))) != 0;
-  return FALSE;
-  }
-
-/* Skip the bitmap. */
+/* Code points < 256 are matched against a bitmap, if one is present. */
 
 if ((*data++ & XCL_MAP) != 0)
+  {
+  if (c < 256)
+    return (((const uint8_t *)data)[c/8] & (1u << (c&7))) != 0;
+  /* Skip bitmap. */
   data += 32 / sizeof(PCRE2_UCHAR);
+  }
 
 /* Match against the list of Unicode properties. We won't ever
 encounter XCL_PROP or XCL_NOTPROP when UTF support is not compiled. */