};
#define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */
-#define ANYOF_CLASSBITMAP_SIZE 4
+#define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */
struct regnode_charclass {
U8 flags;
U8 type;
U16 next_off;
- char bitmap[ANYOF_BITMAP_SIZE];
+ U32 arg1;
+ char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
};
-struct regnode_charclass_class {
- U8 flags;
+struct regnode_charclass_class { /* has [[:blah:]] classes */
+ U8 flags; /* should have ANYOF_CLASS here */
U8 type;
U16 next_off;
- char bitmap[ANYOF_BITMAP_SIZE];
- char classflags[ANYOF_CLASSBITMAP_SIZE];
+ U32 arg1;
+ char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
+ char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */
};
/* XXX fix this description.
#define ARG_VALUE(arg) (arg)
#define ARG__SET(arg,val) ((arg) = (val))
+#undef ARG
+#undef ARG1
+#undef ARG2
+
#define ARG(p) ARG_VALUE(ARG_LOC(p))
#define ARG1(p) ARG_VALUE(ARG1_LOC(p))
#define ARG2(p) ARG_VALUE(ARG2_LOC(p))
#define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
#define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
+#undef NEXT_OFF
+#undef NODE_ALIGN
+
#ifndef lint
# define NEXT_OFF(p) ((p)->next_off)
# define NODE_ALIGN(node)
#define SIZE_ALIGN NODE_ALIGN
+#undef OP
+#undef OPERAND
+#undef MASK
+#undef STRING
+
#define OP(p) ((p)->type)
#define OPERAND(p) (((struct regnode_string *)p)->string)
#define MASK(p) ((char*)OPERAND(p))
#define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
#define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
+#undef NODE_ALIGN
+#undef ARG_LOC
+#undef NEXTOPER
+#undef PREVOPER
+
#define NODE_ALIGN(node)
#define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
#define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
#define REG_MAGIC 0234
-#define SIZE_ONLY (PL_regcode == &PL_regdummy)
+#define SIZE_ONLY (RExC_emit == &PL_regdummy)
/* Flags for node->flags of ANYOF */
-#define ANYOF_CLASS 0x08
-#define ANYOF_INVERT 0x04
-#define ANYOF_FOLD 0x02
-#define ANYOF_LOCALE 0x01
+#define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */
+#define ANYOF_INVERT 0x04
+#define ANYOF_FOLD 0x02
+#define ANYOF_LOCALE 0x01
/* Used for regstclass only */
-#define ANYOF_EOS 0x10 /* Can match an empty string too */
+#define ANYOF_EOS 0x10 /* Can match an empty string too */
+
+/* There is a character or a range past 0xff */
+#define ANYOF_UNICODE 0x20
+#define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */
+
+/* size of node is large (includes class pointer) */
+#define ANYOF_LARGE 0x80
+
+/* Are there any runtime flags on in this node? */
+#define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f)
+
+#define ANYOF_FLAGS_ALL 0xff
/* Character classes for node->classflags of ANYOF */
/* Should be synchronized with a table in regprop() */
#define ANYOF_NXDIGIT 25
#define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */
#define ANYOF_NPSXSPC 27
-#define ANYOF_BLANK 28 /* GNU extension: space and tab */
+#define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */
#define ANYOF_NBLANK 29
#define ANYOF_MAX 32
#define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class))
#define ANYOF_FLAGS(p) ((p)->flags)
-#define ANYOF_FLAGS_ALL 0xff
#define ANYOF_BIT(c) (1 << ((c) & 7))
#define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
-#define REG_SEEN_ZERO_LEN 1
-#define REG_SEEN_LOOKBEHIND 2
-#define REG_SEEN_GPOS 4
-#define REG_SEEN_EVAL 8
+#define REG_SEEN_ZERO_LEN 1
+#define REG_SEEN_LOOKBEHIND 2
+#define REG_SEEN_GPOS 4
+#define REG_SEEN_EVAL 8
+#define REG_SEEN_SANY 16
START_EXTERN_C
EXTCONST U8 PL_simple[];
#else
EXTCONST U8 PL_simple[] = {
- REG_ANY, ANYUTF8, SANY, SANYUTF8, ANYOF, ANYOFUTF8,
- ALNUM, ALNUMUTF8, ALNUML, ALNUMLUTF8,
- NALNUM, NALNUMUTF8, NALNUML, NALNUMLUTF8,
- SPACE, SPACEUTF8, SPACEL, SPACELUTF8,
- NSPACE, NSPACEUTF8, NSPACEL, NSPACELUTF8,
- DIGIT, DIGITUTF8, NDIGIT, NDIGITUTF8, 0
+ REG_ANY, SANY,
+ ANYOF,
+ ALNUM, ALNUML,
+ NALNUM, NALNUML,
+ SPACE, SPACEL,
+ NSPACE, NSPACEL,
+ DIGIT, NDIGIT,
+ 0
};
#endif
I32 *scream_pos; /* Internal iterator of scream. */
} re_scream_pos_data;
+/* .what is a character array with one character for each member of .data
+ * The character describes the function of the corresponding .data item:
+ * f - start-class data for regstclass optimization
+ * n - Root of op tree for (?{EVAL}) item
+ * o - Start op for (?{EVAL}) item
+ * p - Pad for (?{EVAL} item
+ * s - swash for unicode-style character class
+ * 20010712 mjd@plover.com
+ * (Remember to update re_dup() and pregfree() if you add any items.)
+ */
struct reg_data {
U32 count;
U8 *what;