3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, 2005 by Larry Wall and others
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
11 typedef OP OP_4tree; /* Will be redefined later. */
14 #define PERL_ENABLE_TRIE_OPTIMISATION 1
15 #define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1
16 #define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 1
17 #define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0
20 * The "internal use only" fields in regexp.h are present to pass info from
21 * compile to execute that permits the execute phase to run lots faster on
22 * simple cases. They are:
24 * regstart sv that must begin a match; NULL if none obvious
25 * reganch is the match anchored (at beginning-of-line only)?
26 * regmust string (pointer into program) that match must include, or NULL
27 * [regmust changed to SV* for bminstr()--law]
28 * regmlen length of regmust string
29 * [regmlen not used currently]
31 * Regstart and reganch permit very fast decisions on suitable starting points
32 * for a match, cutting down the work a lot. Regmust permits fast rejection
33 * of lines that cannot possibly match. The regmust tests are costly enough
34 * that pregcomp() supplies a regmust only if the r.e. contains something
35 * potentially expensive (at present, the only such thing detected is * or +
36 * at the start of the r.e., which can involve a lot of backup). Regmlen is
37 * supplied because the test in pregexec() needs it and pregcomp() is computing
39 * [regmust is now supplied always. The tests that use regmust have a
40 * heuristic that disables the test if it usually matches.]
42 * [In fact, we now use regmust in many cases to locate where the search
43 * starts in the string, so if regback is >= 0, the regmust search is never
44 * wasted effort. The regback variable says how many characters back from
45 * where regmust matched is the earliest possible start of the match.
46 * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.]
50 * Structure for regexp "program". This is essentially a linear encoding
51 * of a nondeterministic finite-state machine (aka syntax charts or
52 * "railroad normal form" in parsing technology). Each node is an opcode
53 * plus a "next" pointer, possibly plus an operand. "Next" pointers of
54 * all nodes except BRANCH implement concatenation; a "next" pointer with
55 * a BRANCH on both ends of it is connecting two alternatives. (Here we
56 * have one of the subtle syntax dependencies: an individual BRANCH (as
57 * opposed to a collection of them) is never concatenated with anything
58 * because of operator precedence.) The operand of some types of node is
59 * a literal string; for others, it is a node leading into a sub-FSM. In
60 * particular, the operand of a BRANCH node is the first node of the branch.
61 * (NB this is *not* a tree structure: the tail of the branch connects
62 * to the thing following the set of BRANCHes.) The opcodes are defined
63 * in regnodes.h which is generated from regcomp.sym by regcomp.pl.
67 * A node is one char of opcode followed by two chars of "next" pointer.
68 * "Next" pointers are stored as two 8-bit pieces, high order first. The
69 * value is a positive offset from the opcode of the node containing it.
70 * An operand, if any, simply follows the node. (Note that much of the
71 * code generation knows about this implicit relationship.)
73 * Using two bytes for the "next" pointer is vast overkill for most things,
74 * but allows patterns to get big without disasters.
76 * [The "next" pointer is always aligned on an even
77 * boundary, and reads the offset directly as a short. Also, there is no
78 * special test to reverse the sign of BACK pointers since the offset is
82 struct regnode_string {
104 #define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */
105 #define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */
107 /* also used by trie */
108 struct regnode_charclass {
113 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
116 struct regnode_charclass_class { /* has [[:blah:]] classes */
117 U8 flags; /* should have ANYOF_CLASS here */
121 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
122 char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */
125 /* XXX fix this description.
126 Impose a limit of REG_INFTY on various pattern matching operations
127 to limit stack growth and to avoid "infinite" recursions.
129 /* The default size for REG_INFTY is I16_MAX, which is the same as
130 SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits
131 (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is
132 ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is
133 ((1<<63)-1). To limit stack growth to reasonable sizes, supply a
135 --Andy Dougherty 11 June 1998
139 # define REG_INFTY ((1<<15)-1)
144 # define REG_INFTY I16_MAX
147 #define ARG_VALUE(arg) (arg)
148 #define ARG__SET(arg,val) ((arg) = (val))
154 #define ARG(p) ARG_VALUE(ARG_LOC(p))
155 #define ARG1(p) ARG_VALUE(ARG1_LOC(p))
156 #define ARG2(p) ARG_VALUE(ARG2_LOC(p))
158 #define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
159 #define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
160 #define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
165 #define NEXT_OFF(p) ((p)->next_off)
166 #define NODE_ALIGN(node)
167 #define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */
169 #define SIZE_ALIGN NODE_ALIGN
176 #define OP(p) ((p)->type)
177 #define OPERAND(p) (((struct regnode_string *)p)->string)
178 #define MASK(p) ((char*)OPERAND(p))
179 #define STR_LEN(p) (((struct regnode_string *)p)->str_len)
180 #define STRING(p) (((struct regnode_string *)p)->string)
181 #define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
182 #define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
189 #define NODE_ALIGN(node)
190 #define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
191 #define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
192 #define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
195 #define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
196 #define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
198 #define NODE_STEP_B 4
200 #define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
201 #define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
203 #define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
204 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END
205 #define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
206 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END
208 #define REG_MAGIC 0234
210 #define SIZE_ONLY (RExC_emit == &PL_regdummy)
212 /* Flags for node->flags of ANYOF */
214 #define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */
215 #define ANYOF_INVERT 0x04
216 #define ANYOF_FOLD 0x02
217 #define ANYOF_LOCALE 0x01
219 /* Used for regstclass only */
220 #define ANYOF_EOS 0x10 /* Can match an empty string too */
222 /* There is a character or a range past 0xff */
223 #define ANYOF_UNICODE 0x20
224 #define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */
226 /* size of node is large (includes class pointer) */
227 #define ANYOF_LARGE 0x80
229 /* Are there any runtime flags on in this node? */
230 #define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f)
232 #define ANYOF_FLAGS_ALL 0xff
234 /* Character classes for node->classflags of ANYOF */
235 /* Should be synchronized with a table in regprop() */
236 /* 2n should pair with 2n+1 */
238 #define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
239 #define ANYOF_NALNUM 1
240 #define ANYOF_SPACE 2 /* \s */
241 #define ANYOF_NSPACE 3
242 #define ANYOF_DIGIT 4
243 #define ANYOF_NDIGIT 5
244 #define ANYOF_ALNUMC 6 /* isalnum(3), utf8::IsAlnum, ALNUMC */
245 #define ANYOF_NALNUMC 7
246 #define ANYOF_ALPHA 8
247 #define ANYOF_NALPHA 9
248 #define ANYOF_ASCII 10
249 #define ANYOF_NASCII 11
250 #define ANYOF_CNTRL 12
251 #define ANYOF_NCNTRL 13
252 #define ANYOF_GRAPH 14
253 #define ANYOF_NGRAPH 15
254 #define ANYOF_LOWER 16
255 #define ANYOF_NLOWER 17
256 #define ANYOF_PRINT 18
257 #define ANYOF_NPRINT 19
258 #define ANYOF_PUNCT 20
259 #define ANYOF_NPUNCT 21
260 #define ANYOF_UPPER 22
261 #define ANYOF_NUPPER 23
262 #define ANYOF_XDIGIT 24
263 #define ANYOF_NXDIGIT 25
264 #define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */
265 #define ANYOF_NPSXSPC 27
266 #define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */
267 #define ANYOF_NBLANK 29
271 /* Backward source code compatibility. */
273 #define ANYOF_ALNUML ANYOF_ALNUM
274 #define ANYOF_NALNUML ANYOF_NALNUM
275 #define ANYOF_SPACEL ANYOF_SPACE
276 #define ANYOF_NSPACEL ANYOF_NSPACE
278 /* Utility macros for the bitmap and classes of ANYOF */
280 #define ANYOF_SIZE (sizeof(struct regnode_charclass))
281 #define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class))
283 #define ANYOF_FLAGS(p) ((p)->flags)
285 #define ANYOF_BIT(c) (1 << ((c) & 7))
287 #define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3])
288 #define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c))
289 #define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c))
290 #define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c))
292 #define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char)
293 #define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
295 #define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
296 #define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
297 #define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
298 #define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
299 #define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
301 #define ANYOF_BITMAP_SETALL(p) \
302 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
303 #define ANYOF_BITMAP_CLEARALL(p) \
304 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
305 /* Check that all 256 bits are all set. Used in S_cl_is_anything() */
306 #define ANYOF_BITMAP_TESTALLSET(p) \
307 memEQ (ANYOF_BITMAP(p), "\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377", ANYOF_BITMAP_SIZE)
309 #define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode))
310 #define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode))
311 #define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP)
315 * Utility definitions.
318 # define UCHARAT(p) ((int)*(const U8*)(p))
320 # define UCHARAT(p) ((int)*(p)&CHARMASK)
323 #define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
325 #define REG_SEEN_ZERO_LEN 0x00000001
326 #define REG_SEEN_LOOKBEHIND 0x00000002
327 #define REG_SEEN_GPOS 0x00000004
328 #define REG_SEEN_EVAL 0x00000008
329 #define REG_SEEN_CANY 0x00000010
330 #define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */
334 #ifdef PLUGGABLE_RE_EXTENSION
335 #include "re_nodes.h"
337 #include "regnodes.h"
340 /* The following have no fixed length. U8 so we can do strchr() on it. */
342 EXTCONST U8 PL_varies[];
344 EXTCONST U8 PL_varies[] = {
345 BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL,
346 WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, CLUMP, 0
350 /* The following always have a length of 1. U8 we can do strchr() on it. */
351 /* (Note that length 1 means "one character" under UTF8, not "one octet".) */
353 EXTCONST U8 PL_simple[];
355 EXTCONST U8 PL_simple[] = {
367 #ifndef PLUGGABLE_RE_EXTENSION
369 EXTCONST regexp_engine PL_core_reg_engine;
371 EXTCONST regexp_engine PL_core_reg_engine = {
374 Perl_re_intuit_start,
375 Perl_re_intuit_string,
377 #if defined(USE_ITHREADS)
382 #endif /* PLUGGABLE_RE_EXTENSION */
388 /* .what is a character array with one character for each member of .data
389 * The character describes the function of the corresponding .data item:
390 * f - start-class data for regstclass optimization
391 * n - Root of op tree for (?{EVAL}) item
392 * o - Start op for (?{EVAL}) item
393 * p - Pad for (?{EVAL}) item
394 * s - swash for unicode-style character class, and the multicharacter
395 * strings resulting from casefolding the single-character entries
396 * in the character class
398 * T - aho-trie struct
399 * 20010712 mjd@plover.com
400 * (Remember to update re_dup() and pregfree() if you add any items.)
408 struct reg_substr_datum {
411 SV *substr; /* non-utf8 variant */
412 SV *utf8_substr; /* utf8 variant */
416 struct reg_substr_data {
417 struct reg_substr_datum data[3]; /* Actual array */
420 #define anchored_substr substrs->data[0].substr
421 #define anchored_utf8 substrs->data[0].utf8_substr
422 #define anchored_offset substrs->data[0].min_offset
423 #define anchored_end_shift substrs->data[0].end_shift
425 #define float_substr substrs->data[1].substr
426 #define float_utf8 substrs->data[1].utf8_substr
427 #define float_min_offset substrs->data[1].min_offset
428 #define float_max_offset substrs->data[1].max_offset
429 #define float_end_shift substrs->data[1].end_shift
431 #define check_substr substrs->data[2].substr
432 #define check_utf8 substrs->data[2].utf8_substr
433 #define check_offset_min substrs->data[2].min_offset
434 #define check_offset_max substrs->data[2].max_offset
435 #define check_end_shift substrs->data[2].end_shift
439 /* trie related stuff */
441 /* a transition record for the state machine. the
442 check field determines which state "owns" the
443 transition. the char the transition is for is
444 determined by offset from the owning states base
445 field. the next field determines which state
446 is to be transitioned to if any.
448 struct _reg_trie_trans {
453 /* a transition list element for the list based representation */
454 struct _reg_trie_trans_list_elem {
458 typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
460 /* a state for compressed nodes. base is an offset
461 into an array of reg_trie_trans array. If wordnum is
462 nonzero the state is accepting. if base is zero then
463 the state has no children (and will be accepting)
465 struct _reg_trie_state {
469 reg_trie_trans_le* list;
475 typedef struct _reg_trie_state reg_trie_state;
476 typedef struct _reg_trie_trans reg_trie_trans;
479 /* anything in here that needs to be freed later
480 should be dealt with in pregfree */
481 struct _reg_trie_data {
482 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
483 U32 lasttrans; /* last valid transition element */
484 U16 *charmap; /* byte to charid lookup array */
485 HV *widecharmap; /* code points > 255 to charid */
486 reg_trie_state *states; /* state data */
487 reg_trie_trans *trans; /* array of transition elements */
488 char *bitmap; /* stclass bitmap */
489 U32 refcount; /* number of times this trie is referenced */
490 U32 startstate; /* initial state - used for common prefix optimisation */
491 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
492 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
493 U32 *wordlen; /* array of lengths of words */
494 U16 *jump; /* optional 1 indexed array of offsets before tail
495 for the node following a given word. */
496 U16 *nextword; /* optional 1 indexed array to support linked list
497 of duplicate wordnums */
498 U32 laststate; /* Build only */
499 U32 wordcount; /* Build only */
501 STRLEN charcount; /* Build only */
502 AV *words; /* Array of words contained in trie, for dumping */
503 AV *revcharmap; /* Map of each charid back to its character representation */
506 typedef struct _reg_trie_data reg_trie_data;
508 struct _reg_ac_data {
510 reg_trie_state *states;
514 typedef struct _reg_ac_data reg_ac_data;
516 /* ANY_BIT doesnt use the structure, so we can borrow it here.
517 This is simpler than refactoring all of it as wed end up with
518 three different sets... */
520 #define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
521 #define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
522 #define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
523 #define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
524 #define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
526 #define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
527 #define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
530 #define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
531 #define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
533 /* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
534 #define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
535 #define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
536 #define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
539 #define TRIE_CHARCOUNT(trie) ((trie)->charcount)
540 #define TRIE_REVCHARMAP(trie) ((trie)->revcharmap)
542 #define TRIE_CHARCOUNT(trie) (trie_charcount)
543 #define TRIE_REVCHARMAP(trie) (trie_revcharmap)
546 #define RE_TRIE_MAXBUF_INIT 65536
547 #define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
548 #define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
552 RE_DEBUG_FLAGS is used to control what debug output is emitted
553 its divided into three groups of options, some of which interact.
554 The three groups are: Compile, Execute, Extra. There is room for a
555 further group, as currently only the low three bytes are used.
576 If you modify any of these make sure you make corresponding changes to
577 re.pm, especially to the documentation.
583 #define RE_DEBUG_COMPILE_MASK 0x0000FF
584 #define RE_DEBUG_COMPILE_PARSE 0x000001
585 #define RE_DEBUG_COMPILE_OPTIMISE 0x000002
586 #define RE_DEBUG_COMPILE_TRIE 0x000004
587 #define RE_DEBUG_COMPILE_DUMP 0x000008
590 #define RE_DEBUG_EXECUTE_MASK 0x00FF00
591 #define RE_DEBUG_EXECUTE_INTUIT 0x000100
592 #define RE_DEBUG_EXECUTE_MATCH 0x000200
593 #define RE_DEBUG_EXECUTE_TRIE 0x000400
596 #define RE_DEBUG_EXTRA_MASK 0xFF0000
597 #define RE_DEBUG_EXTRA_TRIE 0x010000
598 #define RE_DEBUG_EXTRA_OFFSETS 0x020000
599 #define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
600 #define RE_DEBUG_EXTRA_STATE 0x080000
601 #define RE_DEBUG_EXTRA_OPTIMISE 0x100000
603 #define RE_DEBUG_FLAG(x) (re_debug_flags & x)
605 #define DEBUG_COMPILE_r(x) DEBUG_r( \
606 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
607 #define DEBUG_PARSE_r(x) DEBUG_r( \
608 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
609 #define DEBUG_OPTIMISE_r(x) DEBUG_r( \
610 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
611 #define DEBUG_PARSE_r(x) DEBUG_r( \
612 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
613 #define DEBUG_DUMP_r(x) DEBUG_r( \
614 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
615 #define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
616 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
619 #define DEBUG_EXECUTE_r(x) DEBUG_r( \
620 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
621 #define DEBUG_INTUIT_r(x) DEBUG_r( \
622 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
623 #define DEBUG_MATCH_r(x) DEBUG_r( \
624 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
625 #define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
626 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
629 #define DEBUG_EXTRA_r(x) DEBUG_r( \
630 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
631 #define DEBUG_OFFSETS_r(x) DEBUG_r( \
632 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
633 #define DEBUG_STATE_r(x) DEBUG_r( \
634 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
635 #define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
636 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
637 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
638 #define MJD_OFFSET_DEBUG(x) DEBUG_r( \
639 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
640 Perl_warn_nocontext x )
641 #define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
642 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
643 #define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
644 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
646 #define DEBUG_TRIE_r(x) DEBUG_r( \
647 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
648 | RE_DEBUG_EXECUTE_TRIE )) x )
651 /* get_sv() can return NULL during global destruction. */
652 #define GET_RE_DEBUG_FLAGS DEBUG_r({ \
653 SV * re_debug_flags_sv = NULL; \
654 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
655 if (re_debug_flags_sv) { \
656 if (!SvIOK(re_debug_flags_sv)) \
657 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
658 re_debug_flags=SvIV(re_debug_flags_sv); \
664 #define GET_RE_DEBUG_FLAGS_DECL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS;
666 #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
667 const char * const rpv = \
668 pv_pretty((dsv), (pv), (l), (m), \
669 PL_colors[(c1)],PL_colors[(c2)], \
670 ((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
671 const int rlen = SvCUR(dsv)
673 #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
674 const char * const rpv = \
675 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
676 PL_colors[(c1)],PL_colors[(c2)], \
677 ((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
679 #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
680 const char * const rpv = \
681 pv_pretty((dsv), (pv), (l), (m), \
682 PL_colors[0], PL_colors[1], \
683 ( PERL_PV_PRETTY_QUOTE | PERL_PV_PRETTY_ELIPSES | \
684 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
687 #define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
688 #define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
690 #else /* if not DEBUGGING */
692 #define GET_RE_DEBUG_FLAGS_DECL
693 #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
694 #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
695 #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
696 #define RE_SV_DUMPLEN(ItEm)
697 #define RE_SV_TAIL(ItEm)
699 #endif /* DEBUG RELATED DEFINES */