3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, 2004, 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.
12 * He still hopefully carried some of his gear in his pack: a small tinder-box,
13 * two small shallow pans, the smaller fitting into the larger; inside them a
14 * wooden spoon, a short two-pronged fork and some skewers were stowed; and
15 * hidden at the bottom of the pack in a flat wooden box a dwindling treasure,
19 /* This file contains pp ("push/pop") functions that
20 * execute the opcodes that make up a perl program. A typical pp function
21 * expects to find its arguments on the stack, and usually pushes its
22 * results onto the stack, hence the 'pp' terminology. Each OP structure
23 * contains a pointer to the relevant pp_foo() function.
25 * This particular file just contains pp_pack() and pp_unpack(). See the
26 * other pp*.c files for the rest of the pp_ functions.
31 #define PERL_IN_PP_PACK_C
34 /* Types used by pack/unpack */
36 e_no_len, /* no length */
37 e_number, /* number, [] */
41 typedef struct tempsym {
42 const char* patptr; /* current template char */
43 const char* patend; /* one after last char */
44 const char* grpbeg; /* 1st char of ()-group */
45 const char* grpend; /* end of ()-group */
46 I32 code; /* template code (!<>) */
47 I32 length; /* length/repeat count */
48 howlen_t howlen; /* how length is given */
49 int level; /* () nesting level */
50 U32 flags; /* /=4, comma=2, pack=1 */
51 /* and group modifiers */
52 STRLEN strbeg; /* offset of group start */
53 struct tempsym *previous; /* previous group */
56 #define TEMPSYM_INIT(symptr, p, e, f) \
58 (symptr)->patptr = (p); \
59 (symptr)->patend = (e); \
60 (symptr)->grpbeg = NULL; \
61 (symptr)->grpend = NULL; \
62 (symptr)->grpend = NULL; \
64 (symptr)->length = 0; \
65 (symptr)->howlen = 0; \
66 (symptr)->level = 0; \
67 (symptr)->flags = (f); \
68 (symptr)->strbeg = 0; \
69 (symptr)->previous = NULL; \
73 # define PERL_PACK_CAN_BYTEORDER
74 # define PERL_PACK_CAN_SHRIEKSIGN
80 /* Maximum number of bytes to which a byte can grow due to upgrade */
84 * Offset for integer pack/unpack.
86 * On architectures where I16 and I32 aren't really 16 and 32 bits,
87 * which for now are all Crays, pack and unpack have to play games.
91 * These values are required for portability of pack() output.
92 * If they're not right on your machine, then pack() and unpack()
93 * wouldn't work right anyway; you'll need to apply the Cray hack.
94 * (I'd like to check them with #if, but you can't use sizeof() in
95 * the preprocessor.) --???
98 The appropriate SHORTSIZE, INTSIZE, LONGSIZE, and LONGLONGSIZE
99 defines are now in config.h. --Andy Dougherty April 1998
104 /* CROSSCOMPILE and MULTIARCH are going to affect pp_pack() and pp_unpack().
107 #if U16SIZE > SIZE16 || U32SIZE > SIZE32
108 # if BYTEORDER == 0x1234 || BYTEORDER == 0x12345678 /* little-endian */
109 # define OFF16(p) ((char*)(p))
110 # define OFF32(p) ((char*)(p))
112 # if BYTEORDER == 0x4321 || BYTEORDER == 0x87654321 /* big-endian */
113 # define OFF16(p) ((char*)(p) + (sizeof(U16) - SIZE16))
114 # define OFF32(p) ((char*)(p) + (sizeof(U32) - SIZE32))
116 ++++ bad cray byte order
120 # define OFF16(p) ((char *) (p))
121 # define OFF32(p) ((char *) (p))
124 /* Only to be used inside a loop (see the break) */
125 #define SHIFT16(utf8, s, strend, p, datumtype) STMT_START { \
127 if (!uni_to_bytes(aTHX_ &(s), strend, OFF16(p), SIZE16, datumtype)) break; \
129 Copy(s, OFF16(p), SIZE16, char); \
134 /* Only to be used inside a loop (see the break) */
135 #define SHIFT32(utf8, s, strend, p, datumtype) STMT_START { \
137 if (!uni_to_bytes(aTHX_ &(s), strend, OFF32(p), SIZE32, datumtype)) break; \
139 Copy(s, OFF32(p), SIZE32, char); \
144 #define PUSH16(utf8, cur, p) PUSH_BYTES(utf8, cur, OFF16(p), SIZE16)
145 #define PUSH32(utf8, cur, p) PUSH_BYTES(utf8, cur, OFF32(p), SIZE32)
147 /* Only to be used inside a loop (see the break) */
148 #define SHIFT_VAR(utf8, s, strend, var, datumtype) \
151 if (!uni_to_bytes(aTHX_ &s, strend, \
152 (char *) &var, sizeof(var), datumtype)) break;\
154 Copy(s, (char *) &var, sizeof(var), char); \
159 #define PUSH_VAR(utf8, aptr, var) \
160 PUSH_BYTES(utf8, aptr, &(var), sizeof(var))
162 /* Avoid stack overflow due to pathological templates. 100 should be plenty. */
163 #define MAX_SUB_TEMPLATE_LEVEL 100
165 /* flags (note that type modifiers can also be used as flags!) */
166 #define FLAG_WAS_UTF8 0x40
167 #define FLAG_PARSE_UTF8 0x20 /* Parse as utf8 */
168 #define FLAG_UNPACK_ONLY_ONE 0x10
169 #define FLAG_DO_UTF8 0x08 /* The underlying string is utf8 */
170 #define FLAG_SLASH 0x04
171 #define FLAG_COMMA 0x02
172 #define FLAG_PACK 0x01
175 S_mul128(pTHX_ SV *sv, U8 m)
178 char *s = SvPV(sv, len);
181 if (!strnEQ(s, "0000", 4)) { /* need to grow sv */
182 SV *tmpNew = newSVpvn("0000000000", 10);
184 sv_catsv(tmpNew, sv);
185 SvREFCNT_dec(sv); /* free old sv */
190 while (!*t) /* trailing '\0'? */
193 const U32 i = ((*t - '0') << 7) + m;
194 *(t--) = '0' + (char)(i % 10);
200 /* Explosives and implosives. */
202 #if 'I' == 73 && 'J' == 74
203 /* On an ASCII/ISO kind of system */
204 #define ISUUCHAR(ch) ((ch) >= ' ' && (ch) < 'a')
207 Some other sort of character set - use memchr() so we don't match
210 #define ISUUCHAR(ch) (memchr(PL_uuemap, (ch), sizeof(PL_uuemap)-1) || (ch) == ' ')
214 #define TYPE_IS_SHRIEKING 0x100
215 #define TYPE_IS_BIG_ENDIAN 0x200
216 #define TYPE_IS_LITTLE_ENDIAN 0x400
217 #define TYPE_IS_PACK 0x800
218 #define TYPE_ENDIANNESS_MASK (TYPE_IS_BIG_ENDIAN|TYPE_IS_LITTLE_ENDIAN)
219 #define TYPE_MODIFIERS(t) ((t) & ~0xFF)
220 #define TYPE_NO_MODIFIERS(t) ((t) & 0xFF)
222 #ifdef PERL_PACK_CAN_SHRIEKSIGN
223 # define SHRIEKING_ALLOWED_TYPES "sSiIlLxXnNvV@."
225 # define SHRIEKING_ALLOWED_TYPES "sSiIlLxX"
228 #ifndef PERL_PACK_CAN_BYTEORDER
229 /* Put "can't" first because it is shorter */
230 # define TYPE_ENDIANNESS(t) 0
231 # define TYPE_NO_ENDIANNESS(t) (t)
233 # define ENDIANNESS_ALLOWED_TYPES ""
235 # define DO_BO_UNPACK(var, type)
236 # define DO_BO_PACK(var, type)
237 # define DO_BO_UNPACK_PTR(var, type, pre_cast, post_cast)
238 # define DO_BO_PACK_PTR(var, type, pre_cast, post_cast)
239 # define DO_BO_UNPACK_N(var, type)
240 # define DO_BO_PACK_N(var, type)
241 # define DO_BO_UNPACK_P(var)
242 # define DO_BO_PACK_P(var)
244 #else /* PERL_PACK_CAN_BYTEORDER */
246 # define TYPE_ENDIANNESS(t) ((t) & TYPE_ENDIANNESS_MASK)
247 # define TYPE_NO_ENDIANNESS(t) ((t) & ~TYPE_ENDIANNESS_MASK)
249 # define ENDIANNESS_ALLOWED_TYPES "sSiIlLqQjJfFdDpP("
251 # define DO_BO_UNPACK(var, type) \
253 switch (TYPE_ENDIANNESS(datumtype)) { \
254 case TYPE_IS_BIG_ENDIAN: var = my_betoh ## type (var); break; \
255 case TYPE_IS_LITTLE_ENDIAN: var = my_letoh ## type (var); break; \
260 # define DO_BO_PACK(var, type) \
262 switch (TYPE_ENDIANNESS(datumtype)) { \
263 case TYPE_IS_BIG_ENDIAN: var = my_htobe ## type (var); break; \
264 case TYPE_IS_LITTLE_ENDIAN: var = my_htole ## type (var); break; \
269 # define DO_BO_UNPACK_PTR(var, type, pre_cast, post_cast) \
271 switch (TYPE_ENDIANNESS(datumtype)) { \
272 case TYPE_IS_BIG_ENDIAN: \
273 var = (post_cast*) my_betoh ## type ((pre_cast) var); \
275 case TYPE_IS_LITTLE_ENDIAN: \
276 var = (post_cast *) my_letoh ## type ((pre_cast) var); \
283 # define DO_BO_PACK_PTR(var, type, pre_cast, post_cast) \
285 switch (TYPE_ENDIANNESS(datumtype)) { \
286 case TYPE_IS_BIG_ENDIAN: \
287 var = (post_cast *) my_htobe ## type ((pre_cast) var); \
289 case TYPE_IS_LITTLE_ENDIAN: \
290 var = (post_cast *) my_htole ## type ((pre_cast) var); \
297 # define BO_CANT_DOIT(action, type) \
299 switch (TYPE_ENDIANNESS(datumtype)) { \
300 case TYPE_IS_BIG_ENDIAN: \
301 Perl_croak(aTHX_ "Can't %s big-endian %ss on this " \
302 "platform", #action, #type); \
304 case TYPE_IS_LITTLE_ENDIAN: \
305 Perl_croak(aTHX_ "Can't %s little-endian %ss on this " \
306 "platform", #action, #type); \
313 # if PTRSIZE == INTSIZE
314 # define DO_BO_UNPACK_P(var) DO_BO_UNPACK_PTR(var, i, int, void)
315 # define DO_BO_PACK_P(var) DO_BO_PACK_PTR(var, i, int, void)
316 # define DO_BO_UNPACK_PC(var) DO_BO_UNPACK_PTR(var, i, int, char)
317 # define DO_BO_PACK_PC(var) DO_BO_PACK_PTR(var, i, int, char)
318 # elif PTRSIZE == LONGSIZE
319 # define DO_BO_UNPACK_P(var) DO_BO_UNPACK_PTR(var, l, long, void)
320 # define DO_BO_PACK_P(var) DO_BO_PACK_PTR(var, l, long, void)
321 # define DO_BO_UNPACK_PC(var) DO_BO_UNPACK_PTR(var, l, long, char)
322 # define DO_BO_PACK_PC(var) DO_BO_PACK_PTR(var, l, long, char)
324 # define DO_BO_UNPACK_P(var) BO_CANT_DOIT(unpack, pointer)
325 # define DO_BO_PACK_P(var) BO_CANT_DOIT(pack, pointer)
328 # if defined(my_htolen) && defined(my_letohn) && \
329 defined(my_htoben) && defined(my_betohn)
330 # define DO_BO_UNPACK_N(var, type) \
332 switch (TYPE_ENDIANNESS(datumtype)) { \
333 case TYPE_IS_BIG_ENDIAN: my_betohn(&var, sizeof(type)); break;\
334 case TYPE_IS_LITTLE_ENDIAN: my_letohn(&var, sizeof(type)); break;\
339 # define DO_BO_PACK_N(var, type) \
341 switch (TYPE_ENDIANNESS(datumtype)) { \
342 case TYPE_IS_BIG_ENDIAN: my_htoben(&var, sizeof(type)); break;\
343 case TYPE_IS_LITTLE_ENDIAN: my_htolen(&var, sizeof(type)); break;\
348 # define DO_BO_UNPACK_N(var, type) BO_CANT_DOIT(unpack, type)
349 # define DO_BO_PACK_N(var, type) BO_CANT_DOIT(pack, type)
352 #endif /* PERL_PACK_CAN_BYTEORDER */
354 #define PACK_SIZE_CANNOT_CSUM 0x80
355 #define PACK_SIZE_UNPREDICTABLE 0x40 /* Not a fixed size element */
356 #define PACK_SIZE_MASK 0x3F
358 /* These tables are regenerated by genpacksizetables.pl (and then hand pasted
359 in). You're unlikely ever to need to regenerate them. */
361 #if TYPE_IS_SHRIEKING != 0x100
362 ++++shriek offset should be 256
365 typedef U8 packprops_t;
368 const packprops_t packprops[512] = {
370 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
371 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
372 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
373 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
375 /* C */ sizeof(unsigned char) | PACK_SIZE_UNPREDICTABLE,
376 #if defined(HAS_LONG_DOUBLE) && defined(USE_LONG_DOUBLE)
377 /* D */ LONG_DOUBLESIZE,
384 /* I */ sizeof(unsigned int),
391 #if defined(HAS_QUAD)
392 /* Q */ sizeof(Uquad_t),
399 /* U */ sizeof(char) | PACK_SIZE_UNPREDICTABLE,
401 /* W */ sizeof(unsigned char) | PACK_SIZE_UNPREDICTABLE,
402 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
403 /* c */ sizeof(char),
404 /* d */ sizeof(double),
406 /* f */ sizeof(float),
415 /* p */ sizeof(char *) | PACK_SIZE_CANNOT_CSUM,
416 #if defined(HAS_QUAD)
417 /* q */ sizeof(Quad_t),
425 /* w */ sizeof(char) | PACK_SIZE_UNPREDICTABLE | PACK_SIZE_CANNOT_CSUM,
426 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
427 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
428 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
429 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
430 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
431 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
432 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
433 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
434 0, 0, 0, 0, 0, 0, 0, 0,
436 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
437 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
438 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
439 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
440 0, 0, 0, 0, 0, 0, 0, 0, 0,
441 /* I */ sizeof(unsigned int),
443 /* L */ sizeof(unsigned long),
445 #if defined(PERL_PACK_CAN_SHRIEKSIGN)
451 /* S */ sizeof(unsigned short),
453 #if defined(PERL_PACK_CAN_SHRIEKSIGN)
458 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
462 /* l */ sizeof(long),
464 #if defined(PERL_PACK_CAN_SHRIEKSIGN)
470 /* s */ sizeof(short),
472 #if defined(PERL_PACK_CAN_SHRIEKSIGN)
477 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
478 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
479 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
480 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
481 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
482 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
483 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
484 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
485 0, 0, 0, 0, 0, 0, 0, 0, 0
488 /* EBCDIC (or bust) */
489 const packprops_t packprops[512] = {
491 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
492 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
493 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
494 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
495 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
496 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
497 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
498 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
500 /* c */ sizeof(char),
501 /* d */ sizeof(double),
503 /* f */ sizeof(float),
513 /* p */ sizeof(char *) | PACK_SIZE_CANNOT_CSUM,
514 #if defined(HAS_QUAD)
515 /* q */ sizeof(Quad_t),
519 0, 0, 0, 0, 0, 0, 0, 0, 0,
523 /* w */ sizeof(char) | PACK_SIZE_UNPREDICTABLE | PACK_SIZE_CANNOT_CSUM,
524 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
525 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
526 /* C */ sizeof(unsigned char) | PACK_SIZE_UNPREDICTABLE,
527 #if defined(HAS_LONG_DOUBLE) && defined(USE_LONG_DOUBLE)
528 /* D */ LONG_DOUBLESIZE,
535 /* I */ sizeof(unsigned int),
543 #if defined(HAS_QUAD)
544 /* Q */ sizeof(Uquad_t),
548 0, 0, 0, 0, 0, 0, 0, 0, 0,
551 /* U */ sizeof(char) | PACK_SIZE_UNPREDICTABLE,
553 /* W */ sizeof(unsigned char) | PACK_SIZE_UNPREDICTABLE,
554 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
555 0, 0, 0, 0, 0, 0, 0, 0, 0,
557 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
558 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
559 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
560 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
561 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
562 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
563 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
564 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
565 0, 0, 0, 0, 0, 0, 0, 0, 0,
567 0, 0, 0, 0, 0, 0, 0, 0, 0,
568 /* l */ sizeof(long),
570 #if defined(PERL_PACK_CAN_SHRIEKSIGN)
575 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
576 /* s */ sizeof(short),
578 #if defined(PERL_PACK_CAN_SHRIEKSIGN)
583 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
584 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
586 /* I */ sizeof(unsigned int),
587 0, 0, 0, 0, 0, 0, 0, 0, 0,
588 /* L */ sizeof(unsigned long),
590 #if defined(PERL_PACK_CAN_SHRIEKSIGN)
595 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
596 /* S */ sizeof(unsigned short),
598 #if defined(PERL_PACK_CAN_SHRIEKSIGN)
603 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
604 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
609 uni_to_byte(pTHX_ const char **s, const char *end, I32 datumtype)
613 val = utf8n_to_uvchr((U8 *) *s, end-*s, &retlen,
614 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
615 /* We try to process malformed UTF-8 as much as possible (preferrably with
616 warnings), but these two mean we make no progress in the string and
617 might enter an infinite loop */
618 if (retlen == (STRLEN) -1 || retlen == 0)
619 Perl_croak(aTHX_ "Malformed UTF-8 string in '%c' format in unpack",
620 (int) TYPE_NO_MODIFIERS(datumtype));
622 if (ckWARN(WARN_UNPACK))
623 Perl_warner(aTHX_ packWARN(WARN_UNPACK),
624 "Character in '%c' format wrapped in unpack",
625 (int) TYPE_NO_MODIFIERS(datumtype));
632 #define SHIFT_BYTE(utf8, s, strend, datumtype) ((utf8) ? \
633 uni_to_byte(aTHX_ &(s), (strend), (datumtype)) : \
637 uni_to_bytes(pTHX_ const char **s, const char *end, const char *buf, int buf_len, I32 datumtype)
641 const char *from = *s;
643 const U32 flags = ckWARN(WARN_UTF8) ?
644 UTF8_CHECK_ONLY : (UTF8_CHECK_ONLY | UTF8_ALLOW_ANY);
645 for (;buf_len > 0; buf_len--) {
646 if (from >= end) return FALSE;
647 val = utf8n_to_uvchr((U8 *) from, end-from, &retlen, flags);
648 if (retlen == (STRLEN) -1 || retlen == 0) {
649 from += UTF8SKIP(from);
651 } else from += retlen;
656 *(U8 *)buf++ = (U8)val;
658 /* We have enough characters for the buffer. Did we have problems ? */
661 /* Rewalk the string fragment while warning */
663 const int flags = ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY;
664 for (ptr = *s; ptr < from; ptr += UTF8SKIP(ptr)) {
665 if (ptr >= end) break;
666 utf8n_to_uvuni((U8 *) ptr, end-ptr, &retlen, flags);
668 if (from > end) from = end;
670 if ((bad & 2) && ckWARN(WARN_UNPACK))
671 Perl_warner(aTHX_ packWARN(datumtype & TYPE_IS_PACK ?
672 WARN_PACK : WARN_UNPACK),
673 "Character(s) in '%c' format wrapped in %s",
674 (int) TYPE_NO_MODIFIERS(datumtype),
675 datumtype & TYPE_IS_PACK ? "pack" : "unpack");
682 next_uni_uu(pTHX_ const char **s, const char *end, I32 *out)
686 val = utf8n_to_uvchr((U8 *) *s, end-*s, &retlen, UTF8_CHECK_ONLY);
687 if (val >= 0x100 || !ISUUCHAR(val) ||
688 retlen == (STRLEN) -1 || retlen == 0) {
692 *out = PL_uudmap[val] & 077;
698 bytes_to_uni(pTHX_ const U8 *start, STRLEN len, char **dest) {
699 U8 buffer[UTF8_MAXLEN];
700 const U8 *end = start + len;
702 while (start < end) {
704 uvuni_to_utf8_flags(buffer, NATIVE_TO_UNI(*start), 0) - buffer;
714 Perl_croak(aTHX_ "Perl bug: value %d UTF-8 expands to %d bytes",
722 #define PUSH_BYTES(utf8, cur, buf, len) \
724 if (utf8) bytes_to_uni(aTHX_ (U8 *) buf, len, &(cur)); \
726 Copy(buf, cur, len, char); \
731 #define GROWING(utf8, cat, start, cur, in_len) \
733 STRLEN glen = (in_len); \
734 if (utf8) glen *= UTF8_EXPAND; \
735 if ((cur) + glen >= (start) + SvLEN(cat)) { \
736 (start) = sv_exp_grow(aTHX_ cat, glen); \
737 (cur) = (start) + SvCUR(cat); \
741 #define PUSH_GROWING_BYTES(utf8, cat, start, cur, buf, in_len) \
743 const STRLEN glen = (in_len); \
745 if (utf8) gl *= UTF8_EXPAND; \
746 if ((cur) + gl >= (start) + SvLEN(cat)) { \
748 SvCUR_set((cat), (cur) - (start)); \
749 (start) = sv_exp_grow(aTHX_ cat, gl); \
750 (cur) = (start) + SvCUR(cat); \
752 PUSH_BYTES(utf8, cur, buf, glen); \
755 #define PUSH_BYTE(utf8, s, byte) \
758 const U8 au8 = (byte); \
759 bytes_to_uni(aTHX_ &au8, 1, &(s)); \
760 } else *(U8 *)(s)++ = (byte); \
763 /* Only to be used inside a loop (see the break) */
764 #define NEXT_UNI_VAL(val, cur, str, end, utf8_flags) \
767 if (str >= end) break; \
768 val = utf8n_to_uvchr((U8 *) str, end-str, &retlen, utf8_flags); \
769 if (retlen == (STRLEN) -1 || retlen == 0) { \
771 Perl_croak(aTHX_ "Malformed UTF-8 string in pack"); \
776 static const char *_action( const tempsym_t* symptr )
778 return ( symptr->flags & FLAG_PACK ) ? "pack" : "unpack";
781 /* Returns the sizeof() struct described by pat */
783 S_measure_struct(pTHX_ tempsym_t* symptr)
787 while (next_symbol(symptr)) {
791 switch (symptr->howlen) {
793 Perl_croak(aTHX_ "Within []-length '*' not allowed in %s",
797 /* e_no_len and e_number */
798 len = symptr->length;
802 size = packprops[TYPE_NO_ENDIANNESS(symptr->code)] & PACK_SIZE_MASK;
805 /* endianness doesn't influence the size of a type */
806 switch(TYPE_NO_ENDIANNESS(symptr->code)) {
808 Perl_croak(aTHX_ "Invalid type '%c' in %s",
809 (int)TYPE_NO_MODIFIERS(symptr->code),
811 #ifdef PERL_PACK_CAN_SHRIEKSIGN
812 case '.' | TYPE_IS_SHRIEKING:
813 case '@' | TYPE_IS_SHRIEKING:
818 case 'U': /* XXXX Is it correct? */
821 Perl_croak(aTHX_ "Within []-length '%c' not allowed in %s",
822 (int) TYPE_NO_MODIFIERS(symptr->code),
829 tempsym_t savsym = *symptr;
830 symptr->patptr = savsym.grpbeg;
831 symptr->patend = savsym.grpend;
832 /* XXXX Theoretically, we need to measure many times at
833 different positions, since the subexpression may contain
834 alignment commands, but be not of aligned length.
835 Need to detect this and croak(). */
836 size = measure_struct(symptr);
840 case 'X' | TYPE_IS_SHRIEKING:
841 /* XXXX Is this useful? Then need to treat MEASURE_BACKWARDS.
843 if (!len) /* Avoid division by 0 */
845 len = total % len; /* Assumed: the start is aligned. */
850 Perl_croak(aTHX_ "'X' outside of string in %s", _action( symptr ) );
852 case 'x' | TYPE_IS_SHRIEKING:
853 if (!len) /* Avoid division by 0 */
855 star = total % len; /* Assumed: the start is aligned. */
856 if (star) /* Other portable ways? */
880 size = sizeof(char*);
890 /* locate matching closing parenthesis or bracket
891 * returns char pointer to char after match, or NULL
894 S_group_end(pTHX_ register const char *patptr, register const char *patend, char ender)
896 while (patptr < patend) {
897 const char c = *patptr++;
904 while (patptr < patend && *patptr != '\n')
908 patptr = group_end(patptr, patend, ')') + 1;
910 patptr = group_end(patptr, patend, ']') + 1;
912 Perl_croak(aTHX_ "No group ending character '%c' found in template",
918 /* Convert unsigned decimal number to binary.
919 * Expects a pointer to the first digit and address of length variable
920 * Advances char pointer to 1st non-digit char and returns number
923 S_get_num(pTHX_ register const char *patptr, I32 *lenptr )
925 I32 len = *patptr++ - '0';
926 while (isDIGIT(*patptr)) {
927 if (len >= 0x7FFFFFFF/10)
928 Perl_croak(aTHX_ "pack/unpack repeat count overflow");
929 len = (len * 10) + (*patptr++ - '0');
935 /* The marvellous template parsing routine: Using state stored in *symptr,
936 * locates next template code and count
939 S_next_symbol(pTHX_ tempsym_t* symptr )
941 const char* patptr = symptr->patptr;
942 const char* patend = symptr->patend;
944 symptr->flags &= ~FLAG_SLASH;
946 while (patptr < patend) {
947 if (isSPACE(*patptr))
949 else if (*patptr == '#') {
951 while (patptr < patend && *patptr != '\n')
956 /* We should have found a template code */
957 I32 code = *patptr++ & 0xFF;
958 U32 inherited_modifiers = 0;
960 if (code == ','){ /* grandfather in commas but with a warning */
961 if (((symptr->flags & FLAG_COMMA) == 0) && ckWARN(WARN_UNPACK)){
962 symptr->flags |= FLAG_COMMA;
963 Perl_warner(aTHX_ packWARN(WARN_UNPACK),
964 "Invalid type ',' in %s", _action( symptr ) );
969 /* for '(', skip to ')' */
971 if( isDIGIT(*patptr) || *patptr == '*' || *patptr == '[' )
972 Perl_croak(aTHX_ "()-group starts with a count in %s",
974 symptr->grpbeg = patptr;
975 patptr = 1 + ( symptr->grpend = group_end(patptr, patend, ')') );
976 if( symptr->level >= MAX_SUB_TEMPLATE_LEVEL )
977 Perl_croak(aTHX_ "Too deeply nested ()-groups in %s",
981 /* look for group modifiers to inherit */
982 if (TYPE_ENDIANNESS(symptr->flags)) {
983 if (strchr(ENDIANNESS_ALLOWED_TYPES, TYPE_NO_MODIFIERS(code)))
984 inherited_modifiers |= TYPE_ENDIANNESS(symptr->flags);
987 /* look for modifiers */
988 while (patptr < patend) {
993 modifier = TYPE_IS_SHRIEKING;
994 allowed = SHRIEKING_ALLOWED_TYPES;
996 #ifdef PERL_PACK_CAN_BYTEORDER
998 modifier = TYPE_IS_BIG_ENDIAN;
999 allowed = ENDIANNESS_ALLOWED_TYPES;
1002 modifier = TYPE_IS_LITTLE_ENDIAN;
1003 allowed = ENDIANNESS_ALLOWED_TYPES;
1005 #endif /* PERL_PACK_CAN_BYTEORDER */
1015 if (!strchr(allowed, TYPE_NO_MODIFIERS(code)))
1016 Perl_croak(aTHX_ "'%c' allowed only after types %s in %s", *patptr,
1017 allowed, _action( symptr ) );
1019 if (TYPE_ENDIANNESS(code | modifier) == TYPE_ENDIANNESS_MASK)
1020 Perl_croak(aTHX_ "Can't use both '<' and '>' after type '%c' in %s",
1021 (int) TYPE_NO_MODIFIERS(code), _action( symptr ) );
1022 else if (TYPE_ENDIANNESS(code | modifier | inherited_modifiers) ==
1023 TYPE_ENDIANNESS_MASK)
1024 Perl_croak(aTHX_ "Can't use '%c' in a group with different byte-order in %s",
1025 *patptr, _action( symptr ) );
1027 if (ckWARN(WARN_UNPACK)) {
1028 if (code & modifier)
1029 Perl_warner(aTHX_ packWARN(WARN_UNPACK),
1030 "Duplicate modifier '%c' after '%c' in %s",
1031 *patptr, (int) TYPE_NO_MODIFIERS(code),
1032 _action( symptr ) );
1039 /* inherit modifiers */
1040 code |= inherited_modifiers;
1042 /* look for count and/or / */
1043 if (patptr < patend) {
1044 if (isDIGIT(*patptr)) {
1045 patptr = get_num( patptr, &symptr->length );
1046 symptr->howlen = e_number;
1048 } else if (*patptr == '*') {
1050 symptr->howlen = e_star;
1052 } else if (*patptr == '[') {
1053 const char* lenptr = ++patptr;
1054 symptr->howlen = e_number;
1055 patptr = group_end( patptr, patend, ']' ) + 1;
1056 /* what kind of [] is it? */
1057 if (isDIGIT(*lenptr)) {
1058 lenptr = get_num( lenptr, &symptr->length );
1059 if( *lenptr != ']' )
1060 Perl_croak(aTHX_ "Malformed integer in [] in %s",
1061 _action( symptr ) );
1063 tempsym_t savsym = *symptr;
1064 symptr->patend = patptr-1;
1065 symptr->patptr = lenptr;
1066 savsym.length = measure_struct(symptr);
1070 symptr->howlen = e_no_len;
1075 while (patptr < patend) {
1076 if (isSPACE(*patptr))
1078 else if (*patptr == '#') {
1080 while (patptr < patend && *patptr != '\n')
1082 if (patptr < patend)
1085 if (*patptr == '/') {
1086 symptr->flags |= FLAG_SLASH;
1088 if (patptr < patend &&
1089 (isDIGIT(*patptr) || *patptr == '*' || *patptr == '['))
1090 Perl_croak(aTHX_ "'/' does not take a repeat count in %s",
1091 _action( symptr ) );
1097 /* at end - no count, no / */
1098 symptr->howlen = e_no_len;
1102 symptr->code = code;
1103 symptr->patptr = patptr;
1107 symptr->patptr = patptr;
1112 There is no way to cleanly handle the case where we should process the
1113 string per byte in its upgraded form while it's really in downgraded form
1114 (e.g. estimates like strend-s as an upper bound for the number of
1115 characters left wouldn't work). So if we foresee the need of this
1116 (pattern starts with U or contains U0), we want to work on the encoded
1117 version of the string. Users are advised to upgrade their pack string
1118 themselves if they need to do a lot of unpacks like this on it
1120 /* XXX These can be const */
1122 need_utf8(const char *pat, const char *patend)
1125 while (pat < patend) {
1126 if (pat[0] == '#') {
1128 pat = (const char *) memchr(pat, '\n', patend-pat);
1129 if (!pat) return FALSE;
1130 } else if (pat[0] == 'U') {
1131 if (first || pat[1] == '0') return TRUE;
1132 } else first = FALSE;
1139 first_symbol(const char *pat, const char *patend) {
1140 while (pat < patend) {
1141 if (pat[0] != '#') return pat[0];
1143 pat = (const char *) memchr(pat, '\n', patend-pat);
1151 =for apidoc unpack_str
1153 The engine implementing unpack() Perl function. Note: parameters strbeg, new_s
1154 and ocnt are not used. This call should not be used, use unpackstring instead.
1159 Perl_unpack_str(pTHX_ const char *pat, const char *patend, const char *s, const char *strbeg, const char *strend, char **new_s, I32 ocnt, U32 flags)
1166 if (flags & FLAG_DO_UTF8) flags |= FLAG_WAS_UTF8;
1167 else if (need_utf8(pat, patend)) {
1168 /* We probably should try to avoid this in case a scalar context call
1169 wouldn't get to the "U0" */
1170 STRLEN len = strend - s;
1171 s = (char *) bytes_to_utf8((U8 *) s, &len);
1174 flags |= FLAG_DO_UTF8;
1177 if (first_symbol(pat, patend) != 'U' && (flags & FLAG_DO_UTF8))
1178 flags |= FLAG_PARSE_UTF8;
1180 TEMPSYM_INIT(&sym, pat, patend, flags);
1182 return unpack_rec(&sym, s, s, strend, NULL );
1186 =for apidoc unpackstring
1188 The engine implementing unpack() Perl function. C<unpackstring> puts the
1189 extracted list items on the stack and returns the number of elements.
1190 Issue C<PUTBACK> before and C<SPAGAIN> after the call to this function.
1195 Perl_unpackstring(pTHX_ const char *pat, const char *patend, const char *s, const char *strend, U32 flags)
1199 if (flags & FLAG_DO_UTF8) flags |= FLAG_WAS_UTF8;
1200 else if (need_utf8(pat, patend)) {
1201 /* We probably should try to avoid this in case a scalar context call
1202 wouldn't get to the "U0" */
1203 STRLEN len = strend - s;
1204 s = (char *) bytes_to_utf8((U8 *) s, &len);
1207 flags |= FLAG_DO_UTF8;
1210 if (first_symbol(pat, patend) != 'U' && (flags & FLAG_DO_UTF8))
1211 flags |= FLAG_PARSE_UTF8;
1213 TEMPSYM_INIT(&sym, pat, patend, flags);
1215 return unpack_rec(&sym, s, s, strend, NULL );
1220 S_unpack_rec(pTHX_ tempsym_t* symptr, const char *s, const char *strbeg, const char *strend, const char **new_s )
1224 const I32 start_sp_offset = SP - PL_stack_base;
1230 const int bits_in_uv = CHAR_BIT * sizeof(cuv);
1231 bool beyond = FALSE;
1232 bool explicit_length;
1233 const bool unpack_only_one = (symptr->flags & FLAG_UNPACK_ONLY_ONE) != 0;
1234 bool utf8 = (symptr->flags & FLAG_PARSE_UTF8) ? 1 : 0;
1235 symptr->strbeg = s - strbeg;
1237 while (next_symbol(symptr)) {
1240 I32 datumtype = symptr->code;
1241 /* do first one only unless in list context
1242 / is implemented by unpacking the count, then popping it from the
1243 stack, so must check that we're not in the middle of a / */
1244 if ( unpack_only_one
1245 && (SP - PL_stack_base == start_sp_offset + 1)
1246 && (datumtype != '/') ) /* XXX can this be omitted */
1249 switch (howlen = symptr->howlen) {
1251 len = strend - strbeg; /* long enough */
1254 /* e_no_len and e_number */
1255 len = symptr->length;
1259 explicit_length = TRUE;
1261 beyond = s >= strend;
1263 props = packprops[TYPE_NO_ENDIANNESS(datumtype)];
1265 /* props nonzero means we can process this letter. */
1266 const long size = props & PACK_SIZE_MASK;
1267 const long howmany = (strend - s) / size;
1271 if (!checksum || (props & PACK_SIZE_CANNOT_CSUM)) {
1272 if (len && unpack_only_one) len = 1;
1278 switch(TYPE_NO_ENDIANNESS(datumtype)) {
1280 Perl_croak(aTHX_ "Invalid type '%c' in unpack", (int)TYPE_NO_MODIFIERS(datumtype) );
1283 if (howlen == e_no_len)
1284 len = 16; /* len is not specified */
1292 tempsym_t savsym = *symptr;
1293 const U32 group_modifiers = TYPE_MODIFIERS(datumtype & ~symptr->flags);
1294 symptr->flags |= group_modifiers;
1295 symptr->patend = savsym.grpend;
1296 symptr->previous = &savsym;
1300 symptr->patptr = savsym.grpbeg;
1301 if (utf8) symptr->flags |= FLAG_PARSE_UTF8;
1302 else symptr->flags &= ~FLAG_PARSE_UTF8;
1303 unpack_rec(symptr, s, strbeg, strend, &s);
1304 if (s == strend && savsym.howlen == e_star)
1305 break; /* No way to continue */
1308 savsym.flags = symptr->flags & ~group_modifiers;
1312 #ifdef PERL_PACK_CAN_SHRIEKSIGN
1313 case '.' | TYPE_IS_SHRIEKING:
1318 #ifdef PERL_PACK_CAN_SHRIEKSIGN
1319 const bool u8 = utf8 && !(datumtype & TYPE_IS_SHRIEKING);
1320 #else /* PERL_PACK_CAN_SHRIEKSIGN */
1321 const bool u8 = utf8;
1323 if (howlen == e_star) from = strbeg;
1324 else if (len <= 0) from = s;
1326 tempsym_t *group = symptr;
1328 while (--len && group) group = group->previous;
1329 from = group ? strbeg + group->strbeg : strbeg;
1332 newSVuv( u8 ? (UV) utf8_length((const U8*)from, (const U8*)s) : (UV) (s-from)) :
1333 newSViv(-(u8 ? (IV) utf8_length((const U8*)s, (const U8*)from) : (IV) (from-s)));
1334 XPUSHs(sv_2mortal(sv));
1337 #ifdef PERL_PACK_CAN_SHRIEKSIGN
1338 case '@' | TYPE_IS_SHRIEKING:
1341 s = strbeg + symptr->strbeg;
1342 #ifdef PERL_PACK_CAN_SHRIEKSIGN
1343 if (utf8 && !(datumtype & TYPE_IS_SHRIEKING))
1344 #else /* PERL_PACK_CAN_SHRIEKSIGN */
1350 Perl_croak(aTHX_ "'@' outside of string in unpack");
1355 Perl_croak(aTHX_ "'@' outside of string with malformed UTF-8 in unpack");
1358 Perl_croak(aTHX_ "'@' outside of string in unpack");
1362 case 'X' | TYPE_IS_SHRIEKING:
1363 if (!len) /* Avoid division by 0 */
1366 const char *hop, *last;
1368 hop = last = strbeg;
1370 hop += UTF8SKIP(hop);
1377 Perl_croak(aTHX_ "Malformed UTF-8 string in unpack");
1381 len = (s - strbeg) % len;
1387 Perl_croak(aTHX_ "'X' outside of string in unpack");
1388 while (--s, UTF8_IS_CONTINUATION(*s)) {
1390 Perl_croak(aTHX_ "'X' outside of string in unpack");
1395 if (len > s - strbeg)
1396 Perl_croak(aTHX_ "'X' outside of string in unpack" );
1400 case 'x' | TYPE_IS_SHRIEKING: {
1402 if (!len) /* Avoid division by 0 */
1404 if (utf8) ai32 = utf8_length((U8 *) strbeg, (U8 *) s) % len;
1405 else ai32 = (s - strbeg) % len;
1406 if (ai32 == 0) break;
1414 Perl_croak(aTHX_ "'x' outside of string in unpack");
1419 if (len > strend - s)
1420 Perl_croak(aTHX_ "'x' outside of string in unpack");
1425 Perl_croak(aTHX_ "'/' must follow a numeric type in unpack");
1431 /* Preliminary length estimate is assumed done in 'W' */
1432 if (len > strend - s) len = strend - s;
1438 for (l=len, hop=s; l>0; l--, hop += UTF8SKIP(hop)) {
1439 if (hop >= strend) {
1441 Perl_croak(aTHX_ "Malformed UTF-8 string in unpack");
1446 Perl_croak(aTHX_ "Malformed UTF-8 string in unpack");
1448 } else if (len > strend - s)
1451 if (datumtype == 'Z') {
1452 /* 'Z' strips stuff after first null */
1453 const char *ptr, *end;
1455 for (ptr = s; ptr < end; ptr++) if (*ptr == 0) break;
1456 sv = newSVpvn(s, ptr-s);
1457 if (howlen == e_star) /* exact for 'Z*' */
1458 len = ptr-s + (ptr != strend ? 1 : 0);
1459 } else if (datumtype == 'A') {
1460 /* 'A' strips both nulls and spaces */
1462 if (utf8 && (symptr->flags & FLAG_WAS_UTF8)) {
1463 for (ptr = s+len-1; ptr >= s; ptr--)
1464 if (*ptr != 0 && !UTF8_IS_CONTINUATION(*ptr) &&
1465 !is_utf8_space((U8 *) ptr)) break;
1466 if (ptr >= s) ptr += UTF8SKIP(ptr);
1469 Perl_croak(aTHX_ "Malformed UTF-8 string in unpack");
1471 for (ptr = s+len-1; ptr >= s; ptr--)
1472 if (*ptr != 0 && !isSPACE(*ptr)) break;
1475 sv = newSVpvn(s, ptr-s);
1476 } else sv = newSVpvn(s, len);
1480 /* Undo any upgrade done due to need_utf8() */
1481 if (!(symptr->flags & FLAG_WAS_UTF8))
1482 sv_utf8_downgrade(sv, 0);
1484 XPUSHs(sv_2mortal(sv));
1490 if (howlen == e_star || len > (strend - s) * 8)
1491 len = (strend - s) * 8;
1495 Newz(601, PL_bitcount, 256, char);
1496 for (bits = 1; bits < 256; bits++) {
1497 if (bits & 1) PL_bitcount[bits]++;
1498 if (bits & 2) PL_bitcount[bits]++;
1499 if (bits & 4) PL_bitcount[bits]++;
1500 if (bits & 8) PL_bitcount[bits]++;
1501 if (bits & 16) PL_bitcount[bits]++;
1502 if (bits & 32) PL_bitcount[bits]++;
1503 if (bits & 64) PL_bitcount[bits]++;
1504 if (bits & 128) PL_bitcount[bits]++;
1508 while (len >= 8 && s < strend) {
1509 cuv += PL_bitcount[uni_to_byte(aTHX_ &s, strend, datumtype)];
1514 cuv += PL_bitcount[*(U8 *)s++];
1517 if (len && s < strend) {
1519 bits = SHIFT_BYTE(utf8, s, strend, datumtype);
1520 if (datumtype == 'b')
1522 if (bits & 1) cuv++;
1527 if (bits & 0x80) cuv++;
1534 sv = sv_2mortal(NEWSV(35, len ? len : 1));
1537 if (datumtype == 'b') {
1539 const I32 ai32 = len;
1540 for (len = 0; len < ai32; len++) {
1541 if (len & 7) bits >>= 1;
1543 if (s >= strend) break;
1544 bits = uni_to_byte(aTHX_ &s, strend, datumtype);
1545 } else bits = *(U8 *) s++;
1546 *str++ = bits & 1 ? '1' : '0';
1550 const I32 ai32 = len;
1551 for (len = 0; len < ai32; len++) {
1552 if (len & 7) bits <<= 1;
1554 if (s >= strend) break;
1555 bits = uni_to_byte(aTHX_ &s, strend, datumtype);
1556 } else bits = *(U8 *) s++;
1557 *str++ = bits & 0x80 ? '1' : '0';
1561 SvCUR_set(sv, str - SvPVX_const(sv));
1568 /* Preliminary length estimate, acceptable for utf8 too */
1569 if (howlen == e_star || len > (strend - s) * 2)
1570 len = (strend - s) * 2;
1571 sv = sv_2mortal(NEWSV(35, len ? len : 1));
1574 if (datumtype == 'h') {
1577 for (len = 0; len < ai32; len++) {
1578 if (len & 1) bits >>= 4;
1580 if (s >= strend) break;
1581 bits = uni_to_byte(aTHX_ &s, strend, datumtype);
1582 } else bits = * (U8 *) s++;
1583 *str++ = PL_hexdigit[bits & 15];
1587 const I32 ai32 = len;
1588 for (len = 0; len < ai32; len++) {
1589 if (len & 1) bits <<= 4;
1591 if (s >= strend) break;
1592 bits = uni_to_byte(aTHX_ &s, strend, datumtype);
1593 } else bits = *(U8 *) s++;
1594 *str++ = PL_hexdigit[(bits >> 4) & 15];
1598 SvCUR_set(sv, str - SvPVX_const(sv));
1604 int aint = SHIFT_BYTE(utf8, s, strend, datumtype);
1605 if (aint >= 128) /* fake up signed chars */
1608 PUSHs(sv_2mortal(newSViv((IV)aint)));
1609 else if (checksum > bits_in_uv)
1610 cdouble += (NV)aint;
1619 if (explicit_length && datumtype == 'C')
1620 /* Switch to "character" mode */
1621 utf8 = (symptr->flags & FLAG_DO_UTF8) ? 1 : 0;
1624 if (datumtype == 'C' ?
1625 (symptr->flags & FLAG_DO_UTF8) &&
1626 !(symptr->flags & FLAG_WAS_UTF8) : utf8) {
1627 while (len-- > 0 && s < strend) {
1629 const UV val = utf8n_to_uvchr((U8 *) s, strend-s, &retlen,
1630 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
1631 if (retlen == (STRLEN) -1 || retlen == 0)
1632 Perl_croak(aTHX_ "Malformed UTF-8 string in unpack");
1635 PUSHs(sv_2mortal(newSVuv((UV) val)));
1636 else if (checksum > bits_in_uv)
1637 cdouble += (NV) val;
1641 } else if (!checksum)
1643 const U8 ch = *(U8 *) s++;
1644 PUSHs(sv_2mortal(newSVuv((UV) ch)));
1646 else if (checksum > bits_in_uv)
1647 while (len-- > 0) cdouble += (NV) *(U8 *) s++;
1649 while (len-- > 0) cuv += *(U8 *) s++;
1653 if (explicit_length) {
1654 /* Switch to "bytes in UTF-8" mode */
1655 if (symptr->flags & FLAG_DO_UTF8) utf8 = 0;
1657 /* Should be impossible due to the need_utf8() test */
1658 Perl_croak(aTHX_ "U0 mode on a byte string");
1662 if (len > strend - s) len = strend - s;
1664 if (len && unpack_only_one) len = 1;
1668 while (len-- > 0 && s < strend) {
1672 U8 result[UTF8_MAXLEN];
1673 const char *ptr = s;
1675 /* Bug: warns about bad utf8 even if we are short on bytes
1676 and will break out of the loop */
1677 if (!uni_to_bytes(aTHX_ &ptr, strend, (char *) result, 1,
1680 len = UTF8SKIP(result);
1681 if (!uni_to_bytes(aTHX_ &ptr, strend,
1682 (char *) &result[1], len-1, 'U')) break;
1683 auv = utf8n_to_uvuni(result, len, &retlen, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANYUV);
1686 auv = utf8n_to_uvuni((U8*)s, strend - s, &retlen, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANYUV);
1687 if (retlen == (STRLEN) -1 || retlen == 0)
1688 Perl_croak(aTHX_ "Malformed UTF-8 string in unpack");
1692 PUSHs(sv_2mortal(newSVuv((UV) auv)));
1693 else if (checksum > bits_in_uv)
1694 cdouble += (NV) auv;
1699 case 's' | TYPE_IS_SHRIEKING:
1700 #if SHORTSIZE != SIZE16
1703 SHIFT_VAR(utf8, s, strend, ashort, datumtype);
1704 DO_BO_UNPACK(ashort, s);
1706 PUSHs(sv_2mortal(newSViv((IV)ashort)));
1707 else if (checksum > bits_in_uv)
1708 cdouble += (NV)ashort;
1720 #if U16SIZE > SIZE16
1723 SHIFT16(utf8, s, strend, &ai16, datumtype);
1724 DO_BO_UNPACK(ai16, 16);
1725 #if U16SIZE > SIZE16
1730 PUSHs(sv_2mortal(newSViv((IV)ai16)));
1731 else if (checksum > bits_in_uv)
1732 cdouble += (NV)ai16;
1737 case 'S' | TYPE_IS_SHRIEKING:
1738 #if SHORTSIZE != SIZE16
1740 unsigned short aushort;
1741 SHIFT_VAR(utf8, s, strend, aushort, datumtype);
1742 DO_BO_UNPACK(aushort, s);
1744 PUSHs(sv_2mortal(newSVuv((UV) aushort)));
1745 else if (checksum > bits_in_uv)
1746 cdouble += (NV)aushort;
1759 #if U16SIZE > SIZE16
1762 SHIFT16(utf8, s, strend, &au16, datumtype);
1763 DO_BO_UNPACK(au16, 16);
1765 if (datumtype == 'n')
1766 au16 = PerlSock_ntohs(au16);
1769 if (datumtype == 'v')
1773 PUSHs(sv_2mortal(newSVuv((UV)au16)));
1774 else if (checksum > bits_in_uv)
1775 cdouble += (NV) au16;
1780 #ifdef PERL_PACK_CAN_SHRIEKSIGN
1781 case 'v' | TYPE_IS_SHRIEKING:
1782 case 'n' | TYPE_IS_SHRIEKING:
1785 # if U16SIZE > SIZE16
1788 SHIFT16(utf8, s, strend, &ai16, datumtype);
1790 if (datumtype == ('n' | TYPE_IS_SHRIEKING))
1791 ai16 = (I16) PerlSock_ntohs((U16) ai16);
1792 # endif /* HAS_NTOHS */
1794 if (datumtype == ('v' | TYPE_IS_SHRIEKING))
1795 ai16 = (I16) vtohs((U16) ai16);
1796 # endif /* HAS_VTOHS */
1798 PUSHs(sv_2mortal(newSViv((IV)ai16)));
1799 else if (checksum > bits_in_uv)
1800 cdouble += (NV) ai16;
1805 #endif /* PERL_PACK_CAN_SHRIEKSIGN */
1807 case 'i' | TYPE_IS_SHRIEKING:
1810 SHIFT_VAR(utf8, s, strend, aint, datumtype);
1811 DO_BO_UNPACK(aint, i);
1813 PUSHs(sv_2mortal(newSViv((IV)aint)));
1814 else if (checksum > bits_in_uv)
1815 cdouble += (NV)aint;
1821 case 'I' | TYPE_IS_SHRIEKING:
1824 SHIFT_VAR(utf8, s, strend, auint, datumtype);
1825 DO_BO_UNPACK(auint, i);
1827 PUSHs(sv_2mortal(newSVuv((UV)auint)));
1828 else if (checksum > bits_in_uv)
1829 cdouble += (NV)auint;
1837 SHIFT_VAR(utf8, s, strend, aiv, datumtype);
1838 #if IVSIZE == INTSIZE
1839 DO_BO_UNPACK(aiv, i);
1840 #elif IVSIZE == LONGSIZE
1841 DO_BO_UNPACK(aiv, l);
1842 #elif defined(HAS_QUAD) && IVSIZE == U64SIZE
1843 DO_BO_UNPACK(aiv, 64);
1845 Perl_croak(aTHX_ "'j' not supported on this platform");
1848 PUSHs(sv_2mortal(newSViv(aiv)));
1849 else if (checksum > bits_in_uv)
1858 SHIFT_VAR(utf8, s, strend, auv, datumtype);
1859 #if IVSIZE == INTSIZE
1860 DO_BO_UNPACK(auv, i);
1861 #elif IVSIZE == LONGSIZE
1862 DO_BO_UNPACK(auv, l);
1863 #elif defined(HAS_QUAD) && IVSIZE == U64SIZE
1864 DO_BO_UNPACK(auv, 64);
1866 Perl_croak(aTHX_ "'J' not supported on this platform");
1869 PUSHs(sv_2mortal(newSVuv(auv)));
1870 else if (checksum > bits_in_uv)
1876 case 'l' | TYPE_IS_SHRIEKING:
1877 #if LONGSIZE != SIZE32
1880 SHIFT_VAR(utf8, s, strend, along, datumtype);
1881 DO_BO_UNPACK(along, l);
1883 PUSHs(sv_2mortal(newSViv((IV)along)));
1884 else if (checksum > bits_in_uv)
1885 cdouble += (NV)along;
1896 #if U32SIZE > SIZE32
1899 SHIFT32(utf8, s, strend, &ai32, datumtype);
1900 DO_BO_UNPACK(ai32, 32);
1901 #if U32SIZE > SIZE32
1902 if (ai32 > 2147483647) ai32 -= 4294967296;
1905 PUSHs(sv_2mortal(newSViv((IV)ai32)));
1906 else if (checksum > bits_in_uv)
1907 cdouble += (NV)ai32;
1912 case 'L' | TYPE_IS_SHRIEKING:
1913 #if LONGSIZE != SIZE32
1915 unsigned long aulong;
1916 SHIFT_VAR(utf8, s, strend, aulong, datumtype);
1917 DO_BO_UNPACK(aulong, l);
1919 PUSHs(sv_2mortal(newSVuv((UV)aulong)));
1920 else if (checksum > bits_in_uv)
1921 cdouble += (NV)aulong;
1934 #if U32SIZE > SIZE32
1937 SHIFT32(utf8, s, strend, &au32, datumtype);
1938 DO_BO_UNPACK(au32, 32);
1940 if (datumtype == 'N')
1941 au32 = PerlSock_ntohl(au32);
1944 if (datumtype == 'V')
1948 PUSHs(sv_2mortal(newSVuv((UV)au32)));
1949 else if (checksum > bits_in_uv)
1950 cdouble += (NV)au32;
1955 #ifdef PERL_PACK_CAN_SHRIEKSIGN
1956 case 'V' | TYPE_IS_SHRIEKING:
1957 case 'N' | TYPE_IS_SHRIEKING:
1960 # if U32SIZE > SIZE32
1963 SHIFT32(utf8, s, strend, &ai32, datumtype);
1965 if (datumtype == ('N' | TYPE_IS_SHRIEKING))
1966 ai32 = (I32)PerlSock_ntohl((U32)ai32);
1969 if (datumtype == ('V' | TYPE_IS_SHRIEKING))
1970 ai32 = (I32)vtohl((U32)ai32);
1973 PUSHs(sv_2mortal(newSViv((IV)ai32)));
1974 else if (checksum > bits_in_uv)
1975 cdouble += (NV)ai32;
1980 #endif /* PERL_PACK_CAN_SHRIEKSIGN */
1984 SHIFT_VAR(utf8, s, strend, aptr, datumtype);
1985 DO_BO_UNPACK_PC(aptr);
1986 /* newSVpv generates undef if aptr is NULL */
1987 PUSHs(sv_2mortal(newSVpv(aptr, 0)));
1995 while (len > 0 && s < strend) {
1997 ch = SHIFT_BYTE(utf8, s, strend, datumtype);
1998 auv = (auv << 7) | (ch & 0x7f);
1999 /* UTF8_IS_XXXXX not right here - using constant 0x80 */
2002 PUSHs(sv_2mortal(newSVuv(auv)));
2007 if (++bytes >= sizeof(UV)) { /* promote to string */
2011 sv = Perl_newSVpvf(aTHX_ "%.*"UVf, (int)TYPE_DIGITS(UV), auv);
2012 while (s < strend) {
2013 ch = SHIFT_BYTE(utf8, s, strend, datumtype);
2014 sv = mul128(sv, (U8)(ch & 0x7f));
2024 PUSHs(sv_2mortal(sv));
2029 if ((s >= strend) && bytes)
2030 Perl_croak(aTHX_ "Unterminated compressed integer in unpack");
2034 if (symptr->howlen == e_star)
2035 Perl_croak(aTHX_ "'P' must have an explicit size in unpack");
2037 if (sizeof(char*) <= strend - s) {
2039 SHIFT_VAR(utf8, s, strend, aptr, datumtype);
2040 DO_BO_UNPACK_PC(aptr);
2041 /* newSVpvn generates undef if aptr is NULL */
2042 PUSHs(sv_2mortal(newSVpvn(aptr, len)));
2049 SHIFT_VAR(utf8, s, strend, aquad, datumtype);
2050 DO_BO_UNPACK(aquad, 64);
2052 PUSHs(sv_2mortal(aquad >= IV_MIN && aquad <= IV_MAX ?
2053 newSViv((IV)aquad) : newSVnv((NV)aquad)));
2054 else if (checksum > bits_in_uv)
2055 cdouble += (NV)aquad;
2063 SHIFT_VAR(utf8, s, strend, auquad, datumtype);
2064 DO_BO_UNPACK(auquad, 64);
2066 PUSHs(sv_2mortal(auquad <= UV_MAX ?
2067 newSVuv((UV)auquad):newSVnv((NV)auquad)));
2068 else if (checksum > bits_in_uv)
2069 cdouble += (NV)auquad;
2074 #endif /* HAS_QUAD */
2075 /* float and double added gnb@melba.bby.oz.au 22/11/89 */
2079 SHIFT_VAR(utf8, s, strend, afloat, datumtype);
2080 DO_BO_UNPACK_N(afloat, float);
2082 PUSHs(sv_2mortal(newSVnv((NV)afloat)));
2090 SHIFT_VAR(utf8, s, strend, adouble, datumtype);
2091 DO_BO_UNPACK_N(adouble, double);
2093 PUSHs(sv_2mortal(newSVnv((NV)adouble)));
2101 SHIFT_VAR(utf8, s, strend, anv, datumtype);
2102 DO_BO_UNPACK_N(anv, NV);
2104 PUSHs(sv_2mortal(newSVnv(anv)));
2109 #if defined(HAS_LONG_DOUBLE) && defined(USE_LONG_DOUBLE)
2112 long double aldouble;
2113 SHIFT_VAR(utf8, s, strend, aldouble, datumtype);
2114 DO_BO_UNPACK_N(aldouble, long double);
2116 PUSHs(sv_2mortal(newSVnv((NV)aldouble)));
2118 cdouble += aldouble;
2124 * Initialise the decode mapping. By using a table driven
2125 * algorithm, the code will be character-set independent
2126 * (and just as fast as doing character arithmetic)
2128 if (PL_uudmap['M'] == 0) {
2131 for (i = 0; i < sizeof(PL_uuemap); i += 1)
2132 PL_uudmap[(U8)PL_uuemap[i]] = i;
2134 * Because ' ' and '`' map to the same value,
2135 * we need to decode them both the same.
2140 const STRLEN l = (STRLEN) (strend - s) * 3 / 4;
2141 sv = sv_2mortal(NEWSV(42, l));
2142 if (l) SvPOK_on(sv);
2145 while (next_uni_uu(aTHX_ &s, strend, &len)) {
2151 next_uni_uu(aTHX_ &s, strend, &a);
2152 next_uni_uu(aTHX_ &s, strend, &b);
2153 next_uni_uu(aTHX_ &s, strend, &c);
2154 next_uni_uu(aTHX_ &s, strend, &d);
2155 hunk[0] = (char)((a << 2) | (b >> 4));
2156 hunk[1] = (char)((b << 4) | (c >> 2));
2157 hunk[2] = (char)((c << 6) | d);
2158 sv_catpvn(sv, hunk, (len > 3) ? 3 : len);
2166 /* possible checksum byte */
2167 const char *skip = s+UTF8SKIP(s);
2168 if (skip < strend && *skip == '\n')
2174 while (s < strend && *s > ' ' && ISUUCHAR(*s)) {
2179 len = PL_uudmap[*(U8*)s++] & 077;
2181 if (s < strend && ISUUCHAR(*s))
2182 a = PL_uudmap[*(U8*)s++] & 077;
2185 if (s < strend && ISUUCHAR(*s))
2186 b = PL_uudmap[*(U8*)s++] & 077;
2189 if (s < strend && ISUUCHAR(*s))
2190 c = PL_uudmap[*(U8*)s++] & 077;
2193 if (s < strend && ISUUCHAR(*s))
2194 d = PL_uudmap[*(U8*)s++] & 077;
2197 hunk[0] = (char)((a << 2) | (b >> 4));
2198 hunk[1] = (char)((b << 4) | (c >> 2));
2199 hunk[2] = (char)((c << 6) | d);
2200 sv_catpvn(sv, hunk, (len > 3) ? 3 : len);
2205 else /* possible checksum byte */
2206 if (s + 1 < strend && s[1] == '\n')
2215 if (strchr("fFdD", TYPE_NO_MODIFIERS(datumtype)) ||
2216 (checksum > bits_in_uv &&
2217 strchr("cCsSiIlLnNUWvVqQjJ", TYPE_NO_MODIFIERS(datumtype))) ) {
2220 anv = (NV) (1 << (checksum & 15));
2221 while (checksum >= 16) {
2225 while (cdouble < 0.0)
2227 cdouble = Perl_modf(cdouble / anv, &trouble) * anv;
2228 sv = newSVnv(cdouble);
2231 if (checksum < bits_in_uv) {
2232 UV mask = ((UV)1 << checksum) - 1;
2237 XPUSHs(sv_2mortal(sv));
2241 if (symptr->flags & FLAG_SLASH){
2242 if (SP - PL_stack_base - start_sp_offset <= 0)
2243 Perl_croak(aTHX_ "'/' must follow a numeric type in unpack");
2244 if( next_symbol(symptr) ){
2245 if( symptr->howlen == e_number )
2246 Perl_croak(aTHX_ "Count after length/code in unpack" );
2248 /* ...end of char buffer then no decent length available */
2249 Perl_croak(aTHX_ "length/code after end of string in unpack" );
2251 /* take top of stack (hope it's numeric) */
2254 Perl_croak(aTHX_ "Negative '/' count in unpack" );
2257 Perl_croak(aTHX_ "Code missing after '/' in unpack" );
2259 datumtype = symptr->code;
2260 explicit_length = FALSE;
2268 return SP - PL_stack_base - start_sp_offset;
2275 I32 gimme = GIMME_V;
2278 const char *pat = SvPV(left, llen);
2279 const char *s = SvPV(right, rlen);
2280 const char *strend = s + rlen;
2281 const char *patend = pat + llen;
2285 cnt = unpackstring(pat, patend, s, strend,
2286 ((gimme == G_SCALAR) ? FLAG_UNPACK_ONLY_ONE : 0)
2287 | (DO_UTF8(right) ? FLAG_DO_UTF8 : 0));
2290 if ( !cnt && gimme == G_SCALAR )
2291 PUSHs(&PL_sv_undef);
2296 doencodes(U8 *h, const char *s, I32 len)
2298 *h++ = PL_uuemap[len];
2300 *h++ = PL_uuemap[(077 & (s[0] >> 2))];
2301 *h++ = PL_uuemap[(077 & (((s[0] << 4) & 060) | ((s[1] >> 4) & 017)))];
2302 *h++ = PL_uuemap[(077 & (((s[1] << 2) & 074) | ((s[2] >> 6) & 03)))];
2303 *h++ = PL_uuemap[(077 & (s[2] & 077))];
2308 const char r = (len > 1 ? s[1] : '\0');
2309 *h++ = PL_uuemap[(077 & (s[0] >> 2))];
2310 *h++ = PL_uuemap[(077 & (((s[0] << 4) & 060) | ((r >> 4) & 017)))];
2311 *h++ = PL_uuemap[(077 & ((r << 2) & 074))];
2312 *h++ = PL_uuemap[0];
2319 S_is_an_int(pTHX_ const char *s, STRLEN l)
2322 SV *result = newSVpvn(s, l);
2323 char *result_c = SvPV(result, n_a); /* convenience */
2324 char *out = result_c;
2334 SvREFCNT_dec(result);
2357 SvREFCNT_dec(result);
2363 SvCUR_set(result, out - result_c);
2367 /* pnum must be '\0' terminated */
2369 S_div128(pTHX_ SV *pnum, bool *done)
2372 char *s = SvPV(pnum, len);
2379 const int i = m * 10 + (*t - '0');
2381 r = (i >> 7); /* r < 10 */
2388 SvCUR_set(pnum, (STRLEN) (t - s));
2393 =for apidoc pack_cat
2395 The engine implementing pack() Perl function. Note: parameters next_in_list and
2396 flags are not used. This call should not be used; use packlist instead.
2402 Perl_pack_cat(pTHX_ SV *cat, const char *pat, const char *patend, register SV **beglist, SV **endlist, SV ***next_in_list, U32 flags)
2408 TEMPSYM_INIT(&sym, pat, patend, FLAG_PACK);
2410 (void)pack_rec( cat, &sym, beglist, endlist );
2415 =for apidoc packlist
2417 The engine implementing pack() Perl function.
2423 Perl_packlist(pTHX_ SV *cat, const char *pat, const char *patend, register SV **beglist, SV **endlist )
2428 TEMPSYM_INIT(&sym, pat, patend, FLAG_PACK);
2430 /* We're going to do changes through SvPVX(cat). Make sure it's valid.
2431 Also make sure any UTF8 flag is loaded */
2432 SvPV_force(cat, no_len);
2433 if (DO_UTF8(cat)) sym.flags |= FLAG_PARSE_UTF8 | FLAG_DO_UTF8;
2435 (void)pack_rec( cat, &sym, beglist, endlist );
2438 /* like sv_utf8_upgrade, but also repoint the group start markers */
2440 marked_upgrade(pTHX_ SV *sv, tempsym_t *sym_ptr) {
2443 const char *from_ptr, *from_start, *from_end, **marks, **m;
2444 char *to_start, *to_ptr;
2446 if (SvUTF8(sv)) return;
2448 from_start = SvPVX_const(sv);
2449 from_end = from_start + SvCUR(sv);
2450 for (from_ptr = from_start; from_ptr < from_end; from_ptr++)
2451 if (!NATIVE_IS_INVARIANT(*from_ptr)) break;
2452 if (from_ptr == from_end) {
2453 /* Simple case: no character needs to be changed */
2458 len = (from_end-from_ptr)*UTF8_EXPAND+(from_ptr-from_start)+1;
2459 New('U', to_start, len, char);
2460 Copy(from_start, to_start, from_ptr-from_start, char);
2461 to_ptr = to_start + (from_ptr-from_start);
2463 New('U', marks, sym_ptr->level+2, const char *);
2464 for (group=sym_ptr; group; group = group->previous)
2465 marks[group->level] = from_start + group->strbeg;
2466 marks[sym_ptr->level+1] = from_end+1;
2467 for (m = marks; *m < from_ptr; m++)
2468 *m = to_start + (*m-from_start);
2470 for (;from_ptr < from_end; from_ptr++) {
2471 while (*m == from_ptr) *m++ = to_ptr;
2472 to_ptr = (char *) uvchr_to_utf8((U8 *) to_ptr, *(U8 *) from_ptr);
2476 while (*m == from_ptr) *m++ = to_ptr;
2477 if (m != marks + sym_ptr->level+1) {
2480 Perl_croak(aTHX_ "Assertion: marks beyond string end");
2482 for (group=sym_ptr; group; group = group->previous)
2483 group->strbeg = marks[group->level] - to_start;
2488 SvLEN_set(sv, SvLEN(sv) + SvIVX(sv));
2489 from_start -= SvIVX(sv);
2492 SvFLAGS(sv) &= ~SVf_OOK;
2495 Safefree(from_start);
2496 SvPV_set(sv, to_start);
2497 SvCUR_set(sv, to_ptr - to_start);
2502 /* Exponential string grower. Makes string extension effectively O(n)
2503 needed says how many extra bytes we need (not counting the final '\0')
2504 Only grows the string if there is an actual lack of space
2507 sv_exp_grow(pTHX_ SV *sv, STRLEN needed) {
2508 const STRLEN cur = SvCUR(sv);
2509 const STRLEN len = SvLEN(sv);
2511 if (len - cur > needed) return SvPVX(sv);
2512 extend = needed > len ? needed : len;
2513 return SvGROW(sv, len+extend+1);
2518 S_pack_rec(pTHX_ SV *cat, tempsym_t* symptr, SV **beglist, SV **endlist )
2520 tempsym_t lookahead;
2521 I32 items = endlist - beglist;
2522 bool found = next_symbol(symptr);
2523 bool utf8 = (symptr->flags & FLAG_PARSE_UTF8) ? 1 : 0;
2525 if (symptr->level == 0 && found && symptr->code == 'U') {
2526 marked_upgrade(aTHX_ cat, symptr);
2527 symptr->flags |= FLAG_DO_UTF8;
2530 symptr->strbeg = SvCUR(cat);
2536 SV *lengthcode = Nullsv;
2537 I32 datumtype = symptr->code;
2538 howlen_t howlen = symptr->howlen;
2539 char *start = SvPVX(cat);
2540 char *cur = start + SvCUR(cat);
2542 #define NEXTFROM (lengthcode ? lengthcode : items-- > 0 ? *beglist++ : &PL_sv_no)
2546 len = strchr("@Xxu", TYPE_NO_MODIFIERS(datumtype)) ?
2550 /* e_no_len and e_number */
2551 len = symptr->length;
2556 packprops_t props = packprops[TYPE_NO_ENDIANNESS(datumtype)];
2558 if (props && !(props & PACK_SIZE_UNPREDICTABLE)) {
2559 /* We can process this letter. */
2560 STRLEN size = props & PACK_SIZE_MASK;
2561 GROWING(utf8, cat, start, cur, (STRLEN) len * size);
2565 /* Look ahead for next symbol. Do we have code/code? */
2566 lookahead = *symptr;
2567 found = next_symbol(&lookahead);
2568 if (symptr->flags & FLAG_SLASH) {
2570 if (!found) Perl_croak(aTHX_ "Code missing after '/' in pack");
2571 if (strchr("aAZ", lookahead.code)) {
2572 if (lookahead.howlen == e_number) count = lookahead.length;
2575 count = DO_UTF8(*beglist) ?
2576 sv_len_utf8(*beglist) : sv_len(*beglist);
2578 if (lookahead.code == 'Z') count++;
2581 if (lookahead.howlen == e_number && lookahead.length < items)
2582 count = lookahead.length;
2585 lookahead.howlen = e_number;
2586 lookahead.length = count;
2587 lengthcode = sv_2mortal(newSViv(count));
2590 /* Code inside the switch must take care to properly update
2591 cat (CUR length and '\0' termination) if it updated *cur and
2592 doesn't simply leave using break */
2593 switch(TYPE_NO_ENDIANNESS(datumtype)) {
2595 Perl_croak(aTHX_ "Invalid type '%c' in pack",
2596 (int) TYPE_NO_MODIFIERS(datumtype));
2598 Perl_croak(aTHX_ "'%%' may not be used in pack");
2601 #ifdef PERL_PACK_CAN_SHRIEKSIGN
2602 case '.' | TYPE_IS_SHRIEKING:
2605 if (howlen == e_star) from = start;
2606 else if (len == 0) from = cur;
2608 tempsym_t *group = symptr;
2610 while (--len && group) group = group->previous;
2611 from = group ? start + group->strbeg : start;
2614 len = SvIV(fromstr);
2616 #ifdef PERL_PACK_CAN_SHRIEKSIGN
2617 case '@' | TYPE_IS_SHRIEKING:
2620 from = start + symptr->strbeg;
2622 #ifdef PERL_PACK_CAN_SHRIEKSIGN
2623 if (utf8 && !(datumtype & TYPE_IS_SHRIEKING))
2624 #else /* PERL_PACK_CAN_SHRIEKSIGN */
2628 while (len && from < cur) {
2629 from += UTF8SKIP(from);
2633 Perl_croak(aTHX_ "Malformed UTF-8 string in pack");
2635 /* Here we know from == cur */
2637 GROWING(0, cat, start, cur, len);
2638 Zero(cur, len, char);
2640 } else if (from < cur) {
2643 } else goto no_change;
2651 if (len > 0) goto grow;
2652 if (len == 0) goto no_change;
2659 tempsym_t savsym = *symptr;
2660 U32 group_modifiers = TYPE_MODIFIERS(datumtype & ~symptr->flags);
2661 symptr->flags |= group_modifiers;
2662 symptr->patend = savsym.grpend;
2664 symptr->previous = &lookahead;
2667 if (utf8) symptr->flags |= FLAG_PARSE_UTF8;
2668 else symptr->flags &= ~FLAG_PARSE_UTF8;
2669 was_utf8 = SvUTF8(cat);
2670 symptr->patptr = savsym.grpbeg;
2671 beglist = pack_rec(cat, symptr, beglist, endlist);
2672 if (SvUTF8(cat) != was_utf8)
2673 /* This had better be an upgrade while in utf8==0 mode */
2676 if (savsym.howlen == e_star && beglist == endlist)
2677 break; /* No way to continue */
2679 lookahead.flags = symptr->flags & ~group_modifiers;
2682 case 'X' | TYPE_IS_SHRIEKING:
2683 if (!len) /* Avoid division by 0 */
2690 hop += UTF8SKIP(hop);
2697 Perl_croak(aTHX_ "Malformed UTF-8 string in pack");
2701 len = (cur-start) % len;
2705 if (len < 1) goto no_change;
2709 Perl_croak(aTHX_ "'%c' outside of string in pack",
2710 (int) TYPE_NO_MODIFIERS(datumtype));
2711 while (--cur, UTF8_IS_CONTINUATION(*cur)) {
2713 Perl_croak(aTHX_ "'%c' outside of string in pack",
2714 (int) TYPE_NO_MODIFIERS(datumtype));
2720 if (cur - start < len)
2721 Perl_croak(aTHX_ "'%c' outside of string in pack",
2722 (int) TYPE_NO_MODIFIERS(datumtype));
2725 if (cur < start+symptr->strbeg) {
2726 /* Make sure group starts don't point into the void */
2728 const STRLEN length = cur-start;
2729 for (group = symptr;
2730 group && length < group->strbeg;
2731 group = group->previous) group->strbeg = length;
2732 lookahead.strbeg = length;
2735 case 'x' | TYPE_IS_SHRIEKING: {
2737 if (!len) /* Avoid division by 0 */
2739 if (utf8) ai32 = utf8_length((U8 *) start, (U8 *) cur) % len;
2740 else ai32 = (cur - start) % len;
2741 if (ai32 == 0) goto no_change;
2753 aptr = SvPV(fromstr, fromlen);
2754 if (DO_UTF8(fromstr)) {
2755 const char *end, *s;
2757 if (!utf8 && !SvUTF8(cat)) {
2758 marked_upgrade(aTHX_ cat, symptr);
2759 lookahead.flags |= FLAG_DO_UTF8;
2760 lookahead.strbeg = symptr->strbeg;
2763 cur = start + SvCUR(cat);
2765 if (howlen == e_star) {
2766 if (utf8) goto string_copy;
2770 end = aptr + fromlen;
2771 fromlen = datumtype == 'Z' ? len-1 : len;
2772 while ((I32) fromlen > 0 && s < end) {
2777 Perl_croak(aTHX_ "Malformed UTF-8 string in pack");
2780 if (datumtype == 'Z') len++;
2786 fromlen = len - fromlen;
2787 if (datumtype == 'Z') fromlen--;
2788 if (howlen == e_star) {
2790 if (datumtype == 'Z') len++;
2792 GROWING(0, cat, start, cur, len);
2793 if (!uni_to_bytes(aTHX_ &aptr, end, cur, fromlen,
2794 datumtype | TYPE_IS_PACK))
2795 Perl_croak(aTHX_ "Perl bug: predicted utf8 length not available");
2799 if (howlen == e_star) {
2801 if (datumtype == 'Z') len++;
2803 if (len <= (I32) fromlen) {
2805 if (datumtype == 'Z' && fromlen > 0) fromlen--;
2807 /* assumes a byte expands to at most UTF8_EXPAND bytes on
2809 expected_length <= from_len*UTF8_EXPAND + (len-from_len) */
2810 GROWING(0, cat, start, cur, fromlen*(UTF8_EXPAND-1)+len);
2812 while (fromlen > 0) {
2813 cur = (char *) uvchr_to_utf8((U8 *) cur, * (U8 *) aptr);
2819 if (howlen == e_star) {
2821 if (datumtype == 'Z') len++;
2823 if (len <= (I32) fromlen) {
2825 if (datumtype == 'Z' && fromlen > 0) fromlen--;
2827 GROWING(0, cat, start, cur, len);
2828 Copy(aptr, cur, fromlen, char);
2832 memset(cur, datumtype == 'A' ? ' ' : '\0', len);
2845 str = SvPV(fromstr, fromlen);
2846 end = str + fromlen;
2847 if (DO_UTF8(fromstr)) {
2849 utf8_flags = ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY;
2851 utf8_source = FALSE;
2852 utf8_flags = 0; /* Unused, but keep compilers happy */
2854 if (howlen == e_star) len = fromlen;
2855 field_len = (len+7)/8;
2856 GROWING(utf8, cat, start, cur, field_len);
2857 if (len > (I32)fromlen) len = fromlen;
2860 if (datumtype == 'B')
2864 NEXT_UNI_VAL(val, cur, str, end, utf8_flags);
2866 } else bits |= *str++ & 1;
2867 if (l & 7) bits <<= 1;
2869 PUSH_BYTE(utf8, cur, bits);
2874 /* datumtype == 'b' */
2878 NEXT_UNI_VAL(val, cur, str, end, utf8_flags);
2879 if (val & 1) bits |= 0x80;
2880 } else if (*str++ & 1)
2882 if (l & 7) bits >>= 1;
2884 PUSH_BYTE(utf8, cur, bits);
2890 if (datumtype == 'B')
2891 bits <<= 7 - (l & 7);
2893 bits >>= 7 - (l & 7);
2894 PUSH_BYTE(utf8, cur, bits);
2897 /* Determine how many chars are left in the requested field */
2899 if (howlen == e_star) field_len = 0;
2900 else field_len -= l;
2901 Zero(cur, field_len, char);
2914 str = SvPV(fromstr, fromlen);
2915 end = str + fromlen;
2916 if (DO_UTF8(fromstr)) {
2918 utf8_flags = ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY;
2920 utf8_source = FALSE;
2921 utf8_flags = 0; /* Unused, but keep compilers happy */
2923 if (howlen == e_star) len = fromlen;
2924 field_len = (len+1)/2;
2925 GROWING(utf8, cat, start, cur, field_len);
2926 if (!utf8 && len > (I32)fromlen) len = fromlen;
2929 if (datumtype == 'H')
2933 NEXT_UNI_VAL(val, cur, str, end, utf8_flags);
2934 if (val < 256 && isALPHA(val))
2935 bits |= (val + 9) & 0xf;
2938 } else if (isALPHA(*str))
2939 bits |= (*str++ + 9) & 0xf;
2941 bits |= *str++ & 0xf;
2942 if (l & 1) bits <<= 4;
2944 PUSH_BYTE(utf8, cur, bits);
2952 NEXT_UNI_VAL(val, cur, str, end, utf8_flags);
2953 if (val < 256 && isALPHA(val))
2954 bits |= ((val + 9) & 0xf) << 4;
2956 bits |= (val & 0xf) << 4;
2957 } else if (isALPHA(*str))
2958 bits |= ((*str++ + 9) & 0xf) << 4;
2960 bits |= (*str++ & 0xf) << 4;
2961 if (l & 1) bits >>= 4;
2963 PUSH_BYTE(utf8, cur, bits);
2969 PUSH_BYTE(utf8, cur, bits);
2972 /* Determine how many chars are left in the requested field */
2974 if (howlen == e_star) field_len = 0;
2975 else field_len -= l;
2976 Zero(cur, field_len, char);
2984 aiv = SvIV(fromstr);
2985 if ((-128 > aiv || aiv > 127) &&
2987 Perl_warner(aTHX_ packWARN(WARN_PACK),
2988 "Character in 'c' format wrapped in pack");
2989 PUSH_BYTE(utf8, cur, aiv & 0xff);
2994 utf8 = (symptr->flags & FLAG_DO_UTF8) ? 1 : 0;
2997 GROWING(0, cat, start, cur, len);
3001 aiv = SvIV(fromstr);
3002 if ((0 > aiv || aiv > 0xff) &&
3004 Perl_warner(aTHX_ packWARN(WARN_PACK),
3005 "Character in 'C' format wrapped in pack");
3006 *cur++ = aiv & 0xff;
3011 U8 in_bytes = IN_BYTES;
3013 end = start+SvLEN(cat)-1;
3014 if (utf8) end -= UTF8_MAXLEN-1;
3018 auv = SvUV(fromstr);
3019 if (in_bytes) auv = auv % 0x100;
3024 SvCUR_set(cat, cur - start);
3026 GROWING(0, cat, start, cur, len+UTF8_MAXLEN);
3027 end = start+SvLEN(cat)-UTF8_MAXLEN;
3029 cur = (char *) uvuni_to_utf8_flags((U8 *) cur,
3032 0 : UNICODE_ALLOW_ANY);
3037 SvCUR_set(cat, cur - start);
3038 marked_upgrade(aTHX_ cat, symptr);
3039 lookahead.flags |= FLAG_DO_UTF8;
3040 lookahead.strbeg = symptr->strbeg;
3043 cur = start + SvCUR(cat);
3044 end = start+SvLEN(cat)-UTF8_MAXLEN;
3047 if (ckWARN(WARN_PACK))
3048 Perl_warner(aTHX_ packWARN(WARN_PACK),
3049 "Character in 'W' format wrapped in pack");
3054 SvCUR_set(cat, cur - start);
3055 GROWING(0, cat, start, cur, len+1);
3056 end = start+SvLEN(cat)-1;
3058 *(U8 *) cur++ = (U8)auv;
3067 if (!(symptr->flags & FLAG_DO_UTF8)) {
3068 marked_upgrade(aTHX_ cat, symptr);
3069 lookahead.flags |= FLAG_DO_UTF8;
3070 lookahead.strbeg = symptr->strbeg;
3076 end = start+SvLEN(cat);
3077 if (!utf8) end -= UTF8_MAXLEN;
3081 auv = SvUV(fromstr);
3083 U8 buffer[UTF8_MAXLEN], *endb;
3084 endb = uvuni_to_utf8_flags(buffer, auv,
3086 0 : UNICODE_ALLOW_ANY);
3087 if (cur+(endb-buffer)*UTF8_EXPAND >= end) {
3089 SvCUR_set(cat, cur - start);
3090 GROWING(0, cat, start, cur,
3091 len+(endb-buffer)*UTF8_EXPAND);
3092 end = start+SvLEN(cat);
3094 bytes_to_uni(aTHX_ buffer, endb-buffer, &cur);
3098 SvCUR_set(cat, cur - start);
3099 GROWING(0, cat, start, cur, len+UTF8_MAXLEN);
3100 end = start+SvLEN(cat)-UTF8_MAXLEN;
3102 cur = (char *) uvuni_to_utf8_flags((U8 *) cur, auv,
3104 0 : UNICODE_ALLOW_ANY);
3109 /* Float and double added by gnb@melba.bby.oz.au 22/11/89 */
3115 anv = SvNV(fromstr);
3117 /* VOS does not automatically map a floating-point overflow
3118 during conversion from double to float into infinity, so we
3119 do it by hand. This code should either be generalized for
3120 any OS that needs it, or removed if and when VOS implements
3121 posix-976 (suggestion to support mapping to infinity).
3122 Paul.Green@stratus.com 02-04-02. */
3124 afloat = _float_constants[0]; /* single prec. inf. */
3125 else if (anv < -FLT_MAX)
3126 afloat = _float_constants[0]; /* single prec. inf. */
3127 else afloat = (float) anv;
3129 # if defined(VMS) && !defined(__IEEE_FP)
3130 /* IEEE fp overflow shenanigans are unavailable on VAX and optional
3131 * on Alpha; fake it if we don't have them.
3135 else if (anv < -FLT_MAX)
3137 else afloat = (float)anv;
3139 afloat = (float)anv;
3141 #endif /* __VOS__ */
3142 DO_BO_PACK_N(afloat, float);
3143 PUSH_VAR(utf8, cur, afloat);
3151 anv = SvNV(fromstr);
3153 /* VOS does not automatically map a floating-point overflow
3154 during conversion from long double to double into infinity,
3155 so we do it by hand. This code should either be generalized
3156 for any OS that needs it, or removed if and when VOS
3157 implements posix-976 (suggestion to support mapping to
3158 infinity). Paul.Green@stratus.com 02-04-02. */
3160 adouble = _double_constants[0]; /* double prec. inf. */
3161 else if (anv < -DBL_MAX)
3162 adouble = _double_constants[0]; /* double prec. inf. */
3163 else adouble = (double) anv;
3165 # if defined(VMS) && !defined(__IEEE_FP)
3166 /* IEEE fp overflow shenanigans are unavailable on VAX and optional
3167 * on Alpha; fake it if we don't have them.
3171 else if (anv < -DBL_MAX)
3173 else adouble = (double)anv;
3175 adouble = (double)anv;
3177 #endif /* __VOS__ */
3178 DO_BO_PACK_N(adouble, double);
3179 PUSH_VAR(utf8, cur, adouble);
3184 Zero(&anv, 1, NV); /* can be long double with unused bits */
3187 anv = SvNV(fromstr);
3188 DO_BO_PACK_N(anv, NV);
3189 PUSH_VAR(utf8, cur, anv);
3193 #if defined(HAS_LONG_DOUBLE) && defined(USE_LONG_DOUBLE)
3195 long double aldouble;
3196 /* long doubles can have unused bits, which may be nonzero */
3197 Zero(&aldouble, 1, long double);
3200 aldouble = (long double)SvNV(fromstr);
3201 DO_BO_PACK_N(aldouble, long double);
3202 PUSH_VAR(utf8, cur, aldouble);
3207 #ifdef PERL_PACK_CAN_SHRIEKSIGN
3208 case 'n' | TYPE_IS_SHRIEKING:
3214 ai16 = (I16)SvIV(fromstr);
3216 ai16 = PerlSock_htons(ai16);
3218 PUSH16(utf8, cur, &ai16);
3221 #ifdef PERL_PACK_CAN_SHRIEKSIGN
3222 case 'v' | TYPE_IS_SHRIEKING:
3228 ai16 = (I16)SvIV(fromstr);
3232 PUSH16(utf8, cur, &ai16);
3235 case 'S' | TYPE_IS_SHRIEKING:
3236 #if SHORTSIZE != SIZE16
3238 unsigned short aushort;
3240 aushort = SvUV(fromstr);
3241 DO_BO_PACK(aushort, s);
3242 PUSH_VAR(utf8, cur, aushort);
3252 au16 = (U16)SvUV(fromstr);
3253 DO_BO_PACK(au16, 16);
3254 PUSH16(utf8, cur, &au16);
3257 case 's' | TYPE_IS_SHRIEKING:
3258 #if SHORTSIZE != SIZE16
3262 ashort = SvIV(fromstr);
3263 DO_BO_PACK(ashort, s);
3264 PUSH_VAR(utf8, cur, ashort);
3274 ai16 = (I16)SvIV(fromstr);
3275 DO_BO_PACK(ai16, 16);
3276 PUSH16(utf8, cur, &ai16);
3280 case 'I' | TYPE_IS_SHRIEKING:
3284 auint = SvUV(fromstr);
3285 DO_BO_PACK(auint, i);
3286 PUSH_VAR(utf8, cur, auint);
3293 aiv = SvIV(fromstr);
3294 #if IVSIZE == INTSIZE
3296 #elif IVSIZE == LONGSIZE
3298 #elif defined(HAS_QUAD) && IVSIZE == U64SIZE
3299 DO_BO_PACK(aiv, 64);
3301 Perl_croak(aTHX_ "'j' not supported on this platform");
3303 PUSH_VAR(utf8, cur, aiv);
3310 auv = SvUV(fromstr);
3311 #if UVSIZE == INTSIZE
3313 #elif UVSIZE == LONGSIZE
3315 #elif defined(HAS_QUAD) && UVSIZE == U64SIZE
3316 DO_BO_PACK(auv, 64);
3318 Perl_croak(aTHX_ "'J' not supported on this platform");
3320 PUSH_VAR(utf8, cur, auv);
3327 anv = SvNV(fromstr);
3331 SvCUR_set(cat, cur - start);
3332 Perl_croak(aTHX_ "Cannot compress negative numbers in pack");
3335 /* 0xFFFFFFFFFFFFFFFF may cast to 18446744073709551616.0,
3336 which is == UV_MAX_P1. IOK is fine (instead of UV_only), as
3337 any negative IVs will have already been got by the croak()
3338 above. IOK is untrue for fractions, so we test them
3339 against UV_MAX_P1. */
3340 if (SvIOK(fromstr) || anv < UV_MAX_P1) {
3341 char buf[(sizeof(UV)*CHAR_BIT)/7+1];
3342 char *in = buf + sizeof(buf);
3343 UV auv = SvUV(fromstr);
3346 *--in = (char)((auv & 0x7f) | 0x80);
3349 buf[sizeof(buf) - 1] &= 0x7f; /* clear continue bit */
3350 PUSH_GROWING_BYTES(utf8, cat, start, cur,
3351 in, (buf + sizeof(buf)) - in);
3352 } else if (SvPOKp(fromstr))
3354 else if (SvNOKp(fromstr)) {
3355 /* 10**NV_MAX_10_EXP is the largest power of 10
3356 so 10**(NV_MAX_10_EXP+1) is definately unrepresentable
3357 given 10**(NV_MAX_10_EXP+1) == 128 ** x solve for x:
3358 x = (NV_MAX_10_EXP+1) * log (10) / log (128)
3359 And with that many bytes only Inf can overflow.
3360 Some C compilers are strict about integral constant
3361 expressions so we conservatively divide by a slightly
3362 smaller integer instead of multiplying by the exact
3363 floating-point value.
3365 #ifdef NV_MAX_10_EXP
3366 /* char buf[1 + (int)((NV_MAX_10_EXP + 1) * 0.47456)]; -- invalid C */
3367 char buf[1 + (int)((NV_MAX_10_EXP + 1) / 2)]; /* valid C */
3369 /* char buf[1 + (int)((308 + 1) * 0.47456)]; -- invalid C */
3370 char buf[1 + (int)((308 + 1) / 2)]; /* valid C */
3372 char *in = buf + sizeof(buf);
3374 anv = Perl_floor(anv);
3376 NV next = Perl_floor(anv / 128);
3377 if (in <= buf) /* this cannot happen ;-) */
3378 Perl_croak(aTHX_ "Cannot compress integer in pack");
3379 *--in = (unsigned char)(anv - (next * 128)) | 0x80;
3382 buf[sizeof(buf) - 1] &= 0x7f; /* clear continue bit */
3383 PUSH_GROWING_BYTES(utf8, cat, start, cur,
3384 in, (buf + sizeof(buf)) - in);
3386 char *from, *result, *in;
3392 /* Copy string and check for compliance */
3393 from = SvPV(fromstr, len);
3394 if ((norm = is_an_int(from, len)) == NULL)
3395 Perl_croak(aTHX_ "Can only compress unsigned integers in pack");
3397 New('w', result, len, char);
3400 while (!done) *--in = div128(norm, &done) | 0x80;
3401 result[len - 1] &= 0x7F; /* clear continue bit */
3402 PUSH_GROWING_BYTES(utf8, cat, start, cur,
3403 in, (result + len) - in);
3405 SvREFCNT_dec(norm); /* free norm */
3410 case 'i' | TYPE_IS_SHRIEKING:
3414 aint = SvIV(fromstr);
3415 DO_BO_PACK(aint, i);
3416 PUSH_VAR(utf8, cur, aint);
3419 #ifdef PERL_PACK_CAN_SHRIEKSIGN
3420 case 'N' | TYPE_IS_SHRIEKING:
3426 au32 = SvUV(fromstr);
3428 au32 = PerlSock_htonl(au32);
3430 PUSH32(utf8, cur, &au32);
3433 #ifdef PERL_PACK_CAN_SHRIEKSIGN
3434 case 'V' | TYPE_IS_SHRIEKING:
3440 au32 = SvUV(fromstr);
3444 PUSH32(utf8, cur, &au32);
3447 case 'L' | TYPE_IS_SHRIEKING:
3448 #if LONGSIZE != SIZE32
3450 unsigned long aulong;
3452 aulong = SvUV(fromstr);
3453 DO_BO_PACK(aulong, l);
3454 PUSH_VAR(utf8, cur, aulong);
3464 au32 = SvUV(fromstr);
3465 DO_BO_PACK(au32, 32);
3466 PUSH32(utf8, cur, &au32);
3469 case 'l' | TYPE_IS_SHRIEKING:
3470 #if LONGSIZE != SIZE32
3474 along = SvIV(fromstr);
3475 DO_BO_PACK(along, l);
3476 PUSH_VAR(utf8, cur, along);
3486 ai32 = SvIV(fromstr);
3487 DO_BO_PACK(ai32, 32);
3488 PUSH32(utf8, cur, &ai32);
3496 auquad = (Uquad_t) SvUV(fromstr);
3497 DO_BO_PACK(auquad, 64);
3498 PUSH_VAR(utf8, cur, auquad);
3505 aquad = (Quad_t)SvIV(fromstr);
3506 DO_BO_PACK(aquad, 64);
3507 PUSH_VAR(utf8, cur, aquad);
3510 #endif /* HAS_QUAD */
3512 len = 1; /* assume SV is correct length */
3513 GROWING(utf8, cat, start, cur, sizeof(char *));
3520 SvGETMAGIC(fromstr);
3521 if (!SvOK(fromstr)) aptr = NULL;
3524 /* XXX better yet, could spirit away the string to
3525 * a safe spot and hang on to it until the result
3526 * of pack() (and all copies of the result) are
3529 if (ckWARN(WARN_PACK) &&
3530 (SvTEMP(fromstr) || (SvPADTMP(fromstr) &&
3531 !SvREADONLY(fromstr)))) {
3532 Perl_warner(aTHX_ packWARN(WARN_PACK),
3533 "Attempt to pack pointer to temporary value");
3535 if (SvPOK(fromstr) || SvNIOK(fromstr))
3536 aptr = SvPV_flags(fromstr, n_a, 0);
3538 aptr = SvPV_force_flags(fromstr, n_a, 0);
3540 DO_BO_PACK_PC(aptr);
3541 PUSH_VAR(utf8, cur, aptr);
3545 const char *aptr, *aend;
3549 if (len <= 2) len = 45;
3550 else len = len / 3 * 3;
3552 Perl_warner(aTHX_ packWARN(WARN_PACK),
3553 "Field too wide in 'u' format in pack");
3556 aptr = SvPV(fromstr, fromlen);
3557 from_utf8 = DO_UTF8(fromstr);
3559 aend = aptr + fromlen;
3560 fromlen = sv_len_utf8(fromstr);
3561 } else aend = NULL; /* Unused, but keep compilers happy */
3562 GROWING(utf8, cat, start, cur, (fromlen+2) / 3 * 4 + (fromlen+len-1)/len * 2);
3563 while (fromlen > 0) {
3566 U8 hunk[1+63/3*4+1];
3568 if ((I32)fromlen > len)
3574 if (!uni_to_bytes(aTHX_ &aptr, aend, buffer, todo,
3575 'u' | TYPE_IS_PACK)) {
3577 SvCUR_set(cat, cur - start);
3578 Perl_croak(aTHX_ "Assertion: string is shorter than advertised");
3580 end = doencodes(hunk, buffer, todo);
3582 end = doencodes(hunk, aptr, todo);
3585 PUSH_BYTES(utf8, cur, hunk, end-hunk);
3592 SvCUR_set(cat, cur - start);
3594 *symptr = lookahead;
3603 dSP; dMARK; dORIGMARK; dTARGET;
3604 register SV *cat = TARG;
3606 register const char *pat = SvPVx(*++MARK, fromlen);
3607 register const char *patend = pat + fromlen;
3610 sv_setpvn(cat, "", 0);
3613 packlist(cat, pat, patend, MARK, SP + 1);
3623 * c-indentation-style: bsd
3625 * indent-tabs-mode: t
3628 * ex: set ts=8 sts=4 sw=4 noet: