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 */
2010 sv = Perl_newSVpvf(aTHX_ "%.*"UVf, (int)TYPE_DIGITS(UV), auv);
2011 while (s < strend) {
2012 ch = SHIFT_BYTE(utf8, s, strend, datumtype);
2013 sv = mul128(sv, (U8)(ch & 0x7f));
2019 t = SvPV_nolen_const(sv);
2023 PUSHs(sv_2mortal(sv));
2028 if ((s >= strend) && bytes)
2029 Perl_croak(aTHX_ "Unterminated compressed integer in unpack");
2033 if (symptr->howlen == e_star)
2034 Perl_croak(aTHX_ "'P' must have an explicit size in unpack");
2036 if (sizeof(char*) <= strend - s) {
2038 SHIFT_VAR(utf8, s, strend, aptr, datumtype);
2039 DO_BO_UNPACK_PC(aptr);
2040 /* newSVpvn generates undef if aptr is NULL */
2041 PUSHs(sv_2mortal(newSVpvn(aptr, len)));
2048 SHIFT_VAR(utf8, s, strend, aquad, datumtype);
2049 DO_BO_UNPACK(aquad, 64);
2051 PUSHs(sv_2mortal(aquad >= IV_MIN && aquad <= IV_MAX ?
2052 newSViv((IV)aquad) : newSVnv((NV)aquad)));
2053 else if (checksum > bits_in_uv)
2054 cdouble += (NV)aquad;
2062 SHIFT_VAR(utf8, s, strend, auquad, datumtype);
2063 DO_BO_UNPACK(auquad, 64);
2065 PUSHs(sv_2mortal(auquad <= UV_MAX ?
2066 newSVuv((UV)auquad):newSVnv((NV)auquad)));
2067 else if (checksum > bits_in_uv)
2068 cdouble += (NV)auquad;
2073 #endif /* HAS_QUAD */
2074 /* float and double added gnb@melba.bby.oz.au 22/11/89 */
2078 SHIFT_VAR(utf8, s, strend, afloat, datumtype);
2079 DO_BO_UNPACK_N(afloat, float);
2081 PUSHs(sv_2mortal(newSVnv((NV)afloat)));
2089 SHIFT_VAR(utf8, s, strend, adouble, datumtype);
2090 DO_BO_UNPACK_N(adouble, double);
2092 PUSHs(sv_2mortal(newSVnv((NV)adouble)));
2100 SHIFT_VAR(utf8, s, strend, anv, datumtype);
2101 DO_BO_UNPACK_N(anv, NV);
2103 PUSHs(sv_2mortal(newSVnv(anv)));
2108 #if defined(HAS_LONG_DOUBLE) && defined(USE_LONG_DOUBLE)
2111 long double aldouble;
2112 SHIFT_VAR(utf8, s, strend, aldouble, datumtype);
2113 DO_BO_UNPACK_N(aldouble, long double);
2115 PUSHs(sv_2mortal(newSVnv((NV)aldouble)));
2117 cdouble += aldouble;
2123 * Initialise the decode mapping. By using a table driven
2124 * algorithm, the code will be character-set independent
2125 * (and just as fast as doing character arithmetic)
2127 if (PL_uudmap['M'] == 0) {
2130 for (i = 0; i < sizeof(PL_uuemap); i += 1)
2131 PL_uudmap[(U8)PL_uuemap[i]] = i;
2133 * Because ' ' and '`' map to the same value,
2134 * we need to decode them both the same.
2139 const STRLEN l = (STRLEN) (strend - s) * 3 / 4;
2140 sv = sv_2mortal(NEWSV(42, l));
2141 if (l) SvPOK_on(sv);
2144 while (next_uni_uu(aTHX_ &s, strend, &len)) {
2150 next_uni_uu(aTHX_ &s, strend, &a);
2151 next_uni_uu(aTHX_ &s, strend, &b);
2152 next_uni_uu(aTHX_ &s, strend, &c);
2153 next_uni_uu(aTHX_ &s, strend, &d);
2154 hunk[0] = (char)((a << 2) | (b >> 4));
2155 hunk[1] = (char)((b << 4) | (c >> 2));
2156 hunk[2] = (char)((c << 6) | d);
2157 sv_catpvn(sv, hunk, (len > 3) ? 3 : len);
2165 /* possible checksum byte */
2166 const char *skip = s+UTF8SKIP(s);
2167 if (skip < strend && *skip == '\n')
2173 while (s < strend && *s > ' ' && ISUUCHAR(*s)) {
2178 len = PL_uudmap[*(U8*)s++] & 077;
2180 if (s < strend && ISUUCHAR(*s))
2181 a = PL_uudmap[*(U8*)s++] & 077;
2184 if (s < strend && ISUUCHAR(*s))
2185 b = PL_uudmap[*(U8*)s++] & 077;
2188 if (s < strend && ISUUCHAR(*s))
2189 c = PL_uudmap[*(U8*)s++] & 077;
2192 if (s < strend && ISUUCHAR(*s))
2193 d = PL_uudmap[*(U8*)s++] & 077;
2196 hunk[0] = (char)((a << 2) | (b >> 4));
2197 hunk[1] = (char)((b << 4) | (c >> 2));
2198 hunk[2] = (char)((c << 6) | d);
2199 sv_catpvn(sv, hunk, (len > 3) ? 3 : len);
2204 else /* possible checksum byte */
2205 if (s + 1 < strend && s[1] == '\n')
2214 if (strchr("fFdD", TYPE_NO_MODIFIERS(datumtype)) ||
2215 (checksum > bits_in_uv &&
2216 strchr("cCsSiIlLnNUWvVqQjJ", TYPE_NO_MODIFIERS(datumtype))) ) {
2219 anv = (NV) (1 << (checksum & 15));
2220 while (checksum >= 16) {
2224 while (cdouble < 0.0)
2226 cdouble = Perl_modf(cdouble / anv, &trouble) * anv;
2227 sv = newSVnv(cdouble);
2230 if (checksum < bits_in_uv) {
2231 UV mask = ((UV)1 << checksum) - 1;
2236 XPUSHs(sv_2mortal(sv));
2240 if (symptr->flags & FLAG_SLASH){
2241 if (SP - PL_stack_base - start_sp_offset <= 0)
2242 Perl_croak(aTHX_ "'/' must follow a numeric type in unpack");
2243 if( next_symbol(symptr) ){
2244 if( symptr->howlen == e_number )
2245 Perl_croak(aTHX_ "Count after length/code in unpack" );
2247 /* ...end of char buffer then no decent length available */
2248 Perl_croak(aTHX_ "length/code after end of string in unpack" );
2250 /* take top of stack (hope it's numeric) */
2253 Perl_croak(aTHX_ "Negative '/' count in unpack" );
2256 Perl_croak(aTHX_ "Code missing after '/' in unpack" );
2258 datumtype = symptr->code;
2259 explicit_length = FALSE;
2267 return SP - PL_stack_base - start_sp_offset;
2274 I32 gimme = GIMME_V;
2277 const char *pat = SvPV_const(left, llen);
2278 const char *s = SvPV_const(right, rlen);
2279 const char *strend = s + rlen;
2280 const char *patend = pat + llen;
2284 cnt = unpackstring(pat, patend, s, strend,
2285 ((gimme == G_SCALAR) ? FLAG_UNPACK_ONLY_ONE : 0)
2286 | (DO_UTF8(right) ? FLAG_DO_UTF8 : 0));
2289 if ( !cnt && gimme == G_SCALAR )
2290 PUSHs(&PL_sv_undef);
2295 doencodes(U8 *h, const char *s, I32 len)
2297 *h++ = PL_uuemap[len];
2299 *h++ = PL_uuemap[(077 & (s[0] >> 2))];
2300 *h++ = PL_uuemap[(077 & (((s[0] << 4) & 060) | ((s[1] >> 4) & 017)))];
2301 *h++ = PL_uuemap[(077 & (((s[1] << 2) & 074) | ((s[2] >> 6) & 03)))];
2302 *h++ = PL_uuemap[(077 & (s[2] & 077))];
2307 const char r = (len > 1 ? s[1] : '\0');
2308 *h++ = PL_uuemap[(077 & (s[0] >> 2))];
2309 *h++ = PL_uuemap[(077 & (((s[0] << 4) & 060) | ((r >> 4) & 017)))];
2310 *h++ = PL_uuemap[(077 & ((r << 2) & 074))];
2311 *h++ = PL_uuemap[0];
2318 S_is_an_int(pTHX_ const char *s, STRLEN l)
2320 SV *result = newSVpvn(s, l);
2321 char *result_c = SvPV_nolen(result); /* convenience */
2322 char *out = result_c;
2332 SvREFCNT_dec(result);
2355 SvREFCNT_dec(result);
2361 SvCUR_set(result, out - result_c);
2365 /* pnum must be '\0' terminated */
2367 S_div128(pTHX_ SV *pnum, bool *done)
2370 char *s = SvPV(pnum, len);
2377 const int i = m * 10 + (*t - '0');
2379 r = (i >> 7); /* r < 10 */
2386 SvCUR_set(pnum, (STRLEN) (t - s));
2391 =for apidoc pack_cat
2393 The engine implementing pack() Perl function. Note: parameters next_in_list and
2394 flags are not used. This call should not be used; use packlist instead.
2400 Perl_pack_cat(pTHX_ SV *cat, const char *pat, const char *patend, register SV **beglist, SV **endlist, SV ***next_in_list, U32 flags)
2406 TEMPSYM_INIT(&sym, pat, patend, FLAG_PACK);
2408 (void)pack_rec( cat, &sym, beglist, endlist );
2413 =for apidoc packlist
2415 The engine implementing pack() Perl function.
2421 Perl_packlist(pTHX_ SV *cat, const char *pat, const char *patend, register SV **beglist, SV **endlist )
2426 TEMPSYM_INIT(&sym, pat, patend, FLAG_PACK);
2428 /* We're going to do changes through SvPVX(cat). Make sure it's valid.
2429 Also make sure any UTF8 flag is loaded */
2430 SvPV_force(cat, no_len);
2431 if (DO_UTF8(cat)) sym.flags |= FLAG_PARSE_UTF8 | FLAG_DO_UTF8;
2433 (void)pack_rec( cat, &sym, beglist, endlist );
2436 /* like sv_utf8_upgrade, but also repoint the group start markers */
2438 marked_upgrade(pTHX_ SV *sv, tempsym_t *sym_ptr) {
2441 const char *from_ptr, *from_start, *from_end, **marks, **m;
2442 char *to_start, *to_ptr;
2444 if (SvUTF8(sv)) return;
2446 from_start = SvPVX_const(sv);
2447 from_end = from_start + SvCUR(sv);
2448 for (from_ptr = from_start; from_ptr < from_end; from_ptr++)
2449 if (!NATIVE_IS_INVARIANT(*from_ptr)) break;
2450 if (from_ptr == from_end) {
2451 /* Simple case: no character needs to be changed */
2456 len = (from_end-from_ptr)*UTF8_EXPAND+(from_ptr-from_start)+1;
2457 New('U', to_start, len, char);
2458 Copy(from_start, to_start, from_ptr-from_start, char);
2459 to_ptr = to_start + (from_ptr-from_start);
2461 New('U', marks, sym_ptr->level+2, const char *);
2462 for (group=sym_ptr; group; group = group->previous)
2463 marks[group->level] = from_start + group->strbeg;
2464 marks[sym_ptr->level+1] = from_end+1;
2465 for (m = marks; *m < from_ptr; m++)
2466 *m = to_start + (*m-from_start);
2468 for (;from_ptr < from_end; from_ptr++) {
2469 while (*m == from_ptr) *m++ = to_ptr;
2470 to_ptr = (char *) uvchr_to_utf8((U8 *) to_ptr, *(U8 *) from_ptr);
2474 while (*m == from_ptr) *m++ = to_ptr;
2475 if (m != marks + sym_ptr->level+1) {
2478 Perl_croak(aTHX_ "Assertion: marks beyond string end");
2480 for (group=sym_ptr; group; group = group->previous)
2481 group->strbeg = marks[group->level] - to_start;
2486 SvLEN_set(sv, SvLEN(sv) + SvIVX(sv));
2487 from_start -= SvIVX(sv);
2490 SvFLAGS(sv) &= ~SVf_OOK;
2493 Safefree(from_start);
2494 SvPV_set(sv, to_start);
2495 SvCUR_set(sv, to_ptr - to_start);
2500 /* Exponential string grower. Makes string extension effectively O(n)
2501 needed says how many extra bytes we need (not counting the final '\0')
2502 Only grows the string if there is an actual lack of space
2505 sv_exp_grow(pTHX_ SV *sv, STRLEN needed) {
2506 const STRLEN cur = SvCUR(sv);
2507 const STRLEN len = SvLEN(sv);
2509 if (len - cur > needed) return SvPVX(sv);
2510 extend = needed > len ? needed : len;
2511 return SvGROW(sv, len+extend+1);
2516 S_pack_rec(pTHX_ SV *cat, tempsym_t* symptr, SV **beglist, SV **endlist )
2518 tempsym_t lookahead;
2519 I32 items = endlist - beglist;
2520 bool found = next_symbol(symptr);
2521 bool utf8 = (symptr->flags & FLAG_PARSE_UTF8) ? 1 : 0;
2523 if (symptr->level == 0 && found && symptr->code == 'U') {
2524 marked_upgrade(aTHX_ cat, symptr);
2525 symptr->flags |= FLAG_DO_UTF8;
2528 symptr->strbeg = SvCUR(cat);
2534 SV *lengthcode = Nullsv;
2535 I32 datumtype = symptr->code;
2536 howlen_t howlen = symptr->howlen;
2537 char *start = SvPVX(cat);
2538 char *cur = start + SvCUR(cat);
2540 #define NEXTFROM (lengthcode ? lengthcode : items-- > 0 ? *beglist++ : &PL_sv_no)
2544 len = strchr("@Xxu", TYPE_NO_MODIFIERS(datumtype)) ?
2548 /* e_no_len and e_number */
2549 len = symptr->length;
2554 packprops_t props = packprops[TYPE_NO_ENDIANNESS(datumtype)];
2556 if (props && !(props & PACK_SIZE_UNPREDICTABLE)) {
2557 /* We can process this letter. */
2558 STRLEN size = props & PACK_SIZE_MASK;
2559 GROWING(utf8, cat, start, cur, (STRLEN) len * size);
2563 /* Look ahead for next symbol. Do we have code/code? */
2564 lookahead = *symptr;
2565 found = next_symbol(&lookahead);
2566 if (symptr->flags & FLAG_SLASH) {
2568 if (!found) Perl_croak(aTHX_ "Code missing after '/' in pack");
2569 if (strchr("aAZ", lookahead.code)) {
2570 if (lookahead.howlen == e_number) count = lookahead.length;
2573 count = DO_UTF8(*beglist) ?
2574 sv_len_utf8(*beglist) : sv_len(*beglist);
2576 if (lookahead.code == 'Z') count++;
2579 if (lookahead.howlen == e_number && lookahead.length < items)
2580 count = lookahead.length;
2583 lookahead.howlen = e_number;
2584 lookahead.length = count;
2585 lengthcode = sv_2mortal(newSViv(count));
2588 /* Code inside the switch must take care to properly update
2589 cat (CUR length and '\0' termination) if it updated *cur and
2590 doesn't simply leave using break */
2591 switch(TYPE_NO_ENDIANNESS(datumtype)) {
2593 Perl_croak(aTHX_ "Invalid type '%c' in pack",
2594 (int) TYPE_NO_MODIFIERS(datumtype));
2596 Perl_croak(aTHX_ "'%%' may not be used in pack");
2599 #ifdef PERL_PACK_CAN_SHRIEKSIGN
2600 case '.' | TYPE_IS_SHRIEKING:
2603 if (howlen == e_star) from = start;
2604 else if (len == 0) from = cur;
2606 tempsym_t *group = symptr;
2608 while (--len && group) group = group->previous;
2609 from = group ? start + group->strbeg : start;
2612 len = SvIV(fromstr);
2614 #ifdef PERL_PACK_CAN_SHRIEKSIGN
2615 case '@' | TYPE_IS_SHRIEKING:
2618 from = start + symptr->strbeg;
2620 #ifdef PERL_PACK_CAN_SHRIEKSIGN
2621 if (utf8 && !(datumtype & TYPE_IS_SHRIEKING))
2622 #else /* PERL_PACK_CAN_SHRIEKSIGN */
2626 while (len && from < cur) {
2627 from += UTF8SKIP(from);
2631 Perl_croak(aTHX_ "Malformed UTF-8 string in pack");
2633 /* Here we know from == cur */
2635 GROWING(0, cat, start, cur, len);
2636 Zero(cur, len, char);
2638 } else if (from < cur) {
2641 } else goto no_change;
2649 if (len > 0) goto grow;
2650 if (len == 0) goto no_change;
2657 tempsym_t savsym = *symptr;
2658 U32 group_modifiers = TYPE_MODIFIERS(datumtype & ~symptr->flags);
2659 symptr->flags |= group_modifiers;
2660 symptr->patend = savsym.grpend;
2662 symptr->previous = &lookahead;
2665 if (utf8) symptr->flags |= FLAG_PARSE_UTF8;
2666 else symptr->flags &= ~FLAG_PARSE_UTF8;
2667 was_utf8 = SvUTF8(cat);
2668 symptr->patptr = savsym.grpbeg;
2669 beglist = pack_rec(cat, symptr, beglist, endlist);
2670 if (SvUTF8(cat) != was_utf8)
2671 /* This had better be an upgrade while in utf8==0 mode */
2674 if (savsym.howlen == e_star && beglist == endlist)
2675 break; /* No way to continue */
2677 lookahead.flags = symptr->flags & ~group_modifiers;
2680 case 'X' | TYPE_IS_SHRIEKING:
2681 if (!len) /* Avoid division by 0 */
2688 hop += UTF8SKIP(hop);
2695 Perl_croak(aTHX_ "Malformed UTF-8 string in pack");
2699 len = (cur-start) % len;
2703 if (len < 1) goto no_change;
2707 Perl_croak(aTHX_ "'%c' outside of string in pack",
2708 (int) TYPE_NO_MODIFIERS(datumtype));
2709 while (--cur, UTF8_IS_CONTINUATION(*cur)) {
2711 Perl_croak(aTHX_ "'%c' outside of string in pack",
2712 (int) TYPE_NO_MODIFIERS(datumtype));
2718 if (cur - start < len)
2719 Perl_croak(aTHX_ "'%c' outside of string in pack",
2720 (int) TYPE_NO_MODIFIERS(datumtype));
2723 if (cur < start+symptr->strbeg) {
2724 /* Make sure group starts don't point into the void */
2726 const STRLEN length = cur-start;
2727 for (group = symptr;
2728 group && length < group->strbeg;
2729 group = group->previous) group->strbeg = length;
2730 lookahead.strbeg = length;
2733 case 'x' | TYPE_IS_SHRIEKING: {
2735 if (!len) /* Avoid division by 0 */
2737 if (utf8) ai32 = utf8_length((U8 *) start, (U8 *) cur) % len;
2738 else ai32 = (cur - start) % len;
2739 if (ai32 == 0) goto no_change;
2751 aptr = SvPV_const(fromstr, fromlen);
2752 if (DO_UTF8(fromstr)) {
2753 const char *end, *s;
2755 if (!utf8 && !SvUTF8(cat)) {
2756 marked_upgrade(aTHX_ cat, symptr);
2757 lookahead.flags |= FLAG_DO_UTF8;
2758 lookahead.strbeg = symptr->strbeg;
2761 cur = start + SvCUR(cat);
2763 if (howlen == e_star) {
2764 if (utf8) goto string_copy;
2768 end = aptr + fromlen;
2769 fromlen = datumtype == 'Z' ? len-1 : len;
2770 while ((I32) fromlen > 0 && s < end) {
2775 Perl_croak(aTHX_ "Malformed UTF-8 string in pack");
2778 if (datumtype == 'Z') len++;
2784 fromlen = len - fromlen;
2785 if (datumtype == 'Z') fromlen--;
2786 if (howlen == e_star) {
2788 if (datumtype == 'Z') len++;
2790 GROWING(0, cat, start, cur, len);
2791 if (!uni_to_bytes(aTHX_ &aptr, end, cur, fromlen,
2792 datumtype | TYPE_IS_PACK))
2793 Perl_croak(aTHX_ "Perl bug: predicted utf8 length not available");
2797 if (howlen == e_star) {
2799 if (datumtype == 'Z') len++;
2801 if (len <= (I32) fromlen) {
2803 if (datumtype == 'Z' && fromlen > 0) fromlen--;
2805 /* assumes a byte expands to at most UTF8_EXPAND bytes on
2807 expected_length <= from_len*UTF8_EXPAND + (len-from_len) */
2808 GROWING(0, cat, start, cur, fromlen*(UTF8_EXPAND-1)+len);
2810 while (fromlen > 0) {
2811 cur = (char *) uvchr_to_utf8((U8 *) cur, * (U8 *) aptr);
2817 if (howlen == e_star) {
2819 if (datumtype == 'Z') len++;
2821 if (len <= (I32) fromlen) {
2823 if (datumtype == 'Z' && fromlen > 0) fromlen--;
2825 GROWING(0, cat, start, cur, len);
2826 Copy(aptr, cur, fromlen, char);
2830 memset(cur, datumtype == 'A' ? ' ' : '\0', len);
2843 str = SvPV(fromstr, fromlen);
2844 end = str + fromlen;
2845 if (DO_UTF8(fromstr)) {
2847 utf8_flags = ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY;
2849 utf8_source = FALSE;
2850 utf8_flags = 0; /* Unused, but keep compilers happy */
2852 if (howlen == e_star) len = fromlen;
2853 field_len = (len+7)/8;
2854 GROWING(utf8, cat, start, cur, field_len);
2855 if (len > (I32)fromlen) len = fromlen;
2858 if (datumtype == 'B')
2862 NEXT_UNI_VAL(val, cur, str, end, utf8_flags);
2864 } else bits |= *str++ & 1;
2865 if (l & 7) bits <<= 1;
2867 PUSH_BYTE(utf8, cur, bits);
2872 /* datumtype == 'b' */
2876 NEXT_UNI_VAL(val, cur, str, end, utf8_flags);
2877 if (val & 1) bits |= 0x80;
2878 } else if (*str++ & 1)
2880 if (l & 7) bits >>= 1;
2882 PUSH_BYTE(utf8, cur, bits);
2888 if (datumtype == 'B')
2889 bits <<= 7 - (l & 7);
2891 bits >>= 7 - (l & 7);
2892 PUSH_BYTE(utf8, cur, bits);
2895 /* Determine how many chars are left in the requested field */
2897 if (howlen == e_star) field_len = 0;
2898 else field_len -= l;
2899 Zero(cur, field_len, char);
2905 const char *str, *end;
2912 str = SvPV_const(fromstr, fromlen);
2913 end = str + fromlen;
2914 if (DO_UTF8(fromstr)) {
2916 utf8_flags = ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY;
2918 utf8_source = FALSE;
2919 utf8_flags = 0; /* Unused, but keep compilers happy */
2921 if (howlen == e_star) len = fromlen;
2922 field_len = (len+1)/2;
2923 GROWING(utf8, cat, start, cur, field_len);
2924 if (!utf8 && len > (I32)fromlen) len = fromlen;
2927 if (datumtype == 'H')
2931 NEXT_UNI_VAL(val, cur, str, end, utf8_flags);
2932 if (val < 256 && isALPHA(val))
2933 bits |= (val + 9) & 0xf;
2936 } else if (isALPHA(*str))
2937 bits |= (*str++ + 9) & 0xf;
2939 bits |= *str++ & 0xf;
2940 if (l & 1) bits <<= 4;
2942 PUSH_BYTE(utf8, cur, bits);
2950 NEXT_UNI_VAL(val, cur, str, end, utf8_flags);
2951 if (val < 256 && isALPHA(val))
2952 bits |= ((val + 9) & 0xf) << 4;
2954 bits |= (val & 0xf) << 4;
2955 } else if (isALPHA(*str))
2956 bits |= ((*str++ + 9) & 0xf) << 4;
2958 bits |= (*str++ & 0xf) << 4;
2959 if (l & 1) bits >>= 4;
2961 PUSH_BYTE(utf8, cur, bits);
2967 PUSH_BYTE(utf8, cur, bits);
2970 /* Determine how many chars are left in the requested field */
2972 if (howlen == e_star) field_len = 0;
2973 else field_len -= l;
2974 Zero(cur, field_len, char);
2982 aiv = SvIV(fromstr);
2983 if ((-128 > aiv || aiv > 127) &&
2985 Perl_warner(aTHX_ packWARN(WARN_PACK),
2986 "Character in 'c' format wrapped in pack");
2987 PUSH_BYTE(utf8, cur, aiv & 0xff);
2992 utf8 = (symptr->flags & FLAG_DO_UTF8) ? 1 : 0;
2995 GROWING(0, cat, start, cur, len);
2999 aiv = SvIV(fromstr);
3000 if ((0 > aiv || aiv > 0xff) &&
3002 Perl_warner(aTHX_ packWARN(WARN_PACK),
3003 "Character in 'C' format wrapped in pack");
3004 *cur++ = aiv & 0xff;
3009 U8 in_bytes = IN_BYTES;
3011 end = start+SvLEN(cat)-1;
3012 if (utf8) end -= UTF8_MAXLEN-1;
3016 auv = SvUV(fromstr);
3017 if (in_bytes) auv = auv % 0x100;
3022 SvCUR_set(cat, cur - start);
3024 GROWING(0, cat, start, cur, len+UTF8_MAXLEN);
3025 end = start+SvLEN(cat)-UTF8_MAXLEN;
3027 cur = (char *) uvuni_to_utf8_flags((U8 *) cur,
3030 0 : UNICODE_ALLOW_ANY);
3035 SvCUR_set(cat, cur - start);
3036 marked_upgrade(aTHX_ cat, symptr);
3037 lookahead.flags |= FLAG_DO_UTF8;
3038 lookahead.strbeg = symptr->strbeg;
3041 cur = start + SvCUR(cat);
3042 end = start+SvLEN(cat)-UTF8_MAXLEN;
3045 if (ckWARN(WARN_PACK))
3046 Perl_warner(aTHX_ packWARN(WARN_PACK),
3047 "Character in 'W' format wrapped in pack");
3052 SvCUR_set(cat, cur - start);
3053 GROWING(0, cat, start, cur, len+1);
3054 end = start+SvLEN(cat)-1;
3056 *(U8 *) cur++ = (U8)auv;
3065 if (!(symptr->flags & FLAG_DO_UTF8)) {
3066 marked_upgrade(aTHX_ cat, symptr);
3067 lookahead.flags |= FLAG_DO_UTF8;
3068 lookahead.strbeg = symptr->strbeg;
3074 end = start+SvLEN(cat);
3075 if (!utf8) end -= UTF8_MAXLEN;
3079 auv = SvUV(fromstr);
3081 U8 buffer[UTF8_MAXLEN], *endb;
3082 endb = uvuni_to_utf8_flags(buffer, auv,
3084 0 : UNICODE_ALLOW_ANY);
3085 if (cur+(endb-buffer)*UTF8_EXPAND >= end) {
3087 SvCUR_set(cat, cur - start);
3088 GROWING(0, cat, start, cur,
3089 len+(endb-buffer)*UTF8_EXPAND);
3090 end = start+SvLEN(cat);
3092 bytes_to_uni(aTHX_ buffer, endb-buffer, &cur);
3096 SvCUR_set(cat, cur - start);
3097 GROWING(0, cat, start, cur, len+UTF8_MAXLEN);
3098 end = start+SvLEN(cat)-UTF8_MAXLEN;
3100 cur = (char *) uvuni_to_utf8_flags((U8 *) cur, auv,
3102 0 : UNICODE_ALLOW_ANY);
3107 /* Float and double added by gnb@melba.bby.oz.au 22/11/89 */
3113 anv = SvNV(fromstr);
3115 /* VOS does not automatically map a floating-point overflow
3116 during conversion from double to float into infinity, so we
3117 do it by hand. This code should either be generalized for
3118 any OS that needs it, or removed if and when VOS implements
3119 posix-976 (suggestion to support mapping to infinity).
3120 Paul.Green@stratus.com 02-04-02. */
3122 afloat = _float_constants[0]; /* single prec. inf. */
3123 else if (anv < -FLT_MAX)
3124 afloat = _float_constants[0]; /* single prec. inf. */
3125 else afloat = (float) anv;
3127 # if defined(VMS) && !defined(__IEEE_FP)
3128 /* IEEE fp overflow shenanigans are unavailable on VAX and optional
3129 * on Alpha; fake it if we don't have them.
3133 else if (anv < -FLT_MAX)
3135 else afloat = (float)anv;
3137 afloat = (float)anv;
3139 #endif /* __VOS__ */
3140 DO_BO_PACK_N(afloat, float);
3141 PUSH_VAR(utf8, cur, afloat);
3149 anv = SvNV(fromstr);
3151 /* VOS does not automatically map a floating-point overflow
3152 during conversion from long double to double into infinity,
3153 so we do it by hand. This code should either be generalized
3154 for any OS that needs it, or removed if and when VOS
3155 implements posix-976 (suggestion to support mapping to
3156 infinity). Paul.Green@stratus.com 02-04-02. */
3158 adouble = _double_constants[0]; /* double prec. inf. */
3159 else if (anv < -DBL_MAX)
3160 adouble = _double_constants[0]; /* double prec. inf. */
3161 else adouble = (double) anv;
3163 # if defined(VMS) && !defined(__IEEE_FP)
3164 /* IEEE fp overflow shenanigans are unavailable on VAX and optional
3165 * on Alpha; fake it if we don't have them.
3169 else if (anv < -DBL_MAX)
3171 else adouble = (double)anv;
3173 adouble = (double)anv;
3175 #endif /* __VOS__ */
3176 DO_BO_PACK_N(adouble, double);
3177 PUSH_VAR(utf8, cur, adouble);
3182 Zero(&anv, 1, NV); /* can be long double with unused bits */
3185 anv = SvNV(fromstr);
3186 DO_BO_PACK_N(anv, NV);
3187 PUSH_VAR(utf8, cur, anv);
3191 #if defined(HAS_LONG_DOUBLE) && defined(USE_LONG_DOUBLE)
3193 long double aldouble;
3194 /* long doubles can have unused bits, which may be nonzero */
3195 Zero(&aldouble, 1, long double);
3198 aldouble = (long double)SvNV(fromstr);
3199 DO_BO_PACK_N(aldouble, long double);
3200 PUSH_VAR(utf8, cur, aldouble);
3205 #ifdef PERL_PACK_CAN_SHRIEKSIGN
3206 case 'n' | TYPE_IS_SHRIEKING:
3212 ai16 = (I16)SvIV(fromstr);
3214 ai16 = PerlSock_htons(ai16);
3216 PUSH16(utf8, cur, &ai16);
3219 #ifdef PERL_PACK_CAN_SHRIEKSIGN
3220 case 'v' | TYPE_IS_SHRIEKING:
3226 ai16 = (I16)SvIV(fromstr);
3230 PUSH16(utf8, cur, &ai16);
3233 case 'S' | TYPE_IS_SHRIEKING:
3234 #if SHORTSIZE != SIZE16
3236 unsigned short aushort;
3238 aushort = SvUV(fromstr);
3239 DO_BO_PACK(aushort, s);
3240 PUSH_VAR(utf8, cur, aushort);
3250 au16 = (U16)SvUV(fromstr);
3251 DO_BO_PACK(au16, 16);
3252 PUSH16(utf8, cur, &au16);
3255 case 's' | TYPE_IS_SHRIEKING:
3256 #if SHORTSIZE != SIZE16
3260 ashort = SvIV(fromstr);
3261 DO_BO_PACK(ashort, s);
3262 PUSH_VAR(utf8, cur, ashort);
3272 ai16 = (I16)SvIV(fromstr);
3273 DO_BO_PACK(ai16, 16);
3274 PUSH16(utf8, cur, &ai16);
3278 case 'I' | TYPE_IS_SHRIEKING:
3282 auint = SvUV(fromstr);
3283 DO_BO_PACK(auint, i);
3284 PUSH_VAR(utf8, cur, auint);
3291 aiv = SvIV(fromstr);
3292 #if IVSIZE == INTSIZE
3294 #elif IVSIZE == LONGSIZE
3296 #elif defined(HAS_QUAD) && IVSIZE == U64SIZE
3297 DO_BO_PACK(aiv, 64);
3299 Perl_croak(aTHX_ "'j' not supported on this platform");
3301 PUSH_VAR(utf8, cur, aiv);
3308 auv = SvUV(fromstr);
3309 #if UVSIZE == INTSIZE
3311 #elif UVSIZE == LONGSIZE
3313 #elif defined(HAS_QUAD) && UVSIZE == U64SIZE
3314 DO_BO_PACK(auv, 64);
3316 Perl_croak(aTHX_ "'J' not supported on this platform");
3318 PUSH_VAR(utf8, cur, auv);
3325 anv = SvNV(fromstr);
3329 SvCUR_set(cat, cur - start);
3330 Perl_croak(aTHX_ "Cannot compress negative numbers in pack");
3333 /* 0xFFFFFFFFFFFFFFFF may cast to 18446744073709551616.0,
3334 which is == UV_MAX_P1. IOK is fine (instead of UV_only), as
3335 any negative IVs will have already been got by the croak()
3336 above. IOK is untrue for fractions, so we test them
3337 against UV_MAX_P1. */
3338 if (SvIOK(fromstr) || anv < UV_MAX_P1) {
3339 char buf[(sizeof(UV)*CHAR_BIT)/7+1];
3340 char *in = buf + sizeof(buf);
3341 UV auv = SvUV(fromstr);
3344 *--in = (char)((auv & 0x7f) | 0x80);
3347 buf[sizeof(buf) - 1] &= 0x7f; /* clear continue bit */
3348 PUSH_GROWING_BYTES(utf8, cat, start, cur,
3349 in, (buf + sizeof(buf)) - in);
3350 } else if (SvPOKp(fromstr))
3352 else if (SvNOKp(fromstr)) {
3353 /* 10**NV_MAX_10_EXP is the largest power of 10
3354 so 10**(NV_MAX_10_EXP+1) is definately unrepresentable
3355 given 10**(NV_MAX_10_EXP+1) == 128 ** x solve for x:
3356 x = (NV_MAX_10_EXP+1) * log (10) / log (128)
3357 And with that many bytes only Inf can overflow.
3358 Some C compilers are strict about integral constant
3359 expressions so we conservatively divide by a slightly
3360 smaller integer instead of multiplying by the exact
3361 floating-point value.
3363 #ifdef NV_MAX_10_EXP
3364 /* char buf[1 + (int)((NV_MAX_10_EXP + 1) * 0.47456)]; -- invalid C */
3365 char buf[1 + (int)((NV_MAX_10_EXP + 1) / 2)]; /* valid C */
3367 /* char buf[1 + (int)((308 + 1) * 0.47456)]; -- invalid C */
3368 char buf[1 + (int)((308 + 1) / 2)]; /* valid C */
3370 char *in = buf + sizeof(buf);
3372 anv = Perl_floor(anv);
3374 NV next = Perl_floor(anv / 128);
3375 if (in <= buf) /* this cannot happen ;-) */
3376 Perl_croak(aTHX_ "Cannot compress integer in pack");
3377 *--in = (unsigned char)(anv - (next * 128)) | 0x80;
3380 buf[sizeof(buf) - 1] &= 0x7f; /* clear continue bit */
3381 PUSH_GROWING_BYTES(utf8, cat, start, cur,
3382 in, (buf + sizeof(buf)) - in);
3384 char *from, *result, *in;
3390 /* Copy string and check for compliance */
3391 from = SvPV(fromstr, len);
3392 if ((norm = is_an_int(from, len)) == NULL)
3393 Perl_croak(aTHX_ "Can only compress unsigned integers in pack");
3395 New('w', result, len, char);
3398 while (!done) *--in = div128(norm, &done) | 0x80;
3399 result[len - 1] &= 0x7F; /* clear continue bit */
3400 PUSH_GROWING_BYTES(utf8, cat, start, cur,
3401 in, (result + len) - in);
3403 SvREFCNT_dec(norm); /* free norm */
3408 case 'i' | TYPE_IS_SHRIEKING:
3412 aint = SvIV(fromstr);
3413 DO_BO_PACK(aint, i);
3414 PUSH_VAR(utf8, cur, aint);
3417 #ifdef PERL_PACK_CAN_SHRIEKSIGN
3418 case 'N' | TYPE_IS_SHRIEKING:
3424 au32 = SvUV(fromstr);
3426 au32 = PerlSock_htonl(au32);
3428 PUSH32(utf8, cur, &au32);
3431 #ifdef PERL_PACK_CAN_SHRIEKSIGN
3432 case 'V' | TYPE_IS_SHRIEKING:
3438 au32 = SvUV(fromstr);
3442 PUSH32(utf8, cur, &au32);
3445 case 'L' | TYPE_IS_SHRIEKING:
3446 #if LONGSIZE != SIZE32
3448 unsigned long aulong;
3450 aulong = SvUV(fromstr);
3451 DO_BO_PACK(aulong, l);
3452 PUSH_VAR(utf8, cur, aulong);
3462 au32 = SvUV(fromstr);
3463 DO_BO_PACK(au32, 32);
3464 PUSH32(utf8, cur, &au32);
3467 case 'l' | TYPE_IS_SHRIEKING:
3468 #if LONGSIZE != SIZE32
3472 along = SvIV(fromstr);
3473 DO_BO_PACK(along, l);
3474 PUSH_VAR(utf8, cur, along);
3484 ai32 = SvIV(fromstr);
3485 DO_BO_PACK(ai32, 32);
3486 PUSH32(utf8, cur, &ai32);
3494 auquad = (Uquad_t) SvUV(fromstr);
3495 DO_BO_PACK(auquad, 64);
3496 PUSH_VAR(utf8, cur, auquad);
3503 aquad = (Quad_t)SvIV(fromstr);
3504 DO_BO_PACK(aquad, 64);
3505 PUSH_VAR(utf8, cur, aquad);
3508 #endif /* HAS_QUAD */
3510 len = 1; /* assume SV is correct length */
3511 GROWING(utf8, cat, start, cur, sizeof(char *));
3518 SvGETMAGIC(fromstr);
3519 if (!SvOK(fromstr)) aptr = NULL;
3522 /* XXX better yet, could spirit away the string to
3523 * a safe spot and hang on to it until the result
3524 * of pack() (and all copies of the result) are
3527 if (ckWARN(WARN_PACK) &&
3528 (SvTEMP(fromstr) || (SvPADTMP(fromstr) &&
3529 !SvREADONLY(fromstr)))) {
3530 Perl_warner(aTHX_ packWARN(WARN_PACK),
3531 "Attempt to pack pointer to temporary value");
3533 if (SvPOK(fromstr) || SvNIOK(fromstr))
3534 aptr = SvPV_nomg_const(fromstr, n_a);
3536 aptr = SvPV_force_flags(fromstr, n_a, 0);
3538 DO_BO_PACK_PC(aptr);
3539 PUSH_VAR(utf8, cur, aptr);
3543 const char *aptr, *aend;
3547 if (len <= 2) len = 45;
3548 else len = len / 3 * 3;
3550 Perl_warner(aTHX_ packWARN(WARN_PACK),
3551 "Field too wide in 'u' format in pack");
3554 aptr = SvPV_const(fromstr, fromlen);
3555 from_utf8 = DO_UTF8(fromstr);
3557 aend = aptr + fromlen;
3558 fromlen = sv_len_utf8(fromstr);
3559 } else aend = NULL; /* Unused, but keep compilers happy */
3560 GROWING(utf8, cat, start, cur, (fromlen+2) / 3 * 4 + (fromlen+len-1)/len * 2);
3561 while (fromlen > 0) {
3564 U8 hunk[1+63/3*4+1];
3566 if ((I32)fromlen > len)
3572 if (!uni_to_bytes(aTHX_ &aptr, aend, buffer, todo,
3573 'u' | TYPE_IS_PACK)) {
3575 SvCUR_set(cat, cur - start);
3576 Perl_croak(aTHX_ "Assertion: string is shorter than advertised");
3578 end = doencodes(hunk, buffer, todo);
3580 end = doencodes(hunk, aptr, todo);
3583 PUSH_BYTES(utf8, cur, hunk, end-hunk);
3590 SvCUR_set(cat, cur - start);
3592 *symptr = lookahead;
3601 dSP; dMARK; dORIGMARK; dTARGET;
3602 register SV *cat = TARG;
3604 register const char *pat = SvPVx_const(*++MARK, fromlen);
3605 register const char *patend = pat + fromlen;
3608 sv_setpvn(cat, "", 0);
3611 packlist(cat, pat, patend, MARK, SP + 1);
3621 * c-indentation-style: bsd
3623 * indent-tabs-mode: t
3626 * ex: set ts=8 sts=4 sw=4 noet: