This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Mention Hook::LexWrap in perl5220delta
[perl5.git] / pp_pack.c
CommitLineData
a6ec74c1
JH
1/* pp_pack.c
2 *
1129b882
NC
3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 * 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008 by Larry Wall and others
a6ec74c1
JH
5 *
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.
8 *
9 */
10
d31a8517
AT
11/*
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,
16 * some salt.
4ac71550
TC
17 *
18 * [p.653 of _The Lord of the Rings_, IV/iv: "Of Herbs and Stewed Rabbit"]
d31a8517
AT
19 */
20
166f8a29
DM
21/* This file contains pp ("push/pop") functions that
22 * execute the opcodes that make up a perl program. A typical pp function
23 * expects to find its arguments on the stack, and usually pushes its
24 * results onto the stack, hence the 'pp' terminology. Each OP structure
25 * contains a pointer to the relevant pp_foo() function.
26 *
27 * This particular file just contains pp_pack() and pp_unpack(). See the
28 * other pp*.c files for the rest of the pp_ functions.
29 */
30
a6ec74c1
JH
31#include "EXTERN.h"
32#define PERL_IN_PP_PACK_C
33#include "perl.h"
34
f7fe979e
AL
35/* Types used by pack/unpack */
36typedef enum {
37 e_no_len, /* no length */
38 e_number, /* number, [] */
39 e_star /* asterisk */
40} howlen_t;
41
42typedef struct tempsym {
43 const char* patptr; /* current template char */
44 const char* patend; /* one after last char */
45 const char* grpbeg; /* 1st char of ()-group */
46 const char* grpend; /* end of ()-group */
47 I32 code; /* template code (!<>) */
48 I32 length; /* length/repeat count */
49 howlen_t howlen; /* how length is given */
50 int level; /* () nesting level */
51 U32 flags; /* /=4, comma=2, pack=1 */
52 /* and group modifiers */
53 STRLEN strbeg; /* offset of group start */
54 struct tempsym *previous; /* previous group */
55} tempsym_t;
56
57#define TEMPSYM_INIT(symptr, p, e, f) \
58 STMT_START { \
59 (symptr)->patptr = (p); \
60 (symptr)->patend = (e); \
61 (symptr)->grpbeg = NULL; \
62 (symptr)->grpend = NULL; \
63 (symptr)->grpend = NULL; \
64 (symptr)->code = 0; \
65 (symptr)->length = 0; \
10edeb5d 66 (symptr)->howlen = e_no_len; \
f7fe979e
AL
67 (symptr)->level = 0; \
68 (symptr)->flags = (f); \
69 (symptr)->strbeg = 0; \
70 (symptr)->previous = NULL; \
71 } STMT_END
72
275663fa
TC
73typedef union {
74 NV nv;
75 U8 bytes[sizeof(NV)];
76} NV_bytes;
77
78#if defined(HAS_LONG_DOUBLE) && defined(USE_LONG_DOUBLE)
79typedef union {
80 long double ld;
81 U8 bytes[sizeof(long double)];
82} ld_bytes;
83#endif
84
f337b084
TH
85#ifndef CHAR_BIT
86# define CHAR_BIT 8
7212898e 87#endif
3473cf63
RGS
88/* Maximum number of bytes to which a byte can grow due to upgrade */
89#define UTF8_EXPAND 2
7212898e 90
a6ec74c1 91/*
a6ec74c1
JH
92 * Offset for integer pack/unpack.
93 *
94 * On architectures where I16 and I32 aren't really 16 and 32 bits,
95 * which for now are all Crays, pack and unpack have to play games.
96 */
97
98/*
99 * These values are required for portability of pack() output.
100 * If they're not right on your machine, then pack() and unpack()
101 * wouldn't work right anyway; you'll need to apply the Cray hack.
102 * (I'd like to check them with #if, but you can't use sizeof() in
103 * the preprocessor.) --???
104 */
105/*
106 The appropriate SHORTSIZE, INTSIZE, LONGSIZE, and LONGLONGSIZE
107 defines are now in config.h. --Andy Dougherty April 1998
108 */
109#define SIZE16 2
110#define SIZE32 4
111
112/* CROSSCOMPILE and MULTIARCH are going to affect pp_pack() and pp_unpack().
113 --jhi Feb 1999 */
114
1109a392
MHM
115#if U16SIZE > SIZE16 || U32SIZE > SIZE32
116# if BYTEORDER == 0x1234 || BYTEORDER == 0x12345678 /* little-endian */
08ca2aa3
TH
117# define OFF16(p) ((char*)(p))
118# define OFF32(p) ((char*)(p))
a6ec74c1 119# else
1109a392 120# if BYTEORDER == 0x4321 || BYTEORDER == 0x87654321 /* big-endian */
a6ec74c1
JH
121# define OFF16(p) ((char*)(p) + (sizeof(U16) - SIZE16))
122# define OFF32(p) ((char*)(p) + (sizeof(U32) - SIZE32))
123# else
08ca2aa3 124 ++++ bad cray byte order
a6ec74c1
JH
125# endif
126# endif
a6ec74c1 127#else
08ca2aa3
TH
128# define OFF16(p) ((char *) (p))
129# define OFF32(p) ((char *) (p))
a6ec74c1
JH
130#endif
131
3a88beaa
NC
132#define PUSH16(utf8, cur, p, needs_swap) \
133 PUSH_BYTES(utf8, cur, OFF16(p), SIZE16, needs_swap)
134#define PUSH32(utf8, cur, p, needs_swap) \
135 PUSH_BYTES(utf8, cur, OFF32(p), SIZE32, needs_swap)
08ca2aa3 136
a1219b5e
NC
137#if BYTEORDER == 0x4321 || BYTEORDER == 0x87654321 /* big-endian */
138# define NEEDS_SWAP(d) (TYPE_ENDIANNESS(d) == TYPE_IS_LITTLE_ENDIAN)
139#elif BYTEORDER == 0x1234 || BYTEORDER == 0x12345678 /* little-endian */
140# define NEEDS_SWAP(d) (TYPE_ENDIANNESS(d) == TYPE_IS_BIG_ENDIAN)
141#else
142# error "Unsupported byteorder"
20aa3a38
NC
143 /* Need to add code here to re-instate mixed endian support.
144 NEEDS_SWAP would need to hold a flag indicating which action to
145 take, and S_reverse_copy and the code in uni_to_bytes would need
146 logic adding to deal with any mixed-endian transformations needed.
147 */
a1219b5e
NC
148#endif
149
08ca2aa3 150/* Only to be used inside a loop (see the break) */
aaec8192 151#define SHIFT_BYTES(utf8, s, strend, buf, len, datumtype, needs_swap) \
08ca2aa3 152STMT_START { \
228e69a7 153 if (UNLIKELY(utf8)) { \
f337b084 154 if (!uni_to_bytes(aTHX_ &s, strend, \
275663fa 155 (char *) (buf), len, datumtype)) break; \
08ca2aa3 156 } else { \
228e69a7 157 if (UNLIKELY(needs_swap)) \
20aa3a38
NC
158 S_reverse_copy(s, (char *) (buf), len); \
159 else \
160 Copy(s, (char *) (buf), len, char); \
275663fa 161 s += len; \
08ca2aa3 162 } \
08ca2aa3
TH
163} STMT_END
164
aaec8192
NC
165#define SHIFT16(utf8, s, strend, p, datumtype, needs_swap) \
166 SHIFT_BYTES(utf8, s, strend, OFF16(p), SIZE16, datumtype, needs_swap)
7285e3f4 167
aaec8192
NC
168#define SHIFT32(utf8, s, strend, p, datumtype, needs_swap) \
169 SHIFT_BYTES(utf8, s, strend, OFF32(p), SIZE32, datumtype, needs_swap)
7285e3f4 170
aaec8192
NC
171#define SHIFT_VAR(utf8, s, strend, var, datumtype, needs_swap) \
172 SHIFT_BYTES(utf8, s, strend, &(var), sizeof(var), datumtype, needs_swap)
275663fa 173
3a88beaa
NC
174#define PUSH_VAR(utf8, aptr, var, needs_swap) \
175 PUSH_BYTES(utf8, aptr, &(var), sizeof(var), needs_swap)
f337b084 176
49704364
WL
177/* Avoid stack overflow due to pathological templates. 100 should be plenty. */
178#define MAX_SUB_TEMPLATE_LEVEL 100
179
66c611c5 180/* flags (note that type modifiers can also be used as flags!) */
f337b084
TH
181#define FLAG_WAS_UTF8 0x40
182#define FLAG_PARSE_UTF8 0x20 /* Parse as utf8 */
49704364 183#define FLAG_UNPACK_ONLY_ONE 0x10
f337b084 184#define FLAG_DO_UTF8 0x08 /* The underlying string is utf8 */
49704364
WL
185#define FLAG_SLASH 0x04
186#define FLAG_COMMA 0x02
187#define FLAG_PACK 0x01
188
a6ec74c1
JH
189STATIC SV *
190S_mul128(pTHX_ SV *sv, U8 m)
191{
192 STRLEN len;
193 char *s = SvPV(sv, len);
194 char *t;
a6ec74c1 195
7918f24d
NC
196 PERL_ARGS_ASSERT_MUL128;
197
a6ec74c1 198 if (!strnEQ(s, "0000", 4)) { /* need to grow sv */
396482e1 199 SV * const tmpNew = newSVpvs("0000000000");
a6ec74c1
JH
200
201 sv_catsv(tmpNew, sv);
202 SvREFCNT_dec(sv); /* free old sv */
203 sv = tmpNew;
204 s = SvPV(sv, len);
205 }
206 t = s + len - 1;
207 while (!*t) /* trailing '\0'? */
208 t--;
209 while (t > s) {
f7fe979e 210 const U32 i = ((*t - '0') << 7) + m;
eb160463
GS
211 *(t--) = '0' + (char)(i % 10);
212 m = (char)(i / 10);
a6ec74c1
JH
213 }
214 return (sv);
215}
216
217/* Explosives and implosives. */
218
219#if 'I' == 73 && 'J' == 74
220/* On an ASCII/ISO kind of system */
221#define ISUUCHAR(ch) ((ch) >= ' ' && (ch) < 'a')
222#else
223/*
224 Some other sort of character set - use memchr() so we don't match
225 the null byte.
226 */
227#define ISUUCHAR(ch) (memchr(PL_uuemap, (ch), sizeof(PL_uuemap)-1) || (ch) == ' ')
228#endif
229
66c611c5 230/* type modifiers */
62f95557 231#define TYPE_IS_SHRIEKING 0x100
1109a392
MHM
232#define TYPE_IS_BIG_ENDIAN 0x200
233#define TYPE_IS_LITTLE_ENDIAN 0x400
f337b084 234#define TYPE_IS_PACK 0x800
1109a392 235#define TYPE_ENDIANNESS_MASK (TYPE_IS_BIG_ENDIAN|TYPE_IS_LITTLE_ENDIAN)
66c611c5 236#define TYPE_MODIFIERS(t) ((t) & ~0xFF)
1109a392
MHM
237#define TYPE_NO_MODIFIERS(t) ((t) & 0xFF)
238
7212898e
NC
239# define TYPE_ENDIANNESS(t) ((t) & TYPE_ENDIANNESS_MASK)
240# define TYPE_NO_ENDIANNESS(t) ((t) & ~TYPE_ENDIANNESS_MASK)
241
242# define ENDIANNESS_ALLOWED_TYPES "sSiIlLqQjJfFdDpP("
243
78d46eaa 244#define PACK_SIZE_CANNOT_CSUM 0x80
f337b084 245#define PACK_SIZE_UNPREDICTABLE 0x40 /* Not a fixed size element */
78d46eaa
NC
246#define PACK_SIZE_MASK 0x3F
247
298bc19c 248#include "packsizetables.c"
78d46eaa 249
20aa3a38
NC
250static void
251S_reverse_copy(const char *src, char *dest, STRLEN len)
252{
253 dest += len;
254 while (len--)
255 *--dest = *src++;
256}
257
08ca2aa3 258STATIC U8
f7fe979e 259uni_to_byte(pTHX_ const char **s, const char *end, I32 datumtype)
08ca2aa3 260{
08ca2aa3 261 STRLEN retlen;
0bcc34c2 262 UV val = utf8n_to_uvchr((U8 *) *s, end-*s, &retlen,
f337b084 263 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
486ec47a 264 /* We try to process malformed UTF-8 as much as possible (preferably with
08ca2aa3
TH
265 warnings), but these two mean we make no progress in the string and
266 might enter an infinite loop */
267 if (retlen == (STRLEN) -1 || retlen == 0)
f337b084
TH
268 Perl_croak(aTHX_ "Malformed UTF-8 string in '%c' format in unpack",
269 (int) TYPE_NO_MODIFIERS(datumtype));
08ca2aa3 270 if (val >= 0x100) {
a2a5de95
NC
271 Perl_ck_warner(aTHX_ packWARN(WARN_UNPACK),
272 "Character in '%c' format wrapped in unpack",
273 (int) TYPE_NO_MODIFIERS(datumtype));
08ca2aa3
TH
274 val &= 0xff;
275 }
276 *s += retlen;
fe2774ed 277 return (U8)val;
08ca2aa3
TH
278}
279
f337b084
TH
280#define SHIFT_BYTE(utf8, s, strend, datumtype) ((utf8) ? \
281 uni_to_byte(aTHX_ &(s), (strend), (datumtype)) : \
08ca2aa3
TH
282 *(U8 *)(s)++)
283
284STATIC bool
f7fe979e 285uni_to_bytes(pTHX_ const char **s, const char *end, const char *buf, int buf_len, I32 datumtype)
08ca2aa3
TH
286{
287 UV val;
288 STRLEN retlen;
f7fe979e 289 const char *from = *s;
08ca2aa3 290 int bad = 0;
f7fe979e 291 const U32 flags = ckWARN(WARN_UTF8) ?
08ca2aa3 292 UTF8_CHECK_ONLY : (UTF8_CHECK_ONLY | UTF8_ALLOW_ANY);
20aa3a38
NC
293 const bool needs_swap = NEEDS_SWAP(datumtype);
294
228e69a7 295 if (UNLIKELY(needs_swap))
20aa3a38
NC
296 buf += buf_len;
297
08ca2aa3
TH
298 for (;buf_len > 0; buf_len--) {
299 if (from >= end) return FALSE;
f337b084 300 val = utf8n_to_uvchr((U8 *) from, end-from, &retlen, flags);
08ca2aa3
TH
301 if (retlen == (STRLEN) -1 || retlen == 0) {
302 from += UTF8SKIP(from);
303 bad |= 1;
304 } else from += retlen;
305 if (val >= 0x100) {
306 bad |= 2;
307 val &= 0xff;
308 }
228e69a7 309 if (UNLIKELY(needs_swap))
20aa3a38
NC
310 *(U8 *)--buf = (U8)val;
311 else
312 *(U8 *)buf++ = (U8)val;
08ca2aa3
TH
313 }
314 /* We have enough characters for the buffer. Did we have problems ? */
315 if (bad) {
316 if (bad & 1) {
317 /* Rewalk the string fragment while warning */
f7fe979e 318 const char *ptr;
9e27e96a 319 const int flags = ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY;
08ca2aa3
TH
320 for (ptr = *s; ptr < from; ptr += UTF8SKIP(ptr)) {
321 if (ptr >= end) break;
c80e42f3 322 utf8n_to_uvchr((U8 *) ptr, end-ptr, &retlen, flags);
08ca2aa3
TH
323 }
324 if (from > end) from = end;
325 }
a2a5de95
NC
326 if ((bad & 2))
327 Perl_ck_warner(aTHX_ packWARN(datumtype & TYPE_IS_PACK ?
f337b084 328 WARN_PACK : WARN_UNPACK),
a2a5de95
NC
329 "Character(s) in '%c' format wrapped in %s",
330 (int) TYPE_NO_MODIFIERS(datumtype),
331 datumtype & TYPE_IS_PACK ? "pack" : "unpack");
08ca2aa3
TH
332 }
333 *s = from;
334 return TRUE;
335}
336
337STATIC bool
f7fe979e 338next_uni_uu(pTHX_ const char **s, const char *end, I32 *out)
08ca2aa3 339{
08ca2aa3 340 STRLEN retlen;
0bcc34c2 341 const UV val = utf8n_to_uvchr((U8 *) *s, end-*s, &retlen, UTF8_CHECK_ONLY);
08ca2aa3
TH
342 if (val >= 0x100 || !ISUUCHAR(val) ||
343 retlen == (STRLEN) -1 || retlen == 0) {
344 *out = 0;
345 return FALSE;
346 }
347 *out = PL_uudmap[val] & 077;
f337b084 348 *s += retlen;
08ca2aa3
TH
349 return TRUE;
350}
78d46eaa 351
64844641 352STATIC char *
3a88beaa 353S_bytes_to_uni(const U8 *start, STRLEN len, char *dest, const bool needs_swap) {
7918f24d
NC
354 PERL_ARGS_ASSERT_BYTES_TO_UNI;
355
228e69a7 356 if (UNLIKELY(needs_swap)) {
3a88beaa
NC
357 const U8 *p = start + len;
358 while (p-- > start) {
55d09dc8 359 append_utf8_from_native_byte(*p, (U8 **) & dest);
3a88beaa
NC
360 }
361 } else {
362 const U8 * const end = start + len;
363 while (start < end) {
55d09dc8 364 append_utf8_from_native_byte(*start, (U8 **) & dest);
3a88beaa
NC
365 start++;
366 }
f337b084 367 }
64844641 368 return dest;
f337b084
TH
369}
370
3a88beaa 371#define PUSH_BYTES(utf8, cur, buf, len, needs_swap) \
230e1fce 372STMT_START { \
228e69a7 373 if (UNLIKELY(utf8)) \
3a88beaa 374 (cur) = S_bytes_to_uni((U8 *) buf, len, (cur), needs_swap); \
230e1fce 375 else { \
228e69a7 376 if (UNLIKELY(needs_swap)) \
3a88beaa
NC
377 S_reverse_copy((char *)(buf), cur, len); \
378 else \
379 Copy(buf, cur, len, char); \
230e1fce
NC
380 (cur) += (len); \
381 } \
f337b084
TH
382} STMT_END
383
384#define GROWING(utf8, cat, start, cur, in_len) \
385STMT_START { \
386 STRLEN glen = (in_len); \
3473cf63 387 if (utf8) glen *= UTF8_EXPAND; \
f337b084 388 if ((cur) + glen >= (start) + SvLEN(cat)) { \
0bd48802 389 (start) = sv_exp_grow(cat, glen); \
f337b084
TH
390 (cur) = (start) + SvCUR(cat); \
391 } \
392} STMT_END
393
394#define PUSH_GROWING_BYTES(utf8, cat, start, cur, buf, in_len) \
395STMT_START { \
f7fe979e 396 const STRLEN glen = (in_len); \
f337b084 397 STRLEN gl = glen; \
3473cf63 398 if (utf8) gl *= UTF8_EXPAND; \
f337b084
TH
399 if ((cur) + gl >= (start) + SvLEN(cat)) { \
400 *cur = '\0'; \
b162af07 401 SvCUR_set((cat), (cur) - (start)); \
0bd48802 402 (start) = sv_exp_grow(cat, gl); \
f337b084
TH
403 (cur) = (start) + SvCUR(cat); \
404 } \
3a88beaa 405 PUSH_BYTES(utf8, cur, buf, glen, 0); \
f337b084
TH
406} STMT_END
407
408#define PUSH_BYTE(utf8, s, byte) \
409STMT_START { \
410 if (utf8) { \
f7fe979e 411 const U8 au8 = (byte); \
3a88beaa 412 (s) = S_bytes_to_uni(&au8, 1, (s), 0); \
f337b084
TH
413 } else *(U8 *)(s)++ = (byte); \
414} STMT_END
415
416/* Only to be used inside a loop (see the break) */
417#define NEXT_UNI_VAL(val, cur, str, end, utf8_flags) \
418STMT_START { \
419 STRLEN retlen; \
420 if (str >= end) break; \
421 val = utf8n_to_uvchr((U8 *) str, end-str, &retlen, utf8_flags); \
422 if (retlen == (STRLEN) -1 || retlen == 0) { \
423 *cur = '\0'; \
424 Perl_croak(aTHX_ "Malformed UTF-8 string in pack"); \
425 } \
426 str += retlen; \
427} STMT_END
428
f7fe979e
AL
429static const char *_action( const tempsym_t* symptr )
430{
10edeb5d 431 return (const char *)(( symptr->flags & FLAG_PACK ) ? "pack" : "unpack");
f7fe979e
AL
432}
433
206947d2 434/* Returns the sizeof() struct described by pat */
028d1f6d 435STATIC I32
f337b084 436S_measure_struct(pTHX_ tempsym_t* symptr)
206947d2 437{
f337b084 438 I32 total = 0;
206947d2 439
7918f24d
NC
440 PERL_ARGS_ASSERT_MEASURE_STRUCT;
441
49704364 442 while (next_symbol(symptr)) {
f337b084 443 I32 len;
f7fe979e 444 int size;
f337b084
TH
445
446 switch (symptr->howlen) {
fc241834 447 case e_star:
49704364 448 Perl_croak(aTHX_ "Within []-length '*' not allowed in %s",
f7fe979e 449 _action( symptr ) );
81d52ecd 450
f337b084
TH
451 default:
452 /* e_no_len and e_number */
453 len = symptr->length;
454 break;
49704364
WL
455 }
456
a7a3cfaa 457 size = packprops[TYPE_NO_ENDIANNESS(symptr->code)] & PACK_SIZE_MASK;
80a13697 458 if (!size) {
f7fe979e 459 int star;
80a13697
NC
460 /* endianness doesn't influence the size of a type */
461 switch(TYPE_NO_ENDIANNESS(symptr->code)) {
462 default:
463 Perl_croak(aTHX_ "Invalid type '%c' in %s",
464 (int)TYPE_NO_MODIFIERS(symptr->code),
f7fe979e 465 _action( symptr ) );
28be1210
TH
466 case '.' | TYPE_IS_SHRIEKING:
467 case '@' | TYPE_IS_SHRIEKING:
80a13697 468 case '@':
28be1210 469 case '.':
80a13697
NC
470 case '/':
471 case 'U': /* XXXX Is it correct? */
472 case 'w':
473 case 'u':
474 Perl_croak(aTHX_ "Within []-length '%c' not allowed in %s",
28be1210 475 (int) TYPE_NO_MODIFIERS(symptr->code),
f7fe979e 476 _action( symptr ) );
80a13697
NC
477 case '%':
478 size = 0;
479 break;
480 case '(':
fc241834
RGS
481 {
482 tempsym_t savsym = *symptr;
483 symptr->patptr = savsym.grpbeg;
484 symptr->patend = savsym.grpend;
485 /* XXXX Theoretically, we need to measure many times at
486 different positions, since the subexpression may contain
487 alignment commands, but be not of aligned length.
488 Need to detect this and croak(). */
489 size = measure_struct(symptr);
490 *symptr = savsym;
491 break;
492 }
80a13697
NC
493 case 'X' | TYPE_IS_SHRIEKING:
494 /* XXXX Is this useful? Then need to treat MEASURE_BACKWARDS.
495 */
496 if (!len) /* Avoid division by 0 */
497 len = 1;
498 len = total % len; /* Assumed: the start is aligned. */
924ba076 499 /* FALLTHROUGH */
80a13697
NC
500 case 'X':
501 size = -1;
502 if (total < len)
f7fe979e 503 Perl_croak(aTHX_ "'X' outside of string in %s", _action( symptr ) );
80a13697
NC
504 break;
505 case 'x' | TYPE_IS_SHRIEKING:
506 if (!len) /* Avoid division by 0 */
507 len = 1;
508 star = total % len; /* Assumed: the start is aligned. */
509 if (star) /* Other portable ways? */
510 len = len - star;
511 else
512 len = 0;
924ba076 513 /* FALLTHROUGH */
80a13697
NC
514 case 'x':
515 case 'A':
516 case 'Z':
517 case 'a':
80a13697
NC
518 size = 1;
519 break;
520 case 'B':
521 case 'b':
522 len = (len + 7)/8;
523 size = 1;
524 break;
525 case 'H':
526 case 'h':
527 len = (len + 1)/2;
528 size = 1;
529 break;
78d46eaa 530
80a13697
NC
531 case 'P':
532 len = 1;
533 size = sizeof(char*);
78d46eaa
NC
534 break;
535 }
206947d2
IZ
536 }
537 total += len * size;
538 }
539 return total;
540}
541
49704364
WL
542
543/* locate matching closing parenthesis or bracket
544 * returns char pointer to char after match, or NULL
545 */
f7fe979e 546STATIC const char *
5aaab254 547S_group_end(pTHX_ const char *patptr, const char *patend, char ender)
18529408 548{
7918f24d
NC
549 PERL_ARGS_ASSERT_GROUP_END;
550
49704364 551 while (patptr < patend) {
f7fe979e 552 const char c = *patptr++;
49704364
WL
553
554 if (isSPACE(c))
555 continue;
556 else if (c == ender)
557 return patptr-1;
558 else if (c == '#') {
559 while (patptr < patend && *patptr != '\n')
560 patptr++;
561 continue;
562 } else if (c == '(')
563 patptr = group_end(patptr, patend, ')') + 1;
564 else if (c == '[')
565 patptr = group_end(patptr, patend, ']') + 1;
18529408 566 }
49704364
WL
567 Perl_croak(aTHX_ "No group ending character '%c' found in template",
568 ender);
a25b5927 569 NOT_REACHED; /* NOTREACHED */
18529408
IZ
570}
571
49704364
WL
572
573/* Convert unsigned decimal number to binary.
574 * Expects a pointer to the first digit and address of length variable
575 * Advances char pointer to 1st non-digit char and returns number
fc241834 576 */
f7fe979e 577STATIC const char *
5aaab254 578S_get_num(pTHX_ const char *patptr, I32 *lenptr )
49704364
WL
579{
580 I32 len = *patptr++ - '0';
7918f24d
NC
581
582 PERL_ARGS_ASSERT_GET_NUM;
583
49704364
WL
584 while (isDIGIT(*patptr)) {
585 if (len >= 0x7FFFFFFF/10)
586 Perl_croak(aTHX_ "pack/unpack repeat count overflow");
587 len = (len * 10) + (*patptr++ - '0');
588 }
589 *lenptr = len;
590 return patptr;
591}
592
593/* The marvellous template parsing routine: Using state stored in *symptr,
594 * locates next template code and count
595 */
596STATIC bool
f337b084 597S_next_symbol(pTHX_ tempsym_t* symptr )
18529408 598{
f7fe979e 599 const char* patptr = symptr->patptr;
0bcc34c2 600 const char* const patend = symptr->patend;
49704364 601
7918f24d
NC
602 PERL_ARGS_ASSERT_NEXT_SYMBOL;
603
49704364
WL
604 symptr->flags &= ~FLAG_SLASH;
605
606 while (patptr < patend) {
607 if (isSPACE(*patptr))
608 patptr++;
609 else if (*patptr == '#') {
610 patptr++;
611 while (patptr < patend && *patptr != '\n')
612 patptr++;
613 if (patptr < patend)
614 patptr++;
615 } else {
fc241834 616 /* We should have found a template code */
49704364 617 I32 code = *patptr++ & 0xFF;
66c611c5 618 U32 inherited_modifiers = 0;
49704364
WL
619
620 if (code == ','){ /* grandfather in commas but with a warning */
621 if (((symptr->flags & FLAG_COMMA) == 0) && ckWARN(WARN_UNPACK)){
622 symptr->flags |= FLAG_COMMA;
623 Perl_warner(aTHX_ packWARN(WARN_UNPACK),
f7fe979e 624 "Invalid type ',' in %s", _action( symptr ) );
49704364
WL
625 }
626 continue;
627 }
fc241834 628
49704364 629 /* for '(', skip to ')' */
fc241834 630 if (code == '(') {
49704364
WL
631 if( isDIGIT(*patptr) || *patptr == '*' || *patptr == '[' )
632 Perl_croak(aTHX_ "()-group starts with a count in %s",
f7fe979e 633 _action( symptr ) );
49704364
WL
634 symptr->grpbeg = patptr;
635 patptr = 1 + ( symptr->grpend = group_end(patptr, patend, ')') );
636 if( symptr->level >= MAX_SUB_TEMPLATE_LEVEL )
637 Perl_croak(aTHX_ "Too deeply nested ()-groups in %s",
f7fe979e 638 _action( symptr ) );
49704364
WL
639 }
640
66c611c5
MHM
641 /* look for group modifiers to inherit */
642 if (TYPE_ENDIANNESS(symptr->flags)) {
643 if (strchr(ENDIANNESS_ALLOWED_TYPES, TYPE_NO_MODIFIERS(code)))
644 inherited_modifiers |= TYPE_ENDIANNESS(symptr->flags);
645 }
646
1109a392
MHM
647 /* look for modifiers */
648 while (patptr < patend) {
b7787f18
AL
649 const char *allowed;
650 I32 modifier;
1109a392
MHM
651 switch (*patptr) {
652 case '!':
653 modifier = TYPE_IS_SHRIEKING;
f8e5a5db 654 allowed = "sSiIlLxXnNvV@.";
1109a392
MHM
655 break;
656 case '>':
657 modifier = TYPE_IS_BIG_ENDIAN;
66c611c5 658 allowed = ENDIANNESS_ALLOWED_TYPES;
1109a392
MHM
659 break;
660 case '<':
661 modifier = TYPE_IS_LITTLE_ENDIAN;
66c611c5 662 allowed = ENDIANNESS_ALLOWED_TYPES;
1109a392
MHM
663 break;
664 default:
b7787f18
AL
665 allowed = "";
666 modifier = 0;
1109a392
MHM
667 break;
668 }
66c611c5 669
1109a392
MHM
670 if (modifier == 0)
671 break;
66c611c5 672
1109a392
MHM
673 if (!strchr(allowed, TYPE_NO_MODIFIERS(code)))
674 Perl_croak(aTHX_ "'%c' allowed only after types %s in %s", *patptr,
f7fe979e 675 allowed, _action( symptr ) );
66c611c5
MHM
676
677 if (TYPE_ENDIANNESS(code | modifier) == TYPE_ENDIANNESS_MASK)
1109a392 678 Perl_croak(aTHX_ "Can't use both '<' and '>' after type '%c' in %s",
f7fe979e 679 (int) TYPE_NO_MODIFIERS(code), _action( symptr ) );
66c611c5
MHM
680 else if (TYPE_ENDIANNESS(code | modifier | inherited_modifiers) ==
681 TYPE_ENDIANNESS_MASK)
682 Perl_croak(aTHX_ "Can't use '%c' in a group with different byte-order in %s",
f7fe979e 683 *patptr, _action( symptr ) );
66c611c5 684
a2a5de95
NC
685 if ((code & modifier)) {
686 Perl_ck_warner(aTHX_ packWARN(WARN_UNPACK),
687 "Duplicate modifier '%c' after '%c' in %s",
688 *patptr, (int) TYPE_NO_MODIFIERS(code),
689 _action( symptr ) );
1109a392 690 }
66c611c5 691
1109a392
MHM
692 code |= modifier;
693 patptr++;
49704364
WL
694 }
695
66c611c5
MHM
696 /* inherit modifiers */
697 code |= inherited_modifiers;
698
fc241834 699 /* look for count and/or / */
49704364
WL
700 if (patptr < patend) {
701 if (isDIGIT(*patptr)) {
702 patptr = get_num( patptr, &symptr->length );
703 symptr->howlen = e_number;
704
705 } else if (*patptr == '*') {
706 patptr++;
707 symptr->howlen = e_star;
708
709 } else if (*patptr == '[') {
f7fe979e 710 const char* lenptr = ++patptr;
49704364
WL
711 symptr->howlen = e_number;
712 patptr = group_end( patptr, patend, ']' ) + 1;
713 /* what kind of [] is it? */
714 if (isDIGIT(*lenptr)) {
715 lenptr = get_num( lenptr, &symptr->length );
716 if( *lenptr != ']' )
717 Perl_croak(aTHX_ "Malformed integer in [] in %s",
f7fe979e 718 _action( symptr ) );
49704364
WL
719 } else {
720 tempsym_t savsym = *symptr;
721 symptr->patend = patptr-1;
722 symptr->patptr = lenptr;
723 savsym.length = measure_struct(symptr);
724 *symptr = savsym;
725 }
726 } else {
727 symptr->howlen = e_no_len;
728 symptr->length = 1;
729 }
730
731 /* try to find / */
732 while (patptr < patend) {
733 if (isSPACE(*patptr))
734 patptr++;
735 else if (*patptr == '#') {
736 patptr++;
737 while (patptr < patend && *patptr != '\n')
738 patptr++;
739 if (patptr < patend)
740 patptr++;
741 } else {
66c611c5 742 if (*patptr == '/') {
49704364
WL
743 symptr->flags |= FLAG_SLASH;
744 patptr++;
66c611c5
MHM
745 if (patptr < patend &&
746 (isDIGIT(*patptr) || *patptr == '*' || *patptr == '['))
49704364 747 Perl_croak(aTHX_ "'/' does not take a repeat count in %s",
f7fe979e 748 _action( symptr ) );
49704364
WL
749 }
750 break;
751 }
18529408 752 }
49704364
WL
753 } else {
754 /* at end - no count, no / */
755 symptr->howlen = e_no_len;
756 symptr->length = 1;
757 }
758
759 symptr->code = code;
fc241834 760 symptr->patptr = patptr;
49704364 761 return TRUE;
18529408 762 }
49704364 763 }
fc241834 764 symptr->patptr = patptr;
49704364 765 return FALSE;
18529408
IZ
766}
767
18529408 768/*
fc241834 769 There is no way to cleanly handle the case where we should process the
08ca2aa3 770 string per byte in its upgraded form while it's really in downgraded form
fc241834
RGS
771 (e.g. estimates like strend-s as an upper bound for the number of
772 characters left wouldn't work). So if we foresee the need of this
773 (pattern starts with U or contains U0), we want to work on the encoded
774 version of the string. Users are advised to upgrade their pack string
08ca2aa3
TH
775 themselves if they need to do a lot of unpacks like this on it
776*/
fc241834 777STATIC bool
08ca2aa3
TH
778need_utf8(const char *pat, const char *patend)
779{
780 bool first = TRUE;
7918f24d
NC
781
782 PERL_ARGS_ASSERT_NEED_UTF8;
783
08ca2aa3
TH
784 while (pat < patend) {
785 if (pat[0] == '#') {
786 pat++;
f7fe979e 787 pat = (const char *) memchr(pat, '\n', patend-pat);
08ca2aa3
TH
788 if (!pat) return FALSE;
789 } else if (pat[0] == 'U') {
790 if (first || pat[1] == '0') return TRUE;
791 } else first = FALSE;
792 pat++;
793 }
794 return FALSE;
795}
796
797STATIC char
798first_symbol(const char *pat, const char *patend) {
7918f24d
NC
799 PERL_ARGS_ASSERT_FIRST_SYMBOL;
800
08ca2aa3
TH
801 while (pat < patend) {
802 if (pat[0] != '#') return pat[0];
803 pat++;
f7fe979e 804 pat = (const char *) memchr(pat, '\n', patend-pat);
08ca2aa3
TH
805 if (!pat) return 0;
806 pat++;
807 }
808 return 0;
809}
810
811/*
dcccc8ff
KW
812
813=head1 Pack and Unpack
814
7accc089
JH
815=for apidoc unpackstring
816
21ebfc7a
DM
817The engine implementing the unpack() Perl function.
818
819Using the template pat..patend, this function unpacks the string
820s..strend into a number of mortal SVs, which it pushes onto the perl
821argument (@_) stack (so you will need to issue a C<PUTBACK> before and
72d33970 822C<SPAGAIN> after the call to this function). It returns the number of
21ebfc7a
DM
823pushed elements.
824
825The strend and patend pointers should point to the byte following the last
826character of each string.
827
828Although this function returns its values on the perl argument stack, it
829doesn't take any parameters from that stack (and thus in particular
830there's no need to do a PUSHMARK before calling it, unlike L</call_pv> for
831example).
7accc089
JH
832
833=cut */
834
835I32
f7fe979e 836Perl_unpackstring(pTHX_ const char *pat, const char *patend, const char *s, const char *strend, U32 flags)
7accc089 837{
f7fe979e 838 tempsym_t sym;
08ca2aa3 839
7918f24d
NC
840 PERL_ARGS_ASSERT_UNPACKSTRING;
841
f337b084 842 if (flags & FLAG_DO_UTF8) flags |= FLAG_WAS_UTF8;
08ca2aa3
TH
843 else if (need_utf8(pat, patend)) {
844 /* We probably should try to avoid this in case a scalar context call
845 wouldn't get to the "U0" */
846 STRLEN len = strend - s;
230e1fce 847 s = (char *) bytes_to_utf8((U8 *) s, &len);
08ca2aa3
TH
848 SAVEFREEPV(s);
849 strend = s + len;
f337b084 850 flags |= FLAG_DO_UTF8;
08ca2aa3
TH
851 }
852
f337b084
TH
853 if (first_symbol(pat, patend) != 'U' && (flags & FLAG_DO_UTF8))
854 flags |= FLAG_PARSE_UTF8;
08ca2aa3 855
f7fe979e 856 TEMPSYM_INIT(&sym, pat, patend, flags);
7accc089
JH
857
858 return unpack_rec(&sym, s, s, strend, NULL );
859}
860
4136a0f7 861STATIC I32
f7fe979e 862S_unpack_rec(pTHX_ tempsym_t* symptr, const char *s, const char *strbeg, const char *strend, const char **new_s )
49704364 863{
20b7effb 864 dSP;
3297d27d 865 SV *sv = NULL;
f7fe979e 866 const I32 start_sp_offset = SP - PL_stack_base;
49704364 867 howlen_t howlen;
a6ec74c1 868 I32 checksum = 0;
92d41999 869 UV cuv = 0;
a6ec74c1 870 NV cdouble = 0.0;
f337b084 871 const int bits_in_uv = CHAR_BIT * sizeof(cuv);
49704364 872 bool beyond = FALSE;
21c16052 873 bool explicit_length;
9e27e96a 874 const bool unpack_only_one = (symptr->flags & FLAG_UNPACK_ONLY_ONE) != 0;
f337b084 875 bool utf8 = (symptr->flags & FLAG_PARSE_UTF8) ? 1 : 0;
7918f24d
NC
876
877 PERL_ARGS_ASSERT_UNPACK_REC;
878
28be1210 879 symptr->strbeg = s - strbeg;
49704364 880
49704364 881 while (next_symbol(symptr)) {
a7a3cfaa 882 packprops_t props;
9e27e96a 883 I32 len;
f337b084 884 I32 datumtype = symptr->code;
a1219b5e 885 bool needs_swap;
206947d2 886 /* do first one only unless in list context
08ca2aa3 887 / is implemented by unpacking the count, then popping it from the
206947d2 888 stack, so must check that we're not in the middle of a / */
49704364 889 if ( unpack_only_one
206947d2 890 && (SP - PL_stack_base == start_sp_offset + 1)
49704364 891 && (datumtype != '/') ) /* XXX can this be omitted */
206947d2 892 break;
49704364 893
f337b084 894 switch (howlen = symptr->howlen) {
fc241834
RGS
895 case e_star:
896 len = strend - strbeg; /* long enough */
49704364 897 break;
f337b084
TH
898 default:
899 /* e_no_len and e_number */
900 len = symptr->length;
901 break;
49704364 902 }
18529408 903
21c16052 904 explicit_length = TRUE;
a6ec74c1 905 redo_switch:
49704364 906 beyond = s >= strend;
a7a3cfaa
TH
907
908 props = packprops[TYPE_NO_ENDIANNESS(datumtype)];
909 if (props) {
910 /* props nonzero means we can process this letter. */
9e27e96a
AL
911 const long size = props & PACK_SIZE_MASK;
912 const long howmany = (strend - s) / size;
a7a3cfaa
TH
913 if (len > howmany)
914 len = howmany;
915
916 if (!checksum || (props & PACK_SIZE_CANNOT_CSUM)) {
917 if (len && unpack_only_one) len = 1;
918 EXTEND(SP, len);
919 EXTEND_MORTAL(len);
78d46eaa
NC
920 }
921 }
a7a3cfaa 922
a1219b5e
NC
923 needs_swap = NEEDS_SWAP(datumtype);
924
1109a392 925 switch(TYPE_NO_ENDIANNESS(datumtype)) {
a6ec74c1 926 default:
1109a392 927 Perl_croak(aTHX_ "Invalid type '%c' in unpack", (int)TYPE_NO_MODIFIERS(datumtype) );
49704364 928
a6ec74c1 929 case '%':
49704364 930 if (howlen == e_no_len)
18529408 931 len = 16; /* len is not specified */
a6ec74c1 932 checksum = len;
92d41999 933 cuv = 0;
a6ec74c1 934 cdouble = 0;
18529408 935 continue;
81d52ecd 936
18529408
IZ
937 case '(':
938 {
49704364 939 tempsym_t savsym = *symptr;
9e27e96a 940 const U32 group_modifiers = TYPE_MODIFIERS(datumtype & ~symptr->flags);
66c611c5 941 symptr->flags |= group_modifiers;
49704364 942 symptr->patend = savsym.grpend;
28be1210 943 symptr->previous = &savsym;
49704364 944 symptr->level++;
18529408 945 PUTBACK;
c6f750d1 946 if (len && unpack_only_one) len = 1;
18529408 947 while (len--) {
49704364 948 symptr->patptr = savsym.grpbeg;
f337b084
TH
949 if (utf8) symptr->flags |= FLAG_PARSE_UTF8;
950 else symptr->flags &= ~FLAG_PARSE_UTF8;
08ca2aa3
TH
951 unpack_rec(symptr, s, strbeg, strend, &s);
952 if (s == strend && savsym.howlen == e_star)
49704364 953 break; /* No way to continue */
18529408
IZ
954 }
955 SPAGAIN;
28be1210 956 savsym.flags = symptr->flags & ~group_modifiers;
49704364 957 *symptr = savsym;
18529408
IZ
958 break;
959 }
28be1210 960 case '.' | TYPE_IS_SHRIEKING:
28be1210 961 case '.': {
9e27e96a 962 const char *from;
28be1210 963 SV *sv;
9e27e96a 964 const bool u8 = utf8 && !(datumtype & TYPE_IS_SHRIEKING);
28be1210
TH
965 if (howlen == e_star) from = strbeg;
966 else if (len <= 0) from = s;
967 else {
968 tempsym_t *group = symptr;
969
970 while (--len && group) group = group->previous;
971 from = group ? strbeg + group->strbeg : strbeg;
972 }
973 sv = from <= s ?
00646304
CB
974 newSVuv( u8 ? (UV) utf8_length((const U8*)from, (const U8*)s) : (UV) (s-from)) :
975 newSViv(-(u8 ? (IV) utf8_length((const U8*)s, (const U8*)from) : (IV) (from-s)));
6e449a3a 976 mXPUSHs(sv);
28be1210
TH
977 break;
978 }
28be1210 979 case '@' | TYPE_IS_SHRIEKING:
a6ec74c1 980 case '@':
28be1210 981 s = strbeg + symptr->strbeg;
28be1210 982 if (utf8 && !(datumtype & TYPE_IS_SHRIEKING))
28be1210 983 {
08ca2aa3
TH
984 while (len > 0) {
985 if (s >= strend)
986 Perl_croak(aTHX_ "'@' outside of string in unpack");
987 s += UTF8SKIP(s);
988 len--;
989 }
990 if (s > strend)
991 Perl_croak(aTHX_ "'@' outside of string with malformed UTF-8 in unpack");
992 } else {
28be1210 993 if (strend-s < len)
fc241834 994 Perl_croak(aTHX_ "'@' outside of string in unpack");
28be1210 995 s += len;
08ca2aa3 996 }
a6ec74c1 997 break;
62f95557
IZ
998 case 'X' | TYPE_IS_SHRIEKING:
999 if (!len) /* Avoid division by 0 */
1000 len = 1;
08ca2aa3 1001 if (utf8) {
f7fe979e 1002 const char *hop, *last;
f337b084
TH
1003 I32 l = len;
1004 hop = last = strbeg;
1005 while (hop < s) {
1006 hop += UTF8SKIP(hop);
1007 if (--l == 0) {
08ca2aa3 1008 last = hop;
f337b084
TH
1009 l = len;
1010 }
fc241834 1011 }
f337b084
TH
1012 if (last > s)
1013 Perl_croak(aTHX_ "Malformed UTF-8 string in unpack");
08ca2aa3
TH
1014 s = last;
1015 break;
f337b084
TH
1016 }
1017 len = (s - strbeg) % len;
924ba076 1018 /* FALLTHROUGH */
a6ec74c1 1019 case 'X':
08ca2aa3
TH
1020 if (utf8) {
1021 while (len > 0) {
1022 if (s <= strbeg)
1023 Perl_croak(aTHX_ "'X' outside of string in unpack");
f337b084 1024 while (--s, UTF8_IS_CONTINUATION(*s)) {
08ca2aa3
TH
1025 if (s <= strbeg)
1026 Perl_croak(aTHX_ "'X' outside of string in unpack");
1027 }
1028 len--;
1029 }
1030 } else {
fc241834
RGS
1031 if (len > s - strbeg)
1032 Perl_croak(aTHX_ "'X' outside of string in unpack" );
1033 s -= len;
08ca2aa3 1034 }
a6ec74c1 1035 break;
9e27e96a
AL
1036 case 'x' | TYPE_IS_SHRIEKING: {
1037 I32 ai32;
62f95557
IZ
1038 if (!len) /* Avoid division by 0 */
1039 len = 1;
230e1fce
NC
1040 if (utf8) ai32 = utf8_length((U8 *) strbeg, (U8 *) s) % len;
1041 else ai32 = (s - strbeg) % len;
08ca2aa3
TH
1042 if (ai32 == 0) break;
1043 len -= ai32;
9e27e96a 1044 }
924ba076 1045 /* FALLTHROUGH */
a6ec74c1 1046 case 'x':
08ca2aa3
TH
1047 if (utf8) {
1048 while (len>0) {
1049 if (s >= strend)
1050 Perl_croak(aTHX_ "'x' outside of string in unpack");
1051 s += UTF8SKIP(s);
1052 len--;
1053 }
1054 } else {
fc241834
RGS
1055 if (len > strend - s)
1056 Perl_croak(aTHX_ "'x' outside of string in unpack");
1057 s += len;
f337b084 1058 }
a6ec74c1
JH
1059 break;
1060 case '/':
49704364 1061 Perl_croak(aTHX_ "'/' must follow a numeric type in unpack");
81d52ecd 1062
a6ec74c1
JH
1063 case 'A':
1064 case 'Z':
1065 case 'a':
08ca2aa3
TH
1066 if (checksum) {
1067 /* Preliminary length estimate is assumed done in 'W' */
1068 if (len > strend - s) len = strend - s;
1069 goto W_checksum;
1070 }
1071 if (utf8) {
1072 I32 l;
f7fe979e 1073 const char *hop;
08ca2aa3
TH
1074 for (l=len, hop=s; l>0; l--, hop += UTF8SKIP(hop)) {
1075 if (hop >= strend) {
1076 if (hop > strend)
1077 Perl_croak(aTHX_ "Malformed UTF-8 string in unpack");
1078 break;
fc241834 1079 }
a6ec74c1 1080 }
08ca2aa3
TH
1081 if (hop > strend)
1082 Perl_croak(aTHX_ "Malformed UTF-8 string in unpack");
1083 len = hop - s;
1084 } else if (len > strend - s)
1085 len = strend - s;
1086
1087 if (datumtype == 'Z') {
1088 /* 'Z' strips stuff after first null */
f7fe979e 1089 const char *ptr, *end;
f337b084
TH
1090 end = s + len;
1091 for (ptr = s; ptr < end; ptr++) if (*ptr == 0) break;
08ca2aa3
TH
1092 sv = newSVpvn(s, ptr-s);
1093 if (howlen == e_star) /* exact for 'Z*' */
1094 len = ptr-s + (ptr != strend ? 1 : 0);
1095 } else if (datumtype == 'A') {
1096 /* 'A' strips both nulls and spaces */
f7fe979e 1097 const char *ptr;
18bdf90a
TH
1098 if (utf8 && (symptr->flags & FLAG_WAS_UTF8)) {
1099 for (ptr = s+len-1; ptr >= s; ptr--)
1100 if (*ptr != 0 && !UTF8_IS_CONTINUATION(*ptr) &&
76a77b1b 1101 !isSPACE_utf8(ptr)) break;
18bdf90a
TH
1102 if (ptr >= s) ptr += UTF8SKIP(ptr);
1103 else ptr++;
28be1210 1104 if (ptr > s+len)
18bdf90a
TH
1105 Perl_croak(aTHX_ "Malformed UTF-8 string in unpack");
1106 } else {
1107 for (ptr = s+len-1; ptr >= s; ptr--)
1108 if (*ptr != 0 && !isSPACE(*ptr)) break;
1109 ptr++;
1110 }
08ca2aa3
TH
1111 sv = newSVpvn(s, ptr-s);
1112 } else sv = newSVpvn(s, len);
1113
1114 if (utf8) {
1115 SvUTF8_on(sv);
1116 /* Undo any upgrade done due to need_utf8() */
f337b084 1117 if (!(symptr->flags & FLAG_WAS_UTF8))
08ca2aa3 1118 sv_utf8_downgrade(sv, 0);
a6ec74c1 1119 }
6e449a3a 1120 mXPUSHs(sv);
08ca2aa3 1121 s += len;
a6ec74c1
JH
1122 break;
1123 case 'B':
08ca2aa3
TH
1124 case 'b': {
1125 char *str;
49704364 1126 if (howlen == e_star || len > (strend - s) * 8)
a6ec74c1
JH
1127 len = (strend - s) * 8;
1128 if (checksum) {
f337b084 1129 if (utf8)
08ca2aa3 1130 while (len >= 8 && s < strend) {
f337b084 1131 cuv += PL_bitcount[uni_to_byte(aTHX_ &s, strend, datumtype)];
08ca2aa3
TH
1132 len -= 8;
1133 }
f337b084 1134 else
fc241834 1135 while (len >= 8) {
08ca2aa3 1136 cuv += PL_bitcount[*(U8 *)s++];
fc241834
RGS
1137 len -= 8;
1138 }
08ca2aa3
TH
1139 if (len && s < strend) {
1140 U8 bits;
f337b084
TH
1141 bits = SHIFT_BYTE(utf8, s, strend, datumtype);
1142 if (datumtype == 'b')
a6ec74c1 1143 while (len-- > 0) {
92d41999 1144 if (bits & 1) cuv++;
a6ec74c1
JH
1145 bits >>= 1;
1146 }
f337b084 1147 else
a6ec74c1 1148 while (len-- > 0) {
08ca2aa3 1149 if (bits & 0x80) cuv++;
a6ec74c1
JH
1150 bits <<= 1;
1151 }
fc241834 1152 }
a6ec74c1
JH
1153 break;
1154 }
08ca2aa3 1155
561b68a9 1156 sv = sv_2mortal(newSV(len ? len : 1));
a6ec74c1
JH
1157 SvPOK_on(sv);
1158 str = SvPVX(sv);
1159 if (datumtype == 'b') {
f337b084 1160 U8 bits = 0;
f7fe979e 1161 const I32 ai32 = len;
08ca2aa3
TH
1162 for (len = 0; len < ai32; len++) {
1163 if (len & 7) bits >>= 1;
1164 else if (utf8) {
1165 if (s >= strend) break;
f337b084 1166 bits = uni_to_byte(aTHX_ &s, strend, datumtype);
08ca2aa3
TH
1167 } else bits = *(U8 *) s++;
1168 *str++ = bits & 1 ? '1' : '0';
a6ec74c1 1169 }
08ca2aa3 1170 } else {
f337b084 1171 U8 bits = 0;
f7fe979e 1172 const I32 ai32 = len;
08ca2aa3
TH
1173 for (len = 0; len < ai32; len++) {
1174 if (len & 7) bits <<= 1;
1175 else if (utf8) {
1176 if (s >= strend) break;
f337b084 1177 bits = uni_to_byte(aTHX_ &s, strend, datumtype);
08ca2aa3
TH
1178 } else bits = *(U8 *) s++;
1179 *str++ = bits & 0x80 ? '1' : '0';
a6ec74c1
JH
1180 }
1181 }
1182 *str = '\0';
aa07b2f6 1183 SvCUR_set(sv, str - SvPVX_const(sv));
08ca2aa3 1184 XPUSHs(sv);
a6ec74c1 1185 break;
08ca2aa3 1186 }
a6ec74c1 1187 case 'H':
08ca2aa3 1188 case 'h': {
3297d27d 1189 char *str = NULL;
fc241834 1190 /* Preliminary length estimate, acceptable for utf8 too */
49704364 1191 if (howlen == e_star || len > (strend - s) * 2)
a6ec74c1 1192 len = (strend - s) * 2;
858fe5e1
TC
1193 if (!checksum) {
1194 sv = sv_2mortal(newSV(len ? len : 1));
1195 SvPOK_on(sv);
1196 str = SvPVX(sv);
1197 }
a6ec74c1 1198 if (datumtype == 'h') {
f337b084 1199 U8 bits = 0;
9e27e96a 1200 I32 ai32 = len;
fc241834
RGS
1201 for (len = 0; len < ai32; len++) {
1202 if (len & 1) bits >>= 4;
1203 else if (utf8) {
1204 if (s >= strend) break;
f337b084 1205 bits = uni_to_byte(aTHX_ &s, strend, datumtype);
fc241834 1206 } else bits = * (U8 *) s++;
858fe5e1
TC
1207 if (!checksum)
1208 *str++ = PL_hexdigit[bits & 15];
a6ec74c1 1209 }
08ca2aa3 1210 } else {
f337b084 1211 U8 bits = 0;
f7fe979e 1212 const I32 ai32 = len;
08ca2aa3
TH
1213 for (len = 0; len < ai32; len++) {
1214 if (len & 1) bits <<= 4;
1215 else if (utf8) {
1216 if (s >= strend) break;
f337b084 1217 bits = uni_to_byte(aTHX_ &s, strend, datumtype);
08ca2aa3 1218 } else bits = *(U8 *) s++;
858fe5e1
TC
1219 if (!checksum)
1220 *str++ = PL_hexdigit[(bits >> 4) & 15];
a6ec74c1
JH
1221 }
1222 }
858fe5e1
TC
1223 if (!checksum) {
1224 *str = '\0';
1225 SvCUR_set(sv, str - SvPVX_const(sv));
1226 XPUSHs(sv);
1227 }
a6ec74c1 1228 break;
08ca2aa3 1229 }
1651fc44
ML
1230 case 'C':
1231 if (len == 0) {
1232 if (explicit_length)
1233 /* Switch to "character" mode */
1234 utf8 = (symptr->flags & FLAG_DO_UTF8) ? 1 : 0;
1235 break;
1236 }
924ba076 1237 /* FALLTHROUGH */
a6ec74c1 1238 case 'c':
1651fc44
ML
1239 while (len-- > 0 && s < strend) {
1240 int aint;
1241 if (utf8)
1242 {
1243 STRLEN retlen;
1244 aint = utf8n_to_uvchr((U8 *) s, strend-s, &retlen,
1245 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
1246 if (retlen == (STRLEN) -1 || retlen == 0)
1247 Perl_croak(aTHX_ "Malformed UTF-8 string in unpack");
1248 s += retlen;
1249 }
1250 else
1251 aint = *(U8 *)(s)++;
1252 if (aint >= 128 && datumtype != 'C') /* fake up signed chars */
73cb7263 1253 aint -= 256;
08ca2aa3 1254 if (!checksum)
6e449a3a 1255 mPUSHi(aint);
73cb7263
NC
1256 else if (checksum > bits_in_uv)
1257 cdouble += (NV)aint;
1258 else
1259 cuv += aint;
a6ec74c1
JH
1260 }
1261 break;
08ca2aa3
TH
1262 case 'W':
1263 W_checksum:
1651fc44 1264 if (utf8) {
08ca2aa3 1265 while (len-- > 0 && s < strend) {
08ca2aa3 1266 STRLEN retlen;
f7fe979e 1267 const UV val = utf8n_to_uvchr((U8 *) s, strend-s, &retlen,
f337b084 1268 ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
08ca2aa3
TH
1269 if (retlen == (STRLEN) -1 || retlen == 0)
1270 Perl_croak(aTHX_ "Malformed UTF-8 string in unpack");
1271 s += retlen;
1272 if (!checksum)
6e449a3a 1273 mPUSHu(val);
08ca2aa3
TH
1274 else if (checksum > bits_in_uv)
1275 cdouble += (NV) val;
d6d3e8bd 1276 else
08ca2aa3 1277 cuv += val;
fc241834 1278 }
08ca2aa3 1279 } else if (!checksum)
a6ec74c1 1280 while (len-- > 0) {
f7fe979e 1281 const U8 ch = *(U8 *) s++;
6e449a3a 1282 mPUSHu(ch);
a6ec74c1 1283 }
08ca2aa3
TH
1284 else if (checksum > bits_in_uv)
1285 while (len-- > 0) cdouble += (NV) *(U8 *) s++;
1286 else
1287 while (len-- > 0) cuv += *(U8 *) s++;
a6ec74c1
JH
1288 break;
1289 case 'U':
35bcd338 1290 if (len == 0) {
c5333953 1291 if (explicit_length && howlen != e_star) {
08ca2aa3 1292 /* Switch to "bytes in UTF-8" mode */
f337b084 1293 if (symptr->flags & FLAG_DO_UTF8) utf8 = 0;
08ca2aa3
TH
1294 else
1295 /* Should be impossible due to the need_utf8() test */
1296 Perl_croak(aTHX_ "U0 mode on a byte string");
1297 }
35bcd338
JH
1298 break;
1299 }
08ca2aa3 1300 if (len > strend - s) len = strend - s;
fc241834 1301 if (!checksum) {
08ca2aa3
TH
1302 if (len && unpack_only_one) len = 1;
1303 EXTEND(SP, len);
1304 EXTEND_MORTAL(len);
fc241834 1305 }
08ca2aa3
TH
1306 while (len-- > 0 && s < strend) {
1307 STRLEN retlen;
1308 UV auv;
1309 if (utf8) {
1310 U8 result[UTF8_MAXLEN];
f7fe979e 1311 const char *ptr = s;
08ca2aa3 1312 STRLEN len;
08ca2aa3
TH
1313 /* Bug: warns about bad utf8 even if we are short on bytes
1314 and will break out of the loop */
230e1fce
NC
1315 if (!uni_to_bytes(aTHX_ &ptr, strend, (char *) result, 1,
1316 'U'))
08ca2aa3
TH
1317 break;
1318 len = UTF8SKIP(result);
fc241834 1319 if (!uni_to_bytes(aTHX_ &ptr, strend,
230e1fce 1320 (char *) &result[1], len-1, 'U')) break;
c80e42f3 1321 auv = utf8n_to_uvchr(result, len, &retlen, UTF8_ALLOW_DEFAULT);
08ca2aa3
TH
1322 s = ptr;
1323 } else {
c80e42f3 1324 auv = utf8n_to_uvchr((U8*)s, strend - s, &retlen, UTF8_ALLOW_DEFAULT);
08ca2aa3
TH
1325 if (retlen == (STRLEN) -1 || retlen == 0)
1326 Perl_croak(aTHX_ "Malformed UTF-8 string in unpack");
1327 s += retlen;
1328 }
1329 if (!checksum)
6e449a3a 1330 mPUSHu(auv);
73cb7263 1331 else if (checksum > bits_in_uv)
08ca2aa3 1332 cdouble += (NV) auv;
73cb7263 1333 else
08ca2aa3 1334 cuv += auv;
a6ec74c1
JH
1335 }
1336 break;
49704364
WL
1337 case 's' | TYPE_IS_SHRIEKING:
1338#if SHORTSIZE != SIZE16
73cb7263 1339 while (len-- > 0) {
08ca2aa3 1340 short ashort;
aaec8192 1341 SHIFT_VAR(utf8, s, strend, ashort, datumtype, needs_swap);
08ca2aa3 1342 if (!checksum)
6e449a3a 1343 mPUSHi(ashort);
73cb7263
NC
1344 else if (checksum > bits_in_uv)
1345 cdouble += (NV)ashort;
1346 else
1347 cuv += ashort;
49704364
WL
1348 }
1349 break;
1350#else
924ba076 1351 /* FALLTHROUGH */
a6ec74c1 1352#endif
49704364 1353 case 's':
73cb7263 1354 while (len-- > 0) {
08ca2aa3
TH
1355 I16 ai16;
1356
1357#if U16SIZE > SIZE16
1358 ai16 = 0;
1359#endif
aaec8192 1360 SHIFT16(utf8, s, strend, &ai16, datumtype, needs_swap);
1109a392 1361#if U16SIZE > SIZE16
73cb7263
NC
1362 if (ai16 > 32767)
1363 ai16 -= 65536;
a6ec74c1 1364#endif
08ca2aa3 1365 if (!checksum)
6e449a3a 1366 mPUSHi(ai16);
73cb7263
NC
1367 else if (checksum > bits_in_uv)
1368 cdouble += (NV)ai16;
1369 else
1370 cuv += ai16;
a6ec74c1
JH
1371 }
1372 break;
49704364
WL
1373 case 'S' | TYPE_IS_SHRIEKING:
1374#if SHORTSIZE != SIZE16
73cb7263 1375 while (len-- > 0) {
08ca2aa3 1376 unsigned short aushort;
aaec8192
NC
1377 SHIFT_VAR(utf8, s, strend, aushort, datumtype, needs_swap,
1378 needs_swap);
08ca2aa3 1379 if (!checksum)
6e449a3a 1380 mPUSHu(aushort);
73cb7263
NC
1381 else if (checksum > bits_in_uv)
1382 cdouble += (NV)aushort;
1383 else
1384 cuv += aushort;
49704364
WL
1385 }
1386 break;
1387#else
924ba076 1388 /* FALLTHROUGH */
49704364 1389#endif
a6ec74c1
JH
1390 case 'v':
1391 case 'n':
1392 case 'S':
73cb7263 1393 while (len-- > 0) {
08ca2aa3
TH
1394 U16 au16;
1395#if U16SIZE > SIZE16
1396 au16 = 0;
1397#endif
aaec8192 1398 SHIFT16(utf8, s, strend, &au16, datumtype, needs_swap);
73cb7263
NC
1399 if (datumtype == 'n')
1400 au16 = PerlSock_ntohs(au16);
73cb7263
NC
1401 if (datumtype == 'v')
1402 au16 = vtohs(au16);
08ca2aa3 1403 if (!checksum)
6e449a3a 1404 mPUSHu(au16);
73cb7263 1405 else if (checksum > bits_in_uv)
f337b084 1406 cdouble += (NV) au16;
73cb7263
NC
1407 else
1408 cuv += au16;
a6ec74c1
JH
1409 }
1410 break;
068bd2e7
MHM
1411 case 'v' | TYPE_IS_SHRIEKING:
1412 case 'n' | TYPE_IS_SHRIEKING:
73cb7263 1413 while (len-- > 0) {
08ca2aa3
TH
1414 I16 ai16;
1415# if U16SIZE > SIZE16
1416 ai16 = 0;
1417# endif
aaec8192 1418 SHIFT16(utf8, s, strend, &ai16, datumtype, needs_swap);
e396d235
NC
1419 /* There should never be any byte-swapping here. */
1420 assert(!TYPE_ENDIANNESS(datumtype));
73cb7263 1421 if (datumtype == ('n' | TYPE_IS_SHRIEKING))
08ca2aa3 1422 ai16 = (I16) PerlSock_ntohs((U16) ai16);
73cb7263 1423 if (datumtype == ('v' | TYPE_IS_SHRIEKING))
08ca2aa3 1424 ai16 = (I16) vtohs((U16) ai16);
08ca2aa3 1425 if (!checksum)
6e449a3a 1426 mPUSHi(ai16);
73cb7263 1427 else if (checksum > bits_in_uv)
08ca2aa3 1428 cdouble += (NV) ai16;
73cb7263
NC
1429 else
1430 cuv += ai16;
068bd2e7
MHM
1431 }
1432 break;
a6ec74c1 1433 case 'i':
49704364 1434 case 'i' | TYPE_IS_SHRIEKING:
73cb7263 1435 while (len-- > 0) {
08ca2aa3 1436 int aint;
aaec8192 1437 SHIFT_VAR(utf8, s, strend, aint, datumtype, needs_swap);
08ca2aa3 1438 if (!checksum)
6e449a3a 1439 mPUSHi(aint);
73cb7263
NC
1440 else if (checksum > bits_in_uv)
1441 cdouble += (NV)aint;
1442 else
1443 cuv += aint;
a6ec74c1
JH
1444 }
1445 break;
1446 case 'I':
49704364 1447 case 'I' | TYPE_IS_SHRIEKING:
73cb7263 1448 while (len-- > 0) {
08ca2aa3 1449 unsigned int auint;
aaec8192 1450 SHIFT_VAR(utf8, s, strend, auint, datumtype, needs_swap);
08ca2aa3 1451 if (!checksum)
6e449a3a 1452 mPUSHu(auint);
73cb7263
NC
1453 else if (checksum > bits_in_uv)
1454 cdouble += (NV)auint;
1455 else
1456 cuv += auint;
a6ec74c1
JH
1457 }
1458 break;
92d41999 1459 case 'j':
73cb7263 1460 while (len-- > 0) {
08ca2aa3 1461 IV aiv;
aaec8192 1462 SHIFT_VAR(utf8, s, strend, aiv, datumtype, needs_swap);
08ca2aa3 1463 if (!checksum)
6e449a3a 1464 mPUSHi(aiv);
73cb7263
NC
1465 else if (checksum > bits_in_uv)
1466 cdouble += (NV)aiv;
1467 else
1468 cuv += aiv;
92d41999
JH
1469 }
1470 break;
1471 case 'J':
73cb7263 1472 while (len-- > 0) {
08ca2aa3 1473 UV auv;
aaec8192 1474 SHIFT_VAR(utf8, s, strend, auv, datumtype, needs_swap);
08ca2aa3 1475 if (!checksum)
6e449a3a 1476 mPUSHu(auv);
73cb7263
NC
1477 else if (checksum > bits_in_uv)
1478 cdouble += (NV)auv;
1479 else
1480 cuv += auv;
92d41999
JH
1481 }
1482 break;
49704364
WL
1483 case 'l' | TYPE_IS_SHRIEKING:
1484#if LONGSIZE != SIZE32
73cb7263 1485 while (len-- > 0) {
08ca2aa3 1486 long along;
aaec8192 1487 SHIFT_VAR(utf8, s, strend, along, datumtype, needs_swap);
08ca2aa3 1488 if (!checksum)
6e449a3a 1489 mPUSHi(along);
73cb7263
NC
1490 else if (checksum > bits_in_uv)
1491 cdouble += (NV)along;
1492 else
1493 cuv += along;
49704364
WL
1494 }
1495 break;
1496#else
924ba076 1497 /* FALLTHROUGH */
a6ec74c1 1498#endif
49704364 1499 case 'l':
73cb7263 1500 while (len-- > 0) {
08ca2aa3
TH
1501 I32 ai32;
1502#if U32SIZE > SIZE32
1503 ai32 = 0;
1504#endif
aaec8192 1505 SHIFT32(utf8, s, strend, &ai32, datumtype, needs_swap);
25a9bd2a 1506#if U32SIZE > SIZE32
08ca2aa3 1507 if (ai32 > 2147483647) ai32 -= 4294967296;
a6ec74c1 1508#endif
08ca2aa3 1509 if (!checksum)
6e449a3a 1510 mPUSHi(ai32);
73cb7263
NC
1511 else if (checksum > bits_in_uv)
1512 cdouble += (NV)ai32;
1513 else
1514 cuv += ai32;
a6ec74c1
JH
1515 }
1516 break;
49704364
WL
1517 case 'L' | TYPE_IS_SHRIEKING:
1518#if LONGSIZE != SIZE32
73cb7263 1519 while (len-- > 0) {
08ca2aa3 1520 unsigned long aulong;
aaec8192 1521 SHIFT_VAR(utf8, s, strend, aulong, datumtype, needs_swap);
08ca2aa3 1522 if (!checksum)
6e449a3a 1523 mPUSHu(aulong);
73cb7263
NC
1524 else if (checksum > bits_in_uv)
1525 cdouble += (NV)aulong;
1526 else
1527 cuv += aulong;
49704364
WL
1528 }
1529 break;
1530#else
924ba076 1531 /* FALLTHROUGH */
49704364 1532#endif
a6ec74c1
JH
1533 case 'V':
1534 case 'N':
1535 case 'L':
73cb7263 1536 while (len-- > 0) {
08ca2aa3
TH
1537 U32 au32;
1538#if U32SIZE > SIZE32
1539 au32 = 0;
1540#endif
aaec8192 1541 SHIFT32(utf8, s, strend, &au32, datumtype, needs_swap);
73cb7263
NC
1542 if (datumtype == 'N')
1543 au32 = PerlSock_ntohl(au32);
73cb7263
NC
1544 if (datumtype == 'V')
1545 au32 = vtohl(au32);
08ca2aa3 1546 if (!checksum)
6e449a3a 1547 mPUSHu(au32);
fc241834
RGS
1548 else if (checksum > bits_in_uv)
1549 cdouble += (NV)au32;
1550 else
1551 cuv += au32;
a6ec74c1
JH
1552 }
1553 break;
068bd2e7
MHM
1554 case 'V' | TYPE_IS_SHRIEKING:
1555 case 'N' | TYPE_IS_SHRIEKING:
73cb7263 1556 while (len-- > 0) {
08ca2aa3 1557 I32 ai32;
f8e5a5db 1558#if U32SIZE > SIZE32
08ca2aa3 1559 ai32 = 0;
f8e5a5db 1560#endif
aaec8192 1561 SHIFT32(utf8, s, strend, &ai32, datumtype, needs_swap);
e396d235
NC
1562 /* There should never be any byte swapping here. */
1563 assert(!TYPE_ENDIANNESS(datumtype));
73cb7263
NC
1564 if (datumtype == ('N' | TYPE_IS_SHRIEKING))
1565 ai32 = (I32)PerlSock_ntohl((U32)ai32);
73cb7263
NC
1566 if (datumtype == ('V' | TYPE_IS_SHRIEKING))
1567 ai32 = (I32)vtohl((U32)ai32);
08ca2aa3 1568 if (!checksum)
6e449a3a 1569 mPUSHi(ai32);
73cb7263
NC
1570 else if (checksum > bits_in_uv)
1571 cdouble += (NV)ai32;
1572 else
1573 cuv += ai32;
068bd2e7
MHM
1574 }
1575 break;
a6ec74c1 1576 case 'p':
a6ec74c1 1577 while (len-- > 0) {
f7fe979e 1578 const char *aptr;
aaec8192 1579 SHIFT_VAR(utf8, s, strend, aptr, datumtype, needs_swap);
c4c5f44a 1580 /* newSVpv generates undef if aptr is NULL */
6e449a3a 1581 mPUSHs(newSVpv(aptr, 0));
a6ec74c1
JH
1582 }
1583 break;
1584 case 'w':
a6ec74c1
JH
1585 {
1586 UV auv = 0;
1587 U32 bytes = 0;
fc241834 1588
08ca2aa3
TH
1589 while (len > 0 && s < strend) {
1590 U8 ch;
f337b084 1591 ch = SHIFT_BYTE(utf8, s, strend, datumtype);
08ca2aa3 1592 auv = (auv << 7) | (ch & 0x7f);
a6ec74c1 1593 /* UTF8_IS_XXXXX not right here - using constant 0x80 */
08ca2aa3 1594 if (ch < 0x80) {
a6ec74c1 1595 bytes = 0;
6e449a3a 1596 mPUSHu(auv);
a6ec74c1
JH
1597 len--;
1598 auv = 0;
08ca2aa3 1599 continue;
a6ec74c1 1600 }
08ca2aa3 1601 if (++bytes >= sizeof(UV)) { /* promote to string */
10516c54 1602 const char *t;
a6ec74c1 1603
f5992bc4 1604 sv = Perl_newSVpvf(aTHX_ "%.*"UVuf, (int)TYPE_DIGITS(UV), auv);
a6ec74c1 1605 while (s < strend) {
f337b084 1606 ch = SHIFT_BYTE(utf8, s, strend, datumtype);
08ca2aa3
TH
1607 sv = mul128(sv, (U8)(ch & 0x7f));
1608 if (!(ch & 0x80)) {
a6ec74c1
JH
1609 bytes = 0;
1610 break;
1611 }
1612 }
10516c54 1613 t = SvPV_nolen_const(sv);
a6ec74c1
JH
1614 while (*t == '0')
1615 t++;
1616 sv_chop(sv, t);
6e449a3a 1617 mPUSHs(sv);
a6ec74c1
JH
1618 len--;
1619 auv = 0;
1620 }
1621 }
1622 if ((s >= strend) && bytes)
49704364 1623 Perl_croak(aTHX_ "Unterminated compressed integer in unpack");
a6ec74c1
JH
1624 }
1625 break;
1626 case 'P':
49704364
WL
1627 if (symptr->howlen == e_star)
1628 Perl_croak(aTHX_ "'P' must have an explicit size in unpack");
a6ec74c1 1629 EXTEND(SP, 1);
2d3e0934 1630 if (s + sizeof(char*) <= strend) {
08ca2aa3 1631 char *aptr;
aaec8192 1632 SHIFT_VAR(utf8, s, strend, aptr, datumtype, needs_swap);
fc241834 1633 /* newSVpvn generates undef if aptr is NULL */
59cd0e26 1634 PUSHs(newSVpvn_flags(aptr, len, SVs_TEMP));
08ca2aa3 1635 }
a6ec74c1 1636 break;
c174bf3b 1637#if defined(HAS_QUAD) && IVSIZE >= 8
a6ec74c1 1638 case 'q':
73cb7263 1639 while (len-- > 0) {
08ca2aa3 1640 Quad_t aquad;
aaec8192 1641 SHIFT_VAR(utf8, s, strend, aquad, datumtype, needs_swap);
08ca2aa3 1642 if (!checksum)
c174bf3b 1643 mPUSHs(newSViv((IV)aquad));
73cb7263
NC
1644 else if (checksum > bits_in_uv)
1645 cdouble += (NV)aquad;
1646 else
1647 cuv += aquad;
1648 }
a6ec74c1
JH
1649 break;
1650 case 'Q':
73cb7263 1651 while (len-- > 0) {
08ca2aa3 1652 Uquad_t auquad;
aaec8192 1653 SHIFT_VAR(utf8, s, strend, auquad, datumtype, needs_swap);
08ca2aa3 1654 if (!checksum)
c174bf3b 1655 mPUSHs(newSVuv((UV)auquad));
73cb7263
NC
1656 else if (checksum > bits_in_uv)
1657 cdouble += (NV)auquad;
1658 else
1659 cuv += auquad;
a6ec74c1
JH
1660 }
1661 break;
1640b983 1662#endif
a6ec74c1
JH
1663 /* float and double added gnb@melba.bby.oz.au 22/11/89 */
1664 case 'f':
73cb7263 1665 while (len-- > 0) {
08ca2aa3 1666 float afloat;
aaec8192 1667 SHIFT_VAR(utf8, s, strend, afloat, datumtype, needs_swap);
08ca2aa3 1668 if (!checksum)
6e449a3a 1669 mPUSHn(afloat);
08ca2aa3 1670 else
73cb7263 1671 cdouble += afloat;
fc241834 1672 }
a6ec74c1
JH
1673 break;
1674 case 'd':
73cb7263 1675 while (len-- > 0) {
08ca2aa3 1676 double adouble;
aaec8192 1677 SHIFT_VAR(utf8, s, strend, adouble, datumtype, needs_swap);
08ca2aa3 1678 if (!checksum)
6e449a3a 1679 mPUSHn(adouble);
08ca2aa3 1680 else
73cb7263 1681 cdouble += adouble;
fc241834 1682 }
a6ec74c1 1683 break;
92d41999 1684 case 'F':
73cb7263 1685 while (len-- > 0) {
275663fa 1686 NV_bytes anv;
aaec8192
NC
1687 SHIFT_BYTES(utf8, s, strend, anv.bytes, sizeof(anv.bytes),
1688 datumtype, needs_swap);
08ca2aa3 1689 if (!checksum)
275663fa 1690 mPUSHn(anv.nv);
08ca2aa3 1691 else
275663fa 1692 cdouble += anv.nv;
fc241834 1693 }
92d41999
JH
1694 break;
1695#if defined(HAS_LONG_DOUBLE) && defined(USE_LONG_DOUBLE)
1696 case 'D':
73cb7263 1697 while (len-- > 0) {
275663fa 1698 ld_bytes aldouble;
aaec8192
NC
1699 SHIFT_BYTES(utf8, s, strend, aldouble.bytes,
1700 sizeof(aldouble.bytes), datumtype, needs_swap);
42262fd3
JH
1701 /* The most common long double format, the x86 80-bit
1702 * extended precision, has either 2 or 6 unused bytes,
1703 * which may contain garbage, which may contain
1704 * unintentional data. While we do zero the bytes of
1705 * the long double data in pack(), here in unpack() we
1706 * don't, because it's really hard to envision that
1707 * reading the long double off aldouble would be
1708 * affected the unused bytes.
1709 *
1710 * Note that trying to unpack 'long doubles' of 'long
1711 * doubles' packed in another system is in the general
1712 * case doomed without having more detail. */
08ca2aa3 1713 if (!checksum)
275663fa 1714 mPUSHn(aldouble.ld);
08ca2aa3 1715 else
275663fa 1716 cdouble += aldouble.ld;
92d41999
JH
1717 }
1718 break;
1719#endif
a6ec74c1 1720 case 'u':
858fe5e1 1721 if (!checksum) {
f7fe979e 1722 const STRLEN l = (STRLEN) (strend - s) * 3 / 4;
561b68a9 1723 sv = sv_2mortal(newSV(l));
08ca2aa3
TH
1724 if (l) SvPOK_on(sv);
1725 }
1726 if (utf8) {
1727 while (next_uni_uu(aTHX_ &s, strend, &len)) {
1728 I32 a, b, c, d;
db187877 1729 char hunk[3];
08ca2aa3 1730
08ca2aa3
TH
1731 while (len > 0) {
1732 next_uni_uu(aTHX_ &s, strend, &a);
1733 next_uni_uu(aTHX_ &s, strend, &b);
1734 next_uni_uu(aTHX_ &s, strend, &c);
1735 next_uni_uu(aTHX_ &s, strend, &d);
1736 hunk[0] = (char)((a << 2) | (b >> 4));
1737 hunk[1] = (char)((b << 4) | (c >> 2));
1738 hunk[2] = (char)((c << 6) | d);
858fe5e1
TC
1739 if (!checksum)
1740 sv_catpvn(sv, hunk, (len > 3) ? 3 : len);
08ca2aa3
TH
1741 len -= 3;
1742 }
1743 if (s < strend) {
f7fe979e
AL
1744 if (*s == '\n') {
1745 s++;
1746 }
08ca2aa3
TH
1747 else {
1748 /* possible checksum byte */
f7fe979e
AL
1749 const char *skip = s+UTF8SKIP(s);
1750 if (skip < strend && *skip == '\n')
1751 s = skip+1;
08ca2aa3
TH
1752 }
1753 }
1754 }
1755 } else {
fc241834
RGS
1756 while (s < strend && *s > ' ' && ISUUCHAR(*s)) {
1757 I32 a, b, c, d;
db187877 1758 char hunk[3];
a6ec74c1 1759
fc241834
RGS
1760 len = PL_uudmap[*(U8*)s++] & 077;
1761 while (len > 0) {
1762 if (s < strend && ISUUCHAR(*s))
1763 a = PL_uudmap[*(U8*)s++] & 077;
1764 else
1765 a = 0;
1766 if (s < strend && ISUUCHAR(*s))
1767 b = PL_uudmap[*(U8*)s++] & 077;
1768 else
1769 b = 0;
1770 if (s < strend && ISUUCHAR(*s))
1771 c = PL_uudmap[*(U8*)s++] & 077;
1772 else
1773 c = 0;
1774 if (s < strend && ISUUCHAR(*s))
1775 d = PL_uudmap[*(U8*)s++] & 077;
1776 else
1777 d = 0;
1778 hunk[0] = (char)((a << 2) | (b >> 4));
1779 hunk[1] = (char)((b << 4) | (c >> 2));
1780 hunk[2] = (char)((c << 6) | d);
858fe5e1
TC
1781 if (!checksum)
1782 sv_catpvn(sv, hunk, (len > 3) ? 3 : len);
fc241834
RGS
1783 len -= 3;
1784 }
1785 if (*s == '\n')
1786 s++;
1787 else /* possible checksum byte */
1788 if (s + 1 < strend && s[1] == '\n')
1789 s += 2;
a6ec74c1 1790 }
08ca2aa3 1791 }
858fe5e1
TC
1792 if (!checksum)
1793 XPUSHs(sv);
a6ec74c1
JH
1794 break;
1795 }
49704364 1796
a6ec74c1 1797 if (checksum) {
1109a392 1798 if (strchr("fFdD", TYPE_NO_MODIFIERS(datumtype)) ||
92d41999 1799 (checksum > bits_in_uv &&
08ca2aa3
TH
1800 strchr("cCsSiIlLnNUWvVqQjJ", TYPE_NO_MODIFIERS(datumtype))) ) {
1801 NV trouble, anv;
a6ec74c1 1802
08ca2aa3 1803 anv = (NV) (1 << (checksum & 15));
a6ec74c1
JH
1804 while (checksum >= 16) {
1805 checksum -= 16;
08ca2aa3 1806 anv *= 65536.0;
a6ec74c1 1807 }
a6ec74c1 1808 while (cdouble < 0.0)
08ca2aa3
TH
1809 cdouble += anv;
1810 cdouble = Perl_modf(cdouble / anv, &trouble) * anv;
c4c5f44a 1811 sv = newSVnv(cdouble);
a6ec74c1
JH
1812 }
1813 else {
fa8ec7c1
NC
1814 if (checksum < bits_in_uv) {
1815 UV mask = ((UV)1 << checksum) - 1;
92d41999 1816 cuv &= mask;
a6ec74c1 1817 }
c4c5f44a 1818 sv = newSVuv(cuv);
a6ec74c1 1819 }
6e449a3a 1820 mXPUSHs(sv);
a6ec74c1
JH
1821 checksum = 0;
1822 }
fc241834 1823
49704364
WL
1824 if (symptr->flags & FLAG_SLASH){
1825 if (SP - PL_stack_base - start_sp_offset <= 0)
21361d07 1826 break;
49704364
WL
1827 if( next_symbol(symptr) ){
1828 if( symptr->howlen == e_number )
1829 Perl_croak(aTHX_ "Count after length/code in unpack" );
1830 if( beyond ){
1831 /* ...end of char buffer then no decent length available */
1832 Perl_croak(aTHX_ "length/code after end of string in unpack" );
1833 } else {
1834 /* take top of stack (hope it's numeric) */
1835 len = POPi;
1836 if( len < 0 )
1837 Perl_croak(aTHX_ "Negative '/' count in unpack" );
1838 }
1839 } else {
1840 Perl_croak(aTHX_ "Code missing after '/' in unpack" );
1841 }
1842 datumtype = symptr->code;
21c16052 1843 explicit_length = FALSE;
49704364
WL
1844 goto redo_switch;
1845 }
a6ec74c1 1846 }
49704364 1847
18529408
IZ
1848 if (new_s)
1849 *new_s = s;
1850 PUTBACK;
1851 return SP - PL_stack_base - start_sp_offset;
1852}
1853
1854PP(pp_unpack)
1855{
1856 dSP;
bab9c0ac 1857 dPOPPOPssrl;
18529408
IZ
1858 I32 gimme = GIMME_V;
1859 STRLEN llen;
1860 STRLEN rlen;
5c144d81
NC
1861 const char *pat = SvPV_const(left, llen);
1862 const char *s = SvPV_const(right, rlen);
f7fe979e
AL
1863 const char *strend = s + rlen;
1864 const char *patend = pat + llen;
08ca2aa3 1865 I32 cnt;
18529408
IZ
1866
1867 PUTBACK;
7accc089 1868 cnt = unpackstring(pat, patend, s, strend,
49704364 1869 ((gimme == G_SCALAR) ? FLAG_UNPACK_ONLY_ONE : 0)
f337b084 1870 | (DO_UTF8(right) ? FLAG_DO_UTF8 : 0));
49704364 1871
18529408
IZ
1872 SPAGAIN;
1873 if ( !cnt && gimme == G_SCALAR )
1874 PUSHs(&PL_sv_undef);
a6ec74c1
JH
1875 RETURN;
1876}
1877
f337b084 1878STATIC U8 *
f7fe979e 1879doencodes(U8 *h, const char *s, I32 len)
a6ec74c1 1880{
f337b084 1881 *h++ = PL_uuemap[len];
a6ec74c1 1882 while (len > 2) {
f337b084
TH
1883 *h++ = PL_uuemap[(077 & (s[0] >> 2))];
1884 *h++ = PL_uuemap[(077 & (((s[0] << 4) & 060) | ((s[1] >> 4) & 017)))];
1885 *h++ = PL_uuemap[(077 & (((s[1] << 2) & 074) | ((s[2] >> 6) & 03)))];
1886 *h++ = PL_uuemap[(077 & (s[2] & 077))];
a6ec74c1
JH
1887 s += 3;
1888 len -= 3;
1889 }
1890 if (len > 0) {
f7fe979e 1891 const char r = (len > 1 ? s[1] : '\0');
f337b084
TH
1892 *h++ = PL_uuemap[(077 & (s[0] >> 2))];
1893 *h++ = PL_uuemap[(077 & (((s[0] << 4) & 060) | ((r >> 4) & 017)))];
1894 *h++ = PL_uuemap[(077 & ((r << 2) & 074))];
1895 *h++ = PL_uuemap[0];
a6ec74c1 1896 }
f337b084
TH
1897 *h++ = '\n';
1898 return h;
a6ec74c1
JH
1899}
1900
1901STATIC SV *
f7fe979e 1902S_is_an_int(pTHX_ const char *s, STRLEN l)
a6ec74c1 1903{
8b6e33c7
AL
1904 SV *result = newSVpvn(s, l);
1905 char *const result_c = SvPV_nolen(result); /* convenience */
1906 char *out = result_c;
1907 bool skip = 1;
1908 bool ignore = 0;
a6ec74c1 1909
7918f24d
NC
1910 PERL_ARGS_ASSERT_IS_AN_INT;
1911
a6ec74c1
JH
1912 while (*s) {
1913 switch (*s) {
1914 case ' ':
1915 break;
1916 case '+':
1917 if (!skip) {
1918 SvREFCNT_dec(result);
1919 return (NULL);
1920 }
1921 break;
1922 case '0':
1923 case '1':
1924 case '2':
1925 case '3':
1926 case '4':
1927 case '5':
1928 case '6':
1929 case '7':
1930 case '8':
1931 case '9':
1932 skip = 0;
1933 if (!ignore) {
1934 *(out++) = *s;
1935 }
1936 break;
1937 case '.':
1938 ignore = 1;
1939 break;
1940 default:
1941 SvREFCNT_dec(result);
1942 return (NULL);
1943 }
1944 s++;
1945 }
1946 *(out++) = '\0';
1947 SvCUR_set(result, out - result_c);
1948 return (result);
1949}
1950
1951/* pnum must be '\0' terminated */
1952STATIC int
1953S_div128(pTHX_ SV *pnum, bool *done)
1954{
8b6e33c7
AL
1955 STRLEN len;
1956 char * const s = SvPV(pnum, len);
1957 char *t = s;
1958 int m = 0;
1959
7918f24d
NC
1960 PERL_ARGS_ASSERT_DIV128;
1961
8b6e33c7
AL
1962 *done = 1;
1963 while (*t) {
1964 const int i = m * 10 + (*t - '0');
1965 const int r = (i >> 7); /* r < 10 */
1966 m = i & 0x7F;
1967 if (r) {
1968 *done = 0;
1969 }
1970 *(t++) = '0' + r;
a6ec74c1 1971 }
8b6e33c7
AL
1972 *(t++) = '\0';
1973 SvCUR_set(pnum, (STRLEN) (t - s));
1974 return (m);
a6ec74c1
JH
1975}
1976
18529408 1977/*
7accc089
JH
1978=for apidoc packlist
1979
1980The engine implementing pack() Perl function.
1981
bfce84ec
AL
1982=cut
1983*/
7accc089
JH
1984
1985void
5aaab254 1986Perl_packlist(pTHX_ SV *cat, const char *pat, const char *patend, SV **beglist, SV **endlist )
7accc089 1987{
aadb217d
JH
1988 tempsym_t sym;
1989
7918f24d
NC
1990 PERL_ARGS_ASSERT_PACKLIST;
1991
f7fe979e 1992 TEMPSYM_INIT(&sym, pat, patend, FLAG_PACK);
49704364 1993
f337b084
TH
1994 /* We're going to do changes through SvPVX(cat). Make sure it's valid.
1995 Also make sure any UTF8 flag is loaded */
56eb0262 1996 SvPV_force_nolen(cat);
bfce84ec
AL
1997 if (DO_UTF8(cat))
1998 sym.flags |= FLAG_PARSE_UTF8 | FLAG_DO_UTF8;
f337b084 1999
49704364
WL
2000 (void)pack_rec( cat, &sym, beglist, endlist );
2001}
2002
f337b084
TH
2003/* like sv_utf8_upgrade, but also repoint the group start markers */
2004STATIC void
2005marked_upgrade(pTHX_ SV *sv, tempsym_t *sym_ptr) {
2006 STRLEN len;
2007 tempsym_t *group;
f7fe979e
AL
2008 const char *from_ptr, *from_start, *from_end, **marks, **m;
2009 char *to_start, *to_ptr;
f337b084
TH
2010
2011 if (SvUTF8(sv)) return;
2012
aa07b2f6 2013 from_start = SvPVX_const(sv);
f337b084
TH
2014 from_end = from_start + SvCUR(sv);
2015 for (from_ptr = from_start; from_ptr < from_end; from_ptr++)
6f2d5cbc 2016 if (!NATIVE_BYTE_IS_INVARIANT(*from_ptr)) break;
f337b084
TH
2017 if (from_ptr == from_end) {
2018 /* Simple case: no character needs to be changed */
2019 SvUTF8_on(sv);
2020 return;
2021 }
2022
3473cf63 2023 len = (from_end-from_ptr)*UTF8_EXPAND+(from_ptr-from_start)+1;
a02a5408 2024 Newx(to_start, len, char);
f337b084
TH
2025 Copy(from_start, to_start, from_ptr-from_start, char);
2026 to_ptr = to_start + (from_ptr-from_start);
2027
a02a5408 2028 Newx(marks, sym_ptr->level+2, const char *);
f337b084
TH
2029 for (group=sym_ptr; group; group = group->previous)
2030 marks[group->level] = from_start + group->strbeg;
2031 marks[sym_ptr->level+1] = from_end+1;
2032 for (m = marks; *m < from_ptr; m++)
2033 *m = to_start + (*m-from_start);
2034
2035 for (;from_ptr < from_end; from_ptr++) {
2036 while (*m == from_ptr) *m++ = to_ptr;
230e1fce 2037 to_ptr = (char *) uvchr_to_utf8((U8 *) to_ptr, *(U8 *) from_ptr);
f337b084
TH
2038 }
2039 *to_ptr = 0;
2040
2041 while (*m == from_ptr) *m++ = to_ptr;
2042 if (m != marks + sym_ptr->level+1) {
2043 Safefree(marks);
2044 Safefree(to_start);
5637ef5b
NC
2045 Perl_croak(aTHX_ "panic: marks beyond string end, m=%p, marks=%p, "
2046 "level=%d", m, marks, sym_ptr->level);
f337b084
TH
2047 }
2048 for (group=sym_ptr; group; group = group->previous)
2049 group->strbeg = marks[group->level] - to_start;
2050 Safefree(marks);
2051
2052 if (SvOOK(sv)) {
2053 if (SvIVX(sv)) {
b162af07 2054 SvLEN_set(sv, SvLEN(sv) + SvIVX(sv));
f337b084
TH
2055 from_start -= SvIVX(sv);
2056 SvIV_set(sv, 0);
2057 }
2058 SvFLAGS(sv) &= ~SVf_OOK;
2059 }
2060 if (SvLEN(sv) != 0)
2061 Safefree(from_start);
f880fe2f 2062 SvPV_set(sv, to_start);
b162af07
SP
2063 SvCUR_set(sv, to_ptr - to_start);
2064 SvLEN_set(sv, len);
f337b084
TH
2065 SvUTF8_on(sv);
2066}
2067
2068/* Exponential string grower. Makes string extension effectively O(n)
2069 needed says how many extra bytes we need (not counting the final '\0')
2070 Only grows the string if there is an actual lack of space
2071*/
2072STATIC char *
0bd48802 2073S_sv_exp_grow(pTHX_ SV *sv, STRLEN needed) {
f7fe979e
AL
2074 const STRLEN cur = SvCUR(sv);
2075 const STRLEN len = SvLEN(sv);
f337b084 2076 STRLEN extend;
7918f24d
NC
2077
2078 PERL_ARGS_ASSERT_SV_EXP_GROW;
2079
f337b084
TH
2080 if (len - cur > needed) return SvPVX(sv);
2081 extend = needed > len ? needed : len;
2082 return SvGROW(sv, len+extend+1);
2083}
49704364
WL
2084
2085STATIC
2086SV **
f337b084 2087S_pack_rec(pTHX_ SV *cat, tempsym_t* symptr, SV **beglist, SV **endlist )
49704364 2088{
49704364 2089 tempsym_t lookahead;
f337b084
TH
2090 I32 items = endlist - beglist;
2091 bool found = next_symbol(symptr);
2092 bool utf8 = (symptr->flags & FLAG_PARSE_UTF8) ? 1 : 0;
041457d9 2093 bool warn_utf8 = ckWARN(WARN_UTF8);
81d52ecd 2094 char* from;
f337b084 2095
7918f24d
NC
2096 PERL_ARGS_ASSERT_PACK_REC;
2097
f337b084
TH
2098 if (symptr->level == 0 && found && symptr->code == 'U') {
2099 marked_upgrade(aTHX_ cat, symptr);
2100 symptr->flags |= FLAG_DO_UTF8;
2101 utf8 = 0;
49704364 2102 }
f337b084 2103 symptr->strbeg = SvCUR(cat);
49704364
WL
2104
2105 while (found) {
f337b084
TH
2106 SV *fromstr;
2107 STRLEN fromlen;
2108 I32 len;
a0714e2c 2109 SV *lengthcode = NULL;
49704364 2110 I32 datumtype = symptr->code;
f337b084
TH
2111 howlen_t howlen = symptr->howlen;
2112 char *start = SvPVX(cat);
2113 char *cur = start + SvCUR(cat);
a1219b5e 2114 bool needs_swap;
49704364 2115
f337b084
TH
2116#define NEXTFROM (lengthcode ? lengthcode : items-- > 0 ? *beglist++ : &PL_sv_no)
2117
2118 switch (howlen) {
fc241834 2119 case e_star:
f337b084
TH
2120 len = strchr("@Xxu", TYPE_NO_MODIFIERS(datumtype)) ?
2121 0 : items;
2122 break;
2123 default:
2124 /* e_no_len and e_number */
2125 len = symptr->length;
49704364
WL
2126 break;
2127 }
2128
f337b084 2129 if (len) {
a7a3cfaa 2130 packprops_t props = packprops[TYPE_NO_ENDIANNESS(datumtype)];
f337b084 2131
a7a3cfaa
TH
2132 if (props && !(props & PACK_SIZE_UNPREDICTABLE)) {
2133 /* We can process this letter. */
2134 STRLEN size = props & PACK_SIZE_MASK;
2135 GROWING(utf8, cat, start, cur, (STRLEN) len * size);
2136 }
f337b084
TH
2137 }
2138
49704364
WL
2139 /* Look ahead for next symbol. Do we have code/code? */
2140 lookahead = *symptr;
2141 found = next_symbol(&lookahead);
246f24af
TH
2142 if (symptr->flags & FLAG_SLASH) {
2143 IV count;
f337b084 2144 if (!found) Perl_croak(aTHX_ "Code missing after '/' in pack");
246f24af
TH
2145 if (strchr("aAZ", lookahead.code)) {
2146 if (lookahead.howlen == e_number) count = lookahead.length;
2147 else {
ce399ba6 2148 if (items > 0) {
48a5da33 2149 count = sv_len_utf8(*beglist);
ce399ba6 2150 }
246f24af
TH
2151 else count = 0;
2152 if (lookahead.code == 'Z') count++;
2153 }
2154 } else {
2155 if (lookahead.howlen == e_number && lookahead.length < items)
2156 count = lookahead.length;
2157 else count = items;
2158 }
2159 lookahead.howlen = e_number;
2160 lookahead.length = count;
2161 lengthcode = sv_2mortal(newSViv(count));
a6ec74c1 2162 }
49704364 2163
a1219b5e
NC
2164 needs_swap = NEEDS_SWAP(datumtype);
2165
fc241834
RGS
2166 /* Code inside the switch must take care to properly update
2167 cat (CUR length and '\0' termination) if it updated *cur and
f337b084 2168 doesn't simply leave using break */
1109a392 2169 switch(TYPE_NO_ENDIANNESS(datumtype)) {
a6ec74c1 2170 default:
f337b084
TH
2171 Perl_croak(aTHX_ "Invalid type '%c' in pack",
2172 (int) TYPE_NO_MODIFIERS(datumtype));
a6ec74c1 2173 case '%':
49704364 2174 Perl_croak(aTHX_ "'%%' may not be used in pack");
81d52ecd 2175
28be1210 2176 case '.' | TYPE_IS_SHRIEKING:
28be1210
TH
2177 case '.':
2178 if (howlen == e_star) from = start;
2179 else if (len == 0) from = cur;
2180 else {
2181 tempsym_t *group = symptr;
2182
2183 while (--len && group) group = group->previous;
2184 from = group ? start + group->strbeg : start;
2185 }
2186 fromstr = NEXTFROM;
2187 len = SvIV(fromstr);
2188 goto resize;
28be1210 2189 case '@' | TYPE_IS_SHRIEKING:
a6ec74c1 2190 case '@':
28be1210
TH
2191 from = start + symptr->strbeg;
2192 resize:
28be1210 2193 if (utf8 && !(datumtype & TYPE_IS_SHRIEKING))
28be1210
TH
2194 if (len >= 0) {
2195 while (len && from < cur) {
2196 from += UTF8SKIP(from);
2197 len--;
2198 }
2199 if (from > cur)
2200 Perl_croak(aTHX_ "Malformed UTF-8 string in pack");
2201 if (len) {
2202 /* Here we know from == cur */
2203 grow:
2204 GROWING(0, cat, start, cur, len);
2205 Zero(cur, len, char);
2206 cur += len;
2207 } else if (from < cur) {
2208 len = cur - from;
2209 goto shrink;
2210 } else goto no_change;
2211 } else {
2212 cur = from;
2213 len = -len;
2214 goto utf8_shrink;
f337b084 2215 }
28be1210
TH
2216 else {
2217 len -= cur - from;
f337b084 2218 if (len > 0) goto grow;
28be1210 2219 if (len == 0) goto no_change;
fc241834 2220 len = -len;
28be1210 2221 goto shrink;
f337b084 2222 }
a6ec74c1 2223 break;
81d52ecd 2224
fc241834 2225 case '(': {
49704364 2226 tempsym_t savsym = *symptr;
66c611c5
MHM
2227 U32 group_modifiers = TYPE_MODIFIERS(datumtype & ~symptr->flags);
2228 symptr->flags |= group_modifiers;
49704364
WL
2229 symptr->patend = savsym.grpend;
2230 symptr->level++;
f337b084 2231 symptr->previous = &lookahead;
18529408 2232 while (len--) {
f337b084
TH
2233 U32 was_utf8;
2234 if (utf8) symptr->flags |= FLAG_PARSE_UTF8;
2235 else symptr->flags &= ~FLAG_PARSE_UTF8;
2236 was_utf8 = SvUTF8(cat);
49704364 2237 symptr->patptr = savsym.grpbeg;
f337b084
TH
2238 beglist = pack_rec(cat, symptr, beglist, endlist);
2239 if (SvUTF8(cat) != was_utf8)
2240 /* This had better be an upgrade while in utf8==0 mode */
2241 utf8 = 1;
2242
49704364 2243 if (savsym.howlen == e_star && beglist == endlist)
18529408
IZ
2244 break; /* No way to continue */
2245 }
ee790063 2246 items = endlist - beglist;
f337b084
TH
2247 lookahead.flags = symptr->flags & ~group_modifiers;
2248 goto no_change;
18529408 2249 }
62f95557
IZ
2250 case 'X' | TYPE_IS_SHRIEKING:
2251 if (!len) /* Avoid division by 0 */
2252 len = 1;
f337b084
TH
2253 if (utf8) {
2254 char *hop, *last;
2255 I32 l = len;
2256 hop = last = start;
2257 while (hop < cur) {
2258 hop += UTF8SKIP(hop);
2259 if (--l == 0) {
2260 last = hop;
2261 l = len;
2262 }
2263 }
2264 if (last > cur)
2265 Perl_croak(aTHX_ "Malformed UTF-8 string in pack");
2266 cur = last;
2267 break;
2268 }
2269 len = (cur-start) % len;
924ba076 2270 /* FALLTHROUGH */
a6ec74c1 2271 case 'X':
f337b084
TH
2272 if (utf8) {
2273 if (len < 1) goto no_change;
28be1210 2274 utf8_shrink:
f337b084
TH
2275 while (len > 0) {
2276 if (cur <= start)
28be1210
TH
2277 Perl_croak(aTHX_ "'%c' outside of string in pack",
2278 (int) TYPE_NO_MODIFIERS(datumtype));
f337b084
TH
2279 while (--cur, UTF8_IS_CONTINUATION(*cur)) {
2280 if (cur <= start)
28be1210
TH
2281 Perl_croak(aTHX_ "'%c' outside of string in pack",
2282 (int) TYPE_NO_MODIFIERS(datumtype));
f337b084
TH
2283 }
2284 len--;
2285 }
2286 } else {
fc241834 2287 shrink:
f337b084 2288 if (cur - start < len)
28be1210
TH
2289 Perl_croak(aTHX_ "'%c' outside of string in pack",
2290 (int) TYPE_NO_MODIFIERS(datumtype));
f337b084
TH
2291 cur -= len;
2292 }
2293 if (cur < start+symptr->strbeg) {
2294 /* Make sure group starts don't point into the void */
2295 tempsym_t *group;
9e27e96a 2296 const STRLEN length = cur-start;
f337b084
TH
2297 for (group = symptr;
2298 group && length < group->strbeg;
2299 group = group->previous) group->strbeg = length;
2300 lookahead.strbeg = length;
2301 }
a6ec74c1 2302 break;
fc241834
RGS
2303 case 'x' | TYPE_IS_SHRIEKING: {
2304 I32 ai32;
62f95557
IZ
2305 if (!len) /* Avoid division by 0 */
2306 len = 1;
230e1fce 2307 if (utf8) ai32 = utf8_length((U8 *) start, (U8 *) cur) % len;
fc241834
RGS
2308 else ai32 = (cur - start) % len;
2309 if (ai32 == 0) goto no_change;
2310 len -= ai32;
2311 }
924ba076 2312 /* FALLTHROUGH */
a6ec74c1 2313 case 'x':
f337b084 2314 goto grow;
a6ec74c1
JH
2315 case 'A':
2316 case 'Z':
f337b084 2317 case 'a': {
f7fe979e 2318 const char *aptr;
f337b084 2319
a6ec74c1 2320 fromstr = NEXTFROM;
e62f0680 2321 aptr = SvPV_const(fromstr, fromlen);
f337b084 2322 if (DO_UTF8(fromstr)) {
f7fe979e 2323 const char *end, *s;
f337b084
TH
2324
2325 if (!utf8 && !SvUTF8(cat)) {
2326 marked_upgrade(aTHX_ cat, symptr);
2327 lookahead.flags |= FLAG_DO_UTF8;
2328 lookahead.strbeg = symptr->strbeg;
2329 utf8 = 1;
2330 start = SvPVX(cat);
2331 cur = start + SvCUR(cat);
2332 }
fc241834 2333 if (howlen == e_star) {
f337b084
TH
2334 if (utf8) goto string_copy;
2335 len = fromlen+1;
2336 }
2337 s = aptr;
2338 end = aptr + fromlen;
2339 fromlen = datumtype == 'Z' ? len-1 : len;
2340 while ((I32) fromlen > 0 && s < end) {
2341 s += UTF8SKIP(s);
2342 fromlen--;
2343 }
2344 if (s > end)
2345 Perl_croak(aTHX_ "Malformed UTF-8 string in pack");
2346 if (utf8) {
fc241834 2347 len = fromlen;
f337b084
TH
2348 if (datumtype == 'Z') len++;
2349 fromlen = s-aptr;
2350 len += fromlen;
fc241834 2351
f337b084 2352 goto string_copy;
fc241834 2353 }
f337b084
TH
2354 fromlen = len - fromlen;
2355 if (datumtype == 'Z') fromlen--;
2356 if (howlen == e_star) {
2357 len = fromlen;
2358 if (datumtype == 'Z') len++;
fc241834 2359 }
f337b084 2360 GROWING(0, cat, start, cur, len);
fc241834 2361 if (!uni_to_bytes(aTHX_ &aptr, end, cur, fromlen,
f337b084 2362 datumtype | TYPE_IS_PACK))
5637ef5b
NC
2363 Perl_croak(aTHX_ "panic: predicted utf8 length not available, "
2364 "for '%c', aptr=%p end=%p cur=%p, fromlen=%"UVuf,
2365 (int)datumtype, aptr, end, cur, (UV)fromlen);
f337b084 2366 cur += fromlen;
a6ec74c1 2367 len -= fromlen;
f337b084
TH
2368 } else if (utf8) {
2369 if (howlen == e_star) {
2370 len = fromlen;
2371 if (datumtype == 'Z') len++;
a6ec74c1 2372 }
f337b084
TH
2373 if (len <= (I32) fromlen) {
2374 fromlen = len;
2375 if (datumtype == 'Z' && fromlen > 0) fromlen--;
2376 }
fc241834 2377 /* assumes a byte expands to at most UTF8_EXPAND bytes on
3473cf63
RGS
2378 upgrade, so:
2379 expected_length <= from_len*UTF8_EXPAND + (len-from_len) */
2380 GROWING(0, cat, start, cur, fromlen*(UTF8_EXPAND-1)+len);
f337b084
TH
2381 len -= fromlen;
2382 while (fromlen > 0) {
230e1fce 2383 cur = (char *) uvchr_to_utf8((U8 *) cur, * (U8 *) aptr);
f337b084
TH
2384 aptr++;
2385 fromlen--;
fc241834 2386 }
f337b084
TH
2387 } else {
2388 string_copy:
2389 if (howlen == e_star) {
2390 len = fromlen;
2391 if (datumtype == 'Z') len++;
2392 }
2393 if (len <= (I32) fromlen) {
2394 fromlen = len;
2395 if (datumtype == 'Z' && fromlen > 0) fromlen--;
a6ec74c1 2396 }
f337b084
TH
2397 GROWING(0, cat, start, cur, len);
2398 Copy(aptr, cur, fromlen, char);
2399 cur += fromlen;
2400 len -= fromlen;
a6ec74c1 2401 }
f337b084
TH
2402 memset(cur, datumtype == 'A' ? ' ' : '\0', len);
2403 cur += len;
3c4fb04a 2404 SvTAINT(cat);
a6ec74c1 2405 break;
f337b084 2406 }
a6ec74c1 2407 case 'B':
f337b084 2408 case 'b': {
b83604b4 2409 const char *str, *end;
f337b084
TH
2410 I32 l, field_len;
2411 U8 bits;
2412 bool utf8_source;
2413 U32 utf8_flags;
a6ec74c1 2414
fc241834 2415 fromstr = NEXTFROM;
b83604b4 2416 str = SvPV_const(fromstr, fromlen);
f337b084
TH
2417 end = str + fromlen;
2418 if (DO_UTF8(fromstr)) {
2419 utf8_source = TRUE;
041457d9 2420 utf8_flags = warn_utf8 ? 0 : UTF8_ALLOW_ANY;
f337b084
TH
2421 } else {
2422 utf8_source = FALSE;
2423 utf8_flags = 0; /* Unused, but keep compilers happy */
2424 }
2425 if (howlen == e_star) len = fromlen;
2426 field_len = (len+7)/8;
2427 GROWING(utf8, cat, start, cur, field_len);
2428 if (len > (I32)fromlen) len = fromlen;
2429 bits = 0;
2430 l = 0;
2431 if (datumtype == 'B')
2432 while (l++ < len) {
2433 if (utf8_source) {
95b63a38 2434 UV val = 0;
f337b084
TH
2435 NEXT_UNI_VAL(val, cur, str, end, utf8_flags);
2436 bits |= val & 1;
2437 } else bits |= *str++ & 1;
2438 if (l & 7) bits <<= 1;
fc241834 2439 else {
f337b084
TH
2440 PUSH_BYTE(utf8, cur, bits);
2441 bits = 0;
a6ec74c1
JH
2442 }
2443 }
f337b084
TH
2444 else
2445 /* datumtype == 'b' */
2446 while (l++ < len) {
2447 if (utf8_source) {
95b63a38 2448 UV val = 0;
f337b084
TH
2449 NEXT_UNI_VAL(val, cur, str, end, utf8_flags);
2450 if (val & 1) bits |= 0x80;
2451 } else if (*str++ & 1)
2452 bits |= 0x80;
2453 if (l & 7) bits >>= 1;
fc241834 2454 else {
f337b084
TH
2455 PUSH_BYTE(utf8, cur, bits);
2456 bits = 0;
a6ec74c1
JH
2457 }
2458 }
f337b084
TH
2459 l--;
2460 if (l & 7) {
fc241834 2461 if (datumtype == 'B')
f337b084 2462 bits <<= 7 - (l & 7);
fc241834 2463 else
f337b084
TH
2464 bits >>= 7 - (l & 7);
2465 PUSH_BYTE(utf8, cur, bits);
2466 l += 7;
a6ec74c1 2467 }
f337b084
TH
2468 /* Determine how many chars are left in the requested field */
2469 l /= 8;
2470 if (howlen == e_star) field_len = 0;
2471 else field_len -= l;
2472 Zero(cur, field_len, char);
2473 cur += field_len;
a6ec74c1 2474 break;
f337b084 2475 }
a6ec74c1 2476 case 'H':
f337b084 2477 case 'h': {
10516c54 2478 const char *str, *end;
f337b084
TH
2479 I32 l, field_len;
2480 U8 bits;
2481 bool utf8_source;
2482 U32 utf8_flags;
a6ec74c1 2483
fc241834 2484 fromstr = NEXTFROM;
10516c54 2485 str = SvPV_const(fromstr, fromlen);
f337b084
TH
2486 end = str + fromlen;
2487 if (DO_UTF8(fromstr)) {
2488 utf8_source = TRUE;
041457d9 2489 utf8_flags = warn_utf8 ? 0 : UTF8_ALLOW_ANY;
f337b084
TH
2490 } else {
2491 utf8_source = FALSE;
2492 utf8_flags = 0; /* Unused, but keep compilers happy */
2493 }
2494 if (howlen == e_star) len = fromlen;
2495 field_len = (len+1)/2;
2496 GROWING(utf8, cat, start, cur, field_len);
2497 if (!utf8 && len > (I32)fromlen) len = fromlen;
2498 bits = 0;
2499 l = 0;
2500 if (datumtype == 'H')
2501 while (l++ < len) {
2502 if (utf8_source) {
95b63a38 2503 UV val = 0;
f337b084
TH
2504 NEXT_UNI_VAL(val, cur, str, end, utf8_flags);
2505 if (val < 256 && isALPHA(val))
2506 bits |= (val + 9) & 0xf;
a6ec74c1 2507 else
f337b084
TH
2508 bits |= val & 0xf;
2509 } else if (isALPHA(*str))
2510 bits |= (*str++ + 9) & 0xf;
2511 else
2512 bits |= *str++ & 0xf;
2513 if (l & 1) bits <<= 4;
fc241834 2514 else {
f337b084
TH
2515 PUSH_BYTE(utf8, cur, bits);
2516 bits = 0;
a6ec74c1
JH
2517 }
2518 }
f337b084
TH
2519 else
2520 while (l++ < len) {
2521 if (utf8_source) {
95b63a38 2522 UV val = 0;
f337b084
TH
2523 NEXT_UNI_VAL(val, cur, str, end, utf8_flags);
2524 if (val < 256 && isALPHA(val))
2525 bits |= ((val + 9) & 0xf) << 4;
a6ec74c1 2526 else
f337b084
TH
2527 bits |= (val & 0xf) << 4;
2528 } else if (isALPHA(*str))
2529 bits |= ((*str++ + 9) & 0xf) << 4;
2530 else
2531 bits |= (*str++ & 0xf) << 4;
2532 if (l & 1) bits >>= 4;
fc241834 2533 else {
f337b084
TH
2534 PUSH_BYTE(utf8, cur, bits);
2535 bits = 0;
a6ec74c1 2536 }
fc241834 2537 }
f337b084
TH
2538 l--;
2539 if (l & 1) {
2540 PUSH_BYTE(utf8, cur, bits);
2541 l++;
2542 }
2543 /* Determine how many chars are left in the requested field */
2544 l /= 2;
2545 if (howlen == e_star) field_len = 0;
2546 else field_len -= l;
2547 Zero(cur, field_len, char);
2548 cur += field_len;
2549 break;
fc241834
RGS
2550 }
2551 case 'c':
f337b084
TH
2552 while (len-- > 0) {
2553 IV aiv;
2554 fromstr = NEXTFROM;
1f4ef0f1
JH
2555 if (SvNOK(fromstr) && Perl_isinfnan(SvNV(fromstr))) {
2556 /* 255 is a pretty arbitrary choice, but with
2557 * inf/-inf/nan and 256 bytes there is not much room. */
2558 aiv = 255;
2559 Perl_ck_warner(aTHX_ packWARN(WARN_PACK),
2560 "Character in 'c' format overflow in pack");
2561 }
2562 else
2563 aiv = SvIV(fromstr);
a2a5de95
NC
2564 if ((-128 > aiv || aiv > 127))
2565 Perl_ck_warner(aTHX_ packWARN(WARN_PACK),
2566 "Character in 'c' format wrapped in pack");
585ec06d 2567 PUSH_BYTE(utf8, cur, (U8)(aiv & 0xff));
a6ec74c1
JH
2568 }
2569 break;
2570 case 'C':
f337b084
TH
2571 if (len == 0) {
2572 utf8 = (symptr->flags & FLAG_DO_UTF8) ? 1 : 0;
2573 break;
2574 }
a6ec74c1 2575 while (len-- > 0) {
f337b084 2576 IV aiv;
a6ec74c1 2577 fromstr = NEXTFROM;
1f4ef0f1
JH
2578 if (SvNOK(fromstr) && Perl_isinfnan(SvNV(fromstr))) {
2579 /* See the 'c' case. */
2580 aiv = 255;
2581 Perl_ck_warner(aTHX_ packWARN(WARN_PACK),
2582 "Character in 'C' format overflow in pack");
2583 }
2584 else
2585 aiv = SvIV(fromstr);
a2a5de95
NC
2586 if ((0 > aiv || aiv > 0xff))
2587 Perl_ck_warner(aTHX_ packWARN(WARN_PACK),
2588 "Character in 'C' format wrapped in pack");
1651fc44 2589 PUSH_BYTE(utf8, cur, (U8)(aiv & 0xff));
f337b084 2590 }
fc241834
RGS
2591 break;
2592 case 'W': {
2593 char *end;
670f1322 2594 U8 in_bytes = (U8)IN_BYTES;
fc241834
RGS
2595
2596 end = start+SvLEN(cat)-1;
2597 if (utf8) end -= UTF8_MAXLEN-1;
2598 while (len-- > 0) {
2599 UV auv;
2600 fromstr = NEXTFROM;
2601 auv = SvUV(fromstr);
2602 if (in_bytes) auv = auv % 0x100;
2603 if (utf8) {
2604 W_utf8:
2605 if (cur > end) {
2606 *cur = '\0';
b162af07 2607 SvCUR_set(cat, cur - start);
fc241834
RGS
2608
2609 GROWING(0, cat, start, cur, len+UTF8_MAXLEN);
2610 end = start+SvLEN(cat)-UTF8_MAXLEN;
2611 }
c80e42f3
KW
2612 cur = (char *) uvchr_to_utf8_flags((U8 *) cur,
2613 auv,
041457d9 2614 warn_utf8 ?
230e1fce 2615 0 : UNICODE_ALLOW_ANY);
fc241834
RGS
2616 } else {
2617 if (auv >= 0x100) {
2618 if (!SvUTF8(cat)) {
2619 *cur = '\0';
b162af07 2620 SvCUR_set(cat, cur - start);
fc241834
RGS
2621 marked_upgrade(aTHX_ cat, symptr);
2622 lookahead.flags |= FLAG_DO_UTF8;
2623 lookahead.strbeg = symptr->strbeg;
2624 utf8 = 1;
2625 start = SvPVX(cat);
2626 cur = start + SvCUR(cat);
2627 end = start+SvLEN(cat)-UTF8_MAXLEN;
2628 goto W_utf8;
2629 }
a2a5de95
NC
2630 Perl_ck_warner(aTHX_ packWARN(WARN_PACK),
2631 "Character in 'W' format wrapped in pack");
fc241834
RGS
2632 auv &= 0xff;
2633 }
2634 if (cur >= end) {
2635 *cur = '\0';
b162af07 2636 SvCUR_set(cat, cur - start);
fc241834
RGS
2637 GROWING(0, cat, start, cur, len+1);
2638 end = start+SvLEN(cat)-1;
2639 }
fe2774ed 2640 *(U8 *) cur++ = (U8)auv;
a6ec74c1
JH
2641 }
2642 }
2643 break;
fc241834
RGS
2644 }
2645 case 'U': {
2646 char *end;
2647
2648 if (len == 0) {
2649 if (!(symptr->flags & FLAG_DO_UTF8)) {
2650 marked_upgrade(aTHX_ cat, symptr);
2651 lookahead.flags |= FLAG_DO_UTF8;
2652 lookahead.strbeg = symptr->strbeg;
2653 }
2654 utf8 = 0;
2655 goto no_change;
2656 }
2657
2658 end = start+SvLEN(cat);
2659 if (!utf8) end -= UTF8_MAXLEN;
a6ec74c1 2660 while (len-- > 0) {
fc241834 2661 UV auv;
a6ec74c1 2662 fromstr = NEXTFROM;
fc241834
RGS
2663 auv = SvUV(fromstr);
2664 if (utf8) {
230e1fce 2665 U8 buffer[UTF8_MAXLEN], *endb;
c80e42f3 2666 endb = uvchr_to_utf8_flags(buffer, auv,
041457d9 2667 warn_utf8 ?
fc241834
RGS
2668 0 : UNICODE_ALLOW_ANY);
2669 if (cur+(endb-buffer)*UTF8_EXPAND >= end) {
2670 *cur = '\0';
b162af07 2671 SvCUR_set(cat, cur - start);
fc241834
RGS
2672 GROWING(0, cat, start, cur,
2673 len+(endb-buffer)*UTF8_EXPAND);
2674 end = start+SvLEN(cat);
2675 }
3a88beaa 2676 cur = S_bytes_to_uni(buffer, endb-buffer, cur, 0);
fc241834
RGS
2677 } else {
2678 if (cur >= end) {
2679 *cur = '\0';
b162af07 2680 SvCUR_set(cat, cur - start);
fc241834
RGS
2681 GROWING(0, cat, start, cur, len+UTF8_MAXLEN);
2682 end = start+SvLEN(cat)-UTF8_MAXLEN;
2683 }
c80e42f3 2684 cur = (char *) uvchr_to_utf8_flags((U8 *) cur, auv,
041457d9 2685 warn_utf8 ?
230e1fce 2686 0 : UNICODE_ALLOW_ANY);
fc241834 2687 }
a6ec74c1 2688 }
a6ec74c1 2689 break;
fc241834 2690 }
a6ec74c1
JH
2691 /* Float and double added by gnb@melba.bby.oz.au 22/11/89 */
2692 case 'f':
a6ec74c1 2693 while (len-- > 0) {
f337b084
TH
2694 float afloat;
2695 NV anv;
a6ec74c1 2696 fromstr = NEXTFROM;
f337b084 2697 anv = SvNV(fromstr);
85bba25f 2698# if defined(VMS) && !defined(_IEEE_FP)
f337b084 2699 /* IEEE fp overflow shenanigans are unavailable on VAX and optional
fc241834
RGS
2700 * on Alpha; fake it if we don't have them.
2701 */
f337b084 2702 if (anv > FLT_MAX)
fc241834 2703 afloat = FLT_MAX;
f337b084 2704 else if (anv < -FLT_MAX)
fc241834 2705 afloat = -FLT_MAX;
f337b084 2706 else afloat = (float)anv;
baf3cf9c 2707# else
f337b084 2708 afloat = (float)anv;
baf3cf9c 2709# endif
3a88beaa 2710 PUSH_VAR(utf8, cur, afloat, needs_swap);
a6ec74c1
JH
2711 }
2712 break;
2713 case 'd':
a6ec74c1 2714 while (len-- > 0) {
f337b084
TH
2715 double adouble;
2716 NV anv;
a6ec74c1 2717 fromstr = NEXTFROM;
f337b084 2718 anv = SvNV(fromstr);
85bba25f 2719# if defined(VMS) && !defined(_IEEE_FP)
f337b084 2720 /* IEEE fp overflow shenanigans are unavailable on VAX and optional
fc241834
RGS
2721 * on Alpha; fake it if we don't have them.
2722 */
f337b084 2723 if (anv > DBL_MAX)
fc241834 2724 adouble = DBL_MAX;
f337b084 2725 else if (anv < -DBL_MAX)
fc241834 2726 adouble = -DBL_MAX;
f337b084 2727 else adouble = (double)anv;
baf3cf9c 2728# else
f337b084 2729 adouble = (double)anv;
baf3cf9c 2730# endif
3a88beaa 2731 PUSH_VAR(utf8, cur, adouble, needs_swap);
a6ec74c1
JH
2732 }
2733 break;
fc241834 2734 case 'F': {
275663fa 2735 NV_bytes anv;
1109a392 2736 Zero(&anv, 1, NV); /* can be long double with unused bits */
92d41999
JH
2737 while (len-- > 0) {
2738 fromstr = NEXTFROM;
cd07c537
DM
2739#ifdef __GNUC__
2740 /* to work round a gcc/x86 bug; don't use SvNV */
2741 anv.nv = sv_2nv(fromstr);
2742#else
275663fa 2743 anv.nv = SvNV(fromstr);
cd07c537 2744#endif
3a88beaa 2745 PUSH_BYTES(utf8, cur, anv.bytes, sizeof(anv.bytes), needs_swap);
92d41999
JH
2746 }
2747 break;
fc241834 2748 }
92d41999 2749#if defined(HAS_LONG_DOUBLE) && defined(USE_LONG_DOUBLE)
fc241834 2750 case 'D': {
275663fa 2751 ld_bytes aldouble;
1109a392
MHM
2752 /* long doubles can have unused bits, which may be nonzero */
2753 Zero(&aldouble, 1, long double);
92d41999
JH
2754 while (len-- > 0) {
2755 fromstr = NEXTFROM;
cd07c537
DM
2756# ifdef __GNUC__
2757 /* to work round a gcc/x86 bug; don't use SvNV */
2758 aldouble.ld = (long double)sv_2nv(fromstr);
2759# else
275663fa 2760 aldouble.ld = (long double)SvNV(fromstr);
cd07c537 2761# endif
3a88beaa
NC
2762 PUSH_BYTES(utf8, cur, aldouble.bytes, sizeof(aldouble.bytes),
2763 needs_swap);
92d41999
JH
2764 }
2765 break;
fc241834 2766 }
92d41999 2767#endif
068bd2e7 2768 case 'n' | TYPE_IS_SHRIEKING:
a6ec74c1
JH
2769 case 'n':
2770 while (len-- > 0) {
f337b084 2771 I16 ai16;
a6ec74c1 2772 fromstr = NEXTFROM;
ef108786 2773 ai16 = (I16)SvIV(fromstr);
ef108786 2774 ai16 = PerlSock_htons(ai16);
3a88beaa 2775 PUSH16(utf8, cur, &ai16, FALSE);
a6ec74c1
JH
2776 }
2777 break;
068bd2e7 2778 case 'v' | TYPE_IS_SHRIEKING:
a6ec74c1
JH
2779 case 'v':
2780 while (len-- > 0) {
f337b084 2781 I16 ai16;
a6ec74c1 2782 fromstr = NEXTFROM;
ef108786 2783 ai16 = (I16)SvIV(fromstr);
ef108786 2784 ai16 = htovs(ai16);
3a88beaa 2785 PUSH16(utf8, cur, &ai16, FALSE);
a6ec74c1
JH
2786 }
2787 break;
49704364 2788 case 'S' | TYPE_IS_SHRIEKING:
a6ec74c1 2789#if SHORTSIZE != SIZE16
fc241834 2790 while (len-- > 0) {
f337b084 2791 unsigned short aushort;
fc241834
RGS
2792 fromstr = NEXTFROM;
2793 aushort = SvUV(fromstr);
3a88beaa 2794 PUSH_VAR(utf8, cur, aushort, needs_swap);
fc241834 2795 }
49704364
WL
2796 break;
2797#else
924ba076 2798 /* FALLTHROUGH */
a6ec74c1 2799#endif
49704364 2800 case 'S':
fc241834 2801 while (len-- > 0) {
f337b084 2802 U16 au16;
fc241834
RGS
2803 fromstr = NEXTFROM;
2804 au16 = (U16)SvUV(fromstr);
3a88beaa 2805 PUSH16(utf8, cur, &au16, needs_swap);
a6ec74c1
JH
2806 }
2807 break;
49704364 2808 case 's' | TYPE_IS_SHRIEKING:
a6ec74c1 2809#if SHORTSIZE != SIZE16
fc241834 2810 while (len-- > 0) {
f337b084 2811 short ashort;
fc241834
RGS
2812 fromstr = NEXTFROM;
2813 ashort = SvIV(fromstr);
3a88beaa 2814 PUSH_VAR(utf8, cur, ashort, needs_swap);
a6ec74c1 2815 }
49704364
WL
2816 break;
2817#else
924ba076 2818 /* FALLTHROUGH */
a6ec74c1 2819#endif
49704364
WL
2820 case 's':
2821 while (len-- > 0) {
f337b084 2822 I16 ai16;
49704364 2823 fromstr = NEXTFROM;
ef108786 2824 ai16 = (I16)SvIV(fromstr);
3a88beaa 2825 PUSH16(utf8, cur, &ai16, needs_swap);
a6ec74c1
JH
2826 }
2827 break;
2828 case 'I':
49704364 2829 case 'I' | TYPE_IS_SHRIEKING:
a6ec74c1 2830 while (len-- > 0) {
f337b084 2831 unsigned int auint;
a6ec74c1
JH
2832 fromstr = NEXTFROM;
2833 auint = SvUV(fromstr);
3a88beaa 2834 PUSH_VAR(utf8, cur, auint, needs_swap);
a6ec74c1
JH
2835 }
2836 break;
92d41999
JH
2837 case 'j':
2838 while (len-- > 0) {
f337b084 2839 IV aiv;
92d41999
JH
2840 fromstr = NEXTFROM;
2841 aiv = SvIV(fromstr);
3a88beaa 2842 PUSH_VAR(utf8, cur, aiv, needs_swap);
92d41999
JH
2843 }
2844 break;
2845 case 'J':
2846 while (len-- > 0) {
f337b084 2847 UV auv;
92d41999
JH
2848 fromstr = NEXTFROM;
2849 auv = SvUV(fromstr);
3a88beaa 2850 PUSH_VAR(utf8, cur, auv, needs_swap);
92d41999
JH
2851 }
2852 break;
a6ec74c1
JH
2853 case 'w':
2854 while (len-- > 0) {
f337b084 2855 NV anv;
a6ec74c1 2856 fromstr = NEXTFROM;
15e9f109 2857 anv = SvNV(fromstr);
a6ec74c1 2858
f337b084
TH
2859 if (anv < 0) {
2860 *cur = '\0';
b162af07 2861 SvCUR_set(cat, cur - start);
49704364 2862 Perl_croak(aTHX_ "Cannot compress negative numbers in pack");
f337b084 2863 }
a6ec74c1 2864
196b62db
NC
2865 /* 0xFFFFFFFFFFFFFFFF may cast to 18446744073709551616.0,
2866 which is == UV_MAX_P1. IOK is fine (instead of UV_only), as
2867 any negative IVs will have already been got by the croak()
2868 above. IOK is untrue for fractions, so we test them
2869 against UV_MAX_P1. */
f337b084
TH
2870 if (SvIOK(fromstr) || anv < UV_MAX_P1) {
2871 char buf[(sizeof(UV)*CHAR_BIT)/7+1];
a6ec74c1 2872 char *in = buf + sizeof(buf);
196b62db 2873 UV auv = SvUV(fromstr);
a6ec74c1
JH
2874
2875 do {
eb160463 2876 *--in = (char)((auv & 0x7f) | 0x80);
a6ec74c1
JH
2877 auv >>= 7;
2878 } while (auv);
2879 buf[sizeof(buf) - 1] &= 0x7f; /* clear continue bit */
f337b084
TH
2880 PUSH_GROWING_BYTES(utf8, cat, start, cur,
2881 in, (buf + sizeof(buf)) - in);
2882 } else if (SvPOKp(fromstr))
2883 goto w_string;
a6ec74c1 2884 else if (SvNOKp(fromstr)) {
0258719b 2885 /* 10**NV_MAX_10_EXP is the largest power of 10
486ec47a 2886 so 10**(NV_MAX_10_EXP+1) is definitely unrepresentable
0258719b
NC
2887 given 10**(NV_MAX_10_EXP+1) == 128 ** x solve for x:
2888 x = (NV_MAX_10_EXP+1) * log (10) / log (128)
2889 And with that many bytes only Inf can overflow.
8f8d40ab
PG
2890 Some C compilers are strict about integral constant
2891 expressions so we conservatively divide by a slightly
2892 smaller integer instead of multiplying by the exact
2893 floating-point value.
0258719b
NC
2894 */
2895#ifdef NV_MAX_10_EXP
f337b084 2896 /* char buf[1 + (int)((NV_MAX_10_EXP + 1) * 0.47456)]; -- invalid C */
8f8d40ab 2897 char buf[1 + (int)((NV_MAX_10_EXP + 1) / 2)]; /* valid C */
0258719b 2898#else
f337b084 2899 /* char buf[1 + (int)((308 + 1) * 0.47456)]; -- invalid C */
8f8d40ab 2900 char buf[1 + (int)((308 + 1) / 2)]; /* valid C */
0258719b 2901#endif
a6ec74c1
JH
2902 char *in = buf + sizeof(buf);
2903
8b6e33c7 2904 anv = Perl_floor(anv);
a6ec74c1 2905 do {
8b6e33c7 2906 const NV next = Perl_floor(anv / 128);
a6ec74c1 2907 if (in <= buf) /* this cannot happen ;-) */
49704364 2908 Perl_croak(aTHX_ "Cannot compress integer in pack");
0258719b 2909 *--in = (unsigned char)(anv - (next * 128)) | 0x80;
15e9f109
NC
2910 anv = next;
2911 } while (anv > 0);
a6ec74c1 2912 buf[sizeof(buf) - 1] &= 0x7f; /* clear continue bit */
f337b084
TH
2913 PUSH_GROWING_BYTES(utf8, cat, start, cur,
2914 in, (buf + sizeof(buf)) - in);
2915 } else {
8b6e33c7
AL
2916 const char *from;
2917 char *result, *in;
735b914b
JH
2918 SV *norm;
2919 STRLEN len;
2920 bool done;
2921
f337b084 2922 w_string:
735b914b 2923 /* Copy string and check for compliance */
349d4f2f 2924 from = SvPV_const(fromstr, len);
735b914b 2925 if ((norm = is_an_int(from, len)) == NULL)
49704364 2926 Perl_croak(aTHX_ "Can only compress unsigned integers in pack");
735b914b 2927
a02a5408 2928 Newx(result, len, char);
735b914b
JH
2929 in = result + len;
2930 done = FALSE;
f337b084 2931 while (!done) *--in = div128(norm, &done) | 0x80;
735b914b 2932 result[len - 1] &= 0x7F; /* clear continue bit */
f337b084
TH
2933 PUSH_GROWING_BYTES(utf8, cat, start, cur,
2934 in, (result + len) - in);
735b914b
JH
2935 Safefree(result);
2936 SvREFCNT_dec(norm); /* free norm */
fc241834 2937 }
a6ec74c1
JH
2938 }
2939 break;
2940 case 'i':
49704364 2941 case 'i' | TYPE_IS_SHRIEKING:
a6ec74c1 2942 while (len-- > 0) {
f337b084 2943 int aint;
a6ec74c1
JH
2944 fromstr = NEXTFROM;
2945 aint = SvIV(fromstr);
3a88beaa 2946 PUSH_VAR(utf8, cur, aint, needs_swap);
a6ec74c1
JH
2947 }
2948 break;
068bd2e7 2949 case 'N' | TYPE_IS_SHRIEKING:
a6ec74c1
JH
2950 case 'N':
2951 while (len-- > 0) {
f337b084 2952 U32 au32;
a6ec74c1 2953 fromstr = NEXTFROM;
ef108786 2954 au32 = SvUV(fromstr);
ef108786 2955 au32 = PerlSock_htonl(au32);
3a88beaa 2956 PUSH32(utf8, cur, &au32, FALSE);
a6ec74c1
JH
2957 }
2958 break;
068bd2e7 2959 case 'V' | TYPE_IS_SHRIEKING:
a6ec74c1
JH
2960 case 'V':
2961 while (len-- > 0) {
f337b084 2962 U32 au32;
a6ec74c1 2963 fromstr = NEXTFROM;
ef108786 2964 au32 = SvUV(fromstr);
ef108786 2965 au32 = htovl(au32);
3a88beaa 2966 PUSH32(utf8, cur, &au32, FALSE);
a6ec74c1
JH
2967 }
2968 break;
49704364 2969 case 'L' | TYPE_IS_SHRIEKING:
a6ec74c1 2970#if LONGSIZE != SIZE32
fc241834 2971 while (len-- > 0) {
f337b084 2972 unsigned long aulong;
fc241834
RGS
2973 fromstr = NEXTFROM;
2974 aulong = SvUV(fromstr);
3a88beaa 2975 PUSH_VAR(utf8, cur, aulong, needs_swap);
a6ec74c1 2976 }
49704364
WL
2977 break;
2978#else
2979 /* Fall though! */
a6ec74c1 2980#endif
49704364 2981 case 'L':
fc241834 2982 while (len-- > 0) {
f337b084 2983 U32 au32;
fc241834
RGS
2984 fromstr = NEXTFROM;
2985 au32 = SvUV(fromstr);
3a88beaa 2986 PUSH32(utf8, cur, &au32, needs_swap);
a6ec74c1
JH
2987 }
2988 break;
49704364 2989 case 'l' | TYPE_IS_SHRIEKING:
a6ec74c1 2990#if LONGSIZE != SIZE32
fc241834 2991 while (len-- > 0) {
f337b084 2992 long along;
fc241834
RGS
2993 fromstr = NEXTFROM;
2994 along = SvIV(fromstr);
3a88beaa 2995 PUSH_VAR(utf8, cur, along, needs_swap);
a6ec74c1 2996 }
49704364
WL
2997 break;
2998#else
2999 /* Fall though! */
a6ec74c1 3000#endif
49704364
WL
3001 case 'l':
3002 while (len-- > 0) {
f337b084 3003 I32 ai32;
49704364 3004 fromstr = NEXTFROM;
ef108786 3005 ai32 = SvIV(fromstr);
3a88beaa 3006 PUSH32(utf8, cur, &ai32, needs_swap);
a6ec74c1
JH
3007 }
3008 break;
c174bf3b 3009#if defined(HAS_QUAD) && IVSIZE >= 8
a6ec74c1
JH
3010 case 'Q':
3011 while (len-- > 0) {
f337b084 3012 Uquad_t auquad;
a6ec74c1 3013 fromstr = NEXTFROM;
f337b084 3014 auquad = (Uquad_t) SvUV(fromstr);
3a88beaa 3015 PUSH_VAR(utf8, cur, auquad, needs_swap);
a6ec74c1
JH
3016 }
3017 break;
3018 case 'q':
3019 while (len-- > 0) {
f337b084 3020 Quad_t aquad;
a6ec74c1
JH
3021 fromstr = NEXTFROM;
3022 aquad = (Quad_t)SvIV(fromstr);
3a88beaa 3023 PUSH_VAR(utf8, cur, aquad, needs_swap);
a6ec74c1
JH
3024 }
3025 break;
1640b983 3026#endif
a6ec74c1
JH
3027 case 'P':
3028 len = 1; /* assume SV is correct length */
f337b084 3029 GROWING(utf8, cat, start, cur, sizeof(char *));
924ba076 3030 /* FALLTHROUGH */
a6ec74c1
JH
3031 case 'p':
3032 while (len-- > 0) {
83003860 3033 const char *aptr;
f337b084 3034
a6ec74c1 3035 fromstr = NEXTFROM;
28a4f200
TH
3036 SvGETMAGIC(fromstr);
3037 if (!SvOK(fromstr)) aptr = NULL;
a6ec74c1 3038 else {
a6ec74c1
JH
3039 /* XXX better yet, could spirit away the string to
3040 * a safe spot and hang on to it until the result
3041 * of pack() (and all copies of the result) are
3042 * gone.
3043 */
041457d9 3044 if ((SvTEMP(fromstr) || (SvPADTMP(fromstr) &&
a2a5de95
NC
3045 !SvREADONLY(fromstr)))) {
3046 Perl_ck_warner(aTHX_ packWARN(WARN_PACK),
3047 "Attempt to pack pointer to temporary value");
a6ec74c1
JH
3048 }
3049 if (SvPOK(fromstr) || SvNIOK(fromstr))
2596d9fe 3050 aptr = SvPV_nomg_const_nolen(fromstr);
a6ec74c1 3051 else
2596d9fe 3052 aptr = SvPV_force_flags_nolen(fromstr, 0);
a6ec74c1 3053 }
3a88beaa 3054 PUSH_VAR(utf8, cur, aptr, needs_swap);
a6ec74c1
JH
3055 }
3056 break;
fc241834 3057 case 'u': {
f7fe979e 3058 const char *aptr, *aend;
fc241834 3059 bool from_utf8;
f337b084 3060
a6ec74c1 3061 fromstr = NEXTFROM;
fc241834
RGS
3062 if (len <= 2) len = 45;
3063 else len = len / 3 * 3;
3064 if (len >= 64) {
a2a5de95
NC
3065 Perl_ck_warner(aTHX_ packWARN(WARN_PACK),
3066 "Field too wide in 'u' format in pack");
fc241834
RGS
3067 len = 63;
3068 }
83003860 3069 aptr = SvPV_const(fromstr, fromlen);
fc241834
RGS
3070 from_utf8 = DO_UTF8(fromstr);
3071 if (from_utf8) {
3072 aend = aptr + fromlen;
3f63b0e5 3073 fromlen = sv_len_utf8_nomg(fromstr);
fc241834
RGS
3074 } else aend = NULL; /* Unused, but keep compilers happy */
3075 GROWING(utf8, cat, start, cur, (fromlen+2) / 3 * 4 + (fromlen+len-1)/len * 2);
a6ec74c1 3076 while (fromlen > 0) {
fc241834 3077 U8 *end;
a6ec74c1 3078 I32 todo;
fc241834 3079 U8 hunk[1+63/3*4+1];
a6ec74c1 3080
eb160463 3081 if ((I32)fromlen > len)
a6ec74c1
JH
3082 todo = len;
3083 else
3084 todo = fromlen;
fc241834
RGS
3085 if (from_utf8) {
3086 char buffer[64];
3087 if (!uni_to_bytes(aTHX_ &aptr, aend, buffer, todo,
3088 'u' | TYPE_IS_PACK)) {
3089 *cur = '\0';
b162af07 3090 SvCUR_set(cat, cur - start);
5637ef5b
NC
3091 Perl_croak(aTHX_ "panic: string is shorter than advertised, "
3092 "aptr=%p, aend=%p, buffer=%p, todo=%ld",
3093 aptr, aend, buffer, (long) todo);
fc241834
RGS
3094 }
3095 end = doencodes(hunk, buffer, todo);
3096 } else {
3097 end = doencodes(hunk, aptr, todo);
3098 aptr += todo;
3099 }
3a88beaa 3100 PUSH_BYTES(utf8, cur, hunk, end-hunk, 0);
fc241834
RGS
3101 fromlen -= todo;
3102 }
a6ec74c1
JH
3103 break;
3104 }
f337b084
TH
3105 }
3106 *cur = '\0';
b162af07 3107 SvCUR_set(cat, cur - start);
f337b084 3108 no_change:
49704364 3109 *symptr = lookahead;
a6ec74c1 3110 }
49704364 3111 return beglist;
18529408
IZ
3112}
3113#undef NEXTFROM
3114
3115
3116PP(pp_pack)
3117{
20b7effb 3118 dSP; dMARK; dORIGMARK; dTARGET;
eb578fdb 3119 SV *cat = TARG;
18529408 3120 STRLEN fromlen;
349d4f2f 3121 SV *pat_sv = *++MARK;
eb578fdb
KW
3122 const char *pat = SvPV_const(pat_sv, fromlen);
3123 const char *patend = pat + fromlen;
18529408
IZ
3124
3125 MARK++;
76f68e9b 3126 sv_setpvs(cat, "");
f337b084 3127 SvUTF8_off(cat);
18529408 3128
7accc089 3129 packlist(cat, pat, patend, MARK, SP + 1);
18529408 3130
a6ec74c1
JH
3131 SvSETMAGIC(cat);
3132 SP = ORIGMARK;
3133 PUSHs(cat);
3134 RETURN;
3135}
a6ec74c1 3136
73cb7263
NC
3137/*
3138 * Local variables:
3139 * c-indentation-style: bsd
3140 * c-basic-offset: 4
14d04a33 3141 * indent-tabs-mode: nil
73cb7263
NC
3142 * End:
3143 *
14d04a33 3144 * ex: set ts=8 sts=4 sw=4 et:
37442d52 3145 */