Ruby  1.9.3p551(2014-11-13revision48407)
pack.c
Go to the documentation of this file.
1 /**********************************************************************
2 
3  pack.c -
4 
5  $Author: naruse $
6  created at: Thu Feb 10 15:17:05 JST 1994
7 
8  Copyright (C) 1993-2007 Yukihiro Matsumoto
9 
10 **********************************************************************/
11 
12 #include "ruby/ruby.h"
13 #include "ruby/encoding.h"
14 #include <sys/types.h>
15 #include <ctype.h>
16 #include <errno.h>
17 
18 #define GCC_VERSION_SINCE(major, minor, patchlevel) \
19  (defined(__GNUC__) && !defined(__INTEL_COMPILER) && \
20  ((__GNUC__ > (major)) || \
21  (__GNUC__ == (major) && __GNUC_MINOR__ > (minor)) || \
22  (__GNUC__ == (major) && __GNUC_MINOR__ == (minor) && __GNUC_PATCHLEVEL__ >= (patchlevel))))
23 #if SIZEOF_SHORT != 2 || SIZEOF_LONG != 4
24 # define NATINT_PACK
25 #endif
26 
27 #ifdef DYNAMIC_ENDIAN
28  /* for universal binary of NEXTSTEP and MacOS X */
29  /* useless since autoconf 2.63? */
30  static int
31  is_bigendian(void)
32  {
33  static int init = 0;
34  static int endian_value;
35  char *p;
36 
37  if (init) return endian_value;
38  init = 1;
39  p = (char*)&init;
40  return endian_value = p[0]?0:1;
41  }
42 # define BIGENDIAN_P() (is_bigendian())
43 #elif defined(WORDS_BIGENDIAN)
44 # define BIGENDIAN_P() 1
45 #else
46 # define BIGENDIAN_P() 0
47 #endif
48 
49 #ifdef NATINT_PACK
50 # define NATINT_LEN(type,len) (natint?(int)sizeof(type):(int)(len))
51 #else
52 # define NATINT_LEN(type,len) ((int)sizeof(type))
53 #endif
54 
55 #if SIZEOF_LONG == 8
56 # define INT64toNUM(x) LONG2NUM(x)
57 # define UINT64toNUM(x) ULONG2NUM(x)
58 #elif defined(HAVE_LONG_LONG) && SIZEOF_LONG_LONG == 8
59 # define INT64toNUM(x) LL2NUM(x)
60 # define UINT64toNUM(x) ULL2NUM(x)
61 #endif
62 
63 #define define_swapx(x, xtype) \
64 static xtype \
65 TOKEN_PASTE(swap,x)(xtype z) \
66 { \
67  xtype r; \
68  xtype *zp; \
69  unsigned char *s, *t; \
70  int i; \
71  \
72  zp = xmalloc(sizeof(xtype)); \
73  *zp = z; \
74  s = (unsigned char*)zp; \
75  t = xmalloc(sizeof(xtype)); \
76  for (i=0; i<sizeof(xtype); i++) { \
77  t[sizeof(xtype)-i-1] = s[i]; \
78  } \
79  r = *(xtype *)t; \
80  xfree(t); \
81  xfree(zp); \
82  return r; \
83 }
84 
85 #if GCC_VERSION_SINCE(4,3,0)
86 # define swap32(x) __builtin_bswap32(x)
87 # define swap64(x) __builtin_bswap64(x)
88 #endif
89 
90 #ifndef swap16
91 # define swap16(x) ((((x)&0xFF)<<8) | (((x)>>8)&0xFF))
92 #endif
93 
94 #ifndef swap32
95 # define swap32(x) ((((x)&0xFF)<<24) \
96  |(((x)>>24)&0xFF) \
97  |(((x)&0x0000FF00)<<8) \
98  |(((x)&0x00FF0000)>>8) )
99 #endif
100 
101 #ifndef swap64
102 # ifdef HAVE_INT64_T
103 # define byte_in_64bit(n) ((uint64_t)0xff << (n))
104 # define swap64(x) ((((x)&byte_in_64bit(0))<<56) \
105  |(((x)>>56)&0xFF) \
106  |(((x)&byte_in_64bit(8))<<40) \
107  |(((x)&byte_in_64bit(48))>>40) \
108  |(((x)&byte_in_64bit(16))<<24) \
109  |(((x)&byte_in_64bit(40))>>24) \
110  |(((x)&byte_in_64bit(24))<<8) \
111  |(((x)&byte_in_64bit(32))>>8))
112 # endif
113 #endif
114 
115 #if SIZEOF_SHORT == 2
116 # define swaps(x) swap16(x)
117 #elif SIZEOF_SHORT == 4
118 # define swaps(x) swap32(x)
119 #else
120  define_swapx(s,short)
121 #endif
122 
123 #if SIZEOF_INT == 2
124 # define swapi(x) swap16(x)
125 #elif SIZEOF_INT == 4
126 # define swapi(x) swap32(x)
127 #else
128  define_swapx(i,int)
129 #endif
130 
131 #if SIZEOF_LONG == 4
132 # define swapl(x) swap32(x)
133 #elif SIZEOF_LONG == 8
134 # define swapl(x) swap64(x)
135 #else
136  define_swapx(l,long)
137 #endif
138 
139 #ifdef HAVE_LONG_LONG
140 # if SIZEOF_LONG_LONG == 8
141 # define swapll(x) swap64(x)
142 # else
143  define_swapx(ll,LONG_LONG)
144 # endif
145 #endif
146 
147 #if SIZEOF_FLOAT == 4 && defined(HAVE_INT32_T)
148 # define swapf(x) swap32(x)
149 # define FLOAT_SWAPPER uint32_t
150 #else
151  define_swapx(f,float)
152 #endif
153 
154 #if SIZEOF_DOUBLE == 8 && defined(HAVE_INT64_T)
155 # define swapd(x) swap64(x)
156 # define DOUBLE_SWAPPER uint64_t
157 #elif SIZEOF_DOUBLE == 8 && defined(HAVE_INT32_T)
158  static double
159  swapd(const double d)
160  {
161  double dtmp = d;
162  uint32_t utmp[2];
163  uint32_t utmp0;
164 
165  utmp[0] = 0; utmp[1] = 0;
166  memcpy(utmp,&dtmp,sizeof(double));
167  utmp0 = utmp[0];
168  utmp[0] = swap32(utmp[1]);
169  utmp[1] = swap32(utmp0);
170  memcpy(&dtmp,utmp,sizeof(double));
171  return dtmp;
172  }
173 #else
174  define_swapx(d, double)
175 #endif
176 
177 #undef define_swapx
178 
179 #define rb_ntohf(x) (BIGENDIAN_P()?(x):swapf(x))
180 #define rb_ntohd(x) (BIGENDIAN_P()?(x):swapd(x))
181 #define rb_htonf(x) (BIGENDIAN_P()?(x):swapf(x))
182 #define rb_htond(x) (BIGENDIAN_P()?(x):swapd(x))
183 #define rb_htovf(x) (BIGENDIAN_P()?swapf(x):(x))
184 #define rb_htovd(x) (BIGENDIAN_P()?swapd(x):(x))
185 #define rb_vtohf(x) (BIGENDIAN_P()?swapf(x):(x))
186 #define rb_vtohd(x) (BIGENDIAN_P()?swapd(x):(x))
187 
188 #ifdef FLOAT_SWAPPER
189 # define FLOAT_CONVWITH(y) FLOAT_SWAPPER y;
190 # define HTONF(x,y) (memcpy(&(y),&(x),sizeof(float)), \
191  (y) = rb_htonf((FLOAT_SWAPPER)(y)), \
192  memcpy(&(x),&(y),sizeof(float)), \
193  (x))
194 # define HTOVF(x,y) (memcpy(&(y),&(x),sizeof(float)), \
195  (y) = rb_htovf((FLOAT_SWAPPER)(y)), \
196  memcpy(&(x),&(y),sizeof(float)), \
197  (x))
198 # define NTOHF(x,y) (memcpy(&(y),&(x),sizeof(float)), \
199  (y) = rb_ntohf((FLOAT_SWAPPER)(y)), \
200  memcpy(&(x),&(y),sizeof(float)), \
201  (x))
202 # define VTOHF(x,y) (memcpy(&(y),&(x),sizeof(float)), \
203  (y) = rb_vtohf((FLOAT_SWAPPER)(y)), \
204  memcpy(&(x),&(y),sizeof(float)), \
205  (x))
206 #else
207 # define FLOAT_CONVWITH(y)
208 # define HTONF(x,y) rb_htonf(x)
209 # define HTOVF(x,y) rb_htovf(x)
210 # define NTOHF(x,y) rb_ntohf(x)
211 # define VTOHF(x,y) rb_vtohf(x)
212 #endif
213 
214 #ifdef DOUBLE_SWAPPER
215 # define DOUBLE_CONVWITH(y) DOUBLE_SWAPPER y;
216 # define HTOND(x,y) (memcpy(&(y),&(x),sizeof(double)), \
217  (y) = rb_htond((DOUBLE_SWAPPER)(y)), \
218  memcpy(&(x),&(y),sizeof(double)), \
219  (x))
220 # define HTOVD(x,y) (memcpy(&(y),&(x),sizeof(double)), \
221  (y) = rb_htovd((DOUBLE_SWAPPER)(y)), \
222  memcpy(&(x),&(y),sizeof(double)), \
223  (x))
224 # define NTOHD(x,y) (memcpy(&(y),&(x),sizeof(double)), \
225  (y) = rb_ntohd((DOUBLE_SWAPPER)(y)), \
226  memcpy(&(x),&(y),sizeof(double)), \
227  (x))
228 # define VTOHD(x,y) (memcpy(&(y),&(x),sizeof(double)), \
229  (y) = rb_vtohd((DOUBLE_SWAPPER)(y)), \
230  memcpy(&(x),&(y),sizeof(double)), \
231  (x))
232 #else
233 # define DOUBLE_CONVWITH(y)
234 # define HTOND(x,y) rb_htond(x)
235 # define HTOVD(x,y) rb_htovd(x)
236 # define NTOHD(x,y) rb_ntohd(x)
237 # define VTOHD(x,y) rb_vtohd(x)
238 #endif
239 
240 static unsigned long
241 num2i32(VALUE x)
242 {
243  x = rb_to_int(x); /* is nil OK? (should not) */
244 
245  if (FIXNUM_P(x)) return FIX2LONG(x);
246  if (TYPE(x) == T_BIGNUM) {
247  return rb_big2ulong_pack(x);
248  }
249  rb_raise(rb_eTypeError, "can't convert %s to `integer'", rb_obj_classname(x));
250  return 0; /* not reached */
251 }
252 
253 #define MAX_INTEGER_PACK_SIZE 8
254 /* #define FORCE_BIG_PACK */
255 
256 static const char toofew[] = "too few arguments";
257 
258 static void encodes(VALUE,const char*,long,int,int);
259 static void qpencode(VALUE,VALUE,long);
260 
261 static unsigned long utf8_to_uv(const char*,long*);
262 
263 /*
264  * call-seq:
265  * arr.pack ( aTemplateString ) -> aBinaryString
266  *
267  * Packs the contents of <i>arr</i> into a binary sequence according to
268  * the directives in <i>aTemplateString</i> (see the table below)
269  * Directives ``A,'' ``a,'' and ``Z'' may be followed by a count,
270  * which gives the width of the resulting field. The remaining
271  * directives also may take a count, indicating the number of array
272  * elements to convert. If the count is an asterisk
273  * (``<code>*</code>''), all remaining array elements will be
274  * converted. Any of the directives ``<code>sSiIlL</code>'' may be
275  * followed by an underscore (``<code>_</code>'') or
276  * exclamation mark (``<code>!</code>'') to use the underlying
277  * platform's native size for the specified type; otherwise, they use a
278  * platform-independent size. Spaces are ignored in the template
279  * string. See also <code>String#unpack</code>.
280  *
281  * a = [ "a", "b", "c" ]
282  * n = [ 65, 66, 67 ]
283  * a.pack("A3A3A3") #=> "a b c "
284  * a.pack("a3a3a3") #=> "a\000\000b\000\000c\000\000"
285  * n.pack("ccc") #=> "ABC"
286  *
287  * Directives for +pack+.
288  *
289  * Integer | Array |
290  * Directive | Element | Meaning
291  * ---------------------------------------------------------------------------
292  * C | Integer | 8-bit unsigned (unsigned char)
293  * S | Integer | 16-bit unsigned, native endian (uint16_t)
294  * L | Integer | 32-bit unsigned, native endian (uint32_t)
295  * Q | Integer | 64-bit unsigned, native endian (uint64_t)
296  * | |
297  * c | Integer | 8-bit signed (signed char)
298  * s | Integer | 16-bit signed, native endian (int16_t)
299  * l | Integer | 32-bit signed, native endian (int32_t)
300  * q | Integer | 64-bit signed, native endian (int64_t)
301  * | |
302  * S_, S! | Integer | unsigned short, native endian
303  * I, I_, I! | Integer | unsigned int, native endian
304  * L_, L! | Integer | unsigned long, native endian
305  * | |
306  * s_, s! | Integer | signed short, native endian
307  * i, i_, i! | Integer | signed int, native endian
308  * l_, l! | Integer | signed long, native endian
309  * | |
310  * S> L> Q> | Integer | same as the directives without ">" except
311  * s> l> q> | | big endian
312  * S!> I!> | | (available since Ruby 1.9.3)
313  * L!> | | "S>" is same as "n"
314  * s!> i!> | | "L>" is same as "N"
315  * l!> | |
316  * | |
317  * S< L< Q< | Integer | same as the directives without "<" except
318  * s< l< q< | | little endian
319  * S!< I!< | | (available since Ruby 1.9.3)
320  * L!< | | "S<" is same as "v"
321  * s!< i!< | | "L<" is same as "V"
322  * l!< | |
323  * | |
324  * n | Integer | 16-bit unsigned, network (big-endian) byte order
325  * N | Integer | 32-bit unsigned, network (big-endian) byte order
326  * v | Integer | 16-bit unsigned, VAX (little-endian) byte order
327  * V | Integer | 32-bit unsigned, VAX (little-endian) byte order
328  * | |
329  * U | Integer | UTF-8 character
330  * w | Integer | BER-compressed integer
331  *
332  * Float | |
333  * Directive | | Meaning
334  * ---------------------------------------------------------------------------
335  * D, d | Float | double-precision, native format
336  * F, f | Float | single-precision, native format
337  * E | Float | double-precision, little-endian byte order
338  * e | Float | single-precision, little-endian byte order
339  * G | Float | double-precision, network (big-endian) byte order
340  * g | Float | single-precision, network (big-endian) byte order
341  *
342  * String | |
343  * Directive | | Meaning
344  * ---------------------------------------------------------------------------
345  * A | String | arbitrary binary string (space padded, count is width)
346  * a | String | arbitrary binary string (null padded, count is width)
347  * Z | String | same as ``a'', except that null is added with *
348  * B | String | bit string (MSB first)
349  * b | String | bit string (LSB first)
350  * H | String | hex string (high nibble first)
351  * h | String | hex string (low nibble first)
352  * u | String | UU-encoded string
353  * M | String | quoted printable, MIME encoding (see RFC2045)
354  * m | String | base64 encoded string (see RFC 2045, count is width)
355  * | | (if count is 0, no line feed are added, see RFC 4648)
356  * P | String | pointer to a structure (fixed-length string)
357  * p | String | pointer to a null-terminated string
358  *
359  * Misc. | |
360  * Directive | | Meaning
361  * ---------------------------------------------------------------------------
362  * @ | --- | moves to absolute position
363  * X | --- | back up a byte
364  * x | --- | null byte
365  */
366 
367 static VALUE
369 {
370  static const char nul10[] = "\0\0\0\0\0\0\0\0\0\0";
371  static const char spc10[] = " ";
372  const char *p, *pend;
373  VALUE res, from, associates = 0;
374  char type;
375  long items, len, idx, plen;
376  const char *ptr;
377  int enc_info = 1; /* 0 - BINARY, 1 - US-ASCII, 2 - UTF-8 */
378 #ifdef NATINT_PACK
379  int natint; /* native integer */
380 #endif
381  int signed_p, integer_size, bigendian_p;
382 
383  StringValue(fmt);
384  p = RSTRING_PTR(fmt);
385  pend = p + RSTRING_LEN(fmt);
386  res = rb_str_buf_new(0);
387 
388  items = RARRAY_LEN(ary);
389  idx = 0;
390 
391 #define TOO_FEW (rb_raise(rb_eArgError, toofew), 0)
392 #define THISFROM (items > 0 ? RARRAY_PTR(ary)[idx] : TOO_FEW)
393 #define NEXTFROM (items-- > 0 ? RARRAY_PTR(ary)[idx++] : TOO_FEW)
394 
395  while (p < pend) {
396  int explicit_endian = 0;
397  if (RSTRING_PTR(fmt) + RSTRING_LEN(fmt) != pend) {
398  rb_raise(rb_eRuntimeError, "format string modified");
399  }
400  type = *p++; /* get data type */
401 #ifdef NATINT_PACK
402  natint = 0;
403 #endif
404 
405  if (ISSPACE(type)) continue;
406  if (type == '#') {
407  while ((p < pend) && (*p != '\n')) {
408  p++;
409  }
410  continue;
411  }
412 
413  {
414  static const char natstr[] = "sSiIlL";
415  static const char endstr[] = "sSiIlLqQ";
416 
417  modifiers:
418  switch (*p) {
419  case '_':
420  case '!':
421  if (strchr(natstr, type)) {
422 #ifdef NATINT_PACK
423  natint = 1;
424 #endif
425  p++;
426  }
427  else {
428  rb_raise(rb_eArgError, "'%c' allowed only after types %s", *p, natstr);
429  }
430  goto modifiers;
431 
432  case '<':
433  case '>':
434  if (!strchr(endstr, type)) {
435  rb_raise(rb_eArgError, "'%c' allowed only after types %s", *p, endstr);
436  }
437  if (explicit_endian) {
438  rb_raise(rb_eRangeError, "Can't use both '<' and '>'");
439  }
440  explicit_endian = *p++;
441  goto modifiers;
442  }
443  }
444 
445  if (*p == '*') { /* set data length */
446  len = strchr("@Xxu", type) ? 0
447  : strchr("PMm", type) ? 1
448  : items;
449  p++;
450  }
451  else if (ISDIGIT(*p)) {
452  errno = 0;
453  len = STRTOUL(p, (char**)&p, 10);
454  if (errno) {
455  rb_raise(rb_eRangeError, "pack length too big");
456  }
457  }
458  else {
459  len = 1;
460  }
461 
462  switch (type) {
463  case 'U':
464  /* if encoding is US-ASCII, upgrade to UTF-8 */
465  if (enc_info == 1) enc_info = 2;
466  break;
467  case 'm': case 'M': case 'u':
468  /* keep US-ASCII (do nothing) */
469  break;
470  default:
471  /* fall back to BINARY */
472  enc_info = 0;
473  break;
474  }
475  switch (type) {
476  case 'A': case 'a': case 'Z':
477  case 'B': case 'b':
478  case 'H': case 'h':
479  from = NEXTFROM;
480  if (NIL_P(from)) {
481  ptr = "";
482  plen = 0;
483  }
484  else {
485  StringValue(from);
486  ptr = RSTRING_PTR(from);
487  plen = RSTRING_LEN(from);
488  OBJ_INFECT(res, from);
489  }
490 
491  if (p[-1] == '*')
492  len = plen;
493 
494  switch (type) {
495  case 'a': /* arbitrary binary string (null padded) */
496  case 'A': /* arbitrary binary string (ASCII space padded) */
497  case 'Z': /* null terminated string */
498  if (plen >= len) {
499  rb_str_buf_cat(res, ptr, len);
500  if (p[-1] == '*' && type == 'Z')
501  rb_str_buf_cat(res, nul10, 1);
502  }
503  else {
504  rb_str_buf_cat(res, ptr, plen);
505  len -= plen;
506  while (len >= 10) {
507  rb_str_buf_cat(res, (type == 'A')?spc10:nul10, 10);
508  len -= 10;
509  }
510  rb_str_buf_cat(res, (type == 'A')?spc10:nul10, len);
511  }
512  break;
513 
514  case 'b': /* bit string (ascending) */
515  {
516  int byte = 0;
517  long i, j = 0;
518 
519  if (len > plen) {
520  j = (len - plen + 1)/2;
521  len = plen;
522  }
523  for (i=0; i++ < len; ptr++) {
524  if (*ptr & 1)
525  byte |= 128;
526  if (i & 7)
527  byte >>= 1;
528  else {
529  char c = byte & 0xff;
530  rb_str_buf_cat(res, &c, 1);
531  byte = 0;
532  }
533  }
534  if (len & 7) {
535  char c;
536  byte >>= 7 - (len & 7);
537  c = byte & 0xff;
538  rb_str_buf_cat(res, &c, 1);
539  }
540  len = j;
541  goto grow;
542  }
543  break;
544 
545  case 'B': /* bit string (descending) */
546  {
547  int byte = 0;
548  long i, j = 0;
549 
550  if (len > plen) {
551  j = (len - plen + 1)/2;
552  len = plen;
553  }
554  for (i=0; i++ < len; ptr++) {
555  byte |= *ptr & 1;
556  if (i & 7)
557  byte <<= 1;
558  else {
559  char c = byte & 0xff;
560  rb_str_buf_cat(res, &c, 1);
561  byte = 0;
562  }
563  }
564  if (len & 7) {
565  char c;
566  byte <<= 7 - (len & 7);
567  c = byte & 0xff;
568  rb_str_buf_cat(res, &c, 1);
569  }
570  len = j;
571  goto grow;
572  }
573  break;
574 
575  case 'h': /* hex string (low nibble first) */
576  {
577  int byte = 0;
578  long i, j = 0;
579 
580  if (len > plen) {
581  j = (len + 1) / 2 - (plen + 1) / 2;
582  len = plen;
583  }
584  for (i=0; i++ < len; ptr++) {
585  if (ISALPHA(*ptr))
586  byte |= (((*ptr & 15) + 9) & 15) << 4;
587  else
588  byte |= (*ptr & 15) << 4;
589  if (i & 1)
590  byte >>= 4;
591  else {
592  char c = byte & 0xff;
593  rb_str_buf_cat(res, &c, 1);
594  byte = 0;
595  }
596  }
597  if (len & 1) {
598  char c = byte & 0xff;
599  rb_str_buf_cat(res, &c, 1);
600  }
601  len = j;
602  goto grow;
603  }
604  break;
605 
606  case 'H': /* hex string (high nibble first) */
607  {
608  int byte = 0;
609  long i, j = 0;
610 
611  if (len > plen) {
612  j = (len + 1) / 2 - (plen + 1) / 2;
613  len = plen;
614  }
615  for (i=0; i++ < len; ptr++) {
616  if (ISALPHA(*ptr))
617  byte |= ((*ptr & 15) + 9) & 15;
618  else
619  byte |= *ptr & 15;
620  if (i & 1)
621  byte <<= 4;
622  else {
623  char c = byte & 0xff;
624  rb_str_buf_cat(res, &c, 1);
625  byte = 0;
626  }
627  }
628  if (len & 1) {
629  char c = byte & 0xff;
630  rb_str_buf_cat(res, &c, 1);
631  }
632  len = j;
633  goto grow;
634  }
635  break;
636  }
637  break;
638 
639  case 'c': /* signed char */
640  case 'C': /* unsigned char */
641  while (len-- > 0) {
642  char c;
643 
644  from = NEXTFROM;
645  c = (char)num2i32(from);
646  rb_str_buf_cat(res, &c, sizeof(char));
647  }
648  break;
649 
650  case 's': /* signed short */
651  signed_p = 1;
652  integer_size = NATINT_LEN(short, 2);
653  bigendian_p = BIGENDIAN_P();
654  goto pack_integer;
655 
656  case 'S': /* unsigned short */
657  signed_p = 0;
658  integer_size = NATINT_LEN(short, 2);
659  bigendian_p = BIGENDIAN_P();
660  goto pack_integer;
661 
662  case 'i': /* signed int */
663  signed_p = 1;
664  integer_size = (int)sizeof(int);
665  bigendian_p = BIGENDIAN_P();
666  goto pack_integer;
667 
668  case 'I': /* unsigned int */
669  signed_p = 0;
670  integer_size = (int)sizeof(int);
671  bigendian_p = BIGENDIAN_P();
672  goto pack_integer;
673 
674  case 'l': /* signed long */
675  signed_p = 1;
676  integer_size = NATINT_LEN(long, 4);
677  bigendian_p = BIGENDIAN_P();
678  goto pack_integer;
679 
680  case 'L': /* unsigned long */
681  signed_p = 0;
682  integer_size = NATINT_LEN(long, 4);
683  bigendian_p = BIGENDIAN_P();
684  goto pack_integer;
685 
686  case 'q': /* signed quad (64bit) int */
687  signed_p = 1;
688  integer_size = 8;
689  bigendian_p = BIGENDIAN_P();
690  goto pack_integer;
691 
692  case 'Q': /* unsigned quad (64bit) int */
693  signed_p = 0;
694  integer_size = 8;
695  bigendian_p = BIGENDIAN_P();
696  goto pack_integer;
697 
698  case 'n': /* unsigned short (network byte-order) */
699  signed_p = 0;
700  integer_size = 2;
701  bigendian_p = 1;
702  goto pack_integer;
703 
704  case 'N': /* unsigned long (network byte-order) */
705  signed_p = 0;
706  integer_size = 4;
707  bigendian_p = 1;
708  goto pack_integer;
709 
710  case 'v': /* unsigned short (VAX byte-order) */
711  signed_p = 0;
712  integer_size = 2;
713  bigendian_p = 0;
714  goto pack_integer;
715 
716  case 'V': /* unsigned long (VAX byte-order) */
717  signed_p = 0;
718  integer_size = 4;
719  bigendian_p = 0;
720  goto pack_integer;
721 
722  pack_integer:
723  if (explicit_endian) {
724  bigendian_p = explicit_endian == '>';
725  }
726 
727  switch (integer_size) {
728 #if defined(HAVE_INT16_T) && !defined(FORCE_BIG_PACK)
729  case SIZEOF_INT16_T:
730  while (len-- > 0) {
731  union {
732  int16_t i;
733  char a[sizeof(int16_t)];
734  } v;
735 
736  from = NEXTFROM;
737  v.i = (int16_t)num2i32(from);
738  if (bigendian_p != BIGENDIAN_P()) v.i = swap16(v.i);
739  rb_str_buf_cat(res, v.a, sizeof(int16_t));
740  }
741  break;
742 #endif
743 
744 #if defined(HAVE_INT32_T) && !defined(FORCE_BIG_PACK)
745  case SIZEOF_INT32_T:
746  while (len-- > 0) {
747  union {
748  int32_t i;
749  char a[sizeof(int32_t)];
750  } v;
751 
752  from = NEXTFROM;
753  v.i = (int32_t)num2i32(from);
754  if (bigendian_p != BIGENDIAN_P()) v.i = swap32(v.i);
755  rb_str_buf_cat(res, v.a, sizeof(int32_t));
756  }
757  break;
758 #endif
759 
760 #if defined(HAVE_INT64_T) && SIZEOF_LONG == SIZEOF_INT64_T && !defined(FORCE_BIG_PACK)
761  case SIZEOF_INT64_T:
762  while (len-- > 0) {
763  union {
764  int64_t i;
765  char a[sizeof(int64_t)];
766  } v;
767 
768  from = NEXTFROM;
769  v.i = num2i32(from); /* can return 64bit value if SIZEOF_LONG == SIZEOF_INT64_T */
770  if (bigendian_p != BIGENDIAN_P()) v.i = swap64(v.i);
771  rb_str_buf_cat(res, v.a, sizeof(int64_t));
772  }
773  break;
774 #endif
775 
776  default:
777  if (integer_size > MAX_INTEGER_PACK_SIZE)
778  rb_bug("unexpected intger size for pack: %d", integer_size);
779  while (len-- > 0) {
780  union {
781  unsigned long i[(MAX_INTEGER_PACK_SIZE+SIZEOF_LONG-1)/SIZEOF_LONG];
782  char a[(MAX_INTEGER_PACK_SIZE+SIZEOF_LONG-1)/SIZEOF_LONG*SIZEOF_LONG];
783  } v;
784  int num_longs = (integer_size+SIZEOF_LONG-1)/SIZEOF_LONG;
785  int i;
786 
787  from = NEXTFROM;
788  rb_big_pack(from, v.i, num_longs);
789  if (bigendian_p) {
790  for (i = 0; i < num_longs/2; i++) {
791  unsigned long t = v.i[i];
792  v.i[i] = v.i[num_longs-1-i];
793  v.i[num_longs-1-i] = t;
794  }
795  }
796  if (bigendian_p != BIGENDIAN_P()) {
797  for (i = 0; i < num_longs; i++)
798  v.i[i] = swapl(v.i[i]);
799  }
800  rb_str_buf_cat(res,
801  bigendian_p ?
802  v.a + sizeof(long)*num_longs - integer_size :
803  v.a,
804  integer_size);
805  }
806  break;
807  }
808  break;
809 
810  case 'f': /* single precision float in native format */
811  case 'F': /* ditto */
812  while (len-- > 0) {
813  float f;
814 
815  from = NEXTFROM;
816  f = (float)RFLOAT_VALUE(rb_to_float(from));
817  rb_str_buf_cat(res, (char*)&f, sizeof(float));
818  }
819  break;
820 
821  case 'e': /* single precision float in VAX byte-order */
822  while (len-- > 0) {
823  float f;
824  FLOAT_CONVWITH(ftmp);
825 
826  from = NEXTFROM;
827  f = (float)RFLOAT_VALUE(rb_to_float(from));
828  f = HTOVF(f,ftmp);
829  rb_str_buf_cat(res, (char*)&f, sizeof(float));
830  }
831  break;
832 
833  case 'E': /* double precision float in VAX byte-order */
834  while (len-- > 0) {
835  double d;
836  DOUBLE_CONVWITH(dtmp);
837 
838  from = NEXTFROM;
839  d = RFLOAT_VALUE(rb_to_float(from));
840  d = HTOVD(d,dtmp);
841  rb_str_buf_cat(res, (char*)&d, sizeof(double));
842  }
843  break;
844 
845  case 'd': /* double precision float in native format */
846  case 'D': /* ditto */
847  while (len-- > 0) {
848  double d;
849 
850  from = NEXTFROM;
851  d = RFLOAT_VALUE(rb_to_float(from));
852  rb_str_buf_cat(res, (char*)&d, sizeof(double));
853  }
854  break;
855 
856  case 'g': /* single precision float in network byte-order */
857  while (len-- > 0) {
858  float f;
859  FLOAT_CONVWITH(ftmp);
860 
861  from = NEXTFROM;
862  f = (float)RFLOAT_VALUE(rb_to_float(from));
863  f = HTONF(f,ftmp);
864  rb_str_buf_cat(res, (char*)&f, sizeof(float));
865  }
866  break;
867 
868  case 'G': /* double precision float in network byte-order */
869  while (len-- > 0) {
870  double d;
871  DOUBLE_CONVWITH(dtmp);
872 
873  from = NEXTFROM;
874  d = RFLOAT_VALUE(rb_to_float(from));
875  d = HTOND(d,dtmp);
876  rb_str_buf_cat(res, (char*)&d, sizeof(double));
877  }
878  break;
879 
880  case 'x': /* null byte */
881  grow:
882  while (len >= 10) {
883  rb_str_buf_cat(res, nul10, 10);
884  len -= 10;
885  }
886  rb_str_buf_cat(res, nul10, len);
887  break;
888 
889  case 'X': /* back up byte */
890  shrink:
891  plen = RSTRING_LEN(res);
892  if (plen < len)
893  rb_raise(rb_eArgError, "X outside of string");
894  rb_str_set_len(res, plen - len);
895  break;
896 
897  case '@': /* null fill to absolute position */
898  len -= RSTRING_LEN(res);
899  if (len > 0) goto grow;
900  len = -len;
901  if (len > 0) goto shrink;
902  break;
903 
904  case '%':
905  rb_raise(rb_eArgError, "%% is not supported");
906  break;
907 
908  case 'U': /* Unicode character */
909  while (len-- > 0) {
910  SIGNED_VALUE l;
911  char buf[8];
912  int le;
913 
914  from = NEXTFROM;
915  from = rb_to_int(from);
916  l = NUM2LONG(from);
917  if (l < 0) {
918  rb_raise(rb_eRangeError, "pack(U): value out of range");
919  }
920  le = rb_uv_to_utf8(buf, l);
921  rb_str_buf_cat(res, (char*)buf, le);
922  }
923  break;
924 
925  case 'u': /* uuencoded string */
926  case 'm': /* base64 encoded string */
927  from = NEXTFROM;
928  StringValue(from);
929  ptr = RSTRING_PTR(from);
930  plen = RSTRING_LEN(from);
931 
932  if (len == 0 && type == 'm') {
933  encodes(res, ptr, plen, type, 0);
934  ptr += plen;
935  break;
936  }
937  if (len <= 2)
938  len = 45;
939  else
940  len = len / 3 * 3;
941  while (plen > 0) {
942  long todo;
943 
944  if (plen > len)
945  todo = len;
946  else
947  todo = plen;
948  encodes(res, ptr, todo, type, 1);
949  plen -= todo;
950  ptr += todo;
951  }
952  break;
953 
954  case 'M': /* quoted-printable encoded string */
955  from = rb_obj_as_string(NEXTFROM);
956  if (len <= 1)
957  len = 72;
958  qpencode(res, from, len);
959  break;
960 
961  case 'P': /* pointer to packed byte string */
962  from = THISFROM;
963  if (!NIL_P(from)) {
964  StringValue(from);
965  if (RSTRING_LEN(from) < len) {
966  rb_raise(rb_eArgError, "too short buffer for P(%ld for %ld)",
967  RSTRING_LEN(from), len);
968  }
969  }
970  len = 1;
971  /* FALL THROUGH */
972  case 'p': /* pointer to string */
973  while (len-- > 0) {
974  char *t;
975  from = NEXTFROM;
976  if (NIL_P(from)) {
977  t = 0;
978  }
979  else {
980  t = StringValuePtr(from);
981  }
982  if (!associates) {
983  associates = rb_ary_new();
984  }
985  rb_ary_push(associates, from);
986  rb_obj_taint(from);
987  rb_str_buf_cat(res, (char*)&t, sizeof(char*));
988  }
989  break;
990 
991  case 'w': /* BER compressed integer */
992  while (len-- > 0) {
993  unsigned long ul;
994  VALUE buf = rb_str_new(0, 0);
995  char c, *bufs, *bufe;
996 
997  from = NEXTFROM;
998  if (TYPE(from) == T_BIGNUM) {
999  VALUE big128 = rb_uint2big(128);
1000  while (TYPE(from) == T_BIGNUM) {
1001  from = rb_big_divmod(from, big128);
1002  c = NUM2INT(RARRAY_PTR(from)[1]) | 0x80; /* mod */
1003  rb_str_buf_cat(buf, &c, sizeof(char));
1004  from = RARRAY_PTR(from)[0]; /* div */
1005  }
1006  }
1007 
1008  {
1009  long l = NUM2LONG(from);
1010  if (l < 0) {
1011  rb_raise(rb_eArgError, "can't compress negative numbers");
1012  }
1013  ul = l;
1014  }
1015 
1016  while (ul) {
1017  c = (char)(ul & 0x7f) | 0x80;
1018  rb_str_buf_cat(buf, &c, sizeof(char));
1019  ul >>= 7;
1020  }
1021 
1022  if (RSTRING_LEN(buf)) {
1023  bufs = RSTRING_PTR(buf);
1024  bufe = bufs + RSTRING_LEN(buf) - 1;
1025  *bufs &= 0x7f; /* clear continue bit */
1026  while (bufs < bufe) { /* reverse */
1027  c = *bufs;
1028  *bufs++ = *bufe;
1029  *bufe-- = c;
1030  }
1031  rb_str_buf_cat(res, RSTRING_PTR(buf), RSTRING_LEN(buf));
1032  }
1033  else {
1034  c = 0;
1035  rb_str_buf_cat(res, &c, sizeof(char));
1036  }
1037  }
1038  break;
1039 
1040  default:
1041  break;
1042  }
1043  }
1044 
1045  if (associates) {
1046  rb_str_associate(res, associates);
1047  }
1048  OBJ_INFECT(res, fmt);
1049  switch (enc_info) {
1050  case 1:
1052  break;
1053  case 2:
1055  break;
1056  default:
1057  /* do nothing, keep ASCII-8BIT */
1058  break;
1059  }
1060  return res;
1061 }
1062 
1063 static const char uu_table[] =
1064 "`!\"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_";
1065 static const char b64_table[] =
1066 "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
1067 
1068 static void
1069 encodes(VALUE str, const char *s, long len, int type, int tail_lf)
1070 {
1071  enum {buff_size = 4096, encoded_unit = 4};
1072  char buff[buff_size + 1]; /* +1 for tail_lf */
1073  long i = 0;
1074  const char *trans = type == 'u' ? uu_table : b64_table;
1075  int padding;
1076 
1077  if (type == 'u') {
1078  buff[i++] = (char)len + ' ';
1079  padding = '`';
1080  }
1081  else {
1082  padding = '=';
1083  }
1084  while (len >= 3) {
1085  while (len >= 3 && buff_size-i >= encoded_unit) {
1086  buff[i++] = trans[077 & (*s >> 2)];
1087  buff[i++] = trans[077 & (((*s << 4) & 060) | ((s[1] >> 4) & 017))];
1088  buff[i++] = trans[077 & (((s[1] << 2) & 074) | ((s[2] >> 6) & 03))];
1089  buff[i++] = trans[077 & s[2]];
1090  s += 3;
1091  len -= 3;
1092  }
1093  if (buff_size-i < encoded_unit) {
1094  rb_str_buf_cat(str, buff, i);
1095  i = 0;
1096  }
1097  }
1098 
1099  if (len == 2) {
1100  buff[i++] = trans[077 & (*s >> 2)];
1101  buff[i++] = trans[077 & (((*s << 4) & 060) | ((s[1] >> 4) & 017))];
1102  buff[i++] = trans[077 & (((s[1] << 2) & 074) | (('\0' >> 6) & 03))];
1103  buff[i++] = padding;
1104  }
1105  else if (len == 1) {
1106  buff[i++] = trans[077 & (*s >> 2)];
1107  buff[i++] = trans[077 & (((*s << 4) & 060) | (('\0' >> 4) & 017))];
1108  buff[i++] = padding;
1109  buff[i++] = padding;
1110  }
1111  if (tail_lf) buff[i++] = '\n';
1112  rb_str_buf_cat(str, buff, i);
1113  if ((size_t)i > sizeof(buff)) rb_bug("encodes() buffer overrun");
1114 }
1115 
1116 static const char hex_table[] = "0123456789ABCDEF";
1117 
1118 static void
1119 qpencode(VALUE str, VALUE from, long len)
1120 {
1121  char buff[1024];
1122  long i = 0, n = 0, prev = EOF;
1123  unsigned char *s = (unsigned char*)RSTRING_PTR(from);
1124  unsigned char *send = s + RSTRING_LEN(from);
1125 
1126  while (s < send) {
1127  if ((*s > 126) ||
1128  (*s < 32 && *s != '\n' && *s != '\t') ||
1129  (*s == '=')) {
1130  buff[i++] = '=';
1131  buff[i++] = hex_table[*s >> 4];
1132  buff[i++] = hex_table[*s & 0x0f];
1133  n += 3;
1134  prev = EOF;
1135  }
1136  else if (*s == '\n') {
1137  if (prev == ' ' || prev == '\t') {
1138  buff[i++] = '=';
1139  buff[i++] = *s;
1140  }
1141  buff[i++] = *s;
1142  n = 0;
1143  prev = *s;
1144  }
1145  else {
1146  buff[i++] = *s;
1147  n++;
1148  prev = *s;
1149  }
1150  if (n > len) {
1151  buff[i++] = '=';
1152  buff[i++] = '\n';
1153  n = 0;
1154  prev = '\n';
1155  }
1156  if (i > 1024 - 5) {
1157  rb_str_buf_cat(str, buff, i);
1158  i = 0;
1159  }
1160  s++;
1161  }
1162  if (n > 0) {
1163  buff[i++] = '=';
1164  buff[i++] = '\n';
1165  }
1166  if (i > 0) {
1167  rb_str_buf_cat(str, buff, i);
1168  }
1169 }
1170 
1171 static inline int
1172 hex2num(char c)
1173 {
1174  switch (c) {
1175  case '0': case '1': case '2': case '3': case '4':
1176  case '5': case '6': case '7': case '8': case '9':
1177  return c - '0';
1178  case 'a': case 'b': case 'c':
1179  case 'd': case 'e': case 'f':
1180  return c - 'a' + 10;
1181  case 'A': case 'B': case 'C':
1182  case 'D': case 'E': case 'F':
1183  return c - 'A' + 10;
1184  default:
1185  return -1;
1186  }
1187 }
1188 
1189 #define PACK_LENGTH_ADJUST_SIZE(sz) do { \
1190  tmp_len = 0; \
1191  if (len > (long)((send-s)/(sz))) { \
1192  if (!star) { \
1193  tmp_len = len-(send-s)/(sz); \
1194  } \
1195  len = (send-s)/(sz); \
1196  } \
1197 } while (0)
1198 
1199 #define PACK_ITEM_ADJUST() do { \
1200  if (tmp_len > 0 && !block_p) \
1201  rb_ary_store(ary, RARRAY_LEN(ary)+tmp_len-1, Qnil); \
1202 } while (0)
1203 
1204 static VALUE
1205 infected_str_new(const char *ptr, long len, VALUE str)
1206 {
1207  VALUE s = rb_str_new(ptr, len);
1208 
1209  OBJ_INFECT(s, str);
1210  return s;
1211 }
1212 
1213 /*
1214  * call-seq:
1215  * str.unpack(format) -> anArray
1216  *
1217  * Decodes <i>str</i> (which may contain binary data) according to the
1218  * format string, returning an array of each value extracted. The
1219  * format string consists of a sequence of single-character directives,
1220  * summarized in the table at the end of this entry.
1221  * Each directive may be followed
1222  * by a number, indicating the number of times to repeat with this
1223  * directive. An asterisk (``<code>*</code>'') will use up all
1224  * remaining elements. The directives <code>sSiIlL</code> may each be
1225  * followed by an underscore (``<code>_</code>'') or
1226  * exclamation mark (``<code>!</code>'') to use the underlying
1227  * platform's native size for the specified type; otherwise, it uses a
1228  * platform-independent consistent size. Spaces are ignored in the
1229  * format string. See also <code>Array#pack</code>.
1230  *
1231  * "abc \0\0abc \0\0".unpack('A6Z6') #=> ["abc", "abc "]
1232  * "abc \0\0".unpack('a3a3') #=> ["abc", " \000\000"]
1233  * "abc \0abc \0".unpack('Z*Z*') #=> ["abc ", "abc "]
1234  * "aa".unpack('b8B8') #=> ["10000110", "01100001"]
1235  * "aaa".unpack('h2H2c') #=> ["16", "61", 97]
1236  * "\xfe\xff\xfe\xff".unpack('sS') #=> [-2, 65534]
1237  * "now=20is".unpack('M*') #=> ["now is"]
1238  * "whole".unpack('xax2aX2aX1aX2a') #=> ["h", "e", "l", "l", "o"]
1239  *
1240  * This table summarizes the various formats and the Ruby classes
1241  * returned by each.
1242  *
1243  * Integer | |
1244  * Directive | Returns | Meaning
1245  * -----------------------------------------------------------------
1246  * C | Integer | 8-bit unsigned (unsigned char)
1247  * S | Integer | 16-bit unsigned, native endian (uint16_t)
1248  * L | Integer | 32-bit unsigned, native endian (uint32_t)
1249  * Q | Integer | 64-bit unsigned, native endian (uint64_t)
1250  * | |
1251  * c | Integer | 8-bit signed (signed char)
1252  * s | Integer | 16-bit signed, native endian (int16_t)
1253  * l | Integer | 32-bit signed, native endian (int32_t)
1254  * q | Integer | 64-bit signed, native endian (int64_t)
1255  * | |
1256  * S_, S! | Integer | unsigned short, native endian
1257  * I, I_, I! | Integer | unsigned int, native endian
1258  * L_, L! | Integer | unsigned long, native endian
1259  * | |
1260  * s_, s! | Integer | signed short, native endian
1261  * i, i_, i! | Integer | signed int, native endian
1262  * l_, l! | Integer | signed long, native endian
1263  * | |
1264  * S> L> Q> | Integer | same as the directives without ">" except
1265  * s> l> q> | | big endian
1266  * S!> I!> | | (available since Ruby 1.9.3)
1267  * L!> Q!> | | "S>" is same as "n"
1268  * s!> i!> | | "L>" is same as "N"
1269  * l!> q!> | |
1270  * | |
1271  * S< L< Q< | Integer | same as the directives without "<" except
1272  * s< l< q< | | little endian
1273  * S!< I!< | | (available since Ruby 1.9.3)
1274  * L!< Q!< | | "S<" is same as "v"
1275  * s!< i!< | | "L<" is same as "V"
1276  * l!< q!< | |
1277  * | |
1278  * n | Integer | 16-bit unsigned, network (big-endian) byte order
1279  * N | Integer | 32-bit unsigned, network (big-endian) byte order
1280  * v | Integer | 16-bit unsigned, VAX (little-endian) byte order
1281  * V | Integer | 32-bit unsigned, VAX (little-endian) byte order
1282  * | |
1283  * U | Integer | UTF-8 character
1284  * w | Integer | BER-compressed integer (see Array.pack)
1285  *
1286  * Float | |
1287  * Directive | Returns | Meaning
1288  * -----------------------------------------------------------------
1289  * D, d | Float | double-precision, native format
1290  * F, f | Float | single-precision, native format
1291  * E | Float | double-precision, little-endian byte order
1292  * e | Float | single-precision, little-endian byte order
1293  * G | Float | double-precision, network (big-endian) byte order
1294  * g | Float | single-precision, network (big-endian) byte order
1295  *
1296  * String | |
1297  * Directive | Returns | Meaning
1298  * -----------------------------------------------------------------
1299  * A | String | arbitrary binary string (remove trailing nulls and ASCII spaces)
1300  * a | String | arbitrary binary string
1301  * Z | String | null-terminated string
1302  * B | String | bit string (MSB first)
1303  * b | String | bit string (LSB first)
1304  * H | String | hex string (high nibble first)
1305  * h | String | hex string (low nibble first)
1306  * u | String | UU-encoded string
1307  * M | String | quoted-printable, MIME encoding (see RFC2045)
1308  * m | String | base64 encoded string (RFC 2045) (default)
1309  * | | base64 encoded string (RFC 4648) if followed by 0
1310  * P | String | pointer to a structure (fixed-length string)
1311  * p | String | pointer to a null-terminated string
1312  *
1313  * Misc. | |
1314  * Directive | Returns | Meaning
1315  * -----------------------------------------------------------------
1316  * @ | --- | skip to the offset given by the length argument
1317  * X | --- | skip backward one byte
1318  * x | --- | skip forward one byte
1319  */
1320 
1321 static VALUE
1323 {
1324  static const char hexdigits[] = "0123456789abcdef";
1325  char *s, *send;
1326  char *p, *pend;
1327  VALUE ary;
1328  char type;
1329  long len, tmp_len;
1330  int star;
1331 #ifdef NATINT_PACK
1332  int natint; /* native integer */
1333 #endif
1334  int block_p = rb_block_given_p();
1335  int signed_p, integer_size, bigendian_p;
1336 #define UNPACK_PUSH(item) do {\
1337  VALUE item_val = (item);\
1338  if (block_p) {\
1339  rb_yield(item_val);\
1340  }\
1341  else {\
1342  rb_ary_push(ary, item_val);\
1343  }\
1344  } while (0)
1345 
1346  StringValue(str);
1347  StringValue(fmt);
1348  s = RSTRING_PTR(str);
1349  send = s + RSTRING_LEN(str);
1350  p = RSTRING_PTR(fmt);
1351  pend = p + RSTRING_LEN(fmt);
1352 
1353  ary = block_p ? Qnil : rb_ary_new();
1354  while (p < pend) {
1355  int explicit_endian = 0;
1356  type = *p++;
1357 #ifdef NATINT_PACK
1358  natint = 0;
1359 #endif
1360 
1361  if (ISSPACE(type)) continue;
1362  if (type == '#') {
1363  while ((p < pend) && (*p != '\n')) {
1364  p++;
1365  }
1366  continue;
1367  }
1368 
1369  star = 0;
1370  {
1371  static const char natstr[] = "sSiIlL";
1372  static const char endstr[] = "sSiIlLqQ";
1373 
1374  modifiers:
1375  switch (*p) {
1376  case '_':
1377  case '!':
1378 
1379  if (strchr(natstr, type)) {
1380 #ifdef NATINT_PACK
1381  natint = 1;
1382 #endif
1383  p++;
1384  }
1385  else {
1386  rb_raise(rb_eArgError, "'%c' allowed only after types %s", *p, natstr);
1387  }
1388  goto modifiers;
1389 
1390  case '<':
1391  case '>':
1392  if (!strchr(endstr, type)) {
1393  rb_raise(rb_eArgError, "'%c' allowed only after types %s", *p, endstr);
1394  }
1395  if (explicit_endian) {
1396  rb_raise(rb_eRangeError, "Can't use both '<' and '>'");
1397  }
1398  explicit_endian = *p++;
1399  goto modifiers;
1400  }
1401  }
1402 
1403  if (p >= pend)
1404  len = 1;
1405  else if (*p == '*') {
1406  star = 1;
1407  len = send - s;
1408  p++;
1409  }
1410  else if (ISDIGIT(*p)) {
1411  errno = 0;
1412  len = STRTOUL(p, (char**)&p, 10);
1413  if (errno) {
1414  rb_raise(rb_eRangeError, "pack length too big");
1415  }
1416  }
1417  else {
1418  len = (type != '@');
1419  }
1420 
1421  switch (type) {
1422  case '%':
1423  rb_raise(rb_eArgError, "%% is not supported");
1424  break;
1425 
1426  case 'A':
1427  if (len > send - s) len = send - s;
1428  {
1429  long end = len;
1430  char *t = s + len - 1;
1431 
1432  while (t >= s) {
1433  if (*t != ' ' && *t != '\0') break;
1434  t--; len--;
1435  }
1436  UNPACK_PUSH(infected_str_new(s, len, str));
1437  s += end;
1438  }
1439  break;
1440 
1441  case 'Z':
1442  {
1443  char *t = s;
1444 
1445  if (len > send-s) len = send-s;
1446  while (t < s+len && *t) t++;
1447  UNPACK_PUSH(infected_str_new(s, t-s, str));
1448  if (t < send) t++;
1449  s = star ? t : s+len;
1450  }
1451  break;
1452 
1453  case 'a':
1454  if (len > send - s) len = send - s;
1455  UNPACK_PUSH(infected_str_new(s, len, str));
1456  s += len;
1457  break;
1458 
1459  case 'b':
1460  {
1461  VALUE bitstr;
1462  char *t;
1463  int bits;
1464  long i;
1465 
1466  if (p[-1] == '*' || len > (send - s) * 8)
1467  len = (send - s) * 8;
1468  bits = 0;
1469  UNPACK_PUSH(bitstr = rb_str_new(0, len));
1470  t = RSTRING_PTR(bitstr);
1471  for (i=0; i<len; i++) {
1472  if (i & 7) bits >>= 1;
1473  else bits = *s++;
1474  *t++ = (bits & 1) ? '1' : '0';
1475  }
1476  }
1477  break;
1478 
1479  case 'B':
1480  {
1481  VALUE bitstr;
1482  char *t;
1483  int bits;
1484  long i;
1485 
1486  if (p[-1] == '*' || len > (send - s) * 8)
1487  len = (send - s) * 8;
1488  bits = 0;
1489  UNPACK_PUSH(bitstr = rb_str_new(0, len));
1490  t = RSTRING_PTR(bitstr);
1491  for (i=0; i<len; i++) {
1492  if (i & 7) bits <<= 1;
1493  else bits = *s++;
1494  *t++ = (bits & 128) ? '1' : '0';
1495  }
1496  }
1497  break;
1498 
1499  case 'h':
1500  {
1501  VALUE bitstr;
1502  char *t;
1503  int bits;
1504  long i;
1505 
1506  if (p[-1] == '*' || len > (send - s) * 2)
1507  len = (send - s) * 2;
1508  bits = 0;
1509  UNPACK_PUSH(bitstr = rb_str_new(0, len));
1510  t = RSTRING_PTR(bitstr);
1511  for (i=0; i<len; i++) {
1512  if (i & 1)
1513  bits >>= 4;
1514  else
1515  bits = *s++;
1516  *t++ = hexdigits[bits & 15];
1517  }
1518  }
1519  break;
1520 
1521  case 'H':
1522  {
1523  VALUE bitstr;
1524  char *t;
1525  int bits;
1526  long i;
1527 
1528  if (p[-1] == '*' || len > (send - s) * 2)
1529  len = (send - s) * 2;
1530  bits = 0;
1531  UNPACK_PUSH(bitstr = rb_str_new(0, len));
1532  t = RSTRING_PTR(bitstr);
1533  for (i=0; i<len; i++) {
1534  if (i & 1)
1535  bits <<= 4;
1536  else
1537  bits = *s++;
1538  *t++ = hexdigits[(bits >> 4) & 15];
1539  }
1540  }
1541  break;
1542 
1543  case 'c':
1544  PACK_LENGTH_ADJUST_SIZE(sizeof(char));
1545  while (len-- > 0) {
1546  int c = *s++;
1547  if (c > (char)127) c-=256;
1548  UNPACK_PUSH(INT2FIX(c));
1549  }
1550  PACK_ITEM_ADJUST();
1551  break;
1552 
1553  case 'C':
1554  PACK_LENGTH_ADJUST_SIZE(sizeof(unsigned char));
1555  while (len-- > 0) {
1556  unsigned char c = *s++;
1557  UNPACK_PUSH(INT2FIX(c));
1558  }
1559  PACK_ITEM_ADJUST();
1560  break;
1561 
1562  case 's':
1563  signed_p = 1;
1564  integer_size = NATINT_LEN(short, 2);
1565  bigendian_p = BIGENDIAN_P();
1566  goto unpack_integer;
1567 
1568  case 'S':
1569  signed_p = 0;
1570  integer_size = NATINT_LEN(short, 2);
1571  bigendian_p = BIGENDIAN_P();
1572  goto unpack_integer;
1573 
1574  case 'i':
1575  signed_p = 1;
1576  integer_size = (int)sizeof(int);
1577  bigendian_p = BIGENDIAN_P();
1578  goto unpack_integer;
1579 
1580  case 'I':
1581  signed_p = 0;
1582  integer_size = (int)sizeof(int);
1583  bigendian_p = BIGENDIAN_P();
1584  goto unpack_integer;
1585 
1586  case 'l':
1587  signed_p = 1;
1588  integer_size = NATINT_LEN(long, 4);
1589  bigendian_p = BIGENDIAN_P();
1590  goto unpack_integer;
1591 
1592  case 'L':
1593  signed_p = 0;
1594  integer_size = NATINT_LEN(long, 4);
1595  bigendian_p = BIGENDIAN_P();
1596  goto unpack_integer;
1597 
1598  case 'q':
1599  signed_p = 1;
1600  integer_size = 8;
1601  bigendian_p = BIGENDIAN_P();
1602  goto unpack_integer;
1603 
1604  case 'Q':
1605  signed_p = 0;
1606  integer_size = 8;
1607  bigendian_p = BIGENDIAN_P();
1608  goto unpack_integer;
1609 
1610  case 'n':
1611  signed_p = 0;
1612  integer_size = 2;
1613  bigendian_p = 1;
1614  goto unpack_integer;
1615 
1616  case 'N':
1617  signed_p = 0;
1618  integer_size = 4;
1619  bigendian_p = 1;
1620  goto unpack_integer;
1621 
1622  case 'v':
1623  signed_p = 0;
1624  integer_size = 2;
1625  bigendian_p = 0;
1626  goto unpack_integer;
1627 
1628  case 'V':
1629  signed_p = 0;
1630  integer_size = 4;
1631  bigendian_p = 0;
1632  goto unpack_integer;
1633 
1634  unpack_integer:
1635  if (explicit_endian) {
1636  bigendian_p = explicit_endian == '>';
1637  }
1638 
1639  switch (integer_size) {
1640 #if defined(HAVE_INT16_T) && !defined(FORCE_BIG_PACK)
1641  case SIZEOF_INT16_T:
1642  if (signed_p) {
1643  PACK_LENGTH_ADJUST_SIZE(sizeof(int16_t));
1644  while (len-- > 0) {
1645  union {
1646  int16_t i;
1647  char a[sizeof(int16_t)];
1648  } v;
1649  memcpy(v.a, s, sizeof(int16_t));
1650  if (bigendian_p != BIGENDIAN_P()) v.i = swap16(v.i);
1651  s += sizeof(int16_t);
1652  UNPACK_PUSH(INT2FIX(v.i));
1653  }
1654  PACK_ITEM_ADJUST();
1655  }
1656  else {
1657  PACK_LENGTH_ADJUST_SIZE(sizeof(uint16_t));
1658  while (len-- > 0) {
1659  union {
1660  uint16_t i;
1661  char a[sizeof(uint16_t)];
1662  } v;
1663  memcpy(v.a, s, sizeof(uint16_t));
1664  if (bigendian_p != BIGENDIAN_P()) v.i = swap16(v.i);
1665  s += sizeof(uint16_t);
1666  UNPACK_PUSH(INT2FIX(v.i));
1667  }
1668  PACK_ITEM_ADJUST();
1669  }
1670  break;
1671 #endif
1672 
1673 #if defined(HAVE_INT32_T) && !defined(FORCE_BIG_PACK)
1674  case SIZEOF_INT32_T:
1675  if (signed_p) {
1676  PACK_LENGTH_ADJUST_SIZE(sizeof(int32_t));
1677  while (len-- > 0) {
1678  union {
1679  int32_t i;
1680  char a[sizeof(int32_t)];
1681  } v;
1682  memcpy(v.a, s, sizeof(int32_t));
1683  if (bigendian_p != BIGENDIAN_P()) v.i = swap32(v.i);
1684  s += sizeof(int32_t);
1685  UNPACK_PUSH(INT2NUM(v.i));
1686  }
1687  PACK_ITEM_ADJUST();
1688  }
1689  else {
1691  while (len-- > 0) {
1692  union {
1693  uint32_t i;
1694  char a[sizeof(uint32_t)];
1695  } v;
1696  memcpy(v.a, s, sizeof(uint32_t));
1697  if (bigendian_p != BIGENDIAN_P()) v.i = swap32(v.i);
1698  s += sizeof(uint32_t);
1699  UNPACK_PUSH(UINT2NUM(v.i));
1700  }
1701  PACK_ITEM_ADJUST();
1702  }
1703  break;
1704 #endif
1705 
1706 #if defined(HAVE_INT64_T) && !defined(FORCE_BIG_PACK)
1707  case SIZEOF_INT64_T:
1708  if (signed_p) {
1709  PACK_LENGTH_ADJUST_SIZE(sizeof(int64_t));
1710  while (len-- > 0) {
1711  union {
1712  int64_t i;
1713  char a[sizeof(int64_t)];
1714  } v;
1715  memcpy(v.a, s, sizeof(int64_t));
1716  if (bigendian_p != BIGENDIAN_P()) v.i = swap64(v.i);
1717  s += sizeof(int64_t);
1718  UNPACK_PUSH(INT64toNUM(v.i));
1719  }
1720  PACK_ITEM_ADJUST();
1721  }
1722  else {
1724  while (len-- > 0) {
1725  union {
1726  uint64_t i;
1727  char a[sizeof(uint64_t)];
1728  } v;
1729  memcpy(v.a, s, sizeof(uint64_t));
1730  if (bigendian_p != BIGENDIAN_P()) v.i = swap64(v.i);
1731  s += sizeof(uint64_t);
1732  UNPACK_PUSH(UINT64toNUM(v.i));
1733  }
1734  PACK_ITEM_ADJUST();
1735  }
1736  break;
1737 #endif
1738 
1739  default:
1740  if (integer_size > MAX_INTEGER_PACK_SIZE)
1741  rb_bug("unexpected intger size for pack: %d", integer_size);
1742  PACK_LENGTH_ADJUST_SIZE(integer_size);
1743  while (len-- > 0) {
1744  union {
1745  unsigned long i[(MAX_INTEGER_PACK_SIZE+SIZEOF_LONG)/SIZEOF_LONG];
1746  char a[(MAX_INTEGER_PACK_SIZE+SIZEOF_LONG)/SIZEOF_LONG*SIZEOF_LONG];
1747  } v;
1748  int num_longs = (integer_size+SIZEOF_LONG)/SIZEOF_LONG;
1749  int i;
1750 
1751  if (signed_p && (signed char)s[bigendian_p ? 0 : (integer_size-1)] < 0)
1752  memset(v.a, 0xff, sizeof(long)*num_longs);
1753  else
1754  memset(v.a, 0, sizeof(long)*num_longs);
1755  if (bigendian_p)
1756  memcpy(v.a + sizeof(long)*num_longs - integer_size, s, integer_size);
1757  else
1758  memcpy(v.a, s, integer_size);
1759  if (bigendian_p) {
1760  for (i = 0; i < num_longs/2; i++) {
1761  unsigned long t = v.i[i];
1762  v.i[i] = v.i[num_longs-1-i];
1763  v.i[num_longs-1-i] = t;
1764  }
1765  }
1766  if (bigendian_p != BIGENDIAN_P()) {
1767  for (i = 0; i < num_longs; i++)
1768  v.i[i] = swapl(v.i[i]);
1769  }
1770  s += integer_size;
1771  UNPACK_PUSH(rb_big_unpack(v.i, num_longs));
1772  }
1773  PACK_ITEM_ADJUST();
1774  break;
1775  }
1776  break;
1777 
1778  case 'f':
1779  case 'F':
1780  PACK_LENGTH_ADJUST_SIZE(sizeof(float));
1781  while (len-- > 0) {
1782  float tmp;
1783  memcpy(&tmp, s, sizeof(float));
1784  s += sizeof(float);
1785  UNPACK_PUSH(DBL2NUM((double)tmp));
1786  }
1787  PACK_ITEM_ADJUST();
1788  break;
1789 
1790  case 'e':
1791  PACK_LENGTH_ADJUST_SIZE(sizeof(float));
1792  while (len-- > 0) {
1793  float tmp;
1794  FLOAT_CONVWITH(ftmp);
1795 
1796  memcpy(&tmp, s, sizeof(float));
1797  s += sizeof(float);
1798  tmp = VTOHF(tmp,ftmp);
1799  UNPACK_PUSH(DBL2NUM((double)tmp));
1800  }
1801  PACK_ITEM_ADJUST();
1802  break;
1803 
1804  case 'E':
1805  PACK_LENGTH_ADJUST_SIZE(sizeof(double));
1806  while (len-- > 0) {
1807  double tmp;
1808  DOUBLE_CONVWITH(dtmp);
1809 
1810  memcpy(&tmp, s, sizeof(double));
1811  s += sizeof(double);
1812  tmp = VTOHD(tmp,dtmp);
1813  UNPACK_PUSH(DBL2NUM(tmp));
1814  }
1815  PACK_ITEM_ADJUST();
1816  break;
1817 
1818  case 'D':
1819  case 'd':
1820  PACK_LENGTH_ADJUST_SIZE(sizeof(double));
1821  while (len-- > 0) {
1822  double tmp;
1823  memcpy(&tmp, s, sizeof(double));
1824  s += sizeof(double);
1825  UNPACK_PUSH(DBL2NUM(tmp));
1826  }
1827  PACK_ITEM_ADJUST();
1828  break;
1829 
1830  case 'g':
1831  PACK_LENGTH_ADJUST_SIZE(sizeof(float));
1832  while (len-- > 0) {
1833  float tmp;
1834  FLOAT_CONVWITH(ftmp);
1835 
1836  memcpy(&tmp, s, sizeof(float));
1837  s += sizeof(float);
1838  tmp = NTOHF(tmp,ftmp);
1839  UNPACK_PUSH(DBL2NUM((double)tmp));
1840  }
1841  PACK_ITEM_ADJUST();
1842  break;
1843 
1844  case 'G':
1845  PACK_LENGTH_ADJUST_SIZE(sizeof(double));
1846  while (len-- > 0) {
1847  double tmp;
1848  DOUBLE_CONVWITH(dtmp);
1849 
1850  memcpy(&tmp, s, sizeof(double));
1851  s += sizeof(double);
1852  tmp = NTOHD(tmp,dtmp);
1853  UNPACK_PUSH(DBL2NUM(tmp));
1854  }
1855  PACK_ITEM_ADJUST();
1856  break;
1857 
1858  case 'U':
1859  if (len > send - s) len = send - s;
1860  while (len > 0 && s < send) {
1861  long alen = send - s;
1862  unsigned long l;
1863 
1864  l = utf8_to_uv(s, &alen);
1865  s += alen; len--;
1866  UNPACK_PUSH(ULONG2NUM(l));
1867  }
1868  break;
1869 
1870  case 'u':
1871  {
1872  VALUE buf = infected_str_new(0, (send - s)*3/4, str);
1873  char *ptr = RSTRING_PTR(buf);
1874  long total = 0;
1875 
1876  while (s < send && *s > ' ' && *s < 'a') {
1877  long a,b,c,d;
1878  char hunk[4];
1879 
1880  hunk[3] = '\0';
1881  len = (*s++ - ' ') & 077;
1882  total += len;
1883  if (total > RSTRING_LEN(buf)) {
1884  len -= total - RSTRING_LEN(buf);
1885  total = RSTRING_LEN(buf);
1886  }
1887 
1888  while (len > 0) {
1889  long mlen = len > 3 ? 3 : len;
1890 
1891  if (s < send && *s >= ' ')
1892  a = (*s++ - ' ') & 077;
1893  else
1894  a = 0;
1895  if (s < send && *s >= ' ')
1896  b = (*s++ - ' ') & 077;
1897  else
1898  b = 0;
1899  if (s < send && *s >= ' ')
1900  c = (*s++ - ' ') & 077;
1901  else
1902  c = 0;
1903  if (s < send && *s >= ' ')
1904  d = (*s++ - ' ') & 077;
1905  else
1906  d = 0;
1907  hunk[0] = (char)(a << 2 | b >> 4);
1908  hunk[1] = (char)(b << 4 | c >> 2);
1909  hunk[2] = (char)(c << 6 | d);
1910  memcpy(ptr, hunk, mlen);
1911  ptr += mlen;
1912  len -= mlen;
1913  }
1914  if (*s == '\r') s++;
1915  if (*s == '\n') s++;
1916  else if (s < send && (s+1 == send || s[1] == '\n'))
1917  s += 2; /* possible checksum byte */
1918  }
1919 
1920  rb_str_set_len(buf, total);
1921  UNPACK_PUSH(buf);
1922  }
1923  break;
1924 
1925  case 'm':
1926  {
1927  VALUE buf = infected_str_new(0, (send - s)*3/4, str);
1928  char *ptr = RSTRING_PTR(buf);
1929  int a = -1,b = -1,c = 0,d = 0;
1930  static signed char b64_xtable[256];
1931 
1932  if (b64_xtable['/'] <= 0) {
1933  int i;
1934 
1935  for (i = 0; i < 256; i++) {
1936  b64_xtable[i] = -1;
1937  }
1938  for (i = 0; i < 64; i++) {
1939  b64_xtable[(unsigned char)b64_table[i]] = i;
1940  }
1941  }
1942  if (len == 0) {
1943  while (s < send) {
1944  a = b = c = d = -1;
1945  a = b64_xtable[(unsigned char)*s++];
1946  if (s >= send || a == -1) rb_raise(rb_eArgError, "invalid base64");
1947  b = b64_xtable[(unsigned char)*s++];
1948  if (s >= send || b == -1) rb_raise(rb_eArgError, "invalid base64");
1949  if (*s == '=') {
1950  if (s + 2 == send && *(s + 1) == '=') break;
1951  rb_raise(rb_eArgError, "invalid base64");
1952  }
1953  c = b64_xtable[(unsigned char)*s++];
1954  if (s >= send || c == -1) rb_raise(rb_eArgError, "invalid base64");
1955  if (s + 1 == send && *s == '=') break;
1956  d = b64_xtable[(unsigned char)*s++];
1957  if (d == -1) rb_raise(rb_eArgError, "invalid base64");
1958  *ptr++ = a << 2 | b >> 4;
1959  *ptr++ = b << 4 | c >> 2;
1960  *ptr++ = c << 6 | d;
1961  }
1962  if (c == -1) {
1963  *ptr++ = a << 2 | b >> 4;
1964  if (b & 0xf) rb_raise(rb_eArgError, "invalid base64");
1965  }
1966  else if (d == -1) {
1967  *ptr++ = a << 2 | b >> 4;
1968  *ptr++ = b << 4 | c >> 2;
1969  if (c & 0x3) rb_raise(rb_eArgError, "invalid base64");
1970  }
1971  }
1972  else {
1973  while (s < send) {
1974  a = b = c = d = -1;
1975  while ((a = b64_xtable[(unsigned char)*s]) == -1 && s < send) {s++;}
1976  if (s >= send) break;
1977  s++;
1978  while ((b = b64_xtable[(unsigned char)*s]) == -1 && s < send) {s++;}
1979  if (s >= send) break;
1980  s++;
1981  while ((c = b64_xtable[(unsigned char)*s]) == -1 && s < send) {if (*s == '=') break; s++;}
1982  if (*s == '=' || s >= send) break;
1983  s++;
1984  while ((d = b64_xtable[(unsigned char)*s]) == -1 && s < send) {if (*s == '=') break; s++;}
1985  if (*s == '=' || s >= send) break;
1986  s++;
1987  *ptr++ = a << 2 | b >> 4;
1988  *ptr++ = b << 4 | c >> 2;
1989  *ptr++ = c << 6 | d;
1990  }
1991  if (a != -1 && b != -1) {
1992  if (c == -1 && *s == '=')
1993  *ptr++ = a << 2 | b >> 4;
1994  else if (c != -1 && *s == '=') {
1995  *ptr++ = a << 2 | b >> 4;
1996  *ptr++ = b << 4 | c >> 2;
1997  }
1998  }
1999  }
2000  rb_str_set_len(buf, ptr - RSTRING_PTR(buf));
2001  UNPACK_PUSH(buf);
2002  }
2003  break;
2004 
2005  case 'M':
2006  {
2007  VALUE buf = infected_str_new(0, send - s, str);
2008  char *ptr = RSTRING_PTR(buf), *ss = s;
2009  int c1, c2;
2010 
2011  while (s < send) {
2012  if (*s == '=') {
2013  if (++s == send) break;
2014  if (s+1 < send && *s == '\r' && *(s+1) == '\n')
2015  s++;
2016  if (*s != '\n') {
2017  if ((c1 = hex2num(*s)) == -1) break;
2018  if (++s == send) break;
2019  if ((c2 = hex2num(*s)) == -1) break;
2020  *ptr++ = c1 << 4 | c2;
2021  }
2022  }
2023  else {
2024  *ptr++ = *s;
2025  }
2026  s++;
2027  ss = s;
2028  }
2029  rb_str_set_len(buf, ptr - RSTRING_PTR(buf));
2030  rb_str_buf_cat(buf, ss, send-ss);
2032  UNPACK_PUSH(buf);
2033  }
2034  break;
2035 
2036  case '@':
2037  if (len > RSTRING_LEN(str))
2038  rb_raise(rb_eArgError, "@ outside of string");
2039  s = RSTRING_PTR(str) + len;
2040  break;
2041 
2042  case 'X':
2043  if (len > s - RSTRING_PTR(str))
2044  rb_raise(rb_eArgError, "X outside of string");
2045  s -= len;
2046  break;
2047 
2048  case 'x':
2049  if (len > send - s)
2050  rb_raise(rb_eArgError, "x outside of string");
2051  s += len;
2052  break;
2053 
2054  case 'P':
2055  if (sizeof(char *) <= (size_t)(send - s)) {
2056  VALUE tmp = Qnil;
2057  char *t;
2058 
2059  memcpy(&t, s, sizeof(char *));
2060  s += sizeof(char *);
2061 
2062  if (t) {
2063  VALUE a, *p, *pend;
2064 
2065  if (!(a = rb_str_associated(str))) {
2066  rb_raise(rb_eArgError, "no associated pointer");
2067  }
2068  p = RARRAY_PTR(a);
2069  pend = p + RARRAY_LEN(a);
2070  while (p < pend) {
2071  if (TYPE(*p) == T_STRING && RSTRING_PTR(*p) == t) {
2072  if (len < RSTRING_LEN(*p)) {
2073  tmp = rb_tainted_str_new(t, len);
2074  rb_str_associate(tmp, a);
2075  }
2076  else {
2077  tmp = *p;
2078  }
2079  break;
2080  }
2081  p++;
2082  }
2083  if (p == pend) {
2084  rb_raise(rb_eArgError, "non associated pointer");
2085  }
2086  }
2087  UNPACK_PUSH(tmp);
2088  }
2089  break;
2090 
2091  case 'p':
2092  if (len > (long)((send - s) / sizeof(char *)))
2093  len = (send - s) / sizeof(char *);
2094  while (len-- > 0) {
2095  if ((size_t)(send - s) < sizeof(char *))
2096  break;
2097  else {
2098  VALUE tmp = Qnil;
2099  char *t;
2100 
2101  memcpy(&t, s, sizeof(char *));
2102  s += sizeof(char *);
2103 
2104  if (t) {
2105  VALUE a, *p, *pend;
2106 
2107  if (!(a = rb_str_associated(str))) {
2108  rb_raise(rb_eArgError, "no associated pointer");
2109  }
2110  p = RARRAY_PTR(a);
2111  pend = p + RARRAY_LEN(a);
2112  while (p < pend) {
2113  if (TYPE(*p) == T_STRING && RSTRING_PTR(*p) == t) {
2114  tmp = *p;
2115  break;
2116  }
2117  p++;
2118  }
2119  if (p == pend) {
2120  rb_raise(rb_eArgError, "non associated pointer");
2121  }
2122  }
2123  UNPACK_PUSH(tmp);
2124  }
2125  }
2126  break;
2127 
2128  case 'w':
2129  {
2130  unsigned long ul = 0;
2131  unsigned long ulmask = 0xfeUL << ((sizeof(unsigned long) - 1) * 8);
2132 
2133  while (len > 0 && s < send) {
2134  ul <<= 7;
2135  ul |= (*s & 0x7f);
2136  if (!(*s++ & 0x80)) {
2137  UNPACK_PUSH(ULONG2NUM(ul));
2138  len--;
2139  ul = 0;
2140  }
2141  else if (ul & ulmask) {
2142  VALUE big = rb_uint2big(ul);
2143  VALUE big128 = rb_uint2big(128);
2144  while (s < send) {
2145  big = rb_big_mul(big, big128);
2146  big = rb_big_plus(big, rb_uint2big(*s & 0x7f));
2147  if (!(*s++ & 0x80)) {
2148  UNPACK_PUSH(big);
2149  len--;
2150  ul = 0;
2151  break;
2152  }
2153  }
2154  }
2155  }
2156  }
2157  break;
2158 
2159  default:
2160  break;
2161  }
2162  }
2163 
2164  return ary;
2165 }
2166 
2167 #define BYTEWIDTH 8
2168 
2169 int
2170 rb_uv_to_utf8(char buf[6], unsigned long uv)
2171 {
2172  if (uv <= 0x7f) {
2173  buf[0] = (char)uv;
2174  return 1;
2175  }
2176  if (uv <= 0x7ff) {
2177  buf[0] = (char)((uv>>6)&0xff)|0xc0;
2178  buf[1] = (char)(uv&0x3f)|0x80;
2179  return 2;
2180  }
2181  if (uv <= 0xffff) {
2182  buf[0] = (char)((uv>>12)&0xff)|0xe0;
2183  buf[1] = (char)((uv>>6)&0x3f)|0x80;
2184  buf[2] = (char)(uv&0x3f)|0x80;
2185  return 3;
2186  }
2187  if (uv <= 0x1fffff) {
2188  buf[0] = (char)((uv>>18)&0xff)|0xf0;
2189  buf[1] = (char)((uv>>12)&0x3f)|0x80;
2190  buf[2] = (char)((uv>>6)&0x3f)|0x80;
2191  buf[3] = (char)(uv&0x3f)|0x80;
2192  return 4;
2193  }
2194  if (uv <= 0x3ffffff) {
2195  buf[0] = (char)((uv>>24)&0xff)|0xf8;
2196  buf[1] = (char)((uv>>18)&0x3f)|0x80;
2197  buf[2] = (char)((uv>>12)&0x3f)|0x80;
2198  buf[3] = (char)((uv>>6)&0x3f)|0x80;
2199  buf[4] = (char)(uv&0x3f)|0x80;
2200  return 5;
2201  }
2202  if (uv <= 0x7fffffff) {
2203  buf[0] = (char)((uv>>30)&0xff)|0xfc;
2204  buf[1] = (char)((uv>>24)&0x3f)|0x80;
2205  buf[2] = (char)((uv>>18)&0x3f)|0x80;
2206  buf[3] = (char)((uv>>12)&0x3f)|0x80;
2207  buf[4] = (char)((uv>>6)&0x3f)|0x80;
2208  buf[5] = (char)(uv&0x3f)|0x80;
2209  return 6;
2210  }
2211  rb_raise(rb_eRangeError, "pack(U): value out of range");
2212 }
2213 
2214 static const unsigned long utf8_limits[] = {
2215  0x0, /* 1 */
2216  0x80, /* 2 */
2217  0x800, /* 3 */
2218  0x10000, /* 4 */
2219  0x200000, /* 5 */
2220  0x4000000, /* 6 */
2221  0x80000000, /* 7 */
2222 };
2223 
2224 static unsigned long
2225 utf8_to_uv(const char *p, long *lenp)
2226 {
2227  int c = *p++ & 0xff;
2228  unsigned long uv = c;
2229  long n;
2230 
2231  if (!(uv & 0x80)) {
2232  *lenp = 1;
2233  return uv;
2234  }
2235  if (!(uv & 0x40)) {
2236  *lenp = 1;
2237  rb_raise(rb_eArgError, "malformed UTF-8 character");
2238  }
2239 
2240  if (!(uv & 0x20)) { n = 2; uv &= 0x1f; }
2241  else if (!(uv & 0x10)) { n = 3; uv &= 0x0f; }
2242  else if (!(uv & 0x08)) { n = 4; uv &= 0x07; }
2243  else if (!(uv & 0x04)) { n = 5; uv &= 0x03; }
2244  else if (!(uv & 0x02)) { n = 6; uv &= 0x01; }
2245  else {
2246  *lenp = 1;
2247  rb_raise(rb_eArgError, "malformed UTF-8 character");
2248  }
2249  if (n > *lenp) {
2250  rb_raise(rb_eArgError, "malformed UTF-8 character (expected %ld bytes, given %ld bytes)",
2251  n, *lenp);
2252  }
2253  *lenp = n--;
2254  if (n != 0) {
2255  while (n--) {
2256  c = *p++ & 0xff;
2257  if ((c & 0xc0) != 0x80) {
2258  *lenp -= n + 1;
2259  rb_raise(rb_eArgError, "malformed UTF-8 character");
2260  }
2261  else {
2262  c &= 0x3f;
2263  uv = uv << 6 | c;
2264  }
2265  }
2266  }
2267  n = *lenp - 1;
2268  if (uv < utf8_limits[n]) {
2269  rb_raise(rb_eArgError, "redundant UTF-8 sequence");
2270  }
2271  return uv;
2272 }
2273 
2274 void
2276 {
2277  rb_define_method(rb_cArray, "pack", pack_pack, 1);
2278  rb_define_method(rb_cString, "unpack", pack_unpack, 1);
2279 }
2280