Ruby  1.9.3p448(2013-06-27revision41675)
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  char buff[4096];
1072  long i = 0;
1073  const char *trans = type == 'u' ? uu_table : b64_table;
1074  int padding;
1075 
1076  if (type == 'u') {
1077  buff[i++] = (char)len + ' ';
1078  padding = '`';
1079  }
1080  else {
1081  padding = '=';
1082  }
1083  while (len >= 3) {
1084  while (len >= 3 && sizeof(buff)-i >= 4) {
1085  buff[i++] = trans[077 & (*s >> 2)];
1086  buff[i++] = trans[077 & (((*s << 4) & 060) | ((s[1] >> 4) & 017))];
1087  buff[i++] = trans[077 & (((s[1] << 2) & 074) | ((s[2] >> 6) & 03))];
1088  buff[i++] = trans[077 & s[2]];
1089  s += 3;
1090  len -= 3;
1091  }
1092  if (sizeof(buff)-i < 4) {
1093  rb_str_buf_cat(str, buff, i);
1094  i = 0;
1095  }
1096  }
1097 
1098  if (len == 2) {
1099  buff[i++] = trans[077 & (*s >> 2)];
1100  buff[i++] = trans[077 & (((*s << 4) & 060) | ((s[1] >> 4) & 017))];
1101  buff[i++] = trans[077 & (((s[1] << 2) & 074) | (('\0' >> 6) & 03))];
1102  buff[i++] = padding;
1103  }
1104  else if (len == 1) {
1105  buff[i++] = trans[077 & (*s >> 2)];
1106  buff[i++] = trans[077 & (((*s << 4) & 060) | (('\0' >> 4) & 017))];
1107  buff[i++] = padding;
1108  buff[i++] = padding;
1109  }
1110  if (tail_lf) buff[i++] = '\n';
1111  rb_str_buf_cat(str, buff, i);
1112 }
1113 
1114 static const char hex_table[] = "0123456789ABCDEF";
1115 
1116 static void
1117 qpencode(VALUE str, VALUE from, long len)
1118 {
1119  char buff[1024];
1120  long i = 0, n = 0, prev = EOF;
1121  unsigned char *s = (unsigned char*)RSTRING_PTR(from);
1122  unsigned char *send = s + RSTRING_LEN(from);
1123 
1124  while (s < send) {
1125  if ((*s > 126) ||
1126  (*s < 32 && *s != '\n' && *s != '\t') ||
1127  (*s == '=')) {
1128  buff[i++] = '=';
1129  buff[i++] = hex_table[*s >> 4];
1130  buff[i++] = hex_table[*s & 0x0f];
1131  n += 3;
1132  prev = EOF;
1133  }
1134  else if (*s == '\n') {
1135  if (prev == ' ' || prev == '\t') {
1136  buff[i++] = '=';
1137  buff[i++] = *s;
1138  }
1139  buff[i++] = *s;
1140  n = 0;
1141  prev = *s;
1142  }
1143  else {
1144  buff[i++] = *s;
1145  n++;
1146  prev = *s;
1147  }
1148  if (n > len) {
1149  buff[i++] = '=';
1150  buff[i++] = '\n';
1151  n = 0;
1152  prev = '\n';
1153  }
1154  if (i > 1024 - 5) {
1155  rb_str_buf_cat(str, buff, i);
1156  i = 0;
1157  }
1158  s++;
1159  }
1160  if (n > 0) {
1161  buff[i++] = '=';
1162  buff[i++] = '\n';
1163  }
1164  if (i > 0) {
1165  rb_str_buf_cat(str, buff, i);
1166  }
1167 }
1168 
1169 static inline int
1170 hex2num(char c)
1171 {
1172  switch (c) {
1173  case '0': case '1': case '2': case '3': case '4':
1174  case '5': case '6': case '7': case '8': case '9':
1175  return c - '0';
1176  case 'a': case 'b': case 'c':
1177  case 'd': case 'e': case 'f':
1178  return c - 'a' + 10;
1179  case 'A': case 'B': case 'C':
1180  case 'D': case 'E': case 'F':
1181  return c - 'A' + 10;
1182  default:
1183  return -1;
1184  }
1185 }
1186 
1187 #define PACK_LENGTH_ADJUST_SIZE(sz) do { \
1188  tmp_len = 0; \
1189  if (len > (long)((send-s)/(sz))) { \
1190  if (!star) { \
1191  tmp_len = len-(send-s)/(sz); \
1192  } \
1193  len = (send-s)/(sz); \
1194  } \
1195 } while (0)
1196 
1197 #define PACK_ITEM_ADJUST() do { \
1198  if (tmp_len > 0 && !block_p) \
1199  rb_ary_store(ary, RARRAY_LEN(ary)+tmp_len-1, Qnil); \
1200 } while (0)
1201 
1202 static VALUE
1203 infected_str_new(const char *ptr, long len, VALUE str)
1204 {
1205  VALUE s = rb_str_new(ptr, len);
1206 
1207  OBJ_INFECT(s, str);
1208  return s;
1209 }
1210 
1211 /*
1212  * call-seq:
1213  * str.unpack(format) -> anArray
1214  *
1215  * Decodes <i>str</i> (which may contain binary data) according to the
1216  * format string, returning an array of each value extracted. The
1217  * format string consists of a sequence of single-character directives,
1218  * summarized in the table at the end of this entry.
1219  * Each directive may be followed
1220  * by a number, indicating the number of times to repeat with this
1221  * directive. An asterisk (``<code>*</code>'') will use up all
1222  * remaining elements. The directives <code>sSiIlL</code> may each be
1223  * followed by an underscore (``<code>_</code>'') or
1224  * exclamation mark (``<code>!</code>'') to use the underlying
1225  * platform's native size for the specified type; otherwise, it uses a
1226  * platform-independent consistent size. Spaces are ignored in the
1227  * format string. See also <code>Array#pack</code>.
1228  *
1229  * "abc \0\0abc \0\0".unpack('A6Z6') #=> ["abc", "abc "]
1230  * "abc \0\0".unpack('a3a3') #=> ["abc", " \000\000"]
1231  * "abc \0abc \0".unpack('Z*Z*') #=> ["abc ", "abc "]
1232  * "aa".unpack('b8B8') #=> ["10000110", "01100001"]
1233  * "aaa".unpack('h2H2c') #=> ["16", "61", 97]
1234  * "\xfe\xff\xfe\xff".unpack('sS') #=> [-2, 65534]
1235  * "now=20is".unpack('M*') #=> ["now is"]
1236  * "whole".unpack('xax2aX2aX1aX2a') #=> ["h", "e", "l", "l", "o"]
1237  *
1238  * This table summarizes the various formats and the Ruby classes
1239  * returned by each.
1240  *
1241  * Integer | |
1242  * Directive | Returns | Meaning
1243  * -----------------------------------------------------------------
1244  * C | Integer | 8-bit unsigned (unsigned char)
1245  * S | Integer | 16-bit unsigned, native endian (uint16_t)
1246  * L | Integer | 32-bit unsigned, native endian (uint32_t)
1247  * Q | Integer | 64-bit unsigned, native endian (uint64_t)
1248  * | |
1249  * c | Integer | 8-bit signed (signed char)
1250  * s | Integer | 16-bit signed, native endian (int16_t)
1251  * l | Integer | 32-bit signed, native endian (int32_t)
1252  * q | Integer | 64-bit signed, native endian (int64_t)
1253  * | |
1254  * S_, S! | Integer | unsigned short, native endian
1255  * I, I_, I! | Integer | unsigned int, native endian
1256  * L_, L! | Integer | unsigned long, native endian
1257  * | |
1258  * s_, s! | Integer | signed short, native endian
1259  * i, i_, i! | Integer | signed int, native endian
1260  * l_, l! | Integer | signed long, native endian
1261  * | |
1262  * S> L> Q> | Integer | same as the directives without ">" except
1263  * s> l> q> | | big endian
1264  * S!> I!> | | (available since Ruby 1.9.3)
1265  * L!> Q!> | | "S>" is same as "n"
1266  * s!> i!> | | "L>" is same as "N"
1267  * l!> q!> | |
1268  * | |
1269  * S< L< Q< | Integer | same as the directives without "<" except
1270  * s< l< q< | | little endian
1271  * S!< I!< | | (available since Ruby 1.9.3)
1272  * L!< Q!< | | "S<" is same as "v"
1273  * s!< i!< | | "L<" is same as "V"
1274  * l!< q!< | |
1275  * | |
1276  * n | Integer | 16-bit unsigned, network (big-endian) byte order
1277  * N | Integer | 32-bit unsigned, network (big-endian) byte order
1278  * v | Integer | 16-bit unsigned, VAX (little-endian) byte order
1279  * V | Integer | 32-bit unsigned, VAX (little-endian) byte order
1280  * | |
1281  * U | Integer | UTF-8 character
1282  * w | Integer | BER-compressed integer (see Array.pack)
1283  *
1284  * Float | |
1285  * Directive | Returns | Meaning
1286  * -----------------------------------------------------------------
1287  * D, d | Float | double-precision, native format
1288  * F, f | Float | single-precision, native format
1289  * E | Float | double-precision, little-endian byte order
1290  * e | Float | single-precision, little-endian byte order
1291  * G | Float | double-precision, network (big-endian) byte order
1292  * g | Float | single-precision, network (big-endian) byte order
1293  *
1294  * String | |
1295  * Directive | Returns | Meaning
1296  * -----------------------------------------------------------------
1297  * A | String | arbitrary binary string (remove trailing nulls and ASCII spaces)
1298  * a | String | arbitrary binary string
1299  * Z | String | null-terminated string
1300  * B | String | bit string (MSB first)
1301  * b | String | bit string (LSB first)
1302  * H | String | hex string (high nibble first)
1303  * h | String | hex string (low nibble first)
1304  * u | String | UU-encoded string
1305  * M | String | quoted-printable, MIME encoding (see RFC2045)
1306  * m | String | base64 encoded string (RFC 2045) (default)
1307  * | | base64 encoded string (RFC 4648) if followed by 0
1308  * P | String | pointer to a structure (fixed-length string)
1309  * p | String | pointer to a null-terminated string
1310  *
1311  * Misc. | |
1312  * Directive | Returns | Meaning
1313  * -----------------------------------------------------------------
1314  * @ | --- | skip to the offset given by the length argument
1315  * X | --- | skip backward one byte
1316  * x | --- | skip forward one byte
1317  */
1318 
1319 static VALUE
1321 {
1322  static const char hexdigits[] = "0123456789abcdef";
1323  char *s, *send;
1324  char *p, *pend;
1325  VALUE ary;
1326  char type;
1327  long len, tmp_len;
1328  int star;
1329 #ifdef NATINT_PACK
1330  int natint; /* native integer */
1331 #endif
1332  int block_p = rb_block_given_p();
1333  int signed_p, integer_size, bigendian_p;
1334 #define UNPACK_PUSH(item) do {\
1335  VALUE item_val = (item);\
1336  if (block_p) {\
1337  rb_yield(item_val);\
1338  }\
1339  else {\
1340  rb_ary_push(ary, item_val);\
1341  }\
1342  } while (0)
1343 
1344  StringValue(str);
1345  StringValue(fmt);
1346  s = RSTRING_PTR(str);
1347  send = s + RSTRING_LEN(str);
1348  p = RSTRING_PTR(fmt);
1349  pend = p + RSTRING_LEN(fmt);
1350 
1351  ary = block_p ? Qnil : rb_ary_new();
1352  while (p < pend) {
1353  int explicit_endian = 0;
1354  type = *p++;
1355 #ifdef NATINT_PACK
1356  natint = 0;
1357 #endif
1358 
1359  if (ISSPACE(type)) continue;
1360  if (type == '#') {
1361  while ((p < pend) && (*p != '\n')) {
1362  p++;
1363  }
1364  continue;
1365  }
1366 
1367  star = 0;
1368  {
1369  static const char natstr[] = "sSiIlL";
1370  static const char endstr[] = "sSiIlLqQ";
1371 
1372  modifiers:
1373  switch (*p) {
1374  case '_':
1375  case '!':
1376 
1377  if (strchr(natstr, type)) {
1378 #ifdef NATINT_PACK
1379  natint = 1;
1380 #endif
1381  p++;
1382  }
1383  else {
1384  rb_raise(rb_eArgError, "'%c' allowed only after types %s", *p, natstr);
1385  }
1386  goto modifiers;
1387 
1388  case '<':
1389  case '>':
1390  if (!strchr(endstr, type)) {
1391  rb_raise(rb_eArgError, "'%c' allowed only after types %s", *p, endstr);
1392  }
1393  if (explicit_endian) {
1394  rb_raise(rb_eRangeError, "Can't use both '<' and '>'");
1395  }
1396  explicit_endian = *p++;
1397  goto modifiers;
1398  }
1399  }
1400 
1401  if (p >= pend)
1402  len = 1;
1403  else if (*p == '*') {
1404  star = 1;
1405  len = send - s;
1406  p++;
1407  }
1408  else if (ISDIGIT(*p)) {
1409  errno = 0;
1410  len = STRTOUL(p, (char**)&p, 10);
1411  if (errno) {
1412  rb_raise(rb_eRangeError, "pack length too big");
1413  }
1414  }
1415  else {
1416  len = (type != '@');
1417  }
1418 
1419  switch (type) {
1420  case '%':
1421  rb_raise(rb_eArgError, "%% is not supported");
1422  break;
1423 
1424  case 'A':
1425  if (len > send - s) len = send - s;
1426  {
1427  long end = len;
1428  char *t = s + len - 1;
1429 
1430  while (t >= s) {
1431  if (*t != ' ' && *t != '\0') break;
1432  t--; len--;
1433  }
1434  UNPACK_PUSH(infected_str_new(s, len, str));
1435  s += end;
1436  }
1437  break;
1438 
1439  case 'Z':
1440  {
1441  char *t = s;
1442 
1443  if (len > send-s) len = send-s;
1444  while (t < s+len && *t) t++;
1445  UNPACK_PUSH(infected_str_new(s, t-s, str));
1446  if (t < send) t++;
1447  s = star ? t : s+len;
1448  }
1449  break;
1450 
1451  case 'a':
1452  if (len > send - s) len = send - s;
1453  UNPACK_PUSH(infected_str_new(s, len, str));
1454  s += len;
1455  break;
1456 
1457  case 'b':
1458  {
1459  VALUE bitstr;
1460  char *t;
1461  int bits;
1462  long i;
1463 
1464  if (p[-1] == '*' || len > (send - s) * 8)
1465  len = (send - s) * 8;
1466  bits = 0;
1467  UNPACK_PUSH(bitstr = rb_str_new(0, len));
1468  t = RSTRING_PTR(bitstr);
1469  for (i=0; i<len; i++) {
1470  if (i & 7) bits >>= 1;
1471  else bits = *s++;
1472  *t++ = (bits & 1) ? '1' : '0';
1473  }
1474  }
1475  break;
1476 
1477  case 'B':
1478  {
1479  VALUE bitstr;
1480  char *t;
1481  int bits;
1482  long i;
1483 
1484  if (p[-1] == '*' || len > (send - s) * 8)
1485  len = (send - s) * 8;
1486  bits = 0;
1487  UNPACK_PUSH(bitstr = rb_str_new(0, len));
1488  t = RSTRING_PTR(bitstr);
1489  for (i=0; i<len; i++) {
1490  if (i & 7) bits <<= 1;
1491  else bits = *s++;
1492  *t++ = (bits & 128) ? '1' : '0';
1493  }
1494  }
1495  break;
1496 
1497  case 'h':
1498  {
1499  VALUE bitstr;
1500  char *t;
1501  int bits;
1502  long i;
1503 
1504  if (p[-1] == '*' || len > (send - s) * 2)
1505  len = (send - s) * 2;
1506  bits = 0;
1507  UNPACK_PUSH(bitstr = rb_str_new(0, len));
1508  t = RSTRING_PTR(bitstr);
1509  for (i=0; i<len; i++) {
1510  if (i & 1)
1511  bits >>= 4;
1512  else
1513  bits = *s++;
1514  *t++ = hexdigits[bits & 15];
1515  }
1516  }
1517  break;
1518 
1519  case 'H':
1520  {
1521  VALUE bitstr;
1522  char *t;
1523  int bits;
1524  long i;
1525 
1526  if (p[-1] == '*' || len > (send - s) * 2)
1527  len = (send - s) * 2;
1528  bits = 0;
1529  UNPACK_PUSH(bitstr = rb_str_new(0, len));
1530  t = RSTRING_PTR(bitstr);
1531  for (i=0; i<len; i++) {
1532  if (i & 1)
1533  bits <<= 4;
1534  else
1535  bits = *s++;
1536  *t++ = hexdigits[(bits >> 4) & 15];
1537  }
1538  }
1539  break;
1540 
1541  case 'c':
1542  PACK_LENGTH_ADJUST_SIZE(sizeof(char));
1543  while (len-- > 0) {
1544  int c = *s++;
1545  if (c > (char)127) c-=256;
1546  UNPACK_PUSH(INT2FIX(c));
1547  }
1548  PACK_ITEM_ADJUST();
1549  break;
1550 
1551  case 'C':
1552  PACK_LENGTH_ADJUST_SIZE(sizeof(unsigned char));
1553  while (len-- > 0) {
1554  unsigned char c = *s++;
1555  UNPACK_PUSH(INT2FIX(c));
1556  }
1557  PACK_ITEM_ADJUST();
1558  break;
1559 
1560  case 's':
1561  signed_p = 1;
1562  integer_size = NATINT_LEN(short, 2);
1563  bigendian_p = BIGENDIAN_P();
1564  goto unpack_integer;
1565 
1566  case 'S':
1567  signed_p = 0;
1568  integer_size = NATINT_LEN(short, 2);
1569  bigendian_p = BIGENDIAN_P();
1570  goto unpack_integer;
1571 
1572  case 'i':
1573  signed_p = 1;
1574  integer_size = (int)sizeof(int);
1575  bigendian_p = BIGENDIAN_P();
1576  goto unpack_integer;
1577 
1578  case 'I':
1579  signed_p = 0;
1580  integer_size = (int)sizeof(int);
1581  bigendian_p = BIGENDIAN_P();
1582  goto unpack_integer;
1583 
1584  case 'l':
1585  signed_p = 1;
1586  integer_size = NATINT_LEN(long, 4);
1587  bigendian_p = BIGENDIAN_P();
1588  goto unpack_integer;
1589 
1590  case 'L':
1591  signed_p = 0;
1592  integer_size = NATINT_LEN(long, 4);
1593  bigendian_p = BIGENDIAN_P();
1594  goto unpack_integer;
1595 
1596  case 'q':
1597  signed_p = 1;
1598  integer_size = 8;
1599  bigendian_p = BIGENDIAN_P();
1600  goto unpack_integer;
1601 
1602  case 'Q':
1603  signed_p = 0;
1604  integer_size = 8;
1605  bigendian_p = BIGENDIAN_P();
1606  goto unpack_integer;
1607 
1608  case 'n':
1609  signed_p = 0;
1610  integer_size = 2;
1611  bigendian_p = 1;
1612  goto unpack_integer;
1613 
1614  case 'N':
1615  signed_p = 0;
1616  integer_size = 4;
1617  bigendian_p = 1;
1618  goto unpack_integer;
1619 
1620  case 'v':
1621  signed_p = 0;
1622  integer_size = 2;
1623  bigendian_p = 0;
1624  goto unpack_integer;
1625 
1626  case 'V':
1627  signed_p = 0;
1628  integer_size = 4;
1629  bigendian_p = 0;
1630  goto unpack_integer;
1631 
1632  unpack_integer:
1633  if (explicit_endian) {
1634  bigendian_p = explicit_endian == '>';
1635  }
1636 
1637  switch (integer_size) {
1638 #if defined(HAVE_INT16_T) && !defined(FORCE_BIG_PACK)
1639  case SIZEOF_INT16_T:
1640  if (signed_p) {
1641  PACK_LENGTH_ADJUST_SIZE(sizeof(int16_t));
1642  while (len-- > 0) {
1643  union {
1644  int16_t i;
1645  char a[sizeof(int16_t)];
1646  } v;
1647  memcpy(v.a, s, sizeof(int16_t));
1648  if (bigendian_p != BIGENDIAN_P()) v.i = swap16(v.i);
1649  s += sizeof(int16_t);
1650  UNPACK_PUSH(INT2FIX(v.i));
1651  }
1652  PACK_ITEM_ADJUST();
1653  }
1654  else {
1655  PACK_LENGTH_ADJUST_SIZE(sizeof(uint16_t));
1656  while (len-- > 0) {
1657  union {
1658  uint16_t i;
1659  char a[sizeof(uint16_t)];
1660  } v;
1661  memcpy(v.a, s, sizeof(uint16_t));
1662  if (bigendian_p != BIGENDIAN_P()) v.i = swap16(v.i);
1663  s += sizeof(uint16_t);
1664  UNPACK_PUSH(INT2FIX(v.i));
1665  }
1666  PACK_ITEM_ADJUST();
1667  }
1668  break;
1669 #endif
1670 
1671 #if defined(HAVE_INT32_T) && !defined(FORCE_BIG_PACK)
1672  case SIZEOF_INT32_T:
1673  if (signed_p) {
1674  PACK_LENGTH_ADJUST_SIZE(sizeof(int32_t));
1675  while (len-- > 0) {
1676  union {
1677  int32_t i;
1678  char a[sizeof(int32_t)];
1679  } v;
1680  memcpy(v.a, s, sizeof(int32_t));
1681  if (bigendian_p != BIGENDIAN_P()) v.i = swap32(v.i);
1682  s += sizeof(int32_t);
1683  UNPACK_PUSH(INT2NUM(v.i));
1684  }
1685  PACK_ITEM_ADJUST();
1686  }
1687  else {
1689  while (len-- > 0) {
1690  union {
1691  uint32_t i;
1692  char a[sizeof(uint32_t)];
1693  } v;
1694  memcpy(v.a, s, sizeof(uint32_t));
1695  if (bigendian_p != BIGENDIAN_P()) v.i = swap32(v.i);
1696  s += sizeof(uint32_t);
1697  UNPACK_PUSH(UINT2NUM(v.i));
1698  }
1699  PACK_ITEM_ADJUST();
1700  }
1701  break;
1702 #endif
1703 
1704 #if defined(HAVE_INT64_T) && !defined(FORCE_BIG_PACK)
1705  case SIZEOF_INT64_T:
1706  if (signed_p) {
1707  PACK_LENGTH_ADJUST_SIZE(sizeof(int64_t));
1708  while (len-- > 0) {
1709  union {
1710  int64_t i;
1711  char a[sizeof(int64_t)];
1712  } v;
1713  memcpy(v.a, s, sizeof(int64_t));
1714  if (bigendian_p != BIGENDIAN_P()) v.i = swap64(v.i);
1715  s += sizeof(int64_t);
1716  UNPACK_PUSH(INT64toNUM(v.i));
1717  }
1718  PACK_ITEM_ADJUST();
1719  }
1720  else {
1722  while (len-- > 0) {
1723  union {
1724  uint64_t i;
1725  char a[sizeof(uint64_t)];
1726  } v;
1727  memcpy(v.a, s, sizeof(uint64_t));
1728  if (bigendian_p != BIGENDIAN_P()) v.i = swap64(v.i);
1729  s += sizeof(uint64_t);
1730  UNPACK_PUSH(UINT64toNUM(v.i));
1731  }
1732  PACK_ITEM_ADJUST();
1733  }
1734  break;
1735 #endif
1736 
1737  default:
1738  if (integer_size > MAX_INTEGER_PACK_SIZE)
1739  rb_bug("unexpected intger size for pack: %d", integer_size);
1740  PACK_LENGTH_ADJUST_SIZE(integer_size);
1741  while (len-- > 0) {
1742  union {
1743  unsigned long i[(MAX_INTEGER_PACK_SIZE+SIZEOF_LONG)/SIZEOF_LONG];
1744  char a[(MAX_INTEGER_PACK_SIZE+SIZEOF_LONG)/SIZEOF_LONG*SIZEOF_LONG];
1745  } v;
1746  int num_longs = (integer_size+SIZEOF_LONG)/SIZEOF_LONG;
1747  int i;
1748 
1749  if (signed_p && (signed char)s[bigendian_p ? 0 : (integer_size-1)] < 0)
1750  memset(v.a, 0xff, sizeof(long)*num_longs);
1751  else
1752  memset(v.a, 0, sizeof(long)*num_longs);
1753  if (bigendian_p)
1754  memcpy(v.a + sizeof(long)*num_longs - integer_size, s, integer_size);
1755  else
1756  memcpy(v.a, s, integer_size);
1757  if (bigendian_p) {
1758  for (i = 0; i < num_longs/2; i++) {
1759  unsigned long t = v.i[i];
1760  v.i[i] = v.i[num_longs-1-i];
1761  v.i[num_longs-1-i] = t;
1762  }
1763  }
1764  if (bigendian_p != BIGENDIAN_P()) {
1765  for (i = 0; i < num_longs; i++)
1766  v.i[i] = swapl(v.i[i]);
1767  }
1768  s += integer_size;
1769  UNPACK_PUSH(rb_big_unpack(v.i, num_longs));
1770  }
1771  PACK_ITEM_ADJUST();
1772  break;
1773  }
1774  break;
1775 
1776  case 'f':
1777  case 'F':
1778  PACK_LENGTH_ADJUST_SIZE(sizeof(float));
1779  while (len-- > 0) {
1780  float tmp;
1781  memcpy(&tmp, s, sizeof(float));
1782  s += sizeof(float);
1783  UNPACK_PUSH(DBL2NUM((double)tmp));
1784  }
1785  PACK_ITEM_ADJUST();
1786  break;
1787 
1788  case 'e':
1789  PACK_LENGTH_ADJUST_SIZE(sizeof(float));
1790  while (len-- > 0) {
1791  float tmp;
1792  FLOAT_CONVWITH(ftmp);
1793 
1794  memcpy(&tmp, s, sizeof(float));
1795  s += sizeof(float);
1796  tmp = VTOHF(tmp,ftmp);
1797  UNPACK_PUSH(DBL2NUM((double)tmp));
1798  }
1799  PACK_ITEM_ADJUST();
1800  break;
1801 
1802  case 'E':
1803  PACK_LENGTH_ADJUST_SIZE(sizeof(double));
1804  while (len-- > 0) {
1805  double tmp;
1806  DOUBLE_CONVWITH(dtmp);
1807 
1808  memcpy(&tmp, s, sizeof(double));
1809  s += sizeof(double);
1810  tmp = VTOHD(tmp,dtmp);
1811  UNPACK_PUSH(DBL2NUM(tmp));
1812  }
1813  PACK_ITEM_ADJUST();
1814  break;
1815 
1816  case 'D':
1817  case 'd':
1818  PACK_LENGTH_ADJUST_SIZE(sizeof(double));
1819  while (len-- > 0) {
1820  double tmp;
1821  memcpy(&tmp, s, sizeof(double));
1822  s += sizeof(double);
1823  UNPACK_PUSH(DBL2NUM(tmp));
1824  }
1825  PACK_ITEM_ADJUST();
1826  break;
1827 
1828  case 'g':
1829  PACK_LENGTH_ADJUST_SIZE(sizeof(float));
1830  while (len-- > 0) {
1831  float tmp;
1832  FLOAT_CONVWITH(ftmp);
1833 
1834  memcpy(&tmp, s, sizeof(float));
1835  s += sizeof(float);
1836  tmp = NTOHF(tmp,ftmp);
1837  UNPACK_PUSH(DBL2NUM((double)tmp));
1838  }
1839  PACK_ITEM_ADJUST();
1840  break;
1841 
1842  case 'G':
1843  PACK_LENGTH_ADJUST_SIZE(sizeof(double));
1844  while (len-- > 0) {
1845  double tmp;
1846  DOUBLE_CONVWITH(dtmp);
1847 
1848  memcpy(&tmp, s, sizeof(double));
1849  s += sizeof(double);
1850  tmp = NTOHD(tmp,dtmp);
1851  UNPACK_PUSH(DBL2NUM(tmp));
1852  }
1853  PACK_ITEM_ADJUST();
1854  break;
1855 
1856  case 'U':
1857  if (len > send - s) len = send - s;
1858  while (len > 0 && s < send) {
1859  long alen = send - s;
1860  unsigned long l;
1861 
1862  l = utf8_to_uv(s, &alen);
1863  s += alen; len--;
1864  UNPACK_PUSH(ULONG2NUM(l));
1865  }
1866  break;
1867 
1868  case 'u':
1869  {
1870  VALUE buf = infected_str_new(0, (send - s)*3/4, str);
1871  char *ptr = RSTRING_PTR(buf);
1872  long total = 0;
1873 
1874  while (s < send && *s > ' ' && *s < 'a') {
1875  long a,b,c,d;
1876  char hunk[4];
1877 
1878  hunk[3] = '\0';
1879  len = (*s++ - ' ') & 077;
1880  total += len;
1881  if (total > RSTRING_LEN(buf)) {
1882  len -= total - RSTRING_LEN(buf);
1883  total = RSTRING_LEN(buf);
1884  }
1885 
1886  while (len > 0) {
1887  long mlen = len > 3 ? 3 : len;
1888 
1889  if (s < send && *s >= ' ')
1890  a = (*s++ - ' ') & 077;
1891  else
1892  a = 0;
1893  if (s < send && *s >= ' ')
1894  b = (*s++ - ' ') & 077;
1895  else
1896  b = 0;
1897  if (s < send && *s >= ' ')
1898  c = (*s++ - ' ') & 077;
1899  else
1900  c = 0;
1901  if (s < send && *s >= ' ')
1902  d = (*s++ - ' ') & 077;
1903  else
1904  d = 0;
1905  hunk[0] = (char)(a << 2 | b >> 4);
1906  hunk[1] = (char)(b << 4 | c >> 2);
1907  hunk[2] = (char)(c << 6 | d);
1908  memcpy(ptr, hunk, mlen);
1909  ptr += mlen;
1910  len -= mlen;
1911  }
1912  if (*s == '\r') s++;
1913  if (*s == '\n') s++;
1914  else if (s < send && (s+1 == send || s[1] == '\n'))
1915  s += 2; /* possible checksum byte */
1916  }
1917 
1918  rb_str_set_len(buf, total);
1919  UNPACK_PUSH(buf);
1920  }
1921  break;
1922 
1923  case 'm':
1924  {
1925  VALUE buf = infected_str_new(0, (send - s)*3/4, str);
1926  char *ptr = RSTRING_PTR(buf);
1927  int a = -1,b = -1,c = 0,d = 0;
1928  static signed char b64_xtable[256];
1929 
1930  if (b64_xtable['/'] <= 0) {
1931  int i;
1932 
1933  for (i = 0; i < 256; i++) {
1934  b64_xtable[i] = -1;
1935  }
1936  for (i = 0; i < 64; i++) {
1937  b64_xtable[(unsigned char)b64_table[i]] = i;
1938  }
1939  }
1940  if (len == 0) {
1941  while (s < send) {
1942  a = b = c = d = -1;
1943  a = b64_xtable[(unsigned char)*s++];
1944  if (s >= send || a == -1) rb_raise(rb_eArgError, "invalid base64");
1945  b = b64_xtable[(unsigned char)*s++];
1946  if (s >= send || b == -1) rb_raise(rb_eArgError, "invalid base64");
1947  if (*s == '=') {
1948  if (s + 2 == send && *(s + 1) == '=') break;
1949  rb_raise(rb_eArgError, "invalid base64");
1950  }
1951  c = b64_xtable[(unsigned char)*s++];
1952  if (s >= send || c == -1) rb_raise(rb_eArgError, "invalid base64");
1953  if (s + 1 == send && *s == '=') break;
1954  d = b64_xtable[(unsigned char)*s++];
1955  if (d == -1) rb_raise(rb_eArgError, "invalid base64");
1956  *ptr++ = a << 2 | b >> 4;
1957  *ptr++ = b << 4 | c >> 2;
1958  *ptr++ = c << 6 | d;
1959  }
1960  if (c == -1) {
1961  *ptr++ = a << 2 | b >> 4;
1962  if (b & 0xf) rb_raise(rb_eArgError, "invalid base64");
1963  }
1964  else if (d == -1) {
1965  *ptr++ = a << 2 | b >> 4;
1966  *ptr++ = b << 4 | c >> 2;
1967  if (c & 0x3) rb_raise(rb_eArgError, "invalid base64");
1968  }
1969  }
1970  else {
1971  while (s < send) {
1972  a = b = c = d = -1;
1973  while ((a = b64_xtable[(unsigned char)*s]) == -1 && s < send) {s++;}
1974  if (s >= send) break;
1975  s++;
1976  while ((b = b64_xtable[(unsigned char)*s]) == -1 && s < send) {s++;}
1977  if (s >= send) break;
1978  s++;
1979  while ((c = b64_xtable[(unsigned char)*s]) == -1 && s < send) {if (*s == '=') break; s++;}
1980  if (*s == '=' || s >= send) break;
1981  s++;
1982  while ((d = b64_xtable[(unsigned char)*s]) == -1 && s < send) {if (*s == '=') break; s++;}
1983  if (*s == '=' || s >= send) break;
1984  s++;
1985  *ptr++ = a << 2 | b >> 4;
1986  *ptr++ = b << 4 | c >> 2;
1987  *ptr++ = c << 6 | d;
1988  }
1989  if (a != -1 && b != -1) {
1990  if (c == -1 && *s == '=')
1991  *ptr++ = a << 2 | b >> 4;
1992  else if (c != -1 && *s == '=') {
1993  *ptr++ = a << 2 | b >> 4;
1994  *ptr++ = b << 4 | c >> 2;
1995  }
1996  }
1997  }
1998  rb_str_set_len(buf, ptr - RSTRING_PTR(buf));
1999  UNPACK_PUSH(buf);
2000  }
2001  break;
2002 
2003  case 'M':
2004  {
2005  VALUE buf = infected_str_new(0, send - s, str);
2006  char *ptr = RSTRING_PTR(buf), *ss = s;
2007  int c1, c2;
2008 
2009  while (s < send) {
2010  if (*s == '=') {
2011  if (++s == send) break;
2012  if (s+1 < send && *s == '\r' && *(s+1) == '\n')
2013  s++;
2014  if (*s != '\n') {
2015  if ((c1 = hex2num(*s)) == -1) break;
2016  if (++s == send) break;
2017  if ((c2 = hex2num(*s)) == -1) break;
2018  *ptr++ = c1 << 4 | c2;
2019  }
2020  }
2021  else {
2022  *ptr++ = *s;
2023  }
2024  s++;
2025  ss = s;
2026  }
2027  rb_str_set_len(buf, ptr - RSTRING_PTR(buf));
2028  rb_str_buf_cat(buf, ss, send-ss);
2030  UNPACK_PUSH(buf);
2031  }
2032  break;
2033 
2034  case '@':
2035  if (len > RSTRING_LEN(str))
2036  rb_raise(rb_eArgError, "@ outside of string");
2037  s = RSTRING_PTR(str) + len;
2038  break;
2039 
2040  case 'X':
2041  if (len > s - RSTRING_PTR(str))
2042  rb_raise(rb_eArgError, "X outside of string");
2043  s -= len;
2044  break;
2045 
2046  case 'x':
2047  if (len > send - s)
2048  rb_raise(rb_eArgError, "x outside of string");
2049  s += len;
2050  break;
2051 
2052  case 'P':
2053  if (sizeof(char *) <= (size_t)(send - s)) {
2054  VALUE tmp = Qnil;
2055  char *t;
2056 
2057  memcpy(&t, s, sizeof(char *));
2058  s += sizeof(char *);
2059 
2060  if (t) {
2061  VALUE a, *p, *pend;
2062 
2063  if (!(a = rb_str_associated(str))) {
2064  rb_raise(rb_eArgError, "no associated pointer");
2065  }
2066  p = RARRAY_PTR(a);
2067  pend = p + RARRAY_LEN(a);
2068  while (p < pend) {
2069  if (TYPE(*p) == T_STRING && RSTRING_PTR(*p) == t) {
2070  if (len < RSTRING_LEN(*p)) {
2071  tmp = rb_tainted_str_new(t, len);
2072  rb_str_associate(tmp, a);
2073  }
2074  else {
2075  tmp = *p;
2076  }
2077  break;
2078  }
2079  p++;
2080  }
2081  if (p == pend) {
2082  rb_raise(rb_eArgError, "non associated pointer");
2083  }
2084  }
2085  UNPACK_PUSH(tmp);
2086  }
2087  break;
2088 
2089  case 'p':
2090  if (len > (long)((send - s) / sizeof(char *)))
2091  len = (send - s) / sizeof(char *);
2092  while (len-- > 0) {
2093  if ((size_t)(send - s) < sizeof(char *))
2094  break;
2095  else {
2096  VALUE tmp = Qnil;
2097  char *t;
2098 
2099  memcpy(&t, s, sizeof(char *));
2100  s += sizeof(char *);
2101 
2102  if (t) {
2103  VALUE a, *p, *pend;
2104 
2105  if (!(a = rb_str_associated(str))) {
2106  rb_raise(rb_eArgError, "no associated pointer");
2107  }
2108  p = RARRAY_PTR(a);
2109  pend = p + RARRAY_LEN(a);
2110  while (p < pend) {
2111  if (TYPE(*p) == T_STRING && RSTRING_PTR(*p) == t) {
2112  tmp = *p;
2113  break;
2114  }
2115  p++;
2116  }
2117  if (p == pend) {
2118  rb_raise(rb_eArgError, "non associated pointer");
2119  }
2120  }
2121  UNPACK_PUSH(tmp);
2122  }
2123  }
2124  break;
2125 
2126  case 'w':
2127  {
2128  unsigned long ul = 0;
2129  unsigned long ulmask = 0xfeUL << ((sizeof(unsigned long) - 1) * 8);
2130 
2131  while (len > 0 && s < send) {
2132  ul <<= 7;
2133  ul |= (*s & 0x7f);
2134  if (!(*s++ & 0x80)) {
2135  UNPACK_PUSH(ULONG2NUM(ul));
2136  len--;
2137  ul = 0;
2138  }
2139  else if (ul & ulmask) {
2140  VALUE big = rb_uint2big(ul);
2141  VALUE big128 = rb_uint2big(128);
2142  while (s < send) {
2143  big = rb_big_mul(big, big128);
2144  big = rb_big_plus(big, rb_uint2big(*s & 0x7f));
2145  if (!(*s++ & 0x80)) {
2146  UNPACK_PUSH(big);
2147  len--;
2148  ul = 0;
2149  break;
2150  }
2151  }
2152  }
2153  }
2154  }
2155  break;
2156 
2157  default:
2158  break;
2159  }
2160  }
2161 
2162  return ary;
2163 }
2164 
2165 #define BYTEWIDTH 8
2166 
2167 int
2168 rb_uv_to_utf8(char buf[6], unsigned long uv)
2169 {
2170  if (uv <= 0x7f) {
2171  buf[0] = (char)uv;
2172  return 1;
2173  }
2174  if (uv <= 0x7ff) {
2175  buf[0] = (char)((uv>>6)&0xff)|0xc0;
2176  buf[1] = (char)(uv&0x3f)|0x80;
2177  return 2;
2178  }
2179  if (uv <= 0xffff) {
2180  buf[0] = (char)((uv>>12)&0xff)|0xe0;
2181  buf[1] = (char)((uv>>6)&0x3f)|0x80;
2182  buf[2] = (char)(uv&0x3f)|0x80;
2183  return 3;
2184  }
2185  if (uv <= 0x1fffff) {
2186  buf[0] = (char)((uv>>18)&0xff)|0xf0;
2187  buf[1] = (char)((uv>>12)&0x3f)|0x80;
2188  buf[2] = (char)((uv>>6)&0x3f)|0x80;
2189  buf[3] = (char)(uv&0x3f)|0x80;
2190  return 4;
2191  }
2192  if (uv <= 0x3ffffff) {
2193  buf[0] = (char)((uv>>24)&0xff)|0xf8;
2194  buf[1] = (char)((uv>>18)&0x3f)|0x80;
2195  buf[2] = (char)((uv>>12)&0x3f)|0x80;
2196  buf[3] = (char)((uv>>6)&0x3f)|0x80;
2197  buf[4] = (char)(uv&0x3f)|0x80;
2198  return 5;
2199  }
2200  if (uv <= 0x7fffffff) {
2201  buf[0] = (char)((uv>>30)&0xff)|0xfc;
2202  buf[1] = (char)((uv>>24)&0x3f)|0x80;
2203  buf[2] = (char)((uv>>18)&0x3f)|0x80;
2204  buf[3] = (char)((uv>>12)&0x3f)|0x80;
2205  buf[4] = (char)((uv>>6)&0x3f)|0x80;
2206  buf[5] = (char)(uv&0x3f)|0x80;
2207  return 6;
2208  }
2209  rb_raise(rb_eRangeError, "pack(U): value out of range");
2210 }
2211 
2212 static const unsigned long utf8_limits[] = {
2213  0x0, /* 1 */
2214  0x80, /* 2 */
2215  0x800, /* 3 */
2216  0x10000, /* 4 */
2217  0x200000, /* 5 */
2218  0x4000000, /* 6 */
2219  0x80000000, /* 7 */
2220 };
2221 
2222 static unsigned long
2223 utf8_to_uv(const char *p, long *lenp)
2224 {
2225  int c = *p++ & 0xff;
2226  unsigned long uv = c;
2227  long n;
2228 
2229  if (!(uv & 0x80)) {
2230  *lenp = 1;
2231  return uv;
2232  }
2233  if (!(uv & 0x40)) {
2234  *lenp = 1;
2235  rb_raise(rb_eArgError, "malformed UTF-8 character");
2236  }
2237 
2238  if (!(uv & 0x20)) { n = 2; uv &= 0x1f; }
2239  else if (!(uv & 0x10)) { n = 3; uv &= 0x0f; }
2240  else if (!(uv & 0x08)) { n = 4; uv &= 0x07; }
2241  else if (!(uv & 0x04)) { n = 5; uv &= 0x03; }
2242  else if (!(uv & 0x02)) { n = 6; uv &= 0x01; }
2243  else {
2244  *lenp = 1;
2245  rb_raise(rb_eArgError, "malformed UTF-8 character");
2246  }
2247  if (n > *lenp) {
2248  rb_raise(rb_eArgError, "malformed UTF-8 character (expected %ld bytes, given %ld bytes)",
2249  n, *lenp);
2250  }
2251  *lenp = n--;
2252  if (n != 0) {
2253  while (n--) {
2254  c = *p++ & 0xff;
2255  if ((c & 0xc0) != 0x80) {
2256  *lenp -= n + 1;
2257  rb_raise(rb_eArgError, "malformed UTF-8 character");
2258  }
2259  else {
2260  c &= 0x3f;
2261  uv = uv << 6 | c;
2262  }
2263  }
2264  }
2265  n = *lenp - 1;
2266  if (uv < utf8_limits[n]) {
2267  rb_raise(rb_eArgError, "redundant UTF-8 sequence");
2268  }
2269  return uv;
2270 }
2271 
2272 void
2274 {
2275  rb_define_method(rb_cArray, "pack", pack_pack, 1);
2276  rb_define_method(rb_cString, "unpack", pack_unpack, 1);
2277 }
2278