1 /**********************************************************************
6 created at: Thu Feb 10 15:17:05 JST 1994
8 Copyright (C) 1993-2007 Yukihiro Matsumoto
10 **********************************************************************/
12 #include "ruby/internal/config.h"
17 #include <sys/types.h>
20 #include "internal/array.h"
21 #include "internal/bits.h"
22 #include "internal/string.h"
23 #include "internal/symbol.h"
24 #include "internal/variable.h"
25 #include "ruby/util.h"
30 * It is intentional that the condition for natstr is HAVE_TRUE_LONG_LONG
31 * instead of HAVE_LONG_LONG or LONG_LONG.
32 * This means q! and Q! means always the standard long long type and
33 * causes ArgumentError for platforms which has no long long type,
34 * even if the platform has an implementation specific 64bit type.
35 * This behavior is consistent with the document of pack/unpack.
37 #ifdef HAVE_TRUE_LONG_LONG
38 static const char natstr
[] = "sSiIlLqQjJ";
39 # define endstr natstr
41 static const char natstr
[] = "sSiIlLjJ";
42 static const char endstr
[] = "sSiIlLqQjJ";
45 #ifdef HAVE_TRUE_LONG_LONG
46 /* It is intentional to use long long instead of LONG_LONG. */
47 # define NATINT_LEN_Q NATINT_LEN(long long, 8)
49 # define NATINT_LEN_Q 8
52 #if SIZEOF_SHORT != 2 || SIZEOF_LONG != 4 || (defined(HAVE_TRUE_LONG_LONG) && SIZEOF_LONG_LONG != 8)
57 /* for universal binary of NEXTSTEP and MacOS X */
58 /* useless since autoconf 2.63? */
63 static int endian_value
;
66 if (init
) return endian_value
;
69 return endian_value
= p
[0]?0:1;
71 # define BIGENDIAN_P() (is_bigendian())
72 #elif defined(WORDS_BIGENDIAN)
73 # define BIGENDIAN_P() 1
75 # define BIGENDIAN_P() 0
79 # define NATINT_LEN(type,len) (natint?(int)sizeof(type):(int)(len))
81 # define NATINT_LEN(type,len) ((int)sizeof(type))
94 #define swapf(x) swap32(x)
95 #define swapd(x) swap64(x)
97 #define rb_ntohf(x) (BIGENDIAN_P()?(x):swapf(x))
98 #define rb_ntohd(x) (BIGENDIAN_P()?(x):swapd(x))
99 #define rb_htonf(x) (BIGENDIAN_P()?(x):swapf(x))
100 #define rb_htond(x) (BIGENDIAN_P()?(x):swapd(x))
101 #define rb_htovf(x) (BIGENDIAN_P()?swapf(x):(x))
102 #define rb_htovd(x) (BIGENDIAN_P()?swapd(x):(x))
103 #define rb_vtohf(x) (BIGENDIAN_P()?swapf(x):(x))
104 #define rb_vtohd(x) (BIGENDIAN_P()?swapd(x):(x))
106 #define FLOAT_CONVWITH(x) FLOAT_SWAPPER x;
107 #define HTONF(x) ((x).u = rb_htonf((x).u))
108 #define HTOVF(x) ((x).u = rb_htovf((x).u))
109 #define NTOHF(x) ((x).u = rb_ntohf((x).u))
110 #define VTOHF(x) ((x).u = rb_vtohf((x).u))
112 #define DOUBLE_CONVWITH(x) DOUBLE_SWAPPER x;
113 #define HTOND(x) ((x).u = rb_htond((x).u))
114 #define HTOVD(x) ((x).u = rb_htovd((x).u))
115 #define NTOHD(x) ((x).u = rb_ntohd((x).u))
116 #define VTOHD(x) ((x).u = rb_vtohd((x).u))
118 #define MAX_INTEGER_PACK_SIZE 8
120 static const char toofew
[] = "too few arguments";
122 static void encodes(VALUE
,const char*,long,int,int);
123 static void qpencode(VALUE
,VALUE
,long);
125 static unsigned long utf8_to_uv(const char*,long*);
127 static ID id_associated
;
130 str_associate(VALUE str
, VALUE add
)
132 /* assert(NIL_P(rb_attr_get(str, id_associated))); */
133 rb_ivar_set(str
, id_associated
, add
);
137 str_associated(VALUE str
)
139 VALUE associates
= rb_ivar_lookup(str
, id_associated
, Qfalse
);
141 rb_raise(rb_eArgError
, "no associated pointer");
146 associated_pointer(VALUE associates
, const char *t
)
148 const VALUE
*p
= RARRAY_CONST_PTR(associates
);
149 const VALUE
*pend
= p
+ RARRAY_LEN(associates
);
150 for (; p
< pend
; p
++) {
152 if (RB_TYPE_P(tmp
, T_STRING
) && RSTRING_PTR(tmp
) == t
) return tmp
;
154 rb_raise(rb_eArgError
, "non associated pointer");
155 UNREACHABLE_RETURN(Qnil
);
158 RBIMPL_ATTR_NORETURN()
160 unknown_directive(const char *mode
, char type
, VALUE fmt
)
169 snprintf(unknown
, sizeof(unknown
), "\\x%.2x", type
& 0xff);
171 fmt
= rb_str_quote_unprintable(fmt
);
172 rb_raise(rb_eArgError
, "unknown %s directive '%s' in '%"PRIsVALUE
"'",
177 VALUE_to_float(VALUE obj
)
179 VALUE v
= rb_to_float(obj
);
180 double d
= RFLOAT_VALUE(v
);
185 else if (d
< -FLT_MAX
) {
188 else if (d
<= FLT_MAX
) {
197 pack_pack(rb_execution_context_t
*ec
, VALUE ary
, VALUE fmt
, VALUE buffer
)
199 static const char nul10
[] = "\0\0\0\0\0\0\0\0\0\0";
200 static const char spc10
[] = " ";
201 const char *p
, *pend
;
202 VALUE res
, from
, associates
= 0;
206 int enc_info
= 1; /* 0 - BINARY, 1 - US-ASCII, 2 - UTF-8 */
208 int natint
; /* native integer */
210 int integer_size
, bigendian_p
;
213 rb_must_asciicompat(fmt
);
214 p
= RSTRING_PTR(fmt
);
215 pend
= p
+ RSTRING_LEN(fmt
);
218 res
= rb_str_buf_new(0);
221 if (!RB_TYPE_P(buffer
, T_STRING
))
222 rb_raise(rb_eTypeError
, "buffer must be String, not %s", rb_obj_classname(buffer
));
223 rb_str_modify(buffer
);
229 #define TOO_FEW (rb_raise(rb_eArgError, toofew), 0)
230 #define MORE_ITEM (idx < RARRAY_LEN(ary))
231 #define THISFROM (MORE_ITEM ? RARRAY_AREF(ary, idx) : TOO_FEW)
232 #define NEXTFROM (MORE_ITEM ? RARRAY_AREF(ary, idx++) : TOO_FEW)
235 int explicit_endian
= 0;
236 if (RSTRING_PTR(fmt
) + RSTRING_LEN(fmt
) != pend
) {
237 rb_raise(rb_eRuntimeError
, "format string modified");
239 type
= *p
++; /* get data type */
244 if (ISSPACE(type
)) continue;
246 while ((p
< pend
) && (*p
!= '\n')) {
257 if (strchr(natstr
, type
)) {
264 rb_raise(rb_eArgError
, "'%c' allowed only after types %s", *p
, natstr
);
270 if (!strchr(endstr
, type
)) {
271 rb_raise(rb_eArgError
, "'%c' allowed only after types %s", *p
, endstr
);
273 if (explicit_endian
) {
274 rb_raise(rb_eRangeError
, "Can't use both '<' and '>'");
276 explicit_endian
= *p
++;
281 if (*p
== '*') { /* set data length */
282 len
= strchr("@Xxu", type
) ? 0
283 : strchr("PMm", type
) ? 1
284 : RARRAY_LEN(ary
) - idx
;
287 else if (ISDIGIT(*p
)) {
289 len
= STRTOUL(p
, (char**)&p
, 10);
291 rb_raise(rb_eRangeError
, "pack length too big");
300 /* if encoding is US-ASCII, upgrade to UTF-8 */
301 if (enc_info
== 1) enc_info
= 2;
303 case 'm': case 'M': case 'u':
304 /* keep US-ASCII (do nothing) */
307 /* fall back to BINARY */
312 case 'A': case 'a': case 'Z':
322 ptr
= RSTRING_PTR(from
);
323 plen
= RSTRING_LEN(from
);
330 case 'a': /* arbitrary binary string (null padded) */
331 case 'A': /* arbitrary binary string (ASCII space padded) */
332 case 'Z': /* null terminated string */
334 rb_str_buf_cat(res
, ptr
, len
);
335 if (p
[-1] == '*' && type
== 'Z')
336 rb_str_buf_cat(res
, nul10
, 1);
339 rb_str_buf_cat(res
, ptr
, plen
);
342 rb_str_buf_cat(res
, (type
== 'A')?spc10
:nul10
, 10);
345 rb_str_buf_cat(res
, (type
== 'A')?spc10
:nul10
, len
);
349 #define castchar(from) (char)((from) & 0xff)
351 case 'b': /* bit string (ascending) */
357 j
= (len
- plen
+ 1)/2;
360 for (i
=0; i
++ < len
; ptr
++) {
366 char c
= castchar(byte
);
367 rb_str_buf_cat(res
, &c
, 1);
373 byte
>>= 7 - (len
& 7);
375 rb_str_buf_cat(res
, &c
, 1);
382 case 'B': /* bit string (descending) */
388 j
= (len
- plen
+ 1)/2;
391 for (i
=0; i
++ < len
; ptr
++) {
396 char c
= castchar(byte
);
397 rb_str_buf_cat(res
, &c
, 1);
403 byte
<<= 7 - (len
& 7);
405 rb_str_buf_cat(res
, &c
, 1);
412 case 'h': /* hex string (low nibble first) */
418 j
= (len
+ 1) / 2 - (plen
+ 1) / 2;
421 for (i
=0; i
++ < len
; ptr
++) {
423 byte
|= (((*ptr
& 15) + 9) & 15) << 4;
425 byte
|= (*ptr
& 15) << 4;
429 char c
= castchar(byte
);
430 rb_str_buf_cat(res
, &c
, 1);
435 char c
= castchar(byte
);
436 rb_str_buf_cat(res
, &c
, 1);
443 case 'H': /* hex string (high nibble first) */
449 j
= (len
+ 1) / 2 - (plen
+ 1) / 2;
452 for (i
=0; i
++ < len
; ptr
++) {
454 byte
|= ((*ptr
& 15) + 9) & 15;
460 char c
= castchar(byte
);
461 rb_str_buf_cat(res
, &c
, 1);
466 char c
= castchar(byte
);
467 rb_str_buf_cat(res
, &c
, 1);
476 case 'c': /* signed char */
477 case 'C': /* unsigned char */
479 bigendian_p
= BIGENDIAN_P(); /* not effective */
482 case 's': /* s for int16_t, s! for signed short */
483 case 'S': /* S for uint16_t, S! for unsigned short */
484 integer_size
= NATINT_LEN(short, 2);
485 bigendian_p
= BIGENDIAN_P();
488 case 'i': /* i and i! for signed int */
489 case 'I': /* I and I! for unsigned int */
490 integer_size
= (int)sizeof(int);
491 bigendian_p
= BIGENDIAN_P();
494 case 'l': /* l for int32_t, l! for signed long */
495 case 'L': /* L for uint32_t, L! for unsigned long */
496 integer_size
= NATINT_LEN(long, 4);
497 bigendian_p
= BIGENDIAN_P();
500 case 'q': /* q for int64_t, q! for signed long long */
501 case 'Q': /* Q for uint64_t, Q! for unsigned long long */
502 integer_size
= NATINT_LEN_Q
;
503 bigendian_p
= BIGENDIAN_P();
506 case 'j': /* j for intptr_t */
507 integer_size
= sizeof(intptr_t);
508 bigendian_p
= BIGENDIAN_P();
511 case 'J': /* J for uintptr_t */
512 integer_size
= sizeof(uintptr_t);
513 bigendian_p
= BIGENDIAN_P();
516 case 'n': /* 16 bit (2 bytes) integer (network byte-order) */
521 case 'N': /* 32 bit (4 bytes) integer (network byte-order) */
526 case 'v': /* 16 bit (2 bytes) integer (VAX byte-order) */
531 case 'V': /* 32 bit (4 bytes) integer (VAX byte-order) */
537 if (explicit_endian
) {
538 bigendian_p
= explicit_endian
== '>';
540 if (integer_size
> MAX_INTEGER_PACK_SIZE
)
541 rb_bug("unexpected integer size for pack: %d", integer_size
);
543 char intbuf
[MAX_INTEGER_PACK_SIZE
];
546 rb_integer_pack(from
, intbuf
, integer_size
, 1, 0,
548 (bigendian_p
? INTEGER_PACK_BIG_ENDIAN
: INTEGER_PACK_LITTLE_ENDIAN
));
549 rb_str_buf_cat(res
, intbuf
, integer_size
);
553 case 'f': /* single precision float in native format */
554 case 'F': /* ditto */
559 f
= VALUE_to_float(from
);
560 rb_str_buf_cat(res
, (char*)&f
, sizeof(float));
564 case 'e': /* single precision float in VAX byte-order */
569 tmp
.f
= VALUE_to_float(from
);
571 rb_str_buf_cat(res
, tmp
.buf
, sizeof(float));
575 case 'E': /* double precision float in VAX byte-order */
577 DOUBLE_CONVWITH(tmp
);
579 tmp
.d
= RFLOAT_VALUE(rb_to_float(from
));
581 rb_str_buf_cat(res
, tmp
.buf
, sizeof(double));
585 case 'd': /* double precision float in native format */
586 case 'D': /* ditto */
591 d
= RFLOAT_VALUE(rb_to_float(from
));
592 rb_str_buf_cat(res
, (char*)&d
, sizeof(double));
596 case 'g': /* single precision float in network byte-order */
600 tmp
.f
= VALUE_to_float(from
);
602 rb_str_buf_cat(res
, tmp
.buf
, sizeof(float));
606 case 'G': /* double precision float in network byte-order */
608 DOUBLE_CONVWITH(tmp
);
611 tmp
.d
= RFLOAT_VALUE(rb_to_float(from
));
613 rb_str_buf_cat(res
, tmp
.buf
, sizeof(double));
617 case 'x': /* null byte */
620 rb_str_buf_cat(res
, nul10
, 10);
623 rb_str_buf_cat(res
, nul10
, len
);
626 case 'X': /* back up byte */
628 plen
= RSTRING_LEN(res
);
630 rb_raise(rb_eArgError
, "X outside of string");
631 rb_str_set_len(res
, plen
- len
);
634 case '@': /* null fill to absolute position */
635 len
-= RSTRING_LEN(res
);
636 if (len
> 0) goto grow
;
638 if (len
> 0) goto shrink
;
642 rb_raise(rb_eArgError
, "%% is not supported");
645 case 'U': /* Unicode character */
652 from
= rb_to_int(from
);
655 rb_raise(rb_eRangeError
, "pack(U): value out of range");
657 le
= rb_uv_to_utf8(buf
, l
);
658 rb_str_buf_cat(res
, (char*)buf
, le
);
662 case 'u': /* uuencoded string */
663 case 'm': /* base64 encoded string */
666 ptr
= RSTRING_PTR(from
);
667 plen
= RSTRING_LEN(from
);
669 if (len
== 0 && type
== 'm') {
670 encodes(res
, ptr
, plen
, type
, 0);
676 else if (len
> 63 && type
== 'u')
687 encodes(res
, ptr
, todo
, type
, 1);
693 case 'M': /* quoted-printable encoded string */
694 from
= rb_obj_as_string(NEXTFROM
);
697 qpencode(res
, from
, len
);
700 case 'P': /* pointer to packed byte string */
704 if (RSTRING_LEN(from
) < len
) {
705 rb_raise(rb_eArgError
, "too short buffer for P(%ld for %ld)",
706 RSTRING_LEN(from
), len
);
711 case 'p': /* pointer to string */
719 t
= StringValuePtr(from
);
722 associates
= rb_ary_new();
724 rb_ary_push(associates
, from
);
725 rb_str_buf_cat(res
, (char*)&t
, sizeof(char*));
729 case 'w': /* BER compressed integer */
731 VALUE buf
= rb_str_new(0, 0);
737 from
= rb_to_int(from
);
738 numbytes
= rb_absint_numwords(from
, 7, NULL
);
741 buf
= rb_str_new(NULL
, numbytes
);
743 sign
= rb_integer_pack(from
, RSTRING_PTR(buf
), RSTRING_LEN(buf
), 1, 1, INTEGER_PACK_BIG_ENDIAN
);
746 rb_raise(rb_eArgError
, "can't compress negative numbers");
748 rb_bug("buffer size problem?");
750 cp
= RSTRING_PTR(buf
);
751 while (1 < numbytes
) {
757 rb_str_buf_cat(res
, RSTRING_PTR(buf
), RSTRING_LEN(buf
));
762 unknown_directive("pack", type
, fmt
);
769 str_associate(res
, associates
);
773 ENCODING_CODERANGE_SET(res
, rb_usascii_encindex(), ENC_CODERANGE_7BIT
);
776 rb_enc_set_index(res
, rb_utf8_encindex());
779 /* do nothing, keep ASCII-8BIT */
785 static const char uu_table
[] =
786 "`!\"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_";
787 static const char b64_table
[] =
788 "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
791 encodes(VALUE str
, const char *s0
, long len
, int type
, int tail_lf
)
793 enum {buff_size
= 4096, encoded_unit
= 4, input_unit
= 3};
794 char buff
[buff_size
+ 1]; /* +1 for tail_lf */
796 const char *const trans
= type
== 'u' ? uu_table
: b64_table
;
798 const unsigned char *s
= (const unsigned char *)s0
;
801 buff
[i
++] = (char)len
+ ' ';
807 while (len
>= input_unit
) {
808 while (len
>= input_unit
&& buff_size
-i
>= encoded_unit
) {
809 buff
[i
++] = trans
[077 & (*s
>> 2)];
810 buff
[i
++] = trans
[077 & (((*s
<< 4) & 060) | ((s
[1] >> 4) & 017))];
811 buff
[i
++] = trans
[077 & (((s
[1] << 2) & 074) | ((s
[2] >> 6) & 03))];
812 buff
[i
++] = trans
[077 & s
[2]];
816 if (buff_size
-i
< encoded_unit
) {
817 rb_str_buf_cat(str
, buff
, i
);
823 buff
[i
++] = trans
[077 & (*s
>> 2)];
824 buff
[i
++] = trans
[077 & (((*s
<< 4) & 060) | ((s
[1] >> 4) & 017))];
825 buff
[i
++] = trans
[077 & (((s
[1] << 2) & 074) | (('\0' >> 6) & 03))];
829 buff
[i
++] = trans
[077 & (*s
>> 2)];
830 buff
[i
++] = trans
[077 & (((*s
<< 4) & 060) | (('\0' >> 4) & 017))];
834 if (tail_lf
) buff
[i
++] = '\n';
835 rb_str_buf_cat(str
, buff
, i
);
836 if ((size_t)i
> sizeof(buff
)) rb_bug("encodes() buffer overrun");
839 static const char hex_table
[] = "0123456789ABCDEF";
842 qpencode(VALUE str
, VALUE from
, long len
)
845 long i
= 0, n
= 0, prev
= EOF
;
846 unsigned char *s
= (unsigned char*)RSTRING_PTR(from
);
847 unsigned char *send
= s
+ RSTRING_LEN(from
);
851 (*s
< 32 && *s
!= '\n' && *s
!= '\t') ||
854 buff
[i
++] = hex_table
[*s
>> 4];
855 buff
[i
++] = hex_table
[*s
& 0x0f];
859 else if (*s
== '\n') {
860 if (prev
== ' ' || prev
== '\t') {
880 rb_str_buf_cat(str
, buff
, i
);
890 rb_str_buf_cat(str
, buff
, i
);
898 n
= ruby_digit36_to_number_table
[(unsigned char)c
];
904 #define PACK_LENGTH_ADJUST_SIZE(sz) do { \
906 if (len > (long)((send-s)/(sz))) { \
908 tmp_len = len-(send-s)/(sz); \
910 len = (send-s)/(sz); \
914 #define PACK_ITEM_ADJUST() do { \
915 if (tmp_len > 0 && mode == UNPACK_ARRAY) \
916 rb_ary_store(ary, RARRAY_LEN(ary)+tmp_len-1, Qnil); \
919 /* Workaround for Oracle Developer Studio (Oracle Solaris Studio)
920 * 12.4/12.5/12.6 C compiler optimization bug
921 * with "-xO4" optimization option.
923 #if defined(__SUNPRO_C) && 0x5130 <= __SUNPRO_C && __SUNPRO_C <= 0x5150
924 # define AVOID_CC_BUG volatile
926 # define AVOID_CC_BUG
936 pack_unpack_internal(VALUE str
, VALUE fmt
, enum unpack_mode mode
, long offset
)
938 #define hexdigits ruby_hexdigits
941 VALUE ary
, associates
= Qfalse
;
944 AVOID_CC_BUG
long tmp_len
;
947 int natint
; /* native integer */
949 int signed_p
, integer_size
, bigendian_p
;
950 #define UNPACK_PUSH(item) do {\
951 VALUE item_val = (item);\
952 if ((mode) == UNPACK_BLOCK) {\
955 else if ((mode) == UNPACK_ARRAY) {\
956 rb_ary_push(ary, item_val);\
958 else /* if ((mode) == UNPACK_1) { */ {\
965 rb_must_asciicompat(fmt
);
967 if (offset
< 0) rb_raise(rb_eArgError
, "offset can't be negative");
968 len
= RSTRING_LEN(str
);
969 if (offset
> len
) rb_raise(rb_eArgError
, "offset outside of string");
971 s
= RSTRING_PTR(str
);
975 p
= RSTRING_PTR(fmt
);
976 pend
= p
+ RSTRING_LEN(fmt
);
978 #define UNPACK_FETCH(var, type) (memcpy((var), s, sizeof(type)), s += sizeof(type))
980 ary
= mode
== UNPACK_ARRAY
? rb_ary_new() : Qnil
;
982 int explicit_endian
= 0;
988 if (ISSPACE(type
)) continue;
990 while ((p
< pend
) && (*p
!= '\n')) {
1003 if (strchr(natstr
, type
)) {
1010 rb_raise(rb_eArgError
, "'%c' allowed only after types %s", *p
, natstr
);
1016 if (!strchr(endstr
, type
)) {
1017 rb_raise(rb_eArgError
, "'%c' allowed only after types %s", *p
, endstr
);
1019 if (explicit_endian
) {
1020 rb_raise(rb_eRangeError
, "Can't use both '<' and '>'");
1022 explicit_endian
= *p
++;
1029 else if (*p
== '*') {
1034 else if (ISDIGIT(*p
)) {
1036 len
= STRTOUL(p
, (char**)&p
, 10);
1037 if (len
< 0 || errno
) {
1038 rb_raise(rb_eRangeError
, "pack length too big");
1042 len
= (type
!= '@');
1047 rb_raise(rb_eArgError
, "%% is not supported");
1051 if (len
> send
- s
) len
= send
- s
;
1054 char *t
= s
+ len
- 1;
1057 if (*t
!= ' ' && *t
!= '\0') break;
1060 UNPACK_PUSH(rb_str_new(s
, len
));
1069 if (len
> send
-s
) len
= send
-s
;
1070 while (t
< s
+len
&& *t
) t
++;
1071 UNPACK_PUSH(rb_str_new(s
, t
-s
));
1073 s
= star
? t
: s
+len
;
1078 if (len
> send
- s
) len
= send
- s
;
1079 UNPACK_PUSH(rb_str_new(s
, len
));
1090 if (p
[-1] == '*' || len
> (send
- s
) * 8)
1091 len
= (send
- s
) * 8;
1093 bitstr
= rb_usascii_str_new(0, len
);
1094 t
= RSTRING_PTR(bitstr
);
1095 for (i
=0; i
<len
; i
++) {
1096 if (i
& 7) bits
>>= 1;
1097 else bits
= (unsigned char)*s
++;
1098 *t
++ = (bits
& 1) ? '1' : '0';
1100 UNPACK_PUSH(bitstr
);
1111 if (p
[-1] == '*' || len
> (send
- s
) * 8)
1112 len
= (send
- s
) * 8;
1114 bitstr
= rb_usascii_str_new(0, len
);
1115 t
= RSTRING_PTR(bitstr
);
1116 for (i
=0; i
<len
; i
++) {
1117 if (i
& 7) bits
<<= 1;
1118 else bits
= (unsigned char)*s
++;
1119 *t
++ = (bits
& 128) ? '1' : '0';
1121 UNPACK_PUSH(bitstr
);
1132 if (p
[-1] == '*' || len
> (send
- s
) * 2)
1133 len
= (send
- s
) * 2;
1135 bitstr
= rb_usascii_str_new(0, len
);
1136 t
= RSTRING_PTR(bitstr
);
1137 for (i
=0; i
<len
; i
++) {
1141 bits
= (unsigned char)*s
++;
1142 *t
++ = hexdigits
[bits
& 15];
1144 UNPACK_PUSH(bitstr
);
1155 if (p
[-1] == '*' || len
> (send
- s
) * 2)
1156 len
= (send
- s
) * 2;
1158 bitstr
= rb_usascii_str_new(0, len
);
1159 t
= RSTRING_PTR(bitstr
);
1160 for (i
=0; i
<len
; i
++) {
1164 bits
= (unsigned char)*s
++;
1165 *t
++ = hexdigits
[(bits
>> 4) & 15];
1167 UNPACK_PUSH(bitstr
);
1174 bigendian_p
= BIGENDIAN_P(); /* not effective */
1175 goto unpack_integer
;
1180 bigendian_p
= BIGENDIAN_P(); /* not effective */
1181 goto unpack_integer
;
1185 integer_size
= NATINT_LEN(short, 2);
1186 bigendian_p
= BIGENDIAN_P();
1187 goto unpack_integer
;
1191 integer_size
= NATINT_LEN(short, 2);
1192 bigendian_p
= BIGENDIAN_P();
1193 goto unpack_integer
;
1197 integer_size
= (int)sizeof(int);
1198 bigendian_p
= BIGENDIAN_P();
1199 goto unpack_integer
;
1203 integer_size
= (int)sizeof(int);
1204 bigendian_p
= BIGENDIAN_P();
1205 goto unpack_integer
;
1209 integer_size
= NATINT_LEN(long, 4);
1210 bigendian_p
= BIGENDIAN_P();
1211 goto unpack_integer
;
1215 integer_size
= NATINT_LEN(long, 4);
1216 bigendian_p
= BIGENDIAN_P();
1217 goto unpack_integer
;
1221 integer_size
= NATINT_LEN_Q
;
1222 bigendian_p
= BIGENDIAN_P();
1223 goto unpack_integer
;
1227 integer_size
= NATINT_LEN_Q
;
1228 bigendian_p
= BIGENDIAN_P();
1229 goto unpack_integer
;
1233 integer_size
= sizeof(intptr_t);
1234 bigendian_p
= BIGENDIAN_P();
1235 goto unpack_integer
;
1239 integer_size
= sizeof(uintptr_t);
1240 bigendian_p
= BIGENDIAN_P();
1241 goto unpack_integer
;
1247 goto unpack_integer
;
1253 goto unpack_integer
;
1259 goto unpack_integer
;
1265 goto unpack_integer
;
1268 if (explicit_endian
) {
1269 bigendian_p
= explicit_endian
== '>';
1271 PACK_LENGTH_ADJUST_SIZE(integer_size
);
1273 int flags
= bigendian_p
? INTEGER_PACK_BIG_ENDIAN
: INTEGER_PACK_LITTLE_ENDIAN
;
1276 flags
|= INTEGER_PACK_2COMP
;
1277 val
= rb_integer_unpack(s
, integer_size
, 1, 0, flags
);
1286 PACK_LENGTH_ADJUST_SIZE(sizeof(float));
1289 UNPACK_FETCH(&tmp
, float);
1290 UNPACK_PUSH(DBL2NUM((double)tmp
));
1296 PACK_LENGTH_ADJUST_SIZE(sizeof(float));
1298 FLOAT_CONVWITH(tmp
);
1299 UNPACK_FETCH(tmp
.buf
, float);
1301 UNPACK_PUSH(DBL2NUM(tmp
.f
));
1307 PACK_LENGTH_ADJUST_SIZE(sizeof(double));
1309 DOUBLE_CONVWITH(tmp
);
1310 UNPACK_FETCH(tmp
.buf
, double);
1312 UNPACK_PUSH(DBL2NUM(tmp
.d
));
1319 PACK_LENGTH_ADJUST_SIZE(sizeof(double));
1322 UNPACK_FETCH(&tmp
, double);
1323 UNPACK_PUSH(DBL2NUM(tmp
));
1329 PACK_LENGTH_ADJUST_SIZE(sizeof(float));
1331 FLOAT_CONVWITH(tmp
);
1332 UNPACK_FETCH(tmp
.buf
, float);
1334 UNPACK_PUSH(DBL2NUM(tmp
.f
));
1340 PACK_LENGTH_ADJUST_SIZE(sizeof(double));
1342 DOUBLE_CONVWITH(tmp
);
1343 UNPACK_FETCH(tmp
.buf
, double);
1345 UNPACK_PUSH(DBL2NUM(tmp
.d
));
1351 if (len
> send
- s
) len
= send
- s
;
1352 while (len
> 0 && s
< send
) {
1353 long alen
= send
- s
;
1356 l
= utf8_to_uv(s
, &alen
);
1358 UNPACK_PUSH(ULONG2NUM(l
));
1364 VALUE buf
= rb_str_new(0, (send
- s
)*3/4);
1365 char *ptr
= RSTRING_PTR(buf
);
1368 while (s
< send
&& (unsigned char)*s
> ' ' && (unsigned char)*s
< 'a') {
1372 len
= ((unsigned char)*s
++ - ' ') & 077;
1375 if (total
> RSTRING_LEN(buf
)) {
1376 len
-= total
- RSTRING_LEN(buf
);
1377 total
= RSTRING_LEN(buf
);
1381 long mlen
= len
> 3 ? 3 : len
;
1383 if (s
< send
&& (unsigned char)*s
>= ' ' && (unsigned char)*s
< 'a')
1384 a
= ((unsigned char)*s
++ - ' ') & 077;
1387 if (s
< send
&& (unsigned char)*s
>= ' ' && (unsigned char)*s
< 'a')
1388 b
= ((unsigned char)*s
++ - ' ') & 077;
1391 if (s
< send
&& (unsigned char)*s
>= ' ' && (unsigned char)*s
< 'a')
1392 c
= ((unsigned char)*s
++ - ' ') & 077;
1395 if (s
< send
&& (unsigned char)*s
>= ' ' && (unsigned char)*s
< 'a')
1396 d
= ((unsigned char)*s
++ - ' ') & 077;
1399 hunk
[0] = (char)(a
<< 2 | b
>> 4);
1400 hunk
[1] = (char)(b
<< 4 | c
>> 2);
1401 hunk
[2] = (char)(c
<< 6 | d
);
1402 memcpy(ptr
, hunk
, mlen
);
1406 if (s
< send
&& (unsigned char)*s
!= '\r' && *s
!= '\n')
1407 s
++; /* possible checksum byte */
1408 if (s
< send
&& *s
== '\r') s
++;
1409 if (s
< send
&& *s
== '\n') s
++;
1412 rb_str_set_len(buf
, total
);
1419 VALUE buf
= rb_str_new(0, (send
- s
+ 3)*3/4); /* +3 is for skipping paddings */
1420 char *ptr
= RSTRING_PTR(buf
);
1421 int a
= -1,b
= -1,c
= 0,d
= 0;
1422 static signed char b64_xtable
[256];
1424 if (b64_xtable
['/'] <= 0) {
1427 for (i
= 0; i
< 256; i
++) {
1430 for (i
= 0; i
< 64; i
++) {
1431 b64_xtable
[(unsigned char)b64_table
[i
]] = (char)i
;
1437 a
= b64_xtable
[(unsigned char)*s
++];
1438 if (s
>= send
|| a
== -1) rb_raise(rb_eArgError
, "invalid base64");
1439 b
= b64_xtable
[(unsigned char)*s
++];
1440 if (s
>= send
|| b
== -1) rb_raise(rb_eArgError
, "invalid base64");
1442 if (s
+ 2 == send
&& *(s
+ 1) == '=') break;
1443 rb_raise(rb_eArgError
, "invalid base64");
1445 c
= b64_xtable
[(unsigned char)*s
++];
1446 if (s
>= send
|| c
== -1) rb_raise(rb_eArgError
, "invalid base64");
1447 if (s
+ 1 == send
&& *s
== '=') break;
1448 d
= b64_xtable
[(unsigned char)*s
++];
1449 if (d
== -1) rb_raise(rb_eArgError
, "invalid base64");
1450 *ptr
++ = castchar(a
<< 2 | b
>> 4);
1451 *ptr
++ = castchar(b
<< 4 | c
>> 2);
1452 *ptr
++ = castchar(c
<< 6 | d
);
1455 *ptr
++ = castchar(a
<< 2 | b
>> 4);
1456 if (b
& 0xf) rb_raise(rb_eArgError
, "invalid base64");
1459 *ptr
++ = castchar(a
<< 2 | b
>> 4);
1460 *ptr
++ = castchar(b
<< 4 | c
>> 2);
1461 if (c
& 0x3) rb_raise(rb_eArgError
, "invalid base64");
1467 while ((a
= b64_xtable
[(unsigned char)*s
]) == -1 && s
< send
) {s
++;}
1468 if (s
>= send
) break;
1470 while ((b
= b64_xtable
[(unsigned char)*s
]) == -1 && s
< send
) {s
++;}
1471 if (s
>= send
) break;
1473 while ((c
= b64_xtable
[(unsigned char)*s
]) == -1 && s
< send
) {if (*s
== '=') break; s
++;}
1474 if (*s
== '=' || s
>= send
) break;
1476 while ((d
= b64_xtable
[(unsigned char)*s
]) == -1 && s
< send
) {if (*s
== '=') break; s
++;}
1477 if (*s
== '=' || s
>= send
) break;
1479 *ptr
++ = castchar(a
<< 2 | b
>> 4);
1480 *ptr
++ = castchar(b
<< 4 | c
>> 2);
1481 *ptr
++ = castchar(c
<< 6 | d
);
1484 if (a
!= -1 && b
!= -1) {
1486 *ptr
++ = castchar(a
<< 2 | b
>> 4);
1488 *ptr
++ = castchar(a
<< 2 | b
>> 4);
1489 *ptr
++ = castchar(b
<< 4 | c
>> 2);
1493 rb_str_set_len(buf
, ptr
- RSTRING_PTR(buf
));
1500 VALUE buf
= rb_str_new(0, send
- s
);
1501 char *ptr
= RSTRING_PTR(buf
), *ss
= s
;
1507 if (++s
== send
) break;
1508 if (s
+1 < send
&& *s
== '\r' && *(s
+1) == '\n')
1511 if ((c1
= hex2num(*s
)) == -1) break;
1512 if (++s
== send
) break;
1513 if ((c2
= hex2num(*s
)) == -1) break;
1514 csum
|= *ptr
++ = castchar(c1
<< 4 | c2
);
1518 csum
|= *ptr
++ = *s
;
1523 rb_str_set_len(buf
, ptr
- RSTRING_PTR(buf
));
1524 rb_str_buf_cat(buf
, ss
, send
-ss
);
1525 csum
= ISASCII(csum
) ? ENC_CODERANGE_7BIT
: ENC_CODERANGE_VALID
;
1526 ENCODING_CODERANGE_SET(buf
, rb_ascii8bit_encindex(), csum
);
1532 if (len
> RSTRING_LEN(str
))
1533 rb_raise(rb_eArgError
, "@ outside of string");
1534 s
= RSTRING_PTR(str
) + len
;
1538 if (len
> s
- RSTRING_PTR(str
))
1539 rb_raise(rb_eArgError
, "X outside of string");
1545 rb_raise(rb_eArgError
, "x outside of string");
1550 if (sizeof(char *) <= (size_t)(send
- s
)) {
1554 UNPACK_FETCH(&t
, char *);
1556 if (!associates
) associates
= str_associated(str
);
1557 tmp
= associated_pointer(associates
, t
);
1558 if (len
< RSTRING_LEN(tmp
)) {
1559 tmp
= rb_str_new(t
, len
);
1560 str_associate(tmp
, associates
);
1568 if (len
> (long)((send
- s
) / sizeof(char *)))
1569 len
= (send
- s
) / sizeof(char *);
1571 if ((size_t)(send
- s
) < sizeof(char *))
1577 UNPACK_FETCH(&t
, char *);
1579 if (!associates
) associates
= str_associated(str
);
1580 tmp
= associated_pointer(associates
, t
);
1590 while (len
> 0 && s
< send
) {
1596 UNPACK_PUSH(rb_integer_unpack(s0
, s
-s0
, 1, 1, INTEGER_PACK_BIG_ENDIAN
));
1605 unknown_directive("unpack", type
, fmt
);
1614 pack_unpack(rb_execution_context_t
*ec
, VALUE str
, VALUE fmt
, VALUE offset
)
1616 enum unpack_mode mode
= rb_block_given_p() ? UNPACK_BLOCK
: UNPACK_ARRAY
;
1617 return pack_unpack_internal(str
, fmt
, mode
, RB_NUM2LONG(offset
));
1621 pack_unpack1(rb_execution_context_t
*ec
, VALUE str
, VALUE fmt
, VALUE offset
)
1623 return pack_unpack_internal(str
, fmt
, UNPACK_1
, RB_NUM2LONG(offset
));
1627 rb_uv_to_utf8(char buf
[6], unsigned long uv
)
1634 buf
[0] = castchar(((uv
>>6)&0xff)|0xc0);
1635 buf
[1] = castchar((uv
&0x3f)|0x80);
1639 buf
[0] = castchar(((uv
>>12)&0xff)|0xe0);
1640 buf
[1] = castchar(((uv
>>6)&0x3f)|0x80);
1641 buf
[2] = castchar((uv
&0x3f)|0x80);
1644 if (uv
<= 0x1fffff) {
1645 buf
[0] = castchar(((uv
>>18)&0xff)|0xf0);
1646 buf
[1] = castchar(((uv
>>12)&0x3f)|0x80);
1647 buf
[2] = castchar(((uv
>>6)&0x3f)|0x80);
1648 buf
[3] = castchar((uv
&0x3f)|0x80);
1651 if (uv
<= 0x3ffffff) {
1652 buf
[0] = castchar(((uv
>>24)&0xff)|0xf8);
1653 buf
[1] = castchar(((uv
>>18)&0x3f)|0x80);
1654 buf
[2] = castchar(((uv
>>12)&0x3f)|0x80);
1655 buf
[3] = castchar(((uv
>>6)&0x3f)|0x80);
1656 buf
[4] = castchar((uv
&0x3f)|0x80);
1659 if (uv
<= 0x7fffffff) {
1660 buf
[0] = castchar(((uv
>>30)&0xff)|0xfc);
1661 buf
[1] = castchar(((uv
>>24)&0x3f)|0x80);
1662 buf
[2] = castchar(((uv
>>18)&0x3f)|0x80);
1663 buf
[3] = castchar(((uv
>>12)&0x3f)|0x80);
1664 buf
[4] = castchar(((uv
>>6)&0x3f)|0x80);
1665 buf
[5] = castchar((uv
&0x3f)|0x80);
1668 rb_raise(rb_eRangeError
, "pack(U): value out of range");
1670 UNREACHABLE_RETURN(Qnil
);
1673 static const unsigned long utf8_limits
[] = {
1683 static unsigned long
1684 utf8_to_uv(const char *p
, long *lenp
)
1686 int c
= *p
++ & 0xff;
1687 unsigned long uv
= c
;
1696 rb_raise(rb_eArgError
, "malformed UTF-8 character");
1699 if (!(uv
& 0x20)) { n
= 2; uv
&= 0x1f; }
1700 else if (!(uv
& 0x10)) { n
= 3; uv
&= 0x0f; }
1701 else if (!(uv
& 0x08)) { n
= 4; uv
&= 0x07; }
1702 else if (!(uv
& 0x04)) { n
= 5; uv
&= 0x03; }
1703 else if (!(uv
& 0x02)) { n
= 6; uv
&= 0x01; }
1706 rb_raise(rb_eArgError
, "malformed UTF-8 character");
1709 rb_raise(rb_eArgError
, "malformed UTF-8 character (expected %ld bytes, given %ld bytes)",
1716 if ((c
& 0xc0) != 0x80) {
1718 rb_raise(rb_eArgError
, "malformed UTF-8 character");
1727 if (uv
< utf8_limits
[n
]) {
1728 rb_raise(rb_eArgError
, "redundant UTF-8 sequence");
1733 #include "pack.rbinc"
1738 id_associated
= rb_make_internal_id();