HCoop / bpt / guile.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
guile-backtrace function
[bpt/guile.git] / libguile / bytevectors.c
1 /* Copyright (C) 2009, 2010, 2011, 2012, 2013 Free Software Foundation, Inc.
2 *
3 * This library is free software; you can redistribute it and/or
4 * modify it under the terms of the GNU Lesser General Public License
5 * as published by the Free Software Foundation; either version 3 of
6 * the License, or (at your option) any later version.
7 *
8 * This library is distributed in the hope that it will be useful, but
9 * WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * Lesser General Public License for more details.
12 *
13 * You should have received a copy of the GNU Lesser General Public
14 * License along with this library; if not, write to the Free Software
15 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
16 * 02110-1301 USA
17 */
18
19
20 #ifdef HAVE_CONFIG_H
21 # include <config.h>
22 #endif
23
24 #include <alloca.h>
25 #include <assert.h>
26
27 #include <gmp.h>
28
29 #include "libguile/_scm.h"
30 #include "libguile/extensions.h"
31 #include "libguile/bytevectors.h"
32 #include "libguile/strings.h"
33 #include "libguile/validate.h"
34 #include "libguile/arrays.h"
35 #include "libguile/array-handle.h"
36 #include "libguile/uniform.h"
37 #include "libguile/srfi-4.h"
38
39 #include <byteswap.h>
40 #include <striconveh.h>
41 #include <uniconv.h>
42 #include <unistr.h>
43
44 #ifdef HAVE_LIMITS_H
45 # include <limits.h>
46 #else
47 /* Assuming 32-bit longs. */
48 # define ULONG_MAX 4294967295UL
49 #endif
50
51 #include <string.h>
52
53
54 \f
55 /* Utilities. */
56
57 /* Convenience macros. These are used by the various templates (macros) that
58 are parameterized by integer signedness. */
59 #define INT8_T_signed scm_t_int8
60 #define INT8_T_unsigned scm_t_uint8
61 #define INT16_T_signed scm_t_int16
62 #define INT16_T_unsigned scm_t_uint16
63 #define INT32_T_signed scm_t_int32
64 #define INT32_T_unsigned scm_t_uint32
65 #define is_signed_int8(_x) (((_x) >= -128L) && ((_x) <= 127L))
66 #define is_unsigned_int8(_x) ((_x) <= 255UL)
67 #define is_signed_int16(_x) (((_x) >= -32768L) && ((_x) <= 32767L))
68 #define is_unsigned_int16(_x) ((_x) <= 65535UL)
69 #define is_signed_int32(_x) (((_x) >= -2147483648L) && ((_x) <= 2147483647L))
70 #define is_unsigned_int32(_x) ((_x) <= 4294967295UL)
71 #define SIGNEDNESS_signed 1
72 #define SIGNEDNESS_unsigned 0
73
74 #define INT_TYPE(_size, _sign) INT ## _size ## _T_ ## _sign
75 #define INT_SWAP(_size) bswap_ ## _size
76 #define INT_VALID_P(_size, _sign) is_ ## _sign ## _int ## _size
77 #define SIGNEDNESS(_sign) SIGNEDNESS_ ## _sign
78
79
80 #define INTEGER_ACCESSOR_PROLOGUE(_len, _sign) \
81 size_t c_len, c_index; \
82 _sign char *c_bv; \
83 \
84 SCM_VALIDATE_BYTEVECTOR (1, bv); \
85 c_index = scm_to_uint (index); \
86 \
87 c_len = SCM_BYTEVECTOR_LENGTH (bv); \
88 c_bv = (_sign char *) SCM_BYTEVECTOR_CONTENTS (bv); \
89 \
90 if (SCM_UNLIKELY (c_index + ((_len) >> 3UL) - 1 >= c_len)) \
91 scm_out_of_range (FUNC_NAME, index);
92
93 /* Template for fixed-size integer access (only 8, 16 or 32-bit). */
94 #define INTEGER_REF(_len, _sign) \
95 SCM result; \
96 \
97 INTEGER_ACCESSOR_PROLOGUE (_len, _sign); \
98 SCM_VALIDATE_SYMBOL (3, endianness); \
99 \
100 { \
101 INT_TYPE (_len, _sign) c_result; \
102 \
103 memcpy (&c_result, &c_bv[c_index], (_len) / 8); \
104 if (!scm_is_eq (endianness, scm_i_native_endianness)) \
105 c_result = INT_SWAP (_len) (c_result); \
106 \
107 result = SCM_I_MAKINUM (c_result); \
108 } \
109 \
110 return result;
111
112 /* Template for fixed-size integer access using the native endianness. */
113 #define INTEGER_NATIVE_REF(_len, _sign) \
114 SCM result; \
115 \
116 INTEGER_ACCESSOR_PROLOGUE (_len, _sign); \
117 \
118 { \
119 INT_TYPE (_len, _sign) c_result; \
120 \
121 memcpy (&c_result, &c_bv[c_index], (_len) / 8); \
122 result = SCM_I_MAKINUM (c_result); \
123 } \
124 \
125 return result;
126
127 /* Template for fixed-size integer modification (only 8, 16 or 32-bit). */
128 #define INTEGER_SET(_len, _sign) \
129 INTEGER_ACCESSOR_PROLOGUE (_len, _sign); \
130 SCM_VALIDATE_SYMBOL (3, endianness); \
131 \
132 { \
133 scm_t_signed_bits c_value; \
134 INT_TYPE (_len, _sign) c_value_short; \
135 \
136 if (SCM_UNLIKELY (!SCM_I_INUMP (value))) \
137 scm_wrong_type_arg (FUNC_NAME, 3, value); \
138 \
139 c_value = SCM_I_INUM (value); \
140 if (SCM_UNLIKELY (!INT_VALID_P (_len, _sign) (c_value))) \
141 scm_out_of_range (FUNC_NAME, value); \
142 \
143 c_value_short = (INT_TYPE (_len, _sign)) c_value; \
144 if (!scm_is_eq (endianness, scm_i_native_endianness)) \
145 c_value_short = INT_SWAP (_len) (c_value_short); \
146 \
147 memcpy (&c_bv[c_index], &c_value_short, (_len) / 8); \
148 } \
149 \
150 return SCM_UNSPECIFIED;
151
152 /* Template for fixed-size integer modification using the native
153 endianness. */
154 #define INTEGER_NATIVE_SET(_len, _sign) \
155 INTEGER_ACCESSOR_PROLOGUE (_len, _sign); \
156 \
157 { \
158 scm_t_signed_bits c_value; \
159 INT_TYPE (_len, _sign) c_value_short; \
160 \
161 if (SCM_UNLIKELY (!SCM_I_INUMP (value))) \
162 scm_wrong_type_arg (FUNC_NAME, 3, value); \
163 \
164 c_value = SCM_I_INUM (value); \
165 if (SCM_UNLIKELY (!INT_VALID_P (_len, _sign) (c_value))) \
166 scm_out_of_range (FUNC_NAME, value); \
167 \
168 c_value_short = (INT_TYPE (_len, _sign)) c_value; \
169 \
170 memcpy (&c_bv[c_index], &c_value_short, (_len) / 8); \
171 } \
172 \
173 return SCM_UNSPECIFIED;
174
175
176 \f
177 /* Bytevector type. */
178
179 #define SCM_BYTEVECTOR_HEADER_BYTES \
180 (SCM_BYTEVECTOR_HEADER_SIZE * sizeof (scm_t_bits))
181
182 #define SCM_BYTEVECTOR_SET_LENGTH(_bv, _len) \
183 SCM_SET_CELL_WORD_1 ((_bv), (scm_t_bits) (_len))
184 #define SCM_BYTEVECTOR_SET_CONTENTS(_bv, _contents) \
185 SCM_SET_CELL_WORD_2 ((_bv), (scm_t_bits) (_contents))
186 #define SCM_BYTEVECTOR_SET_CONTIGUOUS_P(bv, contiguous_p) \
187 SCM_SET_BYTEVECTOR_FLAGS ((bv), \
188 SCM_BYTEVECTOR_ELEMENT_TYPE (bv) \
189 | ((contiguous_p) << 8UL))
190
191 #define SCM_BYTEVECTOR_SET_ELEMENT_TYPE(bv, hint) \
192 SCM_SET_BYTEVECTOR_FLAGS ((bv), \
193 (hint) \
194 | (SCM_BYTEVECTOR_CONTIGUOUS_P (bv) << 8UL))
195 #define SCM_BYTEVECTOR_SET_PARENT(_bv, _parent) \
196 SCM_SET_CELL_OBJECT_3 ((_bv), (_parent))
197
198 #define SCM_BYTEVECTOR_TYPE_SIZE(var) \
199 (scm_i_array_element_type_sizes[SCM_BYTEVECTOR_ELEMENT_TYPE (var)]/8)
200 #define SCM_BYTEVECTOR_TYPED_LENGTH(var) \
201 (SCM_BYTEVECTOR_LENGTH (var) / SCM_BYTEVECTOR_TYPE_SIZE (var))
202
203 /* The empty bytevector. */
204 SCM scm_null_bytevector = SCM_UNSPECIFIED;
205
206
207 static inline SCM
208 make_bytevector (size_t len, scm_t_array_element_type element_type)
209 {
210 SCM ret;
211 size_t c_len;
212
213 if (SCM_UNLIKELY (element_type > SCM_ARRAY_ELEMENT_TYPE_LAST
214 || scm_i_array_element_type_sizes[element_type] < 8
215 || len >= (((size_t) -1)
216 / (scm_i_array_element_type_sizes[element_type]/8))))
217 /* This would be an internal Guile programming error */
218 abort ();
219
220 if (SCM_UNLIKELY (len == 0 && element_type == SCM_ARRAY_ELEMENT_TYPE_VU8
221 && SCM_BYTEVECTOR_P (scm_null_bytevector)))
222 ret = scm_null_bytevector;
223 else
224 {
225 signed char *contents;
226
227 c_len = len * (scm_i_array_element_type_sizes[element_type] / 8);
228
229 contents = scm_gc_malloc_pointerless (SCM_BYTEVECTOR_HEADER_BYTES + c_len,
230 SCM_GC_BYTEVECTOR);
231 ret = SCM_PACK_POINTER (contents);
232 contents += SCM_BYTEVECTOR_HEADER_BYTES;
233
234 SCM_BYTEVECTOR_SET_LENGTH (ret, c_len);
235 SCM_BYTEVECTOR_SET_CONTENTS (ret, contents);
236 SCM_BYTEVECTOR_SET_CONTIGUOUS_P (ret, 1);
237 SCM_BYTEVECTOR_SET_ELEMENT_TYPE (ret, element_type);
238 SCM_BYTEVECTOR_SET_PARENT (ret, SCM_BOOL_F);
239 }
240
241 return ret;
242 }
243
244 /* Return a bytevector of LEN elements of type ELEMENT_TYPE, with element
245 values taken from CONTENTS. Assume that the storage for CONTENTS will be
246 automatically reclaimed when it becomes unreachable. */
247 static inline SCM
248 make_bytevector_from_buffer (size_t len, void *contents,
249 scm_t_array_element_type element_type)
250 {
251 SCM ret;
252
253 if (SCM_UNLIKELY (len == 0))
254 ret = make_bytevector (len, element_type);
255 else
256 {
257 size_t c_len;
258
259 ret = SCM_PACK_POINTER (scm_gc_malloc (SCM_BYTEVECTOR_HEADER_BYTES,
260 SCM_GC_BYTEVECTOR));
261
262 c_len = len * (scm_i_array_element_type_sizes[element_type] / 8);
263
264 SCM_BYTEVECTOR_SET_LENGTH (ret, c_len);
265 SCM_BYTEVECTOR_SET_CONTENTS (ret, contents);
266 SCM_BYTEVECTOR_SET_CONTIGUOUS_P (ret, 0);
267 SCM_BYTEVECTOR_SET_ELEMENT_TYPE (ret, element_type);
268 SCM_BYTEVECTOR_SET_PARENT (ret, SCM_BOOL_F);
269 }
270
271 return ret;
272 }
273
274
275 /* Return a new bytevector of size LEN octets. */
276 SCM
277 scm_c_make_bytevector (size_t len)
278 {
279 return make_bytevector (len, SCM_ARRAY_ELEMENT_TYPE_VU8);
280 }
281
282 /* Return a new bytevector of size LEN elements. */
283 SCM
284 scm_i_make_typed_bytevector (size_t len, scm_t_array_element_type element_type)
285 {
286 return make_bytevector (len, element_type);
287 }
288
289 /* Return a bytevector of size LEN made up of CONTENTS. The area
290 pointed to by CONTENTS must be protected from GC somehow: either
291 because it was allocated using `scm_gc_malloc ()', or because it is
292 part of PARENT. */
293 SCM
294 scm_c_take_gc_bytevector (signed char *contents, size_t len, SCM parent)
295 {
296 SCM ret;
297
298 ret = make_bytevector_from_buffer (len, contents, SCM_ARRAY_ELEMENT_TYPE_VU8);
299 SCM_BYTEVECTOR_SET_PARENT (ret, parent);
300
301 return ret;
302 }
303
304 SCM
305 scm_c_take_typed_bytevector (signed char *contents, size_t len,
306 scm_t_array_element_type element_type, SCM parent)
307 {
308 SCM ret;
309
310 ret = make_bytevector_from_buffer (len, contents, element_type);
311 SCM_BYTEVECTOR_SET_PARENT (ret, parent);
312
313 return ret;
314 }
315
316 /* Shrink BV to C_NEW_LEN (which is assumed to be smaller than its current
317 size) and return the new bytevector (possibly different from BV). */
318 SCM
319 scm_c_shrink_bytevector (SCM bv, size_t c_new_len)
320 {
321 SCM new_bv;
322 size_t c_len;
323
324 if (SCM_UNLIKELY (c_new_len % SCM_BYTEVECTOR_TYPE_SIZE (bv)))
325 /* This would be an internal Guile programming error */
326 abort ();
327
328 c_len = SCM_BYTEVECTOR_LENGTH (bv);
329 if (SCM_UNLIKELY (c_new_len > c_len))
330 abort ();
331
332 SCM_BYTEVECTOR_SET_LENGTH (bv, c_new_len);
333
334 if (SCM_BYTEVECTOR_CONTIGUOUS_P (bv))
335 new_bv = PTR2SCM (scm_gc_realloc (SCM2PTR (bv),
336 c_len + SCM_BYTEVECTOR_HEADER_BYTES,
337 c_new_len + SCM_BYTEVECTOR_HEADER_BYTES,
338 SCM_GC_BYTEVECTOR));
339 else
340 {
341 signed char *c_bv;
342
343 c_bv = scm_gc_realloc (SCM_BYTEVECTOR_CONTENTS (bv),
344 c_len, c_new_len, SCM_GC_BYTEVECTOR);
345 SCM_BYTEVECTOR_SET_CONTENTS (bv, c_bv);
346
347 new_bv = bv;
348 }
349
350 return new_bv;
351 }
352
353 int
354 scm_is_bytevector (SCM obj)
355 {
356 return SCM_BYTEVECTOR_P (obj);
357 }
358
359 size_t
360 scm_c_bytevector_length (SCM bv)
361 #define FUNC_NAME "scm_c_bytevector_length"
362 {
363 SCM_VALIDATE_BYTEVECTOR (1, bv);
364
365 return SCM_BYTEVECTOR_LENGTH (bv);
366 }
367 #undef FUNC_NAME
368
369 scm_t_uint8
370 scm_c_bytevector_ref (SCM bv, size_t index)
371 #define FUNC_NAME "scm_c_bytevector_ref"
372 {
373 size_t c_len;
374 const scm_t_uint8 *c_bv;
375
376 SCM_VALIDATE_BYTEVECTOR (1, bv);
377
378 c_len = SCM_BYTEVECTOR_LENGTH (bv);
379 c_bv = (scm_t_uint8 *) SCM_BYTEVECTOR_CONTENTS (bv);
380
381 if (SCM_UNLIKELY (index >= c_len))
382 scm_out_of_range (FUNC_NAME, scm_from_size_t (index));
383
384 return c_bv[index];
385 }
386 #undef FUNC_NAME
387
388 void
389 scm_c_bytevector_set_x (SCM bv, size_t index, scm_t_uint8 value)
390 #define FUNC_NAME "scm_c_bytevector_set_x"
391 {
392 size_t c_len;
393 scm_t_uint8 *c_bv;
394
395 SCM_VALIDATE_BYTEVECTOR (1, bv);
396
397 c_len = SCM_BYTEVECTOR_LENGTH (bv);
398 c_bv = (scm_t_uint8 *) SCM_BYTEVECTOR_CONTENTS (bv);
399
400 if (SCM_UNLIKELY (index >= c_len))
401 scm_out_of_range (FUNC_NAME, scm_from_size_t (index));
402
403 c_bv[index] = value;
404 }
405 #undef FUNC_NAME
406
407
408 \f
409 int
410 scm_i_print_bytevector (SCM bv, SCM port, scm_print_state *pstate SCM_UNUSED)
411 {
412 ssize_t ubnd, inc, i;
413 scm_t_array_handle h;
414
415 scm_array_get_handle (bv, &h);
416
417 scm_putc_unlocked ('#', port);
418 scm_write (scm_array_handle_element_type (&h), port);
419 scm_putc_unlocked ('(', port);
420 for (i = h.dims[0].lbnd, ubnd = h.dims[0].ubnd, inc = h.dims[0].inc;
421 i <= ubnd; i += inc)
422 {
423 if (i > 0)
424 scm_putc_unlocked (' ', port);
425 scm_write (scm_array_handle_ref (&h, i), port);
426 }
427 scm_putc_unlocked (')', port);
428
429 return 1;
430 }
431
432 \f
433 /* General operations. */
434
435 SCM_SYMBOL (scm_sym_big, "big");
436 SCM_SYMBOL (scm_sym_little, "little");
437
438 SCM scm_endianness_big, scm_endianness_little;
439
440 /* Host endianness (a symbol). */
441 SCM scm_i_native_endianness = SCM_UNSPECIFIED;
442
443 /* Byte-swapping. */
444 #ifndef bswap_24
445 # define bswap_24(_x) \
446 ((((_x) & 0xff0000) >> 16) | \
447 (((_x) & 0x00ff00)) | \
448 (((_x) & 0x0000ff) << 16))
449 #endif
450
451
452 SCM_DEFINE (scm_native_endianness, "native-endianness", 0, 0, 0,
453 (void),
454 "Return a symbol denoting the machine's native endianness.")
455 #define FUNC_NAME s_scm_native_endianness
456 {
457 return scm_i_native_endianness;
458 }
459 #undef FUNC_NAME
460
461 SCM_DEFINE (scm_bytevector_p, "bytevector?", 1, 0, 0,
462 (SCM obj),
463 "Return true if @var{obj} is a bytevector.")
464 #define FUNC_NAME s_scm_bytevector_p
465 {
466 return scm_from_bool (scm_is_bytevector (obj));
467 }
468 #undef FUNC_NAME
469
470 SCM_DEFINE (scm_make_bytevector, "make-bytevector", 1, 1, 0,
471 (SCM len, SCM fill),
472 "Return a newly allocated bytevector of @var{len} bytes, "
473 "optionally filled with @var{fill}.")
474 #define FUNC_NAME s_scm_make_bytevector
475 {
476 SCM bv;
477 unsigned c_len;
478 signed char c_fill = '\0';
479
480 SCM_VALIDATE_UINT_COPY (1, len, c_len);
481 if (!scm_is_eq (fill, SCM_UNDEFINED))
482 {
483 int value;
484
485 value = scm_to_int (fill);
486 if (SCM_UNLIKELY ((value < -128) || (value > 255)))
487 scm_out_of_range (FUNC_NAME, fill);
488 c_fill = (signed char) value;
489 }
490
491 bv = make_bytevector (c_len, SCM_ARRAY_ELEMENT_TYPE_VU8);
492 if (!scm_is_eq (fill, SCM_UNDEFINED))
493 {
494 unsigned i;
495 signed char *contents;
496
497 contents = SCM_BYTEVECTOR_CONTENTS (bv);
498 for (i = 0; i < c_len; i++)
499 contents[i] = c_fill;
500 }
501 else
502 memset (SCM_BYTEVECTOR_CONTENTS (bv), 0, c_len);
503
504 return bv;
505 }
506 #undef FUNC_NAME
507
508 SCM_DEFINE (scm_bytevector_length, "bytevector-length", 1, 0, 0,
509 (SCM bv),
510 "Return the length (in bytes) of @var{bv}.")
511 #define FUNC_NAME s_scm_bytevector_length
512 {
513 return scm_from_uint (scm_c_bytevector_length (bv));
514 }
515 #undef FUNC_NAME
516
517 SCM_DEFINE (scm_bytevector_eq_p, "bytevector=?", 2, 0, 0,
518 (SCM bv1, SCM bv2),
519 "Return is @var{bv1} equals to @var{bv2}---i.e., if they "
520 "have the same length and contents.")
521 #define FUNC_NAME s_scm_bytevector_eq_p
522 {
523 SCM result = SCM_BOOL_F;
524 unsigned c_len1, c_len2;
525
526 SCM_VALIDATE_BYTEVECTOR (1, bv1);
527 SCM_VALIDATE_BYTEVECTOR (2, bv2);
528
529 c_len1 = SCM_BYTEVECTOR_LENGTH (bv1);
530 c_len2 = SCM_BYTEVECTOR_LENGTH (bv2);
531
532 if (c_len1 == c_len2 && (SCM_BYTEVECTOR_ELEMENT_TYPE (bv1)
533 == SCM_BYTEVECTOR_ELEMENT_TYPE (bv2)))
534 {
535 signed char *c_bv1, *c_bv2;
536
537 c_bv1 = SCM_BYTEVECTOR_CONTENTS (bv1);
538 c_bv2 = SCM_BYTEVECTOR_CONTENTS (bv2);
539
540 result = scm_from_bool (!memcmp (c_bv1, c_bv2, c_len1));
541 }
542
543 return result;
544 }
545 #undef FUNC_NAME
546
547 SCM_DEFINE (scm_bytevector_fill_x, "bytevector-fill!", 2, 0, 0,
548 (SCM bv, SCM fill),
549 "Fill bytevector @var{bv} with @var{fill}, a byte.")
550 #define FUNC_NAME s_scm_bytevector_fill_x
551 {
552 unsigned c_len, i;
553 signed char *c_bv, c_fill;
554
555 SCM_VALIDATE_BYTEVECTOR (1, bv);
556 c_fill = scm_to_int8 (fill);
557
558 c_len = SCM_BYTEVECTOR_LENGTH (bv);
559 c_bv = SCM_BYTEVECTOR_CONTENTS (bv);
560
561 for (i = 0; i < c_len; i++)
562 c_bv[i] = c_fill;
563
564 return SCM_UNSPECIFIED;
565 }
566 #undef FUNC_NAME
567
568 SCM_DEFINE (scm_bytevector_copy_x, "bytevector-copy!", 5, 0, 0,
569 (SCM source, SCM source_start, SCM target, SCM target_start,
570 SCM len),
571 "Copy @var{len} bytes from @var{source} into @var{target}, "
572 "starting reading from @var{source_start} (a positive index "
573 "within @var{source}) and start writing at "
574 "@var{target_start}.")
575 #define FUNC_NAME s_scm_bytevector_copy_x
576 {
577 unsigned c_len, c_source_len, c_target_len;
578 unsigned c_source_start, c_target_start;
579 signed char *c_source, *c_target;
580
581 SCM_VALIDATE_BYTEVECTOR (1, source);
582 SCM_VALIDATE_BYTEVECTOR (3, target);
583
584 c_len = scm_to_uint (len);
585 c_source_start = scm_to_uint (source_start);
586 c_target_start = scm_to_uint (target_start);
587
588 c_source = SCM_BYTEVECTOR_CONTENTS (source);
589 c_target = SCM_BYTEVECTOR_CONTENTS (target);
590 c_source_len = SCM_BYTEVECTOR_LENGTH (source);
591 c_target_len = SCM_BYTEVECTOR_LENGTH (target);
592
593 if (SCM_UNLIKELY (c_source_start + c_len > c_source_len))
594 scm_out_of_range (FUNC_NAME, source_start);
595 if (SCM_UNLIKELY (c_target_start + c_len > c_target_len))
596 scm_out_of_range (FUNC_NAME, target_start);
597
598 memmove (c_target + c_target_start,
599 c_source + c_source_start,
600 c_len);
601
602 return SCM_UNSPECIFIED;
603 }
604 #undef FUNC_NAME
605
606 SCM_DEFINE (scm_bytevector_copy, "bytevector-copy", 1, 0, 0,
607 (SCM bv),
608 "Return a newly allocated copy of @var{bv}.")
609 #define FUNC_NAME s_scm_bytevector_copy
610 {
611 SCM copy;
612 unsigned c_len;
613 signed char *c_bv, *c_copy;
614
615 SCM_VALIDATE_BYTEVECTOR (1, bv);
616
617 c_len = SCM_BYTEVECTOR_LENGTH (bv);
618 c_bv = SCM_BYTEVECTOR_CONTENTS (bv);
619
620 copy = make_bytevector (c_len, SCM_BYTEVECTOR_ELEMENT_TYPE (bv));
621 c_copy = SCM_BYTEVECTOR_CONTENTS (copy);
622 memcpy (c_copy, c_bv, c_len);
623
624 return copy;
625 }
626 #undef FUNC_NAME
627
628 SCM_DEFINE (scm_uniform_array_to_bytevector, "uniform-array->bytevector",
629 1, 0, 0, (SCM array),
630 "Return a newly allocated bytevector whose contents\n"
631 "will be copied from the uniform array @var{array}.")
632 #define FUNC_NAME s_scm_uniform_array_to_bytevector
633 {
634 SCM contents, ret;
635 size_t len, sz, byte_len;
636 scm_t_array_handle h;
637 const void *elts;
638
639 contents = scm_array_contents (array, SCM_BOOL_T);
640 if (scm_is_false (contents))
641 scm_wrong_type_arg_msg (FUNC_NAME, 0, array, "uniform contiguous array");
642
643 scm_array_get_handle (contents, &h);
644 assert (h.base == 0);
645
646 elts = h.elements;
647 len = h.dims->inc * (h.dims->ubnd - h.dims->lbnd + 1);
648 sz = scm_array_handle_uniform_element_bit_size (&h);
649 if (sz >= 8 && ((sz % 8) == 0))
650 byte_len = len * (sz / 8);
651 else if (sz < 8)
652 /* Elements of sub-byte size (bitvectors) are addressed in 32-bit
653 units. */
654 byte_len = ((len * sz + 31) / 32) * 4;
655 else
656 /* an internal guile error, really */
657 SCM_MISC_ERROR ("uniform elements larger than 8 bits must fill whole bytes", SCM_EOL);
658
659 ret = make_bytevector (byte_len, SCM_ARRAY_ELEMENT_TYPE_VU8);
660 memcpy (SCM_BYTEVECTOR_CONTENTS (ret), elts, byte_len);
661
662 scm_array_handle_release (&h);
663
664 return ret;
665 }
666 #undef FUNC_NAME
667
668 \f
669 /* Operations on bytes and octets. */
670
671 SCM_DEFINE (scm_bytevector_u8_ref, "bytevector-u8-ref", 2, 0, 0,
672 (SCM bv, SCM index),
673 "Return the octet located at @var{index} in @var{bv}.")
674 #define FUNC_NAME s_scm_bytevector_u8_ref
675 {
676 INTEGER_NATIVE_REF (8, unsigned);
677 }
678 #undef FUNC_NAME
679
680 SCM_DEFINE (scm_bytevector_s8_ref, "bytevector-s8-ref", 2, 0, 0,
681 (SCM bv, SCM index),
682 "Return the byte located at @var{index} in @var{bv}.")
683 #define FUNC_NAME s_scm_bytevector_s8_ref
684 {
685 INTEGER_NATIVE_REF (8, signed);
686 }
687 #undef FUNC_NAME
688
689 SCM_DEFINE (scm_bytevector_u8_set_x, "bytevector-u8-set!", 3, 0, 0,
690 (SCM bv, SCM index, SCM value),
691 "Return the octet located at @var{index} in @var{bv}.")
692 #define FUNC_NAME s_scm_bytevector_u8_set_x
693 {
694 INTEGER_NATIVE_SET (8, unsigned);
695 }
696 #undef FUNC_NAME
697
698 SCM_DEFINE (scm_bytevector_s8_set_x, "bytevector-s8-set!", 3, 0, 0,
699 (SCM bv, SCM index, SCM value),
700 "Return the octet located at @var{index} in @var{bv}.")
701 #define FUNC_NAME s_scm_bytevector_s8_set_x
702 {
703 INTEGER_NATIVE_SET (8, signed);
704 }
705 #undef FUNC_NAME
706
707 #undef OCTET_ACCESSOR_PROLOGUE
708
709
710 SCM_DEFINE (scm_bytevector_to_u8_list, "bytevector->u8-list", 1, 0, 0,
711 (SCM bv),
712 "Return a newly allocated list of octets containing the "
713 "contents of @var{bv}.")
714 #define FUNC_NAME s_scm_bytevector_to_u8_list
715 {
716 SCM lst, pair;
717 unsigned c_len, i;
718 unsigned char *c_bv;
719
720 SCM_VALIDATE_BYTEVECTOR (1, bv);
721
722 c_len = SCM_BYTEVECTOR_LENGTH (bv);
723 c_bv = (unsigned char *) SCM_BYTEVECTOR_CONTENTS (bv);
724
725 lst = scm_make_list (scm_from_uint (c_len), SCM_UNSPECIFIED);
726 for (i = 0, pair = lst;
727 i < c_len;
728 i++, pair = SCM_CDR (pair))
729 {
730 SCM_SETCAR (pair, SCM_I_MAKINUM (c_bv[i]));
731 }
732
733 return lst;
734 }
735 #undef FUNC_NAME
736
737 SCM_DEFINE (scm_u8_list_to_bytevector, "u8-list->bytevector", 1, 0, 0,
738 (SCM lst),
739 "Turn @var{lst}, a list of octets, into a bytevector.")
740 #define FUNC_NAME s_scm_u8_list_to_bytevector
741 {
742 SCM bv, item;
743 long c_len, i;
744 unsigned char *c_bv;
745
746 SCM_VALIDATE_LIST_COPYLEN (1, lst, c_len);
747
748 bv = make_bytevector (c_len, SCM_ARRAY_ELEMENT_TYPE_VU8);
749 c_bv = (unsigned char *) SCM_BYTEVECTOR_CONTENTS (bv);
750
751 for (i = 0; i < c_len; lst = SCM_CDR (lst), i++)
752 {
753 item = SCM_CAR (lst);
754
755 if (SCM_LIKELY (SCM_I_INUMP (item)))
756 {
757 scm_t_signed_bits c_item;
758
759 c_item = SCM_I_INUM (item);
760 if (SCM_LIKELY ((c_item >= 0) && (c_item < 256)))
761 c_bv[i] = (unsigned char) c_item;
762 else
763 goto type_error;
764 }
765 else
766 goto type_error;
767 }
768
769 return bv;
770
771 type_error:
772 scm_wrong_type_arg (FUNC_NAME, 1, item);
773
774 return SCM_BOOL_F;
775 }
776 #undef FUNC_NAME
777
778 /* Compute the two's complement of VALUE (a positive integer) on SIZE octets
779 using (2^(SIZE * 8) - VALUE). */
780 static inline void
781 twos_complement (mpz_t value, size_t size)
782 {
783 unsigned long bit_count;
784
785 /* We expect BIT_COUNT to fit in a unsigned long thanks to the range
786 checking on SIZE performed earlier. */
787 bit_count = (unsigned long) size << 3UL;
788
789 if (SCM_LIKELY (bit_count < sizeof (unsigned long)))
790 mpz_ui_sub (value, 1UL << bit_count, value);
791 else
792 {
793 mpz_t max;
794
795 mpz_init (max);
796 mpz_ui_pow_ui (max, 2, bit_count);
797 mpz_sub (value, max, value);
798 mpz_clear (max);
799 }
800 }
801
802 static inline SCM
803 bytevector_large_ref (const char *c_bv, size_t c_size, int signed_p,
804 SCM endianness)
805 {
806 SCM result;
807 mpz_t c_mpz;
808 int c_endianness, negative_p = 0;
809
810 if (signed_p)
811 {
812 if (scm_is_eq (endianness, scm_sym_big))
813 negative_p = c_bv[0] & 0x80;
814 else
815 negative_p = c_bv[c_size - 1] & 0x80;
816 }
817
818 c_endianness = scm_is_eq (endianness, scm_sym_big) ? 1 : -1;
819
820 mpz_init (c_mpz);
821 mpz_import (c_mpz, 1 /* 1 word */, 1 /* word order doesn't matter */,
822 c_size /* word is C_SIZE-byte long */,
823 c_endianness,
824 0 /* nails */, c_bv);
825
826 if (signed_p && negative_p)
827 {
828 twos_complement (c_mpz, c_size);
829 mpz_neg (c_mpz, c_mpz);
830 }
831
832 result = scm_from_mpz (c_mpz);
833 mpz_clear (c_mpz); /* FIXME: Needed? */
834
835 return result;
836 }
837
838 static inline int
839 bytevector_large_set (char *c_bv, size_t c_size, int signed_p,
840 SCM value, SCM endianness)
841 {
842 mpz_t c_mpz;
843 int c_endianness, c_sign, err = 0;
844
845 c_endianness = scm_is_eq (endianness, scm_sym_big) ? 1 : -1;
846
847 mpz_init (c_mpz);
848 scm_to_mpz (value, c_mpz);
849
850 c_sign = mpz_sgn (c_mpz);
851 if (c_sign < 0)
852 {
853 if (SCM_LIKELY (signed_p))
854 {
855 mpz_neg (c_mpz, c_mpz);
856 twos_complement (c_mpz, c_size);
857 }
858 else
859 {
860 err = -1;
861 goto finish;
862 }
863 }
864
865 if (c_sign == 0)
866 /* Zero. */
867 memset (c_bv, 0, c_size);
868 else
869 {
870 size_t word_count, value_size;
871
872 value_size = (mpz_sizeinbase (c_mpz, 2) + (8 * c_size)) / (8 * c_size);
873 if (SCM_UNLIKELY (value_size > c_size))
874 {
875 err = -2;
876 goto finish;
877 }
878
879
880 mpz_export (c_bv, &word_count, 1 /* word order doesn't matter */,
881 c_size, c_endianness,
882 0 /* nails */, c_mpz);
883 if (SCM_UNLIKELY (word_count != 1))
884 /* Shouldn't happen since we already checked with VALUE_SIZE. */
885 abort ();
886 }
887
888 finish:
889 mpz_clear (c_mpz);
890
891 return err;
892 }
893
894 #define GENERIC_INTEGER_ACCESSOR_PROLOGUE(_sign) \
895 unsigned long c_len, c_index, c_size; \
896 char *c_bv; \
897 \
898 SCM_VALIDATE_BYTEVECTOR (1, bv); \
899 c_index = scm_to_ulong (index); \
900 c_size = scm_to_ulong (size); \
901 \
902 c_len = SCM_BYTEVECTOR_LENGTH (bv); \
903 c_bv = (char *) SCM_BYTEVECTOR_CONTENTS (bv); \
904 \
905 /* C_SIZE must have its 3 higher bits set to zero so that \
906 multiplying it by 8 yields a number that fits in an \
907 unsigned long. */ \
908 if (SCM_UNLIKELY ((c_size == 0) || (c_size >= (ULONG_MAX >> 3L)))) \
909 scm_out_of_range (FUNC_NAME, size); \
910 if (SCM_UNLIKELY (c_index + c_size > c_len)) \
911 scm_out_of_range (FUNC_NAME, index);
912
913
914 /* Template of an integer reference function. */
915 #define GENERIC_INTEGER_REF(_sign) \
916 SCM result; \
917 \
918 if (c_size < 3) \
919 { \
920 int swap; \
921 _sign int value; \
922 \
923 swap = !scm_is_eq (endianness, scm_i_native_endianness); \
924 switch (c_size) \
925 { \
926 case 1: \
927 { \
928 _sign char c_value8; \
929 memcpy (&c_value8, c_bv, 1); \
930 value = c_value8; \
931 } \
932 break; \
933 case 2: \
934 { \
935 INT_TYPE (16, _sign) c_value16; \
936 memcpy (&c_value16, c_bv, 2); \
937 if (swap) \
938 value = (INT_TYPE (16, _sign)) bswap_16 (c_value16); \
939 else \
940 value = c_value16; \
941 } \
942 break; \
943 default: \
944 abort (); \
945 } \
946 \
947 result = SCM_I_MAKINUM ((_sign int) value); \
948 } \
949 else \
950 result = bytevector_large_ref ((char *) c_bv, \
951 c_size, SIGNEDNESS (_sign), \
952 endianness); \
953 \
954 return result;
955
956 static inline SCM
957 bytevector_signed_ref (const char *c_bv, size_t c_size, SCM endianness)
958 {
959 GENERIC_INTEGER_REF (signed);
960 }
961
962 static inline SCM
963 bytevector_unsigned_ref (const char *c_bv, size_t c_size, SCM endianness)
964 {
965 GENERIC_INTEGER_REF (unsigned);
966 }
967
968
969 /* Template of an integer assignment function. */
970 #define GENERIC_INTEGER_SET(_sign) \
971 if (c_size < 3) \
972 { \
973 scm_t_signed_bits c_value; \
974 \
975 if (SCM_UNLIKELY (!SCM_I_INUMP (value))) \
976 goto range_error; \
977 \
978 c_value = SCM_I_INUM (value); \
979 switch (c_size) \
980 { \
981 case 1: \
982 if (SCM_LIKELY (INT_VALID_P (8, _sign) (c_value))) \
983 { \
984 _sign char c_value8; \
985 c_value8 = (_sign char) c_value; \
986 memcpy (c_bv, &c_value8, 1); \
987 } \
988 else \
989 goto range_error; \
990 break; \
991 \
992 case 2: \
993 if (SCM_LIKELY (INT_VALID_P (16, _sign) (c_value))) \
994 { \
995 int swap; \
996 INT_TYPE (16, _sign) c_value16; \
997 \
998 swap = !scm_is_eq (endianness, scm_i_native_endianness); \
999 \
1000 if (swap) \
1001 c_value16 = (INT_TYPE (16, _sign)) bswap_16 (c_value); \
1002 else \
1003 c_value16 = c_value; \
1004 \
1005 memcpy (c_bv, &c_value16, 2); \
1006 } \
1007 else \
1008 goto range_error; \
1009 break; \
1010 \
1011 default: \
1012 abort (); \
1013 } \
1014 } \
1015 else \
1016 { \
1017 int err; \
1018 \
1019 err = bytevector_large_set (c_bv, c_size, \
1020 SIGNEDNESS (_sign), \
1021 value, endianness); \
1022 if (err) \
1023 goto range_error; \
1024 } \
1025 \
1026 return; \
1027 \
1028 range_error: \
1029 scm_out_of_range (FUNC_NAME, value); \
1030 return;
1031
1032 static inline void
1033 bytevector_signed_set (char *c_bv, size_t c_size,
1034 SCM value, SCM endianness,
1035 const char *func_name)
1036 #define FUNC_NAME func_name
1037 {
1038 GENERIC_INTEGER_SET (signed);
1039 }
1040 #undef FUNC_NAME
1041
1042 static inline void
1043 bytevector_unsigned_set (char *c_bv, size_t c_size,
1044 SCM value, SCM endianness,
1045 const char *func_name)
1046 #define FUNC_NAME func_name
1047 {
1048 GENERIC_INTEGER_SET (unsigned);
1049 }
1050 #undef FUNC_NAME
1051
1052 #undef GENERIC_INTEGER_SET
1053 #undef GENERIC_INTEGER_REF
1054
1055
1056 SCM_DEFINE (scm_bytevector_uint_ref, "bytevector-uint-ref", 4, 0, 0,
1057 (SCM bv, SCM index, SCM endianness, SCM size),
1058 "Return the @var{size}-octet long unsigned integer at index "
1059 "@var{index} in @var{bv}.")
1060 #define FUNC_NAME s_scm_bytevector_uint_ref
1061 {
1062 GENERIC_INTEGER_ACCESSOR_PROLOGUE (unsigned);
1063
1064 return (bytevector_unsigned_ref (&c_bv[c_index], c_size, endianness));
1065 }
1066 #undef FUNC_NAME
1067
1068 SCM_DEFINE (scm_bytevector_sint_ref, "bytevector-sint-ref", 4, 0, 0,
1069 (SCM bv, SCM index, SCM endianness, SCM size),
1070 "Return the @var{size}-octet long unsigned integer at index "
1071 "@var{index} in @var{bv}.")
1072 #define FUNC_NAME s_scm_bytevector_sint_ref
1073 {
1074 GENERIC_INTEGER_ACCESSOR_PROLOGUE (signed);
1075
1076 return (bytevector_signed_ref (&c_bv[c_index], c_size, endianness));
1077 }
1078 #undef FUNC_NAME
1079
1080 SCM_DEFINE (scm_bytevector_uint_set_x, "bytevector-uint-set!", 5, 0, 0,
1081 (SCM bv, SCM index, SCM value, SCM endianness, SCM size),
1082 "Set the @var{size}-octet long unsigned integer at @var{index} "
1083 "to @var{value}.")
1084 #define FUNC_NAME s_scm_bytevector_uint_set_x
1085 {
1086 GENERIC_INTEGER_ACCESSOR_PROLOGUE (unsigned);
1087
1088 bytevector_unsigned_set (&c_bv[c_index], c_size, value, endianness,
1089 FUNC_NAME);
1090
1091 return SCM_UNSPECIFIED;
1092 }
1093 #undef FUNC_NAME
1094
1095 SCM_DEFINE (scm_bytevector_sint_set_x, "bytevector-sint-set!", 5, 0, 0,
1096 (SCM bv, SCM index, SCM value, SCM endianness, SCM size),
1097 "Set the @var{size}-octet long signed integer at @var{index} "
1098 "to @var{value}.")
1099 #define FUNC_NAME s_scm_bytevector_sint_set_x
1100 {
1101 GENERIC_INTEGER_ACCESSOR_PROLOGUE (signed);
1102
1103 bytevector_signed_set (&c_bv[c_index], c_size, value, endianness,
1104 FUNC_NAME);
1105
1106 return SCM_UNSPECIFIED;
1107 }
1108 #undef FUNC_NAME
1109
1110
1111 \f
1112 /* Operations on integers of arbitrary size. */
1113
1114 #define INTEGERS_TO_LIST(_sign) \
1115 SCM lst, pair; \
1116 size_t i, c_len, c_size; \
1117 \
1118 SCM_VALIDATE_BYTEVECTOR (1, bv); \
1119 SCM_VALIDATE_SYMBOL (2, endianness); \
1120 c_size = scm_to_unsigned_integer (size, 1, (size_t) -1); \
1121 \
1122 c_len = SCM_BYTEVECTOR_LENGTH (bv); \
1123 if (SCM_UNLIKELY (c_len % c_size != 0)) \
1124 scm_wrong_type_arg_msg \
1125 (FUNC_NAME, 0, size, \
1126 "an exact positive integer that divides the bytevector length"); \
1127 else if (SCM_UNLIKELY (c_len == 0)) \
1128 lst = SCM_EOL; \
1129 else \
1130 { \
1131 const char *c_bv; \
1132 \
1133 c_bv = (char *) SCM_BYTEVECTOR_CONTENTS (bv); \
1134 \
1135 lst = scm_make_list (scm_from_size_t (c_len / c_size), \
1136 SCM_UNSPECIFIED); \
1137 for (i = 0, pair = lst; \
1138 i <= c_len - c_size; \
1139 i += c_size, c_bv += c_size, pair = SCM_CDR (pair)) \
1140 { \
1141 SCM_SETCAR (pair, \
1142 bytevector_ ## _sign ## _ref (c_bv, c_size, \
1143 endianness)); \
1144 } \
1145 } \
1146 \
1147 return lst;
1148
1149 SCM_DEFINE (scm_bytevector_to_sint_list, "bytevector->sint-list",
1150 3, 0, 0,
1151 (SCM bv, SCM endianness, SCM size),
1152 "Return a list of signed integers of @var{size} octets "
1153 "representing the contents of @var{bv}.")
1154 #define FUNC_NAME s_scm_bytevector_to_sint_list
1155 {
1156 INTEGERS_TO_LIST (signed);
1157 }
1158 #undef FUNC_NAME
1159
1160 SCM_DEFINE (scm_bytevector_to_uint_list, "bytevector->uint-list",
1161 3, 0, 0,
1162 (SCM bv, SCM endianness, SCM size),
1163 "Return a list of unsigned integers of @var{size} octets "
1164 "representing the contents of @var{bv}.")
1165 #define FUNC_NAME s_scm_bytevector_to_uint_list
1166 {
1167 INTEGERS_TO_LIST (unsigned);
1168 }
1169 #undef FUNC_NAME
1170
1171 #undef INTEGER_TO_LIST
1172
1173
1174 #define INTEGER_LIST_TO_BYTEVECTOR(_sign) \
1175 SCM bv; \
1176 long c_len; \
1177 size_t c_size; \
1178 char *c_bv, *c_bv_ptr; \
1179 \
1180 SCM_VALIDATE_LIST_COPYLEN (1, lst, c_len); \
1181 SCM_VALIDATE_SYMBOL (2, endianness); \
1182 c_size = scm_to_uint (size); \
1183 \
1184 if (SCM_UNLIKELY ((c_size == 0) || (c_size >= (ULONG_MAX >> 3L)))) \
1185 scm_out_of_range (FUNC_NAME, size); \
1186 \
1187 bv = make_bytevector (c_len * c_size, SCM_ARRAY_ELEMENT_TYPE_VU8); \
1188 c_bv = (char *) SCM_BYTEVECTOR_CONTENTS (bv); \
1189 \
1190 for (c_bv_ptr = c_bv; \
1191 !scm_is_null (lst); \
1192 lst = SCM_CDR (lst), c_bv_ptr += c_size) \
1193 { \
1194 bytevector_ ## _sign ## _set (c_bv_ptr, c_size, \
1195 SCM_CAR (lst), endianness, \
1196 FUNC_NAME); \
1197 } \
1198 \
1199 return bv;
1200
1201
1202 SCM_DEFINE (scm_uint_list_to_bytevector, "uint-list->bytevector",
1203 3, 0, 0,
1204 (SCM lst, SCM endianness, SCM size),
1205 "Return a bytevector containing the unsigned integers "
1206 "listed in @var{lst} and encoded on @var{size} octets "
1207 "according to @var{endianness}.")
1208 #define FUNC_NAME s_scm_uint_list_to_bytevector
1209 {
1210 INTEGER_LIST_TO_BYTEVECTOR (unsigned);
1211 }
1212 #undef FUNC_NAME
1213
1214 SCM_DEFINE (scm_sint_list_to_bytevector, "sint-list->bytevector",
1215 3, 0, 0,
1216 (SCM lst, SCM endianness, SCM size),
1217 "Return a bytevector containing the signed integers "
1218 "listed in @var{lst} and encoded on @var{size} octets "
1219 "according to @var{endianness}.")
1220 #define FUNC_NAME s_scm_sint_list_to_bytevector
1221 {
1222 INTEGER_LIST_TO_BYTEVECTOR (signed);
1223 }
1224 #undef FUNC_NAME
1225
1226 #undef INTEGER_LIST_TO_BYTEVECTOR
1227
1228
1229 \f
1230 /* Operations on 16-bit integers. */
1231
1232 SCM_DEFINE (scm_bytevector_u16_ref, "bytevector-u16-ref",
1233 3, 0, 0,
1234 (SCM bv, SCM index, SCM endianness),
1235 "Return the unsigned 16-bit integer from @var{bv} at "
1236 "@var{index}.")
1237 #define FUNC_NAME s_scm_bytevector_u16_ref
1238 {
1239 INTEGER_REF (16, unsigned);
1240 }
1241 #undef FUNC_NAME
1242
1243 SCM_DEFINE (scm_bytevector_s16_ref, "bytevector-s16-ref",
1244 3, 0, 0,
1245 (SCM bv, SCM index, SCM endianness),
1246 "Return the signed 16-bit integer from @var{bv} at "
1247 "@var{index}.")
1248 #define FUNC_NAME s_scm_bytevector_s16_ref
1249 {
1250 INTEGER_REF (16, signed);
1251 }
1252 #undef FUNC_NAME
1253
1254 SCM_DEFINE (scm_bytevector_u16_native_ref, "bytevector-u16-native-ref",
1255 2, 0, 0,
1256 (SCM bv, SCM index),
1257 "Return the unsigned 16-bit integer from @var{bv} at "
1258 "@var{index} using the native endianness.")
1259 #define FUNC_NAME s_scm_bytevector_u16_native_ref
1260 {
1261 INTEGER_NATIVE_REF (16, unsigned);
1262 }
1263 #undef FUNC_NAME
1264
1265 SCM_DEFINE (scm_bytevector_s16_native_ref, "bytevector-s16-native-ref",
1266 2, 0, 0,
1267 (SCM bv, SCM index),
1268 "Return the unsigned 16-bit integer from @var{bv} at "
1269 "@var{index} using the native endianness.")
1270 #define FUNC_NAME s_scm_bytevector_s16_native_ref
1271 {
1272 INTEGER_NATIVE_REF (16, signed);
1273 }
1274 #undef FUNC_NAME
1275
1276 SCM_DEFINE (scm_bytevector_u16_set_x, "bytevector-u16-set!",
1277 4, 0, 0,
1278 (SCM bv, SCM index, SCM value, SCM endianness),
1279 "Store @var{value} in @var{bv} at @var{index} according to "
1280 "@var{endianness}.")
1281 #define FUNC_NAME s_scm_bytevector_u16_set_x
1282 {
1283 INTEGER_SET (16, unsigned);
1284 }
1285 #undef FUNC_NAME
1286
1287 SCM_DEFINE (scm_bytevector_s16_set_x, "bytevector-s16-set!",
1288 4, 0, 0,
1289 (SCM bv, SCM index, SCM value, SCM endianness),
1290 "Store @var{value} in @var{bv} at @var{index} according to "
1291 "@var{endianness}.")
1292 #define FUNC_NAME s_scm_bytevector_s16_set_x
1293 {
1294 INTEGER_SET (16, signed);
1295 }
1296 #undef FUNC_NAME
1297
1298 SCM_DEFINE (scm_bytevector_u16_native_set_x, "bytevector-u16-native-set!",
1299 3, 0, 0,
1300 (SCM bv, SCM index, SCM value),
1301 "Store the unsigned integer @var{value} at index @var{index} "
1302 "of @var{bv} using the native endianness.")
1303 #define FUNC_NAME s_scm_bytevector_u16_native_set_x
1304 {
1305 INTEGER_NATIVE_SET (16, unsigned);
1306 }
1307 #undef FUNC_NAME
1308
1309 SCM_DEFINE (scm_bytevector_s16_native_set_x, "bytevector-s16-native-set!",
1310 3, 0, 0,
1311 (SCM bv, SCM index, SCM value),
1312 "Store the signed integer @var{value} at index @var{index} "
1313 "of @var{bv} using the native endianness.")
1314 #define FUNC_NAME s_scm_bytevector_s16_native_set_x
1315 {
1316 INTEGER_NATIVE_SET (16, signed);
1317 }
1318 #undef FUNC_NAME
1319
1320
1321 \f
1322 /* Operations on 32-bit integers. */
1323
1324 /* Unfortunately, on 32-bit machines `SCM' is not large enough to hold
1325 arbitrary 32-bit integers. Thus we fall back to using the
1326 `large_{ref,set}' variants on 32-bit machines. */
1327
1328 #define LARGE_INTEGER_REF(_len, _sign) \
1329 INTEGER_ACCESSOR_PROLOGUE(_len, _sign); \
1330 SCM_VALIDATE_SYMBOL (3, endianness); \
1331 \
1332 return (bytevector_large_ref ((char *) c_bv + c_index, _len / 8, \
1333 SIGNEDNESS (_sign), endianness));
1334
1335 #define LARGE_INTEGER_SET(_len, _sign) \
1336 int err; \
1337 INTEGER_ACCESSOR_PROLOGUE (_len, _sign); \
1338 SCM_VALIDATE_SYMBOL (4, endianness); \
1339 \
1340 err = bytevector_large_set ((char *) c_bv + c_index, _len / 8, \
1341 SIGNEDNESS (_sign), value, endianness); \
1342 if (SCM_UNLIKELY (err)) \
1343 scm_out_of_range (FUNC_NAME, value); \
1344 \
1345 return SCM_UNSPECIFIED;
1346
1347 #define LARGE_INTEGER_NATIVE_REF(_len, _sign) \
1348 INTEGER_ACCESSOR_PROLOGUE(_len, _sign); \
1349 return (bytevector_large_ref ((char *) c_bv + c_index, _len / 8, \
1350 SIGNEDNESS (_sign), scm_i_native_endianness));
1351
1352 #define LARGE_INTEGER_NATIVE_SET(_len, _sign) \
1353 int err; \
1354 INTEGER_ACCESSOR_PROLOGUE (_len, _sign); \
1355 \
1356 err = bytevector_large_set ((char *) c_bv + c_index, _len / 8, \
1357 SIGNEDNESS (_sign), value, \
1358 scm_i_native_endianness); \
1359 if (SCM_UNLIKELY (err)) \
1360 scm_out_of_range (FUNC_NAME, value); \
1361 \
1362 return SCM_UNSPECIFIED;
1363
1364
1365 SCM_DEFINE (scm_bytevector_u32_ref, "bytevector-u32-ref",
1366 3, 0, 0,
1367 (SCM bv, SCM index, SCM endianness),
1368 "Return the unsigned 32-bit integer from @var{bv} at "
1369 "@var{index}.")
1370 #define FUNC_NAME s_scm_bytevector_u32_ref
1371 {
1372 #if SIZEOF_VOID_P > 4
1373 INTEGER_REF (32, unsigned);
1374 #else
1375 LARGE_INTEGER_REF (32, unsigned);
1376 #endif
1377 }
1378 #undef FUNC_NAME
1379
1380 SCM_DEFINE (scm_bytevector_s32_ref, "bytevector-s32-ref",
1381 3, 0, 0,
1382 (SCM bv, SCM index, SCM endianness),
1383 "Return the signed 32-bit integer from @var{bv} at "
1384 "@var{index}.")
1385 #define FUNC_NAME s_scm_bytevector_s32_ref
1386 {
1387 #if SIZEOF_VOID_P > 4
1388 INTEGER_REF (32, signed);
1389 #else
1390 LARGE_INTEGER_REF (32, signed);
1391 #endif
1392 }
1393 #undef FUNC_NAME
1394
1395 SCM_DEFINE (scm_bytevector_u32_native_ref, "bytevector-u32-native-ref",
1396 2, 0, 0,
1397 (SCM bv, SCM index),
1398 "Return the unsigned 32-bit integer from @var{bv} at "
1399 "@var{index} using the native endianness.")
1400 #define FUNC_NAME s_scm_bytevector_u32_native_ref
1401 {
1402 #if SIZEOF_VOID_P > 4
1403 INTEGER_NATIVE_REF (32, unsigned);
1404 #else
1405 LARGE_INTEGER_NATIVE_REF (32, unsigned);
1406 #endif
1407 }
1408 #undef FUNC_NAME
1409
1410 SCM_DEFINE (scm_bytevector_s32_native_ref, "bytevector-s32-native-ref",
1411 2, 0, 0,
1412 (SCM bv, SCM index),
1413 "Return the unsigned 32-bit integer from @var{bv} at "
1414 "@var{index} using the native endianness.")
1415 #define FUNC_NAME s_scm_bytevector_s32_native_ref
1416 {
1417 #if SIZEOF_VOID_P > 4
1418 INTEGER_NATIVE_REF (32, signed);
1419 #else
1420 LARGE_INTEGER_NATIVE_REF (32, signed);
1421 #endif
1422 }
1423 #undef FUNC_NAME
1424
1425 SCM_DEFINE (scm_bytevector_u32_set_x, "bytevector-u32-set!",
1426 4, 0, 0,
1427 (SCM bv, SCM index, SCM value, SCM endianness),
1428 "Store @var{value} in @var{bv} at @var{index} according to "
1429 "@var{endianness}.")
1430 #define FUNC_NAME s_scm_bytevector_u32_set_x
1431 {
1432 #if SIZEOF_VOID_P > 4
1433 INTEGER_SET (32, unsigned);
1434 #else
1435 LARGE_INTEGER_SET (32, unsigned);
1436 #endif
1437 }
1438 #undef FUNC_NAME
1439
1440 SCM_DEFINE (scm_bytevector_s32_set_x, "bytevector-s32-set!",
1441 4, 0, 0,
1442 (SCM bv, SCM index, SCM value, SCM endianness),
1443 "Store @var{value} in @var{bv} at @var{index} according to "
1444 "@var{endianness}.")
1445 #define FUNC_NAME s_scm_bytevector_s32_set_x
1446 {
1447 #if SIZEOF_VOID_P > 4
1448 INTEGER_SET (32, signed);
1449 #else
1450 LARGE_INTEGER_SET (32, signed);
1451 #endif
1452 }
1453 #undef FUNC_NAME
1454
1455 SCM_DEFINE (scm_bytevector_u32_native_set_x, "bytevector-u32-native-set!",
1456 3, 0, 0,
1457 (SCM bv, SCM index, SCM value),
1458 "Store the unsigned integer @var{value} at index @var{index} "
1459 "of @var{bv} using the native endianness.")
1460 #define FUNC_NAME s_scm_bytevector_u32_native_set_x
1461 {
1462 #if SIZEOF_VOID_P > 4
1463 INTEGER_NATIVE_SET (32, unsigned);
1464 #else
1465 LARGE_INTEGER_NATIVE_SET (32, unsigned);
1466 #endif
1467 }
1468 #undef FUNC_NAME
1469
1470 SCM_DEFINE (scm_bytevector_s32_native_set_x, "bytevector-s32-native-set!",
1471 3, 0, 0,
1472 (SCM bv, SCM index, SCM value),
1473 "Store the signed integer @var{value} at index @var{index} "
1474 "of @var{bv} using the native endianness.")
1475 #define FUNC_NAME s_scm_bytevector_s32_native_set_x
1476 {
1477 #if SIZEOF_VOID_P > 4
1478 INTEGER_NATIVE_SET (32, signed);
1479 #else
1480 LARGE_INTEGER_NATIVE_SET (32, signed);
1481 #endif
1482 }
1483 #undef FUNC_NAME
1484
1485
1486 \f
1487 /* Operations on 64-bit integers. */
1488
1489 /* For 64-bit integers, we use only the `large_{ref,set}' variant. */
1490
1491 SCM_DEFINE (scm_bytevector_u64_ref, "bytevector-u64-ref",
1492 3, 0, 0,
1493 (SCM bv, SCM index, SCM endianness),
1494 "Return the unsigned 64-bit integer from @var{bv} at "
1495 "@var{index}.")
1496 #define FUNC_NAME s_scm_bytevector_u64_ref
1497 {
1498 LARGE_INTEGER_REF (64, unsigned);
1499 }
1500 #undef FUNC_NAME
1501
1502 SCM_DEFINE (scm_bytevector_s64_ref, "bytevector-s64-ref",
1503 3, 0, 0,
1504 (SCM bv, SCM index, SCM endianness),
1505 "Return the signed 64-bit integer from @var{bv} at "
1506 "@var{index}.")
1507 #define FUNC_NAME s_scm_bytevector_s64_ref
1508 {
1509 LARGE_INTEGER_REF (64, signed);
1510 }
1511 #undef FUNC_NAME
1512
1513 SCM_DEFINE (scm_bytevector_u64_native_ref, "bytevector-u64-native-ref",
1514 2, 0, 0,
1515 (SCM bv, SCM index),
1516 "Return the unsigned 64-bit integer from @var{bv} at "
1517 "@var{index} using the native endianness.")
1518 #define FUNC_NAME s_scm_bytevector_u64_native_ref
1519 {
1520 LARGE_INTEGER_NATIVE_REF (64, unsigned);
1521 }
1522 #undef FUNC_NAME
1523
1524 SCM_DEFINE (scm_bytevector_s64_native_ref, "bytevector-s64-native-ref",
1525 2, 0, 0,
1526 (SCM bv, SCM index),
1527 "Return the unsigned 64-bit integer from @var{bv} at "
1528 "@var{index} using the native endianness.")
1529 #define FUNC_NAME s_scm_bytevector_s64_native_ref
1530 {
1531 LARGE_INTEGER_NATIVE_REF (64, signed);
1532 }
1533 #undef FUNC_NAME
1534
1535 SCM_DEFINE (scm_bytevector_u64_set_x, "bytevector-u64-set!",
1536 4, 0, 0,
1537 (SCM bv, SCM index, SCM value, SCM endianness),
1538 "Store @var{value} in @var{bv} at @var{index} according to "
1539 "@var{endianness}.")
1540 #define FUNC_NAME s_scm_bytevector_u64_set_x
1541 {
1542 LARGE_INTEGER_SET (64, unsigned);
1543 }
1544 #undef FUNC_NAME
1545
1546 SCM_DEFINE (scm_bytevector_s64_set_x, "bytevector-s64-set!",
1547 4, 0, 0,
1548 (SCM bv, SCM index, SCM value, SCM endianness),
1549 "Store @var{value} in @var{bv} at @var{index} according to "
1550 "@var{endianness}.")
1551 #define FUNC_NAME s_scm_bytevector_s64_set_x
1552 {
1553 LARGE_INTEGER_SET (64, signed);
1554 }
1555 #undef FUNC_NAME
1556
1557 SCM_DEFINE (scm_bytevector_u64_native_set_x, "bytevector-u64-native-set!",
1558 3, 0, 0,
1559 (SCM bv, SCM index, SCM value),
1560 "Store the unsigned integer @var{value} at index @var{index} "
1561 "of @var{bv} using the native endianness.")
1562 #define FUNC_NAME s_scm_bytevector_u64_native_set_x
1563 {
1564 LARGE_INTEGER_NATIVE_SET (64, unsigned);
1565 }
1566 #undef FUNC_NAME
1567
1568 SCM_DEFINE (scm_bytevector_s64_native_set_x, "bytevector-s64-native-set!",
1569 3, 0, 0,
1570 (SCM bv, SCM index, SCM value),
1571 "Store the signed integer @var{value} at index @var{index} "
1572 "of @var{bv} using the native endianness.")
1573 #define FUNC_NAME s_scm_bytevector_s64_native_set_x
1574 {
1575 LARGE_INTEGER_NATIVE_SET (64, signed);
1576 }
1577 #undef FUNC_NAME
1578
1579
1580 \f
1581 /* Operations on IEEE-754 numbers. */
1582
1583 /* There are two possible word endians, visible in glibc's <ieee754.h>.
1584 However, in R6RS, when the endianness is `little', little endian is
1585 assumed for both the byte order and the word order. This is clear from
1586 Section 2.1 of R6RS-lib (in response to
1587 http://www.r6rs.org/formal-comments/comment-187.txt). */
1588
1589 union scm_ieee754_float
1590 {
1591 float f;
1592 scm_t_uint32 i;
1593 };
1594
1595 union scm_ieee754_double
1596 {
1597 double d;
1598 scm_t_uint64 i;
1599 };
1600
1601
1602 /* Convert to/from a floating-point number with different endianness. This
1603 method is probably not the most efficient but it should be portable. */
1604
1605 static inline void
1606 float_to_foreign_endianness (union scm_ieee754_float *target,
1607 float source)
1608 {
1609 union scm_ieee754_float input;
1610
1611 input.f = source;
1612 target->i = bswap_32 (input.i);
1613 }
1614
1615 static inline float
1616 float_from_foreign_endianness (const union scm_ieee754_float *source)
1617 {
1618 union scm_ieee754_float result;
1619
1620 result.i = bswap_32 (source->i);
1621
1622 return (result.f);
1623 }
1624
1625 static inline void
1626 double_to_foreign_endianness (union scm_ieee754_double *target,
1627 double source)
1628 {
1629 union scm_ieee754_double input;
1630
1631 input.d = source;
1632 target->i = bswap_64 (input.i);
1633 }
1634
1635 static inline double
1636 double_from_foreign_endianness (const union scm_ieee754_double *source)
1637 {
1638 union scm_ieee754_double result;
1639
1640 result.i = bswap_64 (source->i);
1641
1642 return (result.d);
1643 }
1644
1645 /* Template macros to abstract over doubles and floats.
1646 XXX: Guile can only convert to/from doubles. */
1647 #define IEEE754_UNION(_c_type) union scm_ieee754_ ## _c_type
1648 #define IEEE754_TO_SCM(_c_type) scm_from_double
1649 #define IEEE754_FROM_SCM(_c_type) scm_to_double
1650 #define IEEE754_FROM_FOREIGN_ENDIANNESS(_c_type) \
1651 _c_type ## _from_foreign_endianness
1652 #define IEEE754_TO_FOREIGN_ENDIANNESS(_c_type) \
1653 _c_type ## _to_foreign_endianness
1654
1655
1656 /* FIXME: SCM_VALIDATE_REAL rejects integers, etc. grrr */
1657 #define VALIDATE_REAL(pos, v) \
1658 do { \
1659 SCM_ASSERT_TYPE (scm_is_real (v), v, pos, FUNC_NAME, "real"); \
1660 } while (0)
1661
1662 /* Templace getters and setters. */
1663
1664 #define IEEE754_ACCESSOR_PROLOGUE(_type) \
1665 INTEGER_ACCESSOR_PROLOGUE (sizeof (_type) << 3UL, signed);
1666
1667 #define IEEE754_REF(_type) \
1668 _type c_result; \
1669 \
1670 IEEE754_ACCESSOR_PROLOGUE (_type); \
1671 SCM_VALIDATE_SYMBOL (3, endianness); \
1672 \
1673 if (scm_is_eq (endianness, scm_i_native_endianness)) \
1674 memcpy (&c_result, &c_bv[c_index], sizeof (c_result)); \
1675 else \
1676 { \
1677 IEEE754_UNION (_type) c_raw; \
1678 \
1679 memcpy (&c_raw, &c_bv[c_index], sizeof (c_raw)); \
1680 c_result = \
1681 IEEE754_FROM_FOREIGN_ENDIANNESS (_type) (&c_raw); \
1682 } \
1683 \
1684 return (IEEE754_TO_SCM (_type) (c_result));
1685
1686 #define IEEE754_NATIVE_REF(_type) \
1687 _type c_result; \
1688 \
1689 IEEE754_ACCESSOR_PROLOGUE (_type); \
1690 \
1691 memcpy (&c_result, &c_bv[c_index], sizeof (c_result)); \
1692 return (IEEE754_TO_SCM (_type) (c_result));
1693
1694 #define IEEE754_SET(_type) \
1695 _type c_value; \
1696 \
1697 IEEE754_ACCESSOR_PROLOGUE (_type); \
1698 VALIDATE_REAL (3, value); \
1699 SCM_VALIDATE_SYMBOL (4, endianness); \
1700 c_value = IEEE754_FROM_SCM (_type) (value); \
1701 \
1702 if (scm_is_eq (endianness, scm_i_native_endianness)) \
1703 memcpy (&c_bv[c_index], &c_value, sizeof (c_value)); \
1704 else \
1705 { \
1706 IEEE754_UNION (_type) c_raw; \
1707 \
1708 IEEE754_TO_FOREIGN_ENDIANNESS (_type) (&c_raw, c_value); \
1709 memcpy (&c_bv[c_index], &c_raw, sizeof (c_raw)); \
1710 } \
1711 \
1712 return SCM_UNSPECIFIED;
1713
1714 #define IEEE754_NATIVE_SET(_type) \
1715 _type c_value; \
1716 \
1717 IEEE754_ACCESSOR_PROLOGUE (_type); \
1718 VALIDATE_REAL (3, value); \
1719 c_value = IEEE754_FROM_SCM (_type) (value); \
1720 \
1721 memcpy (&c_bv[c_index], &c_value, sizeof (c_value)); \
1722 return SCM_UNSPECIFIED;
1723
1724
1725 /* Single precision. */
1726
1727 SCM_DEFINE (scm_bytevector_ieee_single_ref,
1728 "bytevector-ieee-single-ref",
1729 3, 0, 0,
1730 (SCM bv, SCM index, SCM endianness),
1731 "Return the IEEE-754 single from @var{bv} at "
1732 "@var{index}.")
1733 #define FUNC_NAME s_scm_bytevector_ieee_single_ref
1734 {
1735 IEEE754_REF (float);
1736 }
1737 #undef FUNC_NAME
1738
1739 SCM_DEFINE (scm_bytevector_ieee_single_native_ref,
1740 "bytevector-ieee-single-native-ref",
1741 2, 0, 0,
1742 (SCM bv, SCM index),
1743 "Return the IEEE-754 single from @var{bv} at "
1744 "@var{index} using the native endianness.")
1745 #define FUNC_NAME s_scm_bytevector_ieee_single_native_ref
1746 {
1747 IEEE754_NATIVE_REF (float);
1748 }
1749 #undef FUNC_NAME
1750
1751 SCM_DEFINE (scm_bytevector_ieee_single_set_x,
1752 "bytevector-ieee-single-set!",
1753 4, 0, 0,
1754 (SCM bv, SCM index, SCM value, SCM endianness),
1755 "Store real @var{value} in @var{bv} at @var{index} according to "
1756 "@var{endianness}.")
1757 #define FUNC_NAME s_scm_bytevector_ieee_single_set_x
1758 {
1759 IEEE754_SET (float);
1760 }
1761 #undef FUNC_NAME
1762
1763 SCM_DEFINE (scm_bytevector_ieee_single_native_set_x,
1764 "bytevector-ieee-single-native-set!",
1765 3, 0, 0,
1766 (SCM bv, SCM index, SCM value),
1767 "Store the real @var{value} at index @var{index} "
1768 "of @var{bv} using the native endianness.")
1769 #define FUNC_NAME s_scm_bytevector_ieee_single_native_set_x
1770 {
1771 IEEE754_NATIVE_SET (float);
1772 }
1773 #undef FUNC_NAME
1774
1775
1776 /* Double precision. */
1777
1778 SCM_DEFINE (scm_bytevector_ieee_double_ref,
1779 "bytevector-ieee-double-ref",
1780 3, 0, 0,
1781 (SCM bv, SCM index, SCM endianness),
1782 "Return the IEEE-754 double from @var{bv} at "
1783 "@var{index}.")
1784 #define FUNC_NAME s_scm_bytevector_ieee_double_ref
1785 {
1786 IEEE754_REF (double);
1787 }
1788 #undef FUNC_NAME
1789
1790 SCM_DEFINE (scm_bytevector_ieee_double_native_ref,
1791 "bytevector-ieee-double-native-ref",
1792 2, 0, 0,
1793 (SCM bv, SCM index),
1794 "Return the IEEE-754 double from @var{bv} at "
1795 "@var{index} using the native endianness.")
1796 #define FUNC_NAME s_scm_bytevector_ieee_double_native_ref
1797 {
1798 IEEE754_NATIVE_REF (double);
1799 }
1800 #undef FUNC_NAME
1801
1802 SCM_DEFINE (scm_bytevector_ieee_double_set_x,
1803 "bytevector-ieee-double-set!",
1804 4, 0, 0,
1805 (SCM bv, SCM index, SCM value, SCM endianness),
1806 "Store real @var{value} in @var{bv} at @var{index} according to "
1807 "@var{endianness}.")
1808 #define FUNC_NAME s_scm_bytevector_ieee_double_set_x
1809 {
1810 IEEE754_SET (double);
1811 }
1812 #undef FUNC_NAME
1813
1814 SCM_DEFINE (scm_bytevector_ieee_double_native_set_x,
1815 "bytevector-ieee-double-native-set!",
1816 3, 0, 0,
1817 (SCM bv, SCM index, SCM value),
1818 "Store the real @var{value} at index @var{index} "
1819 "of @var{bv} using the native endianness.")
1820 #define FUNC_NAME s_scm_bytevector_ieee_double_native_set_x
1821 {
1822 IEEE754_NATIVE_SET (double);
1823 }
1824 #undef FUNC_NAME
1825
1826
1827 #undef IEEE754_UNION
1828 #undef IEEE754_TO_SCM
1829 #undef IEEE754_FROM_SCM
1830 #undef IEEE754_FROM_FOREIGN_ENDIANNESS
1831 #undef IEEE754_TO_FOREIGN_ENDIANNESS
1832 #undef IEEE754_REF
1833 #undef IEEE754_NATIVE_REF
1834 #undef IEEE754_SET
1835 #undef IEEE754_NATIVE_SET
1836
1837 \f
1838 /* Operations on strings. */
1839
1840
1841 /* Produce a function that returns the length of a UTF-encoded string. */
1842 #define UTF_STRLEN_FUNCTION(_utf_width) \
1843 static inline size_t \
1844 utf ## _utf_width ## _strlen (const uint ## _utf_width ## _t *str) \
1845 { \
1846 size_t len = 0; \
1847 const uint ## _utf_width ## _t *ptr; \
1848 for (ptr = str; \
1849 *ptr != 0; \
1850 ptr++) \
1851 { \
1852 len++; \
1853 } \
1854 \
1855 return (len * ((_utf_width) / 8)); \
1856 }
1857
1858 UTF_STRLEN_FUNCTION (8)
1859
1860
1861 /* Return the length (in bytes) of STR, a UTF-(UTF_WIDTH) encoded string. */
1862 #define UTF_STRLEN(_utf_width, _str) \
1863 utf ## _utf_width ## _strlen (_str)
1864
1865 /* Return the "portable" name of the UTF encoding of size UTF_WIDTH and
1866 ENDIANNESS (Gnulib's `iconv_open' module guarantees the portability of the
1867 encoding name). */
1868 static inline void
1869 utf_encoding_name (char *name, size_t utf_width, SCM endianness)
1870 {
1871 strcpy (name, "UTF-");
1872 strcat (name, ((utf_width == 8)
1873 ? "8"
1874 : ((utf_width == 16)
1875 ? "16"
1876 : ((utf_width == 32)
1877 ? "32"
1878 : "??"))));
1879 strcat (name,
1880 ((scm_is_eq (endianness, scm_sym_big))
1881 ? "BE"
1882 : ((scm_is_eq (endianness, scm_sym_little))
1883 ? "LE"
1884 : "unknown")));
1885 }
1886
1887 /* Maximum length of a UTF encoding name. */
1888 #define MAX_UTF_ENCODING_NAME_LEN 16
1889
1890 /* Produce the body of a `string->utf' function. */
1891 #define STRING_TO_UTF(_utf_width) \
1892 SCM utf; \
1893 int err; \
1894 char c_utf_name[MAX_UTF_ENCODING_NAME_LEN]; \
1895 char *c_utf = NULL; \
1896 size_t c_strlen, c_utf_len = 0; \
1897 \
1898 SCM_VALIDATE_STRING (1, str); \
1899 if (scm_is_eq (endianness, SCM_UNDEFINED)) \
1900 endianness = scm_sym_big; \
1901 else \
1902 SCM_VALIDATE_SYMBOL (2, endianness); \
1903 \
1904 utf_encoding_name (c_utf_name, (_utf_width), endianness); \
1905 \
1906 c_strlen = scm_i_string_length (str); \
1907 if (scm_i_is_narrow_string (str)) \
1908 { \
1909 err = mem_iconveh (scm_i_string_chars (str), c_strlen, \
1910 "ISO-8859-1", c_utf_name, \
1911 iconveh_question_mark, NULL, \
1912 &c_utf, &c_utf_len); \
1913 if (SCM_UNLIKELY (err)) \
1914 scm_syserror_msg (FUNC_NAME, "failed to convert string: ~A", \
1915 scm_list_1 (str), err); \
1916 } \
1917 else \
1918 { \
1919 const scm_t_wchar *wbuf = scm_i_string_wide_chars (str); \
1920 c_utf = u32_conv_to_encoding (c_utf_name, \
1921 iconveh_question_mark, \
1922 (scm_t_uint32 *) wbuf, \
1923 c_strlen, NULL, NULL, &c_utf_len); \
1924 if (SCM_UNLIKELY (c_utf == NULL)) \
1925 scm_syserror_msg (FUNC_NAME, "failed to convert string: ~A", \
1926 scm_list_1 (str), errno); \
1927 } \
1928 scm_dynwind_begin (0); \
1929 scm_dynwind_free (c_utf); \
1930 utf = make_bytevector (c_utf_len, SCM_ARRAY_ELEMENT_TYPE_VU8); \
1931 memcpy (SCM_BYTEVECTOR_CONTENTS (utf), c_utf, c_utf_len); \
1932 scm_dynwind_end (); \
1933 \
1934 return (utf);
1935
1936
1937
1938 SCM_DEFINE (scm_string_to_utf8, "string->utf8",
1939 1, 0, 0,
1940 (SCM str),
1941 "Return a newly allocated bytevector that contains the UTF-8 "
1942 "encoding of @var{str}.")
1943 #define FUNC_NAME s_scm_string_to_utf8
1944 {
1945 SCM utf;
1946 scm_t_uint8 *c_utf;
1947 size_t c_utf_len = 0;
1948
1949 SCM_VALIDATE_STRING (1, str);
1950
1951 c_utf = (scm_t_uint8 *) scm_to_utf8_stringn (str, &c_utf_len);
1952 utf = make_bytevector (c_utf_len, SCM_ARRAY_ELEMENT_TYPE_VU8);
1953 memcpy (SCM_BYTEVECTOR_CONTENTS (utf), c_utf, c_utf_len);
1954 free (c_utf);
1955
1956 return (utf);
1957 }
1958 #undef FUNC_NAME
1959
1960 SCM_DEFINE (scm_string_to_utf16, "string->utf16",
1961 1, 1, 0,
1962 (SCM str, SCM endianness),
1963 "Return a newly allocated bytevector that contains the UTF-16 "
1964 "encoding of @var{str}.")
1965 #define FUNC_NAME s_scm_string_to_utf16
1966 {
1967 STRING_TO_UTF (16);
1968 }
1969 #undef FUNC_NAME
1970
1971 static void
1972 swap_u32 (scm_t_wchar *vals, size_t len)
1973 {
1974 size_t n;
1975 for (n = 0; n < len; n++)
1976 vals[n] = bswap_32 (vals[n]);
1977 }
1978
1979 SCM_DEFINE (scm_string_to_utf32, "string->utf32",
1980 1, 1, 0,
1981 (SCM str, SCM endianness),
1982 "Return a newly allocated bytevector that contains the UTF-32 "
1983 "encoding of @var{str}.")
1984 #define FUNC_NAME s_scm_string_to_utf32
1985 {
1986 SCM bv;
1987 scm_t_wchar *wchars;
1988 size_t wchar_len, bytes_len;
1989
1990 wchars = scm_to_utf32_stringn (str, &wchar_len);
1991 bytes_len = wchar_len * sizeof (scm_t_wchar);
1992 if (!scm_is_eq (SCM_UNBNDP (endianness) ? scm_endianness_big : endianness,
1993 scm_i_native_endianness))
1994 swap_u32 (wchars, wchar_len);
1995
1996 bv = make_bytevector (bytes_len, SCM_ARRAY_ELEMENT_TYPE_VU8);
1997 memcpy (SCM_BYTEVECTOR_CONTENTS (bv), wchars, bytes_len);
1998 free (wchars);
1999
2000 return bv;
2001 }
2002 #undef FUNC_NAME
2003
2004
2005 /* Produce the body of a function that converts a UTF-encoded bytevector to a
2006 string. */
2007 #define UTF_TO_STRING(_utf_width) \
2008 SCM str = SCM_BOOL_F; \
2009 int err; \
2010 char *c_str = NULL; \
2011 char c_utf_name[MAX_UTF_ENCODING_NAME_LEN]; \
2012 char *c_utf; \
2013 size_t c_strlen = 0, c_utf_len = 0; \
2014 \
2015 SCM_VALIDATE_BYTEVECTOR (1, utf); \
2016 if (scm_is_eq (endianness, SCM_UNDEFINED)) \
2017 endianness = scm_sym_big; \
2018 else \
2019 SCM_VALIDATE_SYMBOL (2, endianness); \
2020 \
2021 c_utf_len = SCM_BYTEVECTOR_LENGTH (utf); \
2022 c_utf = (char *) SCM_BYTEVECTOR_CONTENTS (utf); \
2023 utf_encoding_name (c_utf_name, (_utf_width), endianness); \
2024 \
2025 err = mem_iconveh (c_utf, c_utf_len, \
2026 c_utf_name, "UTF-8", \
2027 iconveh_question_mark, NULL, \
2028 &c_str, &c_strlen); \
2029 if (SCM_UNLIKELY (err)) \
2030 scm_syserror_msg (FUNC_NAME, "failed to convert to string: ~A", \
2031 scm_list_1 (utf), err); \
2032 else \
2033 { \
2034 str = scm_from_utf8_stringn (c_str, c_strlen); \
2035 free (c_str); \
2036 } \
2037 return (str);
2038
2039
2040 SCM_DEFINE (scm_utf8_to_string, "utf8->string",
2041 1, 0, 0,
2042 (SCM utf),
2043 "Return a newly allocate string that contains from the UTF-8-"
2044 "encoded contents of bytevector @var{utf}.")
2045 #define FUNC_NAME s_scm_utf8_to_string
2046 {
2047 SCM str;
2048 const char *c_utf;
2049 size_t c_utf_len = 0;
2050
2051 SCM_VALIDATE_BYTEVECTOR (1, utf);
2052
2053 c_utf_len = SCM_BYTEVECTOR_LENGTH (utf);
2054 c_utf = (char *) SCM_BYTEVECTOR_CONTENTS (utf);
2055 str = scm_from_utf8_stringn (c_utf, c_utf_len);
2056
2057 return (str);
2058 }
2059 #undef FUNC_NAME
2060
2061 SCM_DEFINE (scm_utf16_to_string, "utf16->string",
2062 1, 1, 0,
2063 (SCM utf, SCM endianness),
2064 "Return a newly allocate string that contains from the UTF-16-"
2065 "encoded contents of bytevector @var{utf}.")
2066 #define FUNC_NAME s_scm_utf16_to_string
2067 {
2068 UTF_TO_STRING (16);
2069 }
2070 #undef FUNC_NAME
2071
2072 SCM_DEFINE (scm_utf32_to_string, "utf32->string",
2073 1, 1, 0,
2074 (SCM utf, SCM endianness),
2075 "Return a newly allocate string that contains from the UTF-32-"
2076 "encoded contents of bytevector @var{utf}.")
2077 #define FUNC_NAME s_scm_utf32_to_string
2078 {
2079 UTF_TO_STRING (32);
2080 }
2081 #undef FUNC_NAME
2082
2083 \f
2084 /* Initialization. */
2085
2086 void
2087 scm_bootstrap_bytevectors (void)
2088 {
2089 /* This must be instantiated here because the generalized-vector API may
2090 want to access bytevectors even though `(rnrs bytevectors)' hasn't been
2091 loaded. */
2092 scm_null_bytevector = make_bytevector (0, SCM_ARRAY_ELEMENT_TYPE_VU8);
2093
2094 #ifdef WORDS_BIGENDIAN
2095 scm_i_native_endianness = scm_from_latin1_symbol ("big");
2096 #else
2097 scm_i_native_endianness = scm_from_latin1_symbol ("little");
2098 #endif
2099
2100 scm_c_register_extension ("libguile-" SCM_EFFECTIVE_VERSION,
2101 "scm_init_bytevectors",
2102 (scm_t_extension_init_func) scm_init_bytevectors,
2103 NULL);
2104
2105 scm_i_register_vector_constructor
2106 (scm_i_array_element_types[SCM_ARRAY_ELEMENT_TYPE_VU8],
2107 scm_make_bytevector);
2108 }
2109
2110 void
2111 scm_init_bytevectors (void)
2112 {
2113 #include "libguile/bytevectors.x"
2114
2115 scm_endianness_big = scm_sym_big;
2116 scm_endianness_little = scm_sym_little;
2117 }
Unnamed repository; edit this file 'description' to name the repository.