HCoop / bpt / guile.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge commit 'a7bbba05838cabe2294f498e7008e1c51db6d664'
[bpt/guile.git] / libguile / bytevectors.c
1 /* Copyright (C) 2009-2014 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 {
336 signed char *c_bv;
337
338 c_bv = scm_gc_realloc (SCM2PTR (bv),
339 c_len + SCM_BYTEVECTOR_HEADER_BYTES,
340 c_new_len + SCM_BYTEVECTOR_HEADER_BYTES,
341 SCM_GC_BYTEVECTOR);
342 new_bv = PTR2SCM (c_bv);
343 SCM_BYTEVECTOR_SET_CONTENTS (new_bv, c_bv + SCM_BYTEVECTOR_HEADER_BYTES);
344 }
345 else
346 {
347 signed char *c_bv;
348
349 c_bv = scm_gc_realloc (SCM_BYTEVECTOR_CONTENTS (bv),
350 c_len, c_new_len, SCM_GC_BYTEVECTOR);
351 SCM_BYTEVECTOR_SET_CONTENTS (bv, c_bv);
352
353 new_bv = bv;
354 }
355
356 return new_bv;
357 }
358
359 int
360 scm_is_bytevector (SCM obj)
361 {
362 return SCM_BYTEVECTOR_P (obj);
363 }
364
365 size_t
366 scm_c_bytevector_length (SCM bv)
367 #define FUNC_NAME "scm_c_bytevector_length"
368 {
369 SCM_VALIDATE_BYTEVECTOR (1, bv);
370
371 return SCM_BYTEVECTOR_LENGTH (bv);
372 }
373 #undef FUNC_NAME
374
375 scm_t_uint8
376 scm_c_bytevector_ref (SCM bv, size_t index)
377 #define FUNC_NAME "scm_c_bytevector_ref"
378 {
379 size_t c_len;
380 const scm_t_uint8 *c_bv;
381
382 SCM_VALIDATE_BYTEVECTOR (1, bv);
383
384 c_len = SCM_BYTEVECTOR_LENGTH (bv);
385 c_bv = (scm_t_uint8 *) SCM_BYTEVECTOR_CONTENTS (bv);
386
387 if (SCM_UNLIKELY (index >= c_len))
388 scm_out_of_range (FUNC_NAME, scm_from_size_t (index));
389
390 return c_bv[index];
391 }
392 #undef FUNC_NAME
393
394 void
395 scm_c_bytevector_set_x (SCM bv, size_t index, scm_t_uint8 value)
396 #define FUNC_NAME "scm_c_bytevector_set_x"
397 {
398 size_t c_len;
399 scm_t_uint8 *c_bv;
400
401 SCM_VALIDATE_BYTEVECTOR (1, bv);
402
403 c_len = SCM_BYTEVECTOR_LENGTH (bv);
404 c_bv = (scm_t_uint8 *) SCM_BYTEVECTOR_CONTENTS (bv);
405
406 if (SCM_UNLIKELY (index >= c_len))
407 scm_out_of_range (FUNC_NAME, scm_from_size_t (index));
408
409 c_bv[index] = value;
410 }
411 #undef FUNC_NAME
412
413
414 \f
415 int
416 scm_i_print_bytevector (SCM bv, SCM port, scm_print_state *pstate SCM_UNUSED)
417 {
418 ssize_t ubnd, inc, i;
419 scm_t_array_handle h;
420
421 scm_array_get_handle (bv, &h);
422
423 scm_putc_unlocked ('#', port);
424 scm_write (scm_array_handle_element_type (&h), port);
425 scm_putc_unlocked ('(', port);
426 for (i = h.dims[0].lbnd, ubnd = h.dims[0].ubnd, inc = h.dims[0].inc;
427 i <= ubnd; i += inc)
428 {
429 if (i > 0)
430 scm_putc_unlocked (' ', port);
431 scm_write (scm_array_handle_ref (&h, i), port);
432 }
433 scm_putc_unlocked (')', port);
434
435 return 1;
436 }
437
438 \f
439 /* General operations. */
440
441 SCM_SYMBOL (scm_sym_big, "big");
442 SCM_SYMBOL (scm_sym_little, "little");
443
444 SCM scm_endianness_big, scm_endianness_little;
445
446 /* Host endianness (a symbol). */
447 SCM scm_i_native_endianness = SCM_UNSPECIFIED;
448
449 /* Byte-swapping. */
450 #ifndef bswap_24
451 # define bswap_24(_x) \
452 ((((_x) & 0xff0000) >> 16) | \
453 (((_x) & 0x00ff00)) | \
454 (((_x) & 0x0000ff) << 16))
455 #endif
456
457
458 SCM_DEFINE (scm_native_endianness, "native-endianness", 0, 0, 0,
459 (void),
460 "Return a symbol denoting the machine's native endianness.")
461 #define FUNC_NAME s_scm_native_endianness
462 {
463 return scm_i_native_endianness;
464 }
465 #undef FUNC_NAME
466
467 SCM_DEFINE (scm_bytevector_p, "bytevector?", 1, 0, 0,
468 (SCM obj),
469 "Return true if @var{obj} is a bytevector.")
470 #define FUNC_NAME s_scm_bytevector_p
471 {
472 return scm_from_bool (scm_is_bytevector (obj));
473 }
474 #undef FUNC_NAME
475
476 SCM_DEFINE (scm_make_bytevector, "make-bytevector", 1, 1, 0,
477 (SCM len, SCM fill),
478 "Return a newly allocated bytevector of @var{len} bytes, "
479 "optionally filled with @var{fill}.")
480 #define FUNC_NAME s_scm_make_bytevector
481 {
482 SCM bv;
483 unsigned c_len;
484 signed char c_fill = '\0';
485
486 SCM_VALIDATE_UINT_COPY (1, len, c_len);
487 if (!scm_is_eq (fill, SCM_UNDEFINED))
488 {
489 int value;
490
491 value = scm_to_int (fill);
492 if (SCM_UNLIKELY ((value < -128) || (value > 255)))
493 scm_out_of_range (FUNC_NAME, fill);
494 c_fill = (signed char) value;
495 }
496
497 bv = make_bytevector (c_len, SCM_ARRAY_ELEMENT_TYPE_VU8);
498 if (!scm_is_eq (fill, SCM_UNDEFINED))
499 {
500 unsigned i;
501 signed char *contents;
502
503 contents = SCM_BYTEVECTOR_CONTENTS (bv);
504 for (i = 0; i < c_len; i++)
505 contents[i] = c_fill;
506 }
507 else
508 memset (SCM_BYTEVECTOR_CONTENTS (bv), 0, c_len);
509
510 return bv;
511 }
512 #undef FUNC_NAME
513
514 SCM_DEFINE (scm_bytevector_length, "bytevector-length", 1, 0, 0,
515 (SCM bv),
516 "Return the length (in bytes) of @var{bv}.")
517 #define FUNC_NAME s_scm_bytevector_length
518 {
519 return scm_from_uint (scm_c_bytevector_length (bv));
520 }
521 #undef FUNC_NAME
522
523 SCM_DEFINE (scm_bytevector_eq_p, "bytevector=?", 2, 0, 0,
524 (SCM bv1, SCM bv2),
525 "Return is @var{bv1} equals to @var{bv2}---i.e., if they "
526 "have the same length and contents.")
527 #define FUNC_NAME s_scm_bytevector_eq_p
528 {
529 SCM result = SCM_BOOL_F;
530 unsigned c_len1, c_len2;
531
532 SCM_VALIDATE_BYTEVECTOR (1, bv1);
533 SCM_VALIDATE_BYTEVECTOR (2, bv2);
534
535 c_len1 = SCM_BYTEVECTOR_LENGTH (bv1);
536 c_len2 = SCM_BYTEVECTOR_LENGTH (bv2);
537
538 if (c_len1 == c_len2 && (SCM_BYTEVECTOR_ELEMENT_TYPE (bv1)
539 == SCM_BYTEVECTOR_ELEMENT_TYPE (bv2)))
540 {
541 signed char *c_bv1, *c_bv2;
542
543 c_bv1 = SCM_BYTEVECTOR_CONTENTS (bv1);
544 c_bv2 = SCM_BYTEVECTOR_CONTENTS (bv2);
545
546 result = scm_from_bool (!memcmp (c_bv1, c_bv2, c_len1));
547 }
548
549 return result;
550 }
551 #undef FUNC_NAME
552
553 SCM_DEFINE (scm_bytevector_fill_x, "bytevector-fill!", 2, 0, 0,
554 (SCM bv, SCM fill),
555 "Fill bytevector @var{bv} with @var{fill}, a byte.")
556 #define FUNC_NAME s_scm_bytevector_fill_x
557 {
558 unsigned c_len, i;
559 signed char *c_bv, c_fill;
560
561 SCM_VALIDATE_BYTEVECTOR (1, bv);
562 c_fill = scm_to_int8 (fill);
563
564 c_len = SCM_BYTEVECTOR_LENGTH (bv);
565 c_bv = SCM_BYTEVECTOR_CONTENTS (bv);
566
567 for (i = 0; i < c_len; i++)
568 c_bv[i] = c_fill;
569
570 return SCM_UNSPECIFIED;
571 }
572 #undef FUNC_NAME
573
574 SCM_DEFINE (scm_bytevector_copy_x, "bytevector-copy!", 5, 0, 0,
575 (SCM source, SCM source_start, SCM target, SCM target_start,
576 SCM len),
577 "Copy @var{len} bytes from @var{source} into @var{target}, "
578 "starting reading from @var{source_start} (a positive index "
579 "within @var{source}) and start writing at "
580 "@var{target_start}.")
581 #define FUNC_NAME s_scm_bytevector_copy_x
582 {
583 unsigned c_len, c_source_len, c_target_len;
584 unsigned c_source_start, c_target_start;
585 signed char *c_source, *c_target;
586
587 SCM_VALIDATE_BYTEVECTOR (1, source);
588 SCM_VALIDATE_BYTEVECTOR (3, target);
589
590 c_len = scm_to_uint (len);
591 c_source_start = scm_to_uint (source_start);
592 c_target_start = scm_to_uint (target_start);
593
594 c_source = SCM_BYTEVECTOR_CONTENTS (source);
595 c_target = SCM_BYTEVECTOR_CONTENTS (target);
596 c_source_len = SCM_BYTEVECTOR_LENGTH (source);
597 c_target_len = SCM_BYTEVECTOR_LENGTH (target);
598
599 if (SCM_UNLIKELY (c_source_start + c_len > c_source_len))
600 scm_out_of_range (FUNC_NAME, source_start);
601 if (SCM_UNLIKELY (c_target_start + c_len > c_target_len))
602 scm_out_of_range (FUNC_NAME, target_start);
603
604 memmove (c_target + c_target_start,
605 c_source + c_source_start,
606 c_len);
607
608 return SCM_UNSPECIFIED;
609 }
610 #undef FUNC_NAME
611
612 SCM_DEFINE (scm_bytevector_copy, "bytevector-copy", 1, 0, 0,
613 (SCM bv),
614 "Return a newly allocated copy of @var{bv}.")
615 #define FUNC_NAME s_scm_bytevector_copy
616 {
617 SCM copy;
618 unsigned c_len;
619 signed char *c_bv, *c_copy;
620
621 SCM_VALIDATE_BYTEVECTOR (1, bv);
622
623 c_len = SCM_BYTEVECTOR_LENGTH (bv);
624 c_bv = SCM_BYTEVECTOR_CONTENTS (bv);
625
626 copy = make_bytevector (c_len, SCM_BYTEVECTOR_ELEMENT_TYPE (bv));
627 c_copy = SCM_BYTEVECTOR_CONTENTS (copy);
628 memcpy (c_copy, c_bv, c_len);
629
630 return copy;
631 }
632 #undef FUNC_NAME
633
634 SCM_DEFINE (scm_uniform_array_to_bytevector, "uniform-array->bytevector",
635 1, 0, 0, (SCM array),
636 "Return a newly allocated bytevector whose contents\n"
637 "will be copied from the uniform array @var{array}.")
638 #define FUNC_NAME s_scm_uniform_array_to_bytevector
639 {
640 SCM contents, ret;
641 size_t len, sz, byte_len;
642 scm_t_array_handle h;
643 const void *elts;
644
645 contents = scm_array_contents (array, SCM_BOOL_T);
646 if (scm_is_false (contents))
647 scm_wrong_type_arg_msg (FUNC_NAME, 0, array, "uniform contiguous array");
648
649 scm_array_get_handle (contents, &h);
650 assert (h.base == 0);
651
652 elts = h.elements;
653 len = h.dims->inc * (h.dims->ubnd - h.dims->lbnd + 1);
654 sz = scm_array_handle_uniform_element_bit_size (&h);
655 if (sz >= 8 && ((sz % 8) == 0))
656 byte_len = len * (sz / 8);
657 else if (sz < 8)
658 /* Elements of sub-byte size (bitvectors) are addressed in 32-bit
659 units. */
660 byte_len = ((len * sz + 31) / 32) * 4;
661 else
662 /* an internal guile error, really */
663 SCM_MISC_ERROR ("uniform elements larger than 8 bits must fill whole bytes", SCM_EOL);
664
665 ret = make_bytevector (byte_len, SCM_ARRAY_ELEMENT_TYPE_VU8);
666 memcpy (SCM_BYTEVECTOR_CONTENTS (ret), elts, byte_len);
667
668 scm_array_handle_release (&h);
669
670 return ret;
671 }
672 #undef FUNC_NAME
673
674 \f
675 /* Operations on bytes and octets. */
676
677 SCM_DEFINE (scm_bytevector_u8_ref, "bytevector-u8-ref", 2, 0, 0,
678 (SCM bv, SCM index),
679 "Return the octet located at @var{index} in @var{bv}.")
680 #define FUNC_NAME s_scm_bytevector_u8_ref
681 {
682 INTEGER_NATIVE_REF (8, unsigned);
683 }
684 #undef FUNC_NAME
685
686 SCM_DEFINE (scm_bytevector_s8_ref, "bytevector-s8-ref", 2, 0, 0,
687 (SCM bv, SCM index),
688 "Return the byte located at @var{index} in @var{bv}.")
689 #define FUNC_NAME s_scm_bytevector_s8_ref
690 {
691 INTEGER_NATIVE_REF (8, signed);
692 }
693 #undef FUNC_NAME
694
695 SCM_DEFINE (scm_bytevector_u8_set_x, "bytevector-u8-set!", 3, 0, 0,
696 (SCM bv, SCM index, SCM value),
697 "Return the octet located at @var{index} in @var{bv}.")
698 #define FUNC_NAME s_scm_bytevector_u8_set_x
699 {
700 INTEGER_NATIVE_SET (8, unsigned);
701 }
702 #undef FUNC_NAME
703
704 SCM_DEFINE (scm_bytevector_s8_set_x, "bytevector-s8-set!", 3, 0, 0,
705 (SCM bv, SCM index, SCM value),
706 "Return the octet located at @var{index} in @var{bv}.")
707 #define FUNC_NAME s_scm_bytevector_s8_set_x
708 {
709 INTEGER_NATIVE_SET (8, signed);
710 }
711 #undef FUNC_NAME
712
713 #undef OCTET_ACCESSOR_PROLOGUE
714
715
716 SCM_DEFINE (scm_bytevector_to_u8_list, "bytevector->u8-list", 1, 0, 0,
717 (SCM bv),
718 "Return a newly allocated list of octets containing the "
719 "contents of @var{bv}.")
720 #define FUNC_NAME s_scm_bytevector_to_u8_list
721 {
722 SCM lst, pair;
723 unsigned c_len, i;
724 unsigned char *c_bv;
725
726 SCM_VALIDATE_BYTEVECTOR (1, bv);
727
728 c_len = SCM_BYTEVECTOR_LENGTH (bv);
729 c_bv = (unsigned char *) SCM_BYTEVECTOR_CONTENTS (bv);
730
731 lst = scm_make_list (scm_from_uint (c_len), SCM_UNSPECIFIED);
732 for (i = 0, pair = lst;
733 i < c_len;
734 i++, pair = SCM_CDR (pair))
735 {
736 SCM_SETCAR (pair, SCM_I_MAKINUM (c_bv[i]));
737 }
738
739 return lst;
740 }
741 #undef FUNC_NAME
742
743 SCM_DEFINE (scm_u8_list_to_bytevector, "u8-list->bytevector", 1, 0, 0,
744 (SCM lst),
745 "Turn @var{lst}, a list of octets, into a bytevector.")
746 #define FUNC_NAME s_scm_u8_list_to_bytevector
747 {
748 SCM bv, item;
749 long c_len, i;
750 unsigned char *c_bv;
751
752 SCM_VALIDATE_LIST_COPYLEN (1, lst, c_len);
753
754 bv = make_bytevector (c_len, SCM_ARRAY_ELEMENT_TYPE_VU8);
755 c_bv = (unsigned char *) SCM_BYTEVECTOR_CONTENTS (bv);
756
757 for (i = 0; i < c_len; lst = SCM_CDR (lst), i++)
758 {
759 item = SCM_CAR (lst);
760
761 if (SCM_LIKELY (SCM_I_INUMP (item)))
762 {
763 scm_t_signed_bits c_item;
764
765 c_item = SCM_I_INUM (item);
766 if (SCM_LIKELY ((c_item >= 0) && (c_item < 256)))
767 c_bv[i] = (unsigned char) c_item;
768 else
769 goto type_error;
770 }
771 else
772 goto type_error;
773 }
774
775 return bv;
776
777 type_error:
778 scm_wrong_type_arg (FUNC_NAME, 1, item);
779
780 return SCM_BOOL_F;
781 }
782 #undef FUNC_NAME
783
784 /* Compute the two's complement of VALUE (a positive integer) on SIZE octets
785 using (2^(SIZE * 8) - VALUE). */
786 static inline void
787 twos_complement (mpz_t value, size_t size)
788 {
789 unsigned long bit_count;
790
791 /* We expect BIT_COUNT to fit in a unsigned long thanks to the range
792 checking on SIZE performed earlier. */
793 bit_count = (unsigned long) size << 3UL;
794
795 if (SCM_LIKELY (bit_count < sizeof (unsigned long)))
796 mpz_ui_sub (value, 1UL << bit_count, value);
797 else
798 {
799 mpz_t max;
800
801 mpz_init (max);
802 mpz_ui_pow_ui (max, 2, bit_count);
803 mpz_sub (value, max, value);
804 mpz_clear (max);
805 }
806 }
807
808 static inline SCM
809 bytevector_large_ref (const char *c_bv, size_t c_size, int signed_p,
810 SCM endianness)
811 {
812 SCM result;
813 mpz_t c_mpz;
814 int c_endianness, negative_p = 0;
815
816 if (signed_p)
817 {
818 if (scm_is_eq (endianness, scm_sym_big))
819 negative_p = c_bv[0] & 0x80;
820 else
821 negative_p = c_bv[c_size - 1] & 0x80;
822 }
823
824 c_endianness = scm_is_eq (endianness, scm_sym_big) ? 1 : -1;
825
826 mpz_init (c_mpz);
827 mpz_import (c_mpz, 1 /* 1 word */, 1 /* word order doesn't matter */,
828 c_size /* word is C_SIZE-byte long */,
829 c_endianness,
830 0 /* nails */, c_bv);
831
832 if (signed_p && negative_p)
833 {
834 twos_complement (c_mpz, c_size);
835 mpz_neg (c_mpz, c_mpz);
836 }
837
838 result = scm_from_mpz (c_mpz);
839 mpz_clear (c_mpz); /* FIXME: Needed? */
840
841 return result;
842 }
843
844 static inline int
845 bytevector_large_set (char *c_bv, size_t c_size, int signed_p,
846 SCM value, SCM endianness)
847 {
848 mpz_t c_mpz;
849 int c_endianness, c_sign, err = 0;
850
851 c_endianness = scm_is_eq (endianness, scm_sym_big) ? 1 : -1;
852
853 mpz_init (c_mpz);
854 scm_to_mpz (value, c_mpz);
855
856 c_sign = mpz_sgn (c_mpz);
857 if (c_sign < 0)
858 {
859 if (SCM_LIKELY (signed_p))
860 {
861 mpz_neg (c_mpz, c_mpz);
862 twos_complement (c_mpz, c_size);
863 }
864 else
865 {
866 err = -1;
867 goto finish;
868 }
869 }
870
871 if (c_sign == 0)
872 /* Zero. */
873 memset (c_bv, 0, c_size);
874 else
875 {
876 size_t word_count, value_size;
877
878 value_size = (mpz_sizeinbase (c_mpz, 2) + (8 * c_size)) / (8 * c_size);
879 if (SCM_UNLIKELY (value_size > c_size))
880 {
881 err = -2;
882 goto finish;
883 }
884
885
886 mpz_export (c_bv, &word_count, 1 /* word order doesn't matter */,
887 c_size, c_endianness,
888 0 /* nails */, c_mpz);
889 if (SCM_UNLIKELY (word_count != 1))
890 /* Shouldn't happen since we already checked with VALUE_SIZE. */
891 abort ();
892 }
893
894 finish:
895 mpz_clear (c_mpz);
896
897 return err;
898 }
899
900 #define GENERIC_INTEGER_ACCESSOR_PROLOGUE(_sign) \
901 unsigned long c_len, c_index, c_size; \
902 char *c_bv; \
903 \
904 SCM_VALIDATE_BYTEVECTOR (1, bv); \
905 c_index = scm_to_ulong (index); \
906 c_size = scm_to_ulong (size); \
907 \
908 c_len = SCM_BYTEVECTOR_LENGTH (bv); \
909 c_bv = (char *) SCM_BYTEVECTOR_CONTENTS (bv); \
910 \
911 /* C_SIZE must have its 3 higher bits set to zero so that \
912 multiplying it by 8 yields a number that fits in an \
913 unsigned long. */ \
914 if (SCM_UNLIKELY ((c_size == 0) || (c_size >= (ULONG_MAX >> 3L)))) \
915 scm_out_of_range (FUNC_NAME, size); \
916 if (SCM_UNLIKELY (c_index + c_size > c_len)) \
917 scm_out_of_range (FUNC_NAME, index);
918
919
920 /* Template of an integer reference function. */
921 #define GENERIC_INTEGER_REF(_sign) \
922 SCM result; \
923 \
924 if (c_size < 3) \
925 { \
926 int swap; \
927 _sign int value; \
928 \
929 swap = !scm_is_eq (endianness, scm_i_native_endianness); \
930 switch (c_size) \
931 { \
932 case 1: \
933 { \
934 _sign char c_value8; \
935 memcpy (&c_value8, c_bv, 1); \
936 value = c_value8; \
937 } \
938 break; \
939 case 2: \
940 { \
941 INT_TYPE (16, _sign) c_value16; \
942 memcpy (&c_value16, c_bv, 2); \
943 if (swap) \
944 value = (INT_TYPE (16, _sign)) bswap_16 (c_value16); \
945 else \
946 value = c_value16; \
947 } \
948 break; \
949 default: \
950 abort (); \
951 } \
952 \
953 result = SCM_I_MAKINUM ((_sign int) value); \
954 } \
955 else \
956 result = bytevector_large_ref ((char *) c_bv, \
957 c_size, SIGNEDNESS (_sign), \
958 endianness); \
959 \
960 return result;
961
962 static inline SCM
963 bytevector_signed_ref (const char *c_bv, size_t c_size, SCM endianness)
964 {
965 GENERIC_INTEGER_REF (signed);
966 }
967
968 static inline SCM
969 bytevector_unsigned_ref (const char *c_bv, size_t c_size, SCM endianness)
970 {
971 GENERIC_INTEGER_REF (unsigned);
972 }
973
974
975 /* Template of an integer assignment function. */
976 #define GENERIC_INTEGER_SET(_sign) \
977 if (c_size < 3) \
978 { \
979 scm_t_signed_bits c_value; \
980 \
981 if (SCM_UNLIKELY (!SCM_I_INUMP (value))) \
982 goto range_error; \
983 \
984 c_value = SCM_I_INUM (value); \
985 switch (c_size) \
986 { \
987 case 1: \
988 if (SCM_LIKELY (INT_VALID_P (8, _sign) (c_value))) \
989 { \
990 _sign char c_value8; \
991 c_value8 = (_sign char) c_value; \
992 memcpy (c_bv, &c_value8, 1); \
993 } \
994 else \
995 goto range_error; \
996 break; \
997 \
998 case 2: \
999 if (SCM_LIKELY (INT_VALID_P (16, _sign) (c_value))) \
1000 { \
1001 int swap; \
1002 INT_TYPE (16, _sign) c_value16; \
1003 \
1004 swap = !scm_is_eq (endianness, scm_i_native_endianness); \
1005 \
1006 if (swap) \
1007 c_value16 = (INT_TYPE (16, _sign)) bswap_16 (c_value); \
1008 else \
1009 c_value16 = c_value; \
1010 \
1011 memcpy (c_bv, &c_value16, 2); \
1012 } \
1013 else \
1014 goto range_error; \
1015 break; \
1016 \
1017 default: \
1018 abort (); \
1019 } \
1020 } \
1021 else \
1022 { \
1023 int err; \
1024 \
1025 err = bytevector_large_set (c_bv, c_size, \
1026 SIGNEDNESS (_sign), \
1027 value, endianness); \
1028 if (err) \
1029 goto range_error; \
1030 } \
1031 \
1032 return; \
1033 \
1034 range_error: \
1035 scm_out_of_range (FUNC_NAME, value); \
1036 return;
1037
1038 static inline void
1039 bytevector_signed_set (char *c_bv, size_t c_size,
1040 SCM value, SCM endianness,
1041 const char *func_name)
1042 #define FUNC_NAME func_name
1043 {
1044 GENERIC_INTEGER_SET (signed);
1045 }
1046 #undef FUNC_NAME
1047
1048 static inline void
1049 bytevector_unsigned_set (char *c_bv, size_t c_size,
1050 SCM value, SCM endianness,
1051 const char *func_name)
1052 #define FUNC_NAME func_name
1053 {
1054 GENERIC_INTEGER_SET (unsigned);
1055 }
1056 #undef FUNC_NAME
1057
1058 #undef GENERIC_INTEGER_SET
1059 #undef GENERIC_INTEGER_REF
1060
1061
1062 SCM_DEFINE (scm_bytevector_uint_ref, "bytevector-uint-ref", 4, 0, 0,
1063 (SCM bv, SCM index, SCM endianness, SCM size),
1064 "Return the @var{size}-octet long unsigned integer at index "
1065 "@var{index} in @var{bv}.")
1066 #define FUNC_NAME s_scm_bytevector_uint_ref
1067 {
1068 GENERIC_INTEGER_ACCESSOR_PROLOGUE (unsigned);
1069
1070 return (bytevector_unsigned_ref (&c_bv[c_index], c_size, endianness));
1071 }
1072 #undef FUNC_NAME
1073
1074 SCM_DEFINE (scm_bytevector_sint_ref, "bytevector-sint-ref", 4, 0, 0,
1075 (SCM bv, SCM index, SCM endianness, SCM size),
1076 "Return the @var{size}-octet long unsigned integer at index "
1077 "@var{index} in @var{bv}.")
1078 #define FUNC_NAME s_scm_bytevector_sint_ref
1079 {
1080 GENERIC_INTEGER_ACCESSOR_PROLOGUE (signed);
1081
1082 return (bytevector_signed_ref (&c_bv[c_index], c_size, endianness));
1083 }
1084 #undef FUNC_NAME
1085
1086 SCM_DEFINE (scm_bytevector_uint_set_x, "bytevector-uint-set!", 5, 0, 0,
1087 (SCM bv, SCM index, SCM value, SCM endianness, SCM size),
1088 "Set the @var{size}-octet long unsigned integer at @var{index} "
1089 "to @var{value}.")
1090 #define FUNC_NAME s_scm_bytevector_uint_set_x
1091 {
1092 GENERIC_INTEGER_ACCESSOR_PROLOGUE (unsigned);
1093
1094 bytevector_unsigned_set (&c_bv[c_index], c_size, value, endianness,
1095 FUNC_NAME);
1096
1097 return SCM_UNSPECIFIED;
1098 }
1099 #undef FUNC_NAME
1100
1101 SCM_DEFINE (scm_bytevector_sint_set_x, "bytevector-sint-set!", 5, 0, 0,
1102 (SCM bv, SCM index, SCM value, SCM endianness, SCM size),
1103 "Set the @var{size}-octet long signed integer at @var{index} "
1104 "to @var{value}.")
1105 #define FUNC_NAME s_scm_bytevector_sint_set_x
1106 {
1107 GENERIC_INTEGER_ACCESSOR_PROLOGUE (signed);
1108
1109 bytevector_signed_set (&c_bv[c_index], c_size, value, endianness,
1110 FUNC_NAME);
1111
1112 return SCM_UNSPECIFIED;
1113 }
1114 #undef FUNC_NAME
1115
1116
1117 \f
1118 /* Operations on integers of arbitrary size. */
1119
1120 #define INTEGERS_TO_LIST(_sign) \
1121 SCM lst, pair; \
1122 size_t i, c_len, c_size; \
1123 \
1124 SCM_VALIDATE_BYTEVECTOR (1, bv); \
1125 SCM_VALIDATE_SYMBOL (2, endianness); \
1126 c_size = scm_to_unsigned_integer (size, 1, (size_t) -1); \
1127 \
1128 c_len = SCM_BYTEVECTOR_LENGTH (bv); \
1129 if (SCM_UNLIKELY (c_len % c_size != 0)) \
1130 scm_wrong_type_arg_msg \
1131 (FUNC_NAME, 0, size, \
1132 "an exact positive integer that divides the bytevector length"); \
1133 else if (SCM_UNLIKELY (c_len == 0)) \
1134 lst = SCM_EOL; \
1135 else \
1136 { \
1137 const char *c_bv; \
1138 \
1139 c_bv = (char *) SCM_BYTEVECTOR_CONTENTS (bv); \
1140 \
1141 lst = scm_make_list (scm_from_size_t (c_len / c_size), \
1142 SCM_UNSPECIFIED); \
1143 for (i = 0, pair = lst; \
1144 i <= c_len - c_size; \
1145 i += c_size, c_bv += c_size, pair = SCM_CDR (pair)) \
1146 { \
1147 SCM_SETCAR (pair, \
1148 bytevector_ ## _sign ## _ref (c_bv, c_size, \
1149 endianness)); \
1150 } \
1151 } \
1152 \
1153 return lst;
1154
1155 SCM_DEFINE (scm_bytevector_to_sint_list, "bytevector->sint-list",
1156 3, 0, 0,
1157 (SCM bv, SCM endianness, SCM size),
1158 "Return a list of signed integers of @var{size} octets "
1159 "representing the contents of @var{bv}.")
1160 #define FUNC_NAME s_scm_bytevector_to_sint_list
1161 {
1162 INTEGERS_TO_LIST (signed);
1163 }
1164 #undef FUNC_NAME
1165
1166 SCM_DEFINE (scm_bytevector_to_uint_list, "bytevector->uint-list",
1167 3, 0, 0,
1168 (SCM bv, SCM endianness, SCM size),
1169 "Return a list of unsigned integers of @var{size} octets "
1170 "representing the contents of @var{bv}.")
1171 #define FUNC_NAME s_scm_bytevector_to_uint_list
1172 {
1173 INTEGERS_TO_LIST (unsigned);
1174 }
1175 #undef FUNC_NAME
1176
1177 #undef INTEGER_TO_LIST
1178
1179
1180 #define INTEGER_LIST_TO_BYTEVECTOR(_sign) \
1181 SCM bv; \
1182 long c_len; \
1183 size_t c_size; \
1184 char *c_bv, *c_bv_ptr; \
1185 \
1186 SCM_VALIDATE_LIST_COPYLEN (1, lst, c_len); \
1187 SCM_VALIDATE_SYMBOL (2, endianness); \
1188 c_size = scm_to_uint (size); \
1189 \
1190 if (SCM_UNLIKELY ((c_size == 0) || (c_size >= (ULONG_MAX >> 3L)))) \
1191 scm_out_of_range (FUNC_NAME, size); \
1192 \
1193 bv = make_bytevector (c_len * c_size, SCM_ARRAY_ELEMENT_TYPE_VU8); \
1194 c_bv = (char *) SCM_BYTEVECTOR_CONTENTS (bv); \
1195 \
1196 for (c_bv_ptr = c_bv; \
1197 !scm_is_null (lst); \
1198 lst = SCM_CDR (lst), c_bv_ptr += c_size) \
1199 { \
1200 bytevector_ ## _sign ## _set (c_bv_ptr, c_size, \
1201 SCM_CAR (lst), endianness, \
1202 FUNC_NAME); \
1203 } \
1204 \
1205 return bv;
1206
1207
1208 SCM_DEFINE (scm_uint_list_to_bytevector, "uint-list->bytevector",
1209 3, 0, 0,
1210 (SCM lst, SCM endianness, SCM size),
1211 "Return a bytevector containing the unsigned integers "
1212 "listed in @var{lst} and encoded on @var{size} octets "
1213 "according to @var{endianness}.")
1214 #define FUNC_NAME s_scm_uint_list_to_bytevector
1215 {
1216 INTEGER_LIST_TO_BYTEVECTOR (unsigned);
1217 }
1218 #undef FUNC_NAME
1219
1220 SCM_DEFINE (scm_sint_list_to_bytevector, "sint-list->bytevector",
1221 3, 0, 0,
1222 (SCM lst, SCM endianness, SCM size),
1223 "Return a bytevector containing the signed integers "
1224 "listed in @var{lst} and encoded on @var{size} octets "
1225 "according to @var{endianness}.")
1226 #define FUNC_NAME s_scm_sint_list_to_bytevector
1227 {
1228 INTEGER_LIST_TO_BYTEVECTOR (signed);
1229 }
1230 #undef FUNC_NAME
1231
1232 #undef INTEGER_LIST_TO_BYTEVECTOR
1233
1234
1235 \f
1236 /* Operations on 16-bit integers. */
1237
1238 SCM_DEFINE (scm_bytevector_u16_ref, "bytevector-u16-ref",
1239 3, 0, 0,
1240 (SCM bv, SCM index, SCM endianness),
1241 "Return the unsigned 16-bit integer from @var{bv} at "
1242 "@var{index}.")
1243 #define FUNC_NAME s_scm_bytevector_u16_ref
1244 {
1245 INTEGER_REF (16, unsigned);
1246 }
1247 #undef FUNC_NAME
1248
1249 SCM_DEFINE (scm_bytevector_s16_ref, "bytevector-s16-ref",
1250 3, 0, 0,
1251 (SCM bv, SCM index, SCM endianness),
1252 "Return the signed 16-bit integer from @var{bv} at "
1253 "@var{index}.")
1254 #define FUNC_NAME s_scm_bytevector_s16_ref
1255 {
1256 INTEGER_REF (16, signed);
1257 }
1258 #undef FUNC_NAME
1259
1260 SCM_DEFINE (scm_bytevector_u16_native_ref, "bytevector-u16-native-ref",
1261 2, 0, 0,
1262 (SCM bv, SCM index),
1263 "Return the unsigned 16-bit integer from @var{bv} at "
1264 "@var{index} using the native endianness.")
1265 #define FUNC_NAME s_scm_bytevector_u16_native_ref
1266 {
1267 INTEGER_NATIVE_REF (16, unsigned);
1268 }
1269 #undef FUNC_NAME
1270
1271 SCM_DEFINE (scm_bytevector_s16_native_ref, "bytevector-s16-native-ref",
1272 2, 0, 0,
1273 (SCM bv, SCM index),
1274 "Return the unsigned 16-bit integer from @var{bv} at "
1275 "@var{index} using the native endianness.")
1276 #define FUNC_NAME s_scm_bytevector_s16_native_ref
1277 {
1278 INTEGER_NATIVE_REF (16, signed);
1279 }
1280 #undef FUNC_NAME
1281
1282 SCM_DEFINE (scm_bytevector_u16_set_x, "bytevector-u16-set!",
1283 4, 0, 0,
1284 (SCM bv, SCM index, SCM value, SCM endianness),
1285 "Store @var{value} in @var{bv} at @var{index} according to "
1286 "@var{endianness}.")
1287 #define FUNC_NAME s_scm_bytevector_u16_set_x
1288 {
1289 INTEGER_SET (16, unsigned);
1290 }
1291 #undef FUNC_NAME
1292
1293 SCM_DEFINE (scm_bytevector_s16_set_x, "bytevector-s16-set!",
1294 4, 0, 0,
1295 (SCM bv, SCM index, SCM value, SCM endianness),
1296 "Store @var{value} in @var{bv} at @var{index} according to "
1297 "@var{endianness}.")
1298 #define FUNC_NAME s_scm_bytevector_s16_set_x
1299 {
1300 INTEGER_SET (16, signed);
1301 }
1302 #undef FUNC_NAME
1303
1304 SCM_DEFINE (scm_bytevector_u16_native_set_x, "bytevector-u16-native-set!",
1305 3, 0, 0,
1306 (SCM bv, SCM index, SCM value),
1307 "Store the unsigned integer @var{value} at index @var{index} "
1308 "of @var{bv} using the native endianness.")
1309 #define FUNC_NAME s_scm_bytevector_u16_native_set_x
1310 {
1311 INTEGER_NATIVE_SET (16, unsigned);
1312 }
1313 #undef FUNC_NAME
1314
1315 SCM_DEFINE (scm_bytevector_s16_native_set_x, "bytevector-s16-native-set!",
1316 3, 0, 0,
1317 (SCM bv, SCM index, SCM value),
1318 "Store the signed integer @var{value} at index @var{index} "
1319 "of @var{bv} using the native endianness.")
1320 #define FUNC_NAME s_scm_bytevector_s16_native_set_x
1321 {
1322 INTEGER_NATIVE_SET (16, signed);
1323 }
1324 #undef FUNC_NAME
1325
1326
1327 \f
1328 /* Operations on 32-bit integers. */
1329
1330 /* Unfortunately, on 32-bit machines `SCM' is not large enough to hold
1331 arbitrary 32-bit integers. Thus we fall back to using the
1332 `large_{ref,set}' variants on 32-bit machines. */
1333
1334 #define LARGE_INTEGER_REF(_len, _sign) \
1335 INTEGER_ACCESSOR_PROLOGUE(_len, _sign); \
1336 SCM_VALIDATE_SYMBOL (3, endianness); \
1337 \
1338 return (bytevector_large_ref ((char *) c_bv + c_index, _len / 8, \
1339 SIGNEDNESS (_sign), endianness));
1340
1341 #define LARGE_INTEGER_SET(_len, _sign) \
1342 int err; \
1343 INTEGER_ACCESSOR_PROLOGUE (_len, _sign); \
1344 SCM_VALIDATE_SYMBOL (4, endianness); \
1345 \
1346 err = bytevector_large_set ((char *) c_bv + c_index, _len / 8, \
1347 SIGNEDNESS (_sign), value, endianness); \
1348 if (SCM_UNLIKELY (err)) \
1349 scm_out_of_range (FUNC_NAME, value); \
1350 \
1351 return SCM_UNSPECIFIED;
1352
1353 #define LARGE_INTEGER_NATIVE_REF(_len, _sign) \
1354 INTEGER_ACCESSOR_PROLOGUE(_len, _sign); \
1355 return (bytevector_large_ref ((char *) c_bv + c_index, _len / 8, \
1356 SIGNEDNESS (_sign), scm_i_native_endianness));
1357
1358 #define LARGE_INTEGER_NATIVE_SET(_len, _sign) \
1359 int err; \
1360 INTEGER_ACCESSOR_PROLOGUE (_len, _sign); \
1361 \
1362 err = bytevector_large_set ((char *) c_bv + c_index, _len / 8, \
1363 SIGNEDNESS (_sign), value, \
1364 scm_i_native_endianness); \
1365 if (SCM_UNLIKELY (err)) \
1366 scm_out_of_range (FUNC_NAME, value); \
1367 \
1368 return SCM_UNSPECIFIED;
1369
1370
1371 SCM_DEFINE (scm_bytevector_u32_ref, "bytevector-u32-ref",
1372 3, 0, 0,
1373 (SCM bv, SCM index, SCM endianness),
1374 "Return the unsigned 32-bit integer from @var{bv} at "
1375 "@var{index}.")
1376 #define FUNC_NAME s_scm_bytevector_u32_ref
1377 {
1378 #if SIZEOF_VOID_P > 4
1379 INTEGER_REF (32, unsigned);
1380 #else
1381 LARGE_INTEGER_REF (32, unsigned);
1382 #endif
1383 }
1384 #undef FUNC_NAME
1385
1386 SCM_DEFINE (scm_bytevector_s32_ref, "bytevector-s32-ref",
1387 3, 0, 0,
1388 (SCM bv, SCM index, SCM endianness),
1389 "Return the signed 32-bit integer from @var{bv} at "
1390 "@var{index}.")
1391 #define FUNC_NAME s_scm_bytevector_s32_ref
1392 {
1393 #if SIZEOF_VOID_P > 4
1394 INTEGER_REF (32, signed);
1395 #else
1396 LARGE_INTEGER_REF (32, signed);
1397 #endif
1398 }
1399 #undef FUNC_NAME
1400
1401 SCM_DEFINE (scm_bytevector_u32_native_ref, "bytevector-u32-native-ref",
1402 2, 0, 0,
1403 (SCM bv, SCM index),
1404 "Return the unsigned 32-bit integer from @var{bv} at "
1405 "@var{index} using the native endianness.")
1406 #define FUNC_NAME s_scm_bytevector_u32_native_ref
1407 {
1408 #if SIZEOF_VOID_P > 4
1409 INTEGER_NATIVE_REF (32, unsigned);
1410 #else
1411 LARGE_INTEGER_NATIVE_REF (32, unsigned);
1412 #endif
1413 }
1414 #undef FUNC_NAME
1415
1416 SCM_DEFINE (scm_bytevector_s32_native_ref, "bytevector-s32-native-ref",
1417 2, 0, 0,
1418 (SCM bv, SCM index),
1419 "Return the unsigned 32-bit integer from @var{bv} at "
1420 "@var{index} using the native endianness.")
1421 #define FUNC_NAME s_scm_bytevector_s32_native_ref
1422 {
1423 #if SIZEOF_VOID_P > 4
1424 INTEGER_NATIVE_REF (32, signed);
1425 #else
1426 LARGE_INTEGER_NATIVE_REF (32, signed);
1427 #endif
1428 }
1429 #undef FUNC_NAME
1430
1431 SCM_DEFINE (scm_bytevector_u32_set_x, "bytevector-u32-set!",
1432 4, 0, 0,
1433 (SCM bv, SCM index, SCM value, SCM endianness),
1434 "Store @var{value} in @var{bv} at @var{index} according to "
1435 "@var{endianness}.")
1436 #define FUNC_NAME s_scm_bytevector_u32_set_x
1437 {
1438 #if SIZEOF_VOID_P > 4
1439 INTEGER_SET (32, unsigned);
1440 #else
1441 LARGE_INTEGER_SET (32, unsigned);
1442 #endif
1443 }
1444 #undef FUNC_NAME
1445
1446 SCM_DEFINE (scm_bytevector_s32_set_x, "bytevector-s32-set!",
1447 4, 0, 0,
1448 (SCM bv, SCM index, SCM value, SCM endianness),
1449 "Store @var{value} in @var{bv} at @var{index} according to "
1450 "@var{endianness}.")
1451 #define FUNC_NAME s_scm_bytevector_s32_set_x
1452 {
1453 #if SIZEOF_VOID_P > 4
1454 INTEGER_SET (32, signed);
1455 #else
1456 LARGE_INTEGER_SET (32, signed);
1457 #endif
1458 }
1459 #undef FUNC_NAME
1460
1461 SCM_DEFINE (scm_bytevector_u32_native_set_x, "bytevector-u32-native-set!",
1462 3, 0, 0,
1463 (SCM bv, SCM index, SCM value),
1464 "Store the unsigned integer @var{value} at index @var{index} "
1465 "of @var{bv} using the native endianness.")
1466 #define FUNC_NAME s_scm_bytevector_u32_native_set_x
1467 {
1468 #if SIZEOF_VOID_P > 4
1469 INTEGER_NATIVE_SET (32, unsigned);
1470 #else
1471 LARGE_INTEGER_NATIVE_SET (32, unsigned);
1472 #endif
1473 }
1474 #undef FUNC_NAME
1475
1476 SCM_DEFINE (scm_bytevector_s32_native_set_x, "bytevector-s32-native-set!",
1477 3, 0, 0,
1478 (SCM bv, SCM index, SCM value),
1479 "Store the signed integer @var{value} at index @var{index} "
1480 "of @var{bv} using the native endianness.")
1481 #define FUNC_NAME s_scm_bytevector_s32_native_set_x
1482 {
1483 #if SIZEOF_VOID_P > 4
1484 INTEGER_NATIVE_SET (32, signed);
1485 #else
1486 LARGE_INTEGER_NATIVE_SET (32, signed);
1487 #endif
1488 }
1489 #undef FUNC_NAME
1490
1491
1492 \f
1493 /* Operations on 64-bit integers. */
1494
1495 /* For 64-bit integers, we use only the `large_{ref,set}' variant. */
1496
1497 SCM_DEFINE (scm_bytevector_u64_ref, "bytevector-u64-ref",
1498 3, 0, 0,
1499 (SCM bv, SCM index, SCM endianness),
1500 "Return the unsigned 64-bit integer from @var{bv} at "
1501 "@var{index}.")
1502 #define FUNC_NAME s_scm_bytevector_u64_ref
1503 {
1504 LARGE_INTEGER_REF (64, unsigned);
1505 }
1506 #undef FUNC_NAME
1507
1508 SCM_DEFINE (scm_bytevector_s64_ref, "bytevector-s64-ref",
1509 3, 0, 0,
1510 (SCM bv, SCM index, SCM endianness),
1511 "Return the signed 64-bit integer from @var{bv} at "
1512 "@var{index}.")
1513 #define FUNC_NAME s_scm_bytevector_s64_ref
1514 {
1515 LARGE_INTEGER_REF (64, signed);
1516 }
1517 #undef FUNC_NAME
1518
1519 SCM_DEFINE (scm_bytevector_u64_native_ref, "bytevector-u64-native-ref",
1520 2, 0, 0,
1521 (SCM bv, SCM index),
1522 "Return the unsigned 64-bit integer from @var{bv} at "
1523 "@var{index} using the native endianness.")
1524 #define FUNC_NAME s_scm_bytevector_u64_native_ref
1525 {
1526 LARGE_INTEGER_NATIVE_REF (64, unsigned);
1527 }
1528 #undef FUNC_NAME
1529
1530 SCM_DEFINE (scm_bytevector_s64_native_ref, "bytevector-s64-native-ref",
1531 2, 0, 0,
1532 (SCM bv, SCM index),
1533 "Return the unsigned 64-bit integer from @var{bv} at "
1534 "@var{index} using the native endianness.")
1535 #define FUNC_NAME s_scm_bytevector_s64_native_ref
1536 {
1537 LARGE_INTEGER_NATIVE_REF (64, signed);
1538 }
1539 #undef FUNC_NAME
1540
1541 SCM_DEFINE (scm_bytevector_u64_set_x, "bytevector-u64-set!",
1542 4, 0, 0,
1543 (SCM bv, SCM index, SCM value, SCM endianness),
1544 "Store @var{value} in @var{bv} at @var{index} according to "
1545 "@var{endianness}.")
1546 #define FUNC_NAME s_scm_bytevector_u64_set_x
1547 {
1548 LARGE_INTEGER_SET (64, unsigned);
1549 }
1550 #undef FUNC_NAME
1551
1552 SCM_DEFINE (scm_bytevector_s64_set_x, "bytevector-s64-set!",
1553 4, 0, 0,
1554 (SCM bv, SCM index, SCM value, SCM endianness),
1555 "Store @var{value} in @var{bv} at @var{index} according to "
1556 "@var{endianness}.")
1557 #define FUNC_NAME s_scm_bytevector_s64_set_x
1558 {
1559 LARGE_INTEGER_SET (64, signed);
1560 }
1561 #undef FUNC_NAME
1562
1563 SCM_DEFINE (scm_bytevector_u64_native_set_x, "bytevector-u64-native-set!",
1564 3, 0, 0,
1565 (SCM bv, SCM index, SCM value),
1566 "Store the unsigned integer @var{value} at index @var{index} "
1567 "of @var{bv} using the native endianness.")
1568 #define FUNC_NAME s_scm_bytevector_u64_native_set_x
1569 {
1570 LARGE_INTEGER_NATIVE_SET (64, unsigned);
1571 }
1572 #undef FUNC_NAME
1573
1574 SCM_DEFINE (scm_bytevector_s64_native_set_x, "bytevector-s64-native-set!",
1575 3, 0, 0,
1576 (SCM bv, SCM index, SCM value),
1577 "Store the signed integer @var{value} at index @var{index} "
1578 "of @var{bv} using the native endianness.")
1579 #define FUNC_NAME s_scm_bytevector_s64_native_set_x
1580 {
1581 LARGE_INTEGER_NATIVE_SET (64, signed);
1582 }
1583 #undef FUNC_NAME
1584
1585
1586 \f
1587 /* Operations on IEEE-754 numbers. */
1588
1589 /* There are two possible word endians, visible in glibc's <ieee754.h>.
1590 However, in R6RS, when the endianness is `little', little endian is
1591 assumed for both the byte order and the word order. This is clear from
1592 Section 2.1 of R6RS-lib (in response to
1593 http://www.r6rs.org/formal-comments/comment-187.txt). */
1594
1595 union scm_ieee754_float
1596 {
1597 float f;
1598 scm_t_uint32 i;
1599 };
1600
1601 union scm_ieee754_double
1602 {
1603 double d;
1604 scm_t_uint64 i;
1605 };
1606
1607
1608 /* Convert to/from a floating-point number with different endianness. This
1609 method is probably not the most efficient but it should be portable. */
1610
1611 static inline void
1612 float_to_foreign_endianness (union scm_ieee754_float *target,
1613 float source)
1614 {
1615 union scm_ieee754_float input;
1616
1617 input.f = source;
1618 target->i = bswap_32 (input.i);
1619 }
1620
1621 static inline float
1622 float_from_foreign_endianness (const union scm_ieee754_float *source)
1623 {
1624 union scm_ieee754_float result;
1625
1626 result.i = bswap_32 (source->i);
1627
1628 return (result.f);
1629 }
1630
1631 static inline void
1632 double_to_foreign_endianness (union scm_ieee754_double *target,
1633 double source)
1634 {
1635 union scm_ieee754_double input;
1636
1637 input.d = source;
1638 target->i = bswap_64 (input.i);
1639 }
1640
1641 static inline double
1642 double_from_foreign_endianness (const union scm_ieee754_double *source)
1643 {
1644 union scm_ieee754_double result;
1645
1646 result.i = bswap_64 (source->i);
1647
1648 return (result.d);
1649 }
1650
1651 /* Template macros to abstract over doubles and floats.
1652 XXX: Guile can only convert to/from doubles. */
1653 #define IEEE754_UNION(_c_type) union scm_ieee754_ ## _c_type
1654 #define IEEE754_TO_SCM(_c_type) scm_from_double
1655 #define IEEE754_FROM_SCM(_c_type) scm_to_double
1656 #define IEEE754_FROM_FOREIGN_ENDIANNESS(_c_type) \
1657 _c_type ## _from_foreign_endianness
1658 #define IEEE754_TO_FOREIGN_ENDIANNESS(_c_type) \
1659 _c_type ## _to_foreign_endianness
1660
1661
1662 /* FIXME: SCM_VALIDATE_REAL rejects integers, etc. grrr */
1663 #define VALIDATE_REAL(pos, v) \
1664 do { \
1665 SCM_ASSERT_TYPE (scm_is_real (v), v, pos, FUNC_NAME, "real"); \
1666 } while (0)
1667
1668 /* Templace getters and setters. */
1669
1670 #define IEEE754_ACCESSOR_PROLOGUE(_type) \
1671 INTEGER_ACCESSOR_PROLOGUE (sizeof (_type) << 3UL, signed);
1672
1673 #define IEEE754_REF(_type) \
1674 _type c_result; \
1675 \
1676 IEEE754_ACCESSOR_PROLOGUE (_type); \
1677 SCM_VALIDATE_SYMBOL (3, endianness); \
1678 \
1679 if (scm_is_eq (endianness, scm_i_native_endianness)) \
1680 memcpy (&c_result, &c_bv[c_index], sizeof (c_result)); \
1681 else \
1682 { \
1683 IEEE754_UNION (_type) c_raw; \
1684 \
1685 memcpy (&c_raw, &c_bv[c_index], sizeof (c_raw)); \
1686 c_result = \
1687 IEEE754_FROM_FOREIGN_ENDIANNESS (_type) (&c_raw); \
1688 } \
1689 \
1690 return (IEEE754_TO_SCM (_type) (c_result));
1691
1692 #define IEEE754_NATIVE_REF(_type) \
1693 _type c_result; \
1694 \
1695 IEEE754_ACCESSOR_PROLOGUE (_type); \
1696 \
1697 memcpy (&c_result, &c_bv[c_index], sizeof (c_result)); \
1698 return (IEEE754_TO_SCM (_type) (c_result));
1699
1700 #define IEEE754_SET(_type) \
1701 _type c_value; \
1702 \
1703 IEEE754_ACCESSOR_PROLOGUE (_type); \
1704 VALIDATE_REAL (3, value); \
1705 SCM_VALIDATE_SYMBOL (4, endianness); \
1706 c_value = IEEE754_FROM_SCM (_type) (value); \
1707 \
1708 if (scm_is_eq (endianness, scm_i_native_endianness)) \
1709 memcpy (&c_bv[c_index], &c_value, sizeof (c_value)); \
1710 else \
1711 { \
1712 IEEE754_UNION (_type) c_raw; \
1713 \
1714 IEEE754_TO_FOREIGN_ENDIANNESS (_type) (&c_raw, c_value); \
1715 memcpy (&c_bv[c_index], &c_raw, sizeof (c_raw)); \
1716 } \
1717 \
1718 return SCM_UNSPECIFIED;
1719
1720 #define IEEE754_NATIVE_SET(_type) \
1721 _type c_value; \
1722 \
1723 IEEE754_ACCESSOR_PROLOGUE (_type); \
1724 VALIDATE_REAL (3, value); \
1725 c_value = IEEE754_FROM_SCM (_type) (value); \
1726 \
1727 memcpy (&c_bv[c_index], &c_value, sizeof (c_value)); \
1728 return SCM_UNSPECIFIED;
1729
1730
1731 /* Single precision. */
1732
1733 SCM_DEFINE (scm_bytevector_ieee_single_ref,
1734 "bytevector-ieee-single-ref",
1735 3, 0, 0,
1736 (SCM bv, SCM index, SCM endianness),
1737 "Return the IEEE-754 single from @var{bv} at "
1738 "@var{index}.")
1739 #define FUNC_NAME s_scm_bytevector_ieee_single_ref
1740 {
1741 IEEE754_REF (float);
1742 }
1743 #undef FUNC_NAME
1744
1745 SCM_DEFINE (scm_bytevector_ieee_single_native_ref,
1746 "bytevector-ieee-single-native-ref",
1747 2, 0, 0,
1748 (SCM bv, SCM index),
1749 "Return the IEEE-754 single from @var{bv} at "
1750 "@var{index} using the native endianness.")
1751 #define FUNC_NAME s_scm_bytevector_ieee_single_native_ref
1752 {
1753 IEEE754_NATIVE_REF (float);
1754 }
1755 #undef FUNC_NAME
1756
1757 SCM_DEFINE (scm_bytevector_ieee_single_set_x,
1758 "bytevector-ieee-single-set!",
1759 4, 0, 0,
1760 (SCM bv, SCM index, SCM value, SCM endianness),
1761 "Store real @var{value} in @var{bv} at @var{index} according to "
1762 "@var{endianness}.")
1763 #define FUNC_NAME s_scm_bytevector_ieee_single_set_x
1764 {
1765 IEEE754_SET (float);
1766 }
1767 #undef FUNC_NAME
1768
1769 SCM_DEFINE (scm_bytevector_ieee_single_native_set_x,
1770 "bytevector-ieee-single-native-set!",
1771 3, 0, 0,
1772 (SCM bv, SCM index, SCM value),
1773 "Store the real @var{value} at index @var{index} "
1774 "of @var{bv} using the native endianness.")
1775 #define FUNC_NAME s_scm_bytevector_ieee_single_native_set_x
1776 {
1777 IEEE754_NATIVE_SET (float);
1778 }
1779 #undef FUNC_NAME
1780
1781
1782 /* Double precision. */
1783
1784 SCM_DEFINE (scm_bytevector_ieee_double_ref,
1785 "bytevector-ieee-double-ref",
1786 3, 0, 0,
1787 (SCM bv, SCM index, SCM endianness),
1788 "Return the IEEE-754 double from @var{bv} at "
1789 "@var{index}.")
1790 #define FUNC_NAME s_scm_bytevector_ieee_double_ref
1791 {
1792 IEEE754_REF (double);
1793 }
1794 #undef FUNC_NAME
1795
1796 SCM_DEFINE (scm_bytevector_ieee_double_native_ref,
1797 "bytevector-ieee-double-native-ref",
1798 2, 0, 0,
1799 (SCM bv, SCM index),
1800 "Return the IEEE-754 double from @var{bv} at "
1801 "@var{index} using the native endianness.")
1802 #define FUNC_NAME s_scm_bytevector_ieee_double_native_ref
1803 {
1804 IEEE754_NATIVE_REF (double);
1805 }
1806 #undef FUNC_NAME
1807
1808 SCM_DEFINE (scm_bytevector_ieee_double_set_x,
1809 "bytevector-ieee-double-set!",
1810 4, 0, 0,
1811 (SCM bv, SCM index, SCM value, SCM endianness),
1812 "Store real @var{value} in @var{bv} at @var{index} according to "
1813 "@var{endianness}.")
1814 #define FUNC_NAME s_scm_bytevector_ieee_double_set_x
1815 {
1816 IEEE754_SET (double);
1817 }
1818 #undef FUNC_NAME
1819
1820 SCM_DEFINE (scm_bytevector_ieee_double_native_set_x,
1821 "bytevector-ieee-double-native-set!",
1822 3, 0, 0,
1823 (SCM bv, SCM index, SCM value),
1824 "Store the real @var{value} at index @var{index} "
1825 "of @var{bv} using the native endianness.")
1826 #define FUNC_NAME s_scm_bytevector_ieee_double_native_set_x
1827 {
1828 IEEE754_NATIVE_SET (double);
1829 }
1830 #undef FUNC_NAME
1831
1832
1833 #undef IEEE754_UNION
1834 #undef IEEE754_TO_SCM
1835 #undef IEEE754_FROM_SCM
1836 #undef IEEE754_FROM_FOREIGN_ENDIANNESS
1837 #undef IEEE754_TO_FOREIGN_ENDIANNESS
1838 #undef IEEE754_REF
1839 #undef IEEE754_NATIVE_REF
1840 #undef IEEE754_SET
1841 #undef IEEE754_NATIVE_SET
1842
1843 \f
1844 /* Operations on strings. */
1845
1846
1847 /* Produce a function that returns the length of a UTF-encoded string. */
1848 #define UTF_STRLEN_FUNCTION(_utf_width) \
1849 static inline size_t \
1850 utf ## _utf_width ## _strlen (const uint ## _utf_width ## _t *str) \
1851 { \
1852 size_t len = 0; \
1853 const uint ## _utf_width ## _t *ptr; \
1854 for (ptr = str; \
1855 *ptr != 0; \
1856 ptr++) \
1857 { \
1858 len++; \
1859 } \
1860 \
1861 return (len * ((_utf_width) / 8)); \
1862 }
1863
1864 UTF_STRLEN_FUNCTION (8)
1865
1866
1867 /* Return the length (in bytes) of STR, a UTF-(UTF_WIDTH) encoded string. */
1868 #define UTF_STRLEN(_utf_width, _str) \
1869 utf ## _utf_width ## _strlen (_str)
1870
1871 /* Return the "portable" name of the UTF encoding of size UTF_WIDTH and
1872 ENDIANNESS (Gnulib's `iconv_open' module guarantees the portability of the
1873 encoding name). */
1874 static inline void
1875 utf_encoding_name (char *name, size_t utf_width, SCM endianness)
1876 {
1877 strcpy (name, "UTF-");
1878 strcat (name, ((utf_width == 8)
1879 ? "8"
1880 : ((utf_width == 16)
1881 ? "16"
1882 : ((utf_width == 32)
1883 ? "32"
1884 : "??"))));
1885 strcat (name,
1886 ((scm_is_eq (endianness, scm_sym_big))
1887 ? "BE"
1888 : ((scm_is_eq (endianness, scm_sym_little))
1889 ? "LE"
1890 : "unknown")));
1891 }
1892
1893 /* Maximum length of a UTF encoding name. */
1894 #define MAX_UTF_ENCODING_NAME_LEN 16
1895
1896 /* Produce the body of a `string->utf' function. */
1897 #define STRING_TO_UTF(_utf_width) \
1898 SCM utf; \
1899 int err; \
1900 char c_utf_name[MAX_UTF_ENCODING_NAME_LEN]; \
1901 char *c_utf = NULL; \
1902 size_t c_strlen, c_utf_len = 0; \
1903 \
1904 SCM_VALIDATE_STRING (1, str); \
1905 if (scm_is_eq (endianness, SCM_UNDEFINED)) \
1906 endianness = scm_sym_big; \
1907 else \
1908 SCM_VALIDATE_SYMBOL (2, endianness); \
1909 \
1910 utf_encoding_name (c_utf_name, (_utf_width), endianness); \
1911 \
1912 c_strlen = scm_i_string_length (str); \
1913 if (scm_i_is_narrow_string (str)) \
1914 { \
1915 err = mem_iconveh (scm_i_string_chars (str), c_strlen, \
1916 "ISO-8859-1", c_utf_name, \
1917 iconveh_question_mark, NULL, \
1918 &c_utf, &c_utf_len); \
1919 if (SCM_UNLIKELY (err)) \
1920 scm_syserror_msg (FUNC_NAME, "failed to convert string: ~A", \
1921 scm_list_1 (str), err); \
1922 } \
1923 else \
1924 { \
1925 const scm_t_wchar *wbuf = scm_i_string_wide_chars (str); \
1926 c_utf = u32_conv_to_encoding (c_utf_name, \
1927 iconveh_question_mark, \
1928 (scm_t_uint32 *) wbuf, \
1929 c_strlen, NULL, NULL, &c_utf_len); \
1930 if (SCM_UNLIKELY (c_utf == NULL)) \
1931 scm_syserror_msg (FUNC_NAME, "failed to convert string: ~A", \
1932 scm_list_1 (str), errno); \
1933 } \
1934 scm_dynwind_begin (0); \
1935 scm_dynwind_free (c_utf); \
1936 utf = make_bytevector (c_utf_len, SCM_ARRAY_ELEMENT_TYPE_VU8); \
1937 memcpy (SCM_BYTEVECTOR_CONTENTS (utf), c_utf, c_utf_len); \
1938 scm_dynwind_end (); \
1939 \
1940 return (utf);
1941
1942
1943
1944 SCM_DEFINE (scm_string_to_utf8, "string->utf8",
1945 1, 0, 0,
1946 (SCM str),
1947 "Return a newly allocated bytevector that contains the UTF-8 "
1948 "encoding of @var{str}.")
1949 #define FUNC_NAME s_scm_string_to_utf8
1950 {
1951 SCM utf;
1952 scm_t_uint8 *c_utf;
1953 size_t c_utf_len = 0;
1954
1955 SCM_VALIDATE_STRING (1, str);
1956
1957 c_utf = (scm_t_uint8 *) scm_to_utf8_stringn (str, &c_utf_len);
1958 utf = make_bytevector (c_utf_len, SCM_ARRAY_ELEMENT_TYPE_VU8);
1959 memcpy (SCM_BYTEVECTOR_CONTENTS (utf), c_utf, c_utf_len);
1960 free (c_utf);
1961
1962 return (utf);
1963 }
1964 #undef FUNC_NAME
1965
1966 SCM_DEFINE (scm_string_to_utf16, "string->utf16",
1967 1, 1, 0,
1968 (SCM str, SCM endianness),
1969 "Return a newly allocated bytevector that contains the UTF-16 "
1970 "encoding of @var{str}.")
1971 #define FUNC_NAME s_scm_string_to_utf16
1972 {
1973 STRING_TO_UTF (16);
1974 }
1975 #undef FUNC_NAME
1976
1977 static void
1978 swap_u32 (scm_t_wchar *vals, size_t len)
1979 {
1980 size_t n;
1981 for (n = 0; n < len; n++)
1982 vals[n] = bswap_32 (vals[n]);
1983 }
1984
1985 SCM_DEFINE (scm_string_to_utf32, "string->utf32",
1986 1, 1, 0,
1987 (SCM str, SCM endianness),
1988 "Return a newly allocated bytevector that contains the UTF-32 "
1989 "encoding of @var{str}.")
1990 #define FUNC_NAME s_scm_string_to_utf32
1991 {
1992 SCM bv;
1993 scm_t_wchar *wchars;
1994 size_t wchar_len, bytes_len;
1995
1996 wchars = scm_to_utf32_stringn (str, &wchar_len);
1997 bytes_len = wchar_len * sizeof (scm_t_wchar);
1998 if (!scm_is_eq (SCM_UNBNDP (endianness) ? scm_endianness_big : endianness,
1999 scm_i_native_endianness))
2000 swap_u32 (wchars, wchar_len);
2001
2002 bv = make_bytevector (bytes_len, SCM_ARRAY_ELEMENT_TYPE_VU8);
2003 memcpy (SCM_BYTEVECTOR_CONTENTS (bv), wchars, bytes_len);
2004 free (wchars);
2005
2006 return bv;
2007 }
2008 #undef FUNC_NAME
2009
2010
2011 /* Produce the body of a function that converts a UTF-encoded bytevector to a
2012 string. */
2013 #define UTF_TO_STRING(_utf_width) \
2014 SCM str = SCM_BOOL_F; \
2015 int err; \
2016 char *c_str = NULL; \
2017 char c_utf_name[MAX_UTF_ENCODING_NAME_LEN]; \
2018 char *c_utf; \
2019 size_t c_strlen = 0, c_utf_len = 0; \
2020 \
2021 SCM_VALIDATE_BYTEVECTOR (1, utf); \
2022 if (scm_is_eq (endianness, SCM_UNDEFINED)) \
2023 endianness = scm_sym_big; \
2024 else \
2025 SCM_VALIDATE_SYMBOL (2, endianness); \
2026 \
2027 c_utf_len = SCM_BYTEVECTOR_LENGTH (utf); \
2028 c_utf = (char *) SCM_BYTEVECTOR_CONTENTS (utf); \
2029 utf_encoding_name (c_utf_name, (_utf_width), endianness); \
2030 \
2031 err = mem_iconveh (c_utf, c_utf_len, \
2032 c_utf_name, "UTF-8", \
2033 iconveh_question_mark, NULL, \
2034 &c_str, &c_strlen); \
2035 if (SCM_UNLIKELY (err)) \
2036 scm_syserror_msg (FUNC_NAME, "failed to convert to string: ~A", \
2037 scm_list_1 (utf), err); \
2038 else \
2039 { \
2040 str = scm_from_utf8_stringn (c_str, c_strlen); \
2041 free (c_str); \
2042 } \
2043 return (str);
2044
2045
2046 SCM_DEFINE (scm_utf8_to_string, "utf8->string",
2047 1, 0, 0,
2048 (SCM utf),
2049 "Return a newly allocate string that contains from the UTF-8-"
2050 "encoded contents of bytevector @var{utf}.")
2051 #define FUNC_NAME s_scm_utf8_to_string
2052 {
2053 SCM str;
2054 const char *c_utf;
2055 size_t c_utf_len = 0;
2056
2057 SCM_VALIDATE_BYTEVECTOR (1, utf);
2058
2059 c_utf_len = SCM_BYTEVECTOR_LENGTH (utf);
2060 c_utf = (char *) SCM_BYTEVECTOR_CONTENTS (utf);
2061 str = scm_from_utf8_stringn (c_utf, c_utf_len);
2062
2063 return (str);
2064 }
2065 #undef FUNC_NAME
2066
2067 SCM_DEFINE (scm_utf16_to_string, "utf16->string",
2068 1, 1, 0,
2069 (SCM utf, SCM endianness),
2070 "Return a newly allocate string that contains from the UTF-16-"
2071 "encoded contents of bytevector @var{utf}.")
2072 #define FUNC_NAME s_scm_utf16_to_string
2073 {
2074 UTF_TO_STRING (16);
2075 }
2076 #undef FUNC_NAME
2077
2078 SCM_DEFINE (scm_utf32_to_string, "utf32->string",
2079 1, 1, 0,
2080 (SCM utf, SCM endianness),
2081 "Return a newly allocate string that contains from the UTF-32-"
2082 "encoded contents of bytevector @var{utf}.")
2083 #define FUNC_NAME s_scm_utf32_to_string
2084 {
2085 UTF_TO_STRING (32);
2086 }
2087 #undef FUNC_NAME
2088
2089 \f
2090 /* Initialization. */
2091
2092 void
2093 scm_bootstrap_bytevectors (void)
2094 {
2095 /* This must be instantiated here because the generalized-vector API may
2096 want to access bytevectors even though `(rnrs bytevectors)' hasn't been
2097 loaded. */
2098 scm_null_bytevector = make_bytevector (0, SCM_ARRAY_ELEMENT_TYPE_VU8);
2099
2100 #ifdef WORDS_BIGENDIAN
2101 scm_i_native_endianness = scm_from_latin1_symbol ("big");
2102 #else
2103 scm_i_native_endianness = scm_from_latin1_symbol ("little");
2104 #endif
2105
2106 scm_c_register_extension ("libguile-" SCM_EFFECTIVE_VERSION,
2107 "scm_init_bytevectors",
2108 (scm_t_extension_init_func) scm_init_bytevectors,
2109 NULL);
2110
2111 scm_i_register_vector_constructor
2112 (scm_i_array_element_types[SCM_ARRAY_ELEMENT_TYPE_VU8],
2113 scm_make_bytevector);
2114 }
2115
2116 void
2117 scm_init_bytevectors (void)
2118 {
2119 #include "libguile/bytevectors.x"
2120
2121 scm_endianness_big = scm_sym_big;
2122 scm_endianness_little = scm_sym_little;
2123 }
Unnamed repository; edit this file 'description' to name the repository.