1 | /*
|
---|
2 | ** pack.c - Array#pack, String#unpack
|
---|
3 | */
|
---|
4 |
|
---|
5 | #include "mruby.h"
|
---|
6 | #include "error.h"
|
---|
7 | #include "mruby/array.h"
|
---|
8 | #include "mruby/class.h"
|
---|
9 | #include "mruby/numeric.h"
|
---|
10 | #include "mruby/string.h"
|
---|
11 | #include "mruby/variable.h"
|
---|
12 |
|
---|
13 | #include <ctype.h>
|
---|
14 | #include <errno.h>
|
---|
15 | #include <limits.h>
|
---|
16 | #include <stdio.h>
|
---|
17 | #include <string.h>
|
---|
18 |
|
---|
19 | struct tmpl {
|
---|
20 | mrb_value str;
|
---|
21 | int idx;
|
---|
22 | };
|
---|
23 |
|
---|
24 | enum {
|
---|
25 | PACK_DIR_CHAR, /* C */
|
---|
26 | PACK_DIR_SHORT, /* S */
|
---|
27 | PACK_DIR_LONG, /* L */
|
---|
28 | PACK_DIR_QUAD, /* Q */
|
---|
29 | //PACK_DIR_INT, /* i */
|
---|
30 | //PACK_DIR_VAX,
|
---|
31 | PACK_DIR_UTF8, /* U */
|
---|
32 | //PACK_DIR_BER,
|
---|
33 | PACK_DIR_DOUBLE, /* E */
|
---|
34 | PACK_DIR_FLOAT, /* f */
|
---|
35 | PACK_DIR_STR, /* A */
|
---|
36 | PACK_DIR_HEX, /* h */
|
---|
37 | PACK_DIR_BASE64, /* m */
|
---|
38 | PACK_DIR_NUL, /* x */
|
---|
39 | PACK_DIR_INVALID
|
---|
40 | };
|
---|
41 |
|
---|
42 | enum {
|
---|
43 | PACK_TYPE_INTEGER,
|
---|
44 | PACK_TYPE_FLOAT,
|
---|
45 | PACK_TYPE_STRING,
|
---|
46 | PACK_TYPE_NONE
|
---|
47 | };
|
---|
48 |
|
---|
49 | #define PACK_FLAG_s 0x00000001 /* native size ("_" "!") */
|
---|
50 | #define PACK_FLAG_a 0x00000002 /* null padding ("a") */
|
---|
51 | #define PACK_FLAG_Z 0x00000004 /* append nul char ("z") */
|
---|
52 | #define PACK_FLAG_SIGNED 0x00000008 /* native size ("_" "!") */
|
---|
53 | #define PACK_FLAG_GT 0x00000010 /* big endian (">") */
|
---|
54 | #define PACK_FLAG_LT 0x00000020 /* little endian ("<") */
|
---|
55 | #define PACK_FLAG_WIDTH 0x00000040 /* "count" is "width" */
|
---|
56 | #define PACK_FLAG_LSB 0x00000080 /* LSB / low nibble first */
|
---|
57 | #define PACK_FLAG_COUNT2 0x00000100 /* "count" is special... */
|
---|
58 | #define PACK_FLAG_LITTLEENDIAN 0x00000200 /* little endian actually */
|
---|
59 |
|
---|
60 | #define PACK_BASE64_IGNORE 0xff
|
---|
61 | #define PACK_BASE64_PADDING 0xfe
|
---|
62 |
|
---|
63 | static int littleendian = 0;
|
---|
64 |
|
---|
65 | const static unsigned char base64chars[] =
|
---|
66 | "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
|
---|
67 | static signed char base64_dec_tab[128];
|
---|
68 |
|
---|
69 |
|
---|
70 | static int
|
---|
71 | check_little_endian(void)
|
---|
72 | {
|
---|
73 | unsigned int n = 1;
|
---|
74 | return (*(unsigned char *)&n == 1);
|
---|
75 | }
|
---|
76 |
|
---|
77 | static unsigned int
|
---|
78 | hex2int(unsigned char ch)
|
---|
79 | {
|
---|
80 | if (ch >= '0' && ch <= '9')
|
---|
81 | return ch - '0';
|
---|
82 | else if (ch >= 'A' && ch <= 'F')
|
---|
83 | return 10 + (ch - 'A');
|
---|
84 | else if (ch >= 'a' && ch <= 'f')
|
---|
85 | return 10 + (ch - 'a');
|
---|
86 | else
|
---|
87 | return 0;
|
---|
88 | }
|
---|
89 |
|
---|
90 | static void
|
---|
91 | make_base64_dec_tab(void)
|
---|
92 | {
|
---|
93 | int i;
|
---|
94 | memset(base64_dec_tab, PACK_BASE64_IGNORE, sizeof(base64_dec_tab));
|
---|
95 | for (i = 0; i < 26; i++)
|
---|
96 | base64_dec_tab['A' + i] = i;
|
---|
97 | for (i = 0; i < 26; i++)
|
---|
98 | base64_dec_tab['a' + i] = i + 26;
|
---|
99 | for (i = 0; i < 10; i++)
|
---|
100 | base64_dec_tab['0' + i] = i + 52;
|
---|
101 | base64_dec_tab['+'] = 62;
|
---|
102 | base64_dec_tab['/'] = 63;
|
---|
103 | base64_dec_tab['='] = PACK_BASE64_PADDING;
|
---|
104 | }
|
---|
105 |
|
---|
106 | static mrb_value
|
---|
107 | str_len_ensure(mrb_state *mrb, mrb_value str, int len)
|
---|
108 | {
|
---|
109 | int n = RSTRING_LEN(str);
|
---|
110 | if (len > n) {
|
---|
111 | do {
|
---|
112 | n *= 2;
|
---|
113 | } while (len > n);
|
---|
114 | str = mrb_str_resize(mrb, str, n);
|
---|
115 | }
|
---|
116 | return str;
|
---|
117 | }
|
---|
118 |
|
---|
119 |
|
---|
120 | static int
|
---|
121 | pack_c(mrb_state *mrb, mrb_value o, mrb_value str, mrb_int sidx, unsigned int flags)
|
---|
122 | {
|
---|
123 | str = str_len_ensure(mrb, str, sidx + 1);
|
---|
124 | RSTRING_PTR(str)[sidx] = mrb_fixnum(o);
|
---|
125 | return 1;
|
---|
126 | }
|
---|
127 |
|
---|
128 | static int
|
---|
129 | unpack_c(mrb_state *mrb, const void *src, int srclen, mrb_value ary, unsigned int flags)
|
---|
130 | {
|
---|
131 | if (flags & PACK_FLAG_SIGNED)
|
---|
132 | mrb_ary_push(mrb, ary, mrb_fixnum_value(*(signed char *)src));
|
---|
133 | else
|
---|
134 | mrb_ary_push(mrb, ary, mrb_fixnum_value(*(unsigned char *)src));
|
---|
135 | return 1;
|
---|
136 | }
|
---|
137 |
|
---|
138 | static int
|
---|
139 | pack_s(mrb_state *mrb, mrb_value o, mrb_value str, mrb_int sidx, unsigned int flags)
|
---|
140 | {
|
---|
141 | unsigned short n;
|
---|
142 |
|
---|
143 | str = str_len_ensure(mrb, str, sidx + 2);
|
---|
144 | n = mrb_fixnum(o);
|
---|
145 | if (flags & PACK_FLAG_LITTLEENDIAN) {
|
---|
146 | RSTRING_PTR(str)[sidx+0] = n % 256;
|
---|
147 | RSTRING_PTR(str)[sidx+1] = n / 256;
|
---|
148 | } else {
|
---|
149 | RSTRING_PTR(str)[sidx+0] = n / 256;
|
---|
150 | RSTRING_PTR(str)[sidx+1] = n % 256;
|
---|
151 | }
|
---|
152 | return 2;
|
---|
153 | }
|
---|
154 |
|
---|
155 | static int
|
---|
156 | unpack_s(mrb_state *mrb, const unsigned char *src, int srclen, mrb_value ary, unsigned int flags)
|
---|
157 | {
|
---|
158 | int n;
|
---|
159 |
|
---|
160 | if (flags & PACK_FLAG_LITTLEENDIAN) {
|
---|
161 | n = src[1] * 256 + src[0];
|
---|
162 | } else {
|
---|
163 | n = src[0] * 256 + src[1];
|
---|
164 | }
|
---|
165 | if ((flags & PACK_FLAG_SIGNED) && (n >= 0x8000)) {
|
---|
166 | n -= 0x10000;
|
---|
167 | }
|
---|
168 | mrb_ary_push(mrb, ary, mrb_fixnum_value(n));
|
---|
169 | return 2;
|
---|
170 | }
|
---|
171 |
|
---|
172 | static int
|
---|
173 | pack_l(mrb_state *mrb, mrb_value o, mrb_value str, mrb_int sidx, unsigned int flags)
|
---|
174 | {
|
---|
175 | unsigned long n;
|
---|
176 | str = str_len_ensure(mrb, str, sidx + 4);
|
---|
177 | n = mrb_fixnum(o);
|
---|
178 | if (flags & PACK_FLAG_LITTLEENDIAN) {
|
---|
179 | RSTRING_PTR(str)[sidx+0] = n & 0xff;
|
---|
180 | RSTRING_PTR(str)[sidx+1] = n >> 8;
|
---|
181 | RSTRING_PTR(str)[sidx+2] = n >> 16;
|
---|
182 | RSTRING_PTR(str)[sidx+3] = n >> 24;
|
---|
183 | } else {
|
---|
184 | RSTRING_PTR(str)[sidx+0] = n >> 24;
|
---|
185 | RSTRING_PTR(str)[sidx+1] = n >> 16;
|
---|
186 | RSTRING_PTR(str)[sidx+2] = n >> 8;
|
---|
187 | RSTRING_PTR(str)[sidx+3] = n & 0xff;
|
---|
188 | }
|
---|
189 | return 4;
|
---|
190 | }
|
---|
191 |
|
---|
192 | static int
|
---|
193 | unpack_l(mrb_state *mrb, const unsigned char *src, int srclen, mrb_value ary, unsigned int flags)
|
---|
194 | {
|
---|
195 | char msg[60];
|
---|
196 | uint32_t ul;
|
---|
197 | mrb_int n;
|
---|
198 |
|
---|
199 | if (flags & PACK_FLAG_LITTLEENDIAN) {
|
---|
200 | ul = (uint32_t)src[3] * 256*256*256;
|
---|
201 | ul += (uint32_t)src[2] *256*256;
|
---|
202 | ul += (uint32_t)src[1] *256;
|
---|
203 | ul += (uint32_t)src[0];
|
---|
204 | } else {
|
---|
205 | ul = (uint32_t)src[0] * 256*256*256;
|
---|
206 | ul += (uint32_t)src[1] *256*256;
|
---|
207 | ul += (uint32_t)src[2] *256;
|
---|
208 | ul += (uint32_t)src[3];
|
---|
209 | }
|
---|
210 | if (flags & PACK_FLAG_SIGNED) {
|
---|
211 | int32_t sl = ul;
|
---|
212 | if (!FIXABLE(sl)) {
|
---|
213 | snprintf(msg, sizeof(msg), "cannot unpack to Fixnum: %ld", (long)sl);
|
---|
214 | mrb_raise(mrb, E_RANGE_ERROR, msg);
|
---|
215 | }
|
---|
216 | n = sl;
|
---|
217 | } else {
|
---|
218 | if (!POSFIXABLE(ul)) {
|
---|
219 | snprintf(msg, sizeof(msg), "cannot unpack to Fixnum: %lu", (unsigned long)ul);
|
---|
220 | mrb_raise(mrb, E_RANGE_ERROR, msg);
|
---|
221 | }
|
---|
222 | n = ul;
|
---|
223 | }
|
---|
224 | mrb_ary_push(mrb, ary, mrb_fixnum_value(n));
|
---|
225 | return 4;
|
---|
226 | }
|
---|
227 |
|
---|
228 | static int
|
---|
229 | pack_q(mrb_state *mrb, mrb_value o, mrb_value str, mrb_int sidx, unsigned int flags)
|
---|
230 | {
|
---|
231 | unsigned long long n;
|
---|
232 | str = str_len_ensure(mrb, str, sidx + 8);
|
---|
233 | n = mrb_fixnum(o);
|
---|
234 | if (flags & PACK_FLAG_LITTLEENDIAN) {
|
---|
235 | RSTRING_PTR(str)[sidx+0] = n & 0xff;
|
---|
236 | RSTRING_PTR(str)[sidx+1] = n >> 8;
|
---|
237 | RSTRING_PTR(str)[sidx+2] = n >> 16;
|
---|
238 | RSTRING_PTR(str)[sidx+3] = n >> 24;
|
---|
239 | RSTRING_PTR(str)[sidx+4] = n >> 32;
|
---|
240 | RSTRING_PTR(str)[sidx+5] = n >> 40;
|
---|
241 | RSTRING_PTR(str)[sidx+6] = n >> 48;
|
---|
242 | RSTRING_PTR(str)[sidx+7] = n >> 56;
|
---|
243 | } else {
|
---|
244 | RSTRING_PTR(str)[sidx+0] = n >> 56;
|
---|
245 | RSTRING_PTR(str)[sidx+1] = n >> 48;
|
---|
246 | RSTRING_PTR(str)[sidx+2] = n >> 40;
|
---|
247 | RSTRING_PTR(str)[sidx+3] = n >> 32;
|
---|
248 | RSTRING_PTR(str)[sidx+4] = n >> 24;
|
---|
249 | RSTRING_PTR(str)[sidx+5] = n >> 16;
|
---|
250 | RSTRING_PTR(str)[sidx+6] = n >> 8;
|
---|
251 | RSTRING_PTR(str)[sidx+7] = n & 0xff;
|
---|
252 | }
|
---|
253 | return 8;
|
---|
254 | }
|
---|
255 |
|
---|
256 | static int
|
---|
257 | unpack_q(mrb_state *mrb, const unsigned char *src, int srclen, mrb_value ary, unsigned int flags)
|
---|
258 | {
|
---|
259 | char msg[60];
|
---|
260 | uint64_t ull;
|
---|
261 | int i, pos, step;
|
---|
262 | mrb_int n;
|
---|
263 |
|
---|
264 | if (flags & PACK_FLAG_LITTLEENDIAN) {
|
---|
265 | pos = 7;
|
---|
266 | step = -1;
|
---|
267 | } else {
|
---|
268 | pos = 0;
|
---|
269 | step = 1;
|
---|
270 | }
|
---|
271 | ull = 0;
|
---|
272 | for (i = 0; i < 8; i++) {
|
---|
273 | ull = ull * 256 + (uint64_t)src[pos];
|
---|
274 | pos += step;
|
---|
275 | }
|
---|
276 | if (flags & PACK_FLAG_SIGNED) {
|
---|
277 | int64_t sll = ull;
|
---|
278 | if (!FIXABLE(sll)) {
|
---|
279 | snprintf(msg, sizeof(msg), "cannot unpack to Fixnum: %lld", (long long)sll);
|
---|
280 | mrb_raise(mrb, E_RANGE_ERROR, msg);
|
---|
281 | }
|
---|
282 | n = sll;
|
---|
283 | } else {
|
---|
284 | if (!POSFIXABLE(ull)) {
|
---|
285 | snprintf(msg, sizeof(msg), "cannot unpack to Fixnum: %llu", (unsigned long long)ull);
|
---|
286 | mrb_raise(mrb, E_RANGE_ERROR, msg);
|
---|
287 | }
|
---|
288 | n = ull;
|
---|
289 | }
|
---|
290 | mrb_ary_push(mrb, ary, mrb_fixnum_value(n));
|
---|
291 | return 8;
|
---|
292 | }
|
---|
293 |
|
---|
294 | static int
|
---|
295 | pack_double(mrb_state *mrb, mrb_value o, mrb_value str, mrb_int sidx, unsigned int flags)
|
---|
296 | {
|
---|
297 | int i;
|
---|
298 | double d;
|
---|
299 | uint8_t *buffer = (uint8_t *)&d;
|
---|
300 | str = str_len_ensure(mrb, str, sidx + 8);
|
---|
301 | d = mrb_float(o);
|
---|
302 |
|
---|
303 | if (flags & PACK_FLAG_LITTLEENDIAN) {
|
---|
304 | #ifdef MRB_ENDIAN_BIG
|
---|
305 | for (i = 0; i < 8; ++i) {
|
---|
306 | RSTRING_PTR(str)[sidx + i] = buffer[8 - i - 1];
|
---|
307 | }
|
---|
308 | #else
|
---|
309 | memcpy(RSTRING_PTR(str) + sidx, buffer, 8);
|
---|
310 | #endif
|
---|
311 | } else {
|
---|
312 | #ifdef MRB_ENDIAN_BIG
|
---|
313 | memcpy(RSTRING_PTR(str) + sidx, buffer, 8);
|
---|
314 | #else
|
---|
315 | for (i = 0; i < 8; ++i) {
|
---|
316 | RSTRING_PTR(str)[sidx + i] = buffer[8 - i - 1];
|
---|
317 | }
|
---|
318 | #endif
|
---|
319 | }
|
---|
320 |
|
---|
321 | return 8;
|
---|
322 | }
|
---|
323 |
|
---|
324 | static int
|
---|
325 | unpack_double(mrb_state *mrb, const unsigned char * src, int srclen, mrb_value ary, unsigned int flags)
|
---|
326 | {
|
---|
327 | int i;
|
---|
328 | double d;
|
---|
329 | uint8_t *buffer = (uint8_t *)&d;
|
---|
330 |
|
---|
331 | if (flags & PACK_FLAG_LITTLEENDIAN) {
|
---|
332 | #ifdef MRB_ENDIAN_BIG
|
---|
333 | for (i = 0; i < 8; ++i) {
|
---|
334 | buffer[8 - i - 1] = src[i];
|
---|
335 | }
|
---|
336 | #else
|
---|
337 | memcpy(buffer, src, 8);
|
---|
338 | #endif
|
---|
339 | } else {
|
---|
340 | #ifdef MRB_ENDIAN_BIG
|
---|
341 | memcpy(buffer, src, 8);
|
---|
342 | #else
|
---|
343 | for (i = 0; i < 8; ++i) {
|
---|
344 | buffer[8 - i - 1] = src[i];
|
---|
345 | }
|
---|
346 | #endif
|
---|
347 | }
|
---|
348 | mrb_ary_push(mrb, ary, mrb_float_value(mrb, d));
|
---|
349 |
|
---|
350 | return 8;
|
---|
351 | }
|
---|
352 |
|
---|
353 | static int
|
---|
354 | pack_float(mrb_state *mrb, mrb_value o, mrb_value str, mrb_int sidx, unsigned int flags)
|
---|
355 | {
|
---|
356 | int i;
|
---|
357 | float f;
|
---|
358 | uint8_t *buffer = (uint8_t *)&f;
|
---|
359 | str = str_len_ensure(mrb, str, sidx + 4);
|
---|
360 | f = mrb_float(o);
|
---|
361 |
|
---|
362 | if (flags & PACK_FLAG_LITTLEENDIAN) {
|
---|
363 | #ifdef MRB_ENDIAN_BIG
|
---|
364 | for (i = 0; i < 4; ++i) {
|
---|
365 | RSTRING_PTR(str)[sidx + i] = buffer[4 - i - 1];
|
---|
366 | }
|
---|
367 | #else
|
---|
368 | memcpy(RSTRING_PTR(str) + sidx, buffer, 4);
|
---|
369 | #endif
|
---|
370 | } else {
|
---|
371 | #ifdef MRB_ENDIAN_BIG
|
---|
372 | memcpy(RSTRING_PTR(str) + sidx, buffer, 4);
|
---|
373 | #else
|
---|
374 | for (i = 0; i < 4; ++i) {
|
---|
375 | RSTRING_PTR(str)[sidx + i] = buffer[4 - i - 1];
|
---|
376 | }
|
---|
377 | #endif
|
---|
378 | }
|
---|
379 |
|
---|
380 | return 4;
|
---|
381 | }
|
---|
382 |
|
---|
383 | static int
|
---|
384 | unpack_float(mrb_state *mrb, const unsigned char * src, int srclen, mrb_value ary, unsigned int flags)
|
---|
385 | {
|
---|
386 | int i;
|
---|
387 | float f;
|
---|
388 | uint8_t *buffer = (uint8_t *)&f;
|
---|
389 |
|
---|
390 | if (flags & PACK_FLAG_LITTLEENDIAN) {
|
---|
391 | #ifdef MRB_ENDIAN_BIG
|
---|
392 | for (i = 0; i < 4; ++i) {
|
---|
393 | buffer[4 - i - 1] = src[i];
|
---|
394 | }
|
---|
395 | #else
|
---|
396 | memcpy(buffer, src, 4);
|
---|
397 | #endif
|
---|
398 | } else {
|
---|
399 | #ifdef MRB_ENDIAN_BIG
|
---|
400 | memcpy(buffer, src, 4);
|
---|
401 | #else
|
---|
402 | for (i = 0; i < 4; ++i) {
|
---|
403 | buffer[4 - i - 1] = src[i];
|
---|
404 | }
|
---|
405 | #endif
|
---|
406 | }
|
---|
407 | mrb_ary_push(mrb, ary, mrb_float_value(mrb, f));
|
---|
408 |
|
---|
409 | return 4;
|
---|
410 | }
|
---|
411 |
|
---|
412 | static int
|
---|
413 | pack_utf8(mrb_state *mrb, mrb_value o, mrb_value str, mrb_int sidx, long count, unsigned int flags)
|
---|
414 | {
|
---|
415 | char utf8[4];
|
---|
416 | int len;
|
---|
417 | unsigned long c = 0;
|
---|
418 |
|
---|
419 | if (mrb_float_p(o)) {
|
---|
420 | goto range_error;
|
---|
421 | }
|
---|
422 | c = mrb_fixnum(o);
|
---|
423 |
|
---|
424 | /* Unicode character */
|
---|
425 | /* from mruby-compiler gem */
|
---|
426 | if (c < 0x80) {
|
---|
427 | utf8[0] = (char)c;
|
---|
428 | len = 1;
|
---|
429 | }
|
---|
430 | else if (c < 0x800) {
|
---|
431 | utf8[0] = (char)(0xC0 | (c >> 6));
|
---|
432 | utf8[1] = (char)(0x80 | (c & 0x3F));
|
---|
433 | len = 2;
|
---|
434 | }
|
---|
435 | else if (c < 0x10000) {
|
---|
436 | utf8[0] = (char)(0xE0 | (c >> 12) );
|
---|
437 | utf8[1] = (char)(0x80 | ((c >> 6) & 0x3F));
|
---|
438 | utf8[2] = (char)(0x80 | ( c & 0x3F));
|
---|
439 | len = 3;
|
---|
440 | }
|
---|
441 | else if (c < 0x200000) {
|
---|
442 | utf8[0] = (char)(0xF0 | (c >> 18) );
|
---|
443 | utf8[1] = (char)(0x80 | ((c >> 12) & 0x3F));
|
---|
444 | utf8[2] = (char)(0x80 | ((c >> 6) & 0x3F));
|
---|
445 | utf8[3] = (char)(0x80 | ( c & 0x3F));
|
---|
446 | len = 4;
|
---|
447 | }
|
---|
448 | else {
|
---|
449 | range_error:
|
---|
450 | mrb_raise(mrb, E_RANGE_ERROR, "pack(U): value out of range");
|
---|
451 | }
|
---|
452 |
|
---|
453 | str = str_len_ensure(mrb, str, sidx + len);
|
---|
454 | memcpy(RSTRING_PTR(str) + sidx, utf8, len);
|
---|
455 |
|
---|
456 | return len;
|
---|
457 | }
|
---|
458 |
|
---|
459 | static const unsigned long utf8_limits[] = {
|
---|
460 | 0x0, /* 1 */
|
---|
461 | 0x80, /* 2 */
|
---|
462 | 0x800, /* 3 */
|
---|
463 | 0x10000, /* 4 */
|
---|
464 | 0x200000, /* 5 */
|
---|
465 | 0x4000000, /* 6 */
|
---|
466 | 0x80000000, /* 7 */
|
---|
467 | };
|
---|
468 |
|
---|
469 | static unsigned long
|
---|
470 | utf8_to_uv(mrb_state *mrb, const char *p, long *lenp)
|
---|
471 | {
|
---|
472 | int c = *p++ & 0xff;
|
---|
473 | unsigned long uv = c;
|
---|
474 | long n;
|
---|
475 |
|
---|
476 | if (!(uv & 0x80)) {
|
---|
477 | *lenp = 1;
|
---|
478 | return uv;
|
---|
479 | }
|
---|
480 | if (!(uv & 0x40)) {
|
---|
481 | *lenp = 1;
|
---|
482 | mrb_raise(mrb, E_ARGUMENT_ERROR, "malformed UTF-8 character");
|
---|
483 | }
|
---|
484 |
|
---|
485 | if (!(uv & 0x20)) { n = 2; uv &= 0x1f; }
|
---|
486 | else if (!(uv & 0x10)) { n = 3; uv &= 0x0f; }
|
---|
487 | else if (!(uv & 0x08)) { n = 4; uv &= 0x07; }
|
---|
488 | else if (!(uv & 0x04)) { n = 5; uv &= 0x03; }
|
---|
489 | else if (!(uv & 0x02)) { n = 6; uv &= 0x01; }
|
---|
490 | else {
|
---|
491 | *lenp = 1;
|
---|
492 | mrb_raise(mrb, E_ARGUMENT_ERROR, "malformed UTF-8 character");
|
---|
493 | }
|
---|
494 | if (n > *lenp) {
|
---|
495 | mrb_raisef(mrb, E_ARGUMENT_ERROR, "malformed UTF-8 character (expected %S bytes, given %S bytes)",
|
---|
496 | mrb_fixnum_value(n), mrb_fixnum_value(*lenp));
|
---|
497 | }
|
---|
498 | *lenp = n--;
|
---|
499 | if (n != 0) {
|
---|
500 | while (n--) {
|
---|
501 | c = *p++ & 0xff;
|
---|
502 | if ((c & 0xc0) != 0x80) {
|
---|
503 | *lenp -= n + 1;
|
---|
504 | mrb_raisef(mrb, E_ARGUMENT_ERROR, "malformed UTF-8 character");
|
---|
505 | }
|
---|
506 | else {
|
---|
507 | c &= 0x3f;
|
---|
508 | uv = uv << 6 | c;
|
---|
509 | }
|
---|
510 | }
|
---|
511 | }
|
---|
512 | n = *lenp - 1;
|
---|
513 | if (uv < utf8_limits[n]) {
|
---|
514 | mrb_raisef(mrb, E_ARGUMENT_ERROR, "redundant UTF-8 sequence");
|
---|
515 | }
|
---|
516 | return uv;
|
---|
517 | }
|
---|
518 |
|
---|
519 | static int
|
---|
520 | unpack_utf8(mrb_state *mrb, const unsigned char * src, int srclen, mrb_value ary, unsigned int flags)
|
---|
521 | {
|
---|
522 | unsigned long uv;
|
---|
523 | long lenp = srclen;
|
---|
524 |
|
---|
525 | if (srclen == 0) {
|
---|
526 | return 1;
|
---|
527 | }
|
---|
528 | uv = utf8_to_uv(mrb, (const char *)src, &lenp);
|
---|
529 | mrb_ary_push(mrb, ary, mrb_fixnum_value((mrb_int)uv));
|
---|
530 | return (int)lenp;
|
---|
531 | }
|
---|
532 |
|
---|
533 | static int
|
---|
534 | pack_a(mrb_state *mrb, mrb_value src, mrb_value dst, mrb_int didx, long count, unsigned int flags)
|
---|
535 | {
|
---|
536 | int copylen, slen, padlen;
|
---|
537 | char *dptr, *dptr0, pad, *sptr;
|
---|
538 |
|
---|
539 | sptr = RSTRING_PTR(src);
|
---|
540 | slen = RSTRING_LEN(src);
|
---|
541 |
|
---|
542 | if ((flags & PACK_FLAG_a) || (flags & PACK_FLAG_Z))
|
---|
543 | pad = '\0';
|
---|
544 | else
|
---|
545 | pad = ' ';
|
---|
546 |
|
---|
547 | if (count == 0) {
|
---|
548 | return 0;
|
---|
549 | } else if (count == -1) {
|
---|
550 | copylen = slen;
|
---|
551 | padlen = (flags & PACK_FLAG_Z) ? 1 : 0;
|
---|
552 | } else if (count < slen) {
|
---|
553 | copylen = count;
|
---|
554 | padlen = 0;
|
---|
555 | } else {
|
---|
556 | copylen = slen;
|
---|
557 | padlen = count - slen;
|
---|
558 | }
|
---|
559 |
|
---|
560 | dst = str_len_ensure(mrb, dst, didx + copylen + padlen);
|
---|
561 | dptr0 = dptr = RSTRING_PTR(dst) + didx;
|
---|
562 | memcpy(dptr, sptr, copylen);
|
---|
563 | dptr += copylen;
|
---|
564 | while (padlen-- > 0) {
|
---|
565 | *dptr++ = pad;
|
---|
566 | }
|
---|
567 |
|
---|
568 | return dptr - dptr0;
|
---|
569 | }
|
---|
570 |
|
---|
571 | static int
|
---|
572 | unpack_a(mrb_state *mrb, const void *src, int slen, mrb_value ary, long count, unsigned int flags)
|
---|
573 | {
|
---|
574 | mrb_value dst;
|
---|
575 | const char *cp, *sptr;
|
---|
576 | long copylen;
|
---|
577 |
|
---|
578 | sptr = (const char *)src;
|
---|
579 | if (count != -1 && count < slen) {
|
---|
580 | slen = count;
|
---|
581 | }
|
---|
582 | copylen = slen;
|
---|
583 |
|
---|
584 | if (flags & PACK_FLAG_Z) { /* "Z" */
|
---|
585 | if ((cp = (const char *)memchr(sptr, '\0', slen)) != NULL) {
|
---|
586 | copylen = cp - sptr;
|
---|
587 | if (count == -1) {
|
---|
588 | slen = copylen + 1;
|
---|
589 | }
|
---|
590 | }
|
---|
591 | } else if (!(flags & PACK_FLAG_a)) { /* "A" */
|
---|
592 | while (copylen > 0 && (sptr[copylen - 1] == '\0' || isspace(sptr[copylen - 1]))) {
|
---|
593 | copylen--;
|
---|
594 | }
|
---|
595 | }
|
---|
596 |
|
---|
597 | dst = mrb_str_new(mrb, sptr, copylen);
|
---|
598 | mrb_ary_push(mrb, ary, dst);
|
---|
599 | return slen;
|
---|
600 | }
|
---|
601 |
|
---|
602 |
|
---|
603 | static int
|
---|
604 | pack_h(mrb_state *mrb, mrb_value src, mrb_value dst, mrb_int didx, long count, unsigned int flags)
|
---|
605 | {
|
---|
606 | unsigned int a, ashift, b, bshift;
|
---|
607 | int slen;
|
---|
608 | char *dptr, *dptr0, *sptr;
|
---|
609 |
|
---|
610 | sptr = RSTRING_PTR(src);
|
---|
611 | slen = RSTRING_LEN(src);
|
---|
612 |
|
---|
613 | if (flags & PACK_FLAG_LSB) {
|
---|
614 | ashift = 0;
|
---|
615 | bshift = 4;
|
---|
616 | } else {
|
---|
617 | ashift = 4;
|
---|
618 | bshift = 0;
|
---|
619 | }
|
---|
620 |
|
---|
621 | if (count == -1) {
|
---|
622 | count = slen;
|
---|
623 | } else if (slen > count) {
|
---|
624 | slen = count;
|
---|
625 | }
|
---|
626 |
|
---|
627 | dst = str_len_ensure(mrb, dst, didx + count);
|
---|
628 | dptr = RSTRING_PTR(dst) + didx;
|
---|
629 |
|
---|
630 | dptr0 = dptr;
|
---|
631 | for (; count > 0; count -= 2) {
|
---|
632 | a = b = 0;
|
---|
633 | if (slen > 0) {
|
---|
634 | a = hex2int(*sptr++);
|
---|
635 | slen--;
|
---|
636 | }
|
---|
637 | if (slen > 0) {
|
---|
638 | b = hex2int(*sptr++);
|
---|
639 | slen--;
|
---|
640 | }
|
---|
641 | *dptr++ = (a << ashift) + (b << bshift);
|
---|
642 | }
|
---|
643 |
|
---|
644 | return dptr - dptr0;
|
---|
645 | }
|
---|
646 |
|
---|
647 | static int
|
---|
648 | unpack_h(mrb_state *mrb, const void *src, int slen, mrb_value ary, int count, unsigned int flags)
|
---|
649 | {
|
---|
650 | mrb_value dst;
|
---|
651 | int a, ashift, b, bshift;
|
---|
652 | const char *sptr, *sptr0;
|
---|
653 | char *dptr, *dptr0;
|
---|
654 | const char hexadecimal[] = "0123456789abcdef";
|
---|
655 |
|
---|
656 | if (flags & PACK_FLAG_LSB) {
|
---|
657 | ashift = 0;
|
---|
658 | bshift = 4;
|
---|
659 | } else {
|
---|
660 | ashift = 4;
|
---|
661 | bshift = 0;
|
---|
662 | }
|
---|
663 |
|
---|
664 | sptr = (const char *)src;
|
---|
665 |
|
---|
666 | if (count == -1)
|
---|
667 | count = slen * 2;
|
---|
668 |
|
---|
669 | dst = mrb_str_new(mrb, NULL, count);
|
---|
670 | dptr = RSTRING_PTR(dst);
|
---|
671 |
|
---|
672 | sptr0 = sptr;
|
---|
673 | dptr0 = dptr;
|
---|
674 | while (slen > 0 && count > 0) {
|
---|
675 | a = (*sptr >> ashift) & 0x0f;
|
---|
676 | b = (*sptr >> bshift) & 0x0f;
|
---|
677 | sptr++;
|
---|
678 | slen--;
|
---|
679 |
|
---|
680 | *dptr++ = hexadecimal[a];
|
---|
681 | count--;
|
---|
682 |
|
---|
683 | if (count > 0) {
|
---|
684 | *dptr++ = hexadecimal[b];
|
---|
685 | count--;
|
---|
686 | }
|
---|
687 | }
|
---|
688 |
|
---|
689 | dst = mrb_str_resize(mrb, dst, dptr - dptr0);
|
---|
690 | mrb_ary_push(mrb, ary, dst);
|
---|
691 | return sptr - sptr0;
|
---|
692 | }
|
---|
693 |
|
---|
694 |
|
---|
695 | static int
|
---|
696 | pack_m(mrb_state *mrb, mrb_value src, mrb_value dst, mrb_int didx, long count, unsigned int flags)
|
---|
697 | {
|
---|
698 | mrb_int dstlen;
|
---|
699 | unsigned long l;
|
---|
700 | int column, srclen;
|
---|
701 | char *srcptr, *dstptr, *dstptr0;
|
---|
702 |
|
---|
703 | srcptr = RSTRING_PTR(src);
|
---|
704 | srclen = RSTRING_LEN(src);
|
---|
705 |
|
---|
706 | if (srclen == 0) /* easy case */
|
---|
707 | return 0;
|
---|
708 |
|
---|
709 | if (count != 0 && count < 3) { /* -1, 1 or 2 */
|
---|
710 | count = 45;
|
---|
711 | } else if (count >= 3) {
|
---|
712 | count -= count % 3;
|
---|
713 | }
|
---|
714 |
|
---|
715 | dstlen = srclen / 3 * 4;
|
---|
716 | if (count > 0) {
|
---|
717 | dstlen += (srclen / count) + ((srclen % count) == 0 ? 0 : 1);
|
---|
718 | }
|
---|
719 | dst = str_len_ensure(mrb, dst, didx + dstlen);
|
---|
720 | dstptr = RSTRING_PTR(dst) + didx;
|
---|
721 |
|
---|
722 | dstptr0 = dstptr;
|
---|
723 | for (column = 3; srclen >= 3; srclen -= 3, column += 3) {
|
---|
724 | l = (unsigned char)*srcptr++ << 16;
|
---|
725 | l += (unsigned char)*srcptr++ << 8;
|
---|
726 | l += (unsigned char)*srcptr++;
|
---|
727 |
|
---|
728 | *dstptr++ = base64chars[(l >> 18) & 0x3f];
|
---|
729 | *dstptr++ = base64chars[(l >> 12) & 0x3f];
|
---|
730 | *dstptr++ = base64chars[(l >> 6) & 0x3f];
|
---|
731 | *dstptr++ = base64chars[ l & 0x3f];
|
---|
732 |
|
---|
733 | if (column == count) {
|
---|
734 | *dstptr++ = '\n';
|
---|
735 | column = 0;
|
---|
736 | }
|
---|
737 | }
|
---|
738 | if (srclen == 1) {
|
---|
739 | l = (unsigned char)*srcptr++ << 16;
|
---|
740 | *dstptr++ = base64chars[(l >> 18) & 0x3f];
|
---|
741 | *dstptr++ = base64chars[(l >> 12) & 0x3f];
|
---|
742 | *dstptr++ = '=';
|
---|
743 | *dstptr++ = '=';
|
---|
744 | column += 3;
|
---|
745 | } else if (srclen == 2) {
|
---|
746 | l = (unsigned char)*srcptr++ << 16;
|
---|
747 | l += (unsigned char)*srcptr++ << 8;
|
---|
748 | *dstptr++ = base64chars[(l >> 18) & 0x3f];
|
---|
749 | *dstptr++ = base64chars[(l >> 12) & 0x3f];
|
---|
750 | *dstptr++ = base64chars[(l >> 6) & 0x3f];
|
---|
751 | *dstptr++ = '=';
|
---|
752 | column += 3;
|
---|
753 | }
|
---|
754 | if (column > 0 && count > 0) {
|
---|
755 | *dstptr++ = '\n';
|
---|
756 | }
|
---|
757 |
|
---|
758 | return dstptr - dstptr0;
|
---|
759 | }
|
---|
760 |
|
---|
761 | static int
|
---|
762 | unpack_m(mrb_state *mrb, const void *src, int slen, mrb_value ary, unsigned int flags)
|
---|
763 | {
|
---|
764 | mrb_value dst;
|
---|
765 | int dlen;
|
---|
766 | unsigned long l;
|
---|
767 | int i, padding;
|
---|
768 | unsigned char c, ch[4];
|
---|
769 | const char *sptr, *sptr0;
|
---|
770 | char *dptr, *dptr0;
|
---|
771 |
|
---|
772 | sptr0 = sptr = (const char *)src;
|
---|
773 |
|
---|
774 | dlen = slen / 4 * 3; /* an estimated value - may be shorter */
|
---|
775 | dst = mrb_str_new(mrb, NULL, dlen);
|
---|
776 | dptr0 = dptr = RSTRING_PTR(dst);
|
---|
777 |
|
---|
778 | padding = 0;
|
---|
779 | while (slen >= 4) {
|
---|
780 | for (i = 0; i < 4; i++) {
|
---|
781 | do {
|
---|
782 | if (slen-- == 0)
|
---|
783 | goto done;
|
---|
784 | c = *sptr++;
|
---|
785 | if (c >= sizeof(base64_dec_tab))
|
---|
786 | continue;
|
---|
787 | ch[i] = base64_dec_tab[c];
|
---|
788 | if (ch[i] == PACK_BASE64_PADDING) {
|
---|
789 | ch[i] = 0;
|
---|
790 | padding++;
|
---|
791 | }
|
---|
792 | } while (ch[i] == PACK_BASE64_IGNORE);
|
---|
793 | }
|
---|
794 |
|
---|
795 | l = (ch[0] << 18) + (ch[1] << 12) + (ch[2] << 6) + ch[3];
|
---|
796 |
|
---|
797 | if (padding == 0) {
|
---|
798 | *dptr++ = (l >> 16) & 0xff;
|
---|
799 | *dptr++ = (l >> 8) & 0xff;
|
---|
800 | *dptr++ = l & 0xff;
|
---|
801 | } else if (padding == 1) {
|
---|
802 | *dptr++ = (l >> 16) & 0xff;
|
---|
803 | *dptr++ = (l >> 8) & 0xff;
|
---|
804 | break;
|
---|
805 | } else {
|
---|
806 | *dptr++ = (l >> 16) & 0xff;
|
---|
807 | break;
|
---|
808 | }
|
---|
809 | }
|
---|
810 |
|
---|
811 | done:
|
---|
812 | dst = mrb_str_resize(mrb, dst, dptr - dptr0);
|
---|
813 | mrb_ary_push(mrb, ary, dst);
|
---|
814 | return sptr - sptr0;
|
---|
815 | }
|
---|
816 |
|
---|
817 | static int
|
---|
818 | pack_x(mrb_state *mrb, mrb_value src, mrb_value dst, mrb_int didx, long count, unsigned int flags)
|
---|
819 | {
|
---|
820 | long i;
|
---|
821 |
|
---|
822 | dst = str_len_ensure(mrb, dst, didx + count);
|
---|
823 | for (i = 0; i < count; i++) {
|
---|
824 | RSTRING_PTR(dst)[didx] = '\0';
|
---|
825 | }
|
---|
826 | return count;
|
---|
827 | }
|
---|
828 |
|
---|
829 | static int
|
---|
830 | unpack_x(mrb_state *mrb, const void *src, int slen, mrb_value ary, int count, unsigned int flags)
|
---|
831 | {
|
---|
832 | if (slen < count) {
|
---|
833 | mrb_raise(mrb, E_ARGUMENT_ERROR, "x outside of string");
|
---|
834 | }
|
---|
835 | return count;
|
---|
836 | }
|
---|
837 |
|
---|
838 | static void
|
---|
839 | prepare_tmpl(mrb_state *mrb, struct tmpl *tmpl)
|
---|
840 | {
|
---|
841 | mrb_get_args(mrb, "S", &tmpl->str);
|
---|
842 | tmpl->idx = 0;
|
---|
843 | }
|
---|
844 |
|
---|
845 | static int
|
---|
846 | has_tmpl(const struct tmpl *tmpl)
|
---|
847 | {
|
---|
848 | return (tmpl->idx < RSTRING_LEN(tmpl->str));
|
---|
849 | }
|
---|
850 |
|
---|
851 | static void
|
---|
852 | read_tmpl(mrb_state *mrb, struct tmpl *tmpl, int *dirp, int *typep, int *sizep, long *countp, unsigned int *flagsp)
|
---|
853 | {
|
---|
854 | int ch, dir, t, tlen, type;
|
---|
855 | int size = 0;
|
---|
856 | long count = 1;
|
---|
857 | unsigned int flags = 0;
|
---|
858 | const char *tptr;
|
---|
859 |
|
---|
860 | tptr = RSTRING_PTR(tmpl->str);
|
---|
861 | tlen = RSTRING_LEN(tmpl->str);
|
---|
862 |
|
---|
863 | t = tptr[tmpl->idx++];
|
---|
864 | alias:
|
---|
865 | switch (t) {
|
---|
866 | case 'A':
|
---|
867 | dir = PACK_DIR_STR;
|
---|
868 | type = PACK_TYPE_STRING;
|
---|
869 | flags |= PACK_FLAG_WIDTH | PACK_FLAG_COUNT2;
|
---|
870 | break;
|
---|
871 | case 'a':
|
---|
872 | dir = PACK_DIR_STR;
|
---|
873 | type = PACK_TYPE_STRING;
|
---|
874 | flags |= PACK_FLAG_WIDTH | PACK_FLAG_COUNT2 | PACK_FLAG_a;
|
---|
875 | break;
|
---|
876 | case 'C':
|
---|
877 | dir = PACK_DIR_CHAR;
|
---|
878 | type = PACK_TYPE_INTEGER;
|
---|
879 | size = 1;
|
---|
880 | break;
|
---|
881 | case 'c':
|
---|
882 | dir = PACK_DIR_CHAR;
|
---|
883 | type = PACK_TYPE_INTEGER;
|
---|
884 | size = 1;
|
---|
885 | flags |= PACK_FLAG_SIGNED;
|
---|
886 | break;
|
---|
887 | case 'D': case 'd':
|
---|
888 | dir = PACK_DIR_DOUBLE;
|
---|
889 | type = PACK_TYPE_FLOAT;
|
---|
890 | size = 8;
|
---|
891 | flags |= PACK_FLAG_SIGNED;
|
---|
892 | break;
|
---|
893 | case 'F': case 'f':
|
---|
894 | dir = PACK_DIR_FLOAT;
|
---|
895 | type = PACK_TYPE_FLOAT;
|
---|
896 | size = 4;
|
---|
897 | flags |= PACK_FLAG_SIGNED;
|
---|
898 | break;
|
---|
899 | case 'E':
|
---|
900 | dir = PACK_DIR_DOUBLE;
|
---|
901 | type = PACK_TYPE_FLOAT;
|
---|
902 | size = 8;
|
---|
903 | flags |= PACK_FLAG_SIGNED | PACK_FLAG_LT;
|
---|
904 | break;
|
---|
905 | case 'e':
|
---|
906 | dir = PACK_DIR_FLOAT;
|
---|
907 | type = PACK_TYPE_FLOAT;
|
---|
908 | size = 4;
|
---|
909 | flags |= PACK_FLAG_SIGNED | PACK_FLAG_LT;
|
---|
910 | break;
|
---|
911 | case 'G':
|
---|
912 | dir = PACK_DIR_DOUBLE;
|
---|
913 | type = PACK_TYPE_FLOAT;
|
---|
914 | size = 8;
|
---|
915 | flags |= PACK_FLAG_SIGNED | PACK_FLAG_GT;
|
---|
916 | break;
|
---|
917 | case 'g':
|
---|
918 | dir = PACK_DIR_FLOAT;
|
---|
919 | type = PACK_TYPE_FLOAT;
|
---|
920 | size = 4;
|
---|
921 | flags |= PACK_FLAG_SIGNED | PACK_FLAG_GT;
|
---|
922 | break;
|
---|
923 | case 'H':
|
---|
924 | dir = PACK_DIR_HEX;
|
---|
925 | type = PACK_TYPE_STRING;
|
---|
926 | flags |= PACK_FLAG_COUNT2;
|
---|
927 | break;
|
---|
928 | case 'h':
|
---|
929 | dir = PACK_DIR_HEX;
|
---|
930 | type = PACK_TYPE_STRING;
|
---|
931 | flags |= PACK_FLAG_COUNT2 | PACK_FLAG_LSB;
|
---|
932 | break;
|
---|
933 | case 'I':
|
---|
934 | switch (sizeof(int)) {
|
---|
935 | case 2: t = 'S'; goto alias;
|
---|
936 | case 4: t = 'L'; goto alias;
|
---|
937 | case 8: t = 'Q'; goto alias;
|
---|
938 | default:
|
---|
939 | mrb_raisef(mrb, E_RUNTIME_ERROR, "mruby-pack does not support sizeof(int) == %S", mrb_fixnum_value(sizeof(int)));
|
---|
940 | }
|
---|
941 | break;
|
---|
942 | case 'i':
|
---|
943 | switch (sizeof(int)) {
|
---|
944 | case 2: t = 's'; goto alias;
|
---|
945 | case 4: t = 'l'; goto alias;
|
---|
946 | case 8: t = 'q'; goto alias;
|
---|
947 | default:
|
---|
948 | mrb_raisef(mrb, E_RUNTIME_ERROR, "mruby-pack does not support sizeof(int) == %S", mrb_fixnum_value(sizeof(int)));
|
---|
949 | }
|
---|
950 | break;
|
---|
951 | case 'L':
|
---|
952 | dir = PACK_DIR_LONG;
|
---|
953 | type = PACK_TYPE_INTEGER;
|
---|
954 | size = 4;
|
---|
955 | break;
|
---|
956 | case 'l':
|
---|
957 | dir = PACK_DIR_LONG;
|
---|
958 | type = PACK_TYPE_INTEGER;
|
---|
959 | size = 4;
|
---|
960 | flags |= PACK_FLAG_SIGNED;
|
---|
961 | break;
|
---|
962 | case 'm':
|
---|
963 | dir = PACK_DIR_BASE64;
|
---|
964 | type = PACK_TYPE_STRING;
|
---|
965 | flags |= PACK_FLAG_WIDTH;
|
---|
966 | break;
|
---|
967 | case 'N': /* = "L>" */
|
---|
968 | dir = PACK_DIR_LONG;
|
---|
969 | type = PACK_TYPE_INTEGER;
|
---|
970 | size = 4;
|
---|
971 | flags |= PACK_FLAG_GT;
|
---|
972 | break;
|
---|
973 | case 'n': /* = "S>" */
|
---|
974 | dir = PACK_DIR_SHORT;
|
---|
975 | type = PACK_TYPE_INTEGER;
|
---|
976 | size = 2;
|
---|
977 | flags |= PACK_FLAG_GT;
|
---|
978 | break;
|
---|
979 | case 'Q':
|
---|
980 | dir = PACK_DIR_QUAD;
|
---|
981 | type = PACK_TYPE_INTEGER;
|
---|
982 | size = 8;
|
---|
983 | break;
|
---|
984 | case 'q':
|
---|
985 | dir = PACK_DIR_QUAD;
|
---|
986 | type = PACK_TYPE_INTEGER;
|
---|
987 | size = 8;
|
---|
988 | flags |= PACK_FLAG_SIGNED;
|
---|
989 | break;
|
---|
990 | case 'S':
|
---|
991 | dir = PACK_DIR_SHORT;
|
---|
992 | type = PACK_TYPE_INTEGER;
|
---|
993 | size = 2;
|
---|
994 | break;
|
---|
995 | case 's':
|
---|
996 | dir = PACK_DIR_SHORT;
|
---|
997 | type = PACK_TYPE_INTEGER;
|
---|
998 | size = 2;
|
---|
999 | flags |= PACK_FLAG_SIGNED;
|
---|
1000 | break;
|
---|
1001 | case 'U':
|
---|
1002 | dir = PACK_DIR_UTF8;
|
---|
1003 | type = PACK_TYPE_INTEGER;
|
---|
1004 | break;
|
---|
1005 | case 'V': /* = "L<" */
|
---|
1006 | dir = PACK_DIR_LONG;
|
---|
1007 | type = PACK_TYPE_INTEGER;
|
---|
1008 | size = 4;
|
---|
1009 | flags |= PACK_FLAG_LT;
|
---|
1010 | break;
|
---|
1011 | case 'v': /* = "S<" */
|
---|
1012 | dir = PACK_DIR_SHORT;
|
---|
1013 | type = PACK_TYPE_INTEGER;
|
---|
1014 | size = 2;
|
---|
1015 | flags |= PACK_FLAG_LT;
|
---|
1016 | break;
|
---|
1017 | case 'x':
|
---|
1018 | dir = PACK_DIR_NUL;
|
---|
1019 | type = PACK_TYPE_NONE;
|
---|
1020 | break;
|
---|
1021 | case 'Z':
|
---|
1022 | dir = PACK_DIR_STR;
|
---|
1023 | type = PACK_TYPE_STRING;
|
---|
1024 | flags |= PACK_FLAG_WIDTH | PACK_FLAG_COUNT2 | PACK_FLAG_Z;
|
---|
1025 | break;
|
---|
1026 | default:
|
---|
1027 | dir = PACK_DIR_INVALID;
|
---|
1028 | type = PACK_TYPE_NONE;
|
---|
1029 | break;
|
---|
1030 | }
|
---|
1031 |
|
---|
1032 | /* read suffix [0-9*_!<>] */
|
---|
1033 | while (tmpl->idx < tlen) {
|
---|
1034 | ch = tptr[tmpl->idx++];
|
---|
1035 | if (isdigit(ch)) {
|
---|
1036 | count = ch - '0';
|
---|
1037 | while (tmpl->idx < tlen && isdigit(tptr[tmpl->idx])) {
|
---|
1038 | count = count * 10 + (tptr[tmpl->idx++] - '0');
|
---|
1039 | }
|
---|
1040 | continue; /* special case */
|
---|
1041 | } else if (ch == '*') {
|
---|
1042 | count = -1;
|
---|
1043 | } else if (ch == '_' || ch == '!' || ch == '<' || ch == '>') {
|
---|
1044 | if (strchr("sSiIlLqQ", t) == NULL) {
|
---|
1045 | char ch_str = ch;
|
---|
1046 | mrb_raisef(mrb, E_ARGUMENT_ERROR, "'%S' allowed only after types sSiIlLqQ", mrb_str_new(mrb, &ch_str, 1));
|
---|
1047 | }
|
---|
1048 | if (ch == '_' || ch == '!') {
|
---|
1049 | flags |= PACK_FLAG_s;
|
---|
1050 | } else if (ch == '<') {
|
---|
1051 | flags |= PACK_FLAG_LT;
|
---|
1052 | } else if (ch == '>') {
|
---|
1053 | flags |= PACK_FLAG_GT;
|
---|
1054 | }
|
---|
1055 | } else {
|
---|
1056 | tmpl->idx--;
|
---|
1057 | break;
|
---|
1058 | }
|
---|
1059 | }
|
---|
1060 |
|
---|
1061 | if ((flags & PACK_FLAG_LT) || (!(flags & PACK_FLAG_GT) && littleendian)) {
|
---|
1062 | flags |= PACK_FLAG_LITTLEENDIAN;
|
---|
1063 | }
|
---|
1064 |
|
---|
1065 | *dirp = dir;
|
---|
1066 | *typep = type;
|
---|
1067 | *sizep = size;
|
---|
1068 | *countp = count;
|
---|
1069 | *flagsp = flags;
|
---|
1070 | }
|
---|
1071 |
|
---|
1072 | static mrb_value
|
---|
1073 | mrb_pack_pack(mrb_state *mrb, mrb_value ary)
|
---|
1074 | {
|
---|
1075 | mrb_value o, result;
|
---|
1076 | mrb_int aidx;
|
---|
1077 | struct tmpl tmpl;
|
---|
1078 | long count;
|
---|
1079 | unsigned int flags;
|
---|
1080 | int dir, ridx, size, type;
|
---|
1081 |
|
---|
1082 | prepare_tmpl(mrb, &tmpl);
|
---|
1083 |
|
---|
1084 | result = mrb_str_new(mrb, NULL, 128); /* allocate initial buffer */
|
---|
1085 | aidx = 0;
|
---|
1086 | ridx = 0;
|
---|
1087 | while (has_tmpl(&tmpl)) {
|
---|
1088 | read_tmpl(mrb, &tmpl, &dir, &type, &size, &count, &flags);
|
---|
1089 |
|
---|
1090 | if (dir == PACK_DIR_INVALID)
|
---|
1091 | continue;
|
---|
1092 | else if (dir == PACK_DIR_NUL) {
|
---|
1093 | ridx += pack_x(mrb, mrb_nil_value(), result, ridx, count, flags);
|
---|
1094 | continue;
|
---|
1095 | }
|
---|
1096 |
|
---|
1097 | for (; aidx < RARRAY_LEN(ary); aidx++) {
|
---|
1098 | if (count == 0 && !(flags & PACK_FLAG_WIDTH))
|
---|
1099 | break;
|
---|
1100 |
|
---|
1101 | o = mrb_ary_ref(mrb, ary, aidx);
|
---|
1102 | if (type == PACK_TYPE_INTEGER) {
|
---|
1103 | o = mrb_to_int(mrb, o);
|
---|
1104 | } else if (type == PACK_TYPE_FLOAT) {
|
---|
1105 | if (!mrb_float_p(o)) {
|
---|
1106 | o = mrb_funcall(mrb, o, "to_f", 0);
|
---|
1107 | }
|
---|
1108 | } else if (type == PACK_TYPE_STRING) {
|
---|
1109 | if (!mrb_string_p(o)) {
|
---|
1110 | mrb_raisef(mrb, E_TYPE_ERROR, "can't convert %S into String", mrb_class_path(mrb, mrb_obj_class(mrb, o)));
|
---|
1111 | }
|
---|
1112 | }
|
---|
1113 |
|
---|
1114 | switch (dir) {
|
---|
1115 | case PACK_DIR_CHAR:
|
---|
1116 | ridx += pack_c(mrb, o, result, ridx, flags);
|
---|
1117 | break;
|
---|
1118 | case PACK_DIR_SHORT:
|
---|
1119 | ridx += pack_s(mrb, o, result, ridx, flags);
|
---|
1120 | break;
|
---|
1121 | case PACK_DIR_LONG:
|
---|
1122 | ridx += pack_l(mrb, o, result, ridx, flags);
|
---|
1123 | break;
|
---|
1124 | case PACK_DIR_QUAD:
|
---|
1125 | ridx += pack_q(mrb, o, result, ridx, flags);
|
---|
1126 | break;
|
---|
1127 | case PACK_DIR_BASE64:
|
---|
1128 | ridx += pack_m(mrb, o, result, ridx, count, flags);
|
---|
1129 | break;
|
---|
1130 | case PACK_DIR_HEX:
|
---|
1131 | ridx += pack_h(mrb, o, result, ridx, count, flags);
|
---|
1132 | break;
|
---|
1133 | case PACK_DIR_STR:
|
---|
1134 | ridx += pack_a(mrb, o, result, ridx, count, flags);
|
---|
1135 | break;
|
---|
1136 | case PACK_DIR_DOUBLE:
|
---|
1137 | ridx += pack_double(mrb, o, result, ridx, flags);
|
---|
1138 | break;
|
---|
1139 | case PACK_DIR_FLOAT:
|
---|
1140 | ridx += pack_float(mrb, o, result, ridx, flags);
|
---|
1141 | break;
|
---|
1142 | case PACK_DIR_UTF8:
|
---|
1143 | ridx += pack_utf8(mrb, o, result, ridx, count, flags);
|
---|
1144 | break;
|
---|
1145 | default:
|
---|
1146 | break;
|
---|
1147 | }
|
---|
1148 | if (dir == PACK_DIR_STR) { /* always consumes 1 entry */
|
---|
1149 | aidx++;
|
---|
1150 | break;
|
---|
1151 | }
|
---|
1152 | if (count > 0) {
|
---|
1153 | count--;
|
---|
1154 | }
|
---|
1155 | }
|
---|
1156 | }
|
---|
1157 |
|
---|
1158 | mrb_str_resize(mrb, result, ridx);
|
---|
1159 | return result;
|
---|
1160 | }
|
---|
1161 |
|
---|
1162 | static mrb_value
|
---|
1163 | mrb_pack_unpack(mrb_state *mrb, mrb_value str)
|
---|
1164 | {
|
---|
1165 | mrb_value result;
|
---|
1166 | struct tmpl tmpl;
|
---|
1167 | long count;
|
---|
1168 | unsigned int flags;
|
---|
1169 | int dir, size, srcidx, srclen, type;
|
---|
1170 | const unsigned char *sptr;
|
---|
1171 |
|
---|
1172 | prepare_tmpl(mrb, &tmpl);
|
---|
1173 |
|
---|
1174 | srcidx = 0;
|
---|
1175 | srclen = RSTRING_LEN(str);
|
---|
1176 |
|
---|
1177 | result = mrb_ary_new(mrb);
|
---|
1178 | while (has_tmpl(&tmpl)) {
|
---|
1179 | read_tmpl(mrb, &tmpl, &dir, &type, &size, &count, &flags);
|
---|
1180 |
|
---|
1181 | if (dir == PACK_DIR_INVALID)
|
---|
1182 | continue;
|
---|
1183 | else if (dir == PACK_DIR_NUL) {
|
---|
1184 | srcidx += unpack_x(mrb, sptr, srclen - srcidx, result, count, flags);
|
---|
1185 | continue;
|
---|
1186 | }
|
---|
1187 |
|
---|
1188 | if (flags & PACK_FLAG_COUNT2) {
|
---|
1189 | sptr = (const unsigned char *)RSTRING_PTR(str) + srcidx;
|
---|
1190 | switch (dir) {
|
---|
1191 | case PACK_DIR_HEX:
|
---|
1192 | srcidx += unpack_h(mrb, sptr, srclen - srcidx, result, count, flags);
|
---|
1193 | break;
|
---|
1194 | case PACK_DIR_STR:
|
---|
1195 | srcidx += unpack_a(mrb, sptr, srclen - srcidx, result, count, flags);
|
---|
1196 | break;
|
---|
1197 | }
|
---|
1198 | continue;
|
---|
1199 | }
|
---|
1200 |
|
---|
1201 | while (count != 0) {
|
---|
1202 | if (srclen - srcidx < size) {
|
---|
1203 | while (count-- > 0) {
|
---|
1204 | mrb_ary_push(mrb, result, mrb_nil_value());
|
---|
1205 | }
|
---|
1206 | break;
|
---|
1207 | }
|
---|
1208 |
|
---|
1209 | sptr = (const unsigned char *)RSTRING_PTR(str) + srcidx;
|
---|
1210 | switch (dir) {
|
---|
1211 | case PACK_DIR_CHAR:
|
---|
1212 | srcidx += unpack_c(mrb, sptr, srclen - srcidx, result, flags);
|
---|
1213 | break;
|
---|
1214 | case PACK_DIR_SHORT:
|
---|
1215 | srcidx += unpack_s(mrb, sptr, srclen - srcidx, result, flags);
|
---|
1216 | break;
|
---|
1217 | case PACK_DIR_LONG:
|
---|
1218 | srcidx += unpack_l(mrb, sptr, srclen - srcidx, result, flags);
|
---|
1219 | break;
|
---|
1220 | case PACK_DIR_QUAD:
|
---|
1221 | srcidx += unpack_q(mrb, sptr, srclen - srcidx, result, flags);
|
---|
1222 | break;
|
---|
1223 | case PACK_DIR_BASE64:
|
---|
1224 | srcidx += unpack_m(mrb, sptr, srclen - srcidx, result, flags);
|
---|
1225 | break;
|
---|
1226 | case PACK_DIR_FLOAT:
|
---|
1227 | srcidx += unpack_float(mrb, sptr, srclen - srcidx, result, flags);
|
---|
1228 | break;
|
---|
1229 | case PACK_DIR_DOUBLE:
|
---|
1230 | srcidx += unpack_double(mrb, sptr, srclen - srcidx, result, flags);
|
---|
1231 | break;
|
---|
1232 | case PACK_DIR_UTF8:
|
---|
1233 | srcidx += unpack_utf8(mrb, sptr, srclen - srcidx, result, flags);
|
---|
1234 | break;
|
---|
1235 | default:
|
---|
1236 | mrb_raise(mrb, E_RUNTIME_ERROR, "mruby-pack's bug");
|
---|
1237 | }
|
---|
1238 | if (count > 0) {
|
---|
1239 | count--;
|
---|
1240 | }
|
---|
1241 | }
|
---|
1242 | }
|
---|
1243 |
|
---|
1244 | return result;
|
---|
1245 | }
|
---|
1246 |
|
---|
1247 | void
|
---|
1248 | mrb_mruby_pack_gem_init(mrb_state *mrb)
|
---|
1249 | {
|
---|
1250 | littleendian = check_little_endian();
|
---|
1251 | make_base64_dec_tab();
|
---|
1252 |
|
---|
1253 | mrb_define_method(mrb, mrb->array_class, "pack", mrb_pack_pack, MRB_ARGS_REQ(1));
|
---|
1254 | mrb_define_method(mrb, mrb->string_class, "unpack", mrb_pack_unpack, MRB_ARGS_REQ(1));
|
---|
1255 | }
|
---|
1256 |
|
---|
1257 | void
|
---|
1258 | mrb_mruby_pack_gem_final(mrb_state *mrb)
|
---|
1259 | {
|
---|
1260 | }
|
---|