?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
* include/ruby/encoding.h (rb_econv_t): add fields: in_buf_start,
[ruby-svn.git] / marshal.c
blob0909addc83fe994b0deb1ce08ea32ff6f7386502
1 /**********************************************************************
2
3 marshal.c -
4
5 $Author$
6 created at: Thu Apr 27 16:30:01 JST 1995
7
8 Copyright (C) 1993-2007 Yukihiro Matsumoto
9
10 **********************************************************************/
11
12 #include "ruby/ruby.h"
13 #include "ruby/io.h"
14 #include "ruby/st.h"
15 #include "ruby/util.h"
16 #include "ruby/encoding.h"
17
18 #include <math.h>
19 #ifdef HAVE_FLOAT_H
20 #include <float.h>
21 #endif
22 #ifdef HAVE_IEEEFP_H
23 #include <ieeefp.h>
24 #endif
25
26 #define BITSPERSHORT (2*CHAR_BIT)
27 #define SHORTMASK ((1<<BITSPERSHORT)-1)
28 #define SHORTDN(x) RSHIFT(x,BITSPERSHORT)
29
30 #if SIZEOF_SHORT == SIZEOF_BDIGITS
31 #define SHORTLEN(x) (x)
32 #else
33 static int
34 shortlen(long len, BDIGIT *ds)
35 {
36 BDIGIT num;
37 int offset = 0;
38
39 num = ds[len-1];
40 while (num) {
41 num = SHORTDN(num);
42 offset++;
43 }
44 return (len - 1)*sizeof(BDIGIT)/2 + offset;
45 }
46 #define SHORTLEN(x) shortlen((x),d)
47 #endif
48
49 #define MARSHAL_MAJOR 4
50 #define MARSHAL_MINOR 8
51
52 #define TYPE_NIL '0'
53 #define TYPE_TRUE 'T'
54 #define TYPE_FALSE 'F'
55 #define TYPE_FIXNUM 'i'
56
57 #define TYPE_EXTENDED 'e'
58 #define TYPE_UCLASS 'C'
59 #define TYPE_OBJECT 'o'
60 #define TYPE_DATA 'd'
61 #define TYPE_USERDEF 'u'
62 #define TYPE_USRMARSHAL 'U'
63 #define TYPE_FLOAT 'f'
64 #define TYPE_BIGNUM 'l'
65 #define TYPE_STRING '"'
66 #define TYPE_REGEXP '/'
67 #define TYPE_ARRAY '['
68 #define TYPE_HASH '{'
69 #define TYPE_HASH_DEF '}'
70 #define TYPE_STRUCT 'S'
71 #define TYPE_MODULE_OLD 'M'
72 #define TYPE_CLASS 'c'
73 #define TYPE_MODULE 'm'
74
75 #define TYPE_SYMBOL ':'
76 #define TYPE_SYMLINK ';'
77
78 #define TYPE_IVAR 'I'
79 #define TYPE_LINK '@'
80
81 static ID s_dump, s_load, s_mdump, s_mload;
82 static ID s_dump_data, s_load_data, s_alloc;
83 static ID s_getbyte, s_read, s_write, s_binmode;
84
85 ID rb_id_encoding(void);
86
87 typedef struct {
88 VALUE newclass;
89 VALUE oldclass;
90 VALUE (*dumper)(VALUE);
91 VALUE (*loader)(VALUE, VALUE);
92 } marshal_compat_t;
93
94 static st_table *compat_allocator_tbl;
95 static VALUE compat_allocator_tbl_wrapper;
96
97 static int
98 mark_marshal_compat_i(st_data_t key, st_data_t value)
99 {
100 marshal_compat_t *p = (marshal_compat_t *)value;
101 rb_gc_mark(p->newclass);
102 rb_gc_mark(p->oldclass);
103 return ST_CONTINUE;
104 }
105
106 static void
107 mark_marshal_compat_t(void *tbl)
108 {
109 if (!tbl) return;
110 st_foreach(tbl, mark_marshal_compat_i, 0);
111 }
112
113 void
114 rb_marshal_define_compat(VALUE newclass, VALUE oldclass, VALUE (*dumper)(VALUE), VALUE (*loader)(VALUE, VALUE))
115 {
116 marshal_compat_t *compat;
117 rb_alloc_func_t allocator = rb_get_alloc_func(newclass);
118
119 if (!allocator) {
120 rb_raise(rb_eTypeError, "no allocator");
121 }
122
123 compat = ALLOC(marshal_compat_t);
124 compat->newclass = Qnil;
125 compat->oldclass = Qnil;
126 compat->newclass = newclass;
127 compat->oldclass = oldclass;
128 compat->dumper = dumper;
129 compat->loader = loader;
130
131 st_insert(compat_allocator_tbl, (st_data_t)allocator, (st_data_t)compat);
132 }
133
134 struct dump_arg {
135 VALUE obj;
136 VALUE str, dest;
137 st_table *symbols;
138 st_table *data;
139 int taint;
140 int untrust;
141 st_table *compat_tbl;
142 VALUE wrapper;
143 st_table *encodings;
144 };
145
146 struct dump_call_arg {
147 VALUE obj;
148 struct dump_arg *arg;
149 int limit;
150 };
151
152 static void
153 check_dump_arg(struct dump_arg *arg)
154 {
155 if (!DATA_PTR(arg->wrapper)) {
156 rb_raise(rb_eRuntimeError, "Marshal.dump reentered");
157 }
158 }
159
160 static void
161 mark_dump_arg(void *ptr)
162 {
163 struct dump_arg *p = ptr;
164 if (!ptr)
165 return;
166 rb_mark_set(p->data);
167 rb_mark_hash(p->compat_tbl);
168 }
169
170 static VALUE
171 class2path(VALUE klass)
172 {
173 VALUE path = rb_class_path(klass);
174 char *n = RSTRING_PTR(path);
175
176 if (n[0] == '#') {
177 rb_raise(rb_eTypeError, "can't dump anonymous %s %s",
178 (TYPE(klass) == T_CLASS ? "class" : "module"),
179 n);
180 }
181 if (rb_path2class(n) != rb_class_real(klass)) {
182 rb_raise(rb_eTypeError, "%s can't be referred", n);
183 }
184 return path;
185 }
186
187 static void w_long(long, struct dump_arg*);
188
189 static void
190 w_nbyte(const char *s, int n, struct dump_arg *arg)
191 {
192 VALUE buf = arg->str;
193 rb_str_buf_cat(buf, s, n);
194 if (arg->dest && RSTRING_LEN(buf) >= BUFSIZ) {
195 if (arg->taint) OBJ_TAINT(buf);
196 if (arg->untrust) OBJ_UNTRUST(buf);
197 rb_io_write(arg->dest, buf);
198 rb_str_resize(buf, 0);
199 }
200 }
201
202 static void
203 w_byte(char c, struct dump_arg *arg)
204 {
205 w_nbyte(&c, 1, arg);
206 }
207
208 static void
209 w_bytes(const char *s, int n, struct dump_arg *arg)
210 {
211 w_long(n, arg);
212 w_nbyte(s, n, arg);
213 }
214
215 static void
216 w_short(int x, struct dump_arg *arg)
217 {
218 w_byte((char)((x >> 0) & 0xff), arg);
219 w_byte((char)((x >> 8) & 0xff), arg);
220 }
221
222 static void
223 w_long(long x, struct dump_arg *arg)
224 {
225 char buf[sizeof(long)+1];
226 int i, len = 0;
227
228 #if SIZEOF_LONG > 4
229 if (!(RSHIFT(x, 31) == 0 || RSHIFT(x, 31) == -1)) {
230 /* big long does not fit in 4 bytes */
231 rb_raise(rb_eTypeError, "long too big to dump");
232 }
233 #endif
234
235 if (x == 0) {
236 w_byte(0, arg);
237 return;
238 }
239 if (0 < x && x < 123) {
240 w_byte((char)(x + 5), arg);
241 return;
242 }
243 if (-124 < x && x < 0) {
244 w_byte((char)((x - 5)&0xff), arg);
245 return;
246 }
247 for (i=1;i<sizeof(long)+1;i++) {
248 buf[i] = x & 0xff;
249 x = RSHIFT(x,8);
250 if (x == 0) {
251 buf[0] = i;
252 break;
253 }
254 if (x == -1) {
255 buf[0] = -i;
256 break;
257 }
258 }
259 len = i;
260 for (i=0;i<=len;i++) {
261 w_byte(buf[i], arg);
262 }
263 }
264
265 #ifdef DBL_MANT_DIG
266 #define DECIMAL_MANT (53-16) /* from IEEE754 double precision */
267
268 #if DBL_MANT_DIG > 32
269 #define MANT_BITS 32
270 #elif DBL_MANT_DIG > 24
271 #define MANT_BITS 24
272 #elif DBL_MANT_DIG > 16
273 #define MANT_BITS 16
274 #else
275 #define MANT_BITS 8
276 #endif
277
278 static int
279 save_mantissa(double d, char *buf)
280 {
281 int e, i = 0;
282 unsigned long m;
283 double n;
284
285 d = modf(ldexp(frexp(fabs(d), &e), DECIMAL_MANT), &d);
286 if (d > 0) {
287 buf[i++] = 0;
288 do {
289 d = modf(ldexp(d, MANT_BITS), &n);
290 m = (unsigned long)n;
291 #if MANT_BITS > 24
292 buf[i++] = m >> 24;
293 #endif
294 #if MANT_BITS > 16
295 buf[i++] = m >> 16;
296 #endif
297 #if MANT_BITS > 8
298 buf[i++] = m >> 8;
299 #endif
300 buf[i++] = m;
301 } while (d > 0);
302 while (!buf[i - 1]) --i;
303 }
304 return i;
305 }
306
307 static double
308 load_mantissa(double d, const char *buf, int len)
309 {
310 if (--len > 0 && !*buf++) { /* binary mantissa mark */
311 int e, s = d < 0, dig = 0;
312 unsigned long m;
313
314 modf(ldexp(frexp(fabs(d), &e), DECIMAL_MANT), &d);
315 do {
316 m = 0;
317 switch (len) {
318 default: m = *buf++ & 0xff;
319 #if MANT_BITS > 24
320 case 3: m = (m << 8) | (*buf++ & 0xff);
321 #endif
322 #if MANT_BITS > 16
323 case 2: m = (m << 8) | (*buf++ & 0xff);
324 #endif
325 #if MANT_BITS > 8
326 case 1: m = (m << 8) | (*buf++ & 0xff);
327 #endif
328 }
329 dig -= len < MANT_BITS / 8 ? 8 * (unsigned)len : MANT_BITS;
330 d += ldexp((double)m, dig);
331 } while ((len -= MANT_BITS / 8) > 0);
332 d = ldexp(d, e - DECIMAL_MANT);
333 if (s) d = -d;
334 }
335 return d;
336 }
337 #else
338 #define load_mantissa(d, buf, len) (d)
339 #define save_mantissa(d, buf) 0
340 #endif
341
342 #ifdef DBL_DIG
343 #define FLOAT_DIG (DBL_DIG+2)
344 #else
345 #define FLOAT_DIG 17
346 #endif
347
348 static void
349 w_float(double d, struct dump_arg *arg)
350 {
351 char buf[FLOAT_DIG + (DECIMAL_MANT + 7) / 8 + 10];
352
353 if (isinf(d)) {
354 if (d < 0) strcpy(buf, "-inf");
355 else strcpy(buf, "inf");
356 }
357 else if (isnan(d)) {
358 strcpy(buf, "nan");
359 }
360 else if (d == 0.0) {
361 if (1.0/d < 0) strcpy(buf, "-0");
362 else strcpy(buf, "0");
363 }
364 else {
365 int len;
366
367 /* xxx: should not use system's sprintf(3) */
368 snprintf(buf, sizeof(buf), "%.*g", FLOAT_DIG, d);
369 len = strlen(buf);
370 w_bytes(buf, len + save_mantissa(d, buf + len), arg);
371 return;
372 }
373 w_bytes(buf, strlen(buf), arg);
374 }
375
376 static void
377 w_symbol(ID id, struct dump_arg *arg)
378 {
379 const char *sym;
380 st_data_t num;
381
382 if (st_lookup(arg->symbols, id, &num)) {
383 w_byte(TYPE_SYMLINK, arg);
384 w_long((long)num, arg);
385 }
386 else {
387 sym = rb_id2name(id);
388 if (!sym) {
389 rb_raise(rb_eTypeError, "can't dump anonymous ID %ld", id);
390 }
391 w_byte(TYPE_SYMBOL, arg);
392 w_bytes(sym, strlen(sym), arg);
393 st_add_direct(arg->symbols, id, arg->symbols->num_entries);
394 }
395 }
396
397 static void
398 w_unique(const char *s, struct dump_arg *arg)
399 {
400 if (s[0] == '#') {
401 rb_raise(rb_eTypeError, "can't dump anonymous class %s", s);
402 }
403 w_symbol(rb_intern(s), arg);
404 }
405
406 static void w_object(VALUE,struct dump_arg*,int);
407
408 static int
409 hash_each(VALUE key, VALUE value, struct dump_call_arg *arg)
410 {
411 w_object(key, arg->arg, arg->limit);
412 w_object(value, arg->arg, arg->limit);
413 return ST_CONTINUE;
414 }
415
416 static void
417 w_extended(VALUE klass, struct dump_arg *arg, int check)
418 {
419 const char *path;
420
421 if (check && FL_TEST(klass, FL_SINGLETON)) {
422 if (RCLASS_M_TBL(klass)->num_entries ||
423 (RCLASS_IV_TBL(klass) && RCLASS_IV_TBL(klass)->num_entries > 1)) {
424 rb_raise(rb_eTypeError, "singleton can't be dumped");
425 }
426 klass = RCLASS_SUPER(klass);
427 }
428 while (BUILTIN_TYPE(klass) == T_ICLASS) {
429 path = rb_class2name(RBASIC(klass)->klass);
430 w_byte(TYPE_EXTENDED, arg);
431 w_unique(path, arg);
432 klass = RCLASS_SUPER(klass);
433 }
434 }
435
436 static void
437 w_class(char type, VALUE obj, struct dump_arg *arg, int check)
438 {
439 volatile VALUE p;
440 char *path;
441 st_data_t real_obj;
442 VALUE klass;
443
444 if (st_lookup(arg->compat_tbl, (st_data_t)obj, &real_obj)) {
445 obj = (VALUE)real_obj;
446 }
447 klass = CLASS_OF(obj);
448 w_extended(klass, arg, check);
449 w_byte(type, arg);
450 p = class2path(rb_class_real(klass));
451 path = RSTRING_PTR(p);
452 w_unique(path, arg);
453 }
454
455 static void
456 w_uclass(VALUE obj, VALUE super, struct dump_arg *arg)
457 {
458 VALUE klass = CLASS_OF(obj);
459
460 w_extended(klass, arg, Qtrue);
461 klass = rb_class_real(klass);
462 if (klass != super) {
463 w_byte(TYPE_UCLASS, arg);
464 w_unique(RSTRING_PTR(class2path(klass)), arg);
465 }
466 }
467
468 static int
469 w_obj_each(ID id, VALUE value, struct dump_call_arg *arg)
470 {
471 if (id == rb_id_encoding()) return ST_CONTINUE;
472 w_symbol(id, arg->arg);
473 w_object(value, arg->arg, arg->limit);
474 return ST_CONTINUE;
475 }
476
477 static void
478 w_encoding(VALUE obj, long num, struct dump_call_arg *arg)
479 {
480 int encidx = rb_enc_get_index(obj);
481 rb_encoding *enc = 0;
482 st_data_t name;
483
484 if (encidx <= 0 || !(enc = rb_enc_from_index(encidx))) {
485 w_long(num, arg->arg);
486 return;
487 }
488 w_long(num + 1, arg->arg);
489 w_symbol(rb_id_encoding(), arg->arg);
490 do {
491 if (!arg->arg->encodings)
492 arg->arg->encodings = st_init_strcasetable();
493 else if (st_lookup(arg->arg->encodings, (st_data_t)rb_enc_name(enc), &name))
494 break;
495 name = (st_data_t)rb_str_new2(rb_enc_name(enc));
496 st_insert(arg->arg->encodings, (st_data_t)rb_enc_name(enc), name);
497 } while (0);
498 w_object(name, arg->arg, arg->limit);
499 }
500
501 static void
502 w_ivar(VALUE obj, st_table *tbl, struct dump_call_arg *arg)
503 {
504 long num = tbl ? tbl->num_entries : 0;
505
506 w_encoding(obj, num, arg);
507 if (tbl) {
508 st_foreach_safe(tbl, w_obj_each, (st_data_t)arg);
509 }
510 }
511
512 static void
513 w_objivar(VALUE obj, struct dump_call_arg *arg)
514 {
515 VALUE *ptr;
516 long i, len, num;
517
518 len = ROBJECT_NUMIV(obj);
519 ptr = ROBJECT_IVPTR(obj);
520 num = 0;
521 for (i = 0; i < len; i++)
522 if (ptr[i] != Qundef)
523 num += 1;
524
525 w_encoding(obj, num, arg);
526 if (num != 0) {
527 rb_ivar_foreach(obj, w_obj_each, (st_data_t)arg);
528 }
529 }
530
531 static void
532 w_object(VALUE obj, struct dump_arg *arg, int limit)
533 {
534 struct dump_call_arg c_arg;
535 st_table *ivtbl = 0;
536 st_data_t num;
537 int hasiv = 0;
538 #define has_ivars(obj, ivtbl) ((ivtbl = rb_generic_ivar_table(obj)) != 0 || \
539 (!SPECIAL_CONST_P(obj) && !ENCODING_IS_ASCII8BIT(obj)))
540
541 if (limit == 0) {
542 rb_raise(rb_eArgError, "exceed depth limit");
543 }
544
545 limit--;
546 c_arg.limit = limit;
547 c_arg.arg = arg;
548
549 if (st_lookup(arg->data, obj, &num)) {
550 w_byte(TYPE_LINK, arg);
551 w_long((long)num, arg);
552 return;
553 }
554
555 if ((hasiv = has_ivars(obj, ivtbl)) != 0) {
556 w_byte(TYPE_IVAR, arg);
557 }
558 if (obj == Qnil) {
559 w_byte(TYPE_NIL, arg);
560 }
561 else if (obj == Qtrue) {
562 w_byte(TYPE_TRUE, arg);
563 }
564 else if (obj == Qfalse) {
565 w_byte(TYPE_FALSE, arg);
566 }
567 else if (FIXNUM_P(obj)) {
568 #if SIZEOF_LONG <= 4
569 w_byte(TYPE_FIXNUM, arg);
570 w_long(FIX2INT(obj), arg);
571 #else
572 if (RSHIFT((long)obj, 31) == 0 || RSHIFT((long)obj, 31) == -1) {
573 w_byte(TYPE_FIXNUM, arg);
574 w_long(FIX2LONG(obj), arg);
575 }
576 else {
577 w_object(rb_int2big(FIX2LONG(obj)), arg, limit);
578 }
579 #endif
580 }
581 else if (SYMBOL_P(obj)) {
582 w_symbol(SYM2ID(obj), arg);
583 }
584 else {
585 if (OBJ_TAINTED(obj)) arg->taint = Qtrue;
586 if (OBJ_UNTRUSTED(obj)) arg->untrust = Qtrue;
587
588 if (rb_respond_to(obj, s_mdump)) {
589 volatile VALUE v;
590
591 st_add_direct(arg->data, obj, arg->data->num_entries);
592
593 v = rb_funcall(obj, s_mdump, 0, 0);
594 check_dump_arg(arg);
595 w_class(TYPE_USRMARSHAL, obj, arg, Qfalse);
596 w_object(v, arg, limit);
597 if (hasiv) w_ivar(obj, 0, &c_arg);
598 return;
599 }
600 if (rb_respond_to(obj, s_dump)) {
601 VALUE v;
602 st_table *ivtbl2 = 0;
603 int hasiv2;
604
605 v = rb_funcall(obj, s_dump, 1, INT2NUM(limit));
606 check_dump_arg(arg);
607 if (TYPE(v) != T_STRING) {
608 rb_raise(rb_eTypeError, "_dump() must return string");
609 }
610 if ((hasiv2 = has_ivars(v, ivtbl2)) != 0 && !hasiv) {
611 w_byte(TYPE_IVAR, arg);
612 }
613 w_class(TYPE_USERDEF, obj, arg, Qfalse);
614 w_bytes(RSTRING_PTR(v), RSTRING_LEN(v), arg);
615 if (hasiv2) {
616 w_ivar(v, ivtbl2, &c_arg);
617 }
618 else if (hasiv) {
619 w_ivar(obj, ivtbl, &c_arg);
620 }
621 st_add_direct(arg->data, obj, arg->data->num_entries);
622 return;
623 }
624
625 st_add_direct(arg->data, obj, arg->data->num_entries);
626
627 {
628 st_data_t compat_data;
629 rb_alloc_func_t allocator = rb_get_alloc_func(RBASIC(obj)->klass);
630 if (st_lookup(compat_allocator_tbl,
631 (st_data_t)allocator,
632 &compat_data)) {
633 marshal_compat_t *compat = (marshal_compat_t*)compat_data;
634 VALUE real_obj = obj;
635 obj = compat->dumper(real_obj);
636 st_insert(arg->compat_tbl, (st_data_t)obj, (st_data_t)real_obj);
637 }
638 }
639
640 switch (BUILTIN_TYPE(obj)) {
641 case T_CLASS:
642 if (FL_TEST(obj, FL_SINGLETON)) {
643 rb_raise(rb_eTypeError, "singleton class can't be dumped");
644 }
645 w_byte(TYPE_CLASS, arg);
646 {
647 volatile VALUE path = class2path(obj);
648 w_bytes(RSTRING_PTR(path), RSTRING_LEN(path), arg);
649 }
650 break;
651
652 case T_MODULE:
653 w_byte(TYPE_MODULE, arg);
654 {
655 VALUE path = class2path(obj);
656 w_bytes(RSTRING_PTR(path), RSTRING_LEN(path), arg);
657 }
658 break;
659
660 case T_FLOAT:
661 w_byte(TYPE_FLOAT, arg);
662 w_float(RFLOAT_VALUE(obj), arg);
663 break;
664
665 case T_BIGNUM:
666 w_byte(TYPE_BIGNUM, arg);
667 {
668 char sign = RBIGNUM_SIGN(obj) ? '+' : '-';
669 long len = RBIGNUM_LEN(obj);
670 BDIGIT *d = RBIGNUM_DIGITS(obj);
671
672 w_byte(sign, arg);
673 w_long(SHORTLEN(len), arg); /* w_short? */
674 while (len--) {
675 #if SIZEOF_BDIGITS > SIZEOF_SHORT
676 BDIGIT num = *d;
677 int i;
678
679 for (i=0; i<SIZEOF_BDIGITS; i+=SIZEOF_SHORT) {
680 w_short(num & SHORTMASK, arg);
681 num = SHORTDN(num);
682 if (len == 0 && num == 0) break;
683 }
684 #else
685 w_short(*d, arg);
686 #endif
687 d++;
688 }
689 }
690 break;
691
692 case T_STRING:
693 w_uclass(obj, rb_cString, arg);
694 w_byte(TYPE_STRING, arg);
695 w_bytes(RSTRING_PTR(obj), RSTRING_LEN(obj), arg);
696 break;
697
698 case T_REGEXP:
699 w_uclass(obj, rb_cRegexp, arg);
700 w_byte(TYPE_REGEXP, arg);
701 {
702 int opts = rb_reg_options(obj);
703 w_bytes(RREGEXP_SRC_PTR(obj), RREGEXP_SRC_LEN(obj), arg);
704 w_byte((char)opts, arg);
705 }
706 break;
707
708 case T_ARRAY:
709 w_uclass(obj, rb_cArray, arg);
710 w_byte(TYPE_ARRAY, arg);
711 {
712 long i, len = RARRAY_LEN(obj);
713
714 w_long(len, arg);
715 for (i=0; i<RARRAY_LEN(obj); i++) {
716 w_object(RARRAY_PTR(obj)[i], arg, limit);
717 if (len != RARRAY_LEN(obj)) {
718 rb_raise(rb_eRuntimeError, "array modified during dump");
719 }
720 }
721 }
722 break;
723
724 case T_HASH:
725 w_uclass(obj, rb_cHash, arg);
726 if (NIL_P(RHASH(obj)->ifnone)) {
727 w_byte(TYPE_HASH, arg);
728 }
729 else if (FL_TEST(obj, FL_USER2)) {
730 /* FL_USER2 means HASH_PROC_DEFAULT (see hash.c) */
731 rb_raise(rb_eTypeError, "can't dump hash with default proc");
732 }
733 else {
734 w_byte(TYPE_HASH_DEF, arg);
735 }
736 w_long(RHASH_SIZE(obj), arg);
737 rb_hash_foreach(obj, hash_each, (st_data_t)&c_arg);
738 if (!NIL_P(RHASH(obj)->ifnone)) {
739 w_object(RHASH(obj)->ifnone, arg, limit);
740 }
741 break;
742
743 case T_STRUCT:
744 w_class(TYPE_STRUCT, obj, arg, Qtrue);
745 {
746 long len = RSTRUCT_LEN(obj);
747 VALUE mem;
748 long i;
749
750 w_long(len, arg);
751 mem = rb_struct_members(obj);
752 for (i=0; i<len; i++) {
753 w_symbol(SYM2ID(RARRAY_PTR(mem)[i]), arg);
754 w_object(RSTRUCT_PTR(obj)[i], arg, limit);
755 }
756 }
757 break;
758
759 case T_OBJECT:
760 w_class(TYPE_OBJECT, obj, arg, Qtrue);
761 w_objivar(obj, &c_arg);
762 break;
763
764 case T_DATA:
765 {
766 VALUE v;
767
768 if (!rb_respond_to(obj, s_dump_data)) {
769 rb_raise(rb_eTypeError,
770 "no marshal_dump is defined for class %s",
771 rb_obj_classname(obj));
772 }
773 v = rb_funcall(obj, s_dump_data, 0);
774 check_dump_arg(arg);
775 w_class(TYPE_DATA, obj, arg, Qtrue);
776 w_object(v, arg, limit);
777 }
778 break;
779
780 default:
781 rb_raise(rb_eTypeError, "can't dump %s",
782 rb_obj_classname(obj));
783 break;
784 }
785 }
786 if (hasiv) {
787 w_ivar(obj, ivtbl, &c_arg);
788 }
789 }
790
791 static VALUE
792 dump(struct dump_call_arg *arg)
793 {
794 w_object(arg->obj, arg->arg, arg->limit);
795 if (arg->arg->dest) {
796 rb_io_write(arg->arg->dest, arg->arg->str);
797 rb_str_resize(arg->arg->str, 0);
798 }
799 return 0;
800 }
801
802 static VALUE
803 dump_ensure(struct dump_arg *arg)
804 {
805 if (!DATA_PTR(arg->wrapper)) return 0;
806 st_free_table(arg->symbols);
807 st_free_table(arg->data);
808 st_free_table(arg->compat_tbl);
809 if (arg->encodings) st_free_table(arg->encodings);
810 DATA_PTR(arg->wrapper) = 0;
811 arg->wrapper = 0;
812 if (arg->taint) {
813 OBJ_TAINT(arg->str);
814 }
815 if (arg->untrust) {
816 OBJ_UNTRUST(arg->str);
817 }
818 return 0;
819 }
820
821 /*
822 * call-seq:
823 * dump( obj [, anIO] , limit=--1 ) => anIO
824 *
825 * Serializes obj and all descendent objects. If anIO is
826 * specified, the serialized data will be written to it, otherwise the
827 * data will be returned as a String. If limit is specified, the
828 * traversal of subobjects will be limited to that depth. If limit is
829 * negative, no checking of depth will be performed.
830 *
831 * class Klass
832 * def initialize(str)
833 * @str = str
834 * end
835 * def sayHello
836 * @str
837 * end
838 * end
839 *
840 * (produces no output)
841 *
842 * o = Klass.new("hello\n")
843 * data = Marshal.dump(o)
844 * obj = Marshal.load(data)
845 * obj.sayHello #=> "hello\n"
846 */
847 static VALUE
848 marshal_dump(int argc, VALUE *argv)
849 {
850 VALUE obj, port, a1, a2;
851 int limit = -1;
852 struct dump_arg arg;
853 struct dump_call_arg c_arg;
854
855 port = Qnil;
856 rb_scan_args(argc, argv, "12", &obj, &a1, &a2);
857 if (argc == 3) {
858 if (!NIL_P(a2)) limit = NUM2INT(a2);
859 if (NIL_P(a1)) goto type_error;
860 port = a1;
861 }
862 else if (argc == 2) {
863 if (FIXNUM_P(a1)) limit = FIX2INT(a1);
864 else if (NIL_P(a1)) goto type_error;
865 else port = a1;
866 }
867 arg.dest = 0;
868 if (!NIL_P(port)) {
869 if (!rb_respond_to(port, s_write)) {
870 type_error:
871 rb_raise(rb_eTypeError, "instance of IO needed");
872 }
873 arg.str = rb_str_buf_new(0);
874 arg.dest = port;
875 if (rb_respond_to(port, s_binmode)) {
876 rb_funcall2(port, s_binmode, 0, 0);
877 }
878 }
879 else {
880 port = rb_str_buf_new(0);
881 arg.str = port;
882 }
883
884 arg.symbols = st_init_numtable();
885 arg.data = st_init_numtable();
886 arg.taint = Qfalse;
887 arg.untrust = Qfalse;
888 arg.compat_tbl = st_init_numtable();
889 arg.wrapper = Data_Wrap_Struct(rb_cData, mark_dump_arg, 0, &arg);
890 arg.encodings = 0;
891 c_arg.obj = obj;
892 c_arg.arg = &arg;
893 c_arg.limit = limit;
894
895 w_byte(MARSHAL_MAJOR, &arg);
896 w_byte(MARSHAL_MINOR, &arg);
897
898 rb_ensure(dump, (VALUE)&c_arg, dump_ensure, (VALUE)&arg);
899
900 return port;
901 }
902
903 struct load_arg {
904 VALUE src;
905 long offset;
906 st_table *symbols;
907 VALUE data;
908 VALUE proc;
909 int taint;
910 int untrust;
911 st_table *compat_tbl;
912 VALUE compat_tbl_wrapper;
913 };
914
915 static void
916 check_load_arg(struct load_arg *arg)
917 {
918 if (!DATA_PTR(arg->compat_tbl_wrapper)) {
919 rb_raise(rb_eRuntimeError, "Marshal.load reentered");
920 }
921 }
922
923 static VALUE r_entry(VALUE v, struct load_arg *arg);
924 static VALUE r_object(struct load_arg *arg);
925 static VALUE path2class(const char *path);
926
927 static int
928 r_byte(struct load_arg *arg)
929 {
930 int c;
931
932 if (TYPE(arg->src) == T_STRING) {
933 if (RSTRING_LEN(arg->src) > arg->offset) {
934 c = (unsigned char)RSTRING_PTR(arg->src)[arg->offset++];
935 }
936 else {
937 rb_raise(rb_eArgError, "marshal data too short");
938 }
939 }
940 else {
941 VALUE src = arg->src;
942 VALUE v = rb_funcall2(src, s_getbyte, 0, 0);
943 check_load_arg(arg);
944 if (NIL_P(v)) rb_eof_error();
945 c = (unsigned char)NUM2CHR(v);
946 }
947 return c;
948 }
949
950 static void
951 long_toobig(int size)
952 {
953 rb_raise(rb_eTypeError, "long too big for this architecture (size "
954 STRINGIZE(SIZEOF_LONG)", given %d)", size);
955 }
956
957 #undef SIGN_EXTEND_CHAR
958 #if __STDC__
959 # define SIGN_EXTEND_CHAR(c) ((signed char)(c))
960 #else /* not __STDC__ */
961 /* As in Harbison and Steele. */
962 # define SIGN_EXTEND_CHAR(c) ((((unsigned char)(c)) ^ 128) - 128)
963 #endif
964
965 static long
966 r_long(struct load_arg *arg)
967 {
968 register long x;
969 int c = SIGN_EXTEND_CHAR(r_byte(arg));
970 long i;
971
972 if (c == 0) return 0;
973 if (c > 0) {
974 if (4 < c && c < 128) {
975 return c - 5;
976 }
977 if (c > sizeof(long)) long_toobig(c);
978 x = 0;
979 for (i=0;i<c;i++) {
980 x |= (long)r_byte(arg) << (8*i);
981 }
982 }
983 else {
984 if (-129 < c && c < -4) {
985 return c + 5;
986 }
987 c = -c;
988 if (c > sizeof(long)) long_toobig(c);
989 x = -1;
990 for (i=0;i<c;i++) {
991 x &= ~((long)0xff << (8*i));
992 x |= (long)r_byte(arg) << (8*i);
993 }
994 }
995 return x;
996 }
997
998 #define r_bytes(arg) r_bytes0(r_long(arg), (arg))
999
1000 static VALUE
1001 r_bytes0(long len, struct load_arg *arg)
1002 {
1003 VALUE str;
1004
1005 if (len == 0) return rb_str_new(0, 0);
1006 if (TYPE(arg->src) == T_STRING) {
1007 if (RSTRING_LEN(arg->src) - arg->offset >= len) {
1008 str = rb_str_new(RSTRING_PTR(arg->src)+arg->offset, len);
1009 arg->offset += len;
1010 }
1011 else {
1012 too_short:
1013 rb_raise(rb_eArgError, "marshal data too short");
1014 }
1015 }
1016 else {
1017 VALUE src = arg->src;
1018 VALUE n = LONG2NUM(len);
1019 str = rb_funcall2(src, s_read, 1, &n);
1020 check_load_arg(arg);
1021 if (NIL_P(str)) goto too_short;
1022 StringValue(str);
1023 if (RSTRING_LEN(str) != len) goto too_short;
1024 if (OBJ_TAINTED(str)) arg->taint = Qtrue;
1025 if (OBJ_UNTRUSTED(str)) arg->untrust = Qtrue;
1026 }
1027 return str;
1028 }
1029
1030 static ID
1031 r_symlink(struct load_arg *arg)
1032 {
1033 ID id;
1034 long num = r_long(arg);
1035
1036 if (st_lookup(arg->symbols, num, &id)) {
1037 return id;
1038 }
1039 rb_raise(rb_eArgError, "bad symbol");
1040 }
1041
1042 static ID
1043 r_symreal(struct load_arg *arg)
1044 {
1045 volatile VALUE s = r_bytes(arg);
1046 ID id = rb_intern(RSTRING_PTR(s));
1047
1048 st_insert(arg->symbols, arg->symbols->num_entries, id);
1049
1050 return id;
1051 }
1052
1053 static ID
1054 r_symbol(struct load_arg *arg)
1055 {
1056 int type;
1057
1058 switch ((type = r_byte(arg))) {
1059 case TYPE_SYMBOL:
1060 return r_symreal(arg);
1061 case TYPE_SYMLINK:
1062 return r_symlink(arg);
1063 default:
1064 rb_raise(rb_eArgError, "dump format error(0x%x)", type);
1065 break;
1066 }
1067 }
1068
1069 static const char*
1070 r_unique(struct load_arg *arg)
1071 {
1072 return rb_id2name(r_symbol(arg));
1073 }
1074
1075 static VALUE
1076 r_string(struct load_arg *arg)
1077 {
1078 return r_bytes(arg);
1079 }
1080
1081 static VALUE
1082 r_entry(VALUE v, struct load_arg *arg)
1083 {
1084 st_data_t real_obj = (VALUE)Qundef;
1085 if (st_lookup(arg->compat_tbl, v, &real_obj)) {
1086 rb_hash_aset(arg->data, INT2FIX(RHASH_SIZE(arg->data)), (VALUE)real_obj);
1087 }
1088 else {
1089 rb_hash_aset(arg->data, INT2FIX(RHASH_SIZE(arg->data)), v);
1090 }
1091 if (arg->taint) {
1092 OBJ_TAINT(v);
1093 if ((VALUE)real_obj != Qundef)
1094 OBJ_TAINT((VALUE)real_obj);
1095 }
1096 if (arg->untrust) {
1097 OBJ_UNTRUST(v);
1098 if ((VALUE)real_obj != Qundef)
1099 OBJ_UNTRUST((VALUE)real_obj);
1100 }
1101 return v;
1102 }
1103
1104 static VALUE
1105 r_leave(VALUE v, struct load_arg *arg)
1106 {
1107 st_data_t data;
1108 if (st_lookup(arg->compat_tbl, v, &data)) {
1109 VALUE real_obj = (VALUE)data;
1110 rb_alloc_func_t allocator = rb_get_alloc_func(CLASS_OF(real_obj));
1111 st_data_t key = v;
1112 if (st_lookup(compat_allocator_tbl, (st_data_t)allocator, &data)) {
1113 marshal_compat_t *compat = (marshal_compat_t*)data;
1114 compat->loader(real_obj, v);
1115 }
1116 st_delete(arg->compat_tbl, &key, 0);
1117 v = real_obj;
1118 }
1119 if (arg->proc) {
1120 v = rb_funcall(arg->proc, rb_intern("call"), 1, v);
1121 check_load_arg(arg);
1122 }
1123 return v;
1124 }
1125
1126 static void
1127 r_ivar(VALUE obj, struct load_arg *arg)
1128 {
1129 long len;
1130
1131 len = r_long(arg);
1132 if (len > 0) {
1133 while (len--) {
1134 ID id = r_symbol(arg);
1135 VALUE val = r_object(arg);
1136 if (id == rb_id_encoding()) {
1137 int idx = rb_enc_find_index(StringValueCStr(val));
1138 if (idx > 0) rb_enc_associate_index(obj, idx);
1139 }
1140 else {
1141 rb_ivar_set(obj, id, val);
1142 }
1143 }
1144 }
1145 }
1146
1147 static VALUE
1148 path2class(const char *path)
1149 {
1150 VALUE v = rb_path2class(path);
1151
1152 if (TYPE(v) != T_CLASS) {
1153 rb_raise(rb_eArgError, "%s does not refer class", path);
1154 }
1155 return v;
1156 }
1157
1158 static VALUE
1159 path2module(const char *path)
1160 {
1161 VALUE v = rb_path2class(path);
1162
1163 if (TYPE(v) != T_MODULE) {
1164 rb_raise(rb_eArgError, "%s does not refer module", path);
1165 }
1166 return v;
1167 }
1168
1169 static VALUE
1170 obj_alloc_by_path(const char *path, struct load_arg *arg)
1171 {
1172 VALUE klass;
1173 st_data_t data;
1174 rb_alloc_func_t allocator;
1175
1176 klass = path2class(path);
1177
1178 allocator = rb_get_alloc_func(klass);
1179 if (st_lookup(compat_allocator_tbl, (st_data_t)allocator, &data)) {
1180 marshal_compat_t *compat = (marshal_compat_t*)data;
1181 VALUE real_obj = rb_obj_alloc(klass);
1182 VALUE obj = rb_obj_alloc(compat->oldclass);
1183 st_insert(arg->compat_tbl, (st_data_t)obj, (st_data_t)real_obj);
1184 return obj;
1185 }
1186
1187 return rb_obj_alloc(klass);
1188 }
1189
1190 static VALUE
1191 r_object0(struct load_arg *arg, int *ivp, VALUE extmod)
1192 {
1193 VALUE v = Qnil;
1194 int type = r_byte(arg);
1195 long id;
1196
1197 switch (type) {
1198 case TYPE_LINK:
1199 id = r_long(arg);
1200 v = rb_hash_aref(arg->data, LONG2FIX(id));
1201 check_load_arg(arg);
1202 if (NIL_P(v)) {
1203 rb_raise(rb_eArgError, "dump format error (unlinked)");
1204 }
1205 if (arg->proc) {
1206 v = rb_funcall(arg->proc, rb_intern("call"), 1, v);
1207 check_load_arg(arg);
1208 }
1209 break;
1210
1211 case TYPE_IVAR:
1212 {
1213 int ivar = Qtrue;
1214
1215 v = r_object0(arg, &ivar, extmod);
1216 if (ivar) r_ivar(v, arg);
1217 }
1218 break;
1219
1220 case TYPE_EXTENDED:
1221 {
1222 VALUE m = path2module(r_unique(arg));
1223
1224 if (NIL_P(extmod)) extmod = rb_ary_new2(0);
1225 rb_ary_push(extmod, m);
1226
1227 v = r_object0(arg, 0, extmod);
1228 while (RARRAY_LEN(extmod) > 0) {
1229 m = rb_ary_pop(extmod);
1230 rb_extend_object(v, m);
1231 }
1232 }
1233 break;
1234
1235 case TYPE_UCLASS:
1236 {
1237 VALUE c = path2class(r_unique(arg));
1238
1239 if (FL_TEST(c, FL_SINGLETON)) {
1240 rb_raise(rb_eTypeError, "singleton can't be loaded");
1241 }
1242 v = r_object0(arg, 0, extmod);
1243 if (rb_special_const_p(v) || TYPE(v) == T_OBJECT || TYPE(v) == T_CLASS) {
1244 format_error:
1245 rb_raise(rb_eArgError, "dump format error (user class)");
1246 }
1247 if (TYPE(v) == T_MODULE || !RTEST(rb_class_inherited_p(c, RBASIC(v)->klass))) {
1248 VALUE tmp = rb_obj_alloc(c);
1249
1250 if (TYPE(v) != TYPE(tmp)) goto format_error;
1251 }
1252 RBASIC(v)->klass = c;
1253 }
1254 break;
1255
1256 case TYPE_NIL:
1257 v = Qnil;
1258 v = r_leave(v, arg);
1259 break;
1260
1261 case TYPE_TRUE:
1262 v = Qtrue;
1263 v = r_leave(v, arg);
1264 break;
1265
1266 case TYPE_FALSE:
1267 v = Qfalse;
1268 v = r_leave(v, arg);
1269 break;
1270
1271 case TYPE_FIXNUM:
1272 {
1273 long i = r_long(arg);
1274 v = LONG2FIX(i);
1275 }
1276 v = r_leave(v, arg);
1277 break;
1278
1279 case TYPE_FLOAT:
1280 {
1281 double d, t = 0.0;
1282 VALUE str = r_bytes(arg);
1283 const char *ptr = RSTRING_PTR(str);
1284
1285 if (strcmp(ptr, "nan") == 0) {
1286 d = t / t;
1287 }
1288 else if (strcmp(ptr, "inf") == 0) {
1289 d = 1.0 / t;
1290 }
1291 else if (strcmp(ptr, "-inf") == 0) {
1292 d = -1.0 / t;
1293 }
1294 else {
1295 char *e;
1296 d = strtod(ptr, &e);
1297 d = load_mantissa(d, e, RSTRING_LEN(str) - (e - ptr));
1298 }
1299 v = DOUBLE2NUM(d);
1300 v = r_entry(v, arg);
1301 v = r_leave(v, arg);
1302 }
1303 break;
1304
1305 case TYPE_BIGNUM:
1306 {
1307 long len;
1308 BDIGIT *digits;
1309 volatile VALUE data;
1310
1311 NEWOBJ(big, struct RBignum);
1312 OBJSETUP(big, rb_cBignum, T_BIGNUM);
1313 RBIGNUM_SET_SIGN(big, (r_byte(arg) == '+'));
1314 len = r_long(arg);
1315 data = r_bytes0(len * 2, arg);
1316 #if SIZEOF_BDIGITS == SIZEOF_SHORT
1317 rb_big_resize((VALUE)big, len);
1318 #else
1319 rb_big_resize((VALUE)big, (len + 1) * 2 / sizeof(BDIGIT));
1320 #endif
1321 digits = RBIGNUM_DIGITS(big);
1322 MEMCPY(digits, RSTRING_PTR(data), char, len * 2);
1323 #if SIZEOF_BDIGITS > SIZEOF_SHORT
1324 MEMZERO((char *)digits + len * 2, char,
1325 RBIGNUM_LEN(big) * sizeof(BDIGIT) - len * 2);
1326 #endif
1327 len = RBIGNUM_LEN(big);
1328 while (len > 0) {
1329 unsigned char *p = (unsigned char *)digits;
1330 BDIGIT num = 0;
1331 #if SIZEOF_BDIGITS > SIZEOF_SHORT
1332 int shift = 0;
1333 int i;
1334
1335 for (i=0; i<SIZEOF_BDIGITS; i++) {
1336 num |= (int)p[i] << shift;
1337 shift += 8;
1338 }
1339 #else
1340 num = p[0] | (p[1] << 8);
1341 #endif
1342 *digits++ = num;
1343 len--;
1344 }
1345 v = rb_big_norm((VALUE)big);
1346 v = r_entry(v, arg);
1347 v = r_leave(v, arg);
1348 }
1349 break;
1350
1351 case TYPE_STRING:
1352 v = r_entry(r_string(arg), arg);
1353 v = r_leave(v, arg);
1354 break;
1355
1356 case TYPE_REGEXP:
1357 {
1358 volatile VALUE str = r_bytes(arg);
1359 int options = r_byte(arg);
1360 v = r_entry(rb_reg_new_str(str, options), arg);
1361 v = r_leave(v, arg);
1362 }
1363 break;
1364
1365 case TYPE_ARRAY:
1366 {
1367 volatile long len = r_long(arg); /* gcc 2.7.2.3 -O2 bug?? */
1368
1369 v = rb_ary_new2(len);
1370 v = r_entry(v, arg);
1371 while (len--) {
1372 rb_ary_push(v, r_object(arg));
1373 }
1374 v = r_leave(v, arg);
1375 }
1376 break;
1377
1378 case TYPE_HASH:
1379 case TYPE_HASH_DEF:
1380 {
1381 long len = r_long(arg);
1382
1383 v = rb_hash_new();
1384 v = r_entry(v, arg);
1385 while (len--) {
1386 VALUE key = r_object(arg);
1387 VALUE value = r_object(arg);
1388 rb_hash_aset(v, key, value);
1389 }
1390 if (type == TYPE_HASH_DEF) {
1391 RHASH(v)->ifnone = r_object(arg);
1392 }
1393 v = r_leave(v, arg);
1394 }
1395 break;
1396
1397 case TYPE_STRUCT:
1398 {
1399 VALUE klass, mem;
1400 VALUE values;
1401 volatile long i; /* gcc 2.7.2.3 -O2 bug?? */
1402 long len;
1403 ID slot;
1404
1405 klass = path2class(r_unique(arg));
1406 len = r_long(arg);
1407
1408 v = rb_obj_alloc(klass);
1409 if (TYPE(v) != T_STRUCT) {
1410 rb_raise(rb_eTypeError, "class %s not a struct", rb_class2name(klass));
1411 }
1412 mem = rb_struct_s_members(klass);
1413 if (RARRAY_LEN(mem) != len) {
1414 rb_raise(rb_eTypeError, "struct %s not compatible (struct size differs)",
1415 rb_class2name(klass));
1416 }
1417
1418 v = r_entry(v, arg);
1419 values = rb_ary_new2(len);
1420 for (i=0; i<len; i++) {
1421 slot = r_symbol(arg);
1422
1423 if (RARRAY_PTR(mem)[i] != ID2SYM(slot)) {
1424 rb_raise(rb_eTypeError, "struct %s not compatible (:%s for :%s)",
1425 rb_class2name(klass),
1426 rb_id2name(slot),
1427 rb_id2name(SYM2ID(RARRAY_PTR(mem)[i])));
1428 }
1429 rb_ary_push(values, r_object(arg));
1430 }
1431 rb_struct_initialize(v, values);
1432 v = r_leave(v, arg);
1433 }
1434 break;
1435
1436 case TYPE_USERDEF:
1437 {
1438 VALUE klass = path2class(r_unique(arg));
1439 VALUE data;
1440
1441 if (!rb_respond_to(klass, s_load)) {
1442 rb_raise(rb_eTypeError, "class %s needs to have method `_load'",
1443 rb_class2name(klass));
1444 }
1445 data = r_string(arg);
1446 if (ivp) {
1447 r_ivar(data, arg);
1448 *ivp = Qfalse;
1449 }
1450 v = rb_funcall(klass, s_load, 1, data);
1451 check_load_arg(arg);
1452 v = r_entry(v, arg);
1453 v = r_leave(v, arg);
1454 }
1455 break;
1456
1457 case TYPE_USRMARSHAL:
1458 {
1459 VALUE klass = path2class(r_unique(arg));
1460 VALUE data;
1461
1462 v = rb_obj_alloc(klass);
1463 if (!NIL_P(extmod)) {
1464 while (RARRAY_LEN(extmod) > 0) {
1465 VALUE m = rb_ary_pop(extmod);
1466 rb_extend_object(v, m);
1467 }
1468 }
1469 if (!rb_respond_to(v, s_mload)) {
1470 rb_raise(rb_eTypeError, "instance of %s needs to have method `marshal_load'",
1471 rb_class2name(klass));
1472 }
1473 v = r_entry(v, arg);
1474 data = r_object(arg);
1475 rb_funcall(v, s_mload, 1, data);
1476 check_load_arg(arg);
1477 v = r_leave(v, arg);
1478 }
1479 break;
1480
1481 case TYPE_OBJECT:
1482 {
1483 v = obj_alloc_by_path(r_unique(arg), arg);
1484 if (TYPE(v) != T_OBJECT) {
1485 rb_raise(rb_eArgError, "dump format error");
1486 }
1487 v = r_entry(v, arg);
1488 r_ivar(v, arg);
1489 v = r_leave(v, arg);
1490 }
1491 break;
1492
1493 case TYPE_DATA:
1494 {
1495 VALUE klass = path2class(r_unique(arg));
1496 if (rb_respond_to(klass, s_alloc)) {
1497 static int warn = Qtrue;
1498 if (warn) {
1499 rb_warn("define `allocate' instead of `_alloc'");
1500 warn = Qfalse;
1501 }
1502 v = rb_funcall(klass, s_alloc, 0);
1503 check_load_arg(arg);
1504 }
1505 else {
1506 v = rb_obj_alloc(klass);
1507 }
1508 if (TYPE(v) != T_DATA) {
1509 rb_raise(rb_eArgError, "dump format error");
1510 }
1511 v = r_entry(v, arg);
1512 if (!rb_respond_to(v, s_load_data)) {
1513 rb_raise(rb_eTypeError,
1514 "class %s needs to have instance method `_load_data'",
1515 rb_class2name(klass));
1516 }
1517 rb_funcall(v, s_load_data, 1, r_object0(arg, 0, extmod));
1518 check_load_arg(arg);
1519 v = r_leave(v, arg);
1520 }
1521 break;
1522
1523 case TYPE_MODULE_OLD:
1524 {
1525 volatile VALUE str = r_bytes(arg);
1526
1527 v = rb_path2class(RSTRING_PTR(str));
1528 v = r_entry(v, arg);
1529 v = r_leave(v, arg);
1530 }
1531 break;
1532
1533 case TYPE_CLASS:
1534 {
1535 volatile VALUE str = r_bytes(arg);
1536
1537 v = path2class(RSTRING_PTR(str));
1538 v = r_entry(v, arg);
1539 v = r_leave(v, arg);
1540 }
1541 break;
1542
1543 case TYPE_MODULE:
1544 {
1545 volatile VALUE str = r_bytes(arg);
1546
1547 v = path2module(RSTRING_PTR(str));
1548 v = r_entry(v, arg);
1549 v = r_leave(v, arg);
1550 }
1551 break;
1552
1553 case TYPE_SYMBOL:
1554 v = ID2SYM(r_symreal(arg));
1555 v = r_leave(v, arg);
1556 break;
1557
1558 case TYPE_SYMLINK:
1559 v = ID2SYM(r_symlink(arg));
1560 break;
1561
1562 default:
1563 rb_raise(rb_eArgError, "dump format error(0x%x)", type);
1564 break;
1565 }
1566 return v;
1567 }
1568
1569 static VALUE
1570 r_object(struct load_arg *arg)
1571 {
1572 return r_object0(arg, 0, Qnil);
1573 }
1574
1575 static VALUE
1576 load(struct load_arg *arg)
1577 {
1578 return r_object(arg);
1579 }
1580
1581 static VALUE
1582 load_ensure(struct load_arg *arg)
1583 {
1584 if (!DATA_PTR(arg->compat_tbl_wrapper)) return 0;
1585 st_free_table(arg->symbols);
1586 st_free_table(arg->compat_tbl);
1587 DATA_PTR(arg->compat_tbl_wrapper) = 0;
1588 arg->compat_tbl_wrapper = 0;
1589 return 0;
1590 }
1591
1592 /*
1593 * call-seq:
1594 * load( source [, proc] ) => obj
1595 * restore( source [, proc] ) => obj
1596 *
1597 * Returns the result of converting the serialized data in source into a
1598 * Ruby object (possibly with associated subordinate objects). source
1599 * may be either an instance of IO or an object that responds to
1600 * to_str. If proc is specified, it will be passed each object as it
1601 * is deserialized.
1602 */
1603 static VALUE
1604 marshal_load(int argc, VALUE *argv)
1605 {
1606 VALUE port, proc;
1607 int major, minor;
1608 VALUE v;
1609 struct load_arg arg;
1610
1611 rb_scan_args(argc, argv, "11", &port, &proc);
1612 v = rb_check_string_type(port);
1613 if (!NIL_P(v)) {
1614 arg.taint = OBJ_TAINTED(port); /* original taintedness */
1615 port = v;
1616 }
1617 else if (rb_respond_to(port, s_getbyte) && rb_respond_to(port, s_read)) {
1618 if (rb_respond_to(port, s_binmode)) {
1619 rb_funcall2(port, s_binmode, 0, 0);
1620 }
1621 arg.taint = Qtrue;
1622 }
1623 else {
1624 rb_raise(rb_eTypeError, "instance of IO needed");
1625 }
1626 arg.untrust = OBJ_UNTRUSTED(port);
1627 arg.src = port;
1628 arg.offset = 0;
1629 arg.compat_tbl = st_init_numtable();
1630 arg.compat_tbl_wrapper = Data_Wrap_Struct(rb_cData, rb_mark_tbl, 0, arg.compat_tbl);
1631
1632 major = r_byte(&arg);
1633 minor = r_byte(&arg);
1634 if (major != MARSHAL_MAJOR || minor > MARSHAL_MINOR) {
1635 rb_raise(rb_eTypeError, "incompatible marshal file format (can't be read)\n\
1636 \tformat version %d.%d required; %d.%d given",
1637 MARSHAL_MAJOR, MARSHAL_MINOR, major, minor);
1638 }
1639 if (RTEST(ruby_verbose) && minor != MARSHAL_MINOR) {
1640 rb_warn("incompatible marshal file format (can be read)\n\
1641 \tformat version %d.%d required; %d.%d given",
1642 MARSHAL_MAJOR, MARSHAL_MINOR, major, minor);
1643 }
1644
1645 arg.symbols = st_init_numtable();
1646 arg.data = rb_hash_new();
1647 RBASIC(arg.data)->klass = 0;
1648 if (NIL_P(proc)) arg.proc = 0;
1649 else arg.proc = proc;
1650 v = rb_ensure(load, (VALUE)&arg, load_ensure, (VALUE)&arg);
1651
1652 return v;
1653 }
1654
1655 /*
1656 * The marshaling library converts collections of Ruby objects into a
1657 * byte stream, allowing them to be stored outside the currently
1658 * active script. This data may subsequently be read and the original
1659 * objects reconstituted.
1660 * Marshaled data has major and minor version numbers stored along
1661 * with the object information. In normal use, marshaling can only
1662 * load data written with the same major version number and an equal
1663 * or lower minor version number. If Ruby's ``verbose'' flag is set
1664 * (normally using -d, -v, -w, or --verbose) the major and minor
1665 * numbers must match exactly. Marshal versioning is independent of
1666 * Ruby's version numbers. You can extract the version by reading the
1667 * first two bytes of marshaled data.
1668 *
1669 * str = Marshal.dump("thing")
1670 * RUBY_VERSION #=> "1.9.0"
1671 * str[0].ord #=> 4
1672 * str[1].ord #=> 8
1673 *
1674 * Some objects cannot be dumped: if the objects to be dumped include
1675 * bindings, procedure or method objects, instances of class IO, or
1676 * singleton objects, a TypeError will be raised.
1677 * If your class has special serialization needs (for example, if you
1678 * want to serialize in some specific format), or if it contains
1679 * objects that would otherwise not be serializable, you can implement
1680 * your own serialization strategy by defining two methods, _dump and
1681 * _load:
1682 * The instance method _dump should return a String object containing
1683 * all the information necessary to reconstitute objects of this class
1684 * and all referenced objects up to a maximum depth given as an integer
1685 * parameter (a value of -1 implies that you should disable depth checking).
1686 * The class method _load should take a String and return an object of this class.
1687 */
1688 void
1689 Init_marshal(void)
1690 {
1691 #undef rb_intern
1692 #define rb_intern(str) rb_intern_const(str)
1693
1694 VALUE rb_mMarshal = rb_define_module("Marshal");
1695
1696 s_dump = rb_intern("_dump");
1697 s_load = rb_intern("_load");
1698 s_mdump = rb_intern("marshal_dump");
1699 s_mload = rb_intern("marshal_load");
1700 s_dump_data = rb_intern("_dump_data");
1701 s_load_data = rb_intern("_load_data");
1702 s_alloc = rb_intern("_alloc");
1703 s_getbyte = rb_intern("getbyte");
1704 s_read = rb_intern("read");
1705 s_write = rb_intern("write");
1706 s_binmode = rb_intern("binmode");
1707
1708 rb_define_module_function(rb_mMarshal, "dump", marshal_dump, -1);
1709 rb_define_module_function(rb_mMarshal, "load", marshal_load, -1);
1710 rb_define_module_function(rb_mMarshal, "restore", marshal_load, -1);
1711
1712 rb_define_const(rb_mMarshal, "MAJOR_VERSION", INT2FIX(MARSHAL_MAJOR));
1713 rb_define_const(rb_mMarshal, "MINOR_VERSION", INT2FIX(MARSHAL_MINOR));
1714
1715 compat_allocator_tbl = st_init_numtable();
1716 rb_gc_register_address(&compat_allocator_tbl_wrapper);
1717 compat_allocator_tbl_wrapper =
1718 Data_Wrap_Struct(rb_cData, mark_marshal_compat_t, 0, compat_allocator_tbl);
1719 }
1720
1721 VALUE
1722 rb_marshal_dump(VALUE obj, VALUE port)
1723 {
1724 int argc = 1;
1725 VALUE argv[2];
1726
1727 argv[0] = obj;
1728 argv[1] = port;
1729 if (!NIL_P(port)) argc = 2;
1730 return marshal_dump(argc, argv);
1731 }
1732
1733 VALUE
1734 rb_marshal_load(VALUE port)
1735 {
1736 return marshal_load(1, &port);
1737 }
ruby svn mirror