?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
[ruby/win32ole] Undefine allocator of WIN32OLE_VARIABLE to get rid of warning
[ruby-80x24.org.git] / marshal.c
blobc5327fca6027dc4b644780f3a6431639f321af3d
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/internal/config.h"
13
14 #include <math.h>
15 #ifdef HAVE_FLOAT_H
16 #include <float.h>
17 #endif
18 #ifdef HAVE_IEEEFP_H
19 #include <ieeefp.h>
20 #endif
21
22 #include "encindex.h"
23 #include "id_table.h"
24 #include "internal.h"
25 #include "internal/array.h"
26 #include "internal/bignum.h"
27 #include "internal/class.h"
28 #include "internal/encoding.h"
29 #include "internal/error.h"
30 #include "internal/hash.h"
31 #include "internal/object.h"
32 #include "internal/struct.h"
33 #include "internal/symbol.h"
34 #include "internal/util.h"
35 #include "internal/vm.h"
36 #include "ruby/io.h"
37 #include "ruby/ruby.h"
38 #include "ruby/st.h"
39 #include "ruby/util.h"
40 #include "builtin.h"
41
42 #define BITSPERSHORT (2*CHAR_BIT)
43 #define SHORTMASK ((1<<BITSPERSHORT)-1)
44 #define SHORTDN(x) RSHIFT((x),BITSPERSHORT)
45
46 #if SIZEOF_SHORT == SIZEOF_BDIGIT
47 #define SHORTLEN(x) (x)
48 #else
49 static size_t
50 shortlen(size_t len, BDIGIT *ds)
51 {
52 BDIGIT num;
53 int offset = 0;
54
55 num = ds[len-1];
56 while (num) {
57 num = SHORTDN(num);
58 offset++;
59 }
60 return (len - 1)*SIZEOF_BDIGIT/2 + offset;
61 }
62 #define SHORTLEN(x) shortlen((x),d)
63 #endif
64
65 #define MARSHAL_MAJOR 4
66 #define MARSHAL_MINOR 8
67
68 #define TYPE_NIL '0'
69 #define TYPE_TRUE 'T'
70 #define TYPE_FALSE 'F'
71 #define TYPE_FIXNUM 'i'
72
73 #define TYPE_EXTENDED 'e'
74 #define TYPE_UCLASS 'C'
75 #define TYPE_OBJECT 'o'
76 #define TYPE_DATA 'd'
77 #define TYPE_USERDEF 'u'
78 #define TYPE_USRMARSHAL 'U'
79 #define TYPE_FLOAT 'f'
80 #define TYPE_BIGNUM 'l'
81 #define TYPE_STRING '"'
82 #define TYPE_REGEXP '/'
83 #define TYPE_ARRAY '['
84 #define TYPE_HASH '{'
85 #define TYPE_HASH_DEF '}'
86 #define TYPE_STRUCT 'S'
87 #define TYPE_MODULE_OLD 'M'
88 #define TYPE_CLASS 'c'
89 #define TYPE_MODULE 'm'
90
91 #define TYPE_SYMBOL ':'
92 #define TYPE_SYMLINK ';'
93
94 #define TYPE_IVAR 'I'
95 #define TYPE_LINK '@'
96
97 static ID s_dump, s_load, s_mdump, s_mload;
98 static ID s_dump_data, s_load_data, s_alloc, s_call;
99 static ID s_getbyte, s_read, s_write, s_binmode;
100 static ID s_encoding_short, s_ruby2_keywords_flag;
101
102 #define name_s_dump "_dump"
103 #define name_s_load "_load"
104 #define name_s_mdump "marshal_dump"
105 #define name_s_mload "marshal_load"
106 #define name_s_dump_data "_dump_data"
107 #define name_s_load_data "_load_data"
108 #define name_s_alloc "_alloc"
109 #define name_s_call "call"
110 #define name_s_getbyte "getbyte"
111 #define name_s_read "read"
112 #define name_s_write "write"
113 #define name_s_binmode "binmode"
114 #define name_s_encoding_short "E"
115 #define name_s_ruby2_keywords_flag "K"
116
117 typedef struct {
118 VALUE newclass;
119 VALUE oldclass;
120 VALUE (*dumper)(VALUE);
121 VALUE (*loader)(VALUE, VALUE);
122 } marshal_compat_t;
123
124 static st_table *compat_allocator_tbl;
125 static VALUE compat_allocator_tbl_wrapper;
126 static VALUE rb_marshal_dump_limited(VALUE obj, VALUE port, int limit);
127 static VALUE rb_marshal_load_with_proc(VALUE port, VALUE proc, bool freeze);
128
129 static int
130 mark_marshal_compat_i(st_data_t key, st_data_t value, st_data_t _)
131 {
132 marshal_compat_t *p = (marshal_compat_t *)value;
133 rb_gc_mark(p->newclass);
134 rb_gc_mark(p->oldclass);
135 return ST_CONTINUE;
136 }
137
138 static void
139 mark_marshal_compat_t(void *tbl)
140 {
141 if (!tbl) return;
142 st_foreach(tbl, mark_marshal_compat_i, 0);
143 }
144
145 static st_table *compat_allocator_table(void);
146
147 void
148 rb_marshal_define_compat(VALUE newclass, VALUE oldclass, VALUE (*dumper)(VALUE), VALUE (*loader)(VALUE, VALUE))
149 {
150 marshal_compat_t *compat;
151 rb_alloc_func_t allocator = rb_get_alloc_func(newclass);
152
153 if (!allocator) {
154 rb_raise(rb_eTypeError, "no allocator");
155 }
156
157 compat = ALLOC(marshal_compat_t);
158 compat->newclass = Qnil;
159 compat->oldclass = Qnil;
160 compat->newclass = newclass;
161 compat->oldclass = oldclass;
162 compat->dumper = dumper;
163 compat->loader = loader;
164
165 st_insert(compat_allocator_table(), (st_data_t)allocator, (st_data_t)compat);
166 }
167
168 struct dump_arg {
169 VALUE str, dest;
170 st_table *symbols;
171 st_table *data;
172 st_table *compat_tbl;
173 st_table *encodings;
174 };
175
176 struct dump_call_arg {
177 VALUE obj;
178 struct dump_arg *arg;
179 int limit;
180 };
181
182 static VALUE
183 check_dump_arg(VALUE ret, struct dump_arg *arg, const char *name)
184 {
185 if (!arg->symbols) {
186 rb_raise(rb_eRuntimeError, "Marshal.dump reentered at %s",
187 name);
188 }
189 return ret;
190 }
191
192 static VALUE
193 check_userdump_arg(VALUE obj, ID sym, int argc, const VALUE *argv,
194 struct dump_arg *arg, const char *name)
195 {
196 VALUE ret = rb_funcallv(obj, sym, argc, argv);
197 VALUE klass = CLASS_OF(obj);
198 if (CLASS_OF(ret) == klass) {
199 rb_raise(rb_eRuntimeError, "%"PRIsVALUE"#%s returned same class instance",
200 klass, name);
201 }
202 return check_dump_arg(ret, arg, name);
203 }
204
205 #define dump_funcall(arg, obj, sym, argc, argv) \
206 check_userdump_arg(obj, sym, argc, argv, arg, name_##sym)
207 #define dump_check_funcall(arg, obj, sym, argc, argv) \
208 check_dump_arg(rb_check_funcall(obj, sym, argc, argv), arg, name_##sym)
209
210 static void clear_dump_arg(struct dump_arg *arg);
211
212 static void
213 mark_dump_arg(void *ptr)
214 {
215 struct dump_arg *p = ptr;
216 if (!p->symbols)
217 return;
218 rb_mark_set(p->symbols);
219 rb_mark_set(p->data);
220 rb_mark_hash(p->compat_tbl);
221 rb_gc_mark(p->str);
222 }
223
224 static void
225 free_dump_arg(void *ptr)
226 {
227 clear_dump_arg(ptr);
228 xfree(ptr);
229 }
230
231 static size_t
232 memsize_dump_arg(const void *ptr)
233 {
234 return sizeof(struct dump_arg);
235 }
236
237 static const rb_data_type_t dump_arg_data = {
238 "dump_arg",
239 {mark_dump_arg, free_dump_arg, memsize_dump_arg,},
240 0, 0, RUBY_TYPED_FREE_IMMEDIATELY
241 };
242
243 static VALUE
244 must_not_be_anonymous(const char *type, VALUE path)
245 {
246 char *n = RSTRING_PTR(path);
247
248 if (!rb_enc_asciicompat(rb_enc_get(path))) {
249 /* cannot occur? */
250 rb_raise(rb_eTypeError, "can't dump non-ascii %s name % "PRIsVALUE,
251 type, path);
252 }
253 if (n[0] == '#') {
254 rb_raise(rb_eTypeError, "can't dump anonymous %s % "PRIsVALUE,
255 type, path);
256 }
257 return path;
258 }
259
260 static VALUE
261 class2path(VALUE klass)
262 {
263 VALUE path = rb_class_path(klass);
264
265 must_not_be_anonymous((RB_TYPE_P(klass, T_CLASS) ? "class" : "module"), path);
266 if (rb_path_to_class(path) != rb_class_real(klass)) {
267 rb_raise(rb_eTypeError, "% "PRIsVALUE" can't be referred to", path);
268 }
269 return path;
270 }
271
272 int ruby_marshal_write_long(long x, char *buf);
273 static void w_long(long, struct dump_arg*);
274 static int w_encoding(VALUE encname, struct dump_call_arg *arg);
275 static VALUE encoding_name(VALUE obj, struct dump_arg *arg);
276
277 static void
278 w_nbyte(const char *s, long n, struct dump_arg *arg)
279 {
280 VALUE buf = arg->str;
281 rb_str_buf_cat(buf, s, n);
282 if (arg->dest && RSTRING_LEN(buf) >= BUFSIZ) {
283 rb_io_write(arg->dest, buf);
284 rb_str_resize(buf, 0);
285 }
286 }
287
288 static void
289 w_byte(char c, struct dump_arg *arg)
290 {
291 w_nbyte(&c, 1, arg);
292 }
293
294 static void
295 w_bytes(const char *s, long n, struct dump_arg *arg)
296 {
297 w_long(n, arg);
298 w_nbyte(s, n, arg);
299 }
300
301 #define w_cstr(s, arg) w_bytes((s), strlen(s), (arg))
302
303 static void
304 w_short(int x, struct dump_arg *arg)
305 {
306 w_byte((char)((x >> 0) & 0xff), arg);
307 w_byte((char)((x >> 8) & 0xff), arg);
308 }
309
310 static void
311 w_long(long x, struct dump_arg *arg)
312 {
313 char buf[sizeof(long)+1];
314 int i = ruby_marshal_write_long(x, buf);
315 if (i < 0) {
316 rb_raise(rb_eTypeError, "long too big to dump");
317 }
318 w_nbyte(buf, i, arg);
319 }
320
321 int
322 ruby_marshal_write_long(long x, char *buf)
323 {
324 int i;
325
326 #if SIZEOF_LONG > 4
327 if (!(RSHIFT(x, 31) == 0 || RSHIFT(x, 31) == -1)) {
328 /* big long does not fit in 4 bytes */
329 return -1;
330 }
331 #endif
332
333 if (x == 0) {
334 buf[0] = 0;
335 return 1;
336 }
337 if (0 < x && x < 123) {
338 buf[0] = (char)(x + 5);
339 return 1;
340 }
341 if (-124 < x && x < 0) {
342 buf[0] = (char)((x - 5)&0xff);
343 return 1;
344 }
345 for (i=1;i<(int)sizeof(long)+1;i++) {
346 buf[i] = (char)(x & 0xff);
347 x = RSHIFT(x,8);
348 if (x == 0) {
349 buf[0] = i;
350 break;
351 }
352 if (x == -1) {
353 buf[0] = -i;
354 break;
355 }
356 }
357 return i+1;
358 }
359
360 #ifdef DBL_MANT_DIG
361 #define DECIMAL_MANT (53-16) /* from IEEE754 double precision */
362
363 #if DBL_MANT_DIG > 32
364 #define MANT_BITS 32
365 #elif DBL_MANT_DIG > 24
366 #define MANT_BITS 24
367 #elif DBL_MANT_DIG > 16
368 #define MANT_BITS 16
369 #else
370 #define MANT_BITS 8
371 #endif
372
373 static double
374 load_mantissa(double d, const char *buf, long len)
375 {
376 if (!len) return d;
377 if (--len > 0 && !*buf++) { /* binary mantissa mark */
378 int e, s = d < 0, dig = 0;
379 unsigned long m;
380
381 modf(ldexp(frexp(fabs(d), &e), DECIMAL_MANT), &d);
382 do {
383 m = 0;
384 switch (len) {
385 default: m = *buf++ & 0xff; /* fall through */
386 #if MANT_BITS > 24
387 case 3: m = (m << 8) | (*buf++ & 0xff); /* fall through */
388 #endif
389 #if MANT_BITS > 16
390 case 2: m = (m << 8) | (*buf++ & 0xff); /* fall through */
391 #endif
392 #if MANT_BITS > 8
393 case 1: m = (m << 8) | (*buf++ & 0xff);
394 #endif
395 }
396 dig -= len < MANT_BITS / 8 ? 8 * (unsigned)len : MANT_BITS;
397 d += ldexp((double)m, dig);
398 } while ((len -= MANT_BITS / 8) > 0);
399 d = ldexp(d, e - DECIMAL_MANT);
400 if (s) d = -d;
401 }
402 return d;
403 }
404 #else
405 #define load_mantissa(d, buf, len) (d)
406 #endif
407
408 #ifdef DBL_DIG
409 #define FLOAT_DIG (DBL_DIG+2)
410 #else
411 #define FLOAT_DIG 17
412 #endif
413
414 static void
415 w_float(double d, struct dump_arg *arg)
416 {
417 char buf[FLOAT_DIG + (DECIMAL_MANT + 7) / 8 + 10];
418
419 if (isinf(d)) {
420 if (d < 0) w_cstr("-inf", arg);
421 else w_cstr("inf", arg);
422 }
423 else if (isnan(d)) {
424 w_cstr("nan", arg);
425 }
426 else if (d == 0.0) {
427 if (signbit(d)) w_cstr("-0", arg);
428 else w_cstr("0", arg);
429 }
430 else {
431 int decpt, sign, digs, len = 0;
432 char *e, *p = ruby_dtoa(d, 0, 0, &decpt, &sign, &e);
433 if (sign) buf[len++] = '-';
434 digs = (int)(e - p);
435 if (decpt < -3 || decpt > digs) {
436 buf[len++] = p[0];
437 if (--digs > 0) buf[len++] = '.';
438 memcpy(buf + len, p + 1, digs);
439 len += digs;
440 len += snprintf(buf + len, sizeof(buf) - len, "e%d", decpt - 1);
441 }
442 else if (decpt > 0) {
443 memcpy(buf + len, p, decpt);
444 len += decpt;
445 if ((digs -= decpt) > 0) {
446 buf[len++] = '.';
447 memcpy(buf + len, p + decpt, digs);
448 len += digs;
449 }
450 }
451 else {
452 buf[len++] = '0';
453 buf[len++] = '.';
454 if (decpt) {
455 memset(buf + len, '0', -decpt);
456 len -= decpt;
457 }
458 memcpy(buf + len, p, digs);
459 len += digs;
460 }
461 xfree(p);
462 w_bytes(buf, len, arg);
463 }
464 }
465
466 static void
467 w_symbol(VALUE sym, struct dump_arg *arg)
468 {
469 st_data_t num;
470 VALUE encname;
471
472 if (st_lookup(arg->symbols, sym, &num)) {
473 w_byte(TYPE_SYMLINK, arg);
474 w_long((long)num, arg);
475 }
476 else {
477 const VALUE orig_sym = sym;
478 sym = rb_sym2str(sym);
479 if (!sym) {
480 rb_raise(rb_eTypeError, "can't dump anonymous ID %"PRIdVALUE, sym);
481 }
482 encname = encoding_name(sym, arg);
483 if (NIL_P(encname) ||
484 rb_enc_str_coderange(sym) == ENC_CODERANGE_7BIT) {
485 encname = Qnil;
486 }
487 else {
488 w_byte(TYPE_IVAR, arg);
489 }
490 w_byte(TYPE_SYMBOL, arg);
491 w_bytes(RSTRING_PTR(sym), RSTRING_LEN(sym), arg);
492 st_add_direct(arg->symbols, orig_sym, arg->symbols->num_entries);
493 if (!NIL_P(encname)) {
494 struct dump_call_arg c_arg;
495 c_arg.limit = 1;
496 c_arg.arg = arg;
497 w_long(1L, arg);
498 w_encoding(encname, &c_arg);
499 }
500 }
501 }
502
503 static void
504 w_unique(VALUE s, struct dump_arg *arg)
505 {
506 must_not_be_anonymous("class", s);
507 w_symbol(rb_str_intern(s), arg);
508 }
509
510 static void w_object(VALUE,struct dump_arg*,int);
511
512 static int
513 hash_each(VALUE key, VALUE value, VALUE v)
514 {
515 struct dump_call_arg *arg = (void *)v;
516 w_object(key, arg->arg, arg->limit);
517 w_object(value, arg->arg, arg->limit);
518 return ST_CONTINUE;
519 }
520
521 #define SINGLETON_DUMP_UNABLE_P(klass) \
522 (rb_id_table_size(RCLASS_M_TBL(klass)) > 0 || \
523 (RCLASS_IV_TBL(klass) && RCLASS_IV_TBL(klass)->num_entries > 1))
524
525 static void
526 w_extended(VALUE klass, struct dump_arg *arg, int check)
527 {
528 if (check && FL_TEST(klass, FL_SINGLETON)) {
529 VALUE origin = RCLASS_ORIGIN(klass);
530 if (SINGLETON_DUMP_UNABLE_P(klass) ||
531 (origin != klass && SINGLETON_DUMP_UNABLE_P(origin))) {
532 rb_raise(rb_eTypeError, "singleton can't be dumped");
533 }
534 klass = RCLASS_SUPER(klass);
535 }
536 while (BUILTIN_TYPE(klass) == T_ICLASS) {
537 if (!FL_TEST(klass, RICLASS_IS_ORIGIN) ||
538 BUILTIN_TYPE(RBASIC(klass)->klass) != T_MODULE) {
539 VALUE path = rb_class_name(RBASIC(klass)->klass);
540 w_byte(TYPE_EXTENDED, arg);
541 w_unique(path, arg);
542 }
543 klass = RCLASS_SUPER(klass);
544 }
545 }
546
547 static void
548 w_class(char type, VALUE obj, struct dump_arg *arg, int check)
549 {
550 VALUE path;
551 st_data_t real_obj;
552 VALUE klass;
553
554 if (arg->compat_tbl &&
555 st_lookup(arg->compat_tbl, (st_data_t)obj, &real_obj)) {
556 obj = (VALUE)real_obj;
557 }
558 klass = CLASS_OF(obj);
559 w_extended(klass, arg, check);
560 w_byte(type, arg);
561 path = class2path(rb_class_real(klass));
562 w_unique(path, arg);
563 }
564
565 static void
566 w_uclass(VALUE obj, VALUE super, struct dump_arg *arg)
567 {
568 VALUE klass = CLASS_OF(obj);
569
570 w_extended(klass, arg, TRUE);
571 klass = rb_class_real(klass);
572 if (klass != super) {
573 w_byte(TYPE_UCLASS, arg);
574 w_unique(class2path(klass), arg);
575 }
576 }
577
578 static bool
579 rb_hash_ruby2_keywords_p(VALUE obj)
580 {
581 return (RHASH(obj)->basic.flags & RHASH_PASS_AS_KEYWORDS) != 0;
582 }
583
584 static void
585 rb_hash_ruby2_keywords(VALUE obj)
586 {
587 RHASH(obj)->basic.flags |= RHASH_PASS_AS_KEYWORDS;
588 }
589
590 static inline bool
591 to_be_skipped_id(const ID id)
592 {
593 if (id == s_encoding_short) return true;
594 if (id == s_ruby2_keywords_flag) return true;
595 if (id == rb_id_encoding()) return true;
596 return !rb_id2str(id);
597 }
598
599 struct w_ivar_arg {
600 struct dump_call_arg *dump;
601 st_data_t num_ivar;
602 };
603
604 static int
605 w_obj_each(st_data_t key, st_data_t val, st_data_t a)
606 {
607 ID id = (ID)key;
608 VALUE value = (VALUE)val;
609 struct w_ivar_arg *ivarg = (struct w_ivar_arg *)a;
610 struct dump_call_arg *arg = ivarg->dump;
611
612 if (to_be_skipped_id(id)) {
613 if (id == s_encoding_short) {
614 rb_warn("instance variable `"name_s_encoding_short"' on class %"PRIsVALUE" is not dumped",
615 CLASS_OF(arg->obj));
616 }
617 if (id == s_ruby2_keywords_flag) {
618 rb_warn("instance variable `"name_s_ruby2_keywords_flag"' on class %"PRIsVALUE" is not dumped",
619 CLASS_OF(arg->obj));
620 }
621 return ST_CONTINUE;
622 }
623 if (!ivarg->num_ivar) {
624 rb_raise(rb_eRuntimeError, "instance variable added to %"PRIsVALUE" instance",
625 CLASS_OF(arg->obj));
626 }
627 --ivarg->num_ivar;
628 w_symbol(ID2SYM(id), arg->arg);
629 w_object(value, arg->arg, arg->limit);
630 return ST_CONTINUE;
631 }
632
633 static int
634 obj_count_ivars(st_data_t key, st_data_t val, st_data_t a)
635 {
636 ID id = (ID)key;
637 if (!to_be_skipped_id(id) && UNLIKELY(!++*(st_index_t *)a)) {
638 rb_raise(rb_eRuntimeError, "too many instance variables");
639 }
640 return ST_CONTINUE;
641 }
642
643 static VALUE
644 encoding_name(VALUE obj, struct dump_arg *arg)
645 {
646 if (rb_enc_capable(obj)) {
647 int encidx = rb_enc_get_index(obj);
648 rb_encoding *enc = 0;
649 st_data_t name;
650
651 if (encidx <= 0 || !(enc = rb_enc_from_index(encidx))) {
652 return Qnil;
653 }
654
655 /* special treatment for US-ASCII and UTF-8 */
656 if (encidx == rb_usascii_encindex()) {
657 return Qfalse;
658 }
659 else if (encidx == rb_utf8_encindex()) {
660 return Qtrue;
661 }
662
663 if (arg->encodings ?
664 !st_lookup(arg->encodings, (st_data_t)rb_enc_name(enc), &name) :
665 (arg->encodings = st_init_strcasetable(), 1)) {
666 name = (st_data_t)rb_str_new_cstr(rb_enc_name(enc));
667 st_insert(arg->encodings, (st_data_t)rb_enc_name(enc), name);
668 }
669 return (VALUE)name;
670 }
671 else {
672 return Qnil;
673 }
674 }
675
676 static int
677 w_encoding(VALUE encname, struct dump_call_arg *arg)
678 {
679 int limit = arg->limit;
680 if (limit >= 0) ++limit;
681 switch (encname) {
682 case Qfalse:
683 case Qtrue:
684 w_symbol(ID2SYM(s_encoding_short), arg->arg);
685 w_object(encname, arg->arg, limit);
686 return 1;
687 case Qnil:
688 return 0;
689 }
690 w_symbol(ID2SYM(rb_id_encoding()), arg->arg);
691 w_object(encname, arg->arg, limit);
692 return 1;
693 }
694
695 static st_index_t
696 has_ivars(VALUE obj, VALUE encname, VALUE *ivobj)
697 {
698 st_index_t num = !NIL_P(encname);
699
700 if (SPECIAL_CONST_P(obj)) goto generic;
701 switch (BUILTIN_TYPE(obj)) {
702 case T_OBJECT:
703 case T_CLASS:
704 case T_MODULE:
705 break; /* counted elsewhere */
706 case T_HASH:
707 if (rb_hash_ruby2_keywords_p(obj)) ++num;
708 /* fall through */
709 default:
710 generic:
711 rb_ivar_foreach(obj, obj_count_ivars, (st_data_t)&num);
712 if (num) *ivobj = obj;
713 }
714
715 return num;
716 }
717
718 static void
719 w_ivar_each(VALUE obj, st_index_t num, struct dump_call_arg *arg)
720 {
721 struct w_ivar_arg ivarg = {arg, num};
722 if (!num) return;
723 rb_ivar_foreach(obj, w_obj_each, (st_data_t)&ivarg);
724 if (ivarg.num_ivar) {
725 rb_raise(rb_eRuntimeError, "instance variable removed from %"PRIsVALUE" instance",
726 CLASS_OF(arg->obj));
727 }
728 }
729
730 static void
731 w_ivar(st_index_t num, VALUE ivobj, VALUE encname, struct dump_call_arg *arg)
732 {
733 w_long(num, arg->arg);
734 num -= w_encoding(encname, arg);
735 if (RB_TYPE_P(ivobj, T_HASH) && rb_hash_ruby2_keywords_p(ivobj)) {
736 int limit = arg->limit;
737 if (limit >= 0) ++limit;
738 w_symbol(ID2SYM(s_ruby2_keywords_flag), arg->arg);
739 w_object(Qtrue, arg->arg, limit);
740 num--;
741 }
742 if (ivobj != Qundef && num) {
743 w_ivar_each(ivobj, num, arg);
744 }
745 }
746
747 static void
748 w_objivar(VALUE obj, struct dump_call_arg *arg)
749 {
750 st_data_t num = 0;
751
752 rb_ivar_foreach(obj, obj_count_ivars, (st_data_t)&num);
753 w_long(num, arg->arg);
754 w_ivar_each(obj, num, arg);
755 }
756
757 static void
758 w_object(VALUE obj, struct dump_arg *arg, int limit)
759 {
760 struct dump_call_arg c_arg;
761 VALUE ivobj = Qundef;
762 st_data_t num;
763 st_index_t hasiv = 0;
764 VALUE encname = Qnil;
765
766 if (limit == 0) {
767 rb_raise(rb_eArgError, "exceed depth limit");
768 }
769
770 if (limit > 0) limit--;
771 c_arg.limit = limit;
772 c_arg.arg = arg;
773 c_arg.obj = obj;
774
775 if (st_lookup(arg->data, obj, &num)) {
776 w_byte(TYPE_LINK, arg);
777 w_long((long)num, arg);
778 return;
779 }
780
781 if (NIL_P(obj)) {
782 w_byte(TYPE_NIL, arg);
783 }
784 else if (obj == Qtrue) {
785 w_byte(TYPE_TRUE, arg);
786 }
787 else if (obj == Qfalse) {
788 w_byte(TYPE_FALSE, arg);
789 }
790 else if (FIXNUM_P(obj)) {
791 #if SIZEOF_LONG <= 4
792 w_byte(TYPE_FIXNUM, arg);
793 w_long(FIX2INT(obj), arg);
794 #else
795 if (RSHIFT((long)obj, 31) == 0 || RSHIFT((long)obj, 31) == -1) {
796 w_byte(TYPE_FIXNUM, arg);
797 w_long(FIX2LONG(obj), arg);
798 }
799 else {
800 w_object(rb_int2big(FIX2LONG(obj)), arg, limit);
801 }
802 #endif
803 }
804 else if (SYMBOL_P(obj)) {
805 w_symbol(obj, arg);
806 }
807 else if (FLONUM_P(obj)) {
808 st_add_direct(arg->data, obj, arg->data->num_entries);
809 w_byte(TYPE_FLOAT, arg);
810 w_float(RFLOAT_VALUE(obj), arg);
811 }
812 else {
813 VALUE v;
814
815 if (!RBASIC_CLASS(obj)) {
816 rb_raise(rb_eTypeError, "can't dump internal %s",
817 rb_builtin_type_name(BUILTIN_TYPE(obj)));
818 }
819
820 if (rb_obj_respond_to(obj, s_mdump, TRUE)) {
821 st_add_direct(arg->data, obj, arg->data->num_entries);
822
823 v = dump_funcall(arg, obj, s_mdump, 0, 0);
824 w_class(TYPE_USRMARSHAL, obj, arg, FALSE);
825 w_object(v, arg, limit);
826 return;
827 }
828 if (rb_obj_respond_to(obj, s_dump, TRUE)) {
829 VALUE ivobj2 = Qundef;
830 st_index_t hasiv2;
831 VALUE encname2;
832
833 v = INT2NUM(limit);
834 v = dump_funcall(arg, obj, s_dump, 1, &v);
835 if (!RB_TYPE_P(v, T_STRING)) {
836 rb_raise(rb_eTypeError, "_dump() must return string");
837 }
838 hasiv = has_ivars(obj, (encname = encoding_name(obj, arg)), &ivobj);
839 hasiv2 = has_ivars(v, (encname2 = encoding_name(v, arg)), &ivobj2);
840 if (hasiv2) {
841 hasiv = hasiv2;
842 ivobj = ivobj2;
843 encname = encname2;
844 }
845 if (hasiv) w_byte(TYPE_IVAR, arg);
846 w_class(TYPE_USERDEF, obj, arg, FALSE);
847 w_bytes(RSTRING_PTR(v), RSTRING_LEN(v), arg);
848 if (hasiv) {
849 w_ivar(hasiv, ivobj, encname, &c_arg);
850 }
851 st_add_direct(arg->data, obj, arg->data->num_entries);
852 return;
853 }
854
855 st_add_direct(arg->data, obj, arg->data->num_entries);
856
857 hasiv = has_ivars(obj, (encname = encoding_name(obj, arg)), &ivobj);
858 {
859 st_data_t compat_data;
860 rb_alloc_func_t allocator = rb_get_alloc_func(RBASIC(obj)->klass);
861 if (st_lookup(compat_allocator_tbl,
862 (st_data_t)allocator,
863 &compat_data)) {
864 marshal_compat_t *compat = (marshal_compat_t*)compat_data;
865 VALUE real_obj = obj;
866 obj = compat->dumper(real_obj);
867 if (!arg->compat_tbl) {
868 arg->compat_tbl = rb_init_identtable();
869 }
870 st_insert(arg->compat_tbl, (st_data_t)obj, (st_data_t)real_obj);
871 if (obj != real_obj && ivobj == Qundef) hasiv = 0;
872 }
873 }
874 if (hasiv) w_byte(TYPE_IVAR, arg);
875
876 switch (BUILTIN_TYPE(obj)) {
877 case T_CLASS:
878 if (FL_TEST(obj, FL_SINGLETON)) {
879 rb_raise(rb_eTypeError, "singleton class can't be dumped");
880 }
881 w_byte(TYPE_CLASS, arg);
882 {
883 VALUE path = class2path(obj);
884 w_bytes(RSTRING_PTR(path), RSTRING_LEN(path), arg);
885 RB_GC_GUARD(path);
886 }
887 break;
888
889 case T_MODULE:
890 w_byte(TYPE_MODULE, arg);
891 {
892 VALUE path = class2path(obj);
893 w_bytes(RSTRING_PTR(path), RSTRING_LEN(path), arg);
894 RB_GC_GUARD(path);
895 }
896 break;
897
898 case T_FLOAT:
899 w_byte(TYPE_FLOAT, arg);
900 w_float(RFLOAT_VALUE(obj), arg);
901 break;
902
903 case T_BIGNUM:
904 w_byte(TYPE_BIGNUM, arg);
905 {
906 char sign = BIGNUM_SIGN(obj) ? '+' : '-';
907 size_t len = BIGNUM_LEN(obj);
908 size_t slen;
909 size_t j;
910 BDIGIT *d = BIGNUM_DIGITS(obj);
911
912 slen = SHORTLEN(len);
913 if (LONG_MAX < slen) {
914 rb_raise(rb_eTypeError, "too big Bignum can't be dumped");
915 }
916
917 w_byte(sign, arg);
918 w_long((long)slen, arg);
919 for (j = 0; j < len; j++) {
920 #if SIZEOF_BDIGIT > SIZEOF_SHORT
921 BDIGIT num = *d;
922 int i;
923
924 for (i=0; i<SIZEOF_BDIGIT; i+=SIZEOF_SHORT) {
925 w_short(num & SHORTMASK, arg);
926 num = SHORTDN(num);
927 if (j == len - 1 && num == 0) break;
928 }
929 #else
930 w_short(*d, arg);
931 #endif
932 d++;
933 }
934 }
935 break;
936
937 case T_STRING:
938 w_uclass(obj, rb_cString, arg);
939 w_byte(TYPE_STRING, arg);
940 w_bytes(RSTRING_PTR(obj), RSTRING_LEN(obj), arg);
941 break;
942
943 case T_REGEXP:
944 w_uclass(obj, rb_cRegexp, arg);
945 w_byte(TYPE_REGEXP, arg);
946 {
947 int opts = rb_reg_options(obj);
948 w_bytes(RREGEXP_SRC_PTR(obj), RREGEXP_SRC_LEN(obj), arg);
949 w_byte((char)opts, arg);
950 }
951 break;
952
953 case T_ARRAY:
954 w_uclass(obj, rb_cArray, arg);
955 w_byte(TYPE_ARRAY, arg);
956 {
957 long i, len = RARRAY_LEN(obj);
958
959 w_long(len, arg);
960 for (i=0; i<RARRAY_LEN(obj); i++) {
961 w_object(RARRAY_AREF(obj, i), arg, limit);
962 if (len != RARRAY_LEN(obj)) {
963 rb_raise(rb_eRuntimeError, "array modified during dump");
964 }
965 }
966 }
967 break;
968
969 case T_HASH:
970 w_uclass(obj, rb_cHash, arg);
971 if (rb_hash_compare_by_id_p(obj)) {
972 w_byte(TYPE_UCLASS, arg);
973 w_symbol(rb_sym_intern_ascii_cstr("Hash"), arg);
974 }
975 if (NIL_P(RHASH_IFNONE(obj))) {
976 w_byte(TYPE_HASH, arg);
977 }
978 else if (FL_TEST(obj, RHASH_PROC_DEFAULT)) {
979 rb_raise(rb_eTypeError, "can't dump hash with default proc");
980 }
981 else {
982 w_byte(TYPE_HASH_DEF, arg);
983 }
984 w_long(rb_hash_size_num(obj), arg);
985 rb_hash_foreach(obj, hash_each, (st_data_t)&c_arg);
986 if (!NIL_P(RHASH_IFNONE(obj))) {
987 w_object(RHASH_IFNONE(obj), arg, limit);
988 }
989 break;
990
991 case T_STRUCT:
992 w_class(TYPE_STRUCT, obj, arg, TRUE);
993 {
994 long len = RSTRUCT_LEN(obj);
995 VALUE mem;
996 long i;
997
998 w_long(len, arg);
999 mem = rb_struct_members(obj);
1000 for (i=0; i<len; i++) {
1001 w_symbol(RARRAY_AREF(mem, i), arg);
1002 w_object(RSTRUCT_GET(obj, i), arg, limit);
1003 }
1004 }
1005 break;
1006
1007 case T_OBJECT:
1008 w_class(TYPE_OBJECT, obj, arg, TRUE);
1009 w_objivar(obj, &c_arg);
1010 break;
1011
1012 case T_DATA:
1013 {
1014 VALUE v;
1015
1016 if (!rb_obj_respond_to(obj, s_dump_data, TRUE)) {
1017 rb_raise(rb_eTypeError,
1018 "no _dump_data is defined for class %"PRIsVALUE,
1019 rb_obj_class(obj));
1020 }
1021 v = dump_funcall(arg, obj, s_dump_data, 0, 0);
1022 w_class(TYPE_DATA, obj, arg, TRUE);
1023 w_object(v, arg, limit);
1024 }
1025 break;
1026
1027 default:
1028 rb_raise(rb_eTypeError, "can't dump %"PRIsVALUE,
1029 rb_obj_class(obj));
1030 break;
1031 }
1032 RB_GC_GUARD(obj);
1033 }
1034 if (hasiv) {
1035 w_ivar(hasiv, ivobj, encname, &c_arg);
1036 }
1037 }
1038
1039 static void
1040 clear_dump_arg(struct dump_arg *arg)
1041 {
1042 if (!arg->symbols) return;
1043 st_free_table(arg->symbols);
1044 arg->symbols = 0;
1045 st_free_table(arg->data);
1046 arg->data = 0;
1047 if (arg->compat_tbl) {
1048 st_free_table(arg->compat_tbl);
1049 arg->compat_tbl = 0;
1050 }
1051 if (arg->encodings) {
1052 st_free_table(arg->encodings);
1053 arg->encodings = 0;
1054 }
1055 }
1056
1057 NORETURN(static inline void io_needed(void));
1058 static inline void
1059 io_needed(void)
1060 {
1061 rb_raise(rb_eTypeError, "instance of IO needed");
1062 }
1063
1064 /*
1065 * call-seq:
1066 * dump( obj [, anIO] , limit=-1 ) -> anIO
1067 *
1068 * Serializes obj and all descendant objects. If anIO is
1069 * specified, the serialized data will be written to it, otherwise the
1070 * data will be returned as a String. If limit is specified, the
1071 * traversal of subobjects will be limited to that depth. If limit is
1072 * negative, no checking of depth will be performed.
1073 *
1074 * class Klass
1075 * def initialize(str)
1076 * @str = str
1077 * end
1078 * def say_hello
1079 * @str
1080 * end
1081 * end
1082 *
1083 * (produces no output)
1084 *
1085 * o = Klass.new("hello\n")
1086 * data = Marshal.dump(o)
1087 * obj = Marshal.load(data)
1088 * obj.say_hello #=> "hello\n"
1089 *
1090 * Marshal can't dump following objects:
1091 * * anonymous Class/Module.
1092 * * objects which are related to system (ex: Dir, File::Stat, IO, File, Socket
1093 * and so on)
1094 * * an instance of MatchData, Data, Method, UnboundMethod, Proc, Thread,
1095 * ThreadGroup, Continuation
1096 * * objects which define singleton methods
1097 */
1098 static VALUE
1099 marshal_dump(int argc, VALUE *argv, VALUE _)
1100 {
1101 VALUE obj, port, a1, a2;
1102 int limit = -1;
1103
1104 port = Qnil;
1105 rb_scan_args(argc, argv, "12", &obj, &a1, &a2);
1106 if (argc == 3) {
1107 if (!NIL_P(a2)) limit = NUM2INT(a2);
1108 if (NIL_P(a1)) io_needed();
1109 port = a1;
1110 }
1111 else if (argc == 2) {
1112 if (FIXNUM_P(a1)) limit = FIX2INT(a1);
1113 else if (NIL_P(a1)) io_needed();
1114 else port = a1;
1115 }
1116 return rb_marshal_dump_limited(obj, port, limit);
1117 }
1118
1119 VALUE
1120 rb_marshal_dump_limited(VALUE obj, VALUE port, int limit)
1121 {
1122 struct dump_arg *arg;
1123 VALUE wrapper; /* used to avoid memory leak in case of exception */
1124
1125 wrapper = TypedData_Make_Struct(0, struct dump_arg, &dump_arg_data, arg);
1126 arg->dest = 0;
1127 arg->symbols = st_init_numtable();
1128 arg->data = rb_init_identtable();
1129 arg->compat_tbl = 0;
1130 arg->encodings = 0;
1131 arg->str = rb_str_buf_new(0);
1132 if (!NIL_P(port)) {
1133 if (!rb_respond_to(port, s_write)) {
1134 io_needed();
1135 }
1136 arg->dest = port;
1137 dump_check_funcall(arg, port, s_binmode, 0, 0);
1138 }
1139 else {
1140 port = arg->str;
1141 }
1142
1143 w_byte(MARSHAL_MAJOR, arg);
1144 w_byte(MARSHAL_MINOR, arg);
1145
1146 w_object(obj, arg, limit);
1147 if (arg->dest) {
1148 rb_io_write(arg->dest, arg->str);
1149 rb_str_resize(arg->str, 0);
1150 }
1151 clear_dump_arg(arg);
1152 RB_GC_GUARD(wrapper);
1153
1154 return port;
1155 }
1156
1157 struct load_arg {
1158 VALUE src;
1159 char *buf;
1160 long buflen;
1161 long readable;
1162 long offset;
1163 st_table *symbols;
1164 st_table *data;
1165 st_table *partial_objects;
1166 VALUE proc;
1167 st_table *compat_tbl;
1168 bool freeze;
1169 };
1170
1171 static VALUE
1172 check_load_arg(VALUE ret, struct load_arg *arg, const char *name)
1173 {
1174 if (!arg->symbols) {
1175 rb_raise(rb_eRuntimeError, "Marshal.load reentered at %s",
1176 name);
1177 }
1178 return ret;
1179 }
1180 #define load_funcall(arg, obj, sym, argc, argv) \
1181 check_load_arg(rb_funcallv(obj, sym, argc, argv), arg, name_##sym)
1182
1183 static void clear_load_arg(struct load_arg *arg);
1184
1185 static void
1186 mark_load_arg(void *ptr)
1187 {
1188 struct load_arg *p = ptr;
1189 if (!p->symbols)
1190 return;
1191 rb_mark_tbl(p->symbols);
1192 rb_mark_tbl(p->data);
1193 rb_mark_tbl(p->partial_objects);
1194 rb_mark_hash(p->compat_tbl);
1195 }
1196
1197 static void
1198 free_load_arg(void *ptr)
1199 {
1200 clear_load_arg(ptr);
1201 xfree(ptr);
1202 }
1203
1204 static size_t
1205 memsize_load_arg(const void *ptr)
1206 {
1207 return sizeof(struct load_arg);
1208 }
1209
1210 static const rb_data_type_t load_arg_data = {
1211 "load_arg",
1212 {mark_load_arg, free_load_arg, memsize_load_arg,},
1213 0, 0, RUBY_TYPED_FREE_IMMEDIATELY
1214 };
1215
1216 #define r_entry(v, arg) r_entry0((v), (arg)->data->num_entries, (arg))
1217 static VALUE r_object(struct load_arg *arg);
1218 static VALUE r_symbol(struct load_arg *arg);
1219
1220 NORETURN(static void too_short(void));
1221 static void
1222 too_short(void)
1223 {
1224 rb_raise(rb_eArgError, "marshal data too short");
1225 }
1226
1227 static st_index_t
1228 r_prepare(struct load_arg *arg)
1229 {
1230 st_index_t idx = arg->data->num_entries;
1231
1232 st_insert(arg->data, (st_data_t)idx, (st_data_t)Qundef);
1233 return idx;
1234 }
1235
1236 static unsigned char
1237 r_byte1_buffered(struct load_arg *arg)
1238 {
1239 if (arg->buflen == 0) {
1240 long readable = arg->readable < BUFSIZ ? arg->readable : BUFSIZ;
1241 VALUE str, n = LONG2NUM(readable);
1242
1243 str = load_funcall(arg, arg->src, s_read, 1, &n);
1244 if (NIL_P(str)) too_short();
1245 StringValue(str);
1246 memcpy(arg->buf, RSTRING_PTR(str), RSTRING_LEN(str));
1247 arg->offset = 0;
1248 arg->buflen = RSTRING_LEN(str);
1249 }
1250 arg->buflen--;
1251 return arg->buf[arg->offset++];
1252 }
1253
1254 static int
1255 r_byte(struct load_arg *arg)
1256 {
1257 int c;
1258
1259 if (RB_TYPE_P(arg->src, T_STRING)) {
1260 if (RSTRING_LEN(arg->src) > arg->offset) {
1261 c = (unsigned char)RSTRING_PTR(arg->src)[arg->offset++];
1262 }
1263 else {
1264 too_short();
1265 }
1266 }
1267 else {
1268 if (arg->readable >0 || arg->buflen > 0) {
1269 c = r_byte1_buffered(arg);
1270 }
1271 else {
1272 VALUE v = load_funcall(arg, arg->src, s_getbyte, 0, 0);
1273 if (NIL_P(v)) rb_eof_error();
1274 c = (unsigned char)NUM2CHR(v);
1275 }
1276 }
1277 return c;
1278 }
1279
1280 NORETURN(static void long_toobig(int size));
1281
1282 static void
1283 long_toobig(int size)
1284 {
1285 rb_raise(rb_eTypeError, "long too big for this architecture (size "
1286 STRINGIZE(SIZEOF_LONG)", given %d)", size);
1287 }
1288
1289 static long
1290 r_long(struct load_arg *arg)
1291 {
1292 register long x;
1293 int c = (signed char)r_byte(arg);
1294 long i;
1295
1296 if (c == 0) return 0;
1297 if (c > 0) {
1298 if (4 < c && c < 128) {
1299 return c - 5;
1300 }
1301 if (c > (int)sizeof(long)) long_toobig(c);
1302 x = 0;
1303 for (i=0;i<c;i++) {
1304 x |= (long)r_byte(arg) << (8*i);
1305 }
1306 }
1307 else {
1308 if (-129 < c && c < -4) {
1309 return c + 5;
1310 }
1311 c = -c;
1312 if (c > (int)sizeof(long)) long_toobig(c);
1313 x = -1;
1314 for (i=0;i<c;i++) {
1315 x &= ~((long)0xff << (8*i));
1316 x |= (long)r_byte(arg) << (8*i);
1317 }
1318 }
1319 return x;
1320 }
1321
1322 long
1323 ruby_marshal_read_long(const char **buf, long len)
1324 {
1325 long x;
1326 struct RString src;
1327 struct load_arg arg;
1328 memset(&arg, 0, sizeof(arg));
1329 arg.src = rb_setup_fake_str(&src, *buf, len, 0);
1330 x = r_long(&arg);
1331 *buf += arg.offset;
1332 return x;
1333 }
1334
1335 static VALUE
1336 r_bytes1(long len, struct load_arg *arg)
1337 {
1338 VALUE str, n = LONG2NUM(len);
1339
1340 str = load_funcall(arg, arg->src, s_read, 1, &n);
1341 if (NIL_P(str)) too_short();
1342 StringValue(str);
1343 if (RSTRING_LEN(str) != len) too_short();
1344
1345 return str;
1346 }
1347
1348 static VALUE
1349 r_bytes1_buffered(long len, struct load_arg *arg)
1350 {
1351 VALUE str;
1352
1353 if (len <= arg->buflen) {
1354 str = rb_str_new(arg->buf+arg->offset, len);
1355 arg->offset += len;
1356 arg->buflen -= len;
1357 }
1358 else {
1359 long buflen = arg->buflen;
1360 long readable = arg->readable + 1;
1361 long tmp_len, read_len, need_len = len - buflen;
1362 VALUE tmp, n;
1363
1364 readable = readable < BUFSIZ ? readable : BUFSIZ;
1365 read_len = need_len > readable ? need_len : readable;
1366 n = LONG2NUM(read_len);
1367 tmp = load_funcall(arg, arg->src, s_read, 1, &n);
1368 if (NIL_P(tmp)) too_short();
1369 StringValue(tmp);
1370
1371 tmp_len = RSTRING_LEN(tmp);
1372
1373 if (tmp_len < need_len) too_short();
1374
1375 str = rb_str_new(arg->buf+arg->offset, buflen);
1376 rb_str_cat(str, RSTRING_PTR(tmp), need_len);
1377
1378 if (tmp_len > need_len) {
1379 buflen = tmp_len - need_len;
1380 memcpy(arg->buf, RSTRING_PTR(tmp)+need_len, buflen);
1381 arg->buflen = buflen;
1382 }
1383 else {
1384 arg->buflen = 0;
1385 }
1386 arg->offset = 0;
1387 }
1388
1389 return str;
1390 }
1391
1392 #define r_bytes(arg) r_bytes0(r_long(arg), (arg))
1393
1394 static VALUE
1395 r_bytes0(long len, struct load_arg *arg)
1396 {
1397 VALUE str;
1398
1399 if (len == 0) return rb_str_new(0, 0);
1400 if (RB_TYPE_P(arg->src, T_STRING)) {
1401 if (RSTRING_LEN(arg->src) - arg->offset >= len) {
1402 str = rb_str_new(RSTRING_PTR(arg->src)+arg->offset, len);
1403 arg->offset += len;
1404 }
1405 else {
1406 too_short();
1407 }
1408 }
1409 else {
1410 if (arg->readable > 0 || arg->buflen > 0) {
1411 str = r_bytes1_buffered(len, arg);
1412 }
1413 else {
1414 str = r_bytes1(len, arg);
1415 }
1416 }
1417 return str;
1418 }
1419
1420 static inline int
1421 name_equal(const char *name, size_t nlen, const char *p, long l)
1422 {
1423 if ((size_t)l != nlen || *p != *name) return 0;
1424 return nlen == 1 || memcmp(p+1, name+1, nlen-1) == 0;
1425 }
1426
1427 static int
1428 sym2encidx(VALUE sym, VALUE val)
1429 {
1430 static const char name_encoding[8] = "encoding";
1431 const char *p;
1432 long l;
1433 if (rb_enc_get_index(sym) != ENCINDEX_US_ASCII) return -1;
1434 RSTRING_GETMEM(sym, p, l);
1435 if (l <= 0) return -1;
1436 if (name_equal(name_encoding, sizeof(name_encoding), p, l)) {
1437 int idx = rb_enc_find_index(StringValueCStr(val));
1438 return idx;
1439 }
1440 if (name_equal(name_s_encoding_short, rb_strlen_lit(name_s_encoding_short), p, l)) {
1441 if (val == Qfalse) return rb_usascii_encindex();
1442 else if (val == Qtrue) return rb_utf8_encindex();
1443 /* bogus ignore */
1444 }
1445 return -1;
1446 }
1447
1448 static int
1449 symname_equal(VALUE sym, const char *name, size_t nlen)
1450 {
1451 const char *p;
1452 long l;
1453 if (rb_enc_get_index(sym) != ENCINDEX_US_ASCII) return 0;
1454 RSTRING_GETMEM(sym, p, l);
1455 return name_equal(name, nlen, p, l);
1456 }
1457
1458 #define BUILD_ASSERT_POSITIVE(n) \
1459 /* make 0 negative to workaround the "zero size array" GCC extension, */ \
1460 ((sizeof(char [2*(ssize_t)(n)-1])+1)/2) /* assuming no overflow */
1461 #define symname_equal_lit(sym, sym_name) \
1462 symname_equal(sym, sym_name, BUILD_ASSERT_POSITIVE(rb_strlen_lit(sym_name)))
1463
1464 static VALUE
1465 r_symlink(struct load_arg *arg)
1466 {
1467 st_data_t sym;
1468 long num = r_long(arg);
1469
1470 if (!st_lookup(arg->symbols, num, &sym)) {
1471 rb_raise(rb_eArgError, "bad symbol");
1472 }
1473 return (VALUE)sym;
1474 }
1475
1476 static VALUE
1477 r_symreal(struct load_arg *arg, int ivar)
1478 {
1479 VALUE s = r_bytes(arg);
1480 VALUE sym;
1481 int idx = -1;
1482 st_index_t n = arg->symbols->num_entries;
1483
1484 if (rb_enc_str_asciionly_p(s)) rb_enc_associate_index(s, ENCINDEX_US_ASCII);
1485 st_insert(arg->symbols, (st_data_t)n, (st_data_t)s);
1486 if (ivar) {
1487 long num = r_long(arg);
1488 while (num-- > 0) {
1489 sym = r_symbol(arg);
1490 idx = sym2encidx(sym, r_object(arg));
1491 }
1492 }
1493 if (idx > 0) {
1494 rb_enc_associate_index(s, idx);
1495 if (rb_enc_str_coderange(s) == ENC_CODERANGE_BROKEN) {
1496 rb_raise(rb_eArgError, "invalid byte sequence in %s: %+"PRIsVALUE,
1497 rb_enc_name(rb_enc_from_index(idx)), s);
1498 }
1499 }
1500
1501 return s;
1502 }
1503
1504 static VALUE
1505 r_symbol(struct load_arg *arg)
1506 {
1507 int type, ivar = 0;
1508
1509 again:
1510 switch ((type = r_byte(arg))) {
1511 default:
1512 rb_raise(rb_eArgError, "dump format error for symbol(0x%x)", type);
1513 case TYPE_IVAR:
1514 ivar = 1;
1515 goto again;
1516 case TYPE_SYMBOL:
1517 return r_symreal(arg, ivar);
1518 case TYPE_SYMLINK:
1519 if (ivar) {
1520 rb_raise(rb_eArgError, "dump format error (symlink with encoding)");
1521 }
1522 return r_symlink(arg);
1523 }
1524 }
1525
1526 static VALUE
1527 r_unique(struct load_arg *arg)
1528 {
1529 return r_symbol(arg);
1530 }
1531
1532 static VALUE
1533 r_string(struct load_arg *arg)
1534 {
1535 return r_bytes(arg);
1536 }
1537
1538 static VALUE
1539 r_entry0(VALUE v, st_index_t num, struct load_arg *arg)
1540 {
1541 st_data_t real_obj = (st_data_t)v;
1542 if (arg->compat_tbl) {
1543 /* real_obj is kept if not found */
1544 st_lookup(arg->compat_tbl, v, &real_obj);
1545 }
1546 st_insert(arg->data, num, real_obj);
1547 st_insert(arg->partial_objects, (st_data_t)real_obj, Qtrue);
1548 return v;
1549 }
1550
1551 static VALUE
1552 r_fixup_compat(VALUE v, struct load_arg *arg)
1553 {
1554 st_data_t data;
1555 st_data_t key = (st_data_t)v;
1556 if (arg->compat_tbl && st_delete(arg->compat_tbl, &key, &data)) {
1557 VALUE real_obj = (VALUE)data;
1558 rb_alloc_func_t allocator = rb_get_alloc_func(CLASS_OF(real_obj));
1559 if (st_lookup(compat_allocator_tbl, (st_data_t)allocator, &data)) {
1560 marshal_compat_t *compat = (marshal_compat_t*)data;
1561 compat->loader(real_obj, v);
1562 }
1563 v = real_obj;
1564 }
1565 return v;
1566 }
1567
1568 static VALUE
1569 r_post_proc(VALUE v, struct load_arg *arg)
1570 {
1571 if (arg->proc) {
1572 v = load_funcall(arg, arg->proc, s_call, 1, &v);
1573 }
1574 return v;
1575 }
1576
1577 static VALUE
1578 r_leave(VALUE v, struct load_arg *arg, bool partial)
1579 {
1580 v = r_fixup_compat(v, arg);
1581 if (!partial) {
1582 st_data_t data;
1583 st_data_t key = (st_data_t)v;
1584 st_delete(arg->partial_objects, &key, &data);
1585 if (arg->freeze) {
1586 if (RB_TYPE_P(v, T_MODULE) || RB_TYPE_P(v, T_CLASS)) {
1587 // noop
1588 }
1589 else if (RB_TYPE_P(v, T_STRING)) {
1590 v = rb_str_to_interned_str(v);
1591 }
1592 else {
1593 OBJ_FREEZE(v);
1594 }
1595 }
1596 v = r_post_proc(v, arg);
1597 }
1598 return v;
1599 }
1600
1601 static int
1602 copy_ivar_i(st_data_t key, st_data_t val, st_data_t arg)
1603 {
1604 VALUE obj = (VALUE)arg, value = (VALUE)val;
1605 ID vid = (ID)key;
1606
1607 if (!rb_ivar_defined(obj, vid))
1608 rb_ivar_set(obj, vid, value);
1609 return ST_CONTINUE;
1610 }
1611
1612 static VALUE
1613 r_copy_ivar(VALUE v, VALUE data)
1614 {
1615 rb_ivar_foreach(data, copy_ivar_i, (st_data_t)v);
1616 return v;
1617 }
1618
1619 static void
1620 r_ivar(VALUE obj, int *has_encoding, struct load_arg *arg)
1621 {
1622 long len;
1623
1624 len = r_long(arg);
1625 if (len > 0) {
1626 do {
1627 VALUE sym = r_symbol(arg);
1628 VALUE val = r_object(arg);
1629 int idx = sym2encidx(sym, val);
1630 if (idx >= 0) {
1631 if (rb_enc_capable(obj)) {
1632 rb_enc_associate_index(obj, idx);
1633 }
1634 else {
1635 rb_raise(rb_eArgError, "%"PRIsVALUE" is not enc_capable", obj);
1636 }
1637 if (has_encoding) *has_encoding = TRUE;
1638 }
1639 else if (symname_equal_lit(sym, name_s_ruby2_keywords_flag)) {
1640 if (RB_TYPE_P(obj, T_HASH)) {
1641 rb_hash_ruby2_keywords(obj);
1642 }
1643 else {
1644 rb_raise(rb_eArgError, "ruby2_keywords flag is given but %"PRIsVALUE" is not a Hash", obj);
1645 }
1646 }
1647 else {
1648 rb_ivar_set(obj, rb_intern_str(sym), val);
1649 }
1650 } while (--len > 0);
1651 }
1652 }
1653
1654 static VALUE
1655 path2class(VALUE path)
1656 {
1657 VALUE v = rb_path_to_class(path);
1658
1659 if (!RB_TYPE_P(v, T_CLASS)) {
1660 rb_raise(rb_eArgError, "%"PRIsVALUE" does not refer to class", path);
1661 }
1662 return v;
1663 }
1664
1665 #define path2module(path) must_be_module(rb_path_to_class(path), path)
1666
1667 static VALUE
1668 must_be_module(VALUE v, VALUE path)
1669 {
1670 if (!RB_TYPE_P(v, T_MODULE)) {
1671 rb_raise(rb_eArgError, "%"PRIsVALUE" does not refer to module", path);
1672 }
1673 return v;
1674 }
1675
1676 static VALUE
1677 obj_alloc_by_klass(VALUE klass, struct load_arg *arg, VALUE *oldclass)
1678 {
1679 st_data_t data;
1680 rb_alloc_func_t allocator;
1681
1682 allocator = rb_get_alloc_func(klass);
1683 if (st_lookup(compat_allocator_tbl, (st_data_t)allocator, &data)) {
1684 marshal_compat_t *compat = (marshal_compat_t*)data;
1685 VALUE real_obj = rb_obj_alloc(klass);
1686 VALUE obj = rb_obj_alloc(compat->oldclass);
1687 if (oldclass) *oldclass = compat->oldclass;
1688
1689 if (!arg->compat_tbl) {
1690 arg->compat_tbl = rb_init_identtable();
1691 }
1692 st_insert(arg->compat_tbl, (st_data_t)obj, (st_data_t)real_obj);
1693 return obj;
1694 }
1695
1696 return rb_obj_alloc(klass);
1697 }
1698
1699 static VALUE
1700 obj_alloc_by_path(VALUE path, struct load_arg *arg)
1701 {
1702 return obj_alloc_by_klass(path2class(path), arg, 0);
1703 }
1704
1705 static VALUE
1706 append_extmod(VALUE obj, VALUE extmod)
1707 {
1708 long i = RARRAY_LEN(extmod);
1709 while (i > 0) {
1710 VALUE m = RARRAY_AREF(extmod, --i);
1711 rb_extend_object(obj, m);
1712 }
1713 return obj;
1714 }
1715
1716 #define prohibit_ivar(type, str) do { \
1717 if (!ivp || !*ivp) break; \
1718 rb_raise(rb_eTypeError, \
1719 "can't override instance variable of "type" `%"PRIsVALUE"'", \
1720 (str)); \
1721 } while (0)
1722
1723 static VALUE r_object_for(struct load_arg *arg, bool partial, int *ivp, VALUE extmod, int type);
1724
1725 static VALUE
1726 r_object0(struct load_arg *arg, bool partial, int *ivp, VALUE extmod)
1727 {
1728 int type = r_byte(arg);
1729 return r_object_for(arg, partial, ivp, extmod, type);
1730 }
1731
1732 static VALUE
1733 r_object_for(struct load_arg *arg, bool partial, int *ivp, VALUE extmod, int type)
1734 {
1735 VALUE (*hash_new_with_size)(st_index_t) = rb_hash_new_with_size;
1736 VALUE v = Qnil;
1737 long id;
1738 st_data_t link;
1739
1740 switch (type) {
1741 case TYPE_LINK:
1742 id = r_long(arg);
1743 if (!st_lookup(arg->data, (st_data_t)id, &link)) {
1744 rb_raise(rb_eArgError, "dump format error (unlinked)");
1745 }
1746 v = (VALUE)link;
1747 if (!st_lookup(arg->partial_objects, (st_data_t)v, &link)) {
1748 v = r_post_proc(v, arg);
1749 }
1750 break;
1751
1752 case TYPE_IVAR:
1753 {
1754 int ivar = TRUE;
1755
1756 v = r_object0(arg, true, &ivar, extmod);
1757 if (ivar) r_ivar(v, NULL, arg);
1758 v = r_leave(v, arg, partial);
1759 }
1760 break;
1761
1762 case TYPE_EXTENDED:
1763 {
1764 VALUE path = r_unique(arg);
1765 VALUE m = rb_path_to_class(path);
1766 if (NIL_P(extmod)) extmod = rb_ary_tmp_new(0);
1767
1768 if (RB_TYPE_P(m, T_CLASS)) { /* prepended */
1769 VALUE c;
1770
1771 v = r_object0(arg, true, 0, Qnil);
1772 c = CLASS_OF(v);
1773 if (c != m || FL_TEST(c, FL_SINGLETON)) {
1774 rb_raise(rb_eArgError,
1775 "prepended class %"PRIsVALUE" differs from class %"PRIsVALUE,
1776 path, rb_class_name(c));
1777 }
1778 c = rb_singleton_class(v);
1779 while (RARRAY_LEN(extmod) > 0) {
1780 m = rb_ary_pop(extmod);
1781 rb_prepend_module(c, m);
1782 }
1783 }
1784 else {
1785 must_be_module(m, path);
1786 rb_ary_push(extmod, m);
1787
1788 v = r_object0(arg, true, 0, extmod);
1789 while (RARRAY_LEN(extmod) > 0) {
1790 m = rb_ary_pop(extmod);
1791 rb_extend_object(v, m);
1792 }
1793 }
1794 }
1795 break;
1796
1797 case TYPE_UCLASS:
1798 {
1799 VALUE c = path2class(r_unique(arg));
1800
1801 if (FL_TEST(c, FL_SINGLETON)) {
1802 rb_raise(rb_eTypeError, "singleton can't be loaded");
1803 }
1804 type = r_byte(arg);
1805 if ((c == rb_cHash) &&
1806 /* Hack for compare_by_identify */
1807 (type == TYPE_HASH || type == TYPE_HASH_DEF)) {
1808 hash_new_with_size = rb_ident_hash_new_with_size;
1809 goto type_hash;
1810 }
1811 v = r_object_for(arg, partial, 0, extmod, type);
1812 if (rb_special_const_p(v) || RB_TYPE_P(v, T_OBJECT) || RB_TYPE_P(v, T_CLASS)) {
1813 goto format_error;
1814 }
1815 if (RB_TYPE_P(v, T_MODULE) || !RTEST(rb_class_inherited_p(c, RBASIC(v)->klass))) {
1816 VALUE tmp = rb_obj_alloc(c);
1817
1818 if (TYPE(v) != TYPE(tmp)) goto format_error;
1819 }
1820 RBASIC_SET_CLASS(v, c);
1821 }
1822 break;
1823
1824 format_error:
1825 rb_raise(rb_eArgError, "dump format error (user class)");
1826
1827 case TYPE_NIL:
1828 v = Qnil;
1829 v = r_leave(v, arg, false);
1830 break;
1831
1832 case TYPE_TRUE:
1833 v = Qtrue;
1834 v = r_leave(v, arg, false);
1835 break;
1836
1837 case TYPE_FALSE:
1838 v = Qfalse;
1839 v = r_leave(v, arg, false);
1840 break;
1841
1842 case TYPE_FIXNUM:
1843 {
1844 long i = r_long(arg);
1845 v = LONG2FIX(i);
1846 }
1847 v = r_leave(v, arg, false);
1848 break;
1849
1850 case TYPE_FLOAT:
1851 {
1852 double d;
1853 VALUE str = r_bytes(arg);
1854 const char *ptr = RSTRING_PTR(str);
1855
1856 if (strcmp(ptr, "nan") == 0) {
1857 d = nan("");
1858 }
1859 else if (strcmp(ptr, "inf") == 0) {
1860 d = HUGE_VAL;
1861 }
1862 else if (strcmp(ptr, "-inf") == 0) {
1863 d = -HUGE_VAL;
1864 }
1865 else {
1866 char *e;
1867 d = strtod(ptr, &e);
1868 d = load_mantissa(d, e, RSTRING_LEN(str) - (e - ptr));
1869 }
1870 v = DBL2NUM(d);
1871 v = r_entry(v, arg);
1872 v = r_leave(v, arg, false);
1873 }
1874 break;
1875
1876 case TYPE_BIGNUM:
1877 {
1878 long len;
1879 VALUE data;
1880 int sign;
1881
1882 sign = r_byte(arg);
1883 len = r_long(arg);
1884 data = r_bytes0(len * 2, arg);
1885 v = rb_integer_unpack(RSTRING_PTR(data), len, 2, 0,
1886 INTEGER_PACK_LITTLE_ENDIAN | (sign == '-' ? INTEGER_PACK_NEGATIVE : 0));
1887 rb_str_resize(data, 0L);
1888 v = r_entry(v, arg);
1889 v = r_leave(v, arg, false);
1890 }
1891 break;
1892
1893 case TYPE_STRING:
1894 v = r_entry(r_string(arg), arg);
1895 v = r_leave(v, arg, partial);
1896 break;
1897
1898 case TYPE_REGEXP:
1899 {
1900 VALUE str = r_bytes(arg);
1901 int options = r_byte(arg);
1902 int has_encoding = FALSE;
1903 st_index_t idx = r_prepare(arg);
1904
1905 if (ivp) {
1906 r_ivar(str, &has_encoding, arg);
1907 *ivp = FALSE;
1908 }
1909 if (!has_encoding) {
1910 /* 1.8 compatibility; remove escapes undefined in 1.8 */
1911 char *ptr = RSTRING_PTR(str), *dst = ptr, *src = ptr;
1912 long len = RSTRING_LEN(str);
1913 long bs = 0;
1914 for (; len-- > 0; *dst++ = *src++) {
1915 switch (*src) {
1916 case '\\': bs++; break;
1917 case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
1918 case 'm': case 'o': case 'p': case 'q': case 'u': case 'y':
1919 case 'E': case 'F': case 'H': case 'I': case 'J': case 'K':
1920 case 'L': case 'N': case 'O': case 'P': case 'Q': case 'R':
1921 case 'S': case 'T': case 'U': case 'V': case 'X': case 'Y':
1922 if (bs & 1) --dst;
1923 /* fall through */
1924 default: bs = 0; break;
1925 }
1926 }
1927 rb_str_set_len(str, dst - ptr);
1928 }
1929 v = r_entry0(rb_reg_new_str(str, options), idx, arg);
1930 v = r_leave(v, arg, partial);
1931 }
1932 break;
1933
1934 case TYPE_ARRAY:
1935 {
1936 long len = r_long(arg);
1937
1938 v = rb_ary_new2(len);
1939 v = r_entry(v, arg);
1940 arg->readable += len - 1;
1941 while (len--) {
1942 rb_ary_push(v, r_object(arg));
1943 arg->readable--;
1944 }
1945 v = r_leave(v, arg, partial);
1946 arg->readable++;
1947 }
1948 break;
1949
1950 case TYPE_HASH:
1951 case TYPE_HASH_DEF:
1952 type_hash:
1953 {
1954 long len = r_long(arg);
1955
1956 v = hash_new_with_size(len);
1957 v = r_entry(v, arg);
1958 arg->readable += (len - 1) * 2;
1959 while (len--) {
1960 VALUE key = r_object(arg);
1961 VALUE value = r_object(arg);
1962 rb_hash_aset(v, key, value);
1963 arg->readable -= 2;
1964 }
1965 arg->readable += 2;
1966 if (type == TYPE_HASH_DEF) {
1967 RHASH_SET_IFNONE(v, r_object(arg));
1968 }
1969 v = r_leave(v, arg, partial);
1970 }
1971 break;
1972
1973 case TYPE_STRUCT:
1974 {
1975 VALUE mem, values;
1976 long i;
1977 VALUE slot;
1978 st_index_t idx = r_prepare(arg);
1979 VALUE klass = path2class(r_unique(arg));
1980 long len = r_long(arg);
1981
1982 v = rb_obj_alloc(klass);
1983 if (!RB_TYPE_P(v, T_STRUCT)) {
1984 rb_raise(rb_eTypeError, "class %"PRIsVALUE" not a struct", rb_class_name(klass));
1985 }
1986 mem = rb_struct_s_members(klass);
1987 if (RARRAY_LEN(mem) != len) {
1988 rb_raise(rb_eTypeError, "struct %"PRIsVALUE" not compatible (struct size differs)",
1989 rb_class_name(klass));
1990 }
1991
1992 arg->readable += (len - 1) * 2;
1993 v = r_entry0(v, idx, arg);
1994 values = rb_ary_new2(len);
1995 {
1996 VALUE keywords = Qfalse;
1997 if (RTEST(rb_struct_s_keyword_init(klass))) {
1998 keywords = rb_hash_new();
1999 rb_ary_push(values, keywords);
2000 }
2001
2002 for (i=0; i<len; i++) {
2003 VALUE n = rb_sym2str(RARRAY_AREF(mem, i));
2004 slot = r_symbol(arg);
2005
2006 if (!rb_str_equal(n, slot)) {
2007 rb_raise(rb_eTypeError, "struct %"PRIsVALUE" not compatible (:%"PRIsVALUE" for :%"PRIsVALUE")",
2008 rb_class_name(klass),
2009 slot, n);
2010 }
2011 if (keywords) {
2012 rb_hash_aset(keywords, RARRAY_AREF(mem, i), r_object(arg));
2013 }
2014 else {
2015 rb_ary_push(values, r_object(arg));
2016 }
2017 arg->readable -= 2;
2018 }
2019 }
2020 rb_struct_initialize(v, values);
2021 v = r_leave(v, arg, partial);
2022 arg->readable += 2;
2023 }
2024 break;
2025
2026 case TYPE_USERDEF:
2027 {
2028 VALUE name = r_unique(arg);
2029 VALUE klass = path2class(name);
2030 VALUE data;
2031 st_data_t d;
2032
2033 if (!rb_obj_respond_to(klass, s_load, TRUE)) {
2034 rb_raise(rb_eTypeError, "class %"PRIsVALUE" needs to have method `_load'",
2035 name);
2036 }
2037 data = r_string(arg);
2038 if (ivp) {
2039 r_ivar(data, NULL, arg);
2040 *ivp = FALSE;
2041 }
2042 v = load_funcall(arg, klass, s_load, 1, &data);
2043 v = r_entry(v, arg);
2044 if (st_lookup(compat_allocator_tbl, (st_data_t)rb_get_alloc_func(klass), &d)) {
2045 marshal_compat_t *compat = (marshal_compat_t*)d;
2046 v = compat->loader(klass, v);
2047 }
2048 if (!partial) v = r_post_proc(v, arg);
2049 }
2050 break;
2051
2052 case TYPE_USRMARSHAL:
2053 {
2054 VALUE name = r_unique(arg);
2055 VALUE klass = path2class(name);
2056 VALUE oldclass = 0;
2057 VALUE data;
2058
2059 v = obj_alloc_by_klass(klass, arg, &oldclass);
2060 if (!NIL_P(extmod)) {
2061 /* for the case marshal_load is overridden */
2062 append_extmod(v, extmod);
2063 }
2064 if (!rb_obj_respond_to(v, s_mload, TRUE)) {
2065 rb_raise(rb_eTypeError, "instance of %"PRIsVALUE" needs to have method `marshal_load'",
2066 name);
2067 }
2068 v = r_entry(v, arg);
2069 data = r_object(arg);
2070 load_funcall(arg, v, s_mload, 1, &data);
2071 v = r_fixup_compat(v, arg);
2072 v = r_copy_ivar(v, data);
2073 v = r_post_proc(v, arg);
2074 if (!NIL_P(extmod)) {
2075 if (oldclass) append_extmod(v, extmod);
2076 rb_ary_clear(extmod);
2077 }
2078 }
2079 break;
2080
2081 case TYPE_OBJECT:
2082 {
2083 st_index_t idx = r_prepare(arg);
2084 v = obj_alloc_by_path(r_unique(arg), arg);
2085 if (!RB_TYPE_P(v, T_OBJECT)) {
2086 rb_raise(rb_eArgError, "dump format error");
2087 }
2088 v = r_entry0(v, idx, arg);
2089 r_ivar(v, NULL, arg);
2090 v = r_leave(v, arg, partial);
2091 }
2092 break;
2093
2094 case TYPE_DATA:
2095 {
2096 VALUE name = r_unique(arg);
2097 VALUE klass = path2class(name);
2098 VALUE oldclass = 0;
2099 VALUE r;
2100
2101 v = obj_alloc_by_klass(klass, arg, &oldclass);
2102 if (!RB_TYPE_P(v, T_DATA)) {
2103 rb_raise(rb_eArgError, "dump format error");
2104 }
2105 v = r_entry(v, arg);
2106 if (!rb_obj_respond_to(v, s_load_data, TRUE)) {
2107 rb_raise(rb_eTypeError,
2108 "class %"PRIsVALUE" needs to have instance method `_load_data'",
2109 name);
2110 }
2111 r = r_object0(arg, partial, 0, extmod);
2112 load_funcall(arg, v, s_load_data, 1, &r);
2113 v = r_leave(v, arg, partial);
2114 }
2115 break;
2116
2117 case TYPE_MODULE_OLD:
2118 {
2119 VALUE str = r_bytes(arg);
2120
2121 v = rb_path_to_class(str);
2122 prohibit_ivar("class/module", str);
2123 v = r_entry(v, arg);
2124 v = r_leave(v, arg, partial);
2125 }
2126 break;
2127
2128 case TYPE_CLASS:
2129 {
2130 VALUE str = r_bytes(arg);
2131
2132 v = path2class(str);
2133 prohibit_ivar("class", str);
2134 v = r_entry(v, arg);
2135 v = r_leave(v, arg, partial);
2136 }
2137 break;
2138
2139 case TYPE_MODULE:
2140 {
2141 VALUE str = r_bytes(arg);
2142
2143 v = path2module(str);
2144 prohibit_ivar("module", str);
2145 v = r_entry(v, arg);
2146 v = r_leave(v, arg, partial);
2147 }
2148 break;
2149
2150 case TYPE_SYMBOL:
2151 if (ivp) {
2152 v = r_symreal(arg, *ivp);
2153 *ivp = FALSE;
2154 }
2155 else {
2156 v = r_symreal(arg, 0);
2157 }
2158 v = rb_str_intern(v);
2159 v = r_leave(v, arg, partial);
2160 break;
2161
2162 case TYPE_SYMLINK:
2163 v = rb_str_intern(r_symlink(arg));
2164 break;
2165
2166 default:
2167 rb_raise(rb_eArgError, "dump format error(0x%x)", type);
2168 break;
2169 }
2170
2171 if (v == Qundef) {
2172 rb_raise(rb_eArgError, "dump format error (bad link)");
2173 }
2174
2175 return v;
2176 }
2177
2178 static VALUE
2179 r_object(struct load_arg *arg)
2180 {
2181 return r_object0(arg, false, 0, Qnil);
2182 }
2183
2184 static void
2185 clear_load_arg(struct load_arg *arg)
2186 {
2187 if (arg->buf) {
2188 xfree(arg->buf);
2189 arg->buf = 0;
2190 }
2191 arg->buflen = 0;
2192 arg->offset = 0;
2193 arg->readable = 0;
2194 if (!arg->symbols) return;
2195 st_free_table(arg->symbols);
2196 arg->symbols = 0;
2197 st_free_table(arg->data);
2198 arg->data = 0;
2199 st_free_table(arg->partial_objects);
2200 arg->partial_objects = 0;
2201 if (arg->compat_tbl) {
2202 st_free_table(arg->compat_tbl);
2203 arg->compat_tbl = 0;
2204 }
2205 }
2206
2207 VALUE
2208 rb_marshal_load_with_proc(VALUE port, VALUE proc, bool freeze)
2209 {
2210 int major, minor;
2211 VALUE v;
2212 VALUE wrapper; /* used to avoid memory leak in case of exception */
2213 struct load_arg *arg;
2214
2215 v = rb_check_string_type(port);
2216 if (!NIL_P(v)) {
2217 port = v;
2218 }
2219 else if (rb_respond_to(port, s_getbyte) && rb_respond_to(port, s_read)) {
2220 rb_check_funcall(port, s_binmode, 0, 0);
2221 }
2222 else {
2223 io_needed();
2224 }
2225 wrapper = TypedData_Make_Struct(0, struct load_arg, &load_arg_data, arg);
2226 arg->src = port;
2227 arg->offset = 0;
2228 arg->symbols = st_init_numtable();
2229 arg->data = rb_init_identtable();
2230 arg->partial_objects = rb_init_identtable();
2231 arg->compat_tbl = 0;
2232 arg->proc = 0;
2233 arg->readable = 0;
2234 arg->freeze = freeze;
2235
2236 if (NIL_P(v))
2237 arg->buf = xmalloc(BUFSIZ);
2238 else
2239 arg->buf = 0;
2240
2241 major = r_byte(arg);
2242 minor = r_byte(arg);
2243 if (major != MARSHAL_MAJOR || minor > MARSHAL_MINOR) {
2244 clear_load_arg(arg);
2245 rb_raise(rb_eTypeError, "incompatible marshal file format (can't be read)\n\
2246 \tformat version %d.%d required; %d.%d given",
2247 MARSHAL_MAJOR, MARSHAL_MINOR, major, minor);
2248 }
2249 if (RTEST(ruby_verbose) && minor != MARSHAL_MINOR) {
2250 rb_warn("incompatible marshal file format (can be read)\n\
2251 \tformat version %d.%d required; %d.%d given",
2252 MARSHAL_MAJOR, MARSHAL_MINOR, major, minor);
2253 }
2254
2255 if (!NIL_P(proc)) arg->proc = proc;
2256 v = r_object(arg);
2257 clear_load_arg(arg);
2258 RB_GC_GUARD(wrapper);
2259
2260 return v;
2261 }
2262
2263 static VALUE marshal_load(rb_execution_context_t *ec, VALUE mod, VALUE source, VALUE proc, VALUE freeze)
2264 {
2265 return rb_marshal_load_with_proc(source, proc, RTEST(freeze));
2266 }
2267
2268 #include "marshal.rbinc"
2269
2270 /*
2271 * The marshaling library converts collections of Ruby objects into a
2272 * byte stream, allowing them to be stored outside the currently
2273 * active script. This data may subsequently be read and the original
2274 * objects reconstituted.
2275 *
2276 * Marshaled data has major and minor version numbers stored along
2277 * with the object information. In normal use, marshaling can only
2278 * load data written with the same major version number and an equal
2279 * or lower minor version number. If Ruby's ``verbose'' flag is set
2280 * (normally using -d, -v, -w, or --verbose) the major and minor
2281 * numbers must match exactly. Marshal versioning is independent of
2282 * Ruby's version numbers. You can extract the version by reading the
2283 * first two bytes of marshaled data.
2284 *
2285 * str = Marshal.dump("thing")
2286 * RUBY_VERSION #=> "1.9.0"
2287 * str[0].ord #=> 4
2288 * str[1].ord #=> 8
2289 *
2290 * Some objects cannot be dumped: if the objects to be dumped include
2291 * bindings, procedure or method objects, instances of class IO, or
2292 * singleton objects, a TypeError will be raised.
2293 *
2294 * If your class has special serialization needs (for example, if you
2295 * want to serialize in some specific format), or if it contains
2296 * objects that would otherwise not be serializable, you can implement
2297 * your own serialization strategy.
2298 *
2299 * There are two methods of doing this, your object can define either
2300 * marshal_dump and marshal_load or _dump and _load. marshal_dump will take
2301 * precedence over _dump if both are defined. marshal_dump may result in
2302 * smaller Marshal strings.
2303 *
2304 * == Security considerations
2305 *
2306 * By design, Marshal.load can deserialize almost any class loaded into the
2307 * Ruby process. In many cases this can lead to remote code execution if the
2308 * Marshal data is loaded from an untrusted source.
2309 *
2310 * As a result, Marshal.load is not suitable as a general purpose serialization
2311 * format and you should never unmarshal user supplied input or other untrusted
2312 * data.
2313 *
2314 * If you need to deserialize untrusted data, use JSON or another serialization
2315 * format that is only able to load simple, 'primitive' types such as String,
2316 * Array, Hash, etc. Never allow user input to specify arbitrary types to
2317 * deserialize into.
2318 *
2319 * == marshal_dump and marshal_load
2320 *
2321 * When dumping an object the method marshal_dump will be called.
2322 * marshal_dump must return a result containing the information necessary for
2323 * marshal_load to reconstitute the object. The result can be any object.
2324 *
2325 * When loading an object dumped using marshal_dump the object is first
2326 * allocated then marshal_load is called with the result from marshal_dump.
2327 * marshal_load must recreate the object from the information in the result.
2328 *
2329 * Example:
2330 *
2331 * class MyObj
2332 * def initialize name, version, data
2333 * @name = name
2334 * @version = version
2335 * @data = data
2336 * end
2337 *
2338 * def marshal_dump
2339 * [@name, @version]
2340 * end
2341 *
2342 * def marshal_load array
2343 * @name, @version = array
2344 * end
2345 * end
2346 *
2347 * == _dump and _load
2348 *
2349 * Use _dump and _load when you need to allocate the object you're restoring
2350 * yourself.
2351 *
2352 * When dumping an object the instance method _dump is called with an Integer
2353 * which indicates the maximum depth of objects to dump (a value of -1 implies
2354 * that you should disable depth checking). _dump must return a String
2355 * containing the information necessary to reconstitute the object.
2356 *
2357 * The class method _load should take a String and use it to return an object
2358 * of the same class.
2359 *
2360 * Example:
2361 *
2362 * class MyObj
2363 * def initialize name, version, data
2364 * @name = name
2365 * @version = version
2366 * @data = data
2367 * end
2368 *
2369 * def _dump level
2370 * [@name, @version].join ':'
2371 * end
2372 *
2373 * def self._load args
2374 * new(*args.split(':'))
2375 * end
2376 * end
2377 *
2378 * Since Marshal.dump outputs a string you can have _dump return a Marshal
2379 * string which is Marshal.loaded in _load for complex objects.
2380 */
2381 void
2382 Init_marshal(void)
2383 {
2384 VALUE rb_mMarshal = rb_define_module("Marshal");
2385 #define set_id(sym) sym = rb_intern_const(name_##sym)
2386 set_id(s_dump);
2387 set_id(s_load);
2388 set_id(s_mdump);
2389 set_id(s_mload);
2390 set_id(s_dump_data);
2391 set_id(s_load_data);
2392 set_id(s_alloc);
2393 set_id(s_call);
2394 set_id(s_getbyte);
2395 set_id(s_read);
2396 set_id(s_write);
2397 set_id(s_binmode);
2398 set_id(s_encoding_short);
2399 set_id(s_ruby2_keywords_flag);
2400
2401 rb_define_module_function(rb_mMarshal, "dump", marshal_dump, -1);
2402
2403 /* major version */
2404 rb_define_const(rb_mMarshal, "MAJOR_VERSION", INT2FIX(MARSHAL_MAJOR));
2405 /* minor version */
2406 rb_define_const(rb_mMarshal, "MINOR_VERSION", INT2FIX(MARSHAL_MINOR));
2407 }
2408
2409 static st_table *
2410 compat_allocator_table(void)
2411 {
2412 if (compat_allocator_tbl) return compat_allocator_tbl;
2413 compat_allocator_tbl = st_init_numtable();
2414 #undef RUBY_UNTYPED_DATA_WARNING
2415 #define RUBY_UNTYPED_DATA_WARNING 0
2416 compat_allocator_tbl_wrapper =
2417 Data_Wrap_Struct(0, mark_marshal_compat_t, 0, compat_allocator_tbl);
2418 rb_gc_register_mark_object(compat_allocator_tbl_wrapper);
2419 return compat_allocator_tbl;
2420 }
2421
2422 VALUE
2423 rb_marshal_dump(VALUE obj, VALUE port)
2424 {
2425 return rb_marshal_dump_limited(obj, port, -1);
2426 }
2427
2428 VALUE
2429 rb_marshal_load(VALUE port)
2430 {
2431 return rb_marshal_load_with_proc(port, Qnil, false);
2432 }
Ruby programming language (80x24.org development)