?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Tempfile document updated.
[ruby.git] / marshal.c
blob83134371fa08a30abca5028157878240737da610
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/numeric.h"
32 #include "internal/object.h"
33 #include "internal/struct.h"
34 #include "internal/symbol.h"
35 #include "internal/util.h"
36 #include "internal/vm.h"
37 #include "ruby/io.h"
38 #include "ruby/ruby.h"
39 #include "ruby/st.h"
40 #include "ruby/util.h"
41 #include "builtin.h"
42 #include "shape.h"
43
44 #define BITSPERSHORT (2*CHAR_BIT)
45 #define SHORTMASK ((1<<BITSPERSHORT)-1)
46 #define SHORTDN(x) RSHIFT((x),BITSPERSHORT)
47
48 #if SIZEOF_SHORT == SIZEOF_BDIGIT
49 #define SHORTLEN(x) (x)
50 #else
51 static size_t
52 shortlen(size_t len, BDIGIT *ds)
53 {
54 BDIGIT num;
55 int offset = 0;
56
57 num = ds[len-1];
58 while (num) {
59 num = SHORTDN(num);
60 offset++;
61 }
62 return (len - 1)*SIZEOF_BDIGIT/2 + offset;
63 }
64 #define SHORTLEN(x) shortlen((x),d)
65 #endif
66
67 #define MARSHAL_MAJOR 4
68 #define MARSHAL_MINOR 8
69
70 #define TYPE_NIL '0'
71 #define TYPE_TRUE 'T'
72 #define TYPE_FALSE 'F'
73 #define TYPE_FIXNUM 'i'
74
75 #define TYPE_EXTENDED 'e'
76 #define TYPE_UCLASS 'C'
77 #define TYPE_OBJECT 'o'
78 #define TYPE_DATA 'd'
79 #define TYPE_USERDEF 'u'
80 #define TYPE_USRMARSHAL 'U'
81 #define TYPE_FLOAT 'f'
82 #define TYPE_BIGNUM 'l'
83 #define TYPE_STRING '"'
84 #define TYPE_REGEXP '/'
85 #define TYPE_ARRAY '['
86 #define TYPE_HASH '{'
87 #define TYPE_HASH_DEF '}'
88 #define TYPE_STRUCT 'S'
89 #define TYPE_MODULE_OLD 'M'
90 #define TYPE_CLASS 'c'
91 #define TYPE_MODULE 'm'
92
93 #define TYPE_SYMBOL ':'
94 #define TYPE_SYMLINK ';'
95
96 #define TYPE_IVAR 'I'
97 #define TYPE_LINK '@'
98
99 static ID s_dump, s_load, s_mdump, s_mload;
100 static ID s_dump_data, s_load_data, s_alloc, s_call;
101 static ID s_getbyte, s_read, s_write, s_binmode;
102 static ID s_encoding_short, s_ruby2_keywords_flag;
103
104 #define name_s_dump "_dump"
105 #define name_s_load "_load"
106 #define name_s_mdump "marshal_dump"
107 #define name_s_mload "marshal_load"
108 #define name_s_dump_data "_dump_data"
109 #define name_s_load_data "_load_data"
110 #define name_s_alloc "_alloc"
111 #define name_s_call "call"
112 #define name_s_getbyte "getbyte"
113 #define name_s_read "read"
114 #define name_s_write "write"
115 #define name_s_binmode "binmode"
116 #define name_s_encoding_short "E"
117 #define name_s_ruby2_keywords_flag "K"
118
119 typedef struct {
120 VALUE newclass;
121 VALUE oldclass;
122 VALUE (*dumper)(VALUE);
123 VALUE (*loader)(VALUE, VALUE);
124 } marshal_compat_t;
125
126 static st_table *compat_allocator_tbl;
127 static VALUE compat_allocator_tbl_wrapper;
128 static VALUE rb_marshal_dump_limited(VALUE obj, VALUE port, int limit);
129 static VALUE rb_marshal_load_with_proc(VALUE port, VALUE proc, bool freeze);
130
131 static int
132 mark_marshal_compat_i(st_data_t key, st_data_t value, st_data_t _)
133 {
134 marshal_compat_t *p = (marshal_compat_t *)value;
135 rb_gc_mark(p->newclass);
136 rb_gc_mark(p->oldclass);
137 return ST_CONTINUE;
138 }
139
140 static void
141 mark_marshal_compat_t(void *tbl)
142 {
143 if (!tbl) return;
144 st_foreach(tbl, mark_marshal_compat_i, 0);
145 }
146
147 static st_table *compat_allocator_table(void);
148
149 void
150 rb_marshal_define_compat(VALUE newclass, VALUE oldclass, VALUE (*dumper)(VALUE), VALUE (*loader)(VALUE, VALUE))
151 {
152 marshal_compat_t *compat;
153 rb_alloc_func_t allocator = rb_get_alloc_func(newclass);
154
155 if (!allocator) {
156 rb_raise(rb_eTypeError, "no allocator");
157 }
158
159 compat = ALLOC(marshal_compat_t);
160 compat->newclass = Qnil;
161 compat->oldclass = Qnil;
162 compat->newclass = newclass;
163 compat->oldclass = oldclass;
164 compat->dumper = dumper;
165 compat->loader = loader;
166
167 st_insert(compat_allocator_table(), (st_data_t)allocator, (st_data_t)compat);
168 }
169
170 struct dump_arg {
171 VALUE str, dest;
172 st_table *symbols;
173 st_table *data;
174 st_table *compat_tbl;
175 st_table *encodings;
176 st_index_t num_entries;
177 };
178
179 struct dump_call_arg {
180 VALUE obj;
181 struct dump_arg *arg;
182 int limit;
183 };
184
185 static VALUE
186 check_dump_arg(VALUE ret, struct dump_arg *arg, const char *name)
187 {
188 if (!arg->symbols) {
189 rb_raise(rb_eRuntimeError, "Marshal.dump reentered at %s",
190 name);
191 }
192 return ret;
193 }
194
195 static VALUE
196 check_userdump_arg(VALUE obj, ID sym, int argc, const VALUE *argv,
197 struct dump_arg *arg, const char *name)
198 {
199 VALUE ret = rb_funcallv(obj, sym, argc, argv);
200 VALUE klass = CLASS_OF(obj);
201 if (CLASS_OF(ret) == klass) {
202 rb_raise(rb_eRuntimeError, "%"PRIsVALUE"#%s returned same class instance",
203 klass, name);
204 }
205 return check_dump_arg(ret, arg, name);
206 }
207
208 #define dump_funcall(arg, obj, sym, argc, argv) \
209 check_userdump_arg(obj, sym, argc, argv, arg, name_##sym)
210 #define dump_check_funcall(arg, obj, sym, argc, argv) \
211 check_dump_arg(rb_check_funcall(obj, sym, argc, argv), arg, name_##sym)
212
213 static void clear_dump_arg(struct dump_arg *arg);
214
215 static void
216 mark_dump_arg(void *ptr)
217 {
218 struct dump_arg *p = ptr;
219 if (!p->symbols)
220 return;
221 rb_mark_set(p->symbols);
222 rb_mark_set(p->data);
223 rb_mark_hash(p->compat_tbl);
224 rb_gc_mark(p->str);
225 }
226
227 static void
228 free_dump_arg(void *ptr)
229 {
230 clear_dump_arg(ptr);
231 }
232
233 static size_t
234 memsize_dump_arg(const void *ptr)
235 {
236 const struct dump_arg *p = (struct dump_arg *)ptr;
237 size_t memsize = 0;
238 if (p->symbols) memsize += rb_st_memsize(p->symbols);
239 if (p->data) memsize += rb_st_memsize(p->data);
240 if (p->compat_tbl) memsize += rb_st_memsize(p->compat_tbl);
241 if (p->encodings) memsize += rb_st_memsize(p->encodings);
242 return memsize;
243 }
244
245 static const rb_data_type_t dump_arg_data = {
246 "dump_arg",
247 {mark_dump_arg, free_dump_arg, memsize_dump_arg,},
248 0, 0, RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_EMBEDDABLE
249 };
250
251 static VALUE
252 must_not_be_anonymous(const char *type, VALUE path)
253 {
254 char *n = RSTRING_PTR(path);
255
256 if (!rb_enc_asciicompat(rb_enc_get(path))) {
257 /* cannot occur? */
258 rb_raise(rb_eTypeError, "can't dump non-ascii %s name % "PRIsVALUE,
259 type, path);
260 }
261 if (n[0] == '#') {
262 rb_raise(rb_eTypeError, "can't dump anonymous %s % "PRIsVALUE,
263 type, path);
264 }
265 return path;
266 }
267
268 static VALUE
269 class2path(VALUE klass)
270 {
271 VALUE path = rb_class_path(klass);
272
273 must_not_be_anonymous((RB_TYPE_P(klass, T_CLASS) ? "class" : "module"), path);
274 if (rb_path_to_class(path) != rb_class_real(klass)) {
275 rb_raise(rb_eTypeError, "% "PRIsVALUE" can't be referred to", path);
276 }
277 return path;
278 }
279
280 int ruby_marshal_write_long(long x, char *buf);
281 static void w_long(long, struct dump_arg*);
282 static int w_encoding(VALUE encname, struct dump_call_arg *arg);
283 static VALUE encoding_name(VALUE obj, struct dump_arg *arg);
284
285 static void
286 w_nbyte(const char *s, long n, struct dump_arg *arg)
287 {
288 VALUE buf = arg->str;
289 rb_str_buf_cat(buf, s, n);
290 if (arg->dest && RSTRING_LEN(buf) >= BUFSIZ) {
291 rb_io_write(arg->dest, buf);
292 rb_str_resize(buf, 0);
293 }
294 }
295
296 static void
297 w_byte(char c, struct dump_arg *arg)
298 {
299 w_nbyte(&c, 1, arg);
300 }
301
302 static void
303 w_bytes(const char *s, long n, struct dump_arg *arg)
304 {
305 w_long(n, arg);
306 w_nbyte(s, n, arg);
307 }
308
309 #define w_cstr(s, arg) w_bytes((s), strlen(s), (arg))
310
311 static void
312 w_short(int x, struct dump_arg *arg)
313 {
314 w_byte((char)((x >> 0) & 0xff), arg);
315 w_byte((char)((x >> 8) & 0xff), arg);
316 }
317
318 static void
319 w_long(long x, struct dump_arg *arg)
320 {
321 char buf[sizeof(long)+1];
322 int i = ruby_marshal_write_long(x, buf);
323 if (i < 0) {
324 rb_raise(rb_eTypeError, "long too big to dump");
325 }
326 w_nbyte(buf, i, arg);
327 }
328
329 int
330 ruby_marshal_write_long(long x, char *buf)
331 {
332 int i;
333
334 #if SIZEOF_LONG > 4
335 if (!(RSHIFT(x, 31) == 0 || RSHIFT(x, 31) == -1)) {
336 /* big long does not fit in 4 bytes */
337 return -1;
338 }
339 #endif
340
341 if (x == 0) {
342 buf[0] = 0;
343 return 1;
344 }
345 if (0 < x && x < 123) {
346 buf[0] = (char)(x + 5);
347 return 1;
348 }
349 if (-124 < x && x < 0) {
350 buf[0] = (char)((x - 5)&0xff);
351 return 1;
352 }
353 for (i=1;i<(int)sizeof(long)+1;i++) {
354 buf[i] = (char)(x & 0xff);
355 x = RSHIFT(x,8);
356 if (x == 0) {
357 buf[0] = i;
358 break;
359 }
360 if (x == -1) {
361 buf[0] = -i;
362 break;
363 }
364 }
365 return i+1;
366 }
367
368 #ifdef DBL_MANT_DIG
369 #define DECIMAL_MANT (53-16) /* from IEEE754 double precision */
370
371 #if DBL_MANT_DIG > 32
372 #define MANT_BITS 32
373 #elif DBL_MANT_DIG > 24
374 #define MANT_BITS 24
375 #elif DBL_MANT_DIG > 16
376 #define MANT_BITS 16
377 #else
378 #define MANT_BITS 8
379 #endif
380
381 static double
382 load_mantissa(double d, const char *buf, long len)
383 {
384 if (!len) return d;
385 if (--len > 0 && !*buf++) { /* binary mantissa mark */
386 int e, s = d < 0, dig = 0;
387 unsigned long m;
388
389 modf(ldexp(frexp(fabs(d), &e), DECIMAL_MANT), &d);
390 do {
391 m = 0;
392 switch (len) {
393 default: m = *buf++ & 0xff; /* fall through */
394 #if MANT_BITS > 24
395 case 3: m = (m << 8) | (*buf++ & 0xff); /* fall through */
396 #endif
397 #if MANT_BITS > 16
398 case 2: m = (m << 8) | (*buf++ & 0xff); /* fall through */
399 #endif
400 #if MANT_BITS > 8
401 case 1: m = (m << 8) | (*buf++ & 0xff);
402 #endif
403 }
404 dig -= len < MANT_BITS / 8 ? 8 * (unsigned)len : MANT_BITS;
405 d += ldexp((double)m, dig);
406 } while ((len -= MANT_BITS / 8) > 0);
407 d = ldexp(d, e - DECIMAL_MANT);
408 if (s) d = -d;
409 }
410 return d;
411 }
412 #else
413 #define load_mantissa(d, buf, len) (d)
414 #endif
415
416 #ifdef DBL_DIG
417 #define FLOAT_DIG (DBL_DIG+2)
418 #else
419 #define FLOAT_DIG 17
420 #endif
421
422 static void
423 w_float(double d, struct dump_arg *arg)
424 {
425 char buf[FLOAT_DIG + (DECIMAL_MANT + 7) / 8 + 10];
426
427 if (isinf(d)) {
428 if (d < 0) w_cstr("-inf", arg);
429 else w_cstr("inf", arg);
430 }
431 else if (isnan(d)) {
432 w_cstr("nan", arg);
433 }
434 else if (d == 0.0) {
435 if (signbit(d)) w_cstr("-0", arg);
436 else w_cstr("0", arg);
437 }
438 else {
439 int decpt, sign, digs, len = 0;
440 char *e, *p = ruby_dtoa(d, 0, 0, &decpt, &sign, &e);
441 if (sign) buf[len++] = '-';
442 digs = (int)(e - p);
443 if (decpt < -3 || decpt > digs) {
444 buf[len++] = p[0];
445 if (--digs > 0) buf[len++] = '.';
446 memcpy(buf + len, p + 1, digs);
447 len += digs;
448 len += snprintf(buf + len, sizeof(buf) - len, "e%d", decpt - 1);
449 }
450 else if (decpt > 0) {
451 memcpy(buf + len, p, decpt);
452 len += decpt;
453 if ((digs -= decpt) > 0) {
454 buf[len++] = '.';
455 memcpy(buf + len, p + decpt, digs);
456 len += digs;
457 }
458 }
459 else {
460 buf[len++] = '0';
461 buf[len++] = '.';
462 if (decpt) {
463 memset(buf + len, '0', -decpt);
464 len -= decpt;
465 }
466 memcpy(buf + len, p, digs);
467 len += digs;
468 }
469 free(p);
470 w_bytes(buf, len, arg);
471 }
472 }
473
474 static void
475 w_symbol(VALUE sym, struct dump_arg *arg)
476 {
477 st_data_t num;
478 VALUE encname;
479
480 if (st_lookup(arg->symbols, sym, &num)) {
481 w_byte(TYPE_SYMLINK, arg);
482 w_long((long)num, arg);
483 }
484 else {
485 const VALUE orig_sym = sym;
486 sym = rb_sym2str(sym);
487 if (!sym) {
488 rb_raise(rb_eTypeError, "can't dump anonymous ID %"PRIdVALUE, sym);
489 }
490 encname = encoding_name(sym, arg);
491 if (NIL_P(encname) ||
492 is_ascii_string(sym)) {
493 encname = Qnil;
494 }
495 else {
496 w_byte(TYPE_IVAR, arg);
497 }
498 w_byte(TYPE_SYMBOL, arg);
499 w_bytes(RSTRING_PTR(sym), RSTRING_LEN(sym), arg);
500 st_add_direct(arg->symbols, orig_sym, arg->symbols->num_entries);
501 if (!NIL_P(encname)) {
502 struct dump_call_arg c_arg;
503 c_arg.limit = 1;
504 c_arg.arg = arg;
505 w_long(1L, arg);
506 w_encoding(encname, &c_arg);
507 }
508 }
509 }
510
511 static void
512 w_unique(VALUE s, struct dump_arg *arg)
513 {
514 must_not_be_anonymous("class", s);
515 w_symbol(rb_str_intern(s), arg);
516 }
517
518 static void w_object(VALUE,struct dump_arg*,int);
519
520 static int
521 hash_each(VALUE key, VALUE value, VALUE v)
522 {
523 struct dump_call_arg *arg = (void *)v;
524 w_object(key, arg->arg, arg->limit);
525 w_object(value, arg->arg, arg->limit);
526 return ST_CONTINUE;
527 }
528
529 #define SINGLETON_DUMP_UNABLE_P(klass) \
530 (rb_id_table_size(RCLASS_M_TBL(klass)) > 0 || \
531 rb_ivar_count(klass) > 0)
532
533 static void
534 w_extended(VALUE klass, struct dump_arg *arg, int check)
535 {
536 if (check && RCLASS_SINGLETON_P(klass)) {
537 VALUE origin = RCLASS_ORIGIN(klass);
538 if (SINGLETON_DUMP_UNABLE_P(klass) ||
539 (origin != klass && SINGLETON_DUMP_UNABLE_P(origin))) {
540 rb_raise(rb_eTypeError, "singleton can't be dumped");
541 }
542 klass = RCLASS_SUPER(klass);
543 }
544 while (BUILTIN_TYPE(klass) == T_ICLASS) {
545 if (!FL_TEST(klass, RICLASS_IS_ORIGIN) ||
546 BUILTIN_TYPE(RBASIC(klass)->klass) != T_MODULE) {
547 VALUE path = rb_class_name(RBASIC(klass)->klass);
548 w_byte(TYPE_EXTENDED, arg);
549 w_unique(path, arg);
550 }
551 klass = RCLASS_SUPER(klass);
552 }
553 }
554
555 static void
556 w_class(char type, VALUE obj, struct dump_arg *arg, int check)
557 {
558 VALUE path;
559 st_data_t real_obj;
560 VALUE klass;
561
562 if (arg->compat_tbl &&
563 st_lookup(arg->compat_tbl, (st_data_t)obj, &real_obj)) {
564 obj = (VALUE)real_obj;
565 }
566 klass = CLASS_OF(obj);
567 w_extended(klass, arg, check);
568 w_byte(type, arg);
569 path = class2path(rb_class_real(klass));
570 w_unique(path, arg);
571 }
572
573 static void
574 w_uclass(VALUE obj, VALUE super, struct dump_arg *arg)
575 {
576 VALUE klass = CLASS_OF(obj);
577
578 w_extended(klass, arg, TRUE);
579 klass = rb_class_real(klass);
580 if (klass != super) {
581 w_byte(TYPE_UCLASS, arg);
582 w_unique(class2path(klass), arg);
583 }
584 }
585
586 static bool
587 rb_hash_ruby2_keywords_p(VALUE obj)
588 {
589 return (RHASH(obj)->basic.flags & RHASH_PASS_AS_KEYWORDS) != 0;
590 }
591
592 static void
593 rb_hash_ruby2_keywords(VALUE obj)
594 {
595 RHASH(obj)->basic.flags |= RHASH_PASS_AS_KEYWORDS;
596 }
597
598 static inline bool
599 to_be_skipped_id(const ID id)
600 {
601 if (id == s_encoding_short) return true;
602 if (id == s_ruby2_keywords_flag) return true;
603 if (id == rb_id_encoding()) return true;
604 return !rb_id2str(id);
605 }
606
607 struct w_ivar_arg {
608 struct dump_call_arg *dump;
609 st_data_t num_ivar;
610 };
611
612 static int
613 w_obj_each(ID id, VALUE value, st_data_t a)
614 {
615 struct w_ivar_arg *ivarg = (struct w_ivar_arg *)a;
616 struct dump_call_arg *arg = ivarg->dump;
617
618 if (to_be_skipped_id(id)) {
619 if (id == s_encoding_short) {
620 rb_warn("instance variable '"name_s_encoding_short"' on class %"PRIsVALUE" is not dumped",
621 CLASS_OF(arg->obj));
622 }
623 if (id == s_ruby2_keywords_flag) {
624 rb_warn("instance variable '"name_s_ruby2_keywords_flag"' on class %"PRIsVALUE" is not dumped",
625 CLASS_OF(arg->obj));
626 }
627 return ST_CONTINUE;
628 }
629 --ivarg->num_ivar;
630 w_symbol(ID2SYM(id), arg->arg);
631 w_object(value, arg->arg, arg->limit);
632 return ST_CONTINUE;
633 }
634
635 static int
636 obj_count_ivars(ID id, VALUE val, st_data_t a)
637 {
638 if (!to_be_skipped_id(id) && UNLIKELY(!++*(st_index_t *)a)) {
639 rb_raise(rb_eRuntimeError, "too many instance variables");
640 }
641 return ST_CONTINUE;
642 }
643
644 static VALUE
645 encoding_name(VALUE obj, struct dump_arg *arg)
646 {
647 if (rb_enc_capable(obj)) {
648 int encidx = rb_enc_get_index(obj);
649 rb_encoding *enc = 0;
650 st_data_t name;
651
652 if (encidx <= 0 || !(enc = rb_enc_from_index(encidx))) {
653 return Qnil;
654 }
655
656 /* special treatment for US-ASCII and UTF-8 */
657 if (encidx == rb_usascii_encindex()) {
658 return Qfalse;
659 }
660 else if (encidx == rb_utf8_encindex()) {
661 return Qtrue;
662 }
663
664 if (arg->encodings ?
665 !st_lookup(arg->encodings, (st_data_t)rb_enc_name(enc), &name) :
666 (arg->encodings = st_init_strcasetable(), 1)) {
667 name = (st_data_t)rb_str_new_cstr(rb_enc_name(enc));
668 st_insert(arg->encodings, (st_data_t)rb_enc_name(enc), name);
669 }
670 return (VALUE)name;
671 }
672 else {
673 return Qnil;
674 }
675 }
676
677 static int
678 w_encoding(VALUE encname, struct dump_call_arg *arg)
679 {
680 int limit = arg->limit;
681 if (limit >= 0) ++limit;
682 switch (encname) {
683 case Qfalse:
684 case Qtrue:
685 w_symbol(ID2SYM(s_encoding_short), arg->arg);
686 w_object(encname, arg->arg, limit);
687 return 1;
688 case Qnil:
689 return 0;
690 }
691 w_symbol(ID2SYM(rb_id_encoding()), arg->arg);
692 w_object(encname, arg->arg, limit);
693 return 1;
694 }
695
696 static st_index_t
697 has_ivars(VALUE obj, VALUE encname, VALUE *ivobj)
698 {
699 st_index_t num = !NIL_P(encname);
700
701 if (SPECIAL_CONST_P(obj)) goto generic;
702 switch (BUILTIN_TYPE(obj)) {
703 case T_OBJECT:
704 case T_CLASS:
705 case T_MODULE:
706 break; /* counted elsewhere */
707 case T_HASH:
708 if (rb_hash_ruby2_keywords_p(obj)) ++num;
709 /* fall through */
710 default:
711 generic:
712 rb_ivar_foreach(obj, obj_count_ivars, (st_data_t)&num);
713 if (num) *ivobj = obj;
714 }
715
716 return num;
717 }
718
719 static void
720 w_ivar_each(VALUE obj, st_index_t num, struct dump_call_arg *arg)
721 {
722 shape_id_t shape_id = rb_shape_get_shape_id(arg->obj);
723 struct w_ivar_arg ivarg = {arg, num};
724 if (!num) return;
725 rb_ivar_foreach(obj, w_obj_each, (st_data_t)&ivarg);
726
727 if (shape_id != rb_shape_get_shape_id(arg->obj)) {
728 rb_shape_t * expected_shape = rb_shape_get_shape_by_id(shape_id);
729 rb_shape_t * actual_shape = rb_shape_get_shape(arg->obj);
730
731 // If the shape tree got _shorter_ then we probably removed an IV
732 // If the shape tree got longer, then we probably added an IV.
733 // The exception message might not be accurate when someone adds and
734 // removes the same number of IVs, but they will still get an exception
735 if (rb_shape_depth(expected_shape) > rb_shape_depth(actual_shape)) {
736 rb_raise(rb_eRuntimeError, "instance variable removed from %"PRIsVALUE" instance",
737 CLASS_OF(arg->obj));
738 }
739 else {
740 rb_raise(rb_eRuntimeError, "instance variable added to %"PRIsVALUE" instance",
741 CLASS_OF(arg->obj));
742 }
743 }
744 }
745
746 static void
747 w_ivar(st_index_t num, VALUE ivobj, VALUE encname, struct dump_call_arg *arg)
748 {
749 w_long(num, arg->arg);
750 num -= w_encoding(encname, arg);
751 if (RB_TYPE_P(ivobj, T_HASH) && rb_hash_ruby2_keywords_p(ivobj)) {
752 int limit = arg->limit;
753 if (limit >= 0) ++limit;
754 w_symbol(ID2SYM(s_ruby2_keywords_flag), arg->arg);
755 w_object(Qtrue, arg->arg, limit);
756 num--;
757 }
758 if (!UNDEF_P(ivobj) && num) {
759 w_ivar_each(ivobj, num, arg);
760 }
761 }
762
763 static void
764 w_objivar(VALUE obj, struct dump_call_arg *arg)
765 {
766 st_data_t num = 0;
767
768 rb_ivar_foreach(obj, obj_count_ivars, (st_data_t)&num);
769 w_long(num, arg->arg);
770 w_ivar_each(obj, num, arg);
771 }
772
773 #if SIZEOF_LONG > 4
774 // Optimized dump for fixnum larger than 31-bits
775 static void
776 w_bigfixnum(VALUE obj, struct dump_arg *arg)
777 {
778 RUBY_ASSERT(FIXNUM_P(obj));
779
780 w_byte(TYPE_BIGNUM, arg);
781
782 #if SIZEOF_LONG == SIZEOF_VALUE
783 long num, slen_num;
784 num = FIX2LONG(obj);
785 #else
786 long long num, slen_num;
787 num = NUM2LL(obj);
788 #endif
789
790 char sign = num < 0 ? '-' : '+';
791 w_byte(sign, arg);
792
793 // Guaranteed not to overflow, as FIXNUM is 1-bit less than long
794 if (num < 0) num = -num;
795
796 // calculate the size in shorts
797 int slen = 0;
798 {
799 slen_num = num;
800 while (slen_num) {
801 slen++;
802 slen_num = SHORTDN(slen_num);
803 }
804 }
805
806 RUBY_ASSERT(slen > 0 && slen <= SIZEOF_LONG / 2);
807
808 w_long((long)slen, arg);
809
810 for (int i = 0; i < slen; i++) {
811 w_short(num & SHORTMASK, arg);
812 num = SHORTDN(num);
813 }
814
815 // We aren't adding this object to the link table, but we need to increment
816 // the index.
817 arg->num_entries++;
818
819 RUBY_ASSERT(num == 0);
820 }
821 #endif
822
823 static void
824 w_remember(VALUE obj, struct dump_arg *arg)
825 {
826 st_add_direct(arg->data, obj, arg->num_entries++);
827 }
828
829 static void
830 w_object(VALUE obj, struct dump_arg *arg, int limit)
831 {
832 struct dump_call_arg c_arg;
833 VALUE ivobj = Qundef;
834 st_data_t num;
835 st_index_t hasiv = 0;
836 VALUE encname = Qnil;
837
838 if (limit == 0) {
839 rb_raise(rb_eArgError, "exceed depth limit");
840 }
841
842 if (NIL_P(obj)) {
843 w_byte(TYPE_NIL, arg);
844 }
845 else if (obj == Qtrue) {
846 w_byte(TYPE_TRUE, arg);
847 }
848 else if (obj == Qfalse) {
849 w_byte(TYPE_FALSE, arg);
850 }
851 else if (FIXNUM_P(obj)) {
852 #if SIZEOF_LONG <= 4
853 w_byte(TYPE_FIXNUM, arg);
854 w_long(FIX2INT(obj), arg);
855 #else
856 if (RSHIFT((long)obj, 31) == 0 || RSHIFT((long)obj, 31) == -1) {
857 w_byte(TYPE_FIXNUM, arg);
858 w_long(FIX2LONG(obj), arg);
859 }
860 else {
861 w_bigfixnum(obj, arg);
862 }
863 #endif
864 }
865 else if (SYMBOL_P(obj)) {
866 w_symbol(obj, arg);
867 }
868 else {
869 if (st_lookup(arg->data, obj, &num)) {
870 w_byte(TYPE_LINK, arg);
871 w_long((long)num, arg);
872 return;
873 }
874
875 if (limit > 0) limit--;
876 c_arg.limit = limit;
877 c_arg.arg = arg;
878 c_arg.obj = obj;
879
880 if (FLONUM_P(obj)) {
881 w_remember(obj, arg);
882 w_byte(TYPE_FLOAT, arg);
883 w_float(RFLOAT_VALUE(obj), arg);
884 return;
885 }
886
887 VALUE v;
888
889 if (!RBASIC_CLASS(obj)) {
890 rb_raise(rb_eTypeError, "can't dump internal %s",
891 rb_builtin_type_name(BUILTIN_TYPE(obj)));
892 }
893
894 if (rb_obj_respond_to(obj, s_mdump, TRUE)) {
895 w_remember(obj, arg);
896
897 v = dump_funcall(arg, obj, s_mdump, 0, 0);
898 w_class(TYPE_USRMARSHAL, obj, arg, FALSE);
899 w_object(v, arg, limit);
900 return;
901 }
902 if (rb_obj_respond_to(obj, s_dump, TRUE)) {
903 VALUE ivobj2 = Qundef;
904 st_index_t hasiv2;
905 VALUE encname2;
906
907 v = INT2NUM(limit);
908 v = dump_funcall(arg, obj, s_dump, 1, &v);
909 if (!RB_TYPE_P(v, T_STRING)) {
910 rb_raise(rb_eTypeError, "_dump() must return string");
911 }
912 hasiv = has_ivars(obj, (encname = encoding_name(obj, arg)), &ivobj);
913 hasiv2 = has_ivars(v, (encname2 = encoding_name(v, arg)), &ivobj2);
914 if (hasiv2) {
915 hasiv = hasiv2;
916 ivobj = ivobj2;
917 encname = encname2;
918 }
919 if (hasiv) w_byte(TYPE_IVAR, arg);
920 w_class(TYPE_USERDEF, obj, arg, FALSE);
921 w_bytes(RSTRING_PTR(v), RSTRING_LEN(v), arg);
922 if (hasiv) {
923 w_ivar(hasiv, ivobj, encname, &c_arg);
924 }
925 w_remember(obj, arg);
926 return;
927 }
928
929 w_remember(obj, arg);
930
931 hasiv = has_ivars(obj, (encname = encoding_name(obj, arg)), &ivobj);
932 {
933 st_data_t compat_data;
934 rb_alloc_func_t allocator = rb_get_alloc_func(RBASIC(obj)->klass);
935 if (st_lookup(compat_allocator_tbl,
936 (st_data_t)allocator,
937 &compat_data)) {
938 marshal_compat_t *compat = (marshal_compat_t*)compat_data;
939 VALUE real_obj = obj;
940 obj = compat->dumper(real_obj);
941 if (!arg->compat_tbl) {
942 arg->compat_tbl = rb_init_identtable();
943 }
944 st_insert(arg->compat_tbl, (st_data_t)obj, (st_data_t)real_obj);
945 if (obj != real_obj && UNDEF_P(ivobj)) hasiv = 0;
946 }
947 }
948 if (hasiv) w_byte(TYPE_IVAR, arg);
949
950 switch (BUILTIN_TYPE(obj)) {
951 case T_CLASS:
952 if (FL_TEST(obj, FL_SINGLETON)) {
953 rb_raise(rb_eTypeError, "singleton class can't be dumped");
954 }
955 w_byte(TYPE_CLASS, arg);
956 {
957 VALUE path = class2path(obj);
958 w_bytes(RSTRING_PTR(path), RSTRING_LEN(path), arg);
959 RB_GC_GUARD(path);
960 }
961 break;
962
963 case T_MODULE:
964 w_byte(TYPE_MODULE, arg);
965 {
966 VALUE path = class2path(obj);
967 w_bytes(RSTRING_PTR(path), RSTRING_LEN(path), arg);
968 RB_GC_GUARD(path);
969 }
970 break;
971
972 case T_FLOAT:
973 w_byte(TYPE_FLOAT, arg);
974 w_float(RFLOAT_VALUE(obj), arg);
975 break;
976
977 case T_BIGNUM:
978 w_byte(TYPE_BIGNUM, arg);
979 {
980 char sign = BIGNUM_SIGN(obj) ? '+' : '-';
981 size_t len = BIGNUM_LEN(obj);
982 size_t slen;
983 size_t j;
984 BDIGIT *d = BIGNUM_DIGITS(obj);
985
986 slen = SHORTLEN(len);
987 if (LONG_MAX < slen) {
988 rb_raise(rb_eTypeError, "too big Bignum can't be dumped");
989 }
990
991 w_byte(sign, arg);
992 w_long((long)slen, arg);
993 for (j = 0; j < len; j++) {
994 #if SIZEOF_BDIGIT > SIZEOF_SHORT
995 BDIGIT num = *d;
996 int i;
997
998 for (i=0; i<SIZEOF_BDIGIT; i+=SIZEOF_SHORT) {
999 w_short(num & SHORTMASK, arg);
1000 num = SHORTDN(num);
1001 if (j == len - 1 && num == 0) break;
1002 }
1003 #else
1004 w_short(*d, arg);
1005 #endif
1006 d++;
1007 }
1008 }
1009 break;
1010
1011 case T_STRING:
1012 w_uclass(obj, rb_cString, arg);
1013 w_byte(TYPE_STRING, arg);
1014 w_bytes(RSTRING_PTR(obj), RSTRING_LEN(obj), arg);
1015 break;
1016
1017 case T_REGEXP:
1018 w_uclass(obj, rb_cRegexp, arg);
1019 w_byte(TYPE_REGEXP, arg);
1020 {
1021 int opts = rb_reg_options(obj);
1022 w_bytes(RREGEXP_SRC_PTR(obj), RREGEXP_SRC_LEN(obj), arg);
1023 w_byte((char)opts, arg);
1024 }
1025 break;
1026
1027 case T_ARRAY:
1028 w_uclass(obj, rb_cArray, arg);
1029 w_byte(TYPE_ARRAY, arg);
1030 {
1031 long i, len = RARRAY_LEN(obj);
1032
1033 w_long(len, arg);
1034 for (i=0; i<RARRAY_LEN(obj); i++) {
1035 w_object(RARRAY_AREF(obj, i), arg, limit);
1036 if (len != RARRAY_LEN(obj)) {
1037 rb_raise(rb_eRuntimeError, "array modified during dump");
1038 }
1039 }
1040 }
1041 break;
1042
1043 case T_HASH:
1044 w_uclass(obj, rb_cHash, arg);
1045 if (rb_hash_compare_by_id_p(obj)) {
1046 w_byte(TYPE_UCLASS, arg);
1047 w_symbol(rb_sym_intern_ascii_cstr("Hash"), arg);
1048 }
1049 if (NIL_P(RHASH_IFNONE(obj))) {
1050 w_byte(TYPE_HASH, arg);
1051 }
1052 else if (FL_TEST(obj, RHASH_PROC_DEFAULT)) {
1053 rb_raise(rb_eTypeError, "can't dump hash with default proc");
1054 }
1055 else {
1056 w_byte(TYPE_HASH_DEF, arg);
1057 }
1058 w_long(rb_hash_size_num(obj), arg);
1059 rb_hash_foreach(obj, hash_each, (st_data_t)&c_arg);
1060 if (!NIL_P(RHASH_IFNONE(obj))) {
1061 w_object(RHASH_IFNONE(obj), arg, limit);
1062 }
1063 break;
1064
1065 case T_STRUCT:
1066 w_class(TYPE_STRUCT, obj, arg, TRUE);
1067 {
1068 long len = RSTRUCT_LEN(obj);
1069 VALUE mem;
1070 long i;
1071
1072 w_long(len, arg);
1073 mem = rb_struct_members(obj);
1074 for (i=0; i<len; i++) {
1075 w_symbol(RARRAY_AREF(mem, i), arg);
1076 w_object(RSTRUCT_GET(obj, i), arg, limit);
1077 }
1078 }
1079 break;
1080
1081 case T_OBJECT:
1082 w_class(TYPE_OBJECT, obj, arg, TRUE);
1083 w_objivar(obj, &c_arg);
1084 break;
1085
1086 case T_DATA:
1087 {
1088 VALUE v;
1089
1090 if (!rb_obj_respond_to(obj, s_dump_data, TRUE)) {
1091 rb_raise(rb_eTypeError,
1092 "no _dump_data is defined for class %"PRIsVALUE,
1093 rb_obj_class(obj));
1094 }
1095 v = dump_funcall(arg, obj, s_dump_data, 0, 0);
1096 w_class(TYPE_DATA, obj, arg, TRUE);
1097 w_object(v, arg, limit);
1098 }
1099 break;
1100
1101 default:
1102 rb_raise(rb_eTypeError, "can't dump %"PRIsVALUE,
1103 rb_obj_class(obj));
1104 break;
1105 }
1106 RB_GC_GUARD(obj);
1107 }
1108 if (hasiv) {
1109 w_ivar(hasiv, ivobj, encname, &c_arg);
1110 }
1111 }
1112
1113 static void
1114 clear_dump_arg(struct dump_arg *arg)
1115 {
1116 if (!arg->symbols) return;
1117 st_free_table(arg->symbols);
1118 arg->symbols = 0;
1119 st_free_table(arg->data);
1120 arg->data = 0;
1121 arg->num_entries = 0;
1122 if (arg->compat_tbl) {
1123 st_free_table(arg->compat_tbl);
1124 arg->compat_tbl = 0;
1125 }
1126 if (arg->encodings) {
1127 st_free_table(arg->encodings);
1128 arg->encodings = 0;
1129 }
1130 }
1131
1132 NORETURN(static inline void io_needed(void));
1133 static inline void
1134 io_needed(void)
1135 {
1136 rb_raise(rb_eTypeError, "instance of IO needed");
1137 }
1138
1139 /*
1140 * call-seq:
1141 * dump( obj [, anIO] , limit=-1 ) -> anIO
1142 *
1143 * Serializes obj and all descendant objects. If anIO is
1144 * specified, the serialized data will be written to it, otherwise the
1145 * data will be returned as a String. If limit is specified, the
1146 * traversal of subobjects will be limited to that depth. If limit is
1147 * negative, no checking of depth will be performed.
1148 *
1149 * class Klass
1150 * def initialize(str)
1151 * @str = str
1152 * end
1153 * def say_hello
1154 * @str
1155 * end
1156 * end
1157 *
1158 * (produces no output)
1159 *
1160 * o = Klass.new("hello\n")
1161 * data = Marshal.dump(o)
1162 * obj = Marshal.load(data)
1163 * obj.say_hello #=> "hello\n"
1164 *
1165 * Marshal can't dump following objects:
1166 * * anonymous Class/Module.
1167 * * objects which are related to system (ex: Dir, File::Stat, IO, File, Socket
1168 * and so on)
1169 * * an instance of MatchData, Data, Method, UnboundMethod, Proc, Thread,
1170 * ThreadGroup, Continuation
1171 * * objects which define singleton methods
1172 */
1173 static VALUE
1174 marshal_dump(int argc, VALUE *argv, VALUE _)
1175 {
1176 VALUE obj, port, a1, a2;
1177 int limit = -1;
1178
1179 port = Qnil;
1180 rb_scan_args(argc, argv, "12", &obj, &a1, &a2);
1181 if (argc == 3) {
1182 if (!NIL_P(a2)) limit = NUM2INT(a2);
1183 if (NIL_P(a1)) io_needed();
1184 port = a1;
1185 }
1186 else if (argc == 2) {
1187 if (FIXNUM_P(a1)) limit = FIX2INT(a1);
1188 else if (NIL_P(a1)) io_needed();
1189 else port = a1;
1190 }
1191 return rb_marshal_dump_limited(obj, port, limit);
1192 }
1193
1194 VALUE
1195 rb_marshal_dump_limited(VALUE obj, VALUE port, int limit)
1196 {
1197 struct dump_arg *arg;
1198 VALUE wrapper; /* used to avoid memory leak in case of exception */
1199
1200 wrapper = TypedData_Make_Struct(0, struct dump_arg, &dump_arg_data, arg);
1201 arg->dest = 0;
1202 arg->symbols = st_init_numtable();
1203 arg->data = rb_init_identtable();
1204 arg->num_entries = 0;
1205 arg->compat_tbl = 0;
1206 arg->encodings = 0;
1207 arg->str = rb_str_buf_new(0);
1208 if (!NIL_P(port)) {
1209 if (!rb_respond_to(port, s_write)) {
1210 io_needed();
1211 }
1212 arg->dest = port;
1213 dump_check_funcall(arg, port, s_binmode, 0, 0);
1214 }
1215 else {
1216 port = arg->str;
1217 }
1218
1219 w_byte(MARSHAL_MAJOR, arg);
1220 w_byte(MARSHAL_MINOR, arg);
1221
1222 w_object(obj, arg, limit);
1223 if (arg->dest) {
1224 rb_io_write(arg->dest, arg->str);
1225 rb_str_resize(arg->str, 0);
1226 }
1227 clear_dump_arg(arg);
1228 RB_GC_GUARD(wrapper);
1229
1230 return port;
1231 }
1232
1233 struct load_arg {
1234 VALUE src;
1235 char *buf;
1236 long buflen;
1237 long readable;
1238 long offset;
1239 st_table *symbols;
1240 st_table *data;
1241 st_table *partial_objects;
1242 VALUE proc;
1243 st_table *compat_tbl;
1244 bool freeze;
1245 };
1246
1247 static VALUE
1248 check_load_arg(VALUE ret, struct load_arg *arg, const char *name)
1249 {
1250 if (!arg->symbols) {
1251 rb_raise(rb_eRuntimeError, "Marshal.load reentered at %s",
1252 name);
1253 }
1254 return ret;
1255 }
1256 #define load_funcall(arg, obj, sym, argc, argv) \
1257 check_load_arg(rb_funcallv(obj, sym, argc, argv), arg, name_##sym)
1258
1259 static void clear_load_arg(struct load_arg *arg);
1260
1261 static void
1262 mark_load_arg(void *ptr)
1263 {
1264 struct load_arg *p = ptr;
1265 if (!p->symbols)
1266 return;
1267 rb_mark_tbl(p->symbols);
1268 rb_mark_tbl(p->data);
1269 rb_mark_tbl(p->partial_objects);
1270 rb_mark_hash(p->compat_tbl);
1271 }
1272
1273 static void
1274 free_load_arg(void *ptr)
1275 {
1276 clear_load_arg(ptr);
1277 }
1278
1279 static size_t
1280 memsize_load_arg(const void *ptr)
1281 {
1282 const struct load_arg *p = (struct load_arg *)ptr;
1283 size_t memsize = 0;
1284 if (p->symbols) memsize += rb_st_memsize(p->symbols);
1285 if (p->data) memsize += rb_st_memsize(p->data);
1286 if (p->partial_objects) memsize += rb_st_memsize(p->partial_objects);
1287 if (p->compat_tbl) memsize += rb_st_memsize(p->compat_tbl);
1288 return memsize;
1289 }
1290
1291 static const rb_data_type_t load_arg_data = {
1292 "load_arg",
1293 {mark_load_arg, free_load_arg, memsize_load_arg,},
1294 0, 0, RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_EMBEDDABLE
1295 };
1296
1297 #define r_entry(v, arg) r_entry0((v), (arg)->data->num_entries, (arg))
1298 static VALUE r_object(struct load_arg *arg);
1299 static VALUE r_symbol(struct load_arg *arg);
1300
1301 NORETURN(static void too_short(void));
1302 static void
1303 too_short(void)
1304 {
1305 rb_raise(rb_eArgError, "marshal data too short");
1306 }
1307
1308 static st_index_t
1309 r_prepare(struct load_arg *arg)
1310 {
1311 st_index_t idx = arg->data->num_entries;
1312
1313 st_insert(arg->data, (st_data_t)idx, (st_data_t)Qundef);
1314 return idx;
1315 }
1316
1317 static unsigned char
1318 r_byte1_buffered(struct load_arg *arg)
1319 {
1320 if (arg->buflen == 0) {
1321 long readable = arg->readable < BUFSIZ ? arg->readable : BUFSIZ;
1322 VALUE str, n = LONG2NUM(readable);
1323
1324 str = load_funcall(arg, arg->src, s_read, 1, &n);
1325 if (NIL_P(str)) too_short();
1326 StringValue(str);
1327 memcpy(arg->buf, RSTRING_PTR(str), RSTRING_LEN(str));
1328 arg->offset = 0;
1329 arg->buflen = RSTRING_LEN(str);
1330 }
1331 arg->buflen--;
1332 return arg->buf[arg->offset++];
1333 }
1334
1335 static int
1336 r_byte(struct load_arg *arg)
1337 {
1338 int c;
1339
1340 if (RB_TYPE_P(arg->src, T_STRING)) {
1341 if (RSTRING_LEN(arg->src) > arg->offset) {
1342 c = (unsigned char)RSTRING_PTR(arg->src)[arg->offset++];
1343 }
1344 else {
1345 too_short();
1346 }
1347 }
1348 else {
1349 if (arg->readable >0 || arg->buflen > 0) {
1350 c = r_byte1_buffered(arg);
1351 }
1352 else {
1353 VALUE v = load_funcall(arg, arg->src, s_getbyte, 0, 0);
1354 if (NIL_P(v)) rb_eof_error();
1355 c = (unsigned char)NUM2CHR(v);
1356 }
1357 }
1358 return c;
1359 }
1360
1361 NORETURN(static void long_toobig(int size));
1362
1363 static void
1364 long_toobig(int size)
1365 {
1366 rb_raise(rb_eTypeError, "long too big for this architecture (size "
1367 STRINGIZE(SIZEOF_LONG)", given %d)", size);
1368 }
1369
1370 static long
1371 r_long(struct load_arg *arg)
1372 {
1373 register long x;
1374 int c = (signed char)r_byte(arg);
1375 long i;
1376
1377 if (c == 0) return 0;
1378 if (c > 0) {
1379 if (4 < c && c < 128) {
1380 return c - 5;
1381 }
1382 if (c > (int)sizeof(long)) long_toobig(c);
1383 x = 0;
1384 for (i=0;i<c;i++) {
1385 x |= (long)r_byte(arg) << (8*i);
1386 }
1387 }
1388 else {
1389 if (-129 < c && c < -4) {
1390 return c + 5;
1391 }
1392 c = -c;
1393 if (c > (int)sizeof(long)) long_toobig(c);
1394 x = -1;
1395 for (i=0;i<c;i++) {
1396 x &= ~((long)0xff << (8*i));
1397 x |= (long)r_byte(arg) << (8*i);
1398 }
1399 }
1400 return x;
1401 }
1402
1403 long
1404 ruby_marshal_read_long(const char **buf, long len)
1405 {
1406 long x;
1407 struct RString src;
1408 struct load_arg arg;
1409 memset(&arg, 0, sizeof(arg));
1410 arg.src = rb_setup_fake_str(&src, *buf, len, 0);
1411 x = r_long(&arg);
1412 *buf += arg.offset;
1413 return x;
1414 }
1415
1416 static VALUE
1417 r_bytes1(long len, struct load_arg *arg)
1418 {
1419 VALUE str, n = LONG2NUM(len);
1420
1421 str = load_funcall(arg, arg->src, s_read, 1, &n);
1422 if (NIL_P(str)) too_short();
1423 StringValue(str);
1424 if (RSTRING_LEN(str) != len) too_short();
1425
1426 return str;
1427 }
1428
1429 static VALUE
1430 r_bytes1_buffered(long len, struct load_arg *arg)
1431 {
1432 VALUE str;
1433
1434 if (len <= arg->buflen) {
1435 str = rb_str_new(arg->buf+arg->offset, len);
1436 arg->offset += len;
1437 arg->buflen -= len;
1438 }
1439 else {
1440 long buflen = arg->buflen;
1441 long readable = arg->readable + 1;
1442 long tmp_len, read_len, need_len = len - buflen;
1443 VALUE tmp, n;
1444
1445 readable = readable < BUFSIZ ? readable : BUFSIZ;
1446 read_len = need_len > readable ? need_len : readable;
1447 n = LONG2NUM(read_len);
1448 tmp = load_funcall(arg, arg->src, s_read, 1, &n);
1449 if (NIL_P(tmp)) too_short();
1450 StringValue(tmp);
1451
1452 tmp_len = RSTRING_LEN(tmp);
1453
1454 if (tmp_len < need_len) too_short();
1455
1456 str = rb_str_new(arg->buf+arg->offset, buflen);
1457 rb_str_cat(str, RSTRING_PTR(tmp), need_len);
1458
1459 if (tmp_len > need_len) {
1460 buflen = tmp_len - need_len;
1461 memcpy(arg->buf, RSTRING_PTR(tmp)+need_len, buflen);
1462 arg->buflen = buflen;
1463 }
1464 else {
1465 arg->buflen = 0;
1466 }
1467 arg->offset = 0;
1468 }
1469
1470 return str;
1471 }
1472
1473 #define r_bytes(arg) r_bytes0(r_long(arg), (arg))
1474
1475 static VALUE
1476 r_bytes0(long len, struct load_arg *arg)
1477 {
1478 VALUE str;
1479
1480 if (len == 0) return rb_str_new(0, 0);
1481 if (RB_TYPE_P(arg->src, T_STRING)) {
1482 if (RSTRING_LEN(arg->src) - arg->offset >= len) {
1483 str = rb_str_new(RSTRING_PTR(arg->src)+arg->offset, len);
1484 arg->offset += len;
1485 }
1486 else {
1487 too_short();
1488 }
1489 }
1490 else {
1491 if (arg->readable > 0 || arg->buflen > 0) {
1492 str = r_bytes1_buffered(len, arg);
1493 }
1494 else {
1495 str = r_bytes1(len, arg);
1496 }
1497 }
1498 return str;
1499 }
1500
1501 static inline int
1502 name_equal(const char *name, size_t nlen, const char *p, long l)
1503 {
1504 if ((size_t)l != nlen || *p != *name) return 0;
1505 return nlen == 1 || memcmp(p+1, name+1, nlen-1) == 0;
1506 }
1507
1508 static int
1509 sym2encidx(VALUE sym, VALUE val)
1510 {
1511 static const char name_encoding[8] = "encoding";
1512 const char *p;
1513 long l;
1514 if (rb_enc_get_index(sym) != ENCINDEX_US_ASCII) return -1;
1515 RSTRING_GETMEM(sym, p, l);
1516 if (l <= 0) return -1;
1517 if (name_equal(name_encoding, sizeof(name_encoding), p, l)) {
1518 int idx = rb_enc_find_index(StringValueCStr(val));
1519 return idx;
1520 }
1521 if (name_equal(name_s_encoding_short, rb_strlen_lit(name_s_encoding_short), p, l)) {
1522 if (val == Qfalse) return rb_usascii_encindex();
1523 else if (val == Qtrue) return rb_utf8_encindex();
1524 /* bogus ignore */
1525 }
1526 return -1;
1527 }
1528
1529 static int
1530 symname_equal(VALUE sym, const char *name, size_t nlen)
1531 {
1532 const char *p;
1533 long l;
1534 if (rb_enc_get_index(sym) != ENCINDEX_US_ASCII) return 0;
1535 RSTRING_GETMEM(sym, p, l);
1536 return name_equal(name, nlen, p, l);
1537 }
1538
1539 #define BUILD_ASSERT_POSITIVE(n) \
1540 /* make 0 negative to workaround the "zero size array" GCC extension, */ \
1541 ((sizeof(char [2*(ssize_t)(n)-1])+1)/2) /* assuming no overflow */
1542 #define symname_equal_lit(sym, sym_name) \
1543 symname_equal(sym, sym_name, BUILD_ASSERT_POSITIVE(rb_strlen_lit(sym_name)))
1544
1545 static VALUE
1546 r_symlink(struct load_arg *arg)
1547 {
1548 st_data_t sym;
1549 long num = r_long(arg);
1550
1551 if (!st_lookup(arg->symbols, num, &sym)) {
1552 rb_raise(rb_eArgError, "bad symbol");
1553 }
1554 return (VALUE)sym;
1555 }
1556
1557 static VALUE
1558 r_symreal(struct load_arg *arg, int ivar)
1559 {
1560 VALUE s = r_bytes(arg);
1561 VALUE sym;
1562 int idx = -1;
1563 st_index_t n = arg->symbols->num_entries;
1564
1565 if (rb_enc_str_asciionly_p(s)) rb_enc_associate_index(s, ENCINDEX_US_ASCII);
1566 st_insert(arg->symbols, (st_data_t)n, (st_data_t)s);
1567 if (ivar) {
1568 long num = r_long(arg);
1569 while (num-- > 0) {
1570 sym = r_symbol(arg);
1571 idx = sym2encidx(sym, r_object(arg));
1572 }
1573 }
1574 if (idx > 0) {
1575 rb_enc_associate_index(s, idx);
1576 if (is_broken_string(s)) {
1577 rb_raise(rb_eArgError, "invalid byte sequence in %s: %+"PRIsVALUE,
1578 rb_enc_name(rb_enc_from_index(idx)), s);
1579 }
1580 }
1581
1582 return s;
1583 }
1584
1585 static VALUE
1586 r_symbol(struct load_arg *arg)
1587 {
1588 int type, ivar = 0;
1589
1590 again:
1591 switch ((type = r_byte(arg))) {
1592 default:
1593 rb_raise(rb_eArgError, "dump format error for symbol(0x%x)", type);
1594 case TYPE_IVAR:
1595 ivar = 1;
1596 goto again;
1597 case TYPE_SYMBOL:
1598 return r_symreal(arg, ivar);
1599 case TYPE_SYMLINK:
1600 if (ivar) {
1601 rb_raise(rb_eArgError, "dump format error (symlink with encoding)");
1602 }
1603 return r_symlink(arg);
1604 }
1605 }
1606
1607 static VALUE
1608 r_unique(struct load_arg *arg)
1609 {
1610 return r_symbol(arg);
1611 }
1612
1613 static VALUE
1614 r_string(struct load_arg *arg)
1615 {
1616 return r_bytes(arg);
1617 }
1618
1619 static VALUE
1620 r_entry0(VALUE v, st_index_t num, struct load_arg *arg)
1621 {
1622 st_data_t real_obj = (st_data_t)v;
1623 if (arg->compat_tbl) {
1624 /* real_obj is kept if not found */
1625 st_lookup(arg->compat_tbl, v, &real_obj);
1626 }
1627 st_insert(arg->data, num, real_obj);
1628 st_insert(arg->partial_objects, (st_data_t)real_obj, Qtrue);
1629 return v;
1630 }
1631
1632 static VALUE
1633 r_fixup_compat(VALUE v, struct load_arg *arg)
1634 {
1635 st_data_t data;
1636 st_data_t key = (st_data_t)v;
1637 if (arg->compat_tbl && st_delete(arg->compat_tbl, &key, &data)) {
1638 VALUE real_obj = (VALUE)data;
1639 rb_alloc_func_t allocator = rb_get_alloc_func(CLASS_OF(real_obj));
1640 if (st_lookup(compat_allocator_tbl, (st_data_t)allocator, &data)) {
1641 marshal_compat_t *compat = (marshal_compat_t*)data;
1642 compat->loader(real_obj, v);
1643 }
1644 v = real_obj;
1645 }
1646 return v;
1647 }
1648
1649 static VALUE
1650 r_post_proc(VALUE v, struct load_arg *arg)
1651 {
1652 if (arg->proc) {
1653 v = load_funcall(arg, arg->proc, s_call, 1, &v);
1654 }
1655 return v;
1656 }
1657
1658 static VALUE
1659 r_leave(VALUE v, struct load_arg *arg, bool partial)
1660 {
1661 v = r_fixup_compat(v, arg);
1662 if (!partial) {
1663 st_data_t data;
1664 st_data_t key = (st_data_t)v;
1665 st_delete(arg->partial_objects, &key, &data);
1666 if (arg->freeze) {
1667 if (RB_TYPE_P(v, T_MODULE) || RB_TYPE_P(v, T_CLASS)) {
1668 // noop
1669 }
1670 else if (RB_TYPE_P(v, T_STRING)) {
1671 v = rb_str_to_interned_str(v);
1672 }
1673 else {
1674 OBJ_FREEZE(v);
1675 }
1676 }
1677 v = r_post_proc(v, arg);
1678 }
1679 return v;
1680 }
1681
1682 static int
1683 copy_ivar_i(ID vid, VALUE value, st_data_t arg)
1684 {
1685 VALUE obj = (VALUE)arg;
1686
1687 if (!rb_ivar_defined(obj, vid))
1688 rb_ivar_set(obj, vid, value);
1689 return ST_CONTINUE;
1690 }
1691
1692 static VALUE
1693 r_copy_ivar(VALUE v, VALUE data)
1694 {
1695 rb_ivar_foreach(data, copy_ivar_i, (st_data_t)v);
1696 return v;
1697 }
1698
1699 #define override_ivar_error(type, str) \
1700 rb_raise(rb_eTypeError, \
1701 "can't override instance variable of "type" '%"PRIsVALUE"'", \
1702 (str))
1703
1704 static void
1705 r_ivar(VALUE obj, int *has_encoding, struct load_arg *arg)
1706 {
1707 long len;
1708
1709 len = r_long(arg);
1710 if (len > 0) {
1711 if (RB_TYPE_P(obj, T_MODULE)) {
1712 override_ivar_error("module", rb_mod_name(obj));
1713 }
1714 else if (RB_TYPE_P(obj, T_CLASS)) {
1715 override_ivar_error("class", rb_class_name(obj));
1716 }
1717 do {
1718 VALUE sym = r_symbol(arg);
1719 VALUE val = r_object(arg);
1720 int idx = sym2encidx(sym, val);
1721 if (idx >= 0) {
1722 if (rb_enc_capable(obj)) {
1723 rb_enc_associate_index(obj, idx);
1724 }
1725 else {
1726 rb_raise(rb_eArgError, "%"PRIsVALUE" is not enc_capable", obj);
1727 }
1728 if (has_encoding) *has_encoding = TRUE;
1729 }
1730 else if (symname_equal_lit(sym, name_s_ruby2_keywords_flag)) {
1731 if (RB_TYPE_P(obj, T_HASH)) {
1732 rb_hash_ruby2_keywords(obj);
1733 }
1734 else {
1735 rb_raise(rb_eArgError, "ruby2_keywords flag is given but %"PRIsVALUE" is not a Hash", obj);
1736 }
1737 }
1738 else {
1739 rb_ivar_set(obj, rb_intern_str(sym), val);
1740 }
1741 } while (--len > 0);
1742 }
1743 }
1744
1745 static VALUE
1746 path2class(VALUE path)
1747 {
1748 VALUE v = rb_path_to_class(path);
1749
1750 if (!RB_TYPE_P(v, T_CLASS)) {
1751 rb_raise(rb_eArgError, "%"PRIsVALUE" does not refer to class", path);
1752 }
1753 return v;
1754 }
1755
1756 #define path2module(path) must_be_module(rb_path_to_class(path), path)
1757
1758 static VALUE
1759 must_be_module(VALUE v, VALUE path)
1760 {
1761 if (!RB_TYPE_P(v, T_MODULE)) {
1762 rb_raise(rb_eArgError, "%"PRIsVALUE" does not refer to module", path);
1763 }
1764 return v;
1765 }
1766
1767 static VALUE
1768 obj_alloc_by_klass(VALUE klass, struct load_arg *arg, VALUE *oldclass)
1769 {
1770 st_data_t data;
1771 rb_alloc_func_t allocator;
1772
1773 allocator = rb_get_alloc_func(klass);
1774 if (st_lookup(compat_allocator_tbl, (st_data_t)allocator, &data)) {
1775 marshal_compat_t *compat = (marshal_compat_t*)data;
1776 VALUE real_obj = rb_obj_alloc(klass);
1777 VALUE obj = rb_obj_alloc(compat->oldclass);
1778 if (oldclass) *oldclass = compat->oldclass;
1779
1780 if (!arg->compat_tbl) {
1781 arg->compat_tbl = rb_init_identtable();
1782 }
1783 st_insert(arg->compat_tbl, (st_data_t)obj, (st_data_t)real_obj);
1784 return obj;
1785 }
1786
1787 return rb_obj_alloc(klass);
1788 }
1789
1790 static VALUE
1791 obj_alloc_by_path(VALUE path, struct load_arg *arg)
1792 {
1793 return obj_alloc_by_klass(path2class(path), arg, 0);
1794 }
1795
1796 static VALUE
1797 append_extmod(VALUE obj, VALUE extmod)
1798 {
1799 long i = RARRAY_LEN(extmod);
1800 while (i > 0) {
1801 VALUE m = RARRAY_AREF(extmod, --i);
1802 rb_extend_object(obj, m);
1803 }
1804 return obj;
1805 }
1806
1807 #define prohibit_ivar(type, str) do { \
1808 if (!ivp || !*ivp) break; \
1809 override_ivar_error(type, str); \
1810 } while (0)
1811
1812 static VALUE r_object_for(struct load_arg *arg, bool partial, int *ivp, VALUE extmod, int type);
1813
1814 static VALUE
1815 r_object0(struct load_arg *arg, bool partial, int *ivp, VALUE extmod)
1816 {
1817 int type = r_byte(arg);
1818 return r_object_for(arg, partial, ivp, extmod, type);
1819 }
1820
1821 static VALUE
1822 r_object_for(struct load_arg *arg, bool partial, int *ivp, VALUE extmod, int type)
1823 {
1824 VALUE (*hash_new_with_size)(st_index_t) = rb_hash_new_with_size;
1825 VALUE v = Qnil;
1826 long id;
1827 st_data_t link;
1828
1829 switch (type) {
1830 case TYPE_LINK:
1831 id = r_long(arg);
1832 if (!st_lookup(arg->data, (st_data_t)id, &link)) {
1833 rb_raise(rb_eArgError, "dump format error (unlinked)");
1834 }
1835 v = (VALUE)link;
1836 if (!st_lookup(arg->partial_objects, (st_data_t)v, &link)) {
1837 v = r_post_proc(v, arg);
1838 }
1839 break;
1840
1841 case TYPE_IVAR:
1842 {
1843 int ivar = TRUE;
1844 v = r_object0(arg, true, &ivar, extmod);
1845 if (ivar) r_ivar(v, NULL, arg);
1846 v = r_leave(v, arg, partial);
1847 }
1848 break;
1849
1850 case TYPE_EXTENDED:
1851 {
1852 VALUE path = r_unique(arg);
1853 VALUE m = rb_path_to_class(path);
1854 if (NIL_P(extmod)) extmod = rb_ary_hidden_new(0);
1855
1856 if (RB_TYPE_P(m, T_CLASS)) { /* prepended */
1857 VALUE c;
1858
1859 v = r_object0(arg, true, 0, Qnil);
1860 c = CLASS_OF(v);
1861 if (c != m || FL_TEST(c, FL_SINGLETON)) {
1862 rb_raise(rb_eArgError,
1863 "prepended class %"PRIsVALUE" differs from class %"PRIsVALUE,
1864 path, rb_class_name(c));
1865 }
1866 c = rb_singleton_class(v);
1867 while (RARRAY_LEN(extmod) > 0) {
1868 m = rb_ary_pop(extmod);
1869 rb_prepend_module(c, m);
1870 }
1871 }
1872 else {
1873 must_be_module(m, path);
1874 rb_ary_push(extmod, m);
1875
1876 v = r_object0(arg, true, 0, extmod);
1877 while (RARRAY_LEN(extmod) > 0) {
1878 m = rb_ary_pop(extmod);
1879 rb_extend_object(v, m);
1880 }
1881 }
1882 v = r_leave(v, arg, partial);
1883 }
1884 break;
1885
1886 case TYPE_UCLASS:
1887 {
1888 VALUE c = path2class(r_unique(arg));
1889
1890 if (FL_TEST(c, FL_SINGLETON)) {
1891 rb_raise(rb_eTypeError, "singleton can't be loaded");
1892 }
1893 type = r_byte(arg);
1894 if ((c == rb_cHash) &&
1895 /* Hack for compare_by_identify */
1896 (type == TYPE_HASH || type == TYPE_HASH_DEF)) {
1897 hash_new_with_size = rb_ident_hash_new_with_size;
1898 goto type_hash;
1899 }
1900 v = r_object_for(arg, partial, 0, extmod, type);
1901 if (RB_SPECIAL_CONST_P(v) || RB_TYPE_P(v, T_OBJECT) || RB_TYPE_P(v, T_CLASS)) {
1902 goto format_error;
1903 }
1904 if (RB_TYPE_P(v, T_MODULE) || !RTEST(rb_class_inherited_p(c, RBASIC(v)->klass))) {
1905 VALUE tmp = rb_obj_alloc(c);
1906
1907 if (TYPE(v) != TYPE(tmp)) goto format_error;
1908 }
1909 RBASIC_SET_CLASS(v, c);
1910 }
1911 break;
1912
1913 format_error:
1914 rb_raise(rb_eArgError, "dump format error (user class)");
1915
1916 case TYPE_NIL:
1917 v = Qnil;
1918 v = r_leave(v, arg, false);
1919 break;
1920
1921 case TYPE_TRUE:
1922 v = Qtrue;
1923 v = r_leave(v, arg, false);
1924 break;
1925
1926 case TYPE_FALSE:
1927 v = Qfalse;
1928 v = r_leave(v, arg, false);
1929 break;
1930
1931 case TYPE_FIXNUM:
1932 {
1933 long i = r_long(arg);
1934 v = LONG2FIX(i);
1935 }
1936 v = r_leave(v, arg, false);
1937 break;
1938
1939 case TYPE_FLOAT:
1940 {
1941 double d;
1942 VALUE str = r_bytes(arg);
1943 const char *ptr = RSTRING_PTR(str);
1944
1945 if (strcmp(ptr, "nan") == 0) {
1946 d = nan("");
1947 }
1948 else if (strcmp(ptr, "inf") == 0) {
1949 d = HUGE_VAL;
1950 }
1951 else if (strcmp(ptr, "-inf") == 0) {
1952 d = -HUGE_VAL;
1953 }
1954 else {
1955 char *e;
1956 d = strtod(ptr, &e);
1957 d = load_mantissa(d, e, RSTRING_LEN(str) - (e - ptr));
1958 }
1959 v = DBL2NUM(d);
1960 v = r_entry(v, arg);
1961 v = r_leave(v, arg, false);
1962 }
1963 break;
1964
1965 case TYPE_BIGNUM:
1966 {
1967 long len;
1968 VALUE data;
1969 int sign;
1970
1971 sign = r_byte(arg);
1972 len = r_long(arg);
1973
1974 if (SIZEOF_VALUE >= 8 && len <= 4) {
1975 // Representable within uintptr, likely FIXNUM
1976 VALUE num = 0;
1977 for (int i = 0; i < len; i++) {
1978 num |= (VALUE)r_byte(arg) << (i * 16);
1979 num |= (VALUE)r_byte(arg) << (i * 16 + 8);
1980 }
1981 #if SIZEOF_VALUE == SIZEOF_LONG
1982 v = ULONG2NUM(num);
1983 #else
1984 v = ULL2NUM(num);
1985 #endif
1986 if (sign == '-') {
1987 v = rb_int_uminus(v);
1988 }
1989 }
1990 else {
1991 data = r_bytes0(len * 2, arg);
1992 v = rb_integer_unpack(RSTRING_PTR(data), len, 2, 0,
1993 INTEGER_PACK_LITTLE_ENDIAN | (sign == '-' ? INTEGER_PACK_NEGATIVE : 0));
1994 rb_str_resize(data, 0L);
1995 }
1996 v = r_entry(v, arg);
1997 v = r_leave(v, arg, false);
1998 }
1999 break;
2000
2001 case TYPE_STRING:
2002 v = r_entry(r_string(arg), arg);
2003 v = r_leave(v, arg, partial);
2004 break;
2005
2006 case TYPE_REGEXP:
2007 {
2008 VALUE str = r_bytes(arg);
2009 int options = r_byte(arg);
2010 int has_encoding = FALSE;
2011 st_index_t idx = r_prepare(arg);
2012
2013 if (ivp) {
2014 r_ivar(str, &has_encoding, arg);
2015 *ivp = FALSE;
2016 }
2017 if (!has_encoding) {
2018 /* 1.8 compatibility; remove escapes undefined in 1.8 */
2019 char *ptr = RSTRING_PTR(str), *dst = ptr, *src = ptr;
2020 long len = RSTRING_LEN(str);
2021 long bs = 0;
2022 for (; len-- > 0; *dst++ = *src++) {
2023 switch (*src) {
2024 case '\\': bs++; break;
2025 case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
2026 case 'm': case 'o': case 'p': case 'q': case 'u': case 'y':
2027 case 'E': case 'F': case 'H': case 'I': case 'J': case 'K':
2028 case 'L': case 'N': case 'O': case 'P': case 'Q': case 'R':
2029 case 'S': case 'T': case 'U': case 'V': case 'X': case 'Y':
2030 if (bs & 1) --dst;
2031 /* fall through */
2032 default: bs = 0; break;
2033 }
2034 }
2035 rb_str_set_len(str, dst - ptr);
2036 }
2037 VALUE regexp = rb_reg_new_str(str, options);
2038 r_copy_ivar(regexp, str);
2039
2040 v = r_entry0(regexp, idx, arg);
2041 v = r_leave(v, arg, partial);
2042 }
2043 break;
2044
2045 case TYPE_ARRAY:
2046 {
2047 long len = r_long(arg);
2048
2049 v = rb_ary_new2(len);
2050 v = r_entry(v, arg);
2051 arg->readable += len - 1;
2052 while (len--) {
2053 rb_ary_push(v, r_object(arg));
2054 arg->readable--;
2055 }
2056 v = r_leave(v, arg, partial);
2057 arg->readable++;
2058 }
2059 break;
2060
2061 case TYPE_HASH:
2062 case TYPE_HASH_DEF:
2063 type_hash:
2064 {
2065 long len = r_long(arg);
2066
2067 v = hash_new_with_size(len);
2068 v = r_entry(v, arg);
2069 arg->readable += (len - 1) * 2;
2070 while (len--) {
2071 VALUE key = r_object(arg);
2072 VALUE value = r_object(arg);
2073 rb_hash_aset(v, key, value);
2074 arg->readable -= 2;
2075 }
2076 arg->readable += 2;
2077 if (type == TYPE_HASH_DEF) {
2078 RHASH_SET_IFNONE(v, r_object(arg));
2079 }
2080 v = r_leave(v, arg, partial);
2081 }
2082 break;
2083
2084 case TYPE_STRUCT:
2085 {
2086 VALUE mem, values;
2087 long i;
2088 VALUE slot;
2089 st_index_t idx = r_prepare(arg);
2090 VALUE klass = path2class(r_unique(arg));
2091 long len = r_long(arg);
2092
2093 v = rb_obj_alloc(klass);
2094 if (!RB_TYPE_P(v, T_STRUCT)) {
2095 rb_raise(rb_eTypeError, "class %"PRIsVALUE" not a struct", rb_class_name(klass));
2096 }
2097 mem = rb_struct_s_members(klass);
2098 if (RARRAY_LEN(mem) != len) {
2099 rb_raise(rb_eTypeError, "struct %"PRIsVALUE" not compatible (struct size differs)",
2100 rb_class_name(klass));
2101 }
2102
2103 arg->readable += (len - 1) * 2;
2104 v = r_entry0(v, idx, arg);
2105 values = rb_ary_new2(len);
2106 {
2107 VALUE keywords = Qfalse;
2108 if (RTEST(rb_struct_s_keyword_init(klass))) {
2109 keywords = rb_hash_new();
2110 rb_ary_push(values, keywords);
2111 }
2112
2113 for (i=0; i<len; i++) {
2114 VALUE n = rb_sym2str(RARRAY_AREF(mem, i));
2115 slot = r_symbol(arg);
2116
2117 if (!rb_str_equal(n, slot)) {
2118 rb_raise(rb_eTypeError, "struct %"PRIsVALUE" not compatible (:%"PRIsVALUE" for :%"PRIsVALUE")",
2119 rb_class_name(klass),
2120 slot, n);
2121 }
2122 if (keywords) {
2123 rb_hash_aset(keywords, RARRAY_AREF(mem, i), r_object(arg));
2124 }
2125 else {
2126 rb_ary_push(values, r_object(arg));
2127 }
2128 arg->readable -= 2;
2129 }
2130 }
2131 rb_struct_initialize(v, values);
2132 v = r_leave(v, arg, partial);
2133 arg->readable += 2;
2134 }
2135 break;
2136
2137 case TYPE_USERDEF:
2138 {
2139 VALUE name = r_unique(arg);
2140 VALUE klass = path2class(name);
2141 VALUE data;
2142 st_data_t d;
2143
2144 if (!rb_obj_respond_to(klass, s_load, TRUE)) {
2145 rb_raise(rb_eTypeError, "class %"PRIsVALUE" needs to have method '_load'",
2146 name);
2147 }
2148 data = r_string(arg);
2149 if (ivp) {
2150 r_ivar(data, NULL, arg);
2151 *ivp = FALSE;
2152 }
2153 v = load_funcall(arg, klass, s_load, 1, &data);
2154 v = r_entry(v, arg);
2155 if (st_lookup(compat_allocator_tbl, (st_data_t)rb_get_alloc_func(klass), &d)) {
2156 marshal_compat_t *compat = (marshal_compat_t*)d;
2157 v = compat->loader(klass, v);
2158 }
2159 if (!partial) {
2160 if (arg->freeze) {
2161 OBJ_FREEZE(v);
2162 }
2163 v = r_post_proc(v, arg);
2164 }
2165 }
2166 break;
2167
2168 case TYPE_USRMARSHAL:
2169 {
2170 VALUE name = r_unique(arg);
2171 VALUE klass = path2class(name);
2172 VALUE oldclass = 0;
2173 VALUE data;
2174
2175 v = obj_alloc_by_klass(klass, arg, &oldclass);
2176 if (!NIL_P(extmod)) {
2177 /* for the case marshal_load is overridden */
2178 append_extmod(v, extmod);
2179 }
2180 if (!rb_obj_respond_to(v, s_mload, TRUE)) {
2181 rb_raise(rb_eTypeError, "instance of %"PRIsVALUE" needs to have method 'marshal_load'",
2182 name);
2183 }
2184 v = r_entry(v, arg);
2185 data = r_object(arg);
2186 load_funcall(arg, v, s_mload, 1, &data);
2187 v = r_fixup_compat(v, arg);
2188 v = r_copy_ivar(v, data);
2189 if (arg->freeze) {
2190 OBJ_FREEZE(v);
2191 }
2192 v = r_post_proc(v, arg);
2193 if (!NIL_P(extmod)) {
2194 if (oldclass) append_extmod(v, extmod);
2195 rb_ary_clear(extmod);
2196 }
2197 }
2198 break;
2199
2200 case TYPE_OBJECT:
2201 {
2202 st_index_t idx = r_prepare(arg);
2203 v = obj_alloc_by_path(r_unique(arg), arg);
2204 if (!RB_TYPE_P(v, T_OBJECT)) {
2205 rb_raise(rb_eArgError, "dump format error");
2206 }
2207 v = r_entry0(v, idx, arg);
2208 r_ivar(v, NULL, arg);
2209 v = r_leave(v, arg, partial);
2210 }
2211 break;
2212
2213 case TYPE_DATA:
2214 {
2215 VALUE name = r_unique(arg);
2216 VALUE klass = path2class(name);
2217 VALUE oldclass = 0;
2218 VALUE r;
2219
2220 v = obj_alloc_by_klass(klass, arg, &oldclass);
2221 if (!RB_TYPE_P(v, T_DATA)) {
2222 rb_raise(rb_eArgError, "dump format error");
2223 }
2224 v = r_entry(v, arg);
2225 if (!rb_obj_respond_to(v, s_load_data, TRUE)) {
2226 rb_raise(rb_eTypeError,
2227 "class %"PRIsVALUE" needs to have instance method '_load_data'",
2228 name);
2229 }
2230 r = r_object0(arg, partial, 0, extmod);
2231 load_funcall(arg, v, s_load_data, 1, &r);
2232 v = r_leave(v, arg, partial);
2233 }
2234 break;
2235
2236 case TYPE_MODULE_OLD:
2237 {
2238 VALUE str = r_bytes(arg);
2239
2240 v = rb_path_to_class(str);
2241 prohibit_ivar("class/module", str);
2242 v = r_entry(v, arg);
2243 v = r_leave(v, arg, partial);
2244 }
2245 break;
2246
2247 case TYPE_CLASS:
2248 {
2249 VALUE str = r_bytes(arg);
2250
2251 v = path2class(str);
2252 prohibit_ivar("class", str);
2253 v = r_entry(v, arg);
2254 v = r_leave(v, arg, partial);
2255 }
2256 break;
2257
2258 case TYPE_MODULE:
2259 {
2260 VALUE str = r_bytes(arg);
2261
2262 v = path2module(str);
2263 prohibit_ivar("module", str);
2264 v = r_entry(v, arg);
2265 v = r_leave(v, arg, partial);
2266 }
2267 break;
2268
2269 case TYPE_SYMBOL:
2270 if (ivp) {
2271 v = r_symreal(arg, *ivp);
2272 *ivp = FALSE;
2273 }
2274 else {
2275 v = r_symreal(arg, 0);
2276 }
2277 v = rb_str_intern(v);
2278 v = r_leave(v, arg, partial);
2279 break;
2280
2281 case TYPE_SYMLINK:
2282 v = rb_str_intern(r_symlink(arg));
2283 break;
2284
2285 default:
2286 rb_raise(rb_eArgError, "dump format error(0x%x)", type);
2287 break;
2288 }
2289
2290 if (UNDEF_P(v)) {
2291 rb_raise(rb_eArgError, "dump format error (bad link)");
2292 }
2293
2294 return v;
2295 }
2296
2297 static VALUE
2298 r_object(struct load_arg *arg)
2299 {
2300 return r_object0(arg, false, 0, Qnil);
2301 }
2302
2303 static void
2304 clear_load_arg(struct load_arg *arg)
2305 {
2306 xfree(arg->buf);
2307 arg->buf = NULL;
2308 arg->buflen = 0;
2309 arg->offset = 0;
2310 arg->readable = 0;
2311 if (!arg->symbols) return;
2312 st_free_table(arg->symbols);
2313 arg->symbols = 0;
2314 st_free_table(arg->data);
2315 arg->data = 0;
2316 st_free_table(arg->partial_objects);
2317 arg->partial_objects = 0;
2318 if (arg->compat_tbl) {
2319 st_free_table(arg->compat_tbl);
2320 arg->compat_tbl = 0;
2321 }
2322 }
2323
2324 VALUE
2325 rb_marshal_load_with_proc(VALUE port, VALUE proc, bool freeze)
2326 {
2327 int major, minor;
2328 VALUE v;
2329 VALUE wrapper; /* used to avoid memory leak in case of exception */
2330 struct load_arg *arg;
2331
2332 v = rb_check_string_type(port);
2333 if (!NIL_P(v)) {
2334 port = v;
2335 }
2336 else if (rb_respond_to(port, s_getbyte) && rb_respond_to(port, s_read)) {
2337 rb_check_funcall(port, s_binmode, 0, 0);
2338 }
2339 else {
2340 io_needed();
2341 }
2342 wrapper = TypedData_Make_Struct(0, struct load_arg, &load_arg_data, arg);
2343 arg->src = port;
2344 arg->offset = 0;
2345 arg->symbols = st_init_numtable();
2346 arg->data = rb_init_identtable();
2347 arg->partial_objects = rb_init_identtable();
2348 arg->compat_tbl = 0;
2349 arg->proc = 0;
2350 arg->readable = 0;
2351 arg->freeze = freeze;
2352
2353 if (NIL_P(v))
2354 arg->buf = xmalloc(BUFSIZ);
2355 else
2356 arg->buf = 0;
2357
2358 major = r_byte(arg);
2359 minor = r_byte(arg);
2360 if (major != MARSHAL_MAJOR || minor > MARSHAL_MINOR) {
2361 clear_load_arg(arg);
2362 rb_raise(rb_eTypeError, "incompatible marshal file format (can't be read)\n\
2363 \tformat version %d.%d required; %d.%d given",
2364 MARSHAL_MAJOR, MARSHAL_MINOR, major, minor);
2365 }
2366 if (RTEST(ruby_verbose) && minor != MARSHAL_MINOR) {
2367 rb_warn("incompatible marshal file format (can be read)\n\
2368 \tformat version %d.%d required; %d.%d given",
2369 MARSHAL_MAJOR, MARSHAL_MINOR, major, minor);
2370 }
2371
2372 if (!NIL_P(proc)) arg->proc = proc;
2373 v = r_object(arg);
2374 clear_load_arg(arg);
2375 RB_GC_GUARD(wrapper);
2376
2377 return v;
2378 }
2379
2380 static VALUE
2381 marshal_load(rb_execution_context_t *ec, VALUE mod, VALUE source, VALUE proc, VALUE freeze)
2382 {
2383 return rb_marshal_load_with_proc(source, proc, RTEST(freeze));
2384 }
2385
2386 #include "marshal.rbinc"
2387
2388 /*
2389 * The marshaling library converts collections of Ruby objects into a
2390 * byte stream, allowing them to be stored outside the currently
2391 * active script. This data may subsequently be read and the original
2392 * objects reconstituted.
2393 *
2394 * Marshaled data has major and minor version numbers stored along
2395 * with the object information. In normal use, marshaling can only
2396 * load data written with the same major version number and an equal
2397 * or lower minor version number. If Ruby's ``verbose'' flag is set
2398 * (normally using -d, -v, -w, or --verbose) the major and minor
2399 * numbers must match exactly. Marshal versioning is independent of
2400 * Ruby's version numbers. You can extract the version by reading the
2401 * first two bytes of marshaled data.
2402 *
2403 * str = Marshal.dump("thing")
2404 * RUBY_VERSION #=> "1.9.0"
2405 * str[0].ord #=> 4
2406 * str[1].ord #=> 8
2407 *
2408 * Some objects cannot be dumped: if the objects to be dumped include
2409 * bindings, procedure or method objects, instances of class IO, or
2410 * singleton objects, a TypeError will be raised.
2411 *
2412 * If your class has special serialization needs (for example, if you
2413 * want to serialize in some specific format), or if it contains
2414 * objects that would otherwise not be serializable, you can implement
2415 * your own serialization strategy.
2416 *
2417 * There are two methods of doing this, your object can define either
2418 * marshal_dump and marshal_load or _dump and _load. marshal_dump will take
2419 * precedence over _dump if both are defined. marshal_dump may result in
2420 * smaller Marshal strings.
2421 *
2422 * == Security considerations
2423 *
2424 * By design, Marshal.load can deserialize almost any class loaded into the
2425 * Ruby process. In many cases this can lead to remote code execution if the
2426 * Marshal data is loaded from an untrusted source.
2427 *
2428 * As a result, Marshal.load is not suitable as a general purpose serialization
2429 * format and you should never unmarshal user supplied input or other untrusted
2430 * data.
2431 *
2432 * If you need to deserialize untrusted data, use JSON or another serialization
2433 * format that is only able to load simple, 'primitive' types such as String,
2434 * Array, Hash, etc. Never allow user input to specify arbitrary types to
2435 * deserialize into.
2436 *
2437 * == marshal_dump and marshal_load
2438 *
2439 * When dumping an object the method marshal_dump will be called.
2440 * marshal_dump must return a result containing the information necessary for
2441 * marshal_load to reconstitute the object. The result can be any object.
2442 *
2443 * When loading an object dumped using marshal_dump the object is first
2444 * allocated then marshal_load is called with the result from marshal_dump.
2445 * marshal_load must recreate the object from the information in the result.
2446 *
2447 * Example:
2448 *
2449 * class MyObj
2450 * def initialize name, version, data
2451 * @name = name
2452 * @version = version
2453 * @data = data
2454 * end
2455 *
2456 * def marshal_dump
2457 * [@name, @version]
2458 * end
2459 *
2460 * def marshal_load array
2461 * @name, @version = array
2462 * end
2463 * end
2464 *
2465 * == _dump and _load
2466 *
2467 * Use _dump and _load when you need to allocate the object you're restoring
2468 * yourself.
2469 *
2470 * When dumping an object the instance method _dump is called with an Integer
2471 * which indicates the maximum depth of objects to dump (a value of -1 implies
2472 * that you should disable depth checking). _dump must return a String
2473 * containing the information necessary to reconstitute the object.
2474 *
2475 * The class method _load should take a String and use it to return an object
2476 * of the same class.
2477 *
2478 * Example:
2479 *
2480 * class MyObj
2481 * def initialize name, version, data
2482 * @name = name
2483 * @version = version
2484 * @data = data
2485 * end
2486 *
2487 * def _dump level
2488 * [@name, @version].join ':'
2489 * end
2490 *
2491 * def self._load args
2492 * new(*args.split(':'))
2493 * end
2494 * end
2495 *
2496 * Since Marshal.dump outputs a string you can have _dump return a Marshal
2497 * string which is Marshal.loaded in _load for complex objects.
2498 */
2499 void
2500 Init_marshal(void)
2501 {
2502 VALUE rb_mMarshal = rb_define_module("Marshal");
2503 #define set_id(sym) sym = rb_intern_const(name_##sym)
2504 set_id(s_dump);
2505 set_id(s_load);
2506 set_id(s_mdump);
2507 set_id(s_mload);
2508 set_id(s_dump_data);
2509 set_id(s_load_data);
2510 set_id(s_alloc);
2511 set_id(s_call);
2512 set_id(s_getbyte);
2513 set_id(s_read);
2514 set_id(s_write);
2515 set_id(s_binmode);
2516 set_id(s_encoding_short);
2517 set_id(s_ruby2_keywords_flag);
2518
2519 rb_define_module_function(rb_mMarshal, "dump", marshal_dump, -1);
2520
2521 /* major version */
2522 rb_define_const(rb_mMarshal, "MAJOR_VERSION", INT2FIX(MARSHAL_MAJOR));
2523 /* minor version */
2524 rb_define_const(rb_mMarshal, "MINOR_VERSION", INT2FIX(MARSHAL_MINOR));
2525 }
2526
2527 static int
2528 free_compat_i(st_data_t key, st_data_t value, st_data_t _)
2529 {
2530 xfree((marshal_compat_t *)value);
2531 return ST_CONTINUE;
2532 }
2533
2534 static void
2535 free_compat_allocator_table(void *data)
2536 {
2537 st_foreach(data, free_compat_i, 0);
2538 st_free_table(data);
2539 }
2540
2541 static st_table *
2542 compat_allocator_table(void)
2543 {
2544 if (compat_allocator_tbl) return compat_allocator_tbl;
2545 compat_allocator_tbl = st_init_numtable();
2546 #undef RUBY_UNTYPED_DATA_WARNING
2547 #define RUBY_UNTYPED_DATA_WARNING 0
2548 compat_allocator_tbl_wrapper =
2549 Data_Wrap_Struct(0, mark_marshal_compat_t, free_compat_allocator_table, compat_allocator_tbl);
2550 rb_vm_register_global_object(compat_allocator_tbl_wrapper);
2551 return compat_allocator_tbl;
2552 }
2553
2554 VALUE
2555 rb_marshal_dump(VALUE obj, VALUE port)
2556 {
2557 return rb_marshal_dump_limited(obj, port, -1);
2558 }
2559
2560 VALUE
2561 rb_marshal_load(VALUE port)
2562 {
2563 return rb_marshal_load_with_proc(port, Qnil, false);
2564 }
The Ruby Programming Language