Ruby 3.5.0dev (2025-06-27 revision 3d5619c8b1a76626e0991d758b71afc549829c38)
marshal.c (3d5619c8b1a76626e0991d758b71afc549829c38)
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#include "ruby/internal/attr/nonstring.h"
44
45#define BITSPERSHORT (2*CHAR_BIT)
46#define SHORTMASK ((1<<BITSPERSHORT)-1)
47#define SHORTDN(x) RSHIFT((x),BITSPERSHORT)
48
49#if SIZEOF_SHORT == SIZEOF_BDIGIT
50#define SHORTLEN(x) (x)
51#else
52static size_t
53shortlen(size_t len, BDIGIT *ds)
54{
55 BDIGIT num;
56 int offset = 0;
57
58 num = ds[len-1];
59 while (num) {
60 num = SHORTDN(num);
61 offset++;
62 }
63 return (len - 1)*SIZEOF_BDIGIT/2 + offset;
64}
65#define SHORTLEN(x) shortlen((x),d)
66#endif
67
68#define MARSHAL_MAJOR 4
69#define MARSHAL_MINOR 8
70
71#define TYPE_NIL '0'
72#define TYPE_TRUE 'T'
73#define TYPE_FALSE 'F'
74#define TYPE_FIXNUM 'i'
75
76#define TYPE_EXTENDED 'e'
77#define TYPE_UCLASS 'C'
78#define TYPE_OBJECT 'o'
79#define TYPE_DATA 'd'
80#define TYPE_USERDEF 'u'
81#define TYPE_USRMARSHAL 'U'
82#define TYPE_FLOAT 'f'
83#define TYPE_BIGNUM 'l'
84#define TYPE_STRING '"'
85#define TYPE_REGEXP '/'
86#define TYPE_ARRAY '['
87#define TYPE_HASH '{'
88#define TYPE_HASH_DEF '}'
89#define TYPE_STRUCT 'S'
90#define TYPE_MODULE_OLD 'M'
91#define TYPE_CLASS 'c'
92#define TYPE_MODULE 'm'
93
94#define TYPE_SYMBOL ':'
95#define TYPE_SYMLINK ';'
96
97#define TYPE_IVAR 'I'
98#define TYPE_LINK '@'
99
100static ID s_dump, s_load, s_mdump, s_mload;
101static ID s_dump_data, s_load_data, s_alloc, s_call;
102static ID s_getbyte, s_read, s_write, s_binmode;
103static ID s_encoding_short, s_ruby2_keywords_flag;
104#define s_encoding_long rb_id_encoding()
105
106#define name_s_dump "_dump"
107#define name_s_load "_load"
108#define name_s_mdump "marshal_dump"
109#define name_s_mload "marshal_load"
110#define name_s_dump_data "_dump_data"
111#define name_s_load_data "_load_data"
112#define name_s_alloc "_alloc"
113#define name_s_call "call"
114#define name_s_getbyte "getbyte"
115#define name_s_read "read"
116#define name_s_write "write"
117#define name_s_binmode "binmode"
118#define name_s_encoding_short "E"
119#define name_s_encoding_long "encoding"
120#define name_s_ruby2_keywords_flag "K"
121
122typedef struct {
123 VALUE newclass;
124 VALUE oldclass;
125 VALUE (*dumper)(VALUE);
126 VALUE (*loader)(VALUE, VALUE);
127} marshal_compat_t;
128
129static st_table *compat_allocator_tbl;
130static VALUE compat_allocator_tbl_wrapper;
131static VALUE rb_marshal_dump_limited(VALUE obj, VALUE port, int limit);
132static VALUE rb_marshal_load_with_proc(VALUE port, VALUE proc, bool freeze);
133
134static st_table *compat_allocator_table(void);
135
136void
137rb_marshal_define_compat(VALUE newclass, VALUE oldclass, VALUE (*dumper)(VALUE), VALUE (*loader)(VALUE, VALUE))
138{
139 marshal_compat_t *compat;
140 rb_alloc_func_t allocator = rb_get_alloc_func(newclass);
141
142 if (!allocator) {
143 rb_raise(rb_eTypeError, "no allocator");
144 }
145
146 compat_allocator_table();
147 compat = ALLOC(marshal_compat_t);
148 compat->newclass = newclass;
149 compat->oldclass = oldclass;
150 compat->dumper = dumper;
151 compat->loader = loader;
152
153 st_insert(compat_allocator_table(), (st_data_t)allocator, (st_data_t)compat);
154 RB_OBJ_WRITTEN(compat_allocator_tbl_wrapper, Qundef, newclass);
155 RB_OBJ_WRITTEN(compat_allocator_tbl_wrapper, Qundef, oldclass);
156}
157
158struct dump_arg {
159 VALUE str, dest;
160 st_table *symbols;
161 st_table *data;
162 st_table *compat_tbl;
163 st_table *encodings;
164 st_table *userdefs;
165 st_index_t num_entries;
166};
167
168struct dump_call_arg {
169 VALUE obj;
170 struct dump_arg *arg;
171 int limit;
172};
173
174static VALUE
175check_dump_arg(VALUE ret, struct dump_arg *arg, const char *name)
176{
177 if (!arg->symbols) {
178 rb_raise(rb_eRuntimeError, "Marshal.dump reentered at %s",
179 name);
180 }
181 return ret;
182}
183
184static VALUE
185check_userdump_arg(VALUE obj, ID sym, int argc, const VALUE *argv,
186 struct dump_arg *arg, const char *name)
187{
188 VALUE ret = rb_funcallv(obj, sym, argc, argv);
189 VALUE klass = CLASS_OF(obj);
190 if (CLASS_OF(ret) == klass) {
191 rb_raise(rb_eRuntimeError, "%"PRIsVALUE"#%s returned same class instance",
192 klass, name);
193 }
194 return check_dump_arg(ret, arg, name);
195}
196
197#define dump_funcall(arg, obj, sym, argc, argv) \
198 check_userdump_arg(obj, sym, argc, argv, arg, name_##sym)
199#define dump_check_funcall(arg, obj, sym, argc, argv) \
200 check_dump_arg(rb_check_funcall(obj, sym, argc, argv), arg, name_##sym)
201
202static void clear_dump_arg(struct dump_arg *arg);
203
204static void
205mark_dump_arg(void *ptr)
206{
207 struct dump_arg *p = ptr;
208 if (!p->symbols)
209 return;
210 rb_mark_set(p->symbols);
211 rb_mark_set(p->data);
212 rb_mark_hash(p->compat_tbl);
213 rb_mark_set(p->userdefs);
214 rb_gc_mark(p->str);
215}
216
217static void
218free_dump_arg(void *ptr)
219{
220 clear_dump_arg(ptr);
221}
222
223static size_t
224memsize_dump_arg(const void *ptr)
225{
226 const struct dump_arg *p = (struct dump_arg *)ptr;
227 size_t memsize = 0;
228 if (p->symbols) memsize += rb_st_memsize(p->symbols);
229 if (p->data) memsize += rb_st_memsize(p->data);
230 if (p->compat_tbl) memsize += rb_st_memsize(p->compat_tbl);
231 if (p->userdefs) memsize += rb_st_memsize(p->userdefs);
232 if (p->encodings) memsize += rb_st_memsize(p->encodings);
233 return memsize;
234}
235
236static const rb_data_type_t dump_arg_data = {
237 "dump_arg",
238 {mark_dump_arg, free_dump_arg, memsize_dump_arg,},
239 0, 0, RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_EMBEDDABLE
240};
241
242static VALUE
243must_not_be_anonymous(const char *type, VALUE path)
244{
245 char *n = RSTRING_PTR(path);
246
247 if (!rb_enc_asciicompat(rb_enc_get(path))) {
248 /* cannot occur? */
249 rb_raise(rb_eTypeError, "can't dump non-ascii %s name % "PRIsVALUE,
250 type, path);
251 }
252 if (n[0] == '#') {
253 rb_raise(rb_eTypeError, "can't dump anonymous %s % "PRIsVALUE,
254 type, path);
255 }
256 return path;
257}
258
259static VALUE
260class2path(VALUE klass)
261{
262 VALUE path = rb_class_path(klass);
263
264 must_not_be_anonymous((RB_TYPE_P(klass, T_CLASS) ? "class" : "module"), path);
265 if (rb_path_to_class(path) != rb_class_real(klass)) {
266 rb_raise(rb_eTypeError, "% "PRIsVALUE" can't be referred to", path);
267 }
268 return path;
269}
270
271int ruby_marshal_write_long(long x, char *buf);
272static void w_long(long, struct dump_arg*);
273static int w_encoding(VALUE encname, struct dump_call_arg *arg);
274static VALUE encoding_name(VALUE obj, struct dump_arg *arg);
275
276static void
277w_nbyte(const char *s, long n, struct dump_arg *arg)
278{
279 VALUE buf = arg->str;
280 rb_str_buf_cat(buf, s, n);
281 if (arg->dest && RSTRING_LEN(buf) >= BUFSIZ) {
282 rb_io_write(arg->dest, buf);
283 rb_str_resize(buf, 0);
284 }
285}
286
287static void
288w_byte(char c, struct dump_arg *arg)
289{
290 w_nbyte(&c, 1, arg);
291}
292
293static void
294w_bytes(const char *s, long n, struct dump_arg *arg)
295{
296 w_long(n, arg);
297 w_nbyte(s, n, arg);
298}
299
300#define w_cstr(s, arg) w_bytes((s), strlen(s), (arg))
301
302static void
303w_short(int x, struct dump_arg *arg)
304{
305 w_byte((char)((x >> 0) & 0xff), arg);
306 w_byte((char)((x >> 8) & 0xff), arg);
307}
308
309static void
310w_long(long x, struct dump_arg *arg)
311{
312 char buf[sizeof(long)+1];
313 int i = ruby_marshal_write_long(x, buf);
314 if (i < 0) {
315 rb_raise(rb_eTypeError, "long too big to dump");
316 }
317 w_nbyte(buf, i, arg);
318}
319
320int
321ruby_marshal_write_long(long x, char *buf)
322{
323 int i;
324
325#if SIZEOF_LONG > 4
326 if (!(RSHIFT(x, 31) == 0 || RSHIFT(x, 31) == -1)) {
327 /* big long does not fit in 4 bytes */
328 return -1;
329 }
330#endif
331
332 if (x == 0) {
333 buf[0] = 0;
334 return 1;
335 }
336 if (0 < x && x < 123) {
337 buf[0] = (char)(x + 5);
338 return 1;
339 }
340 if (-124 < x && x < 0) {
341 buf[0] = (char)((x - 5)&0xff);
342 return 1;
343 }
344 for (i=1;i<(int)sizeof(long)+1;i++) {
345 buf[i] = (char)(x & 0xff);
346 x = RSHIFT(x,8);
347 if (x == 0) {
348 buf[0] = i;
349 break;
350 }
351 if (x == -1) {
352 buf[0] = -i;
353 break;
354 }
355 }
356 return i+1;
357}
358
359#ifdef DBL_MANT_DIG
360#define DECIMAL_MANT (53-16) /* from IEEE754 double precision */
361
362#if DBL_MANT_DIG > 32
363#define MANT_BITS 32
364#elif DBL_MANT_DIG > 24
365#define MANT_BITS 24
366#elif DBL_MANT_DIG > 16
367#define MANT_BITS 16
368#else
369#define MANT_BITS 8
370#endif
371
372static double
373load_mantissa(double d, const char *buf, long len)
374{
375 if (!len) return d;
376 if (--len > 0 && !*buf++) { /* binary mantissa mark */
377 int e, s = d < 0, dig = 0;
378 unsigned long m;
379
380 modf(ldexp(frexp(fabs(d), &e), DECIMAL_MANT), &d);
381 do {
382 m = 0;
383 switch (len) {
384 default: m = *buf++ & 0xff; /* fall through */
385#if MANT_BITS > 24
386 case 3: m = (m << 8) | (*buf++ & 0xff); /* fall through */
387#endif
388#if MANT_BITS > 16
389 case 2: m = (m << 8) | (*buf++ & 0xff); /* fall through */
390#endif
391#if MANT_BITS > 8
392 case 1: m = (m << 8) | (*buf++ & 0xff);
393#endif
394 }
395 dig -= len < MANT_BITS / 8 ? 8 * (unsigned)len : MANT_BITS;
396 d += ldexp((double)m, dig);
397 } while ((len -= MANT_BITS / 8) > 0);
398 d = ldexp(d, e - DECIMAL_MANT);
399 if (s) d = -d;
400 }
401 return d;
402}
403#else
404#define load_mantissa(d, buf, len) (d)
405#endif
406
407#ifdef DBL_DIG
408#define FLOAT_DIG (DBL_DIG+2)
409#else
410#define FLOAT_DIG 17
411#endif
412
413static void
414w_float(double d, struct dump_arg *arg)
415{
416 char buf[FLOAT_DIG + (DECIMAL_MANT + 7) / 8 + 10];
417
418 if (isinf(d)) {
419 if (d < 0) w_cstr("-inf", arg);
420 else w_cstr("inf", arg);
421 }
422 else if (isnan(d)) {
423 w_cstr("nan", arg);
424 }
425 else if (d == 0.0) {
426 if (signbit(d)) w_cstr("-0", arg);
427 else w_cstr("0", arg);
428 }
429 else {
430 int decpt, sign, digs, len = 0;
431 char *e, *p = ruby_dtoa(d, 0, 0, &decpt, &sign, &e);
432 if (sign) buf[len++] = '-';
433 digs = (int)(e - p);
434 if (decpt < -3 || decpt > digs) {
435 buf[len++] = p[0];
436 if (--digs > 0) buf[len++] = '.';
437 memcpy(buf + len, p + 1, digs);
438 len += digs;
439 len += snprintf(buf + len, sizeof(buf) - len, "e%d", decpt - 1);
440 }
441 else if (decpt > 0) {
442 memcpy(buf + len, p, decpt);
443 len += decpt;
444 if ((digs -= decpt) > 0) {
445 buf[len++] = '.';
446 memcpy(buf + len, p + decpt, digs);
447 len += digs;
448 }
449 }
450 else {
451 buf[len++] = '0';
452 buf[len++] = '.';
453 if (decpt) {
454 memset(buf + len, '0', -decpt);
455 len -= decpt;
456 }
457 memcpy(buf + len, p, digs);
458 len += digs;
459 }
460 free(p);
461 w_bytes(buf, len, arg);
462 }
463}
464
465
466static VALUE
467w_encivar(VALUE str, struct dump_arg *arg)
468{
469 VALUE encname = encoding_name(str, arg);
470 if (NIL_P(encname) ||
471 is_ascii_string(str)) {
472 return Qnil;
473 }
474 w_byte(TYPE_IVAR, arg);
475 return encname;
476}
477
478static void
479w_encname(VALUE encname, struct dump_arg *arg)
480{
481 if (!NIL_P(encname)) {
482 struct dump_call_arg c_arg;
483 c_arg.limit = 1;
484 c_arg.arg = arg;
485 w_long(1L, arg);
486 w_encoding(encname, &c_arg);
487 }
488}
489
490static void
491w_symbol(VALUE sym, struct dump_arg *arg)
492{
493 st_data_t num;
494 VALUE encname;
495
496 if (st_lookup(arg->symbols, sym, &num)) {
497 w_byte(TYPE_SYMLINK, arg);
498 w_long((long)num, arg);
499 }
500 else {
501 const VALUE orig_sym = sym;
502 sym = rb_sym2str(sym);
503 if (!sym) {
504 rb_raise(rb_eTypeError, "can't dump anonymous ID %"PRIdVALUE, sym);
505 }
506 encname = w_encivar(sym, arg);
507 w_byte(TYPE_SYMBOL, arg);
508 w_bytes(RSTRING_PTR(sym), RSTRING_LEN(sym), arg);
509 st_add_direct(arg->symbols, orig_sym, arg->symbols->num_entries);
510 w_encname(encname, arg);
511 }
512}
513
514static void
515w_unique(VALUE s, struct dump_arg *arg)
516{
517 must_not_be_anonymous("class", s);
518 w_symbol(rb_str_intern(s), arg);
519}
520
521static void w_object(VALUE,struct dump_arg*,int);
522
523static int
524hash_each(VALUE key, VALUE value, VALUE v)
525{
526 struct dump_call_arg *arg = (void *)v;
527 w_object(key, arg->arg, arg->limit);
528 w_object(value, arg->arg, arg->limit);
529 return ST_CONTINUE;
530}
531
532#define SINGLETON_DUMP_UNABLE_P(klass) \
533 (rb_id_table_size(RCLASS_M_TBL(klass)) > 0 || \
534 rb_ivar_count(klass) > 0)
535
536static void
537w_extended(VALUE klass, struct dump_arg *arg, int check)
538{
539 if (check && RCLASS_SINGLETON_P(klass)) {
540 VALUE origin = RCLASS_ORIGIN(klass);
541 if (SINGLETON_DUMP_UNABLE_P(klass) ||
542 (origin != klass && SINGLETON_DUMP_UNABLE_P(origin))) {
543 rb_raise(rb_eTypeError, "singleton can't be dumped");
544 }
545 klass = RCLASS_SUPER(klass);
546 }
547 while (BUILTIN_TYPE(klass) == T_ICLASS) {
548 if (!RICLASS_IS_ORIGIN_P(klass) ||
549 BUILTIN_TYPE(RBASIC(klass)->klass) != T_MODULE) {
550 VALUE path = rb_class_name(RBASIC(klass)->klass);
551 w_byte(TYPE_EXTENDED, arg);
552 w_unique(path, arg);
553 }
554 klass = RCLASS_SUPER(klass);
555 }
556}
557
558static void
559w_class(char type, VALUE obj, struct dump_arg *arg, int check)
560{
561 VALUE path;
562 st_data_t real_obj;
563 VALUE klass;
564
565 if (arg->compat_tbl &&
566 st_lookup(arg->compat_tbl, (st_data_t)obj, &real_obj)) {
567 obj = (VALUE)real_obj;
568 }
569 klass = CLASS_OF(obj);
570 w_extended(klass, arg, check);
571 w_byte(type, arg);
572 path = class2path(rb_class_real(klass));
573 w_unique(path, arg);
574}
575
576static void
577w_uclass(VALUE obj, VALUE super, struct dump_arg *arg)
578{
579 VALUE klass = CLASS_OF(obj);
580
581 w_extended(klass, arg, TRUE);
582 klass = rb_class_real(klass);
583 if (klass != super) {
584 w_byte(TYPE_UCLASS, arg);
585 w_unique(class2path(klass), arg);
586 }
587}
588
589static bool
590rb_hash_ruby2_keywords_p(VALUE obj)
591{
592 return (RHASH(obj)->basic.flags & RHASH_PASS_AS_KEYWORDS) != 0;
593}
594
595static void
596rb_hash_ruby2_keywords(VALUE obj)
597{
598 RHASH(obj)->basic.flags |= RHASH_PASS_AS_KEYWORDS;
599}
600
601/*
602 * if instance variable name `id` is a special name to be skipped,
603 * returns the name of it. otherwise it cannot be dumped (unnamed),
604 * returns `name` as-is. returns NULL for ID that can be dumped.
605 */
606static inline const char *
607skipping_ivar_name(const ID id, const char *name)
608{
609#define IS_SKIPPED_IVAR(idname) \
610 ((id == idname) && (name = name_##idname, true))
611 if (IS_SKIPPED_IVAR(s_encoding_short)) return name;
612 if (IS_SKIPPED_IVAR(s_ruby2_keywords_flag)) return name;
613 if (IS_SKIPPED_IVAR(s_encoding_long)) return name;
614 if (!rb_id2str(id)) return name;
615 return NULL;
616}
617
618struct w_ivar_arg {
619 struct dump_call_arg *dump;
620 st_data_t num_ivar;
621};
622
623static int
624w_obj_each(ID id, VALUE value, st_data_t a)
625{
626 struct w_ivar_arg *ivarg = (struct w_ivar_arg *)a;
627 struct dump_call_arg *arg = ivarg->dump;
628 const char unnamed[] = "", *ivname = skipping_ivar_name(id, unnamed);
629
630 if (ivname) {
631 if (ivname != unnamed) {
632 rb_warn("instance variable '%s' on class %"PRIsVALUE" is not dumped",
633 ivname, CLASS_OF(arg->obj));
634 }
635 return ST_CONTINUE;
636 }
637 --ivarg->num_ivar;
638 w_symbol(ID2SYM(id), arg->arg);
639 w_object(value, arg->arg, arg->limit);
640 return ST_CONTINUE;
641}
642
643static int
644obj_count_ivars(ID id, VALUE val, st_data_t a)
645{
646 if (!skipping_ivar_name(id, "") && UNLIKELY(!++*(st_index_t *)a)) {
647 rb_raise(rb_eRuntimeError, "too many instance variables");
648 }
649 return ST_CONTINUE;
650}
651
652static VALUE
653encoding_name(VALUE obj, struct dump_arg *arg)
654{
655 if (rb_enc_capable(obj)) {
656 int encidx = rb_enc_get_index(obj);
657 rb_encoding *enc = 0;
658 st_data_t name;
659
660 if (encidx <= 0 || !(enc = rb_enc_from_index(encidx))) {
661 return Qnil;
662 }
663
664 /* special treatment for US-ASCII and UTF-8 */
665 if (encidx == rb_usascii_encindex()) {
666 return Qfalse;
667 }
668 else if (encidx == rb_utf8_encindex()) {
669 return Qtrue;
670 }
671
672 if (arg->encodings ?
673 !st_lookup(arg->encodings, (st_data_t)rb_enc_name(enc), &name) :
674 (arg->encodings = st_init_strcasetable(), 1)) {
675 name = (st_data_t)rb_str_new_cstr(rb_enc_name(enc));
676 st_insert(arg->encodings, (st_data_t)rb_enc_name(enc), name);
677 }
678 return (VALUE)name;
679 }
680 else {
681 return Qnil;
682 }
683}
684
685static int
686w_encoding(VALUE encname, struct dump_call_arg *arg)
687{
688 int limit = arg->limit;
689 if (limit >= 0) ++limit;
690 switch (encname) {
691 case Qfalse:
692 case Qtrue:
693 w_symbol(ID2SYM(s_encoding_short), arg->arg);
694 w_object(encname, arg->arg, limit);
695 return 1;
696 case Qnil:
697 return 0;
698 }
699 w_symbol(ID2SYM(rb_id_encoding()), arg->arg);
700 w_object(encname, arg->arg, limit);
701 return 1;
702}
703
704static st_index_t
705has_ivars(VALUE obj, VALUE encname, VALUE *ivobj)
706{
707 st_index_t num = !NIL_P(encname);
708
709 if (SPECIAL_CONST_P(obj)) goto generic;
710 switch (BUILTIN_TYPE(obj)) {
711 case T_OBJECT:
712 case T_CLASS:
713 case T_MODULE:
714 break; /* counted elsewhere */
715 case T_HASH:
716 if (rb_hash_ruby2_keywords_p(obj)) ++num;
717 /* fall through */
718 default:
719 generic:
720 rb_ivar_foreach(obj, obj_count_ivars, (st_data_t)&num);
721 if (num) *ivobj = obj;
722 }
723
724 return num;
725}
726
727static void
728w_ivar_each(VALUE obj, st_index_t num, struct dump_call_arg *arg)
729{
730 shape_id_t shape_id = rb_obj_shape_id(arg->obj);
731 struct w_ivar_arg ivarg = {arg, num};
732 if (!num) return;
733 rb_ivar_foreach(obj, w_obj_each, (st_data_t)&ivarg);
734
735 shape_id_t actual_shape_id = rb_obj_shape_id(arg->obj);
736 if (shape_id != actual_shape_id) {
737 // If the shape tree got _shorter_ then we probably removed an IV
738 // If the shape tree got longer, then we probably added an IV.
739 // The exception message might not be accurate when someone adds and
740 // removes the same number of IVs, but they will still get an exception
741 if (rb_shape_depth(shape_id) > rb_shape_depth(rb_obj_shape_id(arg->obj))) {
742 rb_raise(rb_eRuntimeError, "instance variable removed from %"PRIsVALUE" instance",
743 CLASS_OF(arg->obj));
744 }
745 else {
746 rb_raise(rb_eRuntimeError, "instance variable added to %"PRIsVALUE" instance",
747 CLASS_OF(arg->obj));
748 }
749 }
750}
751
752static void
753w_ivar(st_index_t num, VALUE ivobj, VALUE encname, struct dump_call_arg *arg)
754{
755 w_long(num, arg->arg);
756 num -= w_encoding(encname, arg);
757 if (RB_TYPE_P(ivobj, T_HASH) && rb_hash_ruby2_keywords_p(ivobj)) {
758 int limit = arg->limit;
759 if (limit >= 0) ++limit;
760 w_symbol(ID2SYM(s_ruby2_keywords_flag), arg->arg);
761 w_object(Qtrue, arg->arg, limit);
762 num--;
763 }
764 if (!UNDEF_P(ivobj) && num) {
765 w_ivar_each(ivobj, num, arg);
766 }
767}
768
769static void
770w_objivar(VALUE obj, struct dump_call_arg *arg)
771{
772 st_data_t num = 0;
773
774 rb_ivar_foreach(obj, obj_count_ivars, (st_data_t)&num);
775 w_long(num, arg->arg);
776 w_ivar_each(obj, num, arg);
777}
778
779#if SIZEOF_LONG > 4
780// Optimized dump for fixnum larger than 31-bits
781static void
782w_bigfixnum(VALUE obj, struct dump_arg *arg)
783{
784 RUBY_ASSERT(FIXNUM_P(obj));
785
786 w_byte(TYPE_BIGNUM, arg);
787
788#if SIZEOF_LONG == SIZEOF_VALUE
789 long num, slen_num;
790 num = FIX2LONG(obj);
791#else
792 long long num, slen_num;
793 num = NUM2LL(obj);
794#endif
795
796 char sign = num < 0 ? '-' : '+';
797 w_byte(sign, arg);
798
799 // Guaranteed not to overflow, as FIXNUM is 1-bit less than long
800 if (num < 0) num = -num;
801
802 // calculate the size in shorts
803 int slen = 0;
804 {
805 slen_num = num;
806 while (slen_num) {
807 slen++;
808 slen_num = SHORTDN(slen_num);
809 }
810 }
811
812 RUBY_ASSERT(slen > 0 && slen <= SIZEOF_LONG / 2);
813
814 w_long((long)slen, arg);
815
816 for (int i = 0; i < slen; i++) {
817 w_short(num & SHORTMASK, arg);
818 num = SHORTDN(num);
819 }
820
821 // We aren't adding this object to the link table, but we need to increment
822 // the index.
823 arg->num_entries++;
824
825 RUBY_ASSERT(num == 0);
826}
827#endif
828
829static void
830w_remember(VALUE obj, struct dump_arg *arg)
831{
832 st_add_direct(arg->data, obj, arg->num_entries++);
833}
834
835static void
836w_object(VALUE obj, struct dump_arg *arg, int limit)
837{
838 struct dump_call_arg c_arg;
839 VALUE ivobj = Qundef;
840 st_data_t num;
841 st_index_t hasiv = 0;
842 VALUE encname = Qnil;
843
844 if (limit == 0) {
845 rb_raise(rb_eArgError, "exceed depth limit");
846 }
847
848 if (NIL_P(obj)) {
849 w_byte(TYPE_NIL, arg);
850 }
851 else if (obj == Qtrue) {
852 w_byte(TYPE_TRUE, arg);
853 }
854 else if (obj == Qfalse) {
855 w_byte(TYPE_FALSE, arg);
856 }
857 else if (FIXNUM_P(obj)) {
858#if SIZEOF_LONG <= 4
859 w_byte(TYPE_FIXNUM, arg);
860 w_long(FIX2INT(obj), arg);
861#else
862 if (RSHIFT((long)obj, 31) == 0 || RSHIFT((long)obj, 31) == -1) {
863 w_byte(TYPE_FIXNUM, arg);
864 w_long(FIX2LONG(obj), arg);
865 }
866 else {
867 w_bigfixnum(obj, arg);
868 }
869#endif
870 }
871 else if (SYMBOL_P(obj)) {
872 w_symbol(obj, arg);
873 }
874 else {
875 if (st_lookup(arg->data, obj, &num)) {
876 w_byte(TYPE_LINK, arg);
877 w_long((long)num, arg);
878 return;
879 }
880
881 if (limit > 0) limit--;
882 c_arg.limit = limit;
883 c_arg.arg = arg;
884 c_arg.obj = obj;
885
886 if (FLONUM_P(obj)) {
887 w_remember(obj, arg);
888 w_byte(TYPE_FLOAT, arg);
889 w_float(RFLOAT_VALUE(obj), arg);
890 return;
891 }
892
893 VALUE v;
894
895 if (!RBASIC_CLASS(obj)) {
896 rb_raise(rb_eTypeError, "can't dump internal %s",
897 rb_builtin_type_name(BUILTIN_TYPE(obj)));
898 }
899
900 if (rb_obj_respond_to(obj, s_mdump, TRUE)) {
901 w_remember(obj, arg);
902
903 v = dump_funcall(arg, obj, s_mdump, 0, 0);
904 w_class(TYPE_USRMARSHAL, obj, arg, FALSE);
905 w_object(v, arg, limit);
906 return;
907 }
908 if (rb_obj_respond_to(obj, s_dump, TRUE)) {
909 VALUE ivobj2 = Qundef;
910 st_index_t hasiv2;
911 VALUE encname2;
912
913 if (arg->userdefs && st_is_member(arg->userdefs, (st_data_t)obj)) {
914 rb_raise(rb_eRuntimeError, "can't dump recursive object using _dump()");
915 }
916 v = INT2NUM(limit);
917 v = dump_funcall(arg, obj, s_dump, 1, &v);
918 if (!RB_TYPE_P(v, T_STRING)) {
919 rb_raise(rb_eTypeError, "_dump() must return string");
920 }
921 hasiv = has_ivars(obj, (encname = encoding_name(obj, arg)), &ivobj);
922 hasiv2 = has_ivars(v, (encname2 = encoding_name(v, arg)), &ivobj2);
923 if (hasiv2) {
924 hasiv = hasiv2;
925 ivobj = ivobj2;
926 encname = encname2;
927 }
928 if (hasiv) w_byte(TYPE_IVAR, arg);
929 w_class(TYPE_USERDEF, obj, arg, FALSE);
930 w_bytes(RSTRING_PTR(v), RSTRING_LEN(v), arg);
931 if (hasiv) {
932 st_data_t userdefs = (st_data_t)obj;
933 if (!arg->userdefs) {
934 arg->userdefs = rb_init_identtable();
935 }
936 st_add_direct(arg->userdefs, userdefs, 0);
937 w_ivar(hasiv, ivobj, encname, &c_arg);
938 st_delete(arg->userdefs, &userdefs, NULL);
939 }
940 w_remember(obj, arg);
941 return;
942 }
943
944 w_remember(obj, arg);
945
946 hasiv = has_ivars(obj, (encname = encoding_name(obj, arg)), &ivobj);
947 {
948 st_data_t compat_data;
949 rb_alloc_func_t allocator = rb_get_alloc_func(RBASIC(obj)->klass);
950 if (st_lookup(compat_allocator_tbl,
951 (st_data_t)allocator,
952 &compat_data)) {
953 marshal_compat_t *compat = (marshal_compat_t*)compat_data;
954 VALUE real_obj = obj;
955 obj = compat->dumper(real_obj);
956 if (!arg->compat_tbl) {
957 arg->compat_tbl = rb_init_identtable();
958 }
959 st_insert(arg->compat_tbl, (st_data_t)obj, (st_data_t)real_obj);
960 if (obj != real_obj && UNDEF_P(ivobj)) hasiv = 0;
961 }
962 }
963 if (hasiv) w_byte(TYPE_IVAR, arg);
964
965 switch (BUILTIN_TYPE(obj)) {
966 case T_CLASS:
967 if (FL_TEST(obj, FL_SINGLETON)) {
968 rb_raise(rb_eTypeError, "singleton class can't be dumped");
969 }
970 {
971 VALUE path = class2path(obj);
972 VALUE encname = w_encivar(path, arg);
973 w_byte(TYPE_CLASS, arg);
974 w_bytes(RSTRING_PTR(path), RSTRING_LEN(path), arg);
975 w_encname(encname, arg);
976 RB_GC_GUARD(path);
977 }
978 break;
979
980 case T_MODULE:
981 {
982 VALUE path = class2path(obj);
983 VALUE encname = w_encivar(path, arg);
984 w_byte(TYPE_MODULE, arg);
985 w_bytes(RSTRING_PTR(path), RSTRING_LEN(path), arg);
986 w_encname(encname, arg);
987 RB_GC_GUARD(path);
988 }
989 break;
990
991 case T_FLOAT:
992 w_byte(TYPE_FLOAT, arg);
993 w_float(RFLOAT_VALUE(obj), arg);
994 break;
995
996 case T_BIGNUM:
997 w_byte(TYPE_BIGNUM, arg);
998 {
999 char sign = BIGNUM_SIGN(obj) ? '+' : '-';
1000 size_t len = BIGNUM_LEN(obj);
1001 size_t slen;
1002 size_t j;
1003 BDIGIT *d = BIGNUM_DIGITS(obj);
1004
1005 slen = SHORTLEN(len);
1006 if (LONG_MAX < slen) {
1007 rb_raise(rb_eTypeError, "too big Bignum can't be dumped");
1008 }
1009
1010 w_byte(sign, arg);
1011 w_long((long)slen, arg);
1012 for (j = 0; j < len; j++) {
1013#if SIZEOF_BDIGIT > SIZEOF_SHORT
1014 BDIGIT num = *d;
1015 int i;
1016
1017 for (i=0; i<SIZEOF_BDIGIT; i+=SIZEOF_SHORT) {
1018 w_short(num & SHORTMASK, arg);
1019 num = SHORTDN(num);
1020 if (j == len - 1 && num == 0) break;
1021 }
1022#else
1023 w_short(*d, arg);
1024#endif
1025 d++;
1026 }
1027 }
1028 break;
1029
1030 case T_STRING:
1031 w_uclass(obj, rb_cString, arg);
1032 w_byte(TYPE_STRING, arg);
1033 w_bytes(RSTRING_PTR(obj), RSTRING_LEN(obj), arg);
1034 break;
1035
1036 case T_REGEXP:
1037 w_uclass(obj, rb_cRegexp, arg);
1038 w_byte(TYPE_REGEXP, arg);
1039 {
1040 int opts = rb_reg_options(obj);
1041 w_bytes(RREGEXP_SRC_PTR(obj), RREGEXP_SRC_LEN(obj), arg);
1042 w_byte((char)opts, arg);
1043 }
1044 break;
1045
1046 case T_ARRAY:
1047 w_uclass(obj, rb_cArray, arg);
1048 w_byte(TYPE_ARRAY, arg);
1049 {
1050 long i, len = RARRAY_LEN(obj);
1051
1052 w_long(len, arg);
1053 for (i=0; i<RARRAY_LEN(obj); i++) {
1054 w_object(RARRAY_AREF(obj, i), arg, limit);
1055 if (len != RARRAY_LEN(obj)) {
1056 rb_raise(rb_eRuntimeError, "array modified during dump");
1057 }
1058 }
1059 }
1060 break;
1061
1062 case T_HASH:
1063 w_uclass(obj, rb_cHash, arg);
1064 if (rb_hash_compare_by_id_p(obj)) {
1065 w_byte(TYPE_UCLASS, arg);
1066 w_symbol(rb_sym_intern_ascii_cstr("Hash"), arg);
1067 }
1068 if (NIL_P(RHASH_IFNONE(obj))) {
1069 w_byte(TYPE_HASH, arg);
1070 }
1071 else if (FL_TEST(obj, RHASH_PROC_DEFAULT)) {
1072 rb_raise(rb_eTypeError, "can't dump hash with default proc");
1073 }
1074 else {
1075 w_byte(TYPE_HASH_DEF, arg);
1076 }
1077 w_long(rb_hash_size_num(obj), arg);
1078 rb_hash_foreach(obj, hash_each, (st_data_t)&c_arg);
1079 if (!NIL_P(RHASH_IFNONE(obj))) {
1080 w_object(RHASH_IFNONE(obj), arg, limit);
1081 }
1082 break;
1083
1084 case T_STRUCT:
1085 w_class(TYPE_STRUCT, obj, arg, TRUE);
1086 {
1087 long len = RSTRUCT_LEN(obj);
1088 VALUE mem;
1089 long i;
1090
1091 w_long(len, arg);
1092 mem = rb_struct_members(obj);
1093 for (i=0; i<len; i++) {
1094 w_symbol(RARRAY_AREF(mem, i), arg);
1095 w_object(RSTRUCT_GET(obj, i), arg, limit);
1096 }
1097 }
1098 break;
1099
1100 case T_OBJECT:
1101 w_class(TYPE_OBJECT, obj, arg, TRUE);
1102 w_objivar(obj, &c_arg);
1103 break;
1104
1105 case T_DATA:
1106 {
1107 VALUE v;
1108
1109 if (!rb_obj_respond_to(obj, s_dump_data, TRUE)) {
1110 rb_raise(rb_eTypeError,
1111 "no _dump_data is defined for class %"PRIsVALUE,
1112 rb_obj_class(obj));
1113 }
1114 v = dump_funcall(arg, obj, s_dump_data, 0, 0);
1115 w_class(TYPE_DATA, obj, arg, TRUE);
1116 w_object(v, arg, limit);
1117 }
1118 break;
1119
1120 default:
1121 rb_raise(rb_eTypeError, "can't dump %"PRIsVALUE,
1122 rb_obj_class(obj));
1123 break;
1124 }
1125 RB_GC_GUARD(obj);
1126 }
1127 if (hasiv) {
1128 w_ivar(hasiv, ivobj, encname, &c_arg);
1129 }
1130}
1131
1132static void
1133clear_dump_arg(struct dump_arg *arg)
1134{
1135 if (!arg->symbols) return;
1136 st_free_table(arg->symbols);
1137 arg->symbols = 0;
1138 st_free_table(arg->data);
1139 arg->data = 0;
1140 arg->num_entries = 0;
1141 if (arg->compat_tbl) {
1142 st_free_table(arg->compat_tbl);
1143 arg->compat_tbl = 0;
1144 }
1145 if (arg->encodings) {
1146 st_free_table(arg->encodings);
1147 arg->encodings = 0;
1148 }
1149 if (arg->userdefs) {
1150 st_free_table(arg->userdefs);
1151 arg->userdefs = 0;
1152 }
1153}
1154
1155NORETURN(static inline void io_needed(void));
1156static inline void
1157io_needed(void)
1158{
1159 rb_raise(rb_eTypeError, "instance of IO needed");
1160}
1161
1162/*
1163 * call-seq:
1164 * dump( obj [, anIO] , limit=-1 ) -> anIO
1165 *
1166 * Serializes obj and all descendant objects. If anIO is
1167 * specified, the serialized data will be written to it, otherwise the
1168 * data will be returned as a String. If limit is specified, the
1169 * traversal of subobjects will be limited to that depth. If limit is
1170 * negative, no checking of depth will be performed.
1171 *
1172 * class Klass
1173 * def initialize(str)
1174 * @str = str
1175 * end
1176 * def say_hello
1177 * @str
1178 * end
1179 * end
1180 *
1181 * (produces no output)
1182 *
1183 * o = Klass.new("hello\n")
1184 * data = Marshal.dump(o)
1185 * obj = Marshal.load(data)
1186 * obj.say_hello #=> "hello\n"
1187 *
1188 * Marshal can't dump following objects:
1189 * * anonymous Class/Module.
1190 * * objects which are related to system (ex: Dir, File::Stat, IO, File, Socket
1191 * and so on)
1192 * * an instance of MatchData, Data, Method, UnboundMethod, Proc, Thread,
1193 * ThreadGroup, Continuation
1194 * * objects which define singleton methods
1195 */
1196static VALUE
1197marshal_dump(int argc, VALUE *argv, VALUE _)
1198{
1199 VALUE obj, port, a1, a2;
1200 int limit = -1;
1201
1202 port = Qnil;
1203 rb_scan_args(argc, argv, "12", &obj, &a1, &a2);
1204 if (argc == 3) {
1205 if (!NIL_P(a2)) limit = NUM2INT(a2);
1206 if (NIL_P(a1)) io_needed();
1207 port = a1;
1208 }
1209 else if (argc == 2) {
1210 if (FIXNUM_P(a1)) limit = FIX2INT(a1);
1211 else if (NIL_P(a1)) io_needed();
1212 else port = a1;
1213 }
1214 return rb_marshal_dump_limited(obj, port, limit);
1215}
1216
1217VALUE
1218rb_marshal_dump_limited(VALUE obj, VALUE port, int limit)
1219{
1220 struct dump_arg *arg;
1221 VALUE wrapper; /* used to avoid memory leak in case of exception */
1222
1223 wrapper = TypedData_Make_Struct(0, struct dump_arg, &dump_arg_data, arg);
1224 arg->dest = 0;
1225 arg->symbols = st_init_numtable();
1226 arg->data = rb_init_identtable();
1227 arg->num_entries = 0;
1228 arg->compat_tbl = 0;
1229 arg->encodings = 0;
1230 arg->userdefs = 0;
1231 arg->str = rb_str_buf_new(0);
1232 if (!NIL_P(port)) {
1233 if (!rb_respond_to(port, s_write)) {
1234 io_needed();
1235 }
1236 arg->dest = port;
1237 dump_check_funcall(arg, port, s_binmode, 0, 0);
1238 }
1239 else {
1240 port = arg->str;
1241 }
1242
1243 w_byte(MARSHAL_MAJOR, arg);
1244 w_byte(MARSHAL_MINOR, arg);
1245
1246 w_object(obj, arg, limit);
1247 if (arg->dest) {
1248 rb_io_write(arg->dest, arg->str);
1249 rb_str_resize(arg->str, 0);
1250 }
1251 clear_dump_arg(arg);
1252 RB_GC_GUARD(wrapper);
1253
1254 return port;
1255}
1256
1257struct load_arg {
1258 VALUE src;
1259 char *buf;
1260 long buflen;
1261 long readable;
1262 long offset;
1263 st_table *symbols;
1264 st_table *data;
1265 st_table *partial_objects;
1266 VALUE proc;
1267 st_table *compat_tbl;
1268 bool freeze;
1269};
1270
1271static VALUE
1272check_load_arg(VALUE ret, struct load_arg *arg, const char *name)
1273{
1274 if (!arg->symbols) {
1275 rb_raise(rb_eRuntimeError, "Marshal.load reentered at %s",
1276 name);
1277 }
1278 return ret;
1279}
1280#define load_funcall(arg, obj, sym, argc, argv) \
1281 check_load_arg(rb_funcallv(obj, sym, argc, argv), arg, name_##sym)
1282
1283static void clear_load_arg(struct load_arg *arg);
1284
1285static void
1286mark_load_arg(void *ptr)
1287{
1288 struct load_arg *p = ptr;
1289 if (!p->symbols)
1290 return;
1291 rb_mark_tbl(p->symbols);
1292 rb_mark_tbl(p->data);
1293 rb_mark_tbl(p->partial_objects);
1294 rb_mark_hash(p->compat_tbl);
1295}
1296
1297static void
1298free_load_arg(void *ptr)
1299{
1300 clear_load_arg(ptr);
1301}
1302
1303static size_t
1304memsize_load_arg(const void *ptr)
1305{
1306 const struct load_arg *p = (struct load_arg *)ptr;
1307 size_t memsize = 0;
1308 if (p->symbols) memsize += rb_st_memsize(p->symbols);
1309 if (p->data) memsize += rb_st_memsize(p->data);
1310 if (p->partial_objects) memsize += rb_st_memsize(p->partial_objects);
1311 if (p->compat_tbl) memsize += rb_st_memsize(p->compat_tbl);
1312 return memsize;
1313}
1314
1315static const rb_data_type_t load_arg_data = {
1316 "load_arg",
1317 {mark_load_arg, free_load_arg, memsize_load_arg,},
1318 0, 0, RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_EMBEDDABLE
1319};
1320
1321#define r_entry(v, arg) r_entry0((v), (arg)->data->num_entries, (arg))
1322static VALUE r_object(struct load_arg *arg);
1323static VALUE r_symbol(struct load_arg *arg);
1324
1325NORETURN(static void too_short(void));
1326static void
1327too_short(void)
1328{
1329 rb_raise(rb_eArgError, "marshal data too short");
1330}
1331
1332static st_index_t
1333r_prepare(struct load_arg *arg)
1334{
1335 st_index_t idx = arg->data->num_entries;
1336
1337 st_insert(arg->data, (st_data_t)idx, (st_data_t)Qundef);
1338 return idx;
1339}
1340
1341static unsigned char
1342r_byte1_buffered(struct load_arg *arg)
1343{
1344 if (arg->buflen == 0) {
1345 long readable = arg->readable < BUFSIZ ? arg->readable : BUFSIZ;
1346 VALUE str, n = LONG2NUM(readable);
1347
1348 str = load_funcall(arg, arg->src, s_read, 1, &n);
1349 if (NIL_P(str)) too_short();
1350 StringValue(str);
1351 memcpy(arg->buf, RSTRING_PTR(str), RSTRING_LEN(str));
1352 arg->offset = 0;
1353 arg->buflen = RSTRING_LEN(str);
1354 }
1355 arg->buflen--;
1356 return arg->buf[arg->offset++];
1357}
1358
1359static int
1360r_byte(struct load_arg *arg)
1361{
1362 int c;
1363
1364 if (RB_TYPE_P(arg->src, T_STRING)) {
1365 if (RSTRING_LEN(arg->src) > arg->offset) {
1366 c = (unsigned char)RSTRING_PTR(arg->src)[arg->offset++];
1367 }
1368 else {
1369 too_short();
1370 }
1371 }
1372 else {
1373 if (arg->readable >0 || arg->buflen > 0) {
1374 c = r_byte1_buffered(arg);
1375 }
1376 else {
1377 VALUE v = load_funcall(arg, arg->src, s_getbyte, 0, 0);
1378 if (NIL_P(v)) rb_eof_error();
1379 c = (unsigned char)NUM2CHR(v);
1380 }
1381 }
1382 return c;
1383}
1384
1385NORETURN(static void long_toobig(int size));
1386
1387static void
1388long_toobig(int size)
1389{
1390 rb_raise(rb_eTypeError, "long too big for this architecture (size "
1391 STRINGIZE(SIZEOF_LONG)", given %d)", size);
1392}
1393
1394static long
1395r_long(struct load_arg *arg)
1396{
1397 register long x;
1398 int c = (signed char)r_byte(arg);
1399 long i;
1400
1401 if (c == 0) return 0;
1402 if (c > 0) {
1403 if (4 < c && c < 128) {
1404 return c - 5;
1405 }
1406 if (c > (int)sizeof(long)) long_toobig(c);
1407 x = 0;
1408 for (i=0;i<c;i++) {
1409 x |= (long)r_byte(arg) << (8*i);
1410 }
1411 }
1412 else {
1413 if (-129 < c && c < -4) {
1414 return c + 5;
1415 }
1416 c = -c;
1417 if (c > (int)sizeof(long)) long_toobig(c);
1418 x = -1;
1419 for (i=0;i<c;i++) {
1420 x &= ~((long)0xff << (8*i));
1421 x |= (long)r_byte(arg) << (8*i);
1422 }
1423 }
1424 return x;
1425}
1426
1427long
1428ruby_marshal_read_long(const char **buf, long len)
1429{
1430 long x;
1431 struct RString src;
1432 struct load_arg arg;
1433 memset(&arg, 0, sizeof(arg));
1434 arg.src = rb_setup_fake_str(&src, *buf, len, 0);
1435 x = r_long(&arg);
1436 *buf += arg.offset;
1437 return x;
1438}
1439
1440static VALUE
1441r_bytes1(long len, struct load_arg *arg)
1442{
1443 VALUE str, n = LONG2NUM(len);
1444
1445 str = load_funcall(arg, arg->src, s_read, 1, &n);
1446 if (NIL_P(str)) too_short();
1447 StringValue(str);
1448 if (RSTRING_LEN(str) != len) too_short();
1449
1450 return str;
1451}
1452
1453static VALUE
1454r_bytes1_buffered(long len, struct load_arg *arg)
1455{
1456 VALUE str;
1457
1458 if (len <= arg->buflen) {
1459 str = rb_str_new(arg->buf+arg->offset, len);
1460 arg->offset += len;
1461 arg->buflen -= len;
1462 }
1463 else {
1464 long buflen = arg->buflen;
1465 long readable = arg->readable + 1;
1466 long tmp_len, read_len, need_len = len - buflen;
1467 VALUE tmp, n;
1468
1469 readable = readable < BUFSIZ ? readable : BUFSIZ;
1470 read_len = need_len > readable ? need_len : readable;
1471 n = LONG2NUM(read_len);
1472 tmp = load_funcall(arg, arg->src, s_read, 1, &n);
1473 if (NIL_P(tmp)) too_short();
1474 StringValue(tmp);
1475
1476 tmp_len = RSTRING_LEN(tmp);
1477
1478 if (tmp_len < need_len) too_short();
1479
1480 str = rb_str_new(arg->buf+arg->offset, buflen);
1481 rb_str_cat(str, RSTRING_PTR(tmp), need_len);
1482
1483 if (tmp_len > need_len) {
1484 buflen = tmp_len - need_len;
1485 memcpy(arg->buf, RSTRING_PTR(tmp)+need_len, buflen);
1486 arg->buflen = buflen;
1487 }
1488 else {
1489 arg->buflen = 0;
1490 }
1491 arg->offset = 0;
1492 }
1493
1494 return str;
1495}
1496
1497#define r_bytes(arg) r_bytes0(r_long(arg), (arg))
1498
1499static VALUE
1500r_bytes0(long len, struct load_arg *arg)
1501{
1502 VALUE str;
1503
1504 if (len == 0) return rb_str_new(0, 0);
1505 if (RB_TYPE_P(arg->src, T_STRING)) {
1506 if (RSTRING_LEN(arg->src) - arg->offset >= len) {
1507 str = rb_str_new(RSTRING_PTR(arg->src)+arg->offset, len);
1508 arg->offset += len;
1509 }
1510 else {
1511 too_short();
1512 }
1513 }
1514 else {
1515 if (arg->readable > 0 || arg->buflen > 0) {
1516 str = r_bytes1_buffered(len, arg);
1517 }
1518 else {
1519 str = r_bytes1(len, arg);
1520 }
1521 }
1522 return str;
1523}
1524
1525static inline int
1526name_equal(const char *name, size_t nlen, const char *p, long l)
1527{
1528 if ((size_t)l != nlen || *p != *name) return 0;
1529 return nlen == 1 || memcmp(p+1, name+1, nlen-1) == 0;
1530}
1531
1532static int
1533sym2encidx(VALUE sym, VALUE val)
1534{
1535 RBIMPL_ATTR_NONSTRING() static const char name_encoding[8] = "encoding";
1536 const char *p;
1537 long l;
1538 if (rb_enc_get_index(sym) != ENCINDEX_US_ASCII) return -1;
1539 RSTRING_GETMEM(sym, p, l);
1540 if (l <= 0) return -1;
1541 if (name_equal(name_encoding, sizeof(name_encoding), p, l)) {
1542 int idx = rb_enc_find_index(StringValueCStr(val));
1543 return idx;
1544 }
1545 if (name_equal(name_s_encoding_short, rb_strlen_lit(name_s_encoding_short), p, l)) {
1546 if (val == Qfalse) return rb_usascii_encindex();
1547 else if (val == Qtrue) return rb_utf8_encindex();
1548 /* bogus ignore */
1549 }
1550 return -1;
1551}
1552
1553static int
1554symname_equal(VALUE sym, const char *name, size_t nlen)
1555{
1556 const char *p;
1557 long l;
1558 if (rb_enc_get_index(sym) != ENCINDEX_US_ASCII) return 0;
1559 RSTRING_GETMEM(sym, p, l);
1560 return name_equal(name, nlen, p, l);
1561}
1562
1563#define BUILD_ASSERT_POSITIVE(n) \
1564 /* make 0 negative to workaround the "zero size array" GCC extension, */ \
1565 ((sizeof(char [2*(ssize_t)(n)-1])+1)/2) /* assuming no overflow */
1566#define symname_equal_lit(sym, sym_name) \
1567 symname_equal(sym, sym_name, BUILD_ASSERT_POSITIVE(rb_strlen_lit(sym_name)))
1568
1569static VALUE
1570r_symlink(struct load_arg *arg)
1571{
1572 st_data_t sym;
1573 long num = r_long(arg);
1574
1575 if (!st_lookup(arg->symbols, num, &sym)) {
1576 rb_raise(rb_eArgError, "bad symbol");
1577 }
1578 return (VALUE)sym;
1579}
1580
1581static VALUE
1582r_symreal(struct load_arg *arg, int ivar)
1583{
1584 VALUE s = r_bytes(arg);
1585 VALUE sym;
1586 int idx = -1;
1587 st_index_t n = arg->symbols->num_entries;
1588
1589 if (rb_enc_str_asciionly_p(s)) rb_enc_associate_index(s, ENCINDEX_US_ASCII);
1590 st_insert(arg->symbols, (st_data_t)n, (st_data_t)s);
1591 if (ivar) {
1592 long num = r_long(arg);
1593 while (num-- > 0) {
1594 sym = r_symbol(arg);
1595 idx = sym2encidx(sym, r_object(arg));
1596 }
1597 }
1598 if (idx > 0) {
1599 rb_enc_associate_index(s, idx);
1600 if (is_broken_string(s)) {
1601 rb_raise(rb_eArgError, "invalid byte sequence in %s: %+"PRIsVALUE,
1602 rb_enc_name(rb_enc_from_index(idx)), s);
1603 }
1604 }
1605
1606 return s;
1607}
1608
1609static VALUE
1610r_symbol(struct load_arg *arg)
1611{
1612 int type, ivar = 0;
1613
1614 again:
1615 switch ((type = r_byte(arg))) {
1616 default:
1617 rb_raise(rb_eArgError, "dump format error for symbol(0x%x)", type);
1618 case TYPE_IVAR:
1619 ivar = 1;
1620 goto again;
1621 case TYPE_SYMBOL:
1622 return r_symreal(arg, ivar);
1623 case TYPE_SYMLINK:
1624 if (ivar) {
1625 rb_raise(rb_eArgError, "dump format error (symlink with encoding)");
1626 }
1627 return r_symlink(arg);
1628 }
1629}
1630
1631static VALUE
1632r_unique(struct load_arg *arg)
1633{
1634 return r_symbol(arg);
1635}
1636
1637static VALUE
1638r_string(struct load_arg *arg)
1639{
1640 return r_bytes(arg);
1641}
1642
1643static VALUE
1644r_entry0(VALUE v, st_index_t num, struct load_arg *arg)
1645{
1646 st_data_t real_obj = (st_data_t)v;
1647 if (arg->compat_tbl) {
1648 /* real_obj is kept if not found */
1649 st_lookup(arg->compat_tbl, v, &real_obj);
1650 }
1651 st_insert(arg->data, num, real_obj);
1652 st_insert(arg->partial_objects, (st_data_t)real_obj, Qtrue);
1653 return v;
1654}
1655
1656static VALUE
1657r_fixup_compat(VALUE v, struct load_arg *arg)
1658{
1659 st_data_t data;
1660 st_data_t key = (st_data_t)v;
1661 if (arg->compat_tbl && st_delete(arg->compat_tbl, &key, &data)) {
1662 VALUE real_obj = (VALUE)data;
1663 rb_alloc_func_t allocator = rb_get_alloc_func(CLASS_OF(real_obj));
1664 if (st_lookup(compat_allocator_tbl, (st_data_t)allocator, &data)) {
1665 marshal_compat_t *compat = (marshal_compat_t*)data;
1666 compat->loader(real_obj, v);
1667 }
1668 v = real_obj;
1669 }
1670 return v;
1671}
1672
1673static VALUE
1674r_post_proc(VALUE v, struct load_arg *arg)
1675{
1676 if (arg->proc) {
1677 v = load_funcall(arg, arg->proc, s_call, 1, &v);
1678 }
1679 return v;
1680}
1681
1682static VALUE
1683r_leave(VALUE v, struct load_arg *arg, bool partial)
1684{
1685 v = r_fixup_compat(v, arg);
1686 if (!partial) {
1687 st_data_t data;
1688 st_data_t key = (st_data_t)v;
1689 st_delete(arg->partial_objects, &key, &data);
1690 if (arg->freeze) {
1691 if (RB_TYPE_P(v, T_MODULE) || RB_TYPE_P(v, T_CLASS)) {
1692 // noop
1693 }
1694 else if (RB_TYPE_P(v, T_STRING)) {
1695 v = rb_str_to_interned_str(v);
1696 }
1697 else {
1698 OBJ_FREEZE(v);
1699 }
1700 }
1701 v = r_post_proc(v, arg);
1702 }
1703 return v;
1704}
1705
1706static int
1707copy_ivar_i(ID vid, VALUE value, st_data_t arg)
1708{
1709 VALUE obj = (VALUE)arg;
1710
1711 if (!rb_ivar_defined(obj, vid))
1712 rb_ivar_set(obj, vid, value);
1713 return ST_CONTINUE;
1714}
1715
1716static VALUE
1717r_copy_ivar(VALUE v, VALUE data)
1718{
1719 rb_ivar_foreach(data, copy_ivar_i, (st_data_t)v);
1720 return v;
1721}
1722
1723#define override_ivar_error(type, str) \
1724 rb_raise(rb_eTypeError, \
1725 "can't override instance variable of "type" '%"PRIsVALUE"'", \
1726 (str))
1727
1728static int
1729r_ivar_encoding(VALUE obj, struct load_arg *arg, VALUE sym, VALUE val)
1730{
1731 int idx = sym2encidx(sym, val);
1732 if (idx >= 0) {
1733 if (rb_enc_capable(obj)) {
1734 rb_enc_associate_index(obj, idx);
1735 }
1736 else {
1737 rb_raise(rb_eArgError, "%"PRIsVALUE" is not enc_capable", obj);
1738 }
1739 return TRUE;
1740 }
1741 return FALSE;
1742}
1743
1744static long
1745r_encname(VALUE obj, struct load_arg *arg)
1746{
1747 long len = r_long(arg);
1748 if (len > 0) {
1749 VALUE sym = r_symbol(arg);
1750 VALUE val = r_object(arg);
1751 len -= r_ivar_encoding(obj, arg, sym, val);
1752 }
1753 return len;
1754}
1755
1756static void
1757r_ivar(VALUE obj, int *has_encoding, struct load_arg *arg)
1758{
1759 long len;
1760
1761 len = r_long(arg);
1762 if (len > 0) {
1763 if (RB_TYPE_P(obj, T_MODULE)) {
1764 override_ivar_error("module", rb_mod_name(obj));
1765 }
1766 else if (RB_TYPE_P(obj, T_CLASS)) {
1767 override_ivar_error("class", rb_class_name(obj));
1768 }
1769 do {
1770 VALUE sym = r_symbol(arg);
1771 VALUE val = r_object(arg);
1772 if (r_ivar_encoding(obj, arg, sym, val)) {
1773 if (has_encoding) *has_encoding = TRUE;
1774 }
1775 else if (symname_equal_lit(sym, name_s_ruby2_keywords_flag)) {
1776 if (RB_TYPE_P(obj, T_HASH)) {
1777 rb_hash_ruby2_keywords(obj);
1778 }
1779 else {
1780 rb_raise(rb_eArgError, "ruby2_keywords flag is given but %"PRIsVALUE" is not a Hash", obj);
1781 }
1782 }
1783 else {
1784 rb_ivar_set(obj, rb_intern_str(sym), val);
1785 }
1786 } while (--len > 0);
1787 }
1788}
1789
1790static VALUE
1791path2class(VALUE path)
1792{
1793 VALUE v = rb_path_to_class(path);
1794
1795 if (!RB_TYPE_P(v, T_CLASS)) {
1796 rb_raise(rb_eArgError, "%"PRIsVALUE" does not refer to class", path);
1797 }
1798 return v;
1799}
1800
1801#define path2module(path) must_be_module(rb_path_to_class(path), path)
1802
1803static VALUE
1804must_be_module(VALUE v, VALUE path)
1805{
1806 if (!RB_TYPE_P(v, T_MODULE)) {
1807 rb_raise(rb_eArgError, "%"PRIsVALUE" does not refer to module", path);
1808 }
1809 return v;
1810}
1811
1812static VALUE
1813obj_alloc_by_klass(VALUE klass, struct load_arg *arg, VALUE *oldclass)
1814{
1815 st_data_t data;
1816 rb_alloc_func_t allocator;
1817
1818 allocator = rb_get_alloc_func(klass);
1819 if (st_lookup(compat_allocator_tbl, (st_data_t)allocator, &data)) {
1820 marshal_compat_t *compat = (marshal_compat_t*)data;
1821 VALUE real_obj = rb_obj_alloc(klass);
1822 VALUE obj = rb_obj_alloc(compat->oldclass);
1823 if (oldclass) *oldclass = compat->oldclass;
1824
1825 if (!arg->compat_tbl) {
1826 arg->compat_tbl = rb_init_identtable();
1827 }
1828 st_insert(arg->compat_tbl, (st_data_t)obj, (st_data_t)real_obj);
1829 return obj;
1830 }
1831
1832 return rb_obj_alloc(klass);
1833}
1834
1835static VALUE
1836obj_alloc_by_path(VALUE path, struct load_arg *arg)
1837{
1838 return obj_alloc_by_klass(path2class(path), arg, 0);
1839}
1840
1841static VALUE
1842append_extmod(VALUE obj, VALUE extmod)
1843{
1844 long i = RARRAY_LEN(extmod);
1845 while (i > 0) {
1846 VALUE m = RARRAY_AREF(extmod, --i);
1847 rb_extend_object(obj, m);
1848 }
1849 return obj;
1850}
1851
1852#define prohibit_ivar(type, str) do { \
1853 if (!ivp || !*ivp) break; \
1854 override_ivar_error(type, str); \
1855 } while (0)
1856
1857static VALUE r_object_for(struct load_arg *arg, bool partial, int *ivp, VALUE extmod, int type);
1858
1859static VALUE
1860r_object0(struct load_arg *arg, bool partial, int *ivp, VALUE extmod)
1861{
1862 int type = r_byte(arg);
1863 return r_object_for(arg, partial, ivp, extmod, type);
1864}
1865
1866static VALUE
1867r_object_for(struct load_arg *arg, bool partial, int *ivp, VALUE extmod, int type)
1868{
1869 VALUE (*hash_new_with_size)(st_index_t) = rb_hash_new_with_size;
1870 VALUE v = Qnil;
1871 long id;
1872 st_data_t link;
1873
1874 switch (type) {
1875 case TYPE_LINK:
1876 id = r_long(arg);
1877 if (!st_lookup(arg->data, (st_data_t)id, &link)) {
1878 rb_raise(rb_eArgError, "dump format error (unlinked)");
1879 }
1880 v = (VALUE)link;
1881 if (!st_lookup(arg->partial_objects, (st_data_t)v, &link)) {
1882 v = r_post_proc(v, arg);
1883 }
1884 break;
1885
1886 case TYPE_IVAR:
1887 {
1888 int ivar = TRUE;
1889 v = r_object0(arg, true, &ivar, extmod);
1890 if (ivar) r_ivar(v, NULL, arg);
1891 v = r_leave(v, arg, partial);
1892 }
1893 break;
1894
1895 case TYPE_EXTENDED:
1896 {
1897 VALUE path = r_unique(arg);
1898 VALUE m = rb_path_to_class(path);
1899 if (NIL_P(extmod)) extmod = rb_ary_hidden_new(0);
1900
1901 if (RB_TYPE_P(m, T_CLASS)) { /* prepended */
1902 VALUE c;
1903
1904 v = r_object0(arg, true, 0, Qnil);
1905 c = CLASS_OF(v);
1906 if (c != m || FL_TEST(c, FL_SINGLETON)) {
1907 rb_raise(rb_eArgError,
1908 "prepended class %"PRIsVALUE" differs from class %"PRIsVALUE,
1909 path, rb_class_name(c));
1910 }
1911 c = rb_singleton_class(v);
1912 while (RARRAY_LEN(extmod) > 0) {
1913 m = rb_ary_pop(extmod);
1914 rb_prepend_module(c, m);
1915 }
1916 }
1917 else {
1918 must_be_module(m, path);
1919 rb_ary_push(extmod, m);
1920
1921 v = r_object0(arg, true, 0, extmod);
1922 while (RARRAY_LEN(extmod) > 0) {
1923 m = rb_ary_pop(extmod);
1924 rb_extend_object(v, m);
1925 }
1926 }
1927 v = r_leave(v, arg, partial);
1928 }
1929 break;
1930
1931 case TYPE_UCLASS:
1932 {
1933 VALUE c = path2class(r_unique(arg));
1934
1935 if (FL_TEST(c, FL_SINGLETON)) {
1936 rb_raise(rb_eTypeError, "singleton can't be loaded");
1937 }
1938 type = r_byte(arg);
1939 if ((c == rb_cHash) &&
1940 /* Hack for compare_by_identify */
1941 (type == TYPE_HASH || type == TYPE_HASH_DEF)) {
1942 hash_new_with_size = rb_ident_hash_new_with_size;
1943 goto type_hash;
1944 }
1945 v = r_object_for(arg, partial, 0, extmod, type);
1946 if (RB_SPECIAL_CONST_P(v) || RB_TYPE_P(v, T_OBJECT) || RB_TYPE_P(v, T_CLASS)) {
1947 goto format_error;
1948 }
1949 if (RB_TYPE_P(v, T_MODULE) || !RTEST(rb_class_inherited_p(c, RBASIC(v)->klass))) {
1950 VALUE tmp = rb_obj_alloc(c);
1951
1952 if (TYPE(v) != TYPE(tmp)) goto format_error;
1953 }
1954 RBASIC_SET_CLASS(v, c);
1955 }
1956 break;
1957
1958 format_error:
1959 rb_raise(rb_eArgError, "dump format error (user class)");
1960
1961 case TYPE_NIL:
1962 v = Qnil;
1963 v = r_leave(v, arg, false);
1964 break;
1965
1966 case TYPE_TRUE:
1967 v = Qtrue;
1968 v = r_leave(v, arg, false);
1969 break;
1970
1971 case TYPE_FALSE:
1972 v = Qfalse;
1973 v = r_leave(v, arg, false);
1974 break;
1975
1976 case TYPE_FIXNUM:
1977 {
1978 long i = r_long(arg);
1979 v = LONG2FIX(i);
1980 }
1981 v = r_leave(v, arg, false);
1982 break;
1983
1984 case TYPE_FLOAT:
1985 {
1986 double d;
1987 VALUE str = r_bytes(arg);
1988 const char *ptr = RSTRING_PTR(str);
1989
1990 if (strcmp(ptr, "nan") == 0) {
1991 d = nan("");
1992 }
1993 else if (strcmp(ptr, "inf") == 0) {
1994 d = HUGE_VAL;
1995 }
1996 else if (strcmp(ptr, "-inf") == 0) {
1997 d = -HUGE_VAL;
1998 }
1999 else {
2000 char *e;
2001 d = strtod(ptr, &e);
2002 d = load_mantissa(d, e, RSTRING_LEN(str) - (e - ptr));
2003 }
2004 v = DBL2NUM(d);
2005 v = r_entry(v, arg);
2006 v = r_leave(v, arg, false);
2007 }
2008 break;
2009
2010 case TYPE_BIGNUM:
2011 {
2012 long len;
2013 VALUE data;
2014 int sign;
2015
2016 sign = r_byte(arg);
2017 len = r_long(arg);
2018
2019 if (SIZEOF_VALUE >= 8 && len <= 4) {
2020 // Representable within uintptr, likely FIXNUM
2021 VALUE num = 0;
2022 for (int i = 0; i < len; i++) {
2023 num |= (VALUE)r_byte(arg) << (i * 16);
2024 num |= (VALUE)r_byte(arg) << (i * 16 + 8);
2025 }
2026#if SIZEOF_VALUE == SIZEOF_LONG
2027 v = ULONG2NUM(num);
2028#else
2029 v = ULL2NUM(num);
2030#endif
2031 if (sign == '-') {
2032 v = rb_int_uminus(v);
2033 }
2034 }
2035 else {
2036 data = r_bytes0(len * 2, arg);
2037 v = rb_integer_unpack(RSTRING_PTR(data), len, 2, 0,
2038 INTEGER_PACK_LITTLE_ENDIAN | (sign == '-' ? INTEGER_PACK_NEGATIVE : 0));
2039 rb_str_resize(data, 0L);
2040 }
2041 v = r_entry(v, arg);
2042 v = r_leave(v, arg, false);
2043 }
2044 break;
2045
2046 case TYPE_STRING:
2047 v = r_entry(r_string(arg), arg);
2048 v = r_leave(v, arg, partial);
2049 break;
2050
2051 case TYPE_REGEXP:
2052 {
2053 VALUE str = r_bytes(arg);
2054 int options = r_byte(arg);
2055 int has_encoding = FALSE;
2056 st_index_t idx = r_prepare(arg);
2057
2058 if (ivp) {
2059 r_ivar(str, &has_encoding, arg);
2060 *ivp = FALSE;
2061 }
2062 if (!has_encoding) {
2063 /* 1.8 compatibility; remove escapes undefined in 1.8 */
2064 char *ptr = RSTRING_PTR(str), *dst = ptr, *src = ptr;
2065 long len = RSTRING_LEN(str);
2066 long bs = 0;
2067 for (; len-- > 0; *dst++ = *src++) {
2068 switch (*src) {
2069 case '\\': bs++; break;
2070 case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
2071 case 'm': case 'o': case 'p': case 'q': case 'u': case 'y':
2072 case 'E': case 'F': case 'H': case 'I': case 'J': case 'K':
2073 case 'L': case 'N': case 'O': case 'P': case 'Q': case 'R':
2074 case 'S': case 'T': case 'U': case 'V': case 'X': case 'Y':
2075 if (bs & 1) --dst;
2076 /* fall through */
2077 default: bs = 0; break;
2078 }
2079 }
2080 rb_str_set_len(str, dst - ptr);
2081 }
2082 VALUE regexp = rb_reg_new_str(str, options);
2083 r_copy_ivar(regexp, str);
2084
2085 v = r_entry0(regexp, idx, arg);
2086 v = r_leave(v, arg, partial);
2087 }
2088 break;
2089
2090 case TYPE_ARRAY:
2091 {
2092 long len = r_long(arg);
2093
2094 v = rb_ary_new2(len);
2095 v = r_entry(v, arg);
2096 arg->readable += len - 1;
2097 while (len--) {
2098 rb_ary_push(v, r_object(arg));
2099 arg->readable--;
2100 }
2101 v = r_leave(v, arg, partial);
2102 arg->readable++;
2103 }
2104 break;
2105
2106 case TYPE_HASH:
2107 case TYPE_HASH_DEF:
2108 type_hash:
2109 {
2110 long len = r_long(arg);
2111
2112 v = hash_new_with_size(len);
2113 v = r_entry(v, arg);
2114 arg->readable += (len - 1) * 2;
2115 while (len--) {
2116 VALUE key = r_object(arg);
2117 VALUE value = r_object(arg);
2118 rb_hash_aset(v, key, value);
2119 arg->readable -= 2;
2120 }
2121 arg->readable += 2;
2122 if (type == TYPE_HASH_DEF) {
2123 RHASH_SET_IFNONE(v, r_object(arg));
2124 }
2125 v = r_leave(v, arg, partial);
2126 }
2127 break;
2128
2129 case TYPE_STRUCT:
2130 {
2131 VALUE mem, values;
2132 long i;
2133 VALUE slot;
2134 st_index_t idx = r_prepare(arg);
2135 VALUE klass = path2class(r_unique(arg));
2136 long len = r_long(arg);
2137
2138 v = rb_obj_alloc(klass);
2139 if (!RB_TYPE_P(v, T_STRUCT)) {
2140 rb_raise(rb_eTypeError, "class %"PRIsVALUE" not a struct", rb_class_name(klass));
2141 }
2142 mem = rb_struct_s_members(klass);
2143 if (RARRAY_LEN(mem) != len) {
2144 rb_raise(rb_eTypeError, "struct %"PRIsVALUE" not compatible (struct size differs)",
2145 rb_class_name(klass));
2146 }
2147
2148 arg->readable += (len - 1) * 2;
2149 v = r_entry0(v, idx, arg);
2150 values = rb_ary_new2(len);
2151 {
2152 VALUE keywords = Qfalse;
2153 if (RTEST(rb_struct_s_keyword_init(klass))) {
2154 keywords = rb_hash_new();
2155 rb_ary_push(values, keywords);
2156 }
2157
2158 for (i=0; i<len; i++) {
2159 VALUE n = rb_sym2str(RARRAY_AREF(mem, i));
2160 slot = r_symbol(arg);
2161
2162 if (!rb_str_equal(n, slot)) {
2163 rb_raise(rb_eTypeError, "struct %"PRIsVALUE" not compatible (:%"PRIsVALUE" for :%"PRIsVALUE")",
2164 rb_class_name(klass),
2165 slot, n);
2166 }
2167 if (keywords) {
2168 rb_hash_aset(keywords, RARRAY_AREF(mem, i), r_object(arg));
2169 }
2170 else {
2171 rb_ary_push(values, r_object(arg));
2172 }
2173 arg->readable -= 2;
2174 }
2175 }
2176 rb_struct_initialize(v, values);
2177 v = r_leave(v, arg, partial);
2178 arg->readable += 2;
2179 }
2180 break;
2181
2182 case TYPE_USERDEF:
2183 {
2184 VALUE name = r_unique(arg);
2185 VALUE klass = path2class(name);
2186 VALUE data;
2187 st_data_t d;
2188
2189 if (!rb_obj_respond_to(klass, s_load, TRUE)) {
2190 rb_raise(rb_eTypeError, "class %"PRIsVALUE" needs to have method '_load'",
2191 name);
2192 }
2193 data = r_string(arg);
2194 if (ivp) {
2195 r_ivar(data, NULL, arg);
2196 *ivp = FALSE;
2197 }
2198 v = load_funcall(arg, klass, s_load, 1, &data);
2199 v = r_entry(v, arg);
2200 if (st_lookup(compat_allocator_tbl, (st_data_t)rb_get_alloc_func(klass), &d)) {
2201 marshal_compat_t *compat = (marshal_compat_t*)d;
2202 v = compat->loader(klass, v);
2203 }
2204 if (!partial) {
2205 if (arg->freeze) {
2206 OBJ_FREEZE(v);
2207 }
2208 v = r_post_proc(v, arg);
2209 }
2210 }
2211 break;
2212
2213 case TYPE_USRMARSHAL:
2214 {
2215 VALUE name = r_unique(arg);
2216 VALUE klass = path2class(name);
2217 VALUE oldclass = 0;
2218 VALUE data;
2219
2220 v = obj_alloc_by_klass(klass, arg, &oldclass);
2221 if (!NIL_P(extmod)) {
2222 /* for the case marshal_load is overridden */
2223 append_extmod(v, extmod);
2224 }
2225 if (!rb_obj_respond_to(v, s_mload, TRUE)) {
2226 rb_raise(rb_eTypeError, "instance of %"PRIsVALUE" needs to have method 'marshal_load'",
2227 name);
2228 }
2229 v = r_entry(v, arg);
2230 data = r_object(arg);
2231 load_funcall(arg, v, s_mload, 1, &data);
2232 v = r_fixup_compat(v, arg);
2233 v = r_copy_ivar(v, data);
2234 if (arg->freeze) {
2235 OBJ_FREEZE(v);
2236 }
2237 v = r_post_proc(v, arg);
2238 if (!NIL_P(extmod)) {
2239 if (oldclass) append_extmod(v, extmod);
2240 rb_ary_clear(extmod);
2241 }
2242 }
2243 break;
2244
2245 case TYPE_OBJECT:
2246 {
2247 st_index_t idx = r_prepare(arg);
2248 v = obj_alloc_by_path(r_unique(arg), arg);
2249 if (!RB_TYPE_P(v, T_OBJECT)) {
2250 rb_raise(rb_eArgError, "dump format error");
2251 }
2252 v = r_entry0(v, idx, arg);
2253 r_ivar(v, NULL, arg);
2254 v = r_leave(v, arg, partial);
2255 }
2256 break;
2257
2258 case TYPE_DATA:
2259 {
2260 VALUE name = r_unique(arg);
2261 VALUE klass = path2class(name);
2262 VALUE oldclass = 0;
2263 VALUE r;
2264
2265 v = obj_alloc_by_klass(klass, arg, &oldclass);
2266 if (!RB_TYPE_P(v, T_DATA)) {
2267 rb_raise(rb_eArgError, "dump format error");
2268 }
2269 v = r_entry(v, arg);
2270 if (!rb_obj_respond_to(v, s_load_data, TRUE)) {
2271 rb_raise(rb_eTypeError,
2272 "class %"PRIsVALUE" needs to have instance method '_load_data'",
2273 name);
2274 }
2275 r = r_object0(arg, partial, 0, extmod);
2276 load_funcall(arg, v, s_load_data, 1, &r);
2277 v = r_leave(v, arg, partial);
2278 }
2279 break;
2280
2281 case TYPE_MODULE_OLD:
2282 {
2283 VALUE str = r_bytes(arg);
2284
2285 v = rb_path_to_class(str);
2286 prohibit_ivar("class/module", str);
2287 v = r_entry(v, arg);
2288 v = r_leave(v, arg, partial);
2289 }
2290 break;
2291
2292 case TYPE_CLASS:
2293 {
2294 VALUE str = r_bytes(arg);
2295
2296 if (ivp && *ivp > 0) *ivp = r_encname(str, arg) > 0;
2297 v = path2class(str);
2298 prohibit_ivar("class", str);
2299 v = r_entry(v, arg);
2300 v = r_leave(v, arg, partial);
2301 }
2302 break;
2303
2304 case TYPE_MODULE:
2305 {
2306 VALUE str = r_bytes(arg);
2307
2308 if (ivp && *ivp > 0) *ivp = r_encname(str, arg) > 0;
2309 v = path2module(str);
2310 prohibit_ivar("module", str);
2311 v = r_entry(v, arg);
2312 v = r_leave(v, arg, partial);
2313 }
2314 break;
2315
2316 case TYPE_SYMBOL:
2317 if (ivp) {
2318 v = r_symreal(arg, *ivp);
2319 *ivp = FALSE;
2320 }
2321 else {
2322 v = r_symreal(arg, 0);
2323 }
2324 v = rb_str_intern(v);
2325 v = r_leave(v, arg, partial);
2326 break;
2327
2328 case TYPE_SYMLINK:
2329 v = rb_str_intern(r_symlink(arg));
2330 break;
2331
2332 default:
2333 rb_raise(rb_eArgError, "dump format error(0x%x)", type);
2334 break;
2335 }
2336
2337 if (UNDEF_P(v)) {
2338 rb_raise(rb_eArgError, "dump format error (bad link)");
2339 }
2340
2341 return v;
2342}
2343
2344static VALUE
2345r_object(struct load_arg *arg)
2346{
2347 return r_object0(arg, false, 0, Qnil);
2348}
2349
2350static void
2351clear_load_arg(struct load_arg *arg)
2352{
2353 xfree(arg->buf);
2354 arg->buf = NULL;
2355 arg->buflen = 0;
2356 arg->offset = 0;
2357 arg->readable = 0;
2358 if (!arg->symbols) return;
2359 st_free_table(arg->symbols);
2360 arg->symbols = 0;
2361 st_free_table(arg->data);
2362 arg->data = 0;
2363 st_free_table(arg->partial_objects);
2364 arg->partial_objects = 0;
2365 if (arg->compat_tbl) {
2366 st_free_table(arg->compat_tbl);
2367 arg->compat_tbl = 0;
2368 }
2369}
2370
2371VALUE
2372rb_marshal_load_with_proc(VALUE port, VALUE proc, bool freeze)
2373{
2374 int major, minor;
2375 VALUE v;
2376 VALUE wrapper; /* used to avoid memory leak in case of exception */
2377 struct load_arg *arg;
2378
2379 v = rb_check_string_type(port);
2380 if (!NIL_P(v)) {
2381 port = v;
2382 }
2383 else if (rb_respond_to(port, s_getbyte) && rb_respond_to(port, s_read)) {
2384 rb_check_funcall(port, s_binmode, 0, 0);
2385 }
2386 else {
2387 io_needed();
2388 }
2389 wrapper = TypedData_Make_Struct(0, struct load_arg, &load_arg_data, arg);
2390 arg->src = port;
2391 arg->offset = 0;
2392 arg->symbols = st_init_numtable();
2393 arg->data = rb_init_identtable();
2394 arg->partial_objects = rb_init_identtable();
2395 arg->compat_tbl = 0;
2396 arg->proc = 0;
2397 arg->readable = 0;
2398 arg->freeze = freeze;
2399
2400 if (NIL_P(v))
2401 arg->buf = xmalloc(BUFSIZ);
2402 else
2403 arg->buf = 0;
2404
2405 major = r_byte(arg);
2406 minor = r_byte(arg);
2407 if (major != MARSHAL_MAJOR || minor > MARSHAL_MINOR) {
2408 clear_load_arg(arg);
2409 rb_raise(rb_eTypeError, "incompatible marshal file format (can't be read)\n\
2410\tformat version %d.%d required; %d.%d given",
2411 MARSHAL_MAJOR, MARSHAL_MINOR, major, minor);
2412 }
2413 if (RTEST(ruby_verbose) && minor != MARSHAL_MINOR) {
2414 rb_warn("incompatible marshal file format (can be read)\n\
2415\tformat version %d.%d required; %d.%d given",
2416 MARSHAL_MAJOR, MARSHAL_MINOR, major, minor);
2417 }
2418
2419 if (!NIL_P(proc)) arg->proc = proc;
2420 v = r_object(arg);
2421 clear_load_arg(arg);
2422 RB_GC_GUARD(wrapper);
2423
2424 return v;
2425}
2426
2427static VALUE
2428marshal_load(rb_execution_context_t *ec, VALUE mod, VALUE source, VALUE proc, VALUE freeze)
2429{
2430 return rb_marshal_load_with_proc(source, proc, RTEST(freeze));
2431}
2432
2433#include "marshal.rbinc"
2434
2435/*
2436 * The marshaling library converts collections of Ruby objects into a
2437 * byte stream, allowing them to be stored outside the currently
2438 * active script. This data may subsequently be read and the original
2439 * objects reconstituted.
2440 *
2441 * Marshaled data has major and minor version numbers stored along
2442 * with the object information. In normal use, marshaling can only
2443 * load data written with the same major version number and an equal
2444 * or lower minor version number. If Ruby's ``verbose'' flag is set
2445 * (normally using -d, -v, -w, or --verbose) the major and minor
2446 * numbers must match exactly. Marshal versioning is independent of
2447 * Ruby's version numbers. You can extract the version by reading the
2448 * first two bytes of marshaled data.
2449 *
2450 * str = Marshal.dump("thing")
2451 * RUBY_VERSION #=> "1.9.0"
2452 * str[0].ord #=> 4
2453 * str[1].ord #=> 8
2454 *
2455 * Some objects cannot be dumped: if the objects to be dumped include
2456 * bindings, procedure or method objects, instances of class IO, or
2457 * singleton objects, a TypeError will be raised.
2458 *
2459 * If your class has special serialization needs (for example, if you
2460 * want to serialize in some specific format), or if it contains
2461 * objects that would otherwise not be serializable, you can implement
2462 * your own serialization strategy.
2463 *
2464 * There are two methods of doing this, your object can define either
2465 * marshal_dump and marshal_load or _dump and _load. marshal_dump will take
2466 * precedence over _dump if both are defined. marshal_dump may result in
2467 * smaller Marshal strings.
2468 *
2469 * == Security considerations
2470 *
2471 * By design, Marshal.load can deserialize almost any class loaded into the
2472 * Ruby process. In many cases this can lead to remote code execution if the
2473 * Marshal data is loaded from an untrusted source.
2474 *
2475 * As a result, Marshal.load is not suitable as a general purpose serialization
2476 * format and you should never unmarshal user supplied input or other untrusted
2477 * data.
2478 *
2479 * If you need to deserialize untrusted data, use JSON or another serialization
2480 * format that is only able to load simple, 'primitive' types such as String,
2481 * Array, Hash, etc. Never allow user input to specify arbitrary types to
2482 * deserialize into.
2483 *
2484 * == marshal_dump and marshal_load
2485 *
2486 * When dumping an object the method marshal_dump will be called.
2487 * marshal_dump must return a result containing the information necessary for
2488 * marshal_load to reconstitute the object. The result can be any object.
2489 *
2490 * When loading an object dumped using marshal_dump the object is first
2491 * allocated then marshal_load is called with the result from marshal_dump.
2492 * marshal_load must recreate the object from the information in the result.
2493 *
2494 * Example:
2495 *
2496 * class MyObj
2497 * def initialize name, version, data
2498 * @name = name
2499 * @version = version
2500 * @data = data
2501 * end
2502 *
2503 * def marshal_dump
2504 * [@name, @version]
2505 * end
2506 *
2507 * def marshal_load array
2508 * @name, @version = array
2509 * end
2510 * end
2511 *
2512 * == _dump and _load
2513 *
2514 * Use _dump and _load when you need to allocate the object you're restoring
2515 * yourself.
2516 *
2517 * When dumping an object the instance method _dump is called with an Integer
2518 * which indicates the maximum depth of objects to dump (a value of -1 implies
2519 * that you should disable depth checking). _dump must return a String
2520 * containing the information necessary to reconstitute the object.
2521 *
2522 * The class method _load should take a String and use it to return an object
2523 * of the same class.
2524 *
2525 * Example:
2526 *
2527 * class MyObj
2528 * def initialize name, version, data
2529 * @name = name
2530 * @version = version
2531 * @data = data
2532 * end
2533 *
2534 * def _dump level
2535 * [@name, @version].join ':'
2536 * end
2537 *
2538 * def self._load args
2539 * new(*args.split(':'))
2540 * end
2541 * end
2542 *
2543 * Since Marshal.dump outputs a string you can have _dump return a Marshal
2544 * string which is Marshal.loaded in _load for complex objects.
2545 */
2546void
2547Init_marshal(void)
2548{
2549 VALUE rb_mMarshal = rb_define_module("Marshal");
2550#define set_id(sym) sym = rb_intern_const(name_##sym)
2551 set_id(s_dump);
2552 set_id(s_load);
2553 set_id(s_mdump);
2554 set_id(s_mload);
2555 set_id(s_dump_data);
2556 set_id(s_load_data);
2557 set_id(s_alloc);
2558 set_id(s_call);
2559 set_id(s_getbyte);
2560 set_id(s_read);
2561 set_id(s_write);
2562 set_id(s_binmode);
2563 set_id(s_encoding_short);
2564 set_id(s_ruby2_keywords_flag);
2565
2566 rb_define_module_function(rb_mMarshal, "dump", marshal_dump, -1);
2567
2568 /* major version */
2569 rb_define_const(rb_mMarshal, "MAJOR_VERSION", INT2FIX(MARSHAL_MAJOR));
2570 /* minor version */
2571 rb_define_const(rb_mMarshal, "MINOR_VERSION", INT2FIX(MARSHAL_MINOR));
2572}
2573
2574static int
2575marshal_compat_table_mark_i(st_data_t key, st_data_t value, st_data_t _)
2576{
2577 marshal_compat_t *p = (marshal_compat_t *)value;
2578 rb_gc_mark_movable(p->newclass);
2579 rb_gc_mark_movable(p->oldclass);
2580 return ST_CONTINUE;
2581}
2582
2583static void
2584marshal_compat_table_mark(void *tbl)
2585{
2586 if (!tbl) return;
2587 st_foreach(tbl, marshal_compat_table_mark_i, 0);
2588}
2589
2590static int
2591marshal_compat_table_free_i(st_data_t key, st_data_t value, st_data_t _)
2592{
2593 xfree((marshal_compat_t *)value);
2594 return ST_CONTINUE;
2595}
2596
2597static void
2598marshal_compat_table_free(void *data)
2599{
2600 st_foreach(data, marshal_compat_table_free_i, 0);
2601 st_free_table(data);
2602}
2603
2604static size_t
2605marshal_compat_table_memsize(const void *data)
2606{
2607 return st_memsize(data) + sizeof(marshal_compat_t) * st_table_size(data);
2608}
2609
2610static int
2611marshal_compat_table_compact_i(st_data_t key, st_data_t value, st_data_t _)
2612{
2613 marshal_compat_t *p = (marshal_compat_t *)value;
2614 p->newclass = rb_gc_location(p->newclass);
2615 p->oldclass = rb_gc_location(p->oldclass);
2616 return ST_CONTINUE;
2617}
2618
2619static void
2620marshal_compat_table_compact(void *tbl)
2621{
2622 if (!tbl) return;
2623 st_foreach(tbl, marshal_compat_table_compact_i, 0);
2624}
2625
2626static const rb_data_type_t marshal_compat_type = {
2627 .wrap_struct_name = "marshal_compat_table",
2628 .function = {
2629 .dmark = marshal_compat_table_mark,
2630 .dfree = marshal_compat_table_free,
2631 .dsize = marshal_compat_table_memsize,
2632 .dcompact = marshal_compat_table_compact,
2633 },
2634 .flags = RUBY_TYPED_WB_PROTECTED | RUBY_TYPED_FREE_IMMEDIATELY,
2635};
2636
2637static st_table *
2638compat_allocator_table(void)
2639{
2640 if (compat_allocator_tbl) return compat_allocator_tbl;
2641 compat_allocator_tbl = st_init_numtable();
2642 compat_allocator_tbl_wrapper =
2643 TypedData_Wrap_Struct(0, &marshal_compat_type, compat_allocator_tbl);
2644 rb_vm_register_global_object(compat_allocator_tbl_wrapper);
2645 return compat_allocator_tbl;
2646}
2647
2648VALUE
2649rb_marshal_dump(VALUE obj, VALUE port)
2650{
2651 return rb_marshal_dump_limited(obj, port, -1);
2652}
2653
2654VALUE
2655rb_marshal_load(VALUE port)
2656{
2657 return rb_marshal_load_with_proc(port, Qnil, false);
2658}
builtin.h
Defines RBIMPL_HAS_BUILTIN.
len
int len
Length of the buffer.
Definition io.h:8
nonstring.h
Defines RBIMPL_ATTR_NONSTRING.
Generated by doxygen 1.9.8