Ruby 4.0.6p0 (2026-07-14 revision 03b6d3f8898a28604fe6cb00eae3226b821168f4)
marshal.c
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 struct w_ivar_arg ivarg = {arg, num};
731 if (!num) return;
732 rb_ivar_foreach_buffered(obj, w_obj_each, (st_data_t)&ivarg);
733}
734
735static void
736w_ivar(st_index_t num, VALUE ivobj, VALUE encname, struct dump_call_arg *arg)
737{
738 w_long(num, arg->arg);
739 num -= w_encoding(encname, arg);
740 if (RB_TYPE_P(ivobj, T_HASH) && rb_hash_ruby2_keywords_p(ivobj)) {
741 int limit = arg->limit;
742 if (limit >= 0) ++limit;
743 w_symbol(ID2SYM(s_ruby2_keywords_flag), arg->arg);
744 w_object(Qtrue, arg->arg, limit);
745 num--;
746 }
747 if (!UNDEF_P(ivobj) && num) {
748 w_ivar_each(ivobj, num, arg);
749 }
750}
751
752static void
753w_objivar(VALUE obj, struct dump_call_arg *arg)
754{
755 st_data_t num = 0;
756
757 rb_ivar_foreach(obj, obj_count_ivars, (st_data_t)&num);
758 w_long(num, arg->arg);
759 w_ivar_each(obj, num, arg);
760}
761
762#if SIZEOF_LONG > 4
763// Optimized dump for fixnum larger than 31-bits
764static void
765w_bigfixnum(VALUE obj, struct dump_arg *arg)
766{
767 RUBY_ASSERT(FIXNUM_P(obj));
768
769 w_byte(TYPE_BIGNUM, arg);
770
771#if SIZEOF_LONG == SIZEOF_VALUE
772 long num, slen_num;
773 num = FIX2LONG(obj);
774#else
775 long long num, slen_num;
776 num = NUM2LL(obj);
777#endif
778
779 char sign = num < 0 ? '-' : '+';
780 w_byte(sign, arg);
781
782 // Guaranteed not to overflow, as FIXNUM is 1-bit less than long
783 if (num < 0) num = -num;
784
785 // calculate the size in shorts
786 int slen = 0;
787 {
788 slen_num = num;
789 while (slen_num) {
790 slen++;
791 slen_num = SHORTDN(slen_num);
792 }
793 }
794
795 RUBY_ASSERT(slen > 0 && slen <= SIZEOF_LONG / 2);
796
797 w_long((long)slen, arg);
798
799 for (int i = 0; i < slen; i++) {
800 w_short(num & SHORTMASK, arg);
801 num = SHORTDN(num);
802 }
803
804 // We aren't adding this object to the link table, but we need to increment
805 // the index.
806 arg->num_entries++;
807
808 RUBY_ASSERT(num == 0);
809}
810#endif
811
812static void
813w_remember(VALUE obj, struct dump_arg *arg)
814{
815 st_add_direct(arg->data, obj, arg->num_entries++);
816}
817
818static void
819w_object(VALUE obj, struct dump_arg *arg, int limit)
820{
821 struct dump_call_arg c_arg;
822 VALUE ivobj = Qundef;
823 st_data_t num;
824 st_index_t hasiv = 0;
825 VALUE encname = Qnil;
826
827 if (limit == 0) {
828 rb_raise(rb_eArgError, "exceed depth limit");
829 }
830
831 if (NIL_P(obj)) {
832 w_byte(TYPE_NIL, arg);
833 }
834 else if (obj == Qtrue) {
835 w_byte(TYPE_TRUE, arg);
836 }
837 else if (obj == Qfalse) {
838 w_byte(TYPE_FALSE, arg);
839 }
840 else if (FIXNUM_P(obj)) {
841#if SIZEOF_LONG <= 4
842 w_byte(TYPE_FIXNUM, arg);
843 w_long(FIX2INT(obj), arg);
844#else
845 if (RSHIFT((long)obj, 31) == 0 || RSHIFT((long)obj, 31) == -1) {
846 w_byte(TYPE_FIXNUM, arg);
847 w_long(FIX2LONG(obj), arg);
848 }
849 else {
850 w_bigfixnum(obj, arg);
851 }
852#endif
853 }
854 else if (SYMBOL_P(obj)) {
855 w_symbol(obj, arg);
856 }
857 else {
858 if (st_lookup(arg->data, obj, &num)) {
859 w_byte(TYPE_LINK, arg);
860 w_long((long)num, arg);
861 return;
862 }
863
864 if (limit > 0) limit--;
865 c_arg.limit = limit;
866 c_arg.arg = arg;
867 c_arg.obj = obj;
868
869 if (FLONUM_P(obj)) {
870 w_remember(obj, arg);
871 w_byte(TYPE_FLOAT, arg);
872 w_float(RFLOAT_VALUE(obj), arg);
873 return;
874 }
875
876 VALUE v;
877
878 if (!RBASIC_CLASS(obj)) {
879 rb_raise(rb_eTypeError, "can't dump internal %s",
880 rb_builtin_type_name(BUILTIN_TYPE(obj)));
881 }
882
883 if (rb_obj_respond_to(obj, s_mdump, TRUE)) {
884 w_remember(obj, arg);
885
886 v = dump_funcall(arg, obj, s_mdump, 0, 0);
887 w_class(TYPE_USRMARSHAL, obj, arg, FALSE);
888 w_object(v, arg, limit);
889 return;
890 }
891 if (rb_obj_respond_to(obj, s_dump, TRUE)) {
892 VALUE ivobj2 = Qundef;
893 st_index_t hasiv2;
894 VALUE encname2;
895
896 if (arg->userdefs && st_is_member(arg->userdefs, (st_data_t)obj)) {
897 rb_raise(rb_eRuntimeError, "can't dump recursive object using _dump()");
898 }
899 v = INT2NUM(limit);
900 v = dump_funcall(arg, obj, s_dump, 1, &v);
901 if (!RB_TYPE_P(v, T_STRING)) {
902 rb_raise(rb_eTypeError, "_dump() must return string");
903 }
904 hasiv = has_ivars(obj, (encname = encoding_name(obj, arg)), &ivobj);
905 hasiv2 = has_ivars(v, (encname2 = encoding_name(v, arg)), &ivobj2);
906 if (hasiv2) {
907 hasiv = hasiv2;
908 ivobj = ivobj2;
909 encname = encname2;
910 }
911 if (hasiv) w_byte(TYPE_IVAR, arg);
912 w_class(TYPE_USERDEF, obj, arg, FALSE);
913 w_bytes(RSTRING_PTR(v), RSTRING_LEN(v), arg);
914 if (hasiv) {
915 st_data_t userdefs = (st_data_t)obj;
916 if (!arg->userdefs) {
917 arg->userdefs = rb_init_identtable();
918 }
919 st_add_direct(arg->userdefs, userdefs, 0);
920 w_ivar(hasiv, ivobj, encname, &c_arg);
921 st_delete(arg->userdefs, &userdefs, NULL);
922 }
923 w_remember(obj, arg);
924 return;
925 }
926
927 w_remember(obj, arg);
928
929 hasiv = has_ivars(obj, (encname = encoding_name(obj, arg)), &ivobj);
930 {
931 st_data_t compat_data;
932 rb_alloc_func_t allocator = rb_get_alloc_func(RBASIC(obj)->klass);
933 if (st_lookup(compat_allocator_tbl,
934 (st_data_t)allocator,
935 &compat_data)) {
936 marshal_compat_t *compat = (marshal_compat_t*)compat_data;
937 VALUE real_obj = obj;
938 obj = compat->dumper(real_obj);
939 if (!arg->compat_tbl) {
940 arg->compat_tbl = rb_init_identtable();
941 }
942 st_insert(arg->compat_tbl, (st_data_t)obj, (st_data_t)real_obj);
943 if (obj != real_obj && UNDEF_P(ivobj)) hasiv = 0;
944 }
945 }
946 if (hasiv) w_byte(TYPE_IVAR, arg);
947
948 switch (BUILTIN_TYPE(obj)) {
949 case T_CLASS:
950 if (FL_TEST(obj, FL_SINGLETON)) {
951 rb_raise(rb_eTypeError, "singleton class can't be dumped");
952 }
953 {
954 VALUE path = class2path(obj);
955 VALUE encname = w_encivar(path, arg);
956 w_byte(TYPE_CLASS, arg);
957 w_bytes(RSTRING_PTR(path), RSTRING_LEN(path), arg);
958 w_encname(encname, arg);
959 RB_GC_GUARD(path);
960 }
961 break;
962
963 case T_MODULE:
964 {
965 VALUE path = class2path(obj);
966 VALUE encname = w_encivar(path, arg);
967 w_byte(TYPE_MODULE, arg);
968 w_bytes(RSTRING_PTR(path), RSTRING_LEN(path), arg);
969 w_encname(encname, arg);
970 RB_GC_GUARD(path);
971 }
972 break;
973
974 case T_FLOAT:
975 w_byte(TYPE_FLOAT, arg);
976 w_float(RFLOAT_VALUE(obj), arg);
977 break;
978
979 case T_BIGNUM:
980 w_byte(TYPE_BIGNUM, arg);
981 {
982 char sign = BIGNUM_SIGN(obj) ? '+' : '-';
983 size_t len = BIGNUM_LEN(obj);
984 size_t slen;
985 size_t j;
986 BDIGIT *d = BIGNUM_DIGITS(obj);
987
988 slen = SHORTLEN(len);
989 if (LONG_MAX < slen) {
990 rb_raise(rb_eTypeError, "too big Bignum can't be dumped");
991 }
992
993 w_byte(sign, arg);
994 w_long((long)slen, arg);
995 for (j = 0; j < len; j++) {
996#if SIZEOF_BDIGIT > SIZEOF_SHORT
997 BDIGIT num = *d;
998 int i;
999
1000 for (i=0; i<SIZEOF_BDIGIT; i+=SIZEOF_SHORT) {
1001 w_short(num & SHORTMASK, arg);
1002 num = SHORTDN(num);
1003 if (j == len - 1 && num == 0) break;
1004 }
1005#else
1006 w_short(*d, arg);
1007#endif
1008 d++;
1009 }
1010 }
1011 break;
1012
1013 case T_STRING:
1014 w_uclass(obj, rb_cString, arg);
1015 w_byte(TYPE_STRING, arg);
1016 w_bytes(RSTRING_PTR(obj), RSTRING_LEN(obj), arg);
1017 break;
1018
1019 case T_REGEXP:
1020 w_uclass(obj, rb_cRegexp, arg);
1021 w_byte(TYPE_REGEXP, arg);
1022 {
1023 int opts = rb_reg_options(obj);
1024 w_bytes(RREGEXP_SRC_PTR(obj), RREGEXP_SRC_LEN(obj), arg);
1025 w_byte((char)opts, arg);
1026 }
1027 break;
1028
1029 case T_ARRAY:
1030 w_uclass(obj, rb_cArray, arg);
1031 w_byte(TYPE_ARRAY, arg);
1032 {
1033 long i, len = RARRAY_LEN(obj);
1034
1035 w_long(len, arg);
1036 for (i=0; i<RARRAY_LEN(obj); i++) {
1037 w_object(RARRAY_AREF(obj, i), arg, limit);
1038 if (len != RARRAY_LEN(obj)) {
1039 rb_raise(rb_eRuntimeError, "array modified during dump");
1040 }
1041 }
1042 }
1043 break;
1044
1045 case T_HASH:
1046 w_uclass(obj, rb_cHash, arg);
1047 if (rb_hash_compare_by_id_p(obj)) {
1048 w_byte(TYPE_UCLASS, arg);
1049 w_symbol(rb_sym_intern_ascii_cstr("Hash"), arg);
1050 }
1051 if (NIL_P(RHASH_IFNONE(obj))) {
1052 w_byte(TYPE_HASH, arg);
1053 }
1054 else if (FL_TEST(obj, RHASH_PROC_DEFAULT)) {
1055 rb_raise(rb_eTypeError, "can't dump hash with default proc");
1056 }
1057 else {
1058 w_byte(TYPE_HASH_DEF, arg);
1059 }
1060 w_long(rb_hash_size_num(obj), arg);
1061 rb_hash_foreach(obj, hash_each, (st_data_t)&c_arg);
1062 if (!NIL_P(RHASH_IFNONE(obj))) {
1063 w_object(RHASH_IFNONE(obj), arg, limit);
1064 }
1065 break;
1066
1067 case T_STRUCT:
1068 w_class(TYPE_STRUCT, obj, arg, TRUE);
1069 {
1070 long len = RSTRUCT_LEN(obj);
1071 VALUE mem;
1072 long i;
1073
1074 w_long(len, arg);
1075 mem = rb_struct_members(obj);
1076 for (i=0; i<len; i++) {
1077 w_symbol(RARRAY_AREF(mem, i), arg);
1078 w_object(RSTRUCT_GET(obj, i), arg, limit);
1079 }
1080 }
1081 break;
1082
1083 case T_OBJECT:
1084 w_class(TYPE_OBJECT, obj, arg, TRUE);
1085 w_objivar(obj, &c_arg);
1086 break;
1087
1088 case T_DATA:
1089 {
1090 VALUE v;
1091
1092 if (!rb_obj_respond_to(obj, s_dump_data, TRUE)) {
1093 rb_raise(rb_eTypeError,
1094 "no _dump_data is defined for class %"PRIsVALUE,
1095 rb_obj_class(obj));
1096 }
1097 v = dump_funcall(arg, obj, s_dump_data, 0, 0);
1098 w_class(TYPE_DATA, obj, arg, TRUE);
1099 w_object(v, arg, limit);
1100 }
1101 break;
1102
1103 default:
1104 rb_raise(rb_eTypeError, "can't dump %"PRIsVALUE,
1105 rb_obj_class(obj));
1106 break;
1107 }
1108 RB_GC_GUARD(obj);
1109 }
1110 if (hasiv) {
1111 w_ivar(hasiv, ivobj, encname, &c_arg);
1112 }
1113}
1114
1115static void
1116clear_dump_arg(struct dump_arg *arg)
1117{
1118 if (!arg->symbols) return;
1119 st_free_table(arg->symbols);
1120 arg->symbols = 0;
1121 st_free_table(arg->data);
1122 arg->data = 0;
1123 arg->num_entries = 0;
1124 if (arg->compat_tbl) {
1125 st_free_table(arg->compat_tbl);
1126 arg->compat_tbl = 0;
1127 }
1128 if (arg->encodings) {
1129 st_free_table(arg->encodings);
1130 arg->encodings = 0;
1131 }
1132 if (arg->userdefs) {
1133 st_free_table(arg->userdefs);
1134 arg->userdefs = 0;
1135 }
1136}
1137
1138NORETURN(static inline void io_needed(void));
1139static inline void
1140io_needed(void)
1141{
1142 rb_raise(rb_eTypeError, "instance of IO needed");
1143}
1144
1145/*
1146 * call-seq:
1147 * dump( obj [, anIO] , limit=-1 ) -> anIO
1148 *
1149 * Serializes obj and all descendant objects. If anIO is
1150 * specified, the serialized data will be written to it, otherwise the
1151 * data will be returned as a String. If limit is specified, the
1152 * traversal of subobjects will be limited to that depth. If limit is
1153 * negative, no checking of depth will be performed.
1154 *
1155 * class Klass
1156 * def initialize(str)
1157 * @str = str
1158 * end
1159 * def say_hello
1160 * @str
1161 * end
1162 * end
1163 *
1164 * (produces no output)
1165 *
1166 * o = Klass.new("hello\n")
1167 * data = Marshal.dump(o)
1168 * obj = Marshal.load(data)
1169 * obj.say_hello #=> "hello\n"
1170 *
1171 * Marshal can't dump following objects:
1172 * * anonymous Class/Module.
1173 * * objects which are related to system (ex: Dir, File::Stat, IO, File, Socket
1174 * and so on)
1175 * * an instance of MatchData, Method, UnboundMethod, Proc, Thread,
1176 * ThreadGroup, Continuation
1177 * * objects which define singleton methods
1178 */
1179static VALUE
1180marshal_dump(int argc, VALUE *argv, VALUE _)
1181{
1182 VALUE obj, port, a1, a2;
1183 int limit = -1;
1184
1185 port = Qnil;
1186 rb_scan_args(argc, argv, "12", &obj, &a1, &a2);
1187 if (argc == 3) {
1188 if (!NIL_P(a2)) limit = NUM2INT(a2);
1189 if (NIL_P(a1)) io_needed();
1190 port = a1;
1191 }
1192 else if (argc == 2) {
1193 if (FIXNUM_P(a1)) limit = FIX2INT(a1);
1194 else if (NIL_P(a1)) io_needed();
1195 else port = a1;
1196 }
1197 return rb_marshal_dump_limited(obj, port, limit);
1198}
1199
1200VALUE
1201rb_marshal_dump_limited(VALUE obj, VALUE port, int limit)
1202{
1203 struct dump_arg *arg;
1204 VALUE wrapper; /* used to avoid memory leak in case of exception */
1205
1206 wrapper = TypedData_Make_Struct(0, struct dump_arg, &dump_arg_data, arg);
1207 arg->dest = 0;
1208 arg->symbols = st_init_numtable();
1209 arg->data = rb_init_identtable();
1210 arg->num_entries = 0;
1211 arg->compat_tbl = 0;
1212 arg->encodings = 0;
1213 arg->userdefs = 0;
1214 arg->str = rb_str_buf_new(0);
1215 if (!NIL_P(port)) {
1216 if (!rb_respond_to(port, s_write)) {
1217 io_needed();
1218 }
1219 arg->dest = port;
1220 dump_check_funcall(arg, port, s_binmode, 0, 0);
1221 }
1222 else {
1223 port = arg->str;
1224 }
1225
1226 w_byte(MARSHAL_MAJOR, arg);
1227 w_byte(MARSHAL_MINOR, arg);
1228
1229 w_object(obj, arg, limit);
1230 if (arg->dest) {
1231 rb_io_write(arg->dest, arg->str);
1232 rb_str_resize(arg->str, 0);
1233 }
1234 clear_dump_arg(arg);
1235 RB_GC_GUARD(wrapper);
1236
1237 return port;
1238}
1239
1240struct load_arg {
1241 VALUE src;
1242 char *buf;
1243 long bufsize;
1244 long buflen;
1245 long readable;
1246 long offset;
1247 st_table *symbols;
1248 st_table *data;
1249 st_table *partial_objects;
1250 VALUE proc;
1251 st_table *compat_tbl;
1252 bool freeze;
1253};
1254
1255static VALUE
1256check_load_arg(VALUE ret, struct load_arg *arg, const char *name)
1257{
1258 if (!arg->symbols) {
1259 rb_raise(rb_eRuntimeError, "Marshal.load reentered at %s",
1260 name);
1261 }
1262 return ret;
1263}
1264#define load_funcall(arg, obj, sym, argc, argv) \
1265 check_load_arg(rb_funcallv(obj, sym, argc, argv), arg, name_##sym)
1266
1267static void clear_load_arg(struct load_arg *arg);
1268
1269static void
1270mark_load_arg(void *ptr)
1271{
1272 struct load_arg *p = ptr;
1273 if (!p->symbols)
1274 return;
1275 rb_mark_tbl(p->symbols);
1276 rb_mark_tbl(p->data);
1277 rb_mark_tbl(p->partial_objects);
1278 rb_mark_hash(p->compat_tbl);
1279}
1280
1281static void
1282free_load_arg(void *ptr)
1283{
1284 clear_load_arg(ptr);
1285}
1286
1287static size_t
1288memsize_load_arg(const void *ptr)
1289{
1290 const struct load_arg *p = (struct load_arg *)ptr;
1291 size_t memsize = 0;
1292 if (p->symbols) memsize += rb_st_memsize(p->symbols);
1293 if (p->data) memsize += rb_st_memsize(p->data);
1294 if (p->partial_objects) memsize += rb_st_memsize(p->partial_objects);
1295 if (p->compat_tbl) memsize += rb_st_memsize(p->compat_tbl);
1296 return memsize;
1297}
1298
1299static const rb_data_type_t load_arg_data = {
1300 "load_arg",
1301 {mark_load_arg, free_load_arg, memsize_load_arg,},
1302 0, 0, RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_EMBEDDABLE
1303};
1304
1305#define r_entry(v, arg) r_entry0((v), (arg)->data->num_entries, (arg))
1306static VALUE r_object(struct load_arg *arg);
1307static VALUE r_symbol(struct load_arg *arg);
1308
1309NORETURN(static void too_short(void));
1310static void
1311too_short(void)
1312{
1313 rb_raise(rb_eArgError, "marshal data too short");
1314}
1315
1316static st_index_t
1317r_prepare(struct load_arg *arg)
1318{
1319 st_index_t idx = arg->data->num_entries;
1320
1321 st_insert(arg->data, (st_data_t)idx, (st_data_t)Qundef);
1322 return idx;
1323}
1324
1325static unsigned char
1326r_byte1_buffered(struct load_arg *arg)
1327{
1328 if (arg->buflen == 0) {
1329 long readable = arg->readable < arg->bufsize ? arg->readable : arg->bufsize;
1330 long read_len;
1331 VALUE str, n = LONG2NUM(readable);
1332
1333 str = load_funcall(arg, arg->src, s_read, 1, &n);
1334 if (NIL_P(str)) too_short();
1335 StringValue(str);
1336 read_len = RSTRING_LEN(str);
1337 if (UNLIKELY(read_len < readable)) too_short();
1338 if (UNLIKELY(read_len > arg->bufsize)) {
1339 arg->buf = ruby_sized_realloc_n(arg->buf, read_len, 1, arg->bufsize);
1340 arg->bufsize = read_len;
1341 }
1342 memcpy(arg->buf, RSTRING_PTR(str), read_len);
1343 arg->offset = 0;
1344 arg->buflen = read_len;
1345 RB_GC_GUARD(str);
1346 }
1347 arg->buflen--;
1348 return arg->buf[arg->offset++];
1349}
1350
1351static int
1352r_byte(struct load_arg *arg)
1353{
1354 int c;
1355
1356 if (RB_TYPE_P(arg->src, T_STRING)) {
1357 if (RSTRING_LEN(arg->src) > arg->offset) {
1358 c = (unsigned char)RSTRING_PTR(arg->src)[arg->offset++];
1359 }
1360 else {
1361 too_short();
1362 }
1363 }
1364 else {
1365 if (arg->readable >0 || arg->buflen > 0) {
1366 c = r_byte1_buffered(arg);
1367 }
1368 else {
1369 VALUE v = load_funcall(arg, arg->src, s_getbyte, 0, 0);
1370 if (NIL_P(v)) rb_eof_error();
1371 c = (unsigned char)NUM2CHR(v);
1372 }
1373 }
1374 return c;
1375}
1376
1377NORETURN(static void long_toobig(int size));
1378
1379static void
1380long_toobig(int size)
1381{
1382 rb_raise(rb_eTypeError, "long too big for this architecture (size "
1383 STRINGIZE(SIZEOF_LONG)", given %d)", size);
1384}
1385
1386static long
1387r_long(struct load_arg *arg)
1388{
1389 register long x;
1390 int c = (signed char)r_byte(arg);
1391 long i;
1392
1393 if (c == 0) return 0;
1394 if (c > 0) {
1395 if (4 < c && c < 128) {
1396 return c - 5;
1397 }
1398 if (c > (int)sizeof(long)) long_toobig(c);
1399 x = 0;
1400 for (i=0;i<c;i++) {
1401 x |= (long)r_byte(arg) << (8*i);
1402 }
1403 }
1404 else {
1405 if (-129 < c && c < -4) {
1406 return c + 5;
1407 }
1408 c = -c;
1409 if (c > (int)sizeof(long)) long_toobig(c);
1410 x = -1;
1411 for (i=0;i<c;i++) {
1412 x &= ~((long)0xff << (8*i));
1413 x |= (long)r_byte(arg) << (8*i);
1414 }
1415 }
1416 return x;
1417}
1418
1419long
1420ruby_marshal_read_long(const char **buf, long len)
1421{
1422 long x;
1423 struct RString src = {RBASIC_INIT};
1424 struct load_arg arg;
1425 memset(&arg, 0, sizeof(arg));
1426 arg.src = rb_setup_fake_str(&src, *buf, len, 0);
1427 x = r_long(&arg);
1428 *buf += arg.offset;
1429 return x;
1430}
1431
1432static long
1433r_keep_readable(struct load_arg *arg, long len, size_t size)
1434{
1435 if (UNLIKELY(len < 0)) {
1436 rb_raise(rb_eArgError, "negative length");
1437 }
1438 if (UNLIKELY((unsigned long)len > SIZE_MAX / size || arg->readable >= LONG_MAX - len)) {
1439 rb_raise(rb_eArgError, "marshaled data too big");
1440 }
1441 return len;
1442}
1443
1444static VALUE
1445r_bytes1(long len, struct load_arg *arg)
1446{
1447 VALUE str, n = LONG2NUM(len);
1448
1449 str = load_funcall(arg, arg->src, s_read, 1, &n);
1450 if (NIL_P(str)) too_short();
1451 StringValue(str);
1452 if (RSTRING_LEN(str) != len) too_short();
1453
1454 return str;
1455}
1456
1457static VALUE
1458r_bytes1_buffered(long len, struct load_arg *arg)
1459{
1460 VALUE str;
1461
1462 if (len <= arg->buflen) {
1463 str = rb_str_new(arg->buf+arg->offset, len);
1464 arg->offset += len;
1465 arg->buflen -= len;
1466 }
1467 else {
1468 long buflen = arg->buflen;
1469 long readable = arg->readable + 1;
1470 long tmp_len, read_len, need_len = len - buflen;
1471 VALUE tmp, n;
1472
1473 readable = readable < arg->bufsize ? readable : arg->bufsize;
1474 read_len = need_len > readable ? need_len : readable;
1475 n = LONG2NUM(read_len);
1476 tmp = load_funcall(arg, arg->src, s_read, 1, &n);
1477 if (NIL_P(tmp)) too_short();
1478 StringValue(tmp);
1479
1480 tmp_len = RSTRING_LEN(tmp);
1481
1482 if (tmp_len < need_len) too_short();
1483
1484 str = rb_str_new(arg->buf+arg->offset, buflen);
1485 rb_str_cat(str, RSTRING_PTR(tmp), need_len);
1486
1487 if (tmp_len > need_len) {
1488 buflen = tmp_len - need_len;
1489 memcpy(arg->buf, RSTRING_PTR(tmp)+need_len, buflen);
1490 arg->buflen = buflen;
1491 }
1492 else {
1493 arg->buflen = 0;
1494 }
1495 arg->offset = 0;
1496 }
1497
1498 return str;
1499}
1500
1501#define r_bytes(arg) r_bytes0(r_long(arg), (arg))
1502
1503static VALUE
1504r_bytes0(long len, struct load_arg *arg)
1505{
1506 VALUE str;
1507
1508 if (len == 0) return rb_str_new(0, 0);
1509 if (RB_TYPE_P(arg->src, T_STRING)) {
1510 if (RSTRING_LEN(arg->src) - arg->offset >= len) {
1511 str = rb_str_new(RSTRING_PTR(arg->src)+arg->offset, len);
1512 arg->offset += len;
1513 }
1514 else {
1515 too_short();
1516 }
1517 }
1518 else {
1519 if (arg->readable > 0 || arg->buflen > 0) {
1520 str = r_bytes1_buffered(len, arg);
1521 }
1522 else {
1523 str = r_bytes1(len, arg);
1524 }
1525 }
1526 return str;
1527}
1528
1529static inline int
1530name_equal(const char *name, size_t nlen, const char *p, long l)
1531{
1532 if ((size_t)l != nlen || *p != *name) return 0;
1533 return nlen == 1 || memcmp(p+1, name+1, nlen-1) == 0;
1534}
1535
1536static int
1537sym2encidx(VALUE sym, VALUE val)
1538{
1539 RBIMPL_ATTR_NONSTRING() static const char name_encoding[8] = "encoding";
1540 const char *p;
1541 long l;
1542 if (rb_enc_get_index(sym) != ENCINDEX_US_ASCII) return -1;
1543 RSTRING_GETMEM(sym, p, l);
1544 if (l <= 0) return -1;
1545 if (name_equal(name_encoding, sizeof(name_encoding), p, l)) {
1546 int idx = rb_enc_find_index(StringValueCStr(val));
1547 return idx;
1548 }
1549 if (name_equal(name_s_encoding_short, rb_strlen_lit(name_s_encoding_short), p, l)) {
1550 if (val == Qfalse) return rb_usascii_encindex();
1551 else if (val == Qtrue) return rb_utf8_encindex();
1552 /* bogus ignore */
1553 }
1554 return -1;
1555}
1556
1557static int
1558symname_equal(VALUE sym, const char *name, size_t nlen)
1559{
1560 const char *p;
1561 long l;
1562 if (rb_enc_get_index(sym) != ENCINDEX_US_ASCII) return 0;
1563 RSTRING_GETMEM(sym, p, l);
1564 return name_equal(name, nlen, p, l);
1565}
1566
1567#define BUILD_ASSERT_POSITIVE(n) \
1568 /* make 0 negative to workaround the "zero size array" GCC extension, */ \
1569 ((sizeof(char [2*(ssize_t)(n)-1])+1)/2) /* assuming no overflow */
1570#define symname_equal_lit(sym, sym_name) \
1571 symname_equal(sym, sym_name, BUILD_ASSERT_POSITIVE(rb_strlen_lit(sym_name)))
1572
1573static VALUE
1574r_symlink(struct load_arg *arg)
1575{
1576 st_data_t sym;
1577 long num = r_long(arg);
1578
1579 if (!st_lookup(arg->symbols, num, &sym)) {
1580 rb_raise(rb_eArgError, "bad symbol");
1581 }
1582 return (VALUE)sym;
1583}
1584
1585static VALUE
1586r_symreal(struct load_arg *arg, int ivar)
1587{
1588 VALUE s = r_bytes(arg);
1589 VALUE sym;
1590 int idx = -1;
1591 st_index_t n = arg->symbols->num_entries;
1592
1593 if (rb_enc_str_asciionly_p(s)) rb_enc_associate_index(s, ENCINDEX_US_ASCII);
1594 st_insert(arg->symbols, (st_data_t)n, (st_data_t)s);
1595 if (ivar) {
1596 long num = r_long(arg);
1597 while (num-- > 0) {
1598 sym = r_symbol(arg);
1599 idx = sym2encidx(sym, r_object(arg));
1600 }
1601 }
1602 if (idx > 0) {
1603 rb_enc_associate_index(s, idx);
1604 if (is_broken_string(s)) {
1605 rb_raise(rb_eArgError, "invalid byte sequence in %s: %+"PRIsVALUE,
1606 rb_enc_name(rb_enc_from_index(idx)), s);
1607 }
1608 }
1609
1610 return s;
1611}
1612
1613static VALUE
1614r_symbol(struct load_arg *arg)
1615{
1616 int type, ivar = 0;
1617
1618 again:
1619 switch ((type = r_byte(arg))) {
1620 default:
1621 rb_raise(rb_eArgError, "dump format error for symbol(0x%x)", type);
1622 case TYPE_IVAR:
1623 ivar = 1;
1624 goto again;
1625 case TYPE_SYMBOL:
1626 return r_symreal(arg, ivar);
1627 case TYPE_SYMLINK:
1628 if (ivar) {
1629 rb_raise(rb_eArgError, "dump format error (symlink with encoding)");
1630 }
1631 return r_symlink(arg);
1632 }
1633}
1634
1635static VALUE
1636r_unique(struct load_arg *arg)
1637{
1638 return r_symbol(arg);
1639}
1640
1641static VALUE
1642r_string(struct load_arg *arg)
1643{
1644 return r_bytes(arg);
1645}
1646
1647static VALUE
1648r_entry0(VALUE v, st_index_t num, struct load_arg *arg)
1649{
1650 st_data_t real_obj = (st_data_t)v;
1651 if (arg->compat_tbl) {
1652 /* real_obj is kept if not found */
1653 st_lookup(arg->compat_tbl, v, &real_obj);
1654 }
1655 st_insert(arg->data, num, real_obj);
1656 st_insert(arg->partial_objects, (st_data_t)real_obj, Qtrue);
1657 return v;
1658}
1659
1660static VALUE
1661r_fixup_compat(VALUE v, struct load_arg *arg)
1662{
1663 st_data_t data;
1664 st_data_t key = (st_data_t)v;
1665 if (arg->compat_tbl && st_delete(arg->compat_tbl, &key, &data)) {
1666 VALUE real_obj = (VALUE)data;
1667 rb_alloc_func_t allocator = rb_get_alloc_func(CLASS_OF(real_obj));
1668 if (st_lookup(compat_allocator_tbl, (st_data_t)allocator, &data)) {
1669 marshal_compat_t *compat = (marshal_compat_t*)data;
1670 compat->loader(real_obj, v);
1671 }
1672 v = real_obj;
1673 }
1674 return v;
1675}
1676
1677static VALUE
1678r_post_proc(VALUE v, struct load_arg *arg)
1679{
1680 if (arg->proc) {
1681 v = load_funcall(arg, arg->proc, s_call, 1, &v);
1682 }
1683 return v;
1684}
1685
1686static VALUE
1687r_leave(VALUE v, struct load_arg *arg, bool partial)
1688{
1689 v = r_fixup_compat(v, arg);
1690 if (!partial) {
1691 st_data_t data;
1692 st_data_t key = (st_data_t)v;
1693 st_delete(arg->partial_objects, &key, &data);
1694 if (arg->freeze) {
1695 if (RB_TYPE_P(v, T_MODULE) || RB_TYPE_P(v, T_CLASS)) {
1696 // noop
1697 }
1698 else if (RB_TYPE_P(v, T_STRING)) {
1699 v = rb_str_to_interned_str(v);
1700 }
1701 else {
1702 OBJ_FREEZE(v);
1703 }
1704 }
1705 v = r_post_proc(v, arg);
1706 }
1707 return v;
1708}
1709
1710static int
1711copy_ivar_i(ID vid, VALUE value, st_data_t arg)
1712{
1713 VALUE obj = (VALUE)arg;
1714
1715 if (!rb_ivar_defined(obj, vid))
1716 rb_ivar_set(obj, vid, value);
1717 return ST_CONTINUE;
1718}
1719
1720static VALUE
1721r_copy_ivar(VALUE v, VALUE data)
1722{
1723 rb_ivar_foreach(data, copy_ivar_i, (st_data_t)v);
1724 return v;
1725}
1726
1727#define override_ivar_error(type, str) \
1728 rb_raise(rb_eTypeError, \
1729 "can't override instance variable of "type" '%"PRIsVALUE"'", \
1730 (str))
1731
1732static int
1733r_ivar_encoding(VALUE obj, struct load_arg *arg, VALUE sym, VALUE val)
1734{
1735 int idx = sym2encidx(sym, val);
1736 if (idx >= 0) {
1737 if (rb_enc_capable(obj)) {
1738 rb_enc_associate_index(obj, idx);
1739 }
1740 else {
1741 rb_raise(rb_eArgError, "%"PRIsVALUE" is not enc_capable", obj);
1742 }
1743 return TRUE;
1744 }
1745 return FALSE;
1746}
1747
1748static long
1749r_encname(VALUE obj, struct load_arg *arg)
1750{
1751 long len = r_long(arg);
1752 if (len > 0) {
1753 VALUE sym = r_symbol(arg);
1754 VALUE val = r_object(arg);
1755 len -= r_ivar_encoding(obj, arg, sym, val);
1756 }
1757 return len;
1758}
1759
1760static void
1761r_ivar(VALUE obj, int *has_encoding, struct load_arg *arg)
1762{
1763 long len;
1764
1765 len = r_long(arg);
1766 if (len > 0) {
1767 if (RB_TYPE_P(obj, T_MODULE)) {
1768 override_ivar_error("module", rb_mod_name(obj));
1769 }
1770 else if (RB_TYPE_P(obj, T_CLASS)) {
1771 override_ivar_error("class", rb_class_name(obj));
1772 }
1773 do {
1774 VALUE sym = r_symbol(arg);
1775 VALUE val = r_object(arg);
1776 if (r_ivar_encoding(obj, arg, sym, val)) {
1777 if (has_encoding) *has_encoding = TRUE;
1778 }
1779 else if (symname_equal_lit(sym, name_s_ruby2_keywords_flag)) {
1780 if (RB_TYPE_P(obj, T_HASH)) {
1781 rb_hash_ruby2_keywords(obj);
1782 }
1783 else {
1784 rb_raise(rb_eArgError, "ruby2_keywords flag is given but %"PRIsVALUE" is not a Hash", obj);
1785 }
1786 }
1787 else {
1788 rb_ivar_set(obj, rb_intern_str(sym), val);
1789 }
1790 } while (--len > 0);
1791 }
1792}
1793
1794static VALUE
1795path2class(VALUE path)
1796{
1797 VALUE v = rb_path_to_class(path);
1798
1799 if (!RB_TYPE_P(v, T_CLASS)) {
1800 rb_raise(rb_eArgError, "%"PRIsVALUE" does not refer to class", path);
1801 }
1802 return v;
1803}
1804
1805#define path2module(path) must_be_module(rb_path_to_class(path), path)
1806
1807static VALUE
1808must_be_module(VALUE v, VALUE path)
1809{
1810 if (!RB_TYPE_P(v, T_MODULE)) {
1811 rb_raise(rb_eArgError, "%"PRIsVALUE" does not refer to module", path);
1812 }
1813 return v;
1814}
1815
1816static VALUE
1817obj_alloc_by_klass(VALUE klass, struct load_arg *arg, VALUE *oldclass)
1818{
1819 st_data_t data;
1820 rb_alloc_func_t allocator;
1821
1822 allocator = rb_get_alloc_func(klass);
1823 if (st_lookup(compat_allocator_tbl, (st_data_t)allocator, &data)) {
1824 marshal_compat_t *compat = (marshal_compat_t*)data;
1825 VALUE real_obj = rb_obj_alloc(klass);
1826 VALUE obj = rb_obj_alloc(compat->oldclass);
1827 if (oldclass) *oldclass = compat->oldclass;
1828
1829 if (!arg->compat_tbl) {
1830 arg->compat_tbl = rb_init_identtable();
1831 }
1832 st_insert(arg->compat_tbl, (st_data_t)obj, (st_data_t)real_obj);
1833 return obj;
1834 }
1835
1836 return rb_obj_alloc(klass);
1837}
1838
1839static VALUE
1840obj_alloc_by_path(VALUE path, struct load_arg *arg)
1841{
1842 return obj_alloc_by_klass(path2class(path), arg, 0);
1843}
1844
1845static VALUE
1846append_extmod(VALUE obj, VALUE extmod)
1847{
1848 long i = RARRAY_LEN(extmod);
1849 while (i > 0) {
1850 VALUE m = RARRAY_AREF(extmod, --i);
1851 rb_extend_object(obj, m);
1852 }
1853 return obj;
1854}
1855
1856#define prohibit_ivar(type, str) do { \
1857 if (!ivp || !*ivp) break; \
1858 override_ivar_error(type, str); \
1859 } while (0)
1860
1861static VALUE r_object_for(struct load_arg *arg, bool partial, int *ivp, VALUE extmod, int type);
1862
1863static VALUE
1864r_object0(struct load_arg *arg, bool partial, int *ivp, VALUE extmod)
1865{
1866 int type = r_byte(arg);
1867 return r_object_for(arg, partial, ivp, extmod, type);
1868}
1869
1870static VALUE
1871r_object_for(struct load_arg *arg, bool partial, int *ivp, VALUE extmod, int type)
1872{
1873 VALUE (*hash_new_with_size)(st_index_t) = rb_hash_new_with_size;
1874 VALUE v = Qnil;
1875 long id;
1876 st_data_t link;
1877
1878 switch (type) {
1879 case TYPE_LINK:
1880 id = r_long(arg);
1881 if (!st_lookup(arg->data, (st_data_t)id, &link)) {
1882 rb_raise(rb_eArgError, "dump format error (unlinked)");
1883 }
1884 v = (VALUE)link;
1885 if (!st_lookup(arg->partial_objects, (st_data_t)v, &link)) {
1886 if (arg->freeze && RB_TYPE_P(v, T_STRING)) {
1887 v = rb_str_to_interned_str(v);
1888 }
1889 v = r_post_proc(v, arg);
1890 }
1891 break;
1892
1893 case TYPE_IVAR:
1894 {
1895 int ivar = TRUE;
1896 v = r_object0(arg, true, &ivar, extmod);
1897 if (ivar) r_ivar(v, NULL, arg);
1898 v = r_leave(v, arg, partial);
1899 }
1900 break;
1901
1902 case TYPE_EXTENDED:
1903 {
1904 VALUE path = r_unique(arg);
1905 VALUE m = rb_path_to_class(path);
1906 if (NIL_P(extmod)) extmod = rb_ary_hidden_new(0);
1907
1908 if (RB_TYPE_P(m, T_CLASS)) { /* prepended */
1909 VALUE c;
1910
1911 v = r_object0(arg, true, 0, Qnil);
1912 c = CLASS_OF(v);
1913 if (c != m || FL_TEST(c, FL_SINGLETON)) {
1914 rb_raise(rb_eArgError,
1915 "prepended class %"PRIsVALUE" differs from class %"PRIsVALUE,
1916 path, rb_class_name(c));
1917 }
1918 c = rb_singleton_class(v);
1919 while (RARRAY_LEN(extmod) > 0) {
1920 m = rb_ary_pop(extmod);
1921 rb_prepend_module(c, m);
1922 }
1923 }
1924 else {
1925 must_be_module(m, path);
1926 rb_ary_push(extmod, m);
1927
1928 v = r_object0(arg, true, 0, extmod);
1929 while (RARRAY_LEN(extmod) > 0) {
1930 m = rb_ary_pop(extmod);
1931 rb_extend_object(v, m);
1932 }
1933 }
1934 v = r_leave(v, arg, partial);
1935 }
1936 break;
1937
1938 case TYPE_UCLASS:
1939 {
1940 VALUE c = path2class(r_unique(arg));
1941
1942 if (FL_TEST(c, FL_SINGLETON)) {
1943 rb_raise(rb_eTypeError, "singleton can't be loaded");
1944 }
1945 type = r_byte(arg);
1946 if ((c == rb_cHash) &&
1947 /* Hack for compare_by_identify */
1948 (type == TYPE_HASH || type == TYPE_HASH_DEF)) {
1949 hash_new_with_size = rb_ident_hash_new_with_size;
1950 goto type_hash;
1951 }
1952 v = r_object_for(arg, partial, 0, extmod, type);
1953 if (RB_SPECIAL_CONST_P(v) || RB_TYPE_P(v, T_OBJECT) || RB_TYPE_P(v, T_CLASS)) {
1954 goto format_error;
1955 }
1956 if (RB_TYPE_P(v, T_MODULE) || !RTEST(rb_class_inherited_p(c, RBASIC(v)->klass))) {
1957 VALUE tmp = rb_obj_alloc(c);
1958
1959 if (TYPE(v) != TYPE(tmp)) goto format_error;
1960 }
1961 RBASIC_SET_CLASS(v, c);
1962 }
1963 break;
1964
1965 format_error:
1966 rb_raise(rb_eArgError, "dump format error (user class)");
1967
1968 case TYPE_NIL:
1969 v = Qnil;
1970 v = r_leave(v, arg, false);
1971 break;
1972
1973 case TYPE_TRUE:
1974 v = Qtrue;
1975 v = r_leave(v, arg, false);
1976 break;
1977
1978 case TYPE_FALSE:
1979 v = Qfalse;
1980 v = r_leave(v, arg, false);
1981 break;
1982
1983 case TYPE_FIXNUM:
1984 {
1985 long i = r_long(arg);
1986 v = LONG2FIX(i);
1987 }
1988 v = r_leave(v, arg, false);
1989 break;
1990
1991 case TYPE_FLOAT:
1992 {
1993 double d;
1994 VALUE str = r_bytes(arg);
1995 const char *ptr = RSTRING_PTR(str);
1996
1997 if (strcmp(ptr, "nan") == 0) {
1998 d = nan("");
1999 }
2000 else if (strcmp(ptr, "inf") == 0) {
2001 d = HUGE_VAL;
2002 }
2003 else if (strcmp(ptr, "-inf") == 0) {
2004 d = -HUGE_VAL;
2005 }
2006 else {
2007 char *e;
2008 d = strtod(ptr, &e);
2009 d = load_mantissa(d, e, RSTRING_LEN(str) - (e - ptr));
2010 }
2011 v = DBL2NUM(d);
2012 v = r_entry(v, arg);
2013 v = r_leave(v, arg, false);
2014 }
2015 break;
2016
2017 case TYPE_BIGNUM:
2018 {
2019 long len;
2020 VALUE data;
2021 int sign;
2022
2023 sign = r_byte(arg);
2024 if (sign != '+' && sign != '-') {
2025 rb_raise(rb_eArgError, "invalid Bignum sign");
2026 }
2027 len = r_keep_readable(arg, r_long(arg), 2);
2028
2029 if (SIZEOF_VALUE >= 8 && len <= 4) {
2030 // Representable within uintptr, likely FIXNUM
2031 VALUE num = 0;
2032 for (int i = 0; i < len; i++) {
2033 num |= (VALUE)r_byte(arg) << (i * 16);
2034 num |= (VALUE)r_byte(arg) << (i * 16 + 8);
2035 }
2036#if SIZEOF_VALUE == SIZEOF_LONG
2037 v = ULONG2NUM(num);
2038#else
2039 v = ULL2NUM(num);
2040#endif
2041 if (sign == '-') {
2042 v = rb_int_uminus(v);
2043 }
2044 }
2045 else {
2046 data = r_bytes0(len * 2, arg);
2047 v = rb_integer_unpack(RSTRING_PTR(data), len, 2, 0,
2048 INTEGER_PACK_LITTLE_ENDIAN | (sign == '-' ? INTEGER_PACK_NEGATIVE : 0));
2049 rb_str_resize(data, 0L);
2050 }
2051 v = r_entry(v, arg);
2052 v = r_leave(v, arg, false);
2053 }
2054 break;
2055
2056 case TYPE_STRING:
2057 v = r_entry(r_string(arg), arg);
2058 v = r_leave(v, arg, partial);
2059 break;
2060
2061 case TYPE_REGEXP:
2062 {
2063 VALUE str = r_bytes(arg);
2064 int options = r_byte(arg);
2065 int has_encoding = FALSE;
2066 st_index_t idx = r_prepare(arg);
2067
2068 if (ivp) {
2069 r_ivar(str, &has_encoding, arg);
2070 *ivp = FALSE;
2071 }
2072 if (!has_encoding) {
2073 /* 1.8 compatibility; remove escapes undefined in 1.8 */
2074 char *ptr = RSTRING_PTR(str), *dst = ptr, *src = ptr;
2075 long len = RSTRING_LEN(str);
2076 long bs = 0;
2077 for (; len-- > 0; *dst++ = *src++) {
2078 switch (*src) {
2079 case '\\': bs++; break;
2080 case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
2081 case 'm': case 'o': case 'p': case 'q': case 'u': case 'y':
2082 case 'E': case 'F': case 'H': case 'I': case 'J': case 'K':
2083 case 'L': case 'N': case 'O': case 'P': case 'Q': case 'R':
2084 case 'S': case 'T': case 'U': case 'V': case 'X': case 'Y':
2085 if (bs & 1) --dst;
2086 /* fall through */
2087 default: bs = 0; break;
2088 }
2089 }
2090 rb_str_set_len(str, dst - ptr);
2091 }
2092 VALUE regexp = rb_reg_new_str(str, options);
2093 r_copy_ivar(regexp, str);
2094
2095 v = r_entry0(regexp, idx, arg);
2096 v = r_leave(v, arg, partial);
2097 }
2098 break;
2099
2100 case TYPE_ARRAY:
2101 {
2102 long len = r_keep_readable(arg, r_long(arg), 1);
2103
2104 v = rb_ary_new2(len);
2105 v = r_entry(v, arg);
2106 arg->readable += len - 1;
2107 while (len--) {
2108 rb_ary_push(v, r_object(arg));
2109 arg->readable--;
2110 }
2111 v = r_leave(v, arg, partial);
2112 arg->readable++;
2113 }
2114 break;
2115
2116 case TYPE_HASH:
2117 case TYPE_HASH_DEF:
2118 type_hash:
2119 {
2120 long len = r_keep_readable(arg, r_long(arg), 2);
2121
2122 v = hash_new_with_size(len);
2123 v = r_entry(v, arg);
2124 arg->readable += (len - 1) * 2;
2125 while (len--) {
2126 VALUE key = r_object(arg);
2127 VALUE value = r_object(arg);
2128 rb_hash_aset(v, key, value);
2129 arg->readable -= 2;
2130 }
2131 arg->readable += 2;
2132 if (type == TYPE_HASH_DEF) {
2133 RHASH_SET_IFNONE(v, r_object(arg));
2134 }
2135 v = r_leave(v, arg, partial);
2136 }
2137 break;
2138
2139 case TYPE_STRUCT:
2140 {
2141 VALUE mem, values;
2142 long i;
2143 VALUE slot;
2144 st_index_t idx = r_prepare(arg);
2145 VALUE klass = path2class(r_unique(arg));
2146 long len = r_keep_readable(arg, r_long(arg), 2);
2147
2148 v = rb_obj_alloc(klass);
2149 if (!RB_TYPE_P(v, T_STRUCT)) {
2150 rb_raise(rb_eTypeError, "class %"PRIsVALUE" not a struct", rb_class_name(klass));
2151 }
2152 mem = rb_struct_s_members(klass);
2153 if (RARRAY_LEN(mem) != len) {
2154 rb_raise(rb_eTypeError, "struct %"PRIsVALUE" not compatible (struct size differs)",
2155 rb_class_name(klass));
2156 }
2157
2158 arg->readable += (len - 1) * 2;
2159 v = r_entry0(v, idx, arg);
2160 values = rb_ary_new2(len);
2161 {
2162 VALUE keywords = Qfalse;
2163 if (RTEST(rb_struct_s_keyword_init(klass))) {
2164 keywords = rb_hash_new();
2165 rb_ary_push(values, keywords);
2166 }
2167
2168 for (i=0; i<len; i++) {
2169 VALUE n = rb_sym2str(RARRAY_AREF(mem, i));
2170 slot = r_symbol(arg);
2171
2172 if (!rb_str_equal(n, slot)) {
2173 rb_raise(rb_eTypeError, "struct %"PRIsVALUE" not compatible (:%"PRIsVALUE" for :%"PRIsVALUE")",
2174 rb_class_name(klass),
2175 slot, n);
2176 }
2177 if (keywords) {
2178 rb_hash_aset(keywords, RARRAY_AREF(mem, i), r_object(arg));
2179 }
2180 else {
2181 rb_ary_push(values, r_object(arg));
2182 }
2183 arg->readable -= 2;
2184 }
2185 }
2186 rb_struct_initialize(v, values);
2187 v = r_leave(v, arg, partial);
2188 arg->readable += 2;
2189 }
2190 break;
2191
2192 case TYPE_USERDEF:
2193 {
2194 VALUE name = r_unique(arg);
2195 VALUE klass = path2class(name);
2196 VALUE data;
2197 st_data_t d;
2198
2199 if (!rb_obj_respond_to(klass, s_load, TRUE)) {
2200 rb_raise(rb_eTypeError, "class %"PRIsVALUE" needs to have method '_load'",
2201 name);
2202 }
2203 data = r_string(arg);
2204 if (ivp) {
2205 r_ivar(data, NULL, arg);
2206 *ivp = FALSE;
2207 }
2208 v = load_funcall(arg, klass, s_load, 1, &data);
2209 v = r_entry(v, arg);
2210 if (st_lookup(compat_allocator_tbl, (st_data_t)rb_get_alloc_func(klass), &d)) {
2211 marshal_compat_t *compat = (marshal_compat_t*)d;
2212 v = compat->loader(klass, v);
2213 }
2214 if (!partial) {
2215 if (arg->freeze) {
2216 OBJ_FREEZE(v);
2217 }
2218 v = r_post_proc(v, arg);
2219 }
2220 }
2221 break;
2222
2223 case TYPE_USRMARSHAL:
2224 {
2225 VALUE name = r_unique(arg);
2226 VALUE klass = path2class(name);
2227 VALUE oldclass = 0;
2228 VALUE data;
2229
2230 v = obj_alloc_by_klass(klass, arg, &oldclass);
2231 if (!NIL_P(extmod)) {
2232 /* for the case marshal_load is overridden */
2233 append_extmod(v, extmod);
2234 }
2235 if (!rb_obj_respond_to(v, s_mload, TRUE)) {
2236 rb_raise(rb_eTypeError, "instance of %"PRIsVALUE" needs to have method 'marshal_load'",
2237 name);
2238 }
2239 v = r_entry(v, arg);
2240 data = r_object(arg);
2241 load_funcall(arg, v, s_mload, 1, &data);
2242 v = r_fixup_compat(v, arg);
2243 v = r_copy_ivar(v, data);
2244 if (arg->freeze) {
2245 OBJ_FREEZE(v);
2246 }
2247 v = r_post_proc(v, arg);
2248 if (!NIL_P(extmod)) {
2249 if (oldclass) append_extmod(v, extmod);
2250 rb_ary_clear(extmod);
2251 }
2252 }
2253 break;
2254
2255 case TYPE_OBJECT:
2256 {
2257 st_index_t idx = r_prepare(arg);
2258 v = obj_alloc_by_path(r_unique(arg), arg);
2259 if (!RB_TYPE_P(v, T_OBJECT)) {
2260 rb_raise(rb_eArgError, "dump format error");
2261 }
2262 v = r_entry0(v, idx, arg);
2263 r_ivar(v, NULL, arg);
2264 v = r_leave(v, arg, partial);
2265 }
2266 break;
2267
2268 case TYPE_DATA:
2269 {
2270 VALUE name = r_unique(arg);
2271 VALUE klass = path2class(name);
2272 VALUE oldclass = 0;
2273 VALUE r;
2274
2275 v = obj_alloc_by_klass(klass, arg, &oldclass);
2276 if (!RB_TYPE_P(v, T_DATA)) {
2277 rb_raise(rb_eArgError, "dump format error");
2278 }
2279 v = r_entry(v, arg);
2280 if (!rb_obj_respond_to(v, s_load_data, TRUE)) {
2281 rb_raise(rb_eTypeError,
2282 "class %"PRIsVALUE" needs to have instance method '_load_data'",
2283 name);
2284 }
2285 r = r_object0(arg, partial, 0, extmod);
2286 load_funcall(arg, v, s_load_data, 1, &r);
2287 v = r_leave(v, arg, partial);
2288 }
2289 break;
2290
2291 case TYPE_MODULE_OLD:
2292 {
2293 VALUE str = r_bytes(arg);
2294
2295 v = rb_path_to_class(str);
2296 prohibit_ivar("class/module", str);
2297 v = r_entry(v, arg);
2298 v = r_leave(v, arg, partial);
2299 }
2300 break;
2301
2302 case TYPE_CLASS:
2303 {
2304 VALUE str = r_bytes(arg);
2305
2306 if (ivp && *ivp > 0) *ivp = r_encname(str, arg) > 0;
2307 v = path2class(str);
2308 prohibit_ivar("class", str);
2309 v = r_entry(v, arg);
2310 v = r_leave(v, arg, partial);
2311 }
2312 break;
2313
2314 case TYPE_MODULE:
2315 {
2316 VALUE str = r_bytes(arg);
2317
2318 if (ivp && *ivp > 0) *ivp = r_encname(str, arg) > 0;
2319 v = path2module(str);
2320 prohibit_ivar("module", str);
2321 v = r_entry(v, arg);
2322 v = r_leave(v, arg, partial);
2323 }
2324 break;
2325
2326 case TYPE_SYMBOL:
2327 if (ivp) {
2328 v = r_symreal(arg, *ivp);
2329 *ivp = FALSE;
2330 }
2331 else {
2332 v = r_symreal(arg, 0);
2333 }
2334 v = rb_str_intern(v);
2335 v = r_leave(v, arg, partial);
2336 break;
2337
2338 case TYPE_SYMLINK:
2339 v = rb_str_intern(r_symlink(arg));
2340 break;
2341
2342 default:
2343 rb_raise(rb_eArgError, "dump format error(0x%x)", type);
2344 break;
2345 }
2346
2347 if (UNDEF_P(v)) {
2348 rb_raise(rb_eArgError, "dump format error (bad link)");
2349 }
2350
2351 return v;
2352}
2353
2354static VALUE
2355r_object(struct load_arg *arg)
2356{
2357 return r_object0(arg, false, 0, Qnil);
2358}
2359
2360static void
2361clear_load_arg(struct load_arg *arg)
2362{
2363 xfree(arg->buf);
2364 arg->buf = NULL;
2365 arg->bufsize = 0;
2366 arg->buflen = 0;
2367 arg->offset = 0;
2368 arg->readable = 0;
2369 if (!arg->symbols) return;
2370 st_free_table(arg->symbols);
2371 arg->symbols = 0;
2372 st_free_table(arg->data);
2373 arg->data = 0;
2374 st_free_table(arg->partial_objects);
2375 arg->partial_objects = 0;
2376 if (arg->compat_tbl) {
2377 st_free_table(arg->compat_tbl);
2378 arg->compat_tbl = 0;
2379 }
2380}
2381
2382VALUE
2383rb_marshal_load_with_proc(VALUE port, VALUE proc, bool freeze)
2384{
2385 int major, minor;
2386 VALUE v;
2387 VALUE wrapper; /* used to avoid memory leak in case of exception */
2388 struct load_arg *arg;
2389
2390 v = rb_check_string_type(port);
2391 if (!NIL_P(v)) {
2392 port = v;
2393 }
2394 else if (rb_respond_to(port, s_getbyte) && rb_respond_to(port, s_read)) {
2395 rb_check_funcall(port, s_binmode, 0, 0);
2396 }
2397 else {
2398 io_needed();
2399 }
2400 wrapper = TypedData_Make_Struct(0, struct load_arg, &load_arg_data, arg);
2401 arg->src = port;
2402 arg->offset = 0;
2403 arg->symbols = st_init_numtable();
2404 arg->data = rb_init_identtable();
2405 arg->partial_objects = rb_init_identtable();
2406 arg->compat_tbl = 0;
2407 arg->proc = 0;
2408 arg->readable = 0;
2409 arg->freeze = freeze;
2410
2411 if (NIL_P(v)) {
2412 arg->bufsize = BUFSIZ;
2413 arg->buf = xmalloc(BUFSIZ);
2414 }
2415 else {
2416 arg->bufsize = 0;
2417 arg->buf = 0;
2418 }
2419
2420 major = r_byte(arg);
2421 minor = r_byte(arg);
2422 if (major != MARSHAL_MAJOR || minor > MARSHAL_MINOR) {
2423 clear_load_arg(arg);
2424 rb_raise(rb_eTypeError, "incompatible marshal file format (can't be read)\n\
2425\tformat version %d.%d required; %d.%d given",
2426 MARSHAL_MAJOR, MARSHAL_MINOR, major, minor);
2427 }
2428 if (RTEST(ruby_verbose) && minor != MARSHAL_MINOR) {
2429 rb_warn("incompatible marshal file format (can be read)\n\
2430\tformat version %d.%d required; %d.%d given",
2431 MARSHAL_MAJOR, MARSHAL_MINOR, major, minor);
2432 }
2433
2434 if (!NIL_P(proc)) arg->proc = proc;
2435 v = r_object(arg);
2436 clear_load_arg(arg);
2437 RB_GC_GUARD(wrapper);
2438
2439 return v;
2440}
2441
2442static VALUE
2443marshal_load(rb_execution_context_t *ec, VALUE mod, VALUE source, VALUE proc, VALUE freeze)
2444{
2445 return rb_marshal_load_with_proc(source, proc, RTEST(freeze));
2446}
2447
2448#include "marshal.rbinc"
2449
2450/*
2451 * The marshaling library converts collections of Ruby objects into a
2452 * byte stream, allowing them to be stored outside the currently
2453 * active script. This data may subsequently be read and the original
2454 * objects reconstituted.
2455 *
2456 * Marshaled data has major and minor version numbers stored along
2457 * with the object information. In normal use, marshaling can only
2458 * load data written with the same major version number and an equal
2459 * or lower minor version number. If Ruby's ``verbose'' flag is set
2460 * (normally using -d, -v, -w, or --verbose) the major and minor
2461 * numbers must match exactly. Marshal versioning is independent of
2462 * Ruby's version numbers. You can extract the version by reading the
2463 * first two bytes of marshaled data.
2464 *
2465 * str = Marshal.dump("thing")
2466 * RUBY_VERSION #=> "1.9.0"
2467 * str[0].ord #=> 4
2468 * str[1].ord #=> 8
2469 *
2470 * Some objects cannot be dumped: if the objects to be dumped include
2471 * bindings, procedure or method objects, instances of class IO, or
2472 * singleton objects, a TypeError will be raised.
2473 *
2474 * If your class has special serialization needs (for example, if you
2475 * want to serialize in some specific format), or if it contains
2476 * objects that would otherwise not be serializable, you can implement
2477 * your own serialization strategy.
2478 *
2479 * There are two methods of doing this, your object can define either
2480 * marshal_dump and marshal_load or _dump and _load. marshal_dump will take
2481 * precedence over _dump if both are defined. marshal_dump may result in
2482 * smaller Marshal strings.
2483 *
2484 * == Security considerations
2485 *
2486 * By design, Marshal.load can deserialize almost any class loaded into the
2487 * Ruby process. In many cases this can lead to remote code execution if the
2488 * Marshal data is loaded from an untrusted source.
2489 *
2490 * As a result, Marshal.load is not suitable as a general purpose serialization
2491 * format and you should never unmarshal user supplied input or other untrusted
2492 * data.
2493 *
2494 * If you need to deserialize untrusted data, use JSON or another serialization
2495 * format that is only able to load simple, 'primitive' types such as String,
2496 * Array, Hash, etc. Never allow user input to specify arbitrary types to
2497 * deserialize into.
2498 *
2499 * == marshal_dump and marshal_load
2500 *
2501 * When dumping an object the method marshal_dump will be called.
2502 * marshal_dump must return a result containing the information necessary for
2503 * marshal_load to reconstitute the object. The result can be any object.
2504 *
2505 * When loading an object dumped using marshal_dump the object is first
2506 * allocated then marshal_load is called with the result from marshal_dump.
2507 * marshal_load must recreate the object from the information in the result.
2508 *
2509 * Example:
2510 *
2511 * class MyObj
2512 * def initialize name, version, data
2513 * @name = name
2514 * @version = version
2515 * @data = data
2516 * end
2517 *
2518 * def marshal_dump
2519 * [@name, @version]
2520 * end
2521 *
2522 * def marshal_load array
2523 * @name, @version = array
2524 * end
2525 * end
2526 *
2527 * == _dump and _load
2528 *
2529 * Use _dump and _load when you need to allocate the object you're restoring
2530 * yourself.
2531 *
2532 * When dumping an object the instance method _dump is called with an Integer
2533 * which indicates the maximum depth of objects to dump (a value of -1 implies
2534 * that you should disable depth checking). _dump must return a String
2535 * containing the information necessary to reconstitute the object.
2536 *
2537 * The class method _load should take a String and use it to return an object
2538 * of the same class.
2539 *
2540 * Example:
2541 *
2542 * class MyObj
2543 * def initialize name, version, data
2544 * @name = name
2545 * @version = version
2546 * @data = data
2547 * end
2548 *
2549 * def _dump level
2550 * [@name, @version].join ':'
2551 * end
2552 *
2553 * def self._load args
2554 * new(*args.split(':'))
2555 * end
2556 * end
2557 *
2558 * Since Marshal.dump outputs a string you can have _dump return a Marshal
2559 * string which is Marshal.loaded in _load for complex objects.
2560 */
2561void
2562Init_marshal(void)
2563{
2564 VALUE rb_mMarshal = rb_define_module("Marshal");
2565#define set_id(sym) sym = rb_intern_const(name_##sym)
2566 set_id(s_dump);
2567 set_id(s_load);
2568 set_id(s_mdump);
2569 set_id(s_mload);
2570 set_id(s_dump_data);
2571 set_id(s_load_data);
2572 set_id(s_alloc);
2573 set_id(s_call);
2574 set_id(s_getbyte);
2575 set_id(s_read);
2576 set_id(s_write);
2577 set_id(s_binmode);
2578 set_id(s_encoding_short);
2579 set_id(s_ruby2_keywords_flag);
2580
2581 rb_define_module_function(rb_mMarshal, "dump", marshal_dump, -1);
2582
2583 /* major version */
2584 rb_define_const(rb_mMarshal, "MAJOR_VERSION", INT2FIX(MARSHAL_MAJOR));
2585 /* minor version */
2586 rb_define_const(rb_mMarshal, "MINOR_VERSION", INT2FIX(MARSHAL_MINOR));
2587}
2588
2589static int
2590marshal_compat_table_mark_and_move_i(st_data_t key, st_data_t value, st_data_t _)
2591{
2592 marshal_compat_t *p = (marshal_compat_t *)value;
2593 rb_gc_mark_and_move(&p->newclass);
2594 rb_gc_mark_and_move(&p->oldclass);
2595 return ST_CONTINUE;
2596}
2597
2598static void
2599marshal_compat_table_mark_and_move(void *tbl)
2600{
2601 if (!tbl) return;
2602 st_foreach(tbl, marshal_compat_table_mark_and_move_i, 0);
2603}
2604
2605static int
2606marshal_compat_table_free_i(st_data_t key, st_data_t value, st_data_t _)
2607{
2608 xfree((marshal_compat_t *)value);
2609 return ST_CONTINUE;
2610}
2611
2612static void
2613marshal_compat_table_free(void *data)
2614{
2615 st_foreach(data, marshal_compat_table_free_i, 0);
2616 st_free_table(data);
2617}
2618
2619static size_t
2620marshal_compat_table_memsize(const void *data)
2621{
2622 return st_memsize(data) + sizeof(marshal_compat_t) * st_table_size(data);
2623}
2624
2625static const rb_data_type_t marshal_compat_type = {
2626 .wrap_struct_name = "marshal_compat_table",
2627 .function = {
2628 .dmark = marshal_compat_table_mark_and_move,
2629 .dfree = marshal_compat_table_free,
2630 .dsize = marshal_compat_table_memsize,
2631 .dcompact = marshal_compat_table_mark_and_move,
2632 },
2633 .flags = RUBY_TYPED_WB_PROTECTED | RUBY_TYPED_FREE_IMMEDIATELY,
2634};
2635
2636static st_table *
2637compat_allocator_table(void)
2638{
2639 if (compat_allocator_tbl) return compat_allocator_tbl;
2640 compat_allocator_tbl = st_init_numtable();
2641 compat_allocator_tbl_wrapper =
2642 TypedData_Wrap_Struct(0, &marshal_compat_type, compat_allocator_tbl);
2643 rb_vm_register_global_object(compat_allocator_tbl_wrapper);
2644 return compat_allocator_tbl;
2645}
2646
2647VALUE
2648rb_marshal_dump(VALUE obj, VALUE port)
2649{
2650 return rb_marshal_dump_limited(obj, port, -1);
2651}
2652
2653VALUE
2654rb_marshal_load(VALUE port)
2655{
2656 return rb_marshal_load_with_proc(port, Qnil, false);
2657}
len
int len
Length of the buffer.
Definition io.h:8
nonstring.h
Defines RBIMPL_ATTR_NONSTRING.
Generated by doxygen 1.17.0