Ruby 3.5.0dev (2025-04-03 revision 1dddc6c78b5f6dc6ae18ee04ebe44abfce3b0433)
marshal.c (1dddc6c78b5f6dc6ae18ee04ebe44abfce3b0433)
1/**********************************************************************
2
3 marshal.c -
4
5 $Author$
6 created at: Thu Apr 27 16:30:01 JST 1995
7
8 Copyright (C) 1993-2007 Yukihiro Matsumoto
9
10**********************************************************************/
11
12#include "ruby/internal/config.h"
13
14#include <math.h>
15#ifdef HAVE_FLOAT_H
16#include <float.h>
17#endif
18#ifdef HAVE_IEEEFP_H
19#include <ieeefp.h>
20#endif
21
22#include "encindex.h"
23#include "id_table.h"
24#include "internal.h"
25#include "internal/array.h"
26#include "internal/bignum.h"
27#include "internal/class.h"
28#include "internal/encoding.h"
29#include "internal/error.h"
30#include "internal/hash.h"
31#include "internal/numeric.h"
32#include "internal/object.h"
33#include "internal/struct.h"
34#include "internal/symbol.h"
35#include "internal/util.h"
36#include "internal/vm.h"
37#include "ruby/io.h"
38#include "ruby/ruby.h"
39#include "ruby/st.h"
40#include "ruby/util.h"
41#include "builtin.h"
42#include "shape.h"
43
44#define BITSPERSHORT (2*CHAR_BIT)
45#define SHORTMASK ((1<<BITSPERSHORT)-1)
46#define SHORTDN(x) RSHIFT((x),BITSPERSHORT)
47
48#if SIZEOF_SHORT == SIZEOF_BDIGIT
49#define SHORTLEN(x) (x)
50#else
51static size_t
52shortlen(size_t len, BDIGIT *ds)
53{
54 BDIGIT num;
55 int offset = 0;
56
57 num = ds[len-1];
58 while (num) {
59 num = SHORTDN(num);
60 offset++;
61 }
62 return (len - 1)*SIZEOF_BDIGIT/2 + offset;
63}
64#define SHORTLEN(x) shortlen((x),d)
65#endif
66
67#define MARSHAL_MAJOR 4
68#define MARSHAL_MINOR 8
69
70#define TYPE_NIL '0'
71#define TYPE_TRUE 'T'
72#define TYPE_FALSE 'F'
73#define TYPE_FIXNUM 'i'
74
75#define TYPE_EXTENDED 'e'
76#define TYPE_UCLASS 'C'
77#define TYPE_OBJECT 'o'
78#define TYPE_DATA 'd'
79#define TYPE_USERDEF 'u'
80#define TYPE_USRMARSHAL 'U'
81#define TYPE_FLOAT 'f'
82#define TYPE_BIGNUM 'l'
83#define TYPE_STRING '"'
84#define TYPE_REGEXP '/'
85#define TYPE_ARRAY '['
86#define TYPE_HASH '{'
87#define TYPE_HASH_DEF '}'
88#define TYPE_STRUCT 'S'
89#define TYPE_MODULE_OLD 'M'
90#define TYPE_CLASS 'c'
91#define TYPE_MODULE 'm'
92
93#define TYPE_SYMBOL ':'
94#define TYPE_SYMLINK ';'
95
96#define TYPE_IVAR 'I'
97#define TYPE_LINK '@'
98
99static ID s_dump, s_load, s_mdump, s_mload;
100static ID s_dump_data, s_load_data, s_alloc, s_call;
101static ID s_getbyte, s_read, s_write, s_binmode;
102static ID s_encoding_short, s_ruby2_keywords_flag;
103#define s_encoding_long rb_id_encoding()
104
105#define name_s_dump "_dump"
106#define name_s_load "_load"
107#define name_s_mdump "marshal_dump"
108#define name_s_mload "marshal_load"
109#define name_s_dump_data "_dump_data"
110#define name_s_load_data "_load_data"
111#define name_s_alloc "_alloc"
112#define name_s_call "call"
113#define name_s_getbyte "getbyte"
114#define name_s_read "read"
115#define name_s_write "write"
116#define name_s_binmode "binmode"
117#define name_s_encoding_short "E"
118#define name_s_encoding_long "encoding"
119#define name_s_ruby2_keywords_flag "K"
120
121typedef struct {
122 VALUE newclass;
123 VALUE oldclass;
124 VALUE (*dumper)(VALUE);
125 VALUE (*loader)(VALUE, VALUE);
126} marshal_compat_t;
127
128static st_table *compat_allocator_tbl;
129static VALUE compat_allocator_tbl_wrapper;
130static VALUE rb_marshal_dump_limited(VALUE obj, VALUE port, int limit);
131static VALUE rb_marshal_load_with_proc(VALUE port, VALUE proc, bool freeze);
132
133static st_table *compat_allocator_table(void);
134
135void
136rb_marshal_define_compat(VALUE newclass, VALUE oldclass, VALUE (*dumper)(VALUE), VALUE (*loader)(VALUE, VALUE))
137{
138 marshal_compat_t *compat;
139 rb_alloc_func_t allocator = rb_get_alloc_func(newclass);
140
141 if (!allocator) {
142 rb_raise(rb_eTypeError, "no allocator");
143 }
144
145 compat_allocator_table();
146 compat = ALLOC(marshal_compat_t);
147 RB_OBJ_WRITE(compat_allocator_tbl_wrapper, &compat->newclass, newclass);
148 RB_OBJ_WRITE(compat_allocator_tbl_wrapper, &compat->oldclass, oldclass);
149 compat->dumper = dumper;
150 compat->loader = loader;
151
152 st_insert(compat_allocator_table(), (st_data_t)allocator, (st_data_t)compat);
153}
154
155struct dump_arg {
156 VALUE str, dest;
157 st_table *symbols;
158 st_table *data;
159 st_table *compat_tbl;
160 st_table *encodings;
161 st_table *userdefs;
162 st_index_t num_entries;
163};
164
165struct dump_call_arg {
166 VALUE obj;
167 struct dump_arg *arg;
168 int limit;
169};
170
171static VALUE
172check_dump_arg(VALUE ret, struct dump_arg *arg, const char *name)
173{
174 if (!arg->symbols) {
175 rb_raise(rb_eRuntimeError, "Marshal.dump reentered at %s",
176 name);
177 }
178 return ret;
179}
180
181static VALUE
182check_userdump_arg(VALUE obj, ID sym, int argc, const VALUE *argv,
183 struct dump_arg *arg, const char *name)
184{
185 VALUE ret = rb_funcallv(obj, sym, argc, argv);
186 VALUE klass = CLASS_OF(obj);
187 if (CLASS_OF(ret) == klass) {
188 rb_raise(rb_eRuntimeError, "%"PRIsVALUE"#%s returned same class instance",
189 klass, name);
190 }
191 return check_dump_arg(ret, arg, name);
192}
193
194#define dump_funcall(arg, obj, sym, argc, argv) \
195 check_userdump_arg(obj, sym, argc, argv, arg, name_##sym)
196#define dump_check_funcall(arg, obj, sym, argc, argv) \
197 check_dump_arg(rb_check_funcall(obj, sym, argc, argv), arg, name_##sym)
198
199static void clear_dump_arg(struct dump_arg *arg);
200
201static void
202mark_dump_arg(void *ptr)
203{
204 struct dump_arg *p = ptr;
205 if (!p->symbols)
206 return;
207 rb_mark_set(p->symbols);
208 rb_mark_set(p->data);
209 rb_mark_hash(p->compat_tbl);
210 rb_mark_set(p->userdefs);
211 rb_gc_mark(p->str);
212}
213
214static void
215free_dump_arg(void *ptr)
216{
217 clear_dump_arg(ptr);
218}
219
220static size_t
221memsize_dump_arg(const void *ptr)
222{
223 const struct dump_arg *p = (struct dump_arg *)ptr;
224 size_t memsize = 0;
225 if (p->symbols) memsize += rb_st_memsize(p->symbols);
226 if (p->data) memsize += rb_st_memsize(p->data);
227 if (p->compat_tbl) memsize += rb_st_memsize(p->compat_tbl);
228 if (p->userdefs) memsize += rb_st_memsize(p->userdefs);
229 if (p->encodings) memsize += rb_st_memsize(p->encodings);
230 return memsize;
231}
232
233static const rb_data_type_t dump_arg_data = {
234 "dump_arg",
235 {mark_dump_arg, free_dump_arg, memsize_dump_arg,},
236 0, 0, RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_EMBEDDABLE
237};
238
239static VALUE
240must_not_be_anonymous(const char *type, VALUE path)
241{
242 char *n = RSTRING_PTR(path);
243
244 if (!rb_enc_asciicompat(rb_enc_get(path))) {
245 /* cannot occur? */
246 rb_raise(rb_eTypeError, "can't dump non-ascii %s name % "PRIsVALUE,
247 type, path);
248 }
249 if (n[0] == '#') {
250 rb_raise(rb_eTypeError, "can't dump anonymous %s % "PRIsVALUE,
251 type, path);
252 }
253 return path;
254}
255
256static VALUE
257class2path(VALUE klass)
258{
259 VALUE path = rb_class_path(klass);
260
261 must_not_be_anonymous((RB_TYPE_P(klass, T_CLASS) ? "class" : "module"), path);
262 if (rb_path_to_class(path) != rb_class_real(klass)) {
263 rb_raise(rb_eTypeError, "% "PRIsVALUE" can't be referred to", path);
264 }
265 return path;
266}
267
268int ruby_marshal_write_long(long x, char *buf);
269static void w_long(long, struct dump_arg*);
270static int w_encoding(VALUE encname, struct dump_call_arg *arg);
271static VALUE encoding_name(VALUE obj, struct dump_arg *arg);
272
273static void
274w_nbyte(const char *s, long n, struct dump_arg *arg)
275{
276 VALUE buf = arg->str;
277 rb_str_buf_cat(buf, s, n);
278 if (arg->dest && RSTRING_LEN(buf) >= BUFSIZ) {
279 rb_io_write(arg->dest, buf);
280 rb_str_resize(buf, 0);
281 }
282}
283
284static void
285w_byte(char c, struct dump_arg *arg)
286{
287 w_nbyte(&c, 1, arg);
288}
289
290static void
291w_bytes(const char *s, long n, struct dump_arg *arg)
292{
293 w_long(n, arg);
294 w_nbyte(s, n, arg);
295}
296
297#define w_cstr(s, arg) w_bytes((s), strlen(s), (arg))
298
299static void
300w_short(int x, struct dump_arg *arg)
301{
302 w_byte((char)((x >> 0) & 0xff), arg);
303 w_byte((char)((x >> 8) & 0xff), arg);
304}
305
306static void
307w_long(long x, struct dump_arg *arg)
308{
309 char buf[sizeof(long)+1];
310 int i = ruby_marshal_write_long(x, buf);
311 if (i < 0) {
312 rb_raise(rb_eTypeError, "long too big to dump");
313 }
314 w_nbyte(buf, i, arg);
315}
316
317int
318ruby_marshal_write_long(long x, char *buf)
319{
320 int i;
321
322#if SIZEOF_LONG > 4
323 if (!(RSHIFT(x, 31) == 0 || RSHIFT(x, 31) == -1)) {
324 /* big long does not fit in 4 bytes */
325 return -1;
326 }
327#endif
328
329 if (x == 0) {
330 buf[0] = 0;
331 return 1;
332 }
333 if (0 < x && x < 123) {
334 buf[0] = (char)(x + 5);
335 return 1;
336 }
337 if (-124 < x && x < 0) {
338 buf[0] = (char)((x - 5)&0xff);
339 return 1;
340 }
341 for (i=1;i<(int)sizeof(long)+1;i++) {
342 buf[i] = (char)(x & 0xff);
343 x = RSHIFT(x,8);
344 if (x == 0) {
345 buf[0] = i;
346 break;
347 }
348 if (x == -1) {
349 buf[0] = -i;
350 break;
351 }
352 }
353 return i+1;
354}
355
356#ifdef DBL_MANT_DIG
357#define DECIMAL_MANT (53-16) /* from IEEE754 double precision */
358
359#if DBL_MANT_DIG > 32
360#define MANT_BITS 32
361#elif DBL_MANT_DIG > 24
362#define MANT_BITS 24
363#elif DBL_MANT_DIG > 16
364#define MANT_BITS 16
365#else
366#define MANT_BITS 8
367#endif
368
369static double
370load_mantissa(double d, const char *buf, long len)
371{
372 if (!len) return d;
373 if (--len > 0 && !*buf++) { /* binary mantissa mark */
374 int e, s = d < 0, dig = 0;
375 unsigned long m;
376
377 modf(ldexp(frexp(fabs(d), &e), DECIMAL_MANT), &d);
378 do {
379 m = 0;
380 switch (len) {
381 default: m = *buf++ & 0xff; /* fall through */
382#if MANT_BITS > 24
383 case 3: m = (m << 8) | (*buf++ & 0xff); /* fall through */
384#endif
385#if MANT_BITS > 16
386 case 2: m = (m << 8) | (*buf++ & 0xff); /* fall through */
387#endif
388#if MANT_BITS > 8
389 case 1: m = (m << 8) | (*buf++ & 0xff);
390#endif
391 }
392 dig -= len < MANT_BITS / 8 ? 8 * (unsigned)len : MANT_BITS;
393 d += ldexp((double)m, dig);
394 } while ((len -= MANT_BITS / 8) > 0);
395 d = ldexp(d, e - DECIMAL_MANT);
396 if (s) d = -d;
397 }
398 return d;
399}
400#else
401#define load_mantissa(d, buf, len) (d)
402#endif
403
404#ifdef DBL_DIG
405#define FLOAT_DIG (DBL_DIG+2)
406#else
407#define FLOAT_DIG 17
408#endif
409
410static void
411w_float(double d, struct dump_arg *arg)
412{
413 char buf[FLOAT_DIG + (DECIMAL_MANT + 7) / 8 + 10];
414
415 if (isinf(d)) {
416 if (d < 0) w_cstr("-inf", arg);
417 else w_cstr("inf", arg);
418 }
419 else if (isnan(d)) {
420 w_cstr("nan", arg);
421 }
422 else if (d == 0.0) {
423 if (signbit(d)) w_cstr("-0", arg);
424 else w_cstr("0", arg);
425 }
426 else {
427 int decpt, sign, digs, len = 0;
428 char *e, *p = ruby_dtoa(d, 0, 0, &decpt, &sign, &e);
429 if (sign) buf[len++] = '-';
430 digs = (int)(e - p);
431 if (decpt < -3 || decpt > digs) {
432 buf[len++] = p[0];
433 if (--digs > 0) buf[len++] = '.';
434 memcpy(buf + len, p + 1, digs);
435 len += digs;
436 len += snprintf(buf + len, sizeof(buf) - len, "e%d", decpt - 1);
437 }
438 else if (decpt > 0) {
439 memcpy(buf + len, p, decpt);
440 len += decpt;
441 if ((digs -= decpt) > 0) {
442 buf[len++] = '.';
443 memcpy(buf + len, p + decpt, digs);
444 len += digs;
445 }
446 }
447 else {
448 buf[len++] = '0';
449 buf[len++] = '.';
450 if (decpt) {
451 memset(buf + len, '0', -decpt);
452 len -= decpt;
453 }
454 memcpy(buf + len, p, digs);
455 len += digs;
456 }
457 free(p);
458 w_bytes(buf, len, arg);
459 }
460}
461
462static void
463w_symbol(VALUE sym, struct dump_arg *arg)
464{
465 st_data_t num;
466 VALUE encname;
467
468 if (st_lookup(arg->symbols, sym, &num)) {
469 w_byte(TYPE_SYMLINK, arg);
470 w_long((long)num, arg);
471 }
472 else {
473 const VALUE orig_sym = sym;
474 sym = rb_sym2str(sym);
475 if (!sym) {
476 rb_raise(rb_eTypeError, "can't dump anonymous ID %"PRIdVALUE, sym);
477 }
478 encname = encoding_name(sym, arg);
479 if (NIL_P(encname) ||
480 is_ascii_string(sym)) {
481 encname = Qnil;
482 }
483 else {
484 w_byte(TYPE_IVAR, arg);
485 }
486 w_byte(TYPE_SYMBOL, arg);
487 w_bytes(RSTRING_PTR(sym), RSTRING_LEN(sym), arg);
488 st_add_direct(arg->symbols, orig_sym, arg->symbols->num_entries);
489 if (!NIL_P(encname)) {
490 struct dump_call_arg c_arg;
491 c_arg.limit = 1;
492 c_arg.arg = arg;
493 w_long(1L, arg);
494 w_encoding(encname, &c_arg);
495 }
496 }
497}
498
499static void
500w_unique(VALUE s, struct dump_arg *arg)
501{
502 must_not_be_anonymous("class", s);
503 w_symbol(rb_str_intern(s), arg);
504}
505
506static void w_object(VALUE,struct dump_arg*,int);
507
508static int
509hash_each(VALUE key, VALUE value, VALUE v)
510{
511 struct dump_call_arg *arg = (void *)v;
512 w_object(key, arg->arg, arg->limit);
513 w_object(value, arg->arg, arg->limit);
514 return ST_CONTINUE;
515}
516
517#define SINGLETON_DUMP_UNABLE_P(klass) \
518 (rb_id_table_size(RCLASS_M_TBL(klass)) > 0 || \
519 rb_ivar_count(klass) > 0)
520
521static void
522w_extended(VALUE klass, struct dump_arg *arg, int check)
523{
524 if (check && RCLASS_SINGLETON_P(klass)) {
525 VALUE origin = RCLASS_ORIGIN(klass);
526 if (SINGLETON_DUMP_UNABLE_P(klass) ||
527 (origin != klass && SINGLETON_DUMP_UNABLE_P(origin))) {
528 rb_raise(rb_eTypeError, "singleton can't be dumped");
529 }
530 klass = RCLASS_SUPER(klass);
531 }
532 while (BUILTIN_TYPE(klass) == T_ICLASS) {
533 if (!FL_TEST(klass, RICLASS_IS_ORIGIN) ||
534 BUILTIN_TYPE(RBASIC(klass)->klass) != T_MODULE) {
535 VALUE path = rb_class_name(RBASIC(klass)->klass);
536 w_byte(TYPE_EXTENDED, arg);
537 w_unique(path, arg);
538 }
539 klass = RCLASS_SUPER(klass);
540 }
541}
542
543static void
544w_class(char type, VALUE obj, struct dump_arg *arg, int check)
545{
546 VALUE path;
547 st_data_t real_obj;
548 VALUE klass;
549
550 if (arg->compat_tbl &&
551 st_lookup(arg->compat_tbl, (st_data_t)obj, &real_obj)) {
552 obj = (VALUE)real_obj;
553 }
554 klass = CLASS_OF(obj);
555 w_extended(klass, arg, check);
556 w_byte(type, arg);
557 path = class2path(rb_class_real(klass));
558 w_unique(path, arg);
559}
560
561static void
562w_uclass(VALUE obj, VALUE super, struct dump_arg *arg)
563{
564 VALUE klass = CLASS_OF(obj);
565
566 w_extended(klass, arg, TRUE);
567 klass = rb_class_real(klass);
568 if (klass != super) {
569 w_byte(TYPE_UCLASS, arg);
570 w_unique(class2path(klass), arg);
571 }
572}
573
574static bool
575rb_hash_ruby2_keywords_p(VALUE obj)
576{
577 return (RHASH(obj)->basic.flags & RHASH_PASS_AS_KEYWORDS) != 0;
578}
579
580static void
581rb_hash_ruby2_keywords(VALUE obj)
582{
583 RHASH(obj)->basic.flags |= RHASH_PASS_AS_KEYWORDS;
584}
585
586/*
587 * if instance variable name `id` is a special name to be skipped,
588 * returns the name of it. otherwise it cannot be dumped (unnamed),
589 * returns `name` as-is. returns NULL for ID that can be dumped.
590 */
591static inline const char *
592skipping_ivar_name(const ID id, const char *name)
593{
594#define IS_SKIPPED_IVAR(idname) \
595 ((id == idname) && (name = name_##idname, true))
596 if (IS_SKIPPED_IVAR(s_encoding_short)) return name;
597 if (IS_SKIPPED_IVAR(s_ruby2_keywords_flag)) return name;
598 if (IS_SKIPPED_IVAR(s_encoding_long)) return name;
599 if (!rb_id2str(id)) return name;
600 return NULL;
601}
602
603struct w_ivar_arg {
604 struct dump_call_arg *dump;
605 st_data_t num_ivar;
606};
607
608static int
609w_obj_each(ID id, VALUE value, st_data_t a)
610{
611 struct w_ivar_arg *ivarg = (struct w_ivar_arg *)a;
612 struct dump_call_arg *arg = ivarg->dump;
613 const char unnamed[] = "", *ivname = skipping_ivar_name(id, unnamed);
614
615 if (ivname) {
616 if (ivname != unnamed) {
617 rb_warn("instance variable '%s' on class %"PRIsVALUE" is not dumped",
618 ivname, CLASS_OF(arg->obj));
619 }
620 return ST_CONTINUE;
621 }
622 --ivarg->num_ivar;
623 w_symbol(ID2SYM(id), arg->arg);
624 w_object(value, arg->arg, arg->limit);
625 return ST_CONTINUE;
626}
627
628static int
629obj_count_ivars(ID id, VALUE val, st_data_t a)
630{
631 if (!skipping_ivar_name(id, "") && UNLIKELY(!++*(st_index_t *)a)) {
632 rb_raise(rb_eRuntimeError, "too many instance variables");
633 }
634 return ST_CONTINUE;
635}
636
637static VALUE
638encoding_name(VALUE obj, struct dump_arg *arg)
639{
640 if (rb_enc_capable(obj)) {
641 int encidx = rb_enc_get_index(obj);
642 rb_encoding *enc = 0;
643 st_data_t name;
644
645 if (encidx <= 0 || !(enc = rb_enc_from_index(encidx))) {
646 return Qnil;
647 }
648
649 /* special treatment for US-ASCII and UTF-8 */
650 if (encidx == rb_usascii_encindex()) {
651 return Qfalse;
652 }
653 else if (encidx == rb_utf8_encindex()) {
654 return Qtrue;
655 }
656
657 if (arg->encodings ?
658 !st_lookup(arg->encodings, (st_data_t)rb_enc_name(enc), &name) :
659 (arg->encodings = st_init_strcasetable(), 1)) {
660 name = (st_data_t)rb_str_new_cstr(rb_enc_name(enc));
661 st_insert(arg->encodings, (st_data_t)rb_enc_name(enc), name);
662 }
663 return (VALUE)name;
664 }
665 else {
666 return Qnil;
667 }
668}
669
670static int
671w_encoding(VALUE encname, struct dump_call_arg *arg)
672{
673 int limit = arg->limit;
674 if (limit >= 0) ++limit;
675 switch (encname) {
676 case Qfalse:
677 case Qtrue:
678 w_symbol(ID2SYM(s_encoding_short), arg->arg);
679 w_object(encname, arg->arg, limit);
680 return 1;
681 case Qnil:
682 return 0;
683 }
684 w_symbol(ID2SYM(rb_id_encoding()), arg->arg);
685 w_object(encname, arg->arg, limit);
686 return 1;
687}
688
689static st_index_t
690has_ivars(VALUE obj, VALUE encname, VALUE *ivobj)
691{
692 st_index_t num = !NIL_P(encname);
693
694 if (SPECIAL_CONST_P(obj)) goto generic;
695 switch (BUILTIN_TYPE(obj)) {
696 case T_OBJECT:
697 case T_CLASS:
698 case T_MODULE:
699 break; /* counted elsewhere */
700 case T_HASH:
701 if (rb_hash_ruby2_keywords_p(obj)) ++num;
702 /* fall through */
703 default:
704 generic:
705 rb_ivar_foreach(obj, obj_count_ivars, (st_data_t)&num);
706 if (num) *ivobj = obj;
707 }
708
709 return num;
710}
711
712static void
713w_ivar_each(VALUE obj, st_index_t num, struct dump_call_arg *arg)
714{
715 shape_id_t shape_id = rb_shape_get_shape_id(arg->obj);
716 struct w_ivar_arg ivarg = {arg, num};
717 if (!num) return;
718 rb_ivar_foreach(obj, w_obj_each, (st_data_t)&ivarg);
719
720 if (shape_id != rb_shape_get_shape_id(arg->obj)) {
721 rb_shape_t * expected_shape = rb_shape_get_shape_by_id(shape_id);
722 rb_shape_t * actual_shape = rb_shape_get_shape(arg->obj);
723
724 // If the shape tree got _shorter_ then we probably removed an IV
725 // If the shape tree got longer, then we probably added an IV.
726 // The exception message might not be accurate when someone adds and
727 // removes the same number of IVs, but they will still get an exception
728 if (rb_shape_depth(expected_shape) > rb_shape_depth(actual_shape)) {
729 rb_raise(rb_eRuntimeError, "instance variable removed from %"PRIsVALUE" instance",
730 CLASS_OF(arg->obj));
731 }
732 else {
733 rb_raise(rb_eRuntimeError, "instance variable added to %"PRIsVALUE" instance",
734 CLASS_OF(arg->obj));
735 }
736 }
737}
738
739static void
740w_ivar(st_index_t num, VALUE ivobj, VALUE encname, struct dump_call_arg *arg)
741{
742 w_long(num, arg->arg);
743 num -= w_encoding(encname, arg);
744 if (RB_TYPE_P(ivobj, T_HASH) && rb_hash_ruby2_keywords_p(ivobj)) {
745 int limit = arg->limit;
746 if (limit >= 0) ++limit;
747 w_symbol(ID2SYM(s_ruby2_keywords_flag), arg->arg);
748 w_object(Qtrue, arg->arg, limit);
749 num--;
750 }
751 if (!UNDEF_P(ivobj) && num) {
752 w_ivar_each(ivobj, num, arg);
753 }
754}
755
756static void
757w_objivar(VALUE obj, struct dump_call_arg *arg)
758{
759 st_data_t num = 0;
760
761 rb_ivar_foreach(obj, obj_count_ivars, (st_data_t)&num);
762 w_long(num, arg->arg);
763 w_ivar_each(obj, num, arg);
764}
765
766#if SIZEOF_LONG > 4
767// Optimized dump for fixnum larger than 31-bits
768static void
769w_bigfixnum(VALUE obj, struct dump_arg *arg)
770{
771 RUBY_ASSERT(FIXNUM_P(obj));
772
773 w_byte(TYPE_BIGNUM, arg);
774
775#if SIZEOF_LONG == SIZEOF_VALUE
776 long num, slen_num;
777 num = FIX2LONG(obj);
778#else
779 long long num, slen_num;
780 num = NUM2LL(obj);
781#endif
782
783 char sign = num < 0 ? '-' : '+';
784 w_byte(sign, arg);
785
786 // Guaranteed not to overflow, as FIXNUM is 1-bit less than long
787 if (num < 0) num = -num;
788
789 // calculate the size in shorts
790 int slen = 0;
791 {
792 slen_num = num;
793 while (slen_num) {
794 slen++;
795 slen_num = SHORTDN(slen_num);
796 }
797 }
798
799 RUBY_ASSERT(slen > 0 && slen <= SIZEOF_LONG / 2);
800
801 w_long((long)slen, arg);
802
803 for (int i = 0; i < slen; i++) {
804 w_short(num & SHORTMASK, arg);
805 num = SHORTDN(num);
806 }
807
808 // We aren't adding this object to the link table, but we need to increment
809 // the index.
810 arg->num_entries++;
811
812 RUBY_ASSERT(num == 0);
813}
814#endif
815
816static void
817w_remember(VALUE obj, struct dump_arg *arg)
818{
819 st_add_direct(arg->data, obj, arg->num_entries++);
820}
821
822static void
823w_object(VALUE obj, struct dump_arg *arg, int limit)
824{
825 struct dump_call_arg c_arg;
826 VALUE ivobj = Qundef;
827 st_data_t num;
828 st_index_t hasiv = 0;
829 VALUE encname = Qnil;
830
831 if (limit == 0) {
832 rb_raise(rb_eArgError, "exceed depth limit");
833 }
834
835 if (NIL_P(obj)) {
836 w_byte(TYPE_NIL, arg);
837 }
838 else if (obj == Qtrue) {
839 w_byte(TYPE_TRUE, arg);
840 }
841 else if (obj == Qfalse) {
842 w_byte(TYPE_FALSE, arg);
843 }
844 else if (FIXNUM_P(obj)) {
845#if SIZEOF_LONG <= 4
846 w_byte(TYPE_FIXNUM, arg);
847 w_long(FIX2INT(obj), arg);
848#else
849 if (RSHIFT((long)obj, 31) == 0 || RSHIFT((long)obj, 31) == -1) {
850 w_byte(TYPE_FIXNUM, arg);
851 w_long(FIX2LONG(obj), arg);
852 }
853 else {
854 w_bigfixnum(obj, arg);
855 }
856#endif
857 }
858 else if (SYMBOL_P(obj)) {
859 w_symbol(obj, arg);
860 }
861 else {
862 if (st_lookup(arg->data, obj, &num)) {
863 w_byte(TYPE_LINK, arg);
864 w_long((long)num, arg);
865 return;
866 }
867
868 if (limit > 0) limit--;
869 c_arg.limit = limit;
870 c_arg.arg = arg;
871 c_arg.obj = obj;
872
873 if (FLONUM_P(obj)) {
874 w_remember(obj, arg);
875 w_byte(TYPE_FLOAT, arg);
876 w_float(RFLOAT_VALUE(obj), arg);
877 return;
878 }
879
880 VALUE v;
881
882 if (!RBASIC_CLASS(obj)) {
883 rb_raise(rb_eTypeError, "can't dump internal %s",
884 rb_builtin_type_name(BUILTIN_TYPE(obj)));
885 }
886
887 if (rb_obj_respond_to(obj, s_mdump, TRUE)) {
888 w_remember(obj, arg);
889
890 v = dump_funcall(arg, obj, s_mdump, 0, 0);
891 w_class(TYPE_USRMARSHAL, obj, arg, FALSE);
892 w_object(v, arg, limit);
893 return;
894 }
895 if (rb_obj_respond_to(obj, s_dump, TRUE)) {
896 VALUE ivobj2 = Qundef;
897 st_index_t hasiv2;
898 VALUE encname2;
899
900 if (arg->userdefs && st_is_member(arg->userdefs, (st_data_t)obj)) {
901 rb_raise(rb_eRuntimeError, "can't dump recursive object using _dump()");
902 }
903 v = INT2NUM(limit);
904 v = dump_funcall(arg, obj, s_dump, 1, &v);
905 if (!RB_TYPE_P(v, T_STRING)) {
906 rb_raise(rb_eTypeError, "_dump() must return string");
907 }
908 hasiv = has_ivars(obj, (encname = encoding_name(obj, arg)), &ivobj);
909 hasiv2 = has_ivars(v, (encname2 = encoding_name(v, arg)), &ivobj2);
910 if (hasiv2) {
911 hasiv = hasiv2;
912 ivobj = ivobj2;
913 encname = encname2;
914 }
915 if (hasiv) w_byte(TYPE_IVAR, arg);
916 w_class(TYPE_USERDEF, obj, arg, FALSE);
917 w_bytes(RSTRING_PTR(v), RSTRING_LEN(v), arg);
918 if (hasiv) {
919 st_data_t userdefs = (st_data_t)obj;
920 if (!arg->userdefs) {
921 arg->userdefs = rb_init_identtable();
922 }
923 st_add_direct(arg->userdefs, userdefs, 0);
924 w_ivar(hasiv, ivobj, encname, &c_arg);
925 st_delete(arg->userdefs, &userdefs, NULL);
926 }
927 w_remember(obj, arg);
928 return;
929 }
930
931 w_remember(obj, arg);
932
933 hasiv = has_ivars(obj, (encname = encoding_name(obj, arg)), &ivobj);
934 {
935 st_data_t compat_data;
936 rb_alloc_func_t allocator = rb_get_alloc_func(RBASIC(obj)->klass);
937 if (st_lookup(compat_allocator_tbl,
938 (st_data_t)allocator,
939 &compat_data)) {
940 marshal_compat_t *compat = (marshal_compat_t*)compat_data;
941 VALUE real_obj = obj;
942 obj = compat->dumper(real_obj);
943 if (!arg->compat_tbl) {
944 arg->compat_tbl = rb_init_identtable();
945 }
946 st_insert(arg->compat_tbl, (st_data_t)obj, (st_data_t)real_obj);
947 if (obj != real_obj && UNDEF_P(ivobj)) hasiv = 0;
948 }
949 }
950 if (hasiv) w_byte(TYPE_IVAR, arg);
951
952 switch (BUILTIN_TYPE(obj)) {
953 case T_CLASS:
954 if (FL_TEST(obj, FL_SINGLETON)) {
955 rb_raise(rb_eTypeError, "singleton class can't be dumped");
956 }
957 w_byte(TYPE_CLASS, arg);
958 {
959 VALUE path = class2path(obj);
960 w_bytes(RSTRING_PTR(path), RSTRING_LEN(path), arg);
961 RB_GC_GUARD(path);
962 }
963 break;
964
965 case T_MODULE:
966 w_byte(TYPE_MODULE, arg);
967 {
968 VALUE path = class2path(obj);
969 w_bytes(RSTRING_PTR(path), RSTRING_LEN(path), 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, Data, 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 buflen;
1244 long readable;
1245 long offset;
1246 st_table *symbols;
1247 st_table *data;
1248 st_table *partial_objects;
1249 VALUE proc;
1250 st_table *compat_tbl;
1251 bool freeze;
1252};
1253
1254static VALUE
1255check_load_arg(VALUE ret, struct load_arg *arg, const char *name)
1256{
1257 if (!arg->symbols) {
1258 rb_raise(rb_eRuntimeError, "Marshal.load reentered at %s",
1259 name);
1260 }
1261 return ret;
1262}
1263#define load_funcall(arg, obj, sym, argc, argv) \
1264 check_load_arg(rb_funcallv(obj, sym, argc, argv), arg, name_##sym)
1265
1266static void clear_load_arg(struct load_arg *arg);
1267
1268static void
1269mark_load_arg(void *ptr)
1270{
1271 struct load_arg *p = ptr;
1272 if (!p->symbols)
1273 return;
1274 rb_mark_tbl(p->symbols);
1275 rb_mark_tbl(p->data);
1276 rb_mark_tbl(p->partial_objects);
1277 rb_mark_hash(p->compat_tbl);
1278}
1279
1280static void
1281free_load_arg(void *ptr)
1282{
1283 clear_load_arg(ptr);
1284}
1285
1286static size_t
1287memsize_load_arg(const void *ptr)
1288{
1289 const struct load_arg *p = (struct load_arg *)ptr;
1290 size_t memsize = 0;
1291 if (p->symbols) memsize += rb_st_memsize(p->symbols);
1292 if (p->data) memsize += rb_st_memsize(p->data);
1293 if (p->partial_objects) memsize += rb_st_memsize(p->partial_objects);
1294 if (p->compat_tbl) memsize += rb_st_memsize(p->compat_tbl);
1295 return memsize;
1296}
1297
1298static const rb_data_type_t load_arg_data = {
1299 "load_arg",
1300 {mark_load_arg, free_load_arg, memsize_load_arg,},
1301 0, 0, RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_EMBEDDABLE
1302};
1303
1304#define r_entry(v, arg) r_entry0((v), (arg)->data->num_entries, (arg))
1305static VALUE r_object(struct load_arg *arg);
1306static VALUE r_symbol(struct load_arg *arg);
1307
1308NORETURN(static void too_short(void));
1309static void
1310too_short(void)
1311{
1312 rb_raise(rb_eArgError, "marshal data too short");
1313}
1314
1315static st_index_t
1316r_prepare(struct load_arg *arg)
1317{
1318 st_index_t idx = arg->data->num_entries;
1319
1320 st_insert(arg->data, (st_data_t)idx, (st_data_t)Qundef);
1321 return idx;
1322}
1323
1324static unsigned char
1325r_byte1_buffered(struct load_arg *arg)
1326{
1327 if (arg->buflen == 0) {
1328 long readable = arg->readable < BUFSIZ ? arg->readable : BUFSIZ;
1329 VALUE str, n = LONG2NUM(readable);
1330
1331 str = load_funcall(arg, arg->src, s_read, 1, &n);
1332 if (NIL_P(str)) too_short();
1333 StringValue(str);
1334 memcpy(arg->buf, RSTRING_PTR(str), RSTRING_LEN(str));
1335 arg->offset = 0;
1336 arg->buflen = RSTRING_LEN(str);
1337 }
1338 arg->buflen--;
1339 return arg->buf[arg->offset++];
1340}
1341
1342static int
1343r_byte(struct load_arg *arg)
1344{
1345 int c;
1346
1347 if (RB_TYPE_P(arg->src, T_STRING)) {
1348 if (RSTRING_LEN(arg->src) > arg->offset) {
1349 c = (unsigned char)RSTRING_PTR(arg->src)[arg->offset++];
1350 }
1351 else {
1352 too_short();
1353 }
1354 }
1355 else {
1356 if (arg->readable >0 || arg->buflen > 0) {
1357 c = r_byte1_buffered(arg);
1358 }
1359 else {
1360 VALUE v = load_funcall(arg, arg->src, s_getbyte, 0, 0);
1361 if (NIL_P(v)) rb_eof_error();
1362 c = (unsigned char)NUM2CHR(v);
1363 }
1364 }
1365 return c;
1366}
1367
1368NORETURN(static void long_toobig(int size));
1369
1370static void
1371long_toobig(int size)
1372{
1373 rb_raise(rb_eTypeError, "long too big for this architecture (size "
1374 STRINGIZE(SIZEOF_LONG)", given %d)", size);
1375}
1376
1377static long
1378r_long(struct load_arg *arg)
1379{
1380 register long x;
1381 int c = (signed char)r_byte(arg);
1382 long i;
1383
1384 if (c == 0) return 0;
1385 if (c > 0) {
1386 if (4 < c && c < 128) {
1387 return c - 5;
1388 }
1389 if (c > (int)sizeof(long)) long_toobig(c);
1390 x = 0;
1391 for (i=0;i<c;i++) {
1392 x |= (long)r_byte(arg) << (8*i);
1393 }
1394 }
1395 else {
1396 if (-129 < c && c < -4) {
1397 return c + 5;
1398 }
1399 c = -c;
1400 if (c > (int)sizeof(long)) long_toobig(c);
1401 x = -1;
1402 for (i=0;i<c;i++) {
1403 x &= ~((long)0xff << (8*i));
1404 x |= (long)r_byte(arg) << (8*i);
1405 }
1406 }
1407 return x;
1408}
1409
1410long
1411ruby_marshal_read_long(const char **buf, long len)
1412{
1413 long x;
1414 struct RString src;
1415 struct load_arg arg;
1416 memset(&arg, 0, sizeof(arg));
1417 arg.src = rb_setup_fake_str(&src, *buf, len, 0);
1418 x = r_long(&arg);
1419 *buf += arg.offset;
1420 return x;
1421}
1422
1423static VALUE
1424r_bytes1(long len, struct load_arg *arg)
1425{
1426 VALUE str, n = LONG2NUM(len);
1427
1428 str = load_funcall(arg, arg->src, s_read, 1, &n);
1429 if (NIL_P(str)) too_short();
1430 StringValue(str);
1431 if (RSTRING_LEN(str) != len) too_short();
1432
1433 return str;
1434}
1435
1436static VALUE
1437r_bytes1_buffered(long len, struct load_arg *arg)
1438{
1439 VALUE str;
1440
1441 if (len <= arg->buflen) {
1442 str = rb_str_new(arg->buf+arg->offset, len);
1443 arg->offset += len;
1444 arg->buflen -= len;
1445 }
1446 else {
1447 long buflen = arg->buflen;
1448 long readable = arg->readable + 1;
1449 long tmp_len, read_len, need_len = len - buflen;
1450 VALUE tmp, n;
1451
1452 readable = readable < BUFSIZ ? readable : BUFSIZ;
1453 read_len = need_len > readable ? need_len : readable;
1454 n = LONG2NUM(read_len);
1455 tmp = load_funcall(arg, arg->src, s_read, 1, &n);
1456 if (NIL_P(tmp)) too_short();
1457 StringValue(tmp);
1458
1459 tmp_len = RSTRING_LEN(tmp);
1460
1461 if (tmp_len < need_len) too_short();
1462
1463 str = rb_str_new(arg->buf+arg->offset, buflen);
1464 rb_str_cat(str, RSTRING_PTR(tmp), need_len);
1465
1466 if (tmp_len > need_len) {
1467 buflen = tmp_len - need_len;
1468 memcpy(arg->buf, RSTRING_PTR(tmp)+need_len, buflen);
1469 arg->buflen = buflen;
1470 }
1471 else {
1472 arg->buflen = 0;
1473 }
1474 arg->offset = 0;
1475 }
1476
1477 return str;
1478}
1479
1480#define r_bytes(arg) r_bytes0(r_long(arg), (arg))
1481
1482static VALUE
1483r_bytes0(long len, struct load_arg *arg)
1484{
1485 VALUE str;
1486
1487 if (len == 0) return rb_str_new(0, 0);
1488 if (RB_TYPE_P(arg->src, T_STRING)) {
1489 if (RSTRING_LEN(arg->src) - arg->offset >= len) {
1490 str = rb_str_new(RSTRING_PTR(arg->src)+arg->offset, len);
1491 arg->offset += len;
1492 }
1493 else {
1494 too_short();
1495 }
1496 }
1497 else {
1498 if (arg->readable > 0 || arg->buflen > 0) {
1499 str = r_bytes1_buffered(len, arg);
1500 }
1501 else {
1502 str = r_bytes1(len, arg);
1503 }
1504 }
1505 return str;
1506}
1507
1508static inline int
1509name_equal(const char *name, size_t nlen, const char *p, long l)
1510{
1511 if ((size_t)l != nlen || *p != *name) return 0;
1512 return nlen == 1 || memcmp(p+1, name+1, nlen-1) == 0;
1513}
1514
1515static int
1516sym2encidx(VALUE sym, VALUE val)
1517{
1518 static const char name_encoding[8] = "encoding";
1519 const char *p;
1520 long l;
1521 if (rb_enc_get_index(sym) != ENCINDEX_US_ASCII) return -1;
1522 RSTRING_GETMEM(sym, p, l);
1523 if (l <= 0) return -1;
1524 if (name_equal(name_encoding, sizeof(name_encoding), p, l)) {
1525 int idx = rb_enc_find_index(StringValueCStr(val));
1526 return idx;
1527 }
1528 if (name_equal(name_s_encoding_short, rb_strlen_lit(name_s_encoding_short), p, l)) {
1529 if (val == Qfalse) return rb_usascii_encindex();
1530 else if (val == Qtrue) return rb_utf8_encindex();
1531 /* bogus ignore */
1532 }
1533 return -1;
1534}
1535
1536static int
1537symname_equal(VALUE sym, const char *name, size_t nlen)
1538{
1539 const char *p;
1540 long l;
1541 if (rb_enc_get_index(sym) != ENCINDEX_US_ASCII) return 0;
1542 RSTRING_GETMEM(sym, p, l);
1543 return name_equal(name, nlen, p, l);
1544}
1545
1546#define BUILD_ASSERT_POSITIVE(n) \
1547 /* make 0 negative to workaround the "zero size array" GCC extension, */ \
1548 ((sizeof(char [2*(ssize_t)(n)-1])+1)/2) /* assuming no overflow */
1549#define symname_equal_lit(sym, sym_name) \
1550 symname_equal(sym, sym_name, BUILD_ASSERT_POSITIVE(rb_strlen_lit(sym_name)))
1551
1552static VALUE
1553r_symlink(struct load_arg *arg)
1554{
1555 st_data_t sym;
1556 long num = r_long(arg);
1557
1558 if (!st_lookup(arg->symbols, num, &sym)) {
1559 rb_raise(rb_eArgError, "bad symbol");
1560 }
1561 return (VALUE)sym;
1562}
1563
1564static VALUE
1565r_symreal(struct load_arg *arg, int ivar)
1566{
1567 VALUE s = r_bytes(arg);
1568 VALUE sym;
1569 int idx = -1;
1570 st_index_t n = arg->symbols->num_entries;
1571
1572 if (rb_enc_str_asciionly_p(s)) rb_enc_associate_index(s, ENCINDEX_US_ASCII);
1573 st_insert(arg->symbols, (st_data_t)n, (st_data_t)s);
1574 if (ivar) {
1575 long num = r_long(arg);
1576 while (num-- > 0) {
1577 sym = r_symbol(arg);
1578 idx = sym2encidx(sym, r_object(arg));
1579 }
1580 }
1581 if (idx > 0) {
1582 rb_enc_associate_index(s, idx);
1583 if (is_broken_string(s)) {
1584 rb_raise(rb_eArgError, "invalid byte sequence in %s: %+"PRIsVALUE,
1585 rb_enc_name(rb_enc_from_index(idx)), s);
1586 }
1587 }
1588
1589 return s;
1590}
1591
1592static VALUE
1593r_symbol(struct load_arg *arg)
1594{
1595 int type, ivar = 0;
1596
1597 again:
1598 switch ((type = r_byte(arg))) {
1599 default:
1600 rb_raise(rb_eArgError, "dump format error for symbol(0x%x)", type);
1601 case TYPE_IVAR:
1602 ivar = 1;
1603 goto again;
1604 case TYPE_SYMBOL:
1605 return r_symreal(arg, ivar);
1606 case TYPE_SYMLINK:
1607 if (ivar) {
1608 rb_raise(rb_eArgError, "dump format error (symlink with encoding)");
1609 }
1610 return r_symlink(arg);
1611 }
1612}
1613
1614static VALUE
1615r_unique(struct load_arg *arg)
1616{
1617 return r_symbol(arg);
1618}
1619
1620static VALUE
1621r_string(struct load_arg *arg)
1622{
1623 return r_bytes(arg);
1624}
1625
1626static VALUE
1627r_entry0(VALUE v, st_index_t num, struct load_arg *arg)
1628{
1629 st_data_t real_obj = (st_data_t)v;
1630 if (arg->compat_tbl) {
1631 /* real_obj is kept if not found */
1632 st_lookup(arg->compat_tbl, v, &real_obj);
1633 }
1634 st_insert(arg->data, num, real_obj);
1635 st_insert(arg->partial_objects, (st_data_t)real_obj, Qtrue);
1636 return v;
1637}
1638
1639static VALUE
1640r_fixup_compat(VALUE v, struct load_arg *arg)
1641{
1642 st_data_t data;
1643 st_data_t key = (st_data_t)v;
1644 if (arg->compat_tbl && st_delete(arg->compat_tbl, &key, &data)) {
1645 VALUE real_obj = (VALUE)data;
1646 rb_alloc_func_t allocator = rb_get_alloc_func(CLASS_OF(real_obj));
1647 if (st_lookup(compat_allocator_tbl, (st_data_t)allocator, &data)) {
1648 marshal_compat_t *compat = (marshal_compat_t*)data;
1649 compat->loader(real_obj, v);
1650 }
1651 v = real_obj;
1652 }
1653 return v;
1654}
1655
1656static VALUE
1657r_post_proc(VALUE v, struct load_arg *arg)
1658{
1659 if (arg->proc) {
1660 v = load_funcall(arg, arg->proc, s_call, 1, &v);
1661 }
1662 return v;
1663}
1664
1665static VALUE
1666r_leave(VALUE v, struct load_arg *arg, bool partial)
1667{
1668 v = r_fixup_compat(v, arg);
1669 if (!partial) {
1670 st_data_t data;
1671 st_data_t key = (st_data_t)v;
1672 st_delete(arg->partial_objects, &key, &data);
1673 if (arg->freeze) {
1674 if (RB_TYPE_P(v, T_MODULE) || RB_TYPE_P(v, T_CLASS)) {
1675 // noop
1676 }
1677 else if (RB_TYPE_P(v, T_STRING)) {
1678 v = rb_str_to_interned_str(v);
1679 }
1680 else {
1681 OBJ_FREEZE(v);
1682 }
1683 }
1684 v = r_post_proc(v, arg);
1685 }
1686 return v;
1687}
1688
1689static int
1690copy_ivar_i(ID vid, VALUE value, st_data_t arg)
1691{
1692 VALUE obj = (VALUE)arg;
1693
1694 if (!rb_ivar_defined(obj, vid))
1695 rb_ivar_set(obj, vid, value);
1696 return ST_CONTINUE;
1697}
1698
1699static VALUE
1700r_copy_ivar(VALUE v, VALUE data)
1701{
1702 rb_ivar_foreach(data, copy_ivar_i, (st_data_t)v);
1703 return v;
1704}
1705
1706#define override_ivar_error(type, str) \
1707 rb_raise(rb_eTypeError, \
1708 "can't override instance variable of "type" '%"PRIsVALUE"'", \
1709 (str))
1710
1711static void
1712r_ivar(VALUE obj, int *has_encoding, struct load_arg *arg)
1713{
1714 long len;
1715
1716 len = r_long(arg);
1717 if (len > 0) {
1718 if (RB_TYPE_P(obj, T_MODULE)) {
1719 override_ivar_error("module", rb_mod_name(obj));
1720 }
1721 else if (RB_TYPE_P(obj, T_CLASS)) {
1722 override_ivar_error("class", rb_class_name(obj));
1723 }
1724 do {
1725 VALUE sym = r_symbol(arg);
1726 VALUE val = r_object(arg);
1727 int idx = sym2encidx(sym, val);
1728 if (idx >= 0) {
1729 if (rb_enc_capable(obj)) {
1730 rb_enc_associate_index(obj, idx);
1731 }
1732 else {
1733 rb_raise(rb_eArgError, "%"PRIsVALUE" is not enc_capable", obj);
1734 }
1735 if (has_encoding) *has_encoding = TRUE;
1736 }
1737 else if (symname_equal_lit(sym, name_s_ruby2_keywords_flag)) {
1738 if (RB_TYPE_P(obj, T_HASH)) {
1739 rb_hash_ruby2_keywords(obj);
1740 }
1741 else {
1742 rb_raise(rb_eArgError, "ruby2_keywords flag is given but %"PRIsVALUE" is not a Hash", obj);
1743 }
1744 }
1745 else {
1746 rb_ivar_set(obj, rb_intern_str(sym), val);
1747 }
1748 } while (--len > 0);
1749 }
1750}
1751
1752static VALUE
1753path2class(VALUE path)
1754{
1755 VALUE v = rb_path_to_class(path);
1756
1757 if (!RB_TYPE_P(v, T_CLASS)) {
1758 rb_raise(rb_eArgError, "%"PRIsVALUE" does not refer to class", path);
1759 }
1760 return v;
1761}
1762
1763#define path2module(path) must_be_module(rb_path_to_class(path), path)
1764
1765static VALUE
1766must_be_module(VALUE v, VALUE path)
1767{
1768 if (!RB_TYPE_P(v, T_MODULE)) {
1769 rb_raise(rb_eArgError, "%"PRIsVALUE" does not refer to module", path);
1770 }
1771 return v;
1772}
1773
1774static VALUE
1775obj_alloc_by_klass(VALUE klass, struct load_arg *arg, VALUE *oldclass)
1776{
1777 st_data_t data;
1778 rb_alloc_func_t allocator;
1779
1780 allocator = rb_get_alloc_func(klass);
1781 if (st_lookup(compat_allocator_tbl, (st_data_t)allocator, &data)) {
1782 marshal_compat_t *compat = (marshal_compat_t*)data;
1783 VALUE real_obj = rb_obj_alloc(klass);
1784 VALUE obj = rb_obj_alloc(compat->oldclass);
1785 if (oldclass) *oldclass = compat->oldclass;
1786
1787 if (!arg->compat_tbl) {
1788 arg->compat_tbl = rb_init_identtable();
1789 }
1790 st_insert(arg->compat_tbl, (st_data_t)obj, (st_data_t)real_obj);
1791 return obj;
1792 }
1793
1794 return rb_obj_alloc(klass);
1795}
1796
1797static VALUE
1798obj_alloc_by_path(VALUE path, struct load_arg *arg)
1799{
1800 return obj_alloc_by_klass(path2class(path), arg, 0);
1801}
1802
1803static VALUE
1804append_extmod(VALUE obj, VALUE extmod)
1805{
1806 long i = RARRAY_LEN(extmod);
1807 while (i > 0) {
1808 VALUE m = RARRAY_AREF(extmod, --i);
1809 rb_extend_object(obj, m);
1810 }
1811 return obj;
1812}
1813
1814#define prohibit_ivar(type, str) do { \
1815 if (!ivp || !*ivp) break; \
1816 override_ivar_error(type, str); \
1817 } while (0)
1818
1819static VALUE r_object_for(struct load_arg *arg, bool partial, int *ivp, VALUE extmod, int type);
1820
1821static VALUE
1822r_object0(struct load_arg *arg, bool partial, int *ivp, VALUE extmod)
1823{
1824 int type = r_byte(arg);
1825 return r_object_for(arg, partial, ivp, extmod, type);
1826}
1827
1828static VALUE
1829r_object_for(struct load_arg *arg, bool partial, int *ivp, VALUE extmod, int type)
1830{
1831 VALUE (*hash_new_with_size)(st_index_t) = rb_hash_new_with_size;
1832 VALUE v = Qnil;
1833 long id;
1834 st_data_t link;
1835
1836 switch (type) {
1837 case TYPE_LINK:
1838 id = r_long(arg);
1839 if (!st_lookup(arg->data, (st_data_t)id, &link)) {
1840 rb_raise(rb_eArgError, "dump format error (unlinked)");
1841 }
1842 v = (VALUE)link;
1843 if (!st_lookup(arg->partial_objects, (st_data_t)v, &link)) {
1844 v = r_post_proc(v, arg);
1845 }
1846 break;
1847
1848 case TYPE_IVAR:
1849 {
1850 int ivar = TRUE;
1851 v = r_object0(arg, true, &ivar, extmod);
1852 if (ivar) r_ivar(v, NULL, arg);
1853 v = r_leave(v, arg, partial);
1854 }
1855 break;
1856
1857 case TYPE_EXTENDED:
1858 {
1859 VALUE path = r_unique(arg);
1860 VALUE m = rb_path_to_class(path);
1861 if (NIL_P(extmod)) extmod = rb_ary_hidden_new(0);
1862
1863 if (RB_TYPE_P(m, T_CLASS)) { /* prepended */
1864 VALUE c;
1865
1866 v = r_object0(arg, true, 0, Qnil);
1867 c = CLASS_OF(v);
1868 if (c != m || FL_TEST(c, FL_SINGLETON)) {
1869 rb_raise(rb_eArgError,
1870 "prepended class %"PRIsVALUE" differs from class %"PRIsVALUE,
1871 path, rb_class_name(c));
1872 }
1873 c = rb_singleton_class(v);
1874 while (RARRAY_LEN(extmod) > 0) {
1875 m = rb_ary_pop(extmod);
1876 rb_prepend_module(c, m);
1877 }
1878 }
1879 else {
1880 must_be_module(m, path);
1881 rb_ary_push(extmod, m);
1882
1883 v = r_object0(arg, true, 0, extmod);
1884 while (RARRAY_LEN(extmod) > 0) {
1885 m = rb_ary_pop(extmod);
1886 rb_extend_object(v, m);
1887 }
1888 }
1889 v = r_leave(v, arg, partial);
1890 }
1891 break;
1892
1893 case TYPE_UCLASS:
1894 {
1895 VALUE c = path2class(r_unique(arg));
1896
1897 if (FL_TEST(c, FL_SINGLETON)) {
1898 rb_raise(rb_eTypeError, "singleton can't be loaded");
1899 }
1900 type = r_byte(arg);
1901 if ((c == rb_cHash) &&
1902 /* Hack for compare_by_identify */
1903 (type == TYPE_HASH || type == TYPE_HASH_DEF)) {
1904 hash_new_with_size = rb_ident_hash_new_with_size;
1905 goto type_hash;
1906 }
1907 v = r_object_for(arg, partial, 0, extmod, type);
1908 if (RB_SPECIAL_CONST_P(v) || RB_TYPE_P(v, T_OBJECT) || RB_TYPE_P(v, T_CLASS)) {
1909 goto format_error;
1910 }
1911 if (RB_TYPE_P(v, T_MODULE) || !RTEST(rb_class_inherited_p(c, RBASIC(v)->klass))) {
1912 VALUE tmp = rb_obj_alloc(c);
1913
1914 if (TYPE(v) != TYPE(tmp)) goto format_error;
1915 }
1916 RBASIC_SET_CLASS(v, c);
1917 }
1918 break;
1919
1920 format_error:
1921 rb_raise(rb_eArgError, "dump format error (user class)");
1922
1923 case TYPE_NIL:
1924 v = Qnil;
1925 v = r_leave(v, arg, false);
1926 break;
1927
1928 case TYPE_TRUE:
1929 v = Qtrue;
1930 v = r_leave(v, arg, false);
1931 break;
1932
1933 case TYPE_FALSE:
1934 v = Qfalse;
1935 v = r_leave(v, arg, false);
1936 break;
1937
1938 case TYPE_FIXNUM:
1939 {
1940 long i = r_long(arg);
1941 v = LONG2FIX(i);
1942 }
1943 v = r_leave(v, arg, false);
1944 break;
1945
1946 case TYPE_FLOAT:
1947 {
1948 double d;
1949 VALUE str = r_bytes(arg);
1950 const char *ptr = RSTRING_PTR(str);
1951
1952 if (strcmp(ptr, "nan") == 0) {
1953 d = nan("");
1954 }
1955 else if (strcmp(ptr, "inf") == 0) {
1956 d = HUGE_VAL;
1957 }
1958 else if (strcmp(ptr, "-inf") == 0) {
1959 d = -HUGE_VAL;
1960 }
1961 else {
1962 char *e;
1963 d = strtod(ptr, &e);
1964 d = load_mantissa(d, e, RSTRING_LEN(str) - (e - ptr));
1965 }
1966 v = DBL2NUM(d);
1967 v = r_entry(v, arg);
1968 v = r_leave(v, arg, false);
1969 }
1970 break;
1971
1972 case TYPE_BIGNUM:
1973 {
1974 long len;
1975 VALUE data;
1976 int sign;
1977
1978 sign = r_byte(arg);
1979 len = r_long(arg);
1980
1981 if (SIZEOF_VALUE >= 8 && len <= 4) {
1982 // Representable within uintptr, likely FIXNUM
1983 VALUE num = 0;
1984 for (int i = 0; i < len; i++) {
1985 num |= (VALUE)r_byte(arg) << (i * 16);
1986 num |= (VALUE)r_byte(arg) << (i * 16 + 8);
1987 }
1988#if SIZEOF_VALUE == SIZEOF_LONG
1989 v = ULONG2NUM(num);
1990#else
1991 v = ULL2NUM(num);
1992#endif
1993 if (sign == '-') {
1994 v = rb_int_uminus(v);
1995 }
1996 }
1997 else {
1998 data = r_bytes0(len * 2, arg);
1999 v = rb_integer_unpack(RSTRING_PTR(data), len, 2, 0,
2000 INTEGER_PACK_LITTLE_ENDIAN | (sign == '-' ? INTEGER_PACK_NEGATIVE : 0));
2001 rb_str_resize(data, 0L);
2002 }
2003 v = r_entry(v, arg);
2004 v = r_leave(v, arg, false);
2005 }
2006 break;
2007
2008 case TYPE_STRING:
2009 v = r_entry(r_string(arg), arg);
2010 v = r_leave(v, arg, partial);
2011 break;
2012
2013 case TYPE_REGEXP:
2014 {
2015 VALUE str = r_bytes(arg);
2016 int options = r_byte(arg);
2017 int has_encoding = FALSE;
2018 st_index_t idx = r_prepare(arg);
2019
2020 if (ivp) {
2021 r_ivar(str, &has_encoding, arg);
2022 *ivp = FALSE;
2023 }
2024 if (!has_encoding) {
2025 /* 1.8 compatibility; remove escapes undefined in 1.8 */
2026 char *ptr = RSTRING_PTR(str), *dst = ptr, *src = ptr;
2027 long len = RSTRING_LEN(str);
2028 long bs = 0;
2029 for (; len-- > 0; *dst++ = *src++) {
2030 switch (*src) {
2031 case '\\': bs++; break;
2032 case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
2033 case 'm': case 'o': case 'p': case 'q': case 'u': case 'y':
2034 case 'E': case 'F': case 'H': case 'I': case 'J': case 'K':
2035 case 'L': case 'N': case 'O': case 'P': case 'Q': case 'R':
2036 case 'S': case 'T': case 'U': case 'V': case 'X': case 'Y':
2037 if (bs & 1) --dst;
2038 /* fall through */
2039 default: bs = 0; break;
2040 }
2041 }
2042 rb_str_set_len(str, dst - ptr);
2043 }
2044 VALUE regexp = rb_reg_new_str(str, options);
2045 r_copy_ivar(regexp, str);
2046
2047 v = r_entry0(regexp, idx, arg);
2048 v = r_leave(v, arg, partial);
2049 }
2050 break;
2051
2052 case TYPE_ARRAY:
2053 {
2054 long len = r_long(arg);
2055
2056 v = rb_ary_new2(len);
2057 v = r_entry(v, arg);
2058 arg->readable += len - 1;
2059 while (len--) {
2060 rb_ary_push(v, r_object(arg));
2061 arg->readable--;
2062 }
2063 v = r_leave(v, arg, partial);
2064 arg->readable++;
2065 }
2066 break;
2067
2068 case TYPE_HASH:
2069 case TYPE_HASH_DEF:
2070 type_hash:
2071 {
2072 long len = r_long(arg);
2073
2074 v = hash_new_with_size(len);
2075 v = r_entry(v, arg);
2076 arg->readable += (len - 1) * 2;
2077 while (len--) {
2078 VALUE key = r_object(arg);
2079 VALUE value = r_object(arg);
2080 rb_hash_aset(v, key, value);
2081 arg->readable -= 2;
2082 }
2083 arg->readable += 2;
2084 if (type == TYPE_HASH_DEF) {
2085 RHASH_SET_IFNONE(v, r_object(arg));
2086 }
2087 v = r_leave(v, arg, partial);
2088 }
2089 break;
2090
2091 case TYPE_STRUCT:
2092 {
2093 VALUE mem, values;
2094 long i;
2095 VALUE slot;
2096 st_index_t idx = r_prepare(arg);
2097 VALUE klass = path2class(r_unique(arg));
2098 long len = r_long(arg);
2099
2100 v = rb_obj_alloc(klass);
2101 if (!RB_TYPE_P(v, T_STRUCT)) {
2102 rb_raise(rb_eTypeError, "class %"PRIsVALUE" not a struct", rb_class_name(klass));
2103 }
2104 mem = rb_struct_s_members(klass);
2105 if (RARRAY_LEN(mem) != len) {
2106 rb_raise(rb_eTypeError, "struct %"PRIsVALUE" not compatible (struct size differs)",
2107 rb_class_name(klass));
2108 }
2109
2110 arg->readable += (len - 1) * 2;
2111 v = r_entry0(v, idx, arg);
2112 values = rb_ary_new2(len);
2113 {
2114 VALUE keywords = Qfalse;
2115 if (RTEST(rb_struct_s_keyword_init(klass))) {
2116 keywords = rb_hash_new();
2117 rb_ary_push(values, keywords);
2118 }
2119
2120 for (i=0; i<len; i++) {
2121 VALUE n = rb_sym2str(RARRAY_AREF(mem, i));
2122 slot = r_symbol(arg);
2123
2124 if (!rb_str_equal(n, slot)) {
2125 rb_raise(rb_eTypeError, "struct %"PRIsVALUE" not compatible (:%"PRIsVALUE" for :%"PRIsVALUE")",
2126 rb_class_name(klass),
2127 slot, n);
2128 }
2129 if (keywords) {
2130 rb_hash_aset(keywords, RARRAY_AREF(mem, i), r_object(arg));
2131 }
2132 else {
2133 rb_ary_push(values, r_object(arg));
2134 }
2135 arg->readable -= 2;
2136 }
2137 }
2138 rb_struct_initialize(v, values);
2139 v = r_leave(v, arg, partial);
2140 arg->readable += 2;
2141 }
2142 break;
2143
2144 case TYPE_USERDEF:
2145 {
2146 VALUE name = r_unique(arg);
2147 VALUE klass = path2class(name);
2148 VALUE data;
2149 st_data_t d;
2150
2151 if (!rb_obj_respond_to(klass, s_load, TRUE)) {
2152 rb_raise(rb_eTypeError, "class %"PRIsVALUE" needs to have method '_load'",
2153 name);
2154 }
2155 data = r_string(arg);
2156 if (ivp) {
2157 r_ivar(data, NULL, arg);
2158 *ivp = FALSE;
2159 }
2160 v = load_funcall(arg, klass, s_load, 1, &data);
2161 v = r_entry(v, arg);
2162 if (st_lookup(compat_allocator_tbl, (st_data_t)rb_get_alloc_func(klass), &d)) {
2163 marshal_compat_t *compat = (marshal_compat_t*)d;
2164 v = compat->loader(klass, v);
2165 }
2166 if (!partial) {
2167 if (arg->freeze) {
2168 OBJ_FREEZE(v);
2169 }
2170 v = r_post_proc(v, arg);
2171 }
2172 }
2173 break;
2174
2175 case TYPE_USRMARSHAL:
2176 {
2177 VALUE name = r_unique(arg);
2178 VALUE klass = path2class(name);
2179 VALUE oldclass = 0;
2180 VALUE data;
2181
2182 v = obj_alloc_by_klass(klass, arg, &oldclass);
2183 if (!NIL_P(extmod)) {
2184 /* for the case marshal_load is overridden */
2185 append_extmod(v, extmod);
2186 }
2187 if (!rb_obj_respond_to(v, s_mload, TRUE)) {
2188 rb_raise(rb_eTypeError, "instance of %"PRIsVALUE" needs to have method 'marshal_load'",
2189 name);
2190 }
2191 v = r_entry(v, arg);
2192 data = r_object(arg);
2193 load_funcall(arg, v, s_mload, 1, &data);
2194 v = r_fixup_compat(v, arg);
2195 v = r_copy_ivar(v, data);
2196 if (arg->freeze) {
2197 OBJ_FREEZE(v);
2198 }
2199 v = r_post_proc(v, arg);
2200 if (!NIL_P(extmod)) {
2201 if (oldclass) append_extmod(v, extmod);
2202 rb_ary_clear(extmod);
2203 }
2204 }
2205 break;
2206
2207 case TYPE_OBJECT:
2208 {
2209 st_index_t idx = r_prepare(arg);
2210 v = obj_alloc_by_path(r_unique(arg), arg);
2211 if (!RB_TYPE_P(v, T_OBJECT)) {
2212 rb_raise(rb_eArgError, "dump format error");
2213 }
2214 v = r_entry0(v, idx, arg);
2215 r_ivar(v, NULL, arg);
2216 v = r_leave(v, arg, partial);
2217 }
2218 break;
2219
2220 case TYPE_DATA:
2221 {
2222 VALUE name = r_unique(arg);
2223 VALUE klass = path2class(name);
2224 VALUE oldclass = 0;
2225 VALUE r;
2226
2227 v = obj_alloc_by_klass(klass, arg, &oldclass);
2228 if (!RB_TYPE_P(v, T_DATA)) {
2229 rb_raise(rb_eArgError, "dump format error");
2230 }
2231 v = r_entry(v, arg);
2232 if (!rb_obj_respond_to(v, s_load_data, TRUE)) {
2233 rb_raise(rb_eTypeError,
2234 "class %"PRIsVALUE" needs to have instance method '_load_data'",
2235 name);
2236 }
2237 r = r_object0(arg, partial, 0, extmod);
2238 load_funcall(arg, v, s_load_data, 1, &r);
2239 v = r_leave(v, arg, partial);
2240 }
2241 break;
2242
2243 case TYPE_MODULE_OLD:
2244 {
2245 VALUE str = r_bytes(arg);
2246
2247 v = rb_path_to_class(str);
2248 prohibit_ivar("class/module", str);
2249 v = r_entry(v, arg);
2250 v = r_leave(v, arg, partial);
2251 }
2252 break;
2253
2254 case TYPE_CLASS:
2255 {
2256 VALUE str = r_bytes(arg);
2257
2258 v = path2class(str);
2259 prohibit_ivar("class", str);
2260 v = r_entry(v, arg);
2261 v = r_leave(v, arg, partial);
2262 }
2263 break;
2264
2265 case TYPE_MODULE:
2266 {
2267 VALUE str = r_bytes(arg);
2268
2269 v = path2module(str);
2270 prohibit_ivar("module", str);
2271 v = r_entry(v, arg);
2272 v = r_leave(v, arg, partial);
2273 }
2274 break;
2275
2276 case TYPE_SYMBOL:
2277 if (ivp) {
2278 v = r_symreal(arg, *ivp);
2279 *ivp = FALSE;
2280 }
2281 else {
2282 v = r_symreal(arg, 0);
2283 }
2284 v = rb_str_intern(v);
2285 v = r_leave(v, arg, partial);
2286 break;
2287
2288 case TYPE_SYMLINK:
2289 v = rb_str_intern(r_symlink(arg));
2290 break;
2291
2292 default:
2293 rb_raise(rb_eArgError, "dump format error(0x%x)", type);
2294 break;
2295 }
2296
2297 if (UNDEF_P(v)) {
2298 rb_raise(rb_eArgError, "dump format error (bad link)");
2299 }
2300
2301 return v;
2302}
2303
2304static VALUE
2305r_object(struct load_arg *arg)
2306{
2307 return r_object0(arg, false, 0, Qnil);
2308}
2309
2310static void
2311clear_load_arg(struct load_arg *arg)
2312{
2313 xfree(arg->buf);
2314 arg->buf = NULL;
2315 arg->buflen = 0;
2316 arg->offset = 0;
2317 arg->readable = 0;
2318 if (!arg->symbols) return;
2319 st_free_table(arg->symbols);
2320 arg->symbols = 0;
2321 st_free_table(arg->data);
2322 arg->data = 0;
2323 st_free_table(arg->partial_objects);
2324 arg->partial_objects = 0;
2325 if (arg->compat_tbl) {
2326 st_free_table(arg->compat_tbl);
2327 arg->compat_tbl = 0;
2328 }
2329}
2330
2331VALUE
2332rb_marshal_load_with_proc(VALUE port, VALUE proc, bool freeze)
2333{
2334 int major, minor;
2335 VALUE v;
2336 VALUE wrapper; /* used to avoid memory leak in case of exception */
2337 struct load_arg *arg;
2338
2339 v = rb_check_string_type(port);
2340 if (!NIL_P(v)) {
2341 port = v;
2342 }
2343 else if (rb_respond_to(port, s_getbyte) && rb_respond_to(port, s_read)) {
2344 rb_check_funcall(port, s_binmode, 0, 0);
2345 }
2346 else {
2347 io_needed();
2348 }
2349 wrapper = TypedData_Make_Struct(0, struct load_arg, &load_arg_data, arg);
2350 arg->src = port;
2351 arg->offset = 0;
2352 arg->symbols = st_init_numtable();
2353 arg->data = rb_init_identtable();
2354 arg->partial_objects = rb_init_identtable();
2355 arg->compat_tbl = 0;
2356 arg->proc = 0;
2357 arg->readable = 0;
2358 arg->freeze = freeze;
2359
2360 if (NIL_P(v))
2361 arg->buf = xmalloc(BUFSIZ);
2362 else
2363 arg->buf = 0;
2364
2365 major = r_byte(arg);
2366 minor = r_byte(arg);
2367 if (major != MARSHAL_MAJOR || minor > MARSHAL_MINOR) {
2368 clear_load_arg(arg);
2369 rb_raise(rb_eTypeError, "incompatible marshal file format (can't be read)\n\
2370\tformat version %d.%d required; %d.%d given",
2371 MARSHAL_MAJOR, MARSHAL_MINOR, major, minor);
2372 }
2373 if (RTEST(ruby_verbose) && minor != MARSHAL_MINOR) {
2374 rb_warn("incompatible marshal file format (can be read)\n\
2375\tformat version %d.%d required; %d.%d given",
2376 MARSHAL_MAJOR, MARSHAL_MINOR, major, minor);
2377 }
2378
2379 if (!NIL_P(proc)) arg->proc = proc;
2380 v = r_object(arg);
2381 clear_load_arg(arg);
2382 RB_GC_GUARD(wrapper);
2383
2384 return v;
2385}
2386
2387static VALUE
2388marshal_load(rb_execution_context_t *ec, VALUE mod, VALUE source, VALUE proc, VALUE freeze)
2389{
2390 return rb_marshal_load_with_proc(source, proc, RTEST(freeze));
2391}
2392
2393#include "marshal.rbinc"
2394
2395/*
2396 * The marshaling library converts collections of Ruby objects into a
2397 * byte stream, allowing them to be stored outside the currently
2398 * active script. This data may subsequently be read and the original
2399 * objects reconstituted.
2400 *
2401 * Marshaled data has major and minor version numbers stored along
2402 * with the object information. In normal use, marshaling can only
2403 * load data written with the same major version number and an equal
2404 * or lower minor version number. If Ruby's ``verbose'' flag is set
2405 * (normally using -d, -v, -w, or --verbose) the major and minor
2406 * numbers must match exactly. Marshal versioning is independent of
2407 * Ruby's version numbers. You can extract the version by reading the
2408 * first two bytes of marshaled data.
2409 *
2410 * str = Marshal.dump("thing")
2411 * RUBY_VERSION #=> "1.9.0"
2412 * str[0].ord #=> 4
2413 * str[1].ord #=> 8
2414 *
2415 * Some objects cannot be dumped: if the objects to be dumped include
2416 * bindings, procedure or method objects, instances of class IO, or
2417 * singleton objects, a TypeError will be raised.
2418 *
2419 * If your class has special serialization needs (for example, if you
2420 * want to serialize in some specific format), or if it contains
2421 * objects that would otherwise not be serializable, you can implement
2422 * your own serialization strategy.
2423 *
2424 * There are two methods of doing this, your object can define either
2425 * marshal_dump and marshal_load or _dump and _load. marshal_dump will take
2426 * precedence over _dump if both are defined. marshal_dump may result in
2427 * smaller Marshal strings.
2428 *
2429 * == Security considerations
2430 *
2431 * By design, Marshal.load can deserialize almost any class loaded into the
2432 * Ruby process. In many cases this can lead to remote code execution if the
2433 * Marshal data is loaded from an untrusted source.
2434 *
2435 * As a result, Marshal.load is not suitable as a general purpose serialization
2436 * format and you should never unmarshal user supplied input or other untrusted
2437 * data.
2438 *
2439 * If you need to deserialize untrusted data, use JSON or another serialization
2440 * format that is only able to load simple, 'primitive' types such as String,
2441 * Array, Hash, etc. Never allow user input to specify arbitrary types to
2442 * deserialize into.
2443 *
2444 * == marshal_dump and marshal_load
2445 *
2446 * When dumping an object the method marshal_dump will be called.
2447 * marshal_dump must return a result containing the information necessary for
2448 * marshal_load to reconstitute the object. The result can be any object.
2449 *
2450 * When loading an object dumped using marshal_dump the object is first
2451 * allocated then marshal_load is called with the result from marshal_dump.
2452 * marshal_load must recreate the object from the information in the result.
2453 *
2454 * Example:
2455 *
2456 * class MyObj
2457 * def initialize name, version, data
2458 * @name = name
2459 * @version = version
2460 * @data = data
2461 * end
2462 *
2463 * def marshal_dump
2464 * [@name, @version]
2465 * end
2466 *
2467 * def marshal_load array
2468 * @name, @version = array
2469 * end
2470 * end
2471 *
2472 * == _dump and _load
2473 *
2474 * Use _dump and _load when you need to allocate the object you're restoring
2475 * yourself.
2476 *
2477 * When dumping an object the instance method _dump is called with an Integer
2478 * which indicates the maximum depth of objects to dump (a value of -1 implies
2479 * that you should disable depth checking). _dump must return a String
2480 * containing the information necessary to reconstitute the object.
2481 *
2482 * The class method _load should take a String and use it to return an object
2483 * of the same class.
2484 *
2485 * Example:
2486 *
2487 * class MyObj
2488 * def initialize name, version, data
2489 * @name = name
2490 * @version = version
2491 * @data = data
2492 * end
2493 *
2494 * def _dump level
2495 * [@name, @version].join ':'
2496 * end
2497 *
2498 * def self._load args
2499 * new(*args.split(':'))
2500 * end
2501 * end
2502 *
2503 * Since Marshal.dump outputs a string you can have _dump return a Marshal
2504 * string which is Marshal.loaded in _load for complex objects.
2505 */
2506void
2507Init_marshal(void)
2508{
2509 VALUE rb_mMarshal = rb_define_module("Marshal");
2510#define set_id(sym) sym = rb_intern_const(name_##sym)
2511 set_id(s_dump);
2512 set_id(s_load);
2513 set_id(s_mdump);
2514 set_id(s_mload);
2515 set_id(s_dump_data);
2516 set_id(s_load_data);
2517 set_id(s_alloc);
2518 set_id(s_call);
2519 set_id(s_getbyte);
2520 set_id(s_read);
2521 set_id(s_write);
2522 set_id(s_binmode);
2523 set_id(s_encoding_short);
2524 set_id(s_ruby2_keywords_flag);
2525
2526 rb_define_module_function(rb_mMarshal, "dump", marshal_dump, -1);
2527
2528 /* major version */
2529 rb_define_const(rb_mMarshal, "MAJOR_VERSION", INT2FIX(MARSHAL_MAJOR));
2530 /* minor version */
2531 rb_define_const(rb_mMarshal, "MINOR_VERSION", INT2FIX(MARSHAL_MINOR));
2532}
2533
2534static int
2535marshal_compat_table_mark_i(st_data_t key, st_data_t value, st_data_t _)
2536{
2537 marshal_compat_t *p = (marshal_compat_t *)value;
2538 rb_gc_mark_movable(p->newclass);
2539 rb_gc_mark_movable(p->oldclass);
2540 return ST_CONTINUE;
2541}
2542
2543static void
2544marshal_compat_table_mark(void *tbl)
2545{
2546 if (!tbl) return;
2547 st_foreach(tbl, marshal_compat_table_mark_i, 0);
2548}
2549
2550static int
2551marshal_compat_table_free_i(st_data_t key, st_data_t value, st_data_t _)
2552{
2553 xfree((marshal_compat_t *)value);
2554 return ST_CONTINUE;
2555}
2556
2557static void
2558marshal_compat_table_free(void *data)
2559{
2560 st_foreach(data, marshal_compat_table_free_i, 0);
2561 st_free_table(data);
2562}
2563
2564static size_t
2565marshal_compat_table_memsize(const void *data)
2566{
2567 return st_memsize(data) + sizeof(marshal_compat_t) * st_table_size(data);
2568}
2569
2570static int
2571marshal_compat_table_compact_i(st_data_t key, st_data_t value, st_data_t _)
2572{
2573 marshal_compat_t *p = (marshal_compat_t *)value;
2574 p->newclass = rb_gc_location(p->newclass);
2575 p->oldclass = rb_gc_location(p->oldclass);
2576 return ST_CONTINUE;
2577}
2578
2579static void
2580marshal_compat_table_compact(void *tbl)
2581{
2582 if (!tbl) return;
2583 st_foreach(tbl, marshal_compat_table_compact_i, 0);
2584}
2585
2586static const rb_data_type_t marshal_compat_type = {
2587 .wrap_struct_name = "marshal_compat_table",
2588 .function = {
2589 .dmark = marshal_compat_table_mark,
2590 .dfree = marshal_compat_table_free,
2591 .dsize = marshal_compat_table_memsize,
2592 .dcompact = marshal_compat_table_compact,
2593 },
2594 .flags = RUBY_TYPED_WB_PROTECTED | RUBY_TYPED_FREE_IMMEDIATELY,
2595};
2596
2597static st_table *
2598compat_allocator_table(void)
2599{
2600 if (compat_allocator_tbl) return compat_allocator_tbl;
2601 compat_allocator_tbl = st_init_numtable();
2602 compat_allocator_tbl_wrapper =
2603 TypedData_Wrap_Struct(0, &marshal_compat_type, compat_allocator_tbl);
2604 rb_vm_register_global_object(compat_allocator_tbl_wrapper);
2605 return compat_allocator_tbl;
2606}
2607
2608VALUE
2609rb_marshal_dump(VALUE obj, VALUE port)
2610{
2611 return rb_marshal_dump_limited(obj, port, -1);
2612}
2613
2614VALUE
2615rb_marshal_load(VALUE port)
2616{
2617 return rb_marshal_load_with_proc(port, Qnil, false);
2618}
builtin.h
Defines RBIMPL_HAS_BUILTIN.
len
int len
Length of the buffer.
Definition io.h:8
Generated by doxygen 1.9.8