1
0
Fork 0
mirror of https://github.com/ruby/ruby.git synced 2022-11-09 12:17:21 -05:00
ruby--ruby/marshal.c

1905 lines
44 KiB
C
Raw Normal View History

/**********************************************************************
marshal.c -
$Author$
created at: Thu Apr 27 16:30:01 JST 1995
* encoding.c: provide basic features for M17N. * parse.y: encoding aware parsing. * parse.y (pragma_encoding): encoding specification pragma. * parse.y (rb_intern3): encoding specified symbols. * string.c (rb_str_length): length based on characters. for older behavior, bytesize method added. * string.c (rb_str_index_m): index based on characters. rindex as well. * string.c (succ_char): encoding aware succeeding string. * string.c (rb_str_reverse): reverse based on characters. * string.c (rb_str_inspect): encoding aware string description. * string.c (rb_str_upcase_bang): encoding aware case conversion. downcase, capitalize, swapcase as well. * string.c (rb_str_tr_bang): tr based on characters. delete, squeeze, tr_s, count as well. * string.c (rb_str_split_m): split based on characters. * string.c (rb_str_each_line): encoding aware each_line. * string.c (rb_str_each_char): added. iteration based on characters. * string.c (rb_str_strip_bang): encoding aware whitespace stripping. lstrip, rstrip as well. * string.c (rb_str_justify): encoding aware justifying (ljust, rjust, center). * string.c (str_encoding): get encoding attribute from a string. * re.c (rb_reg_initialize): encoding aware regular expression * sprintf.c (rb_str_format): formatting (i.e. length count) based on characters. * io.c (rb_io_getc): getc to return one-character string. for older behavior, getbyte method added. * ext/stringio/stringio.c (strio_getc): ditto. * io.c (rb_io_ungetc): allow pushing arbitrary string at the current reading point. * ext/stringio/stringio.c (strio_ungetc): ditto. * ext/strscan/strscan.c: encoding support. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@13261 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2007-08-24 23:29:39 -04:00
Copyright (C) 1993-2007 Yukihiro Matsumoto
**********************************************************************/
#include "ruby/ruby.h"
#include "ruby/io.h"
#include "ruby/st.h"
#include "ruby/util.h"
#include "ruby/encoding.h"
#include <math.h>
#ifdef HAVE_FLOAT_H
#include <float.h>
#endif
#ifdef HAVE_IEEEFP_H
#include <ieeefp.h>
#endif
#define BITSPERSHORT (2*CHAR_BIT)
#define SHORTMASK ((1<<BITSPERSHORT)-1)
#define SHORTDN(x) RSHIFT(x,BITSPERSHORT)
#if SIZEOF_SHORT == SIZEOF_BDIGITS
#define SHORTLEN(x) (x)
#else
static long
shortlen(long len, BDIGIT *ds)
{
BDIGIT num;
int offset = 0;
num = ds[len-1];
while (num) {
num = SHORTDN(num);
offset++;
}
return (len - 1)*sizeof(BDIGIT)/2 + offset;
}
#define SHORTLEN(x) shortlen((x),d)
#endif
#define MARSHAL_MAJOR 4
#define MARSHAL_MINOR 8
#define TYPE_NIL '0'
#define TYPE_TRUE 'T'
#define TYPE_FALSE 'F'
#define TYPE_FIXNUM 'i'
#define TYPE_EXTENDED 'e'
#define TYPE_UCLASS 'C'
#define TYPE_OBJECT 'o'
#define TYPE_DATA 'd'
#define TYPE_USERDEF 'u'
#define TYPE_USRMARSHAL 'U'
#define TYPE_FLOAT 'f'
#define TYPE_BIGNUM 'l'
#define TYPE_STRING '"'
#define TYPE_REGEXP '/'
#define TYPE_ARRAY '['
#define TYPE_HASH '{'
#define TYPE_HASH_DEF '}'
#define TYPE_STRUCT 'S'
#define TYPE_MODULE_OLD 'M'
#define TYPE_CLASS 'c'
#define TYPE_MODULE 'm'
#define TYPE_SYMBOL ':'
#define TYPE_SYMLINK ';'
#define TYPE_IVAR 'I'
#define TYPE_LINK '@'
static ID s_dump, s_load, s_mdump, s_mload;
static ID s_dump_data, s_load_data, s_alloc, s_call;
static ID s_getbyte, s_read, s_write, s_binmode;
ID rb_id_encoding(void);
typedef struct {
VALUE newclass;
VALUE oldclass;
VALUE (*dumper)(VALUE);
VALUE (*loader)(VALUE, VALUE);
} marshal_compat_t;
static st_table *compat_allocator_tbl;
static VALUE compat_allocator_tbl_wrapper;
static int
mark_marshal_compat_i(st_data_t key, st_data_t value)
{
marshal_compat_t *p = (marshal_compat_t *)value;
rb_gc_mark(p->newclass);
rb_gc_mark(p->oldclass);
return ST_CONTINUE;
}
static void
mark_marshal_compat_t(void *tbl)
{
if (!tbl) return;
st_foreach(tbl, mark_marshal_compat_i, 0);
}
void
rb_marshal_define_compat(VALUE newclass, VALUE oldclass, VALUE (*dumper)(VALUE), VALUE (*loader)(VALUE, VALUE))
{
marshal_compat_t *compat;
rb_alloc_func_t allocator = rb_get_alloc_func(newclass);
if (!allocator) {
rb_raise(rb_eTypeError, "no allocator");
}
compat = ALLOC(marshal_compat_t);
compat->newclass = Qnil;
compat->oldclass = Qnil;
compat->newclass = newclass;
compat->oldclass = oldclass;
compat->dumper = dumper;
compat->loader = loader;
st_insert(compat_allocator_tbl, (st_data_t)allocator, (st_data_t)compat);
}
struct dump_arg {
VALUE str, dest;
st_table *symbols;
st_table *data;
int taint;
int untrust;
st_table *compat_tbl;
st_table *encodings;
};
struct dump_call_arg {
VALUE obj;
struct dump_arg *arg;
int limit;
};
static void
check_dump_arg(struct dump_arg *arg, ID sym)
{
if (!arg->symbols) {
rb_raise(rb_eRuntimeError, "Marshal.dump reentered at %s",
rb_id2name(sym));
}
}
static void clear_dump_arg(struct dump_arg *arg);
static void
mark_dump_arg(void *ptr)
{
struct dump_arg *p = ptr;
if (!p->symbols)
return;
rb_mark_set(p->data);
rb_mark_hash(p->compat_tbl);
rb_gc_mark(p->str);
}
static void
free_dump_arg(void *ptr)
{
clear_dump_arg(ptr);
xfree(ptr);
}
static size_t
memsize_dump_arg(const void *ptr)
{
return ptr ? sizeof(struct dump_arg) : 0;
}
static const rb_data_type_t dump_arg_data = {
"dump_arg",
mark_dump_arg, free_dump_arg, memsize_dump_arg
};
static const char *
must_not_be_anonymous(const char *type, VALUE path)
{
char *n = RSTRING_PTR(path);
if (!rb_enc_asciicompat(rb_enc_get(path))) {
/* cannot occur? */
rb_raise(rb_eTypeError, "can't dump non-ascii %s name", type);
}
if (n[0] == '#') {
rb_raise(rb_eTypeError, "can't dump anonymous %s %.*s", type,
(int)RSTRING_LEN(path), n);
}
return n;
}
static VALUE
class2path(VALUE klass)
{
VALUE path = rb_class_path(klass);
const char *n;
n = must_not_be_anonymous((TYPE(klass) == T_CLASS ? "class" : "module"), path);
if (rb_path_to_class(path) != rb_class_real(klass)) {
rb_raise(rb_eTypeError, "%s can't be referred to", n);
}
return path;
}
static void w_long(long, struct dump_arg*);
static void w_encoding(VALUE obj, long num, struct dump_call_arg *arg);
static void
w_nbyte(const char *s, long n, struct dump_arg *arg)
{
VALUE buf = arg->str;
rb_str_buf_cat(buf, s, n);
if (arg->dest && RSTRING_LEN(buf) >= BUFSIZ) {
if (arg->taint) OBJ_TAINT(buf);
if (arg->untrust) OBJ_UNTRUST(buf);
rb_io_write(arg->dest, buf);
rb_str_resize(buf, 0);
}
}
static void
w_byte(char c, struct dump_arg *arg)
{
w_nbyte(&c, 1, arg);
}
static void
w_bytes(const char *s, long n, struct dump_arg *arg)
{
w_long(n, arg);
w_nbyte(s, n, arg);
}
static void
w_short(int x, struct dump_arg *arg)
{
w_byte((char)((x >> 0) & 0xff), arg);
w_byte((char)((x >> 8) & 0xff), arg);
}
static void
w_long(long x, struct dump_arg *arg)
{
char buf[sizeof(long)+1];
int i, len = 0;
#if SIZEOF_LONG > 4
if (!(RSHIFT(x, 31) == 0 || RSHIFT(x, 31) == -1)) {
/* big long does not fit in 4 bytes */
rb_raise(rb_eTypeError, "long too big to dump");
}
#endif
if (x == 0) {
w_byte(0, arg);
return;
}
if (0 < x && x < 123) {
w_byte((char)(x + 5), arg);
return;
}
if (-124 < x && x < 0) {
w_byte((char)((x - 5)&0xff), arg);
return;
}
for (i=1;i<(int)sizeof(long)+1;i++) {
buf[i] = (char)(x & 0xff);
x = RSHIFT(x,8);
if (x == 0) {
buf[0] = i;
break;
}
if (x == -1) {
buf[0] = -i;
break;
}
}
len = i;
for (i=0;i<=len;i++) {
w_byte(buf[i], arg);
}
}
#ifdef DBL_MANT_DIG
#define DECIMAL_MANT (53-16) /* from IEEE754 double precision */
#if DBL_MANT_DIG > 32
#define MANT_BITS 32
#elif DBL_MANT_DIG > 24
#define MANT_BITS 24
#elif DBL_MANT_DIG > 16
#define MANT_BITS 16
#else
#define MANT_BITS 8
#endif
static int
save_mantissa(double d, char *buf)
{
int e, i = 0;
unsigned long m;
double n;
d = modf(ldexp(frexp(fabs(d), &e), DECIMAL_MANT), &d);
if (d > 0) {
buf[i++] = 0;
do {
d = modf(ldexp(d, MANT_BITS), &n);
m = (unsigned long)n;
#if MANT_BITS > 24
buf[i++] = (char)(m >> 24);
#endif
#if MANT_BITS > 16
buf[i++] = (char)(m >> 16);
#endif
#if MANT_BITS > 8
buf[i++] = (char)(m >> 8);
#endif
buf[i++] = (char)m;
} while (d > 0);
while (!buf[i - 1]) --i;
}
return i;
}
static double
load_mantissa(double d, const char *buf, long len)
{
if (!len) return d;
if (--len > 0 && !*buf++) { /* binary mantissa mark */
int e, s = d < 0, dig = 0;
unsigned long m;
modf(ldexp(frexp(fabs(d), &e), DECIMAL_MANT), &d);
do {
m = 0;
switch (len) {
default: m = *buf++ & 0xff;
#if MANT_BITS > 24
case 3: m = (m << 8) | (*buf++ & 0xff);
#endif
#if MANT_BITS > 16
case 2: m = (m << 8) | (*buf++ & 0xff);
#endif
#if MANT_BITS > 8
case 1: m = (m << 8) | (*buf++ & 0xff);
#endif
}
dig -= len < MANT_BITS / 8 ? 8 * (unsigned)len : MANT_BITS;
d += ldexp((double)m, dig);
} while ((len -= MANT_BITS / 8) > 0);
d = ldexp(d, e - DECIMAL_MANT);
if (s) d = -d;
}
return d;
}
#else
#define load_mantissa(d, buf, len) (d)
#define save_mantissa(d, buf) 0
#endif
#ifdef DBL_DIG
#define FLOAT_DIG (DBL_DIG+2)
#else
#define FLOAT_DIG 17
#endif
static void
w_float(double d, struct dump_arg *arg)
{
char buf[FLOAT_DIG + (DECIMAL_MANT + 7) / 8 + 10];
if (isinf(d)) {
if (d < 0) strcpy(buf, "-inf");
else strcpy(buf, "inf");
}
else if (isnan(d)) {
strcpy(buf, "nan");
}
else if (d == 0.0) {
if (1.0/d < 0) strcpy(buf, "-0");
else strcpy(buf, "0");
}
else {
size_t len;
/* xxx: should not use system's sprintf(3) */
snprintf(buf, sizeof(buf), "%.*g", FLOAT_DIG, d);
len = strlen(buf);
w_bytes(buf, len + save_mantissa(d, buf + len), arg);
return;
}
w_bytes(buf, strlen(buf), arg);
}
static void
w_symbol(ID id, struct dump_arg *arg)
{
VALUE sym;
st_data_t num;
int encidx = -1;
if (st_lookup(arg->symbols, id, &num)) {
w_byte(TYPE_SYMLINK, arg);
w_long((long)num, arg);
}
else {
sym = rb_id2str(id);
if (!sym) {
rb_raise(rb_eTypeError, "can't dump anonymous ID %ld", id);
}
encidx = rb_enc_get_index(sym);
if (encidx == rb_usascii_encindex() ||
rb_enc_str_coderange(sym) == ENC_CODERANGE_7BIT) {
encidx = -1;
}
else {
w_byte(TYPE_IVAR, arg);
}
w_byte(TYPE_SYMBOL, arg);
w_bytes(RSTRING_PTR(sym), RSTRING_LEN(sym), arg);
st_add_direct(arg->symbols, id, arg->symbols->num_entries);
if (encidx != -1) {
struct dump_call_arg c_arg;
c_arg.limit = 1;
c_arg.arg = arg;
w_encoding(sym, 0, &c_arg);
}
}
}
static void
w_unique(VALUE s, struct dump_arg *arg)
{
must_not_be_anonymous("class", s);
w_symbol(rb_intern_str(s), arg);
}
static void w_object(VALUE,struct dump_arg*,int);
static int
hash_each(VALUE key, VALUE value, struct dump_call_arg *arg)
{
w_object(key, arg->arg, arg->limit);
w_object(value, arg->arg, arg->limit);
return ST_CONTINUE;
}
static void
w_extended(VALUE klass, struct dump_arg *arg, int check)
{
if (check && FL_TEST(klass, FL_SINGLETON)) {
if (RCLASS_M_TBL(klass)->num_entries ||
(RCLASS_IV_TBL(klass) && RCLASS_IV_TBL(klass)->num_entries > 1)) {
rb_raise(rb_eTypeError, "singleton can't be dumped");
}
klass = RCLASS_SUPER(klass);
}
while (BUILTIN_TYPE(klass) == T_ICLASS) {
VALUE path = rb_class_name(RBASIC(klass)->klass);
w_byte(TYPE_EXTENDED, arg);
w_unique(path, arg);
klass = RCLASS_SUPER(klass);
}
}
static void
w_class(char type, VALUE obj, struct dump_arg *arg, int check)
{
VALUE path;
st_data_t real_obj;
VALUE klass;
if (st_lookup(arg->compat_tbl, (st_data_t)obj, &real_obj)) {
obj = (VALUE)real_obj;
}
klass = CLASS_OF(obj);
* gc.c (Init_stack): stack region is far smaller than usual if pthread is used. * marshal.c (w_extended): singleton methods should not be checked when dumping via marshal_dump() or _dump(). [ruby-talk:85909] * file.c (getcwdofdrv): avoid using getcwd() directly, use my_getcwd() instead. * merged NeXT, OpenStep, Rhapsody ports patch from Eric Sunshine <sunshine@sunshineco.com>. [ruby-core:01596] * marshal.c (w_object): LINK check earlier than anything else, i.e. do not dump TYPE_IVAR for already dumped objects. (ruby-bugs PR#1220) * eval.c (rb_eval): call "inherited" only when a new class is generated; not on reopening. * eval.c (eval): prepend error position in evaluating string to * configure.in: revived NextStep, OpenStep, and Rhapsody ports which had become unbuildable; enhanced --enable-fat-binary option so that it accepts a list of desired architectures (rather than assuming a fixed list), or defaults to a platform-appropriate list if user does not provide an explicit list; made the default list of architectures for MAB (fat binary) more comprehensive; now uses -fno-common even when building the interpreter (in addition to using it for extensions), thus allowing the interpreter to be embedded into a plugin module of an external project (in addition to allowing embedding directly into an application); added checks for <netinet/in_systm.h> (needed by `socket' extension) and getcwd(); now ensures that -I/usr/local/include is employed when extensions' extconf.rb scripts invoke have_header() since extension checks on NextStep and OpenStep will fail without it if the desired resource resides in the /usr/local tree; fixed formatting of --help message. * Makefile.in: $(LIBRUBY_A) rule now deletes the archive before invoking $(AR) since `ar' on Apple/NeXT can not "update" MAB archives (see configure's --enable-fat-binary option); added rule for new missing/getcwd.c. * defines.h: fixed endian handling during MAB build (see configure's --enable-fat-binary option) to ensure that all portions of the project see the correct WORDS_BIGENDIAN value (some extension modules were getting the wrong endian setting); added missing constants GETPGRP_VOID, WNOHANG, WUNTRACED, X_OK, and type pid_t for NextStep and OpenStep; removed unnecessary and problematic HAVE_SYS_WAIT_H define in NeXT section. * dir.c: do not allow NAMLEN() macro to trust dirent::d_namlen on NextStep since, on some installations, this value always resolves uselessly to zero. * dln.c: added error reporting to NextStep extension loader since the previous behavior of failing silently was not useful; now ensures that NSLINKMODULE_OPTION_BINDNOW compatibility constant is defined for OpenStep and Rhapsody; no longer includes <mach-o/dyld.h> twice on Rhapsody since this header lacks multiple-include protection, which resulted in "redefinition" compilation errors. * main.c: also create hard reference to objc_msgSend() on NeXT platforms (in addition to Apple platforms). * lib/mkmf.rb: now exports XCFLAGS from configure script to extension makefiles so that extensions can be built MAB (see configure's --enable-fat-binary option); also utilize XCFLAGS in cc_command() (but not cpp_command() because MAB flags are incompatible with direct invocation of `cpp'). * ext/curses/extconf.rb: now additionally checks for presence of these curses functions which are not present on NextStep or Openstep: bkgd(), bkgdset(), color(), curs(), getbkgd(), init(), scrl(), set(), setscrreg(), wattroff(), wattron(), wattrset(), wbkgd(), wbkgdset(), wscrl(), wsetscrreg() * ext/curses/curses.c: added appropriate #ifdef's for additional set of curses functions now checked by extconf.rb; fixed curses_bkgd() and window_bkgd() to correctly return boolean result rather than numeric result; fixed window_getbkgd() to correctly signal an error by returning nil rather than -1. * ext/etc/etc.c: setup_passwd() and setup_group() now check for null pointers before invoking rb_tainted_str_new2() upon fields extracted from `struct passwd' and `struct group' since null pointers in some fields are common on NextStep/OpenStep (especially so for the `pw_comment' field) and rb_tainted_str_new2() throws an exception when it receives a null pointer. * ext/pty/pty.c: include "util.h" for strdup()/ruby_strdup() for platforms such as NextStep and OpenStep which lack strdup(). * ext/socket/getaddrinfo.c: cast first argument of getservbyname(), gethostbyaddr(), and gethostbyname() from (const char*) to non-const (char*) for older platforms such as NextStep and OpenStep. * ext/socket/socket.c: include "util.h" for strdup()/ruby_strdup() for platforms such as NextStep and OpenStep which lack strdup(); include <netinet/in_systm.h> if present for NextStep and OpenStep; cast first argument of gethostbyaddr() and getservbyname() from (const char*) to non-const (char*) for older platforms. * ext/syslog/syslog.c: include "util.h" for strdup()/ruby_strdup() for platforms such as NextStep and OpenStep which lack strdup(). git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@5002 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2003-11-21 23:00:03 -05:00
w_extended(klass, arg, check);
w_byte(type, arg);
path = class2path(rb_class_real(klass));
w_unique(path, arg);
}
static void
w_uclass(VALUE obj, VALUE super, struct dump_arg *arg)
{
VALUE klass = CLASS_OF(obj);
w_extended(klass, arg, TRUE);
klass = rb_class_real(klass);
if (klass != super) {
w_byte(TYPE_UCLASS, arg);
w_unique(class2path(klass), arg);
}
}
static int
w_obj_each(ID id, VALUE value, struct dump_call_arg *arg)
{
if (id == rb_id_encoding()) return ST_CONTINUE;
if (id == rb_intern("E")) return ST_CONTINUE;
w_symbol(id, arg->arg);
w_object(value, arg->arg, arg->limit);
return ST_CONTINUE;
}
static void
w_encoding(VALUE obj, long num, struct dump_call_arg *arg)
{
int encidx = rb_enc_get_index(obj);
rb_encoding *enc = 0;
st_data_t name;
if (encidx <= 0 || !(enc = rb_enc_from_index(encidx))) {
w_long(num, arg->arg);
return;
}
w_long(num + 1, arg->arg);
/* special treatment for US-ASCII and UTF-8 */
if (encidx == rb_usascii_encindex()) {
w_symbol(rb_intern("E"), arg->arg);
w_object(Qfalse, arg->arg, arg->limit + 1);
return;
}
else if (encidx == rb_utf8_encindex()) {
w_symbol(rb_intern("E"), arg->arg);
w_object(Qtrue, arg->arg, arg->limit + 1);
return;
}
w_symbol(rb_id_encoding(), arg->arg);
do {
if (!arg->arg->encodings)
arg->arg->encodings = st_init_strcasetable();
else if (st_lookup(arg->arg->encodings, (st_data_t)rb_enc_name(enc), &name))
break;
name = (st_data_t)rb_str_new2(rb_enc_name(enc));
st_insert(arg->arg->encodings, (st_data_t)rb_enc_name(enc), name);
} while (0);
w_object(name, arg->arg, arg->limit + 1);
}
static void
w_ivar(VALUE obj, st_table *tbl, struct dump_call_arg *arg)
{
long num = tbl ? tbl->num_entries : 0;
w_encoding(obj, num, arg);
if (tbl) {
st_foreach_safe(tbl, w_obj_each, (st_data_t)arg);
}
}
static void
w_objivar(VALUE obj, struct dump_call_arg *arg)
{
VALUE *ptr;
long i, len, num;
len = ROBJECT_NUMIV(obj);
ptr = ROBJECT_IVPTR(obj);
num = 0;
for (i = 0; i < len; i++)
if (ptr[i] != Qundef)
num += 1;
w_encoding(obj, num, arg);
if (num != 0) {
rb_ivar_foreach(obj, w_obj_each, (st_data_t)arg);
}
}
static void
w_object(VALUE obj, struct dump_arg *arg, int limit)
{
struct dump_call_arg c_arg;
st_table *ivtbl = 0;
* gc.c (Init_stack): stack region is far smaller than usual if pthread is used. * marshal.c (w_extended): singleton methods should not be checked when dumping via marshal_dump() or _dump(). [ruby-talk:85909] * file.c (getcwdofdrv): avoid using getcwd() directly, use my_getcwd() instead. * merged NeXT, OpenStep, Rhapsody ports patch from Eric Sunshine <sunshine@sunshineco.com>. [ruby-core:01596] * marshal.c (w_object): LINK check earlier than anything else, i.e. do not dump TYPE_IVAR for already dumped objects. (ruby-bugs PR#1220) * eval.c (rb_eval): call "inherited" only when a new class is generated; not on reopening. * eval.c (eval): prepend error position in evaluating string to * configure.in: revived NextStep, OpenStep, and Rhapsody ports which had become unbuildable; enhanced --enable-fat-binary option so that it accepts a list of desired architectures (rather than assuming a fixed list), or defaults to a platform-appropriate list if user does not provide an explicit list; made the default list of architectures for MAB (fat binary) more comprehensive; now uses -fno-common even when building the interpreter (in addition to using it for extensions), thus allowing the interpreter to be embedded into a plugin module of an external project (in addition to allowing embedding directly into an application); added checks for <netinet/in_systm.h> (needed by `socket' extension) and getcwd(); now ensures that -I/usr/local/include is employed when extensions' extconf.rb scripts invoke have_header() since extension checks on NextStep and OpenStep will fail without it if the desired resource resides in the /usr/local tree; fixed formatting of --help message. * Makefile.in: $(LIBRUBY_A) rule now deletes the archive before invoking $(AR) since `ar' on Apple/NeXT can not "update" MAB archives (see configure's --enable-fat-binary option); added rule for new missing/getcwd.c. * defines.h: fixed endian handling during MAB build (see configure's --enable-fat-binary option) to ensure that all portions of the project see the correct WORDS_BIGENDIAN value (some extension modules were getting the wrong endian setting); added missing constants GETPGRP_VOID, WNOHANG, WUNTRACED, X_OK, and type pid_t for NextStep and OpenStep; removed unnecessary and problematic HAVE_SYS_WAIT_H define in NeXT section. * dir.c: do not allow NAMLEN() macro to trust dirent::d_namlen on NextStep since, on some installations, this value always resolves uselessly to zero. * dln.c: added error reporting to NextStep extension loader since the previous behavior of failing silently was not useful; now ensures that NSLINKMODULE_OPTION_BINDNOW compatibility constant is defined for OpenStep and Rhapsody; no longer includes <mach-o/dyld.h> twice on Rhapsody since this header lacks multiple-include protection, which resulted in "redefinition" compilation errors. * main.c: also create hard reference to objc_msgSend() on NeXT platforms (in addition to Apple platforms). * lib/mkmf.rb: now exports XCFLAGS from configure script to extension makefiles so that extensions can be built MAB (see configure's --enable-fat-binary option); also utilize XCFLAGS in cc_command() (but not cpp_command() because MAB flags are incompatible with direct invocation of `cpp'). * ext/curses/extconf.rb: now additionally checks for presence of these curses functions which are not present on NextStep or Openstep: bkgd(), bkgdset(), color(), curs(), getbkgd(), init(), scrl(), set(), setscrreg(), wattroff(), wattron(), wattrset(), wbkgd(), wbkgdset(), wscrl(), wsetscrreg() * ext/curses/curses.c: added appropriate #ifdef's for additional set of curses functions now checked by extconf.rb; fixed curses_bkgd() and window_bkgd() to correctly return boolean result rather than numeric result; fixed window_getbkgd() to correctly signal an error by returning nil rather than -1. * ext/etc/etc.c: setup_passwd() and setup_group() now check for null pointers before invoking rb_tainted_str_new2() upon fields extracted from `struct passwd' and `struct group' since null pointers in some fields are common on NextStep/OpenStep (especially so for the `pw_comment' field) and rb_tainted_str_new2() throws an exception when it receives a null pointer. * ext/pty/pty.c: include "util.h" for strdup()/ruby_strdup() for platforms such as NextStep and OpenStep which lack strdup(). * ext/socket/getaddrinfo.c: cast first argument of getservbyname(), gethostbyaddr(), and gethostbyname() from (const char*) to non-const (char*) for older platforms such as NextStep and OpenStep. * ext/socket/socket.c: include "util.h" for strdup()/ruby_strdup() for platforms such as NextStep and OpenStep which lack strdup(); include <netinet/in_systm.h> if present for NextStep and OpenStep; cast first argument of gethostbyaddr() and getservbyname() from (const char*) to non-const (char*) for older platforms. * ext/syslog/syslog.c: include "util.h" for strdup()/ruby_strdup() for platforms such as NextStep and OpenStep which lack strdup(). git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@5002 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2003-11-21 23:00:03 -05:00
st_data_t num;
int hasiv = 0;
#define has_ivars(obj, ivtbl) ((ivtbl = rb_generic_ivar_table(obj)) != 0 || \
(!SPECIAL_CONST_P(obj) && !ENCODING_IS_ASCII8BIT(obj)))
if (limit == 0) {
rb_raise(rb_eArgError, "exceed depth limit");
}
limit--;
c_arg.limit = limit;
c_arg.arg = arg;
* gc.c (Init_stack): stack region is far smaller than usual if pthread is used. * marshal.c (w_extended): singleton methods should not be checked when dumping via marshal_dump() or _dump(). [ruby-talk:85909] * file.c (getcwdofdrv): avoid using getcwd() directly, use my_getcwd() instead. * merged NeXT, OpenStep, Rhapsody ports patch from Eric Sunshine <sunshine@sunshineco.com>. [ruby-core:01596] * marshal.c (w_object): LINK check earlier than anything else, i.e. do not dump TYPE_IVAR for already dumped objects. (ruby-bugs PR#1220) * eval.c (rb_eval): call "inherited" only when a new class is generated; not on reopening. * eval.c (eval): prepend error position in evaluating string to * configure.in: revived NextStep, OpenStep, and Rhapsody ports which had become unbuildable; enhanced --enable-fat-binary option so that it accepts a list of desired architectures (rather than assuming a fixed list), or defaults to a platform-appropriate list if user does not provide an explicit list; made the default list of architectures for MAB (fat binary) more comprehensive; now uses -fno-common even when building the interpreter (in addition to using it for extensions), thus allowing the interpreter to be embedded into a plugin module of an external project (in addition to allowing embedding directly into an application); added checks for <netinet/in_systm.h> (needed by `socket' extension) and getcwd(); now ensures that -I/usr/local/include is employed when extensions' extconf.rb scripts invoke have_header() since extension checks on NextStep and OpenStep will fail without it if the desired resource resides in the /usr/local tree; fixed formatting of --help message. * Makefile.in: $(LIBRUBY_A) rule now deletes the archive before invoking $(AR) since `ar' on Apple/NeXT can not "update" MAB archives (see configure's --enable-fat-binary option); added rule for new missing/getcwd.c. * defines.h: fixed endian handling during MAB build (see configure's --enable-fat-binary option) to ensure that all portions of the project see the correct WORDS_BIGENDIAN value (some extension modules were getting the wrong endian setting); added missing constants GETPGRP_VOID, WNOHANG, WUNTRACED, X_OK, and type pid_t for NextStep and OpenStep; removed unnecessary and problematic HAVE_SYS_WAIT_H define in NeXT section. * dir.c: do not allow NAMLEN() macro to trust dirent::d_namlen on NextStep since, on some installations, this value always resolves uselessly to zero. * dln.c: added error reporting to NextStep extension loader since the previous behavior of failing silently was not useful; now ensures that NSLINKMODULE_OPTION_BINDNOW compatibility constant is defined for OpenStep and Rhapsody; no longer includes <mach-o/dyld.h> twice on Rhapsody since this header lacks multiple-include protection, which resulted in "redefinition" compilation errors. * main.c: also create hard reference to objc_msgSend() on NeXT platforms (in addition to Apple platforms). * lib/mkmf.rb: now exports XCFLAGS from configure script to extension makefiles so that extensions can be built MAB (see configure's --enable-fat-binary option); also utilize XCFLAGS in cc_command() (but not cpp_command() because MAB flags are incompatible with direct invocation of `cpp'). * ext/curses/extconf.rb: now additionally checks for presence of these curses functions which are not present on NextStep or Openstep: bkgd(), bkgdset(), color(), curs(), getbkgd(), init(), scrl(), set(), setscrreg(), wattroff(), wattron(), wattrset(), wbkgd(), wbkgdset(), wscrl(), wsetscrreg() * ext/curses/curses.c: added appropriate #ifdef's for additional set of curses functions now checked by extconf.rb; fixed curses_bkgd() and window_bkgd() to correctly return boolean result rather than numeric result; fixed window_getbkgd() to correctly signal an error by returning nil rather than -1. * ext/etc/etc.c: setup_passwd() and setup_group() now check for null pointers before invoking rb_tainted_str_new2() upon fields extracted from `struct passwd' and `struct group' since null pointers in some fields are common on NextStep/OpenStep (especially so for the `pw_comment' field) and rb_tainted_str_new2() throws an exception when it receives a null pointer. * ext/pty/pty.c: include "util.h" for strdup()/ruby_strdup() for platforms such as NextStep and OpenStep which lack strdup(). * ext/socket/getaddrinfo.c: cast first argument of getservbyname(), gethostbyaddr(), and gethostbyname() from (const char*) to non-const (char*) for older platforms such as NextStep and OpenStep. * ext/socket/socket.c: include "util.h" for strdup()/ruby_strdup() for platforms such as NextStep and OpenStep which lack strdup(); include <netinet/in_systm.h> if present for NextStep and OpenStep; cast first argument of gethostbyaddr() and getservbyname() from (const char*) to non-const (char*) for older platforms. * ext/syslog/syslog.c: include "util.h" for strdup()/ruby_strdup() for platforms such as NextStep and OpenStep which lack strdup(). git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@5002 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2003-11-21 23:00:03 -05:00
if (st_lookup(arg->data, obj, &num)) {
w_byte(TYPE_LINK, arg);
w_long((long)num, arg);
return;
}
if ((hasiv = has_ivars(obj, ivtbl)) != 0) {
w_byte(TYPE_IVAR, arg);
}
if (obj == Qnil) {
w_byte(TYPE_NIL, arg);
}
else if (obj == Qtrue) {
w_byte(TYPE_TRUE, arg);
}
else if (obj == Qfalse) {
w_byte(TYPE_FALSE, arg);
}
else if (FIXNUM_P(obj)) {
#if SIZEOF_LONG <= 4
w_byte(TYPE_FIXNUM, arg);
w_long(FIX2INT(obj), arg);
#else
if (RSHIFT((long)obj, 31) == 0 || RSHIFT((long)obj, 31) == -1) {
w_byte(TYPE_FIXNUM, arg);
w_long(FIX2LONG(obj), arg);
}
else {
w_object(rb_int2big(FIX2LONG(obj)), arg, limit);
}
#endif
}
else if (SYMBOL_P(obj)) {
w_symbol(SYM2ID(obj), arg);
}
else {
if (OBJ_TAINTED(obj)) arg->taint = TRUE;
if (OBJ_UNTRUSTED(obj)) arg->untrust = TRUE;
if (rb_respond_to(obj, s_mdump)) {
volatile VALUE v;
st_add_direct(arg->data, obj, arg->data->num_entries);
v = rb_funcall(obj, s_mdump, 0, 0);
check_dump_arg(arg, s_mdump);
w_class(TYPE_USRMARSHAL, obj, arg, FALSE);
w_object(v, arg, limit);
if (hasiv) w_ivar(obj, ivtbl, &c_arg);
return;
}
if (rb_respond_to(obj, s_dump)) {
VALUE v;
st_table *ivtbl2 = 0;
int hasiv2;
v = rb_funcall(obj, s_dump, 1, INT2NUM(limit));
check_dump_arg(arg, s_dump);
if (TYPE(v) != T_STRING) {
rb_raise(rb_eTypeError, "_dump() must return string");
}
if ((hasiv2 = has_ivars(v, ivtbl2)) != 0 && !hasiv) {
w_byte(TYPE_IVAR, arg);
}
w_class(TYPE_USERDEF, obj, arg, FALSE);
w_bytes(RSTRING_PTR(v), RSTRING_LEN(v), arg);
if (hasiv2) {
w_ivar(v, ivtbl2, &c_arg);
}
else if (hasiv) {
w_ivar(obj, ivtbl, &c_arg);
}
st_add_direct(arg->data, obj, arg->data->num_entries);
return;
}
st_add_direct(arg->data, obj, arg->data->num_entries);
{
st_data_t compat_data;
rb_alloc_func_t allocator = rb_get_alloc_func(RBASIC(obj)->klass);
if (st_lookup(compat_allocator_tbl,
(st_data_t)allocator,
&compat_data)) {
marshal_compat_t *compat = (marshal_compat_t*)compat_data;
VALUE real_obj = obj;
obj = compat->dumper(real_obj);
st_insert(arg->compat_tbl, (st_data_t)obj, (st_data_t)real_obj);
}
}
switch (BUILTIN_TYPE(obj)) {
case T_CLASS:
if (FL_TEST(obj, FL_SINGLETON)) {
rb_raise(rb_eTypeError, "singleton class can't be dumped");
}
w_byte(TYPE_CLASS, arg);
{
volatile VALUE path = class2path(obj);
w_bytes(RSTRING_PTR(path), RSTRING_LEN(path), arg);
}
break;
case T_MODULE:
w_byte(TYPE_MODULE, arg);
{
VALUE path = class2path(obj);
w_bytes(RSTRING_PTR(path), RSTRING_LEN(path), arg);
}
break;
case T_FLOAT:
w_byte(TYPE_FLOAT, arg);
w_float(RFLOAT_VALUE(obj), arg);
break;
case T_BIGNUM:
w_byte(TYPE_BIGNUM, arg);
{
char sign = RBIGNUM_SIGN(obj) ? '+' : '-';
long len = RBIGNUM_LEN(obj);
BDIGIT *d = RBIGNUM_DIGITS(obj);
w_byte(sign, arg);
w_long(SHORTLEN(len), arg); /* w_short? */
while (len--) {
#if SIZEOF_BDIGITS > SIZEOF_SHORT
BDIGIT num = *d;
int i;
for (i=0; i<SIZEOF_BDIGITS; i+=SIZEOF_SHORT) {
w_short(num & SHORTMASK, arg);
num = SHORTDN(num);
if (len == 0 && num == 0) break;
}
#else
w_short(*d, arg);
#endif
d++;
}
}
break;
case T_STRING:
w_uclass(obj, rb_cString, arg);
w_byte(TYPE_STRING, arg);
w_bytes(RSTRING_PTR(obj), RSTRING_LEN(obj), arg);
break;
case T_REGEXP:
w_uclass(obj, rb_cRegexp, arg);
w_byte(TYPE_REGEXP, arg);
{
int opts = rb_reg_options(obj);
w_bytes(RREGEXP_SRC_PTR(obj), RREGEXP_SRC_LEN(obj), arg);
w_byte((char)opts, arg);
}
break;
case T_ARRAY:
w_uclass(obj, rb_cArray, arg);
w_byte(TYPE_ARRAY, arg);
{
long i, len = RARRAY_LEN(obj);
w_long(len, arg);
for (i=0; i<RARRAY_LEN(obj); i++) {
w_object(RARRAY_PTR(obj)[i], arg, limit);
if (len != RARRAY_LEN(obj)) {
rb_raise(rb_eRuntimeError, "array modified during dump");
}
}
}
break;
case T_HASH:
w_uclass(obj, rb_cHash, arg);
if (NIL_P(RHASH_IFNONE(obj))) {
w_byte(TYPE_HASH, arg);
}
else if (FL_TEST(obj, FL_USER2)) {
/* FL_USER2 means HASH_PROC_DEFAULT (see hash.c) */
* array.c: replace rb_protect_inspect() and rb_inspecting_p() by rb_exec_recursive() in eval.c. * eval.c (rb_exec_recursive): new function. * array.c (rb_ary_join): use rb_exec_recursive(). * array.c (rb_ary_inspect, rb_ary_hash): ditto. * file.c (rb_file_join): ditto. * hash.c (rb_hash_inspect, rb_hash_to_s, rb_hash_hash): ditto. * io.c (rb_io_puts): ditto. * object.c (rb_obj_inspect): ditto * struct.c (rb_struct_inspect): ditto. * lib/set.rb (SortedSet::setup): a hack to shut up warning. [ruby-talk:132866] * lib/time.rb (Time::strptime): add new function. inspired by [ruby-talk:132815]. * lib/parsedate.rb (ParseDate::strptime): ditto. * regparse.c: move st_*_strend() functions from st.c. fixed some potential memory leaks. * exception error messages updated. [ruby-core:04497] * ext/socket/socket.c (Init_socket): add bunch of Socket constants. Patch from Sam Roberts <sroberts@uniserve.com>. [ruby-core:04409] * array.c (rb_ary_s_create): no need for negative argc check. [ruby-core:04463] * array.c (rb_ary_unshift_m): ditto. * lib/xmlrpc/parser.rb (XMLRPC::FaultException): make it subclass of StandardError class, not Exception class. [ruby-core:04429] * parse.y (fcall_gen): lvar(arg) will be evaluated as lvar.call(arg) when lvar is a defined local variable. [new] * object.c (rb_class_initialize): call inherited method before calling initializing block. * eval.c (rb_thread_start_1): initialize newly pushed frame. * lib/open3.rb (Open3::popen3): $? should not be EXIT_FAILURE. fixed: [ruby-core:04444] * eval.c (is_defined): NODE_IASGN is an assignment. * ext/readline/readline.c (Readline.readline): use rl_outstream and rl_instream. [ruby-dev:25699] * ext/etc/etc.c (Init_etc): sGroup needs HAVE_ST_GR_PASSWD check [ruby-dev:25675] * misc/ruby-mode.el: [ruby-core:04415] * lib/rdoc/generators/html_generator.rb: [ruby-core:04412] * lib/rdoc/generators/ri_generator.rb: ditto. * struct.c (make_struct): fixed: [ruby-core:04402] * ext/curses/curses.c (window_color_set): [ruby-core:04393] * ext/socket/socket.c (Init_socket): SO_REUSEPORT added. [ruby-talk:130092] * object.c: [ruby-doc:818] * parse.y (open_args): fix too verbose warnings for the space before argument parentheses. [ruby-dev:25492] * parse.y (parser_yylex): ditto. * parse.y (parser_yylex): the first expression in the parentheses should not be a command. [ruby-dev:25492] * lib/irb/context.rb (IRB::Context::initialize): [ruby-core:04330] * object.c (Init_Object): remove Object#type. [ruby-core:04335] * st.c (st_foreach): report success/failure by return value. [ruby-Bugs-1396] * parse.y: forgot to initialize parser struct. [ruby-dev:25492] * parse.y (parser_yylex): no tLABEL on EXPR_BEG. [ruby-talk:127711] * document updates - [ruby-core:04296], [ruby-core:04301], [ruby-core:04302], [ruby-core:04307] * dir.c (rb_push_glob): should work for NUL delimited patterns. * dir.c (rb_glob2): should aware of offset in the pattern. * string.c (rb_str_new4): should propagate taintedness. * env.h: rename member names in struct FRAME; last_func -> callee, orig_func -> this_func, last_class -> this_class. * struct.c (rb_struct_set): use original method name, not callee name, to retrieve member slot. [ruby-core:04268] * time.c (time_strftime): protect from format modification from GC finalizers. * object.c (Init_Object): remove rb_obj_id_obsolete() * eval.c (rb_mod_define_method): incomplete subclass check. [ruby-dev:25464] * gc.c (rb_data_object_alloc): klass may be NULL. [ruby-list:40498] * bignum.c (rb_big_rand): should return positive random number. [ruby-dev:25401] * bignum.c (rb_big_rand): do not use rb_big_modulo to generate random bignums. [ruby-dev:25396] * variable.c (rb_autoload): [ruby-dev:25373] * eval.c (svalue_to_avalue): [ruby-dev:25366] * string.c (rb_str_justify): [ruby-dev:25367] * io.c (rb_f_select): [ruby-dev:25312] * ext/socket/socket.c (sock_s_getservbyport): [ruby-talk:124072] * struct.c (make_struct): [ruby-dev:25249] * dir.c (dir_open_dir): new function. [ruby-dev:25242] * io.c (rb_f_open): add type check for return value from to_open. * lib/pstore.rb (PStore#transaction): Use the empty content when a file is not found. [ruby-dev:24561] git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@8068 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2005-03-04 01:47:45 -05:00
rb_raise(rb_eTypeError, "can't dump hash with default proc");
}
else {
w_byte(TYPE_HASH_DEF, arg);
}
w_long(RHASH_SIZE(obj), arg);
rb_hash_foreach(obj, hash_each, (st_data_t)&c_arg);
if (!NIL_P(RHASH_IFNONE(obj))) {
w_object(RHASH_IFNONE(obj), arg, limit);
}
break;
case T_STRUCT:
w_class(TYPE_STRUCT, obj, arg, TRUE);
{
long len = RSTRUCT_LEN(obj);
VALUE mem;
long i;
w_long(len, arg);
mem = rb_struct_members(obj);
for (i=0; i<len; i++) {
w_symbol(SYM2ID(RARRAY_PTR(mem)[i]), arg);
w_object(RSTRUCT_PTR(obj)[i], arg, limit);
}
}
break;
case T_OBJECT:
w_class(TYPE_OBJECT, obj, arg, TRUE);
w_objivar(obj, &c_arg);
break;
case T_DATA:
{
VALUE v;
if (!rb_respond_to(obj, s_dump_data)) {
rb_raise(rb_eTypeError,
* gc.c (Init_stack): stack region is far smaller than usual if pthread is used. * marshal.c (w_extended): singleton methods should not be checked when dumping via marshal_dump() or _dump(). [ruby-talk:85909] * file.c (getcwdofdrv): avoid using getcwd() directly, use my_getcwd() instead. * merged NeXT, OpenStep, Rhapsody ports patch from Eric Sunshine <sunshine@sunshineco.com>. [ruby-core:01596] * marshal.c (w_object): LINK check earlier than anything else, i.e. do not dump TYPE_IVAR for already dumped objects. (ruby-bugs PR#1220) * eval.c (rb_eval): call "inherited" only when a new class is generated; not on reopening. * eval.c (eval): prepend error position in evaluating string to * configure.in: revived NextStep, OpenStep, and Rhapsody ports which had become unbuildable; enhanced --enable-fat-binary option so that it accepts a list of desired architectures (rather than assuming a fixed list), or defaults to a platform-appropriate list if user does not provide an explicit list; made the default list of architectures for MAB (fat binary) more comprehensive; now uses -fno-common even when building the interpreter (in addition to using it for extensions), thus allowing the interpreter to be embedded into a plugin module of an external project (in addition to allowing embedding directly into an application); added checks for <netinet/in_systm.h> (needed by `socket' extension) and getcwd(); now ensures that -I/usr/local/include is employed when extensions' extconf.rb scripts invoke have_header() since extension checks on NextStep and OpenStep will fail without it if the desired resource resides in the /usr/local tree; fixed formatting of --help message. * Makefile.in: $(LIBRUBY_A) rule now deletes the archive before invoking $(AR) since `ar' on Apple/NeXT can not "update" MAB archives (see configure's --enable-fat-binary option); added rule for new missing/getcwd.c. * defines.h: fixed endian handling during MAB build (see configure's --enable-fat-binary option) to ensure that all portions of the project see the correct WORDS_BIGENDIAN value (some extension modules were getting the wrong endian setting); added missing constants GETPGRP_VOID, WNOHANG, WUNTRACED, X_OK, and type pid_t for NextStep and OpenStep; removed unnecessary and problematic HAVE_SYS_WAIT_H define in NeXT section. * dir.c: do not allow NAMLEN() macro to trust dirent::d_namlen on NextStep since, on some installations, this value always resolves uselessly to zero. * dln.c: added error reporting to NextStep extension loader since the previous behavior of failing silently was not useful; now ensures that NSLINKMODULE_OPTION_BINDNOW compatibility constant is defined for OpenStep and Rhapsody; no longer includes <mach-o/dyld.h> twice on Rhapsody since this header lacks multiple-include protection, which resulted in "redefinition" compilation errors. * main.c: also create hard reference to objc_msgSend() on NeXT platforms (in addition to Apple platforms). * lib/mkmf.rb: now exports XCFLAGS from configure script to extension makefiles so that extensions can be built MAB (see configure's --enable-fat-binary option); also utilize XCFLAGS in cc_command() (but not cpp_command() because MAB flags are incompatible with direct invocation of `cpp'). * ext/curses/extconf.rb: now additionally checks for presence of these curses functions which are not present on NextStep or Openstep: bkgd(), bkgdset(), color(), curs(), getbkgd(), init(), scrl(), set(), setscrreg(), wattroff(), wattron(), wattrset(), wbkgd(), wbkgdset(), wscrl(), wsetscrreg() * ext/curses/curses.c: added appropriate #ifdef's for additional set of curses functions now checked by extconf.rb; fixed curses_bkgd() and window_bkgd() to correctly return boolean result rather than numeric result; fixed window_getbkgd() to correctly signal an error by returning nil rather than -1. * ext/etc/etc.c: setup_passwd() and setup_group() now check for null pointers before invoking rb_tainted_str_new2() upon fields extracted from `struct passwd' and `struct group' since null pointers in some fields are common on NextStep/OpenStep (especially so for the `pw_comment' field) and rb_tainted_str_new2() throws an exception when it receives a null pointer. * ext/pty/pty.c: include "util.h" for strdup()/ruby_strdup() for platforms such as NextStep and OpenStep which lack strdup(). * ext/socket/getaddrinfo.c: cast first argument of getservbyname(), gethostbyaddr(), and gethostbyname() from (const char*) to non-const (char*) for older platforms such as NextStep and OpenStep. * ext/socket/socket.c: include "util.h" for strdup()/ruby_strdup() for platforms such as NextStep and OpenStep which lack strdup(); include <netinet/in_systm.h> if present for NextStep and OpenStep; cast first argument of gethostbyaddr() and getservbyname() from (const char*) to non-const (char*) for older platforms. * ext/syslog/syslog.c: include "util.h" for strdup()/ruby_strdup() for platforms such as NextStep and OpenStep which lack strdup(). git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@5002 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2003-11-21 23:00:03 -05:00
"no marshal_dump is defined for class %s",
rb_obj_classname(obj));
}
v = rb_funcall(obj, s_dump_data, 0);
check_dump_arg(arg, s_dump_data);
w_class(TYPE_DATA, obj, arg, TRUE);
w_object(v, arg, limit);
}
break;
default:
rb_raise(rb_eTypeError, "can't dump %s",
rb_obj_classname(obj));
break;
}
}
if (hasiv) {
w_ivar(obj, ivtbl, &c_arg);
}
}
static void
clear_dump_arg(struct dump_arg *arg)
{
if (!arg->symbols) return;
st_free_table(arg->symbols);
arg->symbols = 0;
st_free_table(arg->data);
arg->data = 0;
st_free_table(arg->compat_tbl);
arg->compat_tbl = 0;
if (arg->encodings) {
st_free_table(arg->encodings);
arg->encodings = 0;
}
if (arg->taint) {
OBJ_TAINT(arg->str);
}
if (arg->untrust) {
OBJ_UNTRUST(arg->str);
}
}
/*
* call-seq:
* dump( obj [, anIO] , limit=--1 ) => anIO
*
* Serializes obj and all descendant objects. If anIO is
* specified, the serialized data will be written to it, otherwise the
* data will be returned as a String. If limit is specified, the
* traversal of subobjects will be limited to that depth. If limit is
* negative, no checking of depth will be performed.
*
* class Klass
* def initialize(str)
* @str = str
* end
* def sayHello
* @str
* end
* end
*
* (produces no output)
*
* o = Klass.new("hello\n")
* data = Marshal.dump(o)
* obj = Marshal.load(data)
* obj.sayHello #=> "hello\n"
*
* Marshal can't dump following objects:
* * anonymous Class/Module.
* * objects which related to its system (ex: Dir, File::Stat, IO, File, Socket and so on)
* * an instance of MatchData, Data, Method, UnboundMethod, Proc, Thread, ThreadGroup, Continuation
* * objects which defines singleton methods
*/
static VALUE
marshal_dump(int argc, VALUE *argv)
{
VALUE obj, port, a1, a2;
int limit = -1;
struct dump_arg *arg;
volatile VALUE wrapper;
port = Qnil;
rb_scan_args(argc, argv, "12", &obj, &a1, &a2);
if (argc == 3) {
if (!NIL_P(a2)) limit = NUM2INT(a2);
if (NIL_P(a1)) goto type_error;
port = a1;
}
else if (argc == 2) {
if (FIXNUM_P(a1)) limit = FIX2INT(a1);
else if (NIL_P(a1)) goto type_error;
else port = a1;
}
wrapper = TypedData_Make_Struct(rb_cData, struct dump_arg, &dump_arg_data, arg);
arg->dest = 0;
arg->symbols = st_init_numtable();
arg->data = st_init_numtable();
arg->taint = FALSE;
arg->untrust = FALSE;
arg->compat_tbl = st_init_numtable();
arg->encodings = 0;
arg->str = rb_str_buf_new(0);
if (!NIL_P(port)) {
if (!rb_respond_to(port, s_write)) {
type_error:
rb_raise(rb_eTypeError, "instance of IO needed");
}
arg->dest = port;
if (rb_respond_to(port, s_binmode)) {
rb_funcall2(port, s_binmode, 0, 0);
check_dump_arg(arg, s_binmode);
}
}
else {
port = arg->str;
}
w_byte(MARSHAL_MAJOR, arg);
w_byte(MARSHAL_MINOR, arg);
w_object(obj, arg, limit);
if (arg->dest) {
rb_io_write(arg->dest, arg->str);
rb_str_resize(arg->str, 0);
}
clear_dump_arg(arg);
RB_GC_GUARD(wrapper);
return port;
}
struct load_arg {
VALUE src;
long offset;
st_table *symbols;
st_table *data;
VALUE proc;
int taint;
int untrust;
st_table *compat_tbl;
};
static void
check_load_arg(struct load_arg *arg, ID sym)
{
if (!arg->symbols) {
rb_raise(rb_eRuntimeError, "Marshal.load reentered at %s",
rb_id2name(sym));
}
}
static void clear_load_arg(struct load_arg *arg);
static void
mark_load_arg(void *ptr)
{
struct load_arg *p = ptr;
if (!p->symbols)
return;
rb_mark_tbl(p->data);
rb_mark_hash(p->compat_tbl);
}
static void
free_load_arg(void *ptr)
{
clear_load_arg(ptr);
xfree(ptr);
}
static size_t
memsize_load_arg(const void *ptr)
{
return ptr ? sizeof(struct load_arg) : 0;
}
static const rb_data_type_t load_arg_data = {
"load_arg",
mark_load_arg, free_load_arg, memsize_load_arg
};
#define r_entry(v, arg) r_entry0(v, (arg)->data->num_entries, arg)
static VALUE r_entry0(VALUE v, st_index_t num, struct load_arg *arg);
static VALUE r_object(struct load_arg *arg);
static ID r_symbol(struct load_arg *arg);
static VALUE path2class(VALUE path);
static st_index_t
r_prepare(struct load_arg *arg)
{
st_index_t idx = arg->data->num_entries;
st_insert(arg->data, (st_data_t)idx, (st_data_t)Qundef);
return idx;
}
static int
r_byte(struct load_arg *arg)
{
int c;
if (TYPE(arg->src) == T_STRING) {
if (RSTRING_LEN(arg->src) > arg->offset) {
c = (unsigned char)RSTRING_PTR(arg->src)[arg->offset++];
}
else {
rb_raise(rb_eArgError, "marshal data too short");
}
}
else {
VALUE src = arg->src;
VALUE v = rb_funcall2(src, s_getbyte, 0, 0);
check_load_arg(arg, s_getbyte);
if (NIL_P(v)) rb_eof_error();
c = (unsigned char)NUM2CHR(v);
}
return c;
}
static void
long_toobig(int size)
{
rb_raise(rb_eTypeError, "long too big for this architecture (size "
STRINGIZE(SIZEOF_LONG)", given %d)", size);
}
#undef SIGN_EXTEND_CHAR
#if __STDC__
# define SIGN_EXTEND_CHAR(c) ((signed char)(c))
#else /* not __STDC__ */
/* As in Harbison and Steele. */
# define SIGN_EXTEND_CHAR(c) ((((unsigned char)(c)) ^ 128) - 128)
#endif
static long
r_long(struct load_arg *arg)
{
register long x;
int c = SIGN_EXTEND_CHAR(r_byte(arg));
long i;
if (c == 0) return 0;
if (c > 0) {
if (4 < c && c < 128) {
return c - 5;
}
if (c > (int)sizeof(long)) long_toobig(c);
x = 0;
for (i=0;i<c;i++) {
x |= (long)r_byte(arg) << (8*i);
}
}
else {
if (-129 < c && c < -4) {
return c + 5;
}
c = -c;
if (c > (int)sizeof(long)) long_toobig(c);
x = -1;
for (i=0;i<c;i++) {
x &= ~((long)0xff << (8*i));
x |= (long)r_byte(arg) << (8*i);
}
}
return x;
}
#define r_bytes(arg) r_bytes0(r_long(arg), (arg))
static VALUE
r_bytes0(long len, struct load_arg *arg)
{
VALUE str;
if (len == 0) return rb_str_new(0, 0);
if (TYPE(arg->src) == T_STRING) {
if (RSTRING_LEN(arg->src) - arg->offset >= len) {
str = rb_str_new(RSTRING_PTR(arg->src)+arg->offset, len);
arg->offset += len;
}
else {
too_short:
rb_raise(rb_eArgError, "marshal data too short");
}
}
else {
VALUE src = arg->src;
VALUE n = LONG2NUM(len);
str = rb_funcall2(src, s_read, 1, &n);
check_load_arg(arg, s_read);
if (NIL_P(str)) goto too_short;
StringValue(str);
if (RSTRING_LEN(str) != len) goto too_short;
if (OBJ_TAINTED(str)) arg->taint = TRUE;
if (OBJ_UNTRUSTED(str)) arg->untrust = TRUE;
}
return str;
}
static int
id2encidx(ID id, VALUE val, struct load_arg *arg)
{
if (id == rb_id_encoding()) {
int idx = rb_enc_find_index(StringValueCStr(val));
return idx;
}
else if (id == rb_intern("E")) {
if (val == Qfalse) return rb_usascii_encindex();
else if (val == Qtrue) return rb_utf8_encindex();
/* bogus ignore */
}
return -1;
}
static ID
r_symlink(struct load_arg *arg)
{
ID id;
long num = r_long(arg);
if (st_lookup(arg->symbols, num, &id)) {
return id;
}
rb_raise(rb_eArgError, "bad symbol");
}
static ID
r_symreal(struct load_arg *arg, int ivar)
{
volatile VALUE s = r_bytes(arg);
ID id;
int idx = -1;
st_index_t n = arg->symbols->num_entries;
st_insert(arg->symbols, (st_data_t)n, (st_data_t)0);
if (ivar) {
long num = r_long(arg);
while (num-- > 0) {
id = r_symbol(arg);
idx = id2encidx(id, r_object(arg), arg);
}
}
if (idx < 0) idx = rb_usascii_encindex();
rb_enc_associate_index(s, idx);
id = rb_intern_str(s);
st_insert(arg->symbols, (st_data_t)n, (st_data_t)id);
return id;
}
static ID
r_symbol(struct load_arg *arg)
{
int type, ivar = 0;
again:
switch ((type = r_byte(arg))) {
case TYPE_IVAR:
ivar = 1;
goto again;
case TYPE_SYMBOL:
return r_symreal(arg, ivar);
case TYPE_SYMLINK:
if (ivar) {
rb_raise(rb_eArgError, "dump format error (symlink with encoding)");
}
return r_symlink(arg);
default:
rb_raise(rb_eArgError, "dump format error for symbol(0x%x)", type);
break;
}
}
static VALUE
r_unique(struct load_arg *arg)
{
return rb_id2str(r_symbol(arg));
}
static VALUE
r_string(struct load_arg *arg)
{
return r_bytes(arg);
}
static VALUE
r_entry0(VALUE v, st_index_t num, struct load_arg *arg)
{
st_data_t real_obj = (VALUE)Qundef;
if (st_lookup(arg->compat_tbl, v, &real_obj)) {
st_insert(arg->data, num, (st_data_t)real_obj);
}
else {
st_insert(arg->data, num, (st_data_t)v);
}
if (arg->taint) {
OBJ_TAINT(v);
if ((VALUE)real_obj != Qundef)
OBJ_TAINT((VALUE)real_obj);
}
if (arg->untrust) {
OBJ_UNTRUST(v);
if ((VALUE)real_obj != Qundef)
OBJ_UNTRUST((VALUE)real_obj);
}
return v;
}
static VALUE
r_leave(VALUE v, struct load_arg *arg)
{
st_data_t data;
if (st_lookup(arg->compat_tbl, v, &data)) {
VALUE real_obj = (VALUE)data;
rb_alloc_func_t allocator = rb_get_alloc_func(CLASS_OF(real_obj));
st_data_t key = v;
if (st_lookup(compat_allocator_tbl, (st_data_t)allocator, &data)) {
marshal_compat_t *compat = (marshal_compat_t*)data;
compat->loader(real_obj, v);
}
st_delete(arg->compat_tbl, &key, 0);
v = real_obj;
}
if (arg->proc) {
v = rb_funcall(arg->proc, s_call, 1, v);
check_load_arg(arg, s_call);
}
return v;
}
static void
r_ivar(VALUE obj, int *has_encoding, struct load_arg *arg)
{
long len;
len = r_long(arg);
if (len > 0) {
do {
ID id = r_symbol(arg);
VALUE val = r_object(arg);
int idx = id2encidx(id, val, arg);
if (idx >= 0) {
rb_enc_associate_index(obj, idx);
if (has_encoding) *has_encoding = TRUE;
}
else {
rb_ivar_set(obj, id, val);
}
} while (--len > 0);
}
}
static VALUE
path2class(VALUE path)
{
VALUE v = rb_path_to_class(path);
if (TYPE(v) != T_CLASS) {
rb_raise(rb_eArgError, "%.*s does not refer to class",
(int)RSTRING_LEN(path), RSTRING_PTR(path));
}
return v;
}
static VALUE
path2module(VALUE path)
{
VALUE v = rb_path_to_class(path);
if (TYPE(v) != T_MODULE) {
rb_raise(rb_eArgError, "%.*s does not refer to module",
(int)RSTRING_LEN(path), RSTRING_PTR(path));
}
return v;
}
static VALUE
obj_alloc_by_path(VALUE path, struct load_arg *arg)
{
VALUE klass;
st_data_t data;
rb_alloc_func_t allocator;
klass = path2class(path);
allocator = rb_get_alloc_func(klass);
if (st_lookup(compat_allocator_tbl, (st_data_t)allocator, &data)) {
marshal_compat_t *compat = (marshal_compat_t*)data;
VALUE real_obj = rb_obj_alloc(klass);
VALUE obj = rb_obj_alloc(compat->oldclass);
st_insert(arg->compat_tbl, (st_data_t)obj, (st_data_t)real_obj);
return obj;
}
return rb_obj_alloc(klass);
}
static VALUE
r_object0(struct load_arg *arg, int *ivp, VALUE extmod)
{
VALUE v = Qnil;
int type = r_byte(arg);
long id;
st_data_t link;
switch (type) {
case TYPE_LINK:
id = r_long(arg);
if (!st_lookup(arg->data, (st_data_t)id, &link)) {
rb_raise(rb_eArgError, "dump format error (unlinked)");
}
v = (VALUE)link;
if (arg->proc) {
v = rb_funcall(arg->proc, s_call, 1, v);
check_load_arg(arg, s_call);
}
break;
case TYPE_IVAR:
{
int ivar = TRUE;
v = r_object0(arg, &ivar, extmod);
if (ivar) r_ivar(v, NULL, arg);
}
break;
case TYPE_EXTENDED:
{
VALUE m = path2module(r_unique(arg));
if (NIL_P(extmod)) extmod = rb_ary_new2(0);
rb_ary_push(extmod, m);
v = r_object0(arg, 0, extmod);
while (RARRAY_LEN(extmod) > 0) {
m = rb_ary_pop(extmod);
rb_extend_object(v, m);
}
}
break;
case TYPE_UCLASS:
{
VALUE c = path2class(r_unique(arg));
if (FL_TEST(c, FL_SINGLETON)) {
rb_raise(rb_eTypeError, "singleton can't be loaded");
}
v = r_object0(arg, 0, extmod);
if (rb_special_const_p(v) || TYPE(v) == T_OBJECT || TYPE(v) == T_CLASS) {
format_error:
rb_raise(rb_eArgError, "dump format error (user class)");
}
if (TYPE(v) == T_MODULE || !RTEST(rb_class_inherited_p(c, RBASIC(v)->klass))) {
VALUE tmp = rb_obj_alloc(c);
if (TYPE(v) != TYPE(tmp)) goto format_error;
}
RBASIC(v)->klass = c;
}
break;
case TYPE_NIL:
v = Qnil;
v = r_leave(v, arg);
break;
case TYPE_TRUE:
v = Qtrue;
v = r_leave(v, arg);
break;
case TYPE_FALSE:
v = Qfalse;
v = r_leave(v, arg);
break;
case TYPE_FIXNUM:
{
long i = r_long(arg);
v = LONG2FIX(i);
}
v = r_leave(v, arg);
break;
case TYPE_FLOAT:
{
double d;
VALUE str = r_bytes(arg);
const char *ptr = RSTRING_PTR(str);
if (strcmp(ptr, "nan") == 0) {
d = NAN;
}
else if (strcmp(ptr, "inf") == 0) {
d = INFINITY;
}
else if (strcmp(ptr, "-inf") == 0) {
d = -INFINITY;
}
else {
char *e;
d = strtod(ptr, &e);
d = load_mantissa(d, e, RSTRING_LEN(str) - (e - ptr));
}
v = DBL2NUM(d);
v = r_entry(v, arg);
v = r_leave(v, arg);
}
break;
case TYPE_BIGNUM:
{
long len;
BDIGIT *digits;
volatile VALUE data;
NEWOBJ(big, struct RBignum);
OBJSETUP(big, rb_cBignum, T_BIGNUM);
RBIGNUM_SET_SIGN(big, (r_byte(arg) == '+'));
len = r_long(arg);
data = r_bytes0(len * 2, arg);
#if SIZEOF_BDIGITS == SIZEOF_SHORT
rb_big_resize((VALUE)big, len);
#else
rb_big_resize((VALUE)big, (len + 1) * 2 / sizeof(BDIGIT));
#endif
digits = RBIGNUM_DIGITS(big);
MEMCPY(digits, RSTRING_PTR(data), char, len * 2);
#if SIZEOF_BDIGITS > SIZEOF_SHORT
MEMZERO((char *)digits + len * 2, char,
RBIGNUM_LEN(big) * sizeof(BDIGIT) - len * 2);
#endif
len = RBIGNUM_LEN(big);
while (len > 0) {
unsigned char *p = (unsigned char *)digits;
BDIGIT num = 0;
#if SIZEOF_BDIGITS > SIZEOF_SHORT
int shift = 0;
int i;
for (i=0; i<SIZEOF_BDIGITS; i++) {
num |= (int)p[i] << shift;
shift += 8;
}
#else
num = p[0] | (p[1] << 8);
#endif
*digits++ = num;
len--;
}
v = rb_big_norm((VALUE)big);
v = r_entry(v, arg);
v = r_leave(v, arg);
}
break;
case TYPE_STRING:
v = r_entry(r_string(arg), arg);
v = r_leave(v, arg);
break;
case TYPE_REGEXP:
{
volatile VALUE str = r_bytes(arg);
int options = r_byte(arg);
int has_encoding = FALSE;
v = r_entry(rb_reg_alloc(), arg);
if (ivp) {
r_ivar(str, &has_encoding, arg);
*ivp = FALSE;
}
if (!has_encoding) {
/* 1.8 compatibility; remove escapes undefined in 1.8 */
char *ptr = RSTRING_PTR(str), *dst = ptr, *src = ptr;
long len = RSTRING_LEN(str);
long bs = 0;
for (; len-- > 0; *dst++ = *src++) {
switch (*src) {
case '\\': bs++; break;
case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
case 'm': case 'o': case 'p': case 'q': case 'u': case 'y':
case 'E': case 'F': case 'H': case 'I': case 'J': case 'K':
case 'L': case 'N': case 'O': case 'P': case 'Q': case 'R':
case 'S': case 'T': case 'U': case 'V': case 'X': case 'Y':
if (bs & 1) --dst;
default: bs = 0; break;
}
}
rb_str_set_len(str, dst - ptr);
}
v = rb_reg_init_str(v, str, options);
v = r_leave(v, arg);
}
break;
case TYPE_ARRAY:
{
volatile long len = r_long(arg); /* gcc 2.7.2.3 -O2 bug?? */
v = rb_ary_new2(len);
v = r_entry(v, arg);
while (len--) {
rb_ary_push(v, r_object(arg));
}
v = r_leave(v, arg);
}
break;
case TYPE_HASH:
case TYPE_HASH_DEF:
{
long len = r_long(arg);
v = rb_hash_new();
v = r_entry(v, arg);
while (len--) {
VALUE key = r_object(arg);
VALUE value = r_object(arg);
rb_hash_aset(v, key, value);
}
if (type == TYPE_HASH_DEF) {
RHASH_IFNONE(v) = r_object(arg);
}
v = r_leave(v, arg);
}
break;
case TYPE_STRUCT:
{
VALUE mem, values;
volatile long i; /* gcc 2.7.2.3 -O2 bug?? */
ID slot;
st_index_t idx = r_prepare(arg);
VALUE klass = path2class(r_unique(arg));
long len = r_long(arg);
v = rb_obj_alloc(klass);
if (TYPE(v) != T_STRUCT) {
rb_raise(rb_eTypeError, "class %s not a struct", rb_class2name(klass));
}
mem = rb_struct_s_members(klass);
if (RARRAY_LEN(mem) != len) {
rb_raise(rb_eTypeError, "struct %s not compatible (struct size differs)",
rb_class2name(klass));
}
v = r_entry0(v, idx, arg);
values = rb_ary_new2(len);
for (i=0; i<len; i++) {
slot = r_symbol(arg);
if (RARRAY_PTR(mem)[i] != ID2SYM(slot)) {
rb_raise(rb_eTypeError, "struct %s not compatible (:%s for :%s)",
rb_class2name(klass),
rb_id2name(slot),
rb_id2name(SYM2ID(RARRAY_PTR(mem)[i])));
}
rb_ary_push(values, r_object(arg));
}
rb_struct_initialize(v, values);
v = r_leave(v, arg);
}
break;
case TYPE_USERDEF:
{
VALUE klass = path2class(r_unique(arg));
VALUE data;
if (!rb_respond_to(klass, s_load)) {
rb_raise(rb_eTypeError, "class %s needs to have method `_load'",
rb_class2name(klass));
}
data = r_string(arg);
if (ivp) {
r_ivar(data, NULL, arg);
*ivp = FALSE;
}
v = rb_funcall(klass, s_load, 1, data);
check_load_arg(arg, s_load);
v = r_entry(v, arg);
v = r_leave(v, arg);
}
break;
case TYPE_USRMARSHAL:
{
VALUE klass = path2class(r_unique(arg));
VALUE data;
v = rb_obj_alloc(klass);
if (!NIL_P(extmod)) {
while (RARRAY_LEN(extmod) > 0) {
VALUE m = rb_ary_pop(extmod);
rb_extend_object(v, m);
}
}
if (!rb_respond_to(v, s_mload)) {
rb_raise(rb_eTypeError, "instance of %s needs to have method `marshal_load'",
rb_class2name(klass));
}
v = r_entry(v, arg);
data = r_object(arg);
rb_funcall(v, s_mload, 1, data);
check_load_arg(arg, s_mload);
v = r_leave(v, arg);
}
break;
case TYPE_OBJECT:
{
st_index_t idx = r_prepare(arg);
v = obj_alloc_by_path(r_unique(arg), arg);
if (TYPE(v) != T_OBJECT) {
rb_raise(rb_eArgError, "dump format error");
}
v = r_entry0(v, idx, arg);
r_ivar(v, NULL, arg);
v = r_leave(v, arg);
}
break;
case TYPE_DATA:
{
VALUE klass = path2class(r_unique(arg));
if (rb_respond_to(klass, s_alloc)) {
static int warn = TRUE;
if (warn) {
rb_warn("define `allocate' instead of `_alloc'");
warn = FALSE;
}
v = rb_funcall(klass, s_alloc, 0);
check_load_arg(arg, s_alloc);
}
else {
v = rb_obj_alloc(klass);
}
if (TYPE(v) != T_DATA) {
rb_raise(rb_eArgError, "dump format error");
}
v = r_entry(v, arg);
if (!rb_respond_to(v, s_load_data)) {
rb_raise(rb_eTypeError,
"class %s needs to have instance method `_load_data'",
rb_class2name(klass));
}
rb_funcall(v, s_load_data, 1, r_object0(arg, 0, extmod));
check_load_arg(arg, s_load_data);
v = r_leave(v, arg);
}
break;
case TYPE_MODULE_OLD:
{
volatile VALUE str = r_bytes(arg);
v = rb_path_to_class(str);
v = r_entry(v, arg);
v = r_leave(v, arg);
}
break;
case TYPE_CLASS:
{
volatile VALUE str = r_bytes(arg);
v = path2class(str);
v = r_entry(v, arg);
v = r_leave(v, arg);
}
break;
case TYPE_MODULE:
{
volatile VALUE str = r_bytes(arg);
v = path2module(str);
v = r_entry(v, arg);
v = r_leave(v, arg);
}
break;
case TYPE_SYMBOL:
if (ivp) {
v = ID2SYM(r_symreal(arg, *ivp));
*ivp = FALSE;
}
else {
v = ID2SYM(r_symreal(arg, 0));
}
v = r_leave(v, arg);
break;
case TYPE_SYMLINK:
v = ID2SYM(r_symlink(arg));
break;
default:
rb_raise(rb_eArgError, "dump format error(0x%x)", type);
break;
}
return v;
}
static VALUE
r_object(struct load_arg *arg)
{
return r_object0(arg, 0, Qnil);
}
static void
clear_load_arg(struct load_arg *arg)
{
if (!arg->symbols) return;
st_free_table(arg->symbols);
arg->symbols = 0;
st_free_table(arg->data);
arg->data = 0;
st_free_table(arg->compat_tbl);
arg->compat_tbl = 0;
}
/*
* call-seq:
* load( source [, proc] ) => obj
* restore( source [, proc] ) => obj
*
* Returns the result of converting the serialized data in source into a
* Ruby object (possibly with associated subordinate objects). source
* may be either an instance of IO or an object that responds to
* to_str. If proc is specified, it will be passed each object as it
* is deserialized.
*/
static VALUE
marshal_load(int argc, VALUE *argv)
{
VALUE port, proc;
int major, minor, taint = FALSE;
VALUE v;
volatile VALUE wrapper;
struct load_arg *arg;
rb_scan_args(argc, argv, "11", &port, &proc);
v = rb_check_string_type(port);
if (!NIL_P(v)) {
taint = OBJ_TAINTED(port); /* original taintedness */
port = v;
}
else if (rb_respond_to(port, s_getbyte) && rb_respond_to(port, s_read)) {
if (rb_respond_to(port, s_binmode)) {
rb_funcall2(port, s_binmode, 0, 0);
}
taint = TRUE;
}
else {
rb_raise(rb_eTypeError, "instance of IO needed");
}
wrapper = TypedData_Make_Struct(rb_cData, struct load_arg, &load_arg_data, arg);
arg->taint = taint;
arg->untrust = OBJ_UNTRUSTED(port);
arg->src = port;
arg->offset = 0;
arg->symbols = st_init_numtable();
arg->data = st_init_numtable();
arg->compat_tbl = st_init_numtable();
arg->proc = 0;
major = r_byte(arg);
minor = r_byte(arg);
if (major != MARSHAL_MAJOR || minor > MARSHAL_MINOR) {
clear_load_arg(arg);
rb_raise(rb_eTypeError, "incompatible marshal file format (can't be read)\n\
\tformat version %d.%d required; %d.%d given",
MARSHAL_MAJOR, MARSHAL_MINOR, major, minor);
}
if (RTEST(ruby_verbose) && minor != MARSHAL_MINOR) {
rb_warn("incompatible marshal file format (can be read)\n\
\tformat version %d.%d required; %d.%d given",
MARSHAL_MAJOR, MARSHAL_MINOR, major, minor);
}
if (!NIL_P(proc)) arg->proc = proc;
v = r_object(arg);
clear_load_arg(arg);
RB_GC_GUARD(wrapper);
return v;
}
/*
* The marshaling library converts collections of Ruby objects into a
* byte stream, allowing them to be stored outside the currently
* active script. This data may subsequently be read and the original
* objects reconstituted.
* Marshaled data has major and minor version numbers stored along
* with the object information. In normal use, marshaling can only
* load data written with the same major version number and an equal
* or lower minor version number. If Ruby's ``verbose'' flag is set
* (normally using -d, -v, -w, or --verbose) the major and minor
* numbers must match exactly. Marshal versioning is independent of
* Ruby's version numbers. You can extract the version by reading the
* first two bytes of marshaled data.
*
* str = Marshal.dump("thing")
* RUBY_VERSION #=> "1.9.0"
* str[0].ord #=> 4
* str[1].ord #=> 8
*
* Some objects cannot be dumped: if the objects to be dumped include
* bindings, procedure or method objects, instances of class IO, or
* singleton objects, a TypeError will be raised.
* If your class has special serialization needs (for example, if you
* want to serialize in some specific format), or if it contains
* objects that would otherwise not be serializable, you can implement
* your own serialization strategy by defining two methods, _dump and
* _load:
* The instance method _dump should return a String object containing
* all the information necessary to reconstitute objects of this class
* and all referenced objects up to a maximum depth given as an integer
* parameter (a value of -1 implies that you should disable depth checking).
* The class method _load should take a String and return an object of this class.
*/
void
Init_marshal(void)
{
#undef rb_intern
#define rb_intern(str) rb_intern_const(str)
VALUE rb_mMarshal = rb_define_module("Marshal");
s_dump = rb_intern("_dump");
s_load = rb_intern("_load");
s_mdump = rb_intern("marshal_dump");
s_mload = rb_intern("marshal_load");
s_dump_data = rb_intern("_dump_data");
s_load_data = rb_intern("_load_data");
s_alloc = rb_intern("_alloc");
s_call = rb_intern("call");
s_getbyte = rb_intern("getbyte");
s_read = rb_intern("read");
s_write = rb_intern("write");
s_binmode = rb_intern("binmode");
rb_define_module_function(rb_mMarshal, "dump", marshal_dump, -1);
rb_define_module_function(rb_mMarshal, "load", marshal_load, -1);
rb_define_module_function(rb_mMarshal, "restore", marshal_load, -1);
rb_define_const(rb_mMarshal, "MAJOR_VERSION", INT2FIX(MARSHAL_MAJOR));
rb_define_const(rb_mMarshal, "MINOR_VERSION", INT2FIX(MARSHAL_MINOR));
compat_allocator_tbl = st_init_numtable();
compat_allocator_tbl_wrapper =
Data_Wrap_Struct(rb_cData, mark_marshal_compat_t, 0, compat_allocator_tbl);
rb_gc_register_mark_object(compat_allocator_tbl_wrapper);
}
VALUE
rb_marshal_dump(VALUE obj, VALUE port)
{
int argc = 1;
VALUE argv[2];
argv[0] = obj;
argv[1] = port;
if (!NIL_P(port)) argc = 2;
return marshal_dump(argc, argv);
}
VALUE
rb_marshal_load(VALUE port)
{
return marshal_load(1, &port);
}