mongo-ruby-driver/ext/cbson/cbson.c

951 lines
32 KiB
C
Raw Normal View History

2009-03-04 22:38:06 +00:00
/*
2010-03-01 17:09:21 +00:00
* Copyright 2009-2010 10gen, Inc.
2009-03-04 22:38:06 +00:00
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
/*
* This file contains C implementations of some of the functions needed by the
* bson module. If possible, these implementations should be used to speed up
* BSON encoding and decoding.
*/
2009-03-03 22:07:22 +00:00
#include "ruby.h"
/* Ensure compatibility with early releases of Ruby 1.8.5 */
#ifndef RSTRING_PTR
# define RSTRING_PTR(v) RSTRING(v)->ptr
#endif
#ifndef RSTRING_LEN
# define RSTRING_LEN(v) RSTRING(v)->len
#endif
#ifndef RARRAY_PTR
# define RARRAY_PTR(v) RARRAY(v)->ptr
#endif
#ifndef RARRAY_LEN
# define RARRAY_LEN(v) RARRAY(v)->len
#endif
#if HAVE_RUBY_ST_H
#include "ruby/st.h"
#endif
#if HAVE_ST_H
#include "st.h"
#endif
#if HAVE_RUBY_REGEX_H
#include "ruby/regex.h"
#endif
#if HAVE_REGEX_H
#include "regex.h"
#endif
#include <string.h>
#include <math.h>
#include <unistd.h>
#include <time.h>
2009-03-03 22:07:22 +00:00
#include "version.h"
#include "buffer.h"
#include "encoding_helpers.h"
#define SAFE_WRITE(buffer, data, size) \
if (buffer_write((buffer), (data), (size)) != 0) \
rb_raise(rb_eNoMemError, "failed to allocate memory in buffer.c")
#define SAFE_WRITE_AT_POS(buffer, position, data, size) \
if (buffer_write_at_position((buffer), (position), (data), (size)) != 0) \
2009-12-17 14:47:27 +00:00
rb_raise(rb_eRuntimeError, "invalid write at position in buffer.c")
#define MAX_HOSTNAME_LENGTH 256
2009-03-03 22:07:22 +00:00
2009-03-10 15:48:06 +00:00
static VALUE Binary;
2009-03-10 17:19:18 +00:00
static VALUE ObjectID;
static VALUE ObjectId;
2009-03-10 18:58:20 +00:00
static VALUE DBRef;
2009-03-12 20:12:52 +00:00
static VALUE Code;
2010-01-20 23:50:07 +00:00
static VALUE MinKey;
static VALUE MaxKey;
static VALUE Regexp;
2009-03-10 19:41:28 +00:00
static VALUE OrderedHash;
2010-04-05 22:24:31 +00:00
static VALUE InvalidKeyName;
static VALUE InvalidStringEncoding;
static VALUE InvalidDocument;
static VALUE DigestMD5;
2009-03-10 15:48:06 +00:00
#if HAVE_RUBY_ENCODING_H
#include "ruby/encoding.h"
#define STR_NEW(p,n) \
({ \
VALUE _str = rb_enc_str_new((p), (n), rb_utf8_encoding()); \
rb_encoding* internal_encoding = rb_default_internal_encoding(); \
if (internal_encoding) { \
_str = rb_str_export_to_enc(_str, internal_encoding); \
} \
_str; \
})
/* MUST call TO_UTF8 before calling write_utf8. */
#define TO_UTF8(string) rb_str_export_to_enc((string), rb_utf8_encoding())
static void write_utf8(buffer_t buffer, VALUE string, char check_null) {
result_t status = check_string(RSTRING_PTR(string), RSTRING_LEN(string),
0, check_null);
if (status == HAS_NULL) {
buffer_free(buffer);
rb_raise(InvalidDocument, "Key names / regex patterns must not contain the NULL byte");
}
SAFE_WRITE(buffer, RSTRING_PTR(string), RSTRING_LEN(string));
}
#else
#define STR_NEW(p,n) rb_str_new((p), (n))
/* MUST call TO_UTF8 before calling write_utf8. */
#define TO_UTF8(string) (string)
static void write_utf8(buffer_t buffer, VALUE string, char check_null) {
result_t status = check_string(RSTRING_PTR(string), RSTRING_LEN(string),
1, check_null);
if (status == HAS_NULL) {
buffer_free(buffer);
rb_raise(InvalidDocument, "Key names / regex patterns must not contain the NULL byte");
} else if (status == NOT_UTF_8) {
buffer_free(buffer);
rb_raise(InvalidStringEncoding, "String not valid UTF-8");
}
SAFE_WRITE(buffer, RSTRING_PTR(string), RSTRING_LEN(string));
}
#endif
// this sucks. but for some reason these moved around between 1.8 and 1.9
#ifdef ONIGURUMA_H
#define IGNORECASE ONIG_OPTION_IGNORECASE
#define MULTILINE ONIG_OPTION_MULTILINE
#define EXTENDED ONIG_OPTION_EXTEND
#else
#define IGNORECASE RE_OPTION_IGNORECASE
#define MULTILINE RE_OPTION_MULTILINE
#define EXTENDED RE_OPTION_EXTENDED
#endif
/* TODO we ought to check that the malloc or asprintf was successful
* and raise an exception if not. */
/* TODO maybe we can use something more portable like vsnprintf instead
* of this hack. And share it with the Python extension ;) */
/* If we don't have ASPRINTF, there are two possibilities:
* either use _scprintf and _snprintf on for Windows or
* use snprintf for solaris. */
#ifndef HAVE_ASPRINTF
#ifdef _MSC_VER
#define INT2STRING(buffer, i) \
{ \
int vslength = _scprintf("%d", i) + 1; \
*buffer = malloc(vslength); \
_snprintf(*buffer, vslength, "%d", i); \
}
#else
#define INT2STRING(buffer, i) \
{ \
int vslength = snprintf(NULL, 0, "%d", i) + 1; \
*buffer = malloc(vslength); \
snprintf(*buffer, vslength, "%d", i); \
}
#endif
#else
#define INT2STRING(buffer, i) asprintf(buffer, "%d", i);
#endif
#ifndef RREGEXP_SRC
#define RREGEXP_SRC(r) rb_str_new(RREGEXP((r))->str, RREGEXP((r))->len)
#endif
// rubinius compatibility
#ifndef RREGEXP_OPTIONS
#define RREGEXP_OPTIONS(r) RREGEXP(value)->ptr->options
#endif
2009-03-04 15:05:50 +00:00
static char zero = 0;
2009-03-04 16:07:22 +00:00
static char one = 1;
2009-03-05 18:16:34 +00:00
static int cmp_char(const void* a, const void* b) {
return *(char*)a - *(char*)b;
}
2010-01-28 23:44:50 +00:00
static void write_doc(buffer_t buffer, VALUE hash, VALUE check_keys, VALUE move_id);
2010-02-27 17:22:34 +00:00
static int write_element_with_id(VALUE key, VALUE value, VALUE extra);
static int write_element_without_id(VALUE key, VALUE value, VALUE extra);
static VALUE elements_to_hash(const char* buffer, int max);
2009-03-04 15:05:50 +00:00
static VALUE pack_extra(buffer_t buffer, VALUE check_keys) {
2009-09-04 14:22:28 +00:00
return rb_ary_new3(2, LL2NUM((long long)buffer), check_keys);
}
static void write_name_and_type(buffer_t buffer, VALUE name, char type) {
SAFE_WRITE(buffer, &type, 1);
name = TO_UTF8(name);
write_utf8(buffer, name, 1);
SAFE_WRITE(buffer, &zero, 1);
2009-03-04 15:05:50 +00:00
}
2010-02-27 17:22:34 +00:00
static int write_element(VALUE key, VALUE value, VALUE extra, int allow_id) {
buffer_t buffer = (buffer_t)NUM2LL(rb_ary_entry(extra, 0));
VALUE check_keys = rb_ary_entry(extra, 1);
2009-03-04 15:05:50 +00:00
2009-03-05 15:27:19 +00:00
if (TYPE(key) == T_SYMBOL) {
// TODO better way to do this... ?
key = rb_str_new2(rb_id2name(SYM2ID(key)));
}
if (TYPE(key) != T_STRING) {
2009-12-02 22:01:49 +00:00
buffer_free(buffer);
2009-03-05 15:27:19 +00:00
rb_raise(rb_eTypeError, "keys must be strings or symbols");
}
2010-02-27 17:22:34 +00:00
if (allow_id == 0 && strcmp("_id", RSTRING_PTR(key)) == 0) {
2009-03-05 20:48:40 +00:00
return ST_CONTINUE;
}
if (check_keys == Qtrue) {
int i;
if (RSTRING_LEN(key) > 0 && RSTRING_PTR(key)[0] == '$') {
2009-12-02 22:01:49 +00:00
buffer_free(buffer);
2010-04-05 22:24:31 +00:00
rb_raise(InvalidKeyName, "key must not start with '$'");
}
for (i = 0; i < RSTRING_LEN(key); i++) {
if (RSTRING_PTR(key)[i] == '.') {
2009-12-02 22:01:49 +00:00
buffer_free(buffer);
2010-04-05 22:24:31 +00:00
rb_raise(InvalidKeyName, "key must not contain '.'");
}
}
}
2009-03-04 15:05:50 +00:00
switch(TYPE(value)) {
2009-03-05 22:03:49 +00:00
case T_BIGNUM:
2009-03-04 16:07:22 +00:00
case T_FIXNUM:
2009-03-10 17:19:18 +00:00
{
2009-08-04 14:39:25 +00:00
if (rb_funcall(value, rb_intern(">"), 1, LL2NUM(9223372036854775807LL)) == Qtrue ||
2010-01-11 23:01:36 +00:00
rb_funcall(value, rb_intern("<"), 1, LL2NUM(-9223372036854775808ULL)) == Qtrue) {
2009-12-02 22:01:49 +00:00
buffer_free(buffer);
2009-08-04 14:39:25 +00:00
rb_raise(rb_eRangeError, "MongoDB can only handle 8-byte ints");
}
if (rb_funcall(value, rb_intern(">"), 1, INT2NUM(2147483647L)) == Qtrue ||
rb_funcall(value, rb_intern("<"), 1, INT2NUM(-2147483648L)) == Qtrue) {
long long ll_value;
write_name_and_type(buffer, key, 0x12);
ll_value = NUM2LL(value);
SAFE_WRITE(buffer, (char*)&ll_value, 8);
2009-08-04 14:39:25 +00:00
} else {
int int_value;
write_name_and_type(buffer, key, 0x10);
int_value = NUM2LL(value);
SAFE_WRITE(buffer, (char*)&int_value, 4);
}
2009-03-10 17:19:18 +00:00
break;
}
2009-03-04 16:07:22 +00:00
case T_TRUE:
2009-03-10 17:19:18 +00:00
{
write_name_and_type(buffer, key, 0x08);
SAFE_WRITE(buffer, &one, 1);
2009-03-10 17:19:18 +00:00
break;
}
2009-03-04 16:07:22 +00:00
case T_FALSE:
2009-03-10 17:19:18 +00:00
{
write_name_and_type(buffer, key, 0x08);
SAFE_WRITE(buffer, &zero, 1);
2009-03-10 17:19:18 +00:00
break;
}
2009-03-04 16:07:22 +00:00
case T_FLOAT:
2009-03-10 17:19:18 +00:00
{
double d = NUM2DBL(value);
write_name_and_type(buffer, key, 0x01);
SAFE_WRITE(buffer, (char*)&d, 8);
2009-03-10 17:19:18 +00:00
break;
}
2009-03-04 16:07:22 +00:00
case T_NIL:
2009-03-10 17:19:18 +00:00
{
write_name_and_type(buffer, key, 0x0A);
break;
}
2009-03-04 16:07:22 +00:00
case T_HASH:
2009-03-10 17:19:18 +00:00
{
write_name_and_type(buffer, key, 0x03);
2010-01-28 23:44:50 +00:00
write_doc(buffer, value, check_keys, Qfalse);
2009-03-10 17:19:18 +00:00
break;
}
2009-03-04 16:07:22 +00:00
case T_ARRAY:
2009-03-10 17:19:18 +00:00
{
buffer_position length_location, start_position, obj_length;
int items, i;
VALUE* values;
2009-03-10 17:19:18 +00:00
write_name_and_type(buffer, key, 0x04);
start_position = buffer_get_position(buffer);
2009-03-04 16:07:22 +00:00
2009-03-10 17:19:18 +00:00
// save space for length
length_location = buffer_save_space(buffer, 4);
if (length_location == -1) {
rb_raise(rb_eNoMemError, "failed to allocate memory in buffer.c");
}
2009-03-04 16:07:22 +00:00
items = RARRAY_LEN(value);
values = RARRAY_PTR(value);
2009-03-10 17:19:18 +00:00
for(i = 0; i < items; i++) {
char* name;
VALUE key;
INT2STRING(&name, i);
key = rb_str_new2(name);
2010-02-27 17:22:34 +00:00
write_element_with_id(key, values[i], pack_extra(buffer, check_keys));
2009-03-10 17:19:18 +00:00
free(name);
}
2009-03-04 16:07:22 +00:00
2009-03-10 17:19:18 +00:00
// write null byte and fill in length
SAFE_WRITE(buffer, &zero, 1);
obj_length = buffer_get_position(buffer) - start_position;
SAFE_WRITE_AT_POS(buffer, length_location, (const char*)&obj_length, 4);
2009-03-10 17:19:18 +00:00
break;
}
2009-03-04 15:05:50 +00:00
case T_STRING:
2009-03-10 17:19:18 +00:00
{
if (strcmp(rb_obj_classname(value),
2010-04-05 14:39:55 +00:00
"BSON::Code") == 0) {
buffer_position length_location, start_position, total_length;
int length;
2009-03-12 20:12:52 +00:00
write_name_and_type(buffer, key, 0x0F);
start_position = buffer_get_position(buffer);
length_location = buffer_save_space(buffer, 4);
if (length_location == -1) {
rb_raise(rb_eNoMemError, "failed to allocate memory in buffer.c");
}
2009-03-12 20:12:52 +00:00
length = RSTRING_LEN(value) + 1;
SAFE_WRITE(buffer, (char*)&length, 4);
SAFE_WRITE(buffer, RSTRING_PTR(value), length - 1);
SAFE_WRITE(buffer, &zero, 1);
2010-01-28 23:44:50 +00:00
write_doc(buffer, rb_funcall(value, rb_intern("scope"), 0), Qfalse, Qfalse);
2009-03-12 20:12:52 +00:00
total_length = buffer_get_position(buffer) - start_position;
SAFE_WRITE_AT_POS(buffer, length_location, (const char*)&total_length, 4);
2009-03-12 20:12:52 +00:00
break;
2009-03-10 17:19:18 +00:00
} else {
int length;
write_name_and_type(buffer, key, 0x02);
value = TO_UTF8(value);
length = RSTRING_LEN(value) + 1;
SAFE_WRITE(buffer, (char*)&length, 4);
write_utf8(buffer, value, 0);
SAFE_WRITE(buffer, &zero, 1);
2009-03-12 20:12:52 +00:00
break;
2009-03-10 17:19:18 +00:00
}
2009-03-05 21:08:10 +00:00
}
2009-03-05 16:09:52 +00:00
case T_SYMBOL:
2009-03-10 17:19:18 +00:00
{
const char* str_value = rb_id2name(SYM2ID(value));
int length = strlen(str_value) + 1;
write_name_and_type(buffer, key, 0x0E);
SAFE_WRITE(buffer, (char*)&length, 4);
SAFE_WRITE(buffer, str_value, length);
2009-03-10 17:19:18 +00:00
break;
}
2009-03-04 22:38:06 +00:00
case T_OBJECT:
{
// TODO there has to be a better way to do these checks...
const char* cls = rb_obj_classname(value);
2010-04-05 14:39:55 +00:00
if (strcmp(cls, "BSON::Binary") == 0 ||
2009-03-05 14:22:55 +00:00
strcmp(cls, "ByteBuffer") == 0) {
const char subtype = strcmp(cls, "ByteBuffer") ?
(const char)FIX2INT(rb_funcall(value, rb_intern("subtype"), 0)) : 2;
2009-03-04 22:38:06 +00:00
VALUE string_data = rb_funcall(value, rb_intern("to_s"), 0);
int length = RSTRING_LEN(string_data);
write_name_and_type(buffer, key, 0x05);
2009-03-04 22:38:06 +00:00
if (subtype == 2) {
const int other_length = length + 4;
SAFE_WRITE(buffer, (const char*)&other_length, 4);
SAFE_WRITE(buffer, &subtype, 1);
2009-03-04 22:38:06 +00:00
}
SAFE_WRITE(buffer, (const char*)&length, 4);
2009-03-04 22:38:06 +00:00
if (subtype != 2) {
SAFE_WRITE(buffer, &subtype, 1);
2009-03-04 22:38:06 +00:00
}
SAFE_WRITE(buffer, RSTRING_PTR(string_data), length);
2009-03-04 22:38:06 +00:00
break;
}
if ((strcmp(cls, "BSON::ObjectId") == 0) || (strcmp(cls, "BSON::ObjectID") == 0)) {
2009-03-05 16:38:08 +00:00
VALUE as_array = rb_funcall(value, rb_intern("to_a"), 0);
int i;
write_name_and_type(buffer, key, 0x07);
2009-03-05 16:38:08 +00:00
for (i = 0; i < 12; i++) {
char byte = (char)FIX2INT(RARRAY_PTR(as_array)[i]);
SAFE_WRITE(buffer, &byte, 1);
2009-03-05 16:38:08 +00:00
}
break;
}
2010-04-05 14:39:55 +00:00
if (strcmp(cls, "BSON::DBRef") == 0) {
buffer_position length_location, start_position, obj_length;
VALUE ns, oid;
2009-03-18 17:54:33 +00:00
write_name_and_type(buffer, key, 0x03);
start_position = buffer_get_position(buffer);
2009-03-18 17:54:33 +00:00
// save space for length
length_location = buffer_save_space(buffer, 4);
if (length_location == -1) {
rb_raise(rb_eNoMemError, "failed to allocate memory in buffer.c");
}
2009-03-05 16:45:04 +00:00
ns = rb_funcall(value, rb_intern("namespace"), 0);
2010-02-27 17:22:34 +00:00
write_element_with_id(rb_str_new2("$ref"), ns, pack_extra(buffer, Qfalse));
oid = rb_funcall(value, rb_intern("object_id"), 0);
2010-02-27 17:22:34 +00:00
write_element_with_id(rb_str_new2("$id"), oid, pack_extra(buffer, Qfalse));
2009-03-05 16:45:04 +00:00
2009-03-18 17:54:33 +00:00
// write null byte and fill in length
SAFE_WRITE(buffer, &zero, 1);
obj_length = buffer_get_position(buffer) - start_position;
SAFE_WRITE_AT_POS(buffer, length_location, (const char*)&obj_length, 4);
2009-03-05 16:45:04 +00:00
break;
}
2010-04-05 14:39:55 +00:00
if (strcmp(cls, "BSON::MaxKey") == 0) {
2010-01-20 23:50:07 +00:00
write_name_and_type(buffer, key, 0x7f);
break;
}
2010-04-05 14:39:55 +00:00
if (strcmp(cls, "BSON::MinKey") == 0) {
2010-01-20 23:50:07 +00:00
write_name_and_type(buffer, key, 0xff);
break;
}
if (strcmp(cls, "DateTime") == 0 || strcmp(cls, "Date") == 0 || strcmp(cls, "ActiveSupport::TimeWithZone") == 0) {
buffer_free(buffer);
rb_raise(InvalidDocument, "%s is not currently supported; use a UTC Time instance instead.", cls);
break;
}
if(strcmp(cls, "Complex") == 0 || strcmp(cls, "Rational") == 0 || strcmp(cls, "BigDecimal") == 0) {
buffer_free(buffer);
rb_raise(InvalidDocument, "Cannot serialize the Numeric type %s as BSON; only Bignum, Fixnum, and Float are supported.", cls);
break;
}
buffer_free(buffer);
rb_raise(InvalidDocument, "Cannot serialize an object of class %s into BSON.", cls);
break;
2009-03-04 22:38:06 +00:00
}
2009-03-05 15:58:56 +00:00
case T_DATA:
{
const char* cls = rb_obj_classname(value);
2009-03-05 15:58:56 +00:00
if (strcmp(cls, "Time") == 0) {
double t = NUM2DBL(rb_funcall(value, rb_intern("to_f"), 0));
long long time_since_epoch = (long long)round(t * 1000);
write_name_and_type(buffer, key, 0x09);
SAFE_WRITE(buffer, (const char*)&time_since_epoch, 8);
2009-03-05 15:58:56 +00:00
break;
}
if(strcmp(cls, "BigDecimal") == 0) {
buffer_free(buffer);
rb_raise(InvalidDocument, "Cannot serialize the Numeric type %s as BSON; only Bignum, Fixnum, and Float are supported.", cls);
break;
}
buffer_free(buffer);
rb_raise(InvalidDocument, "Cannot serialize an object of class %s into BSON.", cls);
break;
2009-03-05 15:58:56 +00:00
}
2009-03-05 18:16:34 +00:00
case T_REGEXP:
2009-03-10 17:19:18 +00:00
{
VALUE pattern = RREGEXP_SRC(value);
long flags = RREGEXP_OPTIONS(value);
VALUE has_extra;
write_name_and_type(buffer, key, 0x0B);
pattern = TO_UTF8(pattern);
write_utf8(buffer, pattern, 1);
SAFE_WRITE(buffer, &zero, 1);
2009-03-05 18:16:34 +00:00
if (flags & IGNORECASE) {
2009-03-10 17:19:18 +00:00
char ignorecase = 'i';
SAFE_WRITE(buffer, &ignorecase, 1);
2009-03-10 17:19:18 +00:00
}
if (flags & MULTILINE) {
2009-03-10 17:19:18 +00:00
char multiline = 'm';
SAFE_WRITE(buffer, &multiline, 1);
2009-03-10 17:19:18 +00:00
}
if (flags & EXTENDED) {
2009-03-10 17:19:18 +00:00
char extended = 'x';
SAFE_WRITE(buffer, &extended, 1);
2009-03-10 17:19:18 +00:00
}
2009-03-05 18:16:34 +00:00
has_extra = rb_funcall(value, rb_intern("respond_to?"), 1, rb_str_new2("extra_options_str"));
2009-03-10 17:19:18 +00:00
if (TYPE(has_extra) == T_TRUE) {
VALUE extra = rb_funcall(value, rb_intern("extra_options_str"), 0);
buffer_position old_position = buffer_get_position(buffer);
SAFE_WRITE(buffer, RSTRING_PTR(extra), RSTRING_LEN(extra));
qsort(buffer_get_buffer(buffer) + old_position, RSTRING_LEN(extra), sizeof(char), cmp_char);
2009-03-10 17:19:18 +00:00
}
SAFE_WRITE(buffer, &zero, 1);
2009-03-05 18:16:34 +00:00
2009-03-10 17:19:18 +00:00
break;
}
2009-03-04 15:05:50 +00:00
default:
2009-03-10 17:19:18 +00:00
{
const char* cls = rb_obj_classname(value);
2009-12-02 22:01:49 +00:00
buffer_free(buffer);
rb_raise(InvalidDocument, "Cannot serialize an object of class %s (type %d) into BSON.", cls, TYPE(value));
2009-03-10 17:19:18 +00:00
break;
}
2009-03-04 15:05:50 +00:00
}
return ST_CONTINUE;
}
2010-02-27 17:22:34 +00:00
static int write_element_without_id(VALUE key, VALUE value, VALUE extra) {
return write_element(key, value, extra, 0);
}
static int write_element_with_id(VALUE key, VALUE value, VALUE extra) {
return write_element(key, value, extra, 1);
2009-03-05 20:48:40 +00:00
}
2010-01-28 23:44:50 +00:00
static void write_doc(buffer_t buffer, VALUE hash, VALUE check_keys, VALUE move_id) {
buffer_position start_position = buffer_get_position(buffer);
buffer_position length_location = buffer_save_space(buffer, 4);
buffer_position length;
2010-01-28 23:44:50 +00:00
int allow_id;
2010-02-27 17:22:34 +00:00
int (*write_function)(VALUE, VALUE, VALUE) = NULL;
VALUE id_str = rb_str_new2("_id");
VALUE id_sym = ID2SYM(rb_intern("_id"));
if (length_location == -1) {
rb_raise(rb_eNoMemError, "failed to allocate memory in buffer.c");
}
2010-02-27 17:22:34 +00:00
// write '_id' first if move_id is true. then don't allow an id to be written.
2010-01-28 23:44:50 +00:00
if(move_id == Qtrue) {
allow_id = 0;
if (rb_funcall(hash, rb_intern("has_key?"), 1, id_str) == Qtrue) {
VALUE id = rb_hash_aref(hash, id_str);
2010-02-27 17:22:34 +00:00
write_element_with_id(id_str, id, pack_extra(buffer, check_keys));
2010-01-28 23:44:50 +00:00
} else if (rb_funcall(hash, rb_intern("has_key?"), 1, id_sym) == Qtrue) {
VALUE id = rb_hash_aref(hash, id_sym);
2010-02-27 17:22:34 +00:00
write_element_with_id(id_sym, id, pack_extra(buffer, check_keys));
2010-01-28 23:44:50 +00:00
}
}
else {
allow_id = 1;
if (strcmp(rb_obj_classname(hash), "Hash") == 0) {
2010-02-27 17:22:34 +00:00
if ((rb_funcall(hash, rb_intern("has_key?"), 1, id_str) == Qtrue) &&
(rb_funcall(hash, rb_intern("has_key?"), 1, id_sym) == Qtrue)) {
VALUE oid_sym = rb_hash_delete(hash, id_sym);
rb_funcall(hash, rb_intern("[]="), 2, id_str, oid_sym);
}
2010-01-28 23:44:50 +00:00
}
2009-03-05 20:48:40 +00:00
}
2010-02-27 17:22:34 +00:00
if(allow_id == 1) {
write_function = write_element_with_id;
}
else {
write_function = write_element_without_id;
}
// we have to check for an OrderedHash and handle that specially
2010-05-07 01:25:18 +00:00
if (strcmp(rb_obj_classname(hash), "BSON::OrderedHash") == 0) {
VALUE keys = rb_funcall(hash, rb_intern("keys"), 0);
int i;
2010-02-27 17:22:34 +00:00
for(i = 0; i < RARRAY_LEN(keys); i++) {
VALUE key = RARRAY_PTR(keys)[i];
VALUE value = rb_hash_aref(hash, key);
2010-02-27 17:22:34 +00:00
write_function(key, value, pack_extra(buffer, check_keys));
}
} else if (strcmp(rb_obj_classname(hash), "Hash") == 0) {
2010-02-27 17:22:34 +00:00
rb_hash_foreach(hash, write_function, pack_extra(buffer, check_keys));
} else {
buffer_free(buffer);
char* cls = rb_obj_classname(hash);
rb_raise(InvalidDocument, "BSON.serialize takes a Hash but got a %s", cls);
}
2009-03-04 15:05:50 +00:00
// write null byte and fill in length
SAFE_WRITE(buffer, &zero, 1);
length = buffer_get_position(buffer) - start_position;
// make sure that length doesn't exceed 4MB
if (length > 4 * 1024 * 1024) {
2009-12-16 19:14:15 +00:00
buffer_free(buffer);
rb_raise(InvalidDocument, "Document too large: BSON documents are limited to 4MB.");
return;
}
2010-01-11 23:01:36 +00:00
SAFE_WRITE_AT_POS(buffer, length_location, (const char*)&length, 4);
2009-03-04 15:05:50 +00:00
}
2010-01-28 23:44:50 +00:00
static VALUE method_serialize(VALUE self, VALUE doc, VALUE check_keys, VALUE move_id) {
VALUE result;
buffer_t buffer = buffer_new();
if (buffer == NULL) {
rb_raise(rb_eNoMemError, "failed to allocate memory in buffer.c");
}
2009-03-04 15:05:50 +00:00
2010-01-28 23:44:50 +00:00
write_doc(buffer, doc, check_keys, move_id);
2009-03-04 15:05:50 +00:00
result = rb_str_new(buffer_get_buffer(buffer), buffer_get_position(buffer));
if (buffer_free(buffer) != 0) {
rb_raise(rb_eRuntimeError, "failed to free buffer");
}
2009-03-04 15:05:50 +00:00
return result;
2009-03-03 22:07:22 +00:00
}
2009-03-10 14:23:42 +00:00
static VALUE get_value(const char* buffer, int* position, int type) {
VALUE value;
switch (type) {
2010-01-20 23:50:07 +00:00
case -1:
{
value = rb_class_new_instance(0, NULL, MinKey);
break;
}
2009-03-10 14:23:42 +00:00
case 1:
{
double d;
memcpy(&d, buffer + *position, 8);
value = rb_float_new(d);
2009-03-10 14:23:42 +00:00
*position += 8;
break;
}
case 2:
2009-03-10 15:48:06 +00:00
case 13:
2009-03-10 14:23:42 +00:00
{
int value_length;
value_length = *(int*)(buffer + *position) - 1;
2009-03-10 14:23:42 +00:00
*position += 4;
value = STR_NEW(buffer + *position, value_length);
2009-03-10 14:23:42 +00:00
*position += value_length + 1;
break;
}
case 3:
{
int size;
memcpy(&size, buffer + *position, 4);
2009-03-18 17:54:33 +00:00
if (strcmp(buffer + *position + 5, "$ref") == 0) { // DBRef
int offset = *position + 10;
2009-03-18 17:54:33 +00:00
VALUE argv[2];
int collection_length = *(int*)(buffer + offset) - 1;
char id_type;
offset += 4;
argv[0] = STR_NEW(buffer + offset, collection_length);
2009-03-18 17:54:33 +00:00
offset += collection_length + 1;
id_type = buffer[offset];
2009-03-18 17:54:33 +00:00
offset += 5;
argv[1] = get_value(buffer, &offset, (int)id_type);
value = rb_class_new_instance(2, argv, DBRef);
} else {
value = elements_to_hash(buffer + *position + 4, size - 5);
}
*position += size;
break;
}
case 4:
{
int size, end;
memcpy(&size, buffer + *position, 4);
end = *position + size - 1;
*position += 4;
value = rb_ary_new();
while (*position < end) {
int type = (int)buffer[(*position)++];
int key_size = strlen(buffer + *position);
VALUE to_append;
*position += key_size + 1; // just skip the key, they're in order.
to_append = get_value(buffer, position, type);
rb_ary_push(value, to_append);
}
(*position)++;
break;
}
2009-03-10 15:48:06 +00:00
case 5:
{
int length, subtype;
VALUE data, st;
VALUE argv[2];
2009-03-10 15:48:06 +00:00
memcpy(&length, buffer + *position, 4);
subtype = (unsigned char)buffer[*position + 4];
2009-03-10 15:48:06 +00:00
if (subtype == 2) {
data = rb_str_new(buffer + *position + 9, length - 4);
} else {
data = rb_str_new(buffer + *position + 5, length);
}
st = INT2FIX(subtype);
argv[0] = data;
argv[1] = st;
2009-03-10 15:48:06 +00:00
value = rb_class_new_instance(2, argv, Binary);
*position += length + 5;
break;
}
2009-03-10 17:19:18 +00:00
case 6:
{
value = Qnil;
2009-03-10 17:19:18 +00:00
break;
}
case 7:
{
VALUE str = rb_str_new(buffer + *position, 12);
VALUE oid = rb_funcall(str, rb_intern("unpack"), 1, rb_str_new2("C*"));
value = rb_class_new_instance(1, &oid, ObjectId);
2009-03-10 17:19:18 +00:00
*position += 12;
break;
}
2009-03-10 14:23:42 +00:00
case 8:
{
value = buffer[(*position)++] ? Qtrue : Qfalse;
break;
}
2009-03-10 17:19:18 +00:00
case 9:
{
long long millis;
memcpy(&millis, buffer + *position, 8);
2010-08-25 15:27:38 +00:00
value = rb_time_new(millis / 1000, (millis % 1000) * 1000);
value = rb_funcall(value, rb_intern("utc"), 0);
2009-03-10 17:19:18 +00:00
*position += 8;
break;
}
case 10:
{
value = Qnil;
break;
}
2009-03-10 19:33:03 +00:00
case 11:
{
int pattern_length = strlen(buffer + *position);
VALUE pattern = STR_NEW(buffer + *position, pattern_length);
int flags_length, flags = 0, i = 0;
VALUE argv[3];
2009-03-10 19:33:03 +00:00
*position += pattern_length + 1;
flags_length = strlen(buffer + *position);
2009-03-10 19:33:03 +00:00
for (i = 0; i < flags_length; i++) {
char flag = buffer[*position + i];
2009-03-11 14:49:30 +00:00
if (flag == 'i') {
flags |= IGNORECASE;
2009-03-11 14:49:30 +00:00
}
else if (flag == 'm') {
flags |= MULTILINE;
2009-03-11 14:49:30 +00:00
}
else if (flag == 'x') {
flags |= EXTENDED;
2009-03-10 19:33:03 +00:00
}
}
argv[0] = pattern;
argv[1] = INT2FIX(flags);
2010-04-07 20:32:25 +00:00
value = rb_class_new_instance(2, argv, Regexp);
2009-03-10 19:33:03 +00:00
*position += flags_length + 1;
break;
}
2009-03-10 18:58:20 +00:00
case 12:
{
int collection_length;
VALUE collection, str, oid, id, argv[2];
collection_length = *(int*)(buffer + *position) - 1;
2009-03-10 18:58:20 +00:00
*position += 4;
collection = STR_NEW(buffer + *position, collection_length);
2009-03-10 18:58:20 +00:00
*position += collection_length + 1;
str = rb_str_new(buffer + *position, 12);
oid = rb_funcall(str, rb_intern("unpack"), 1, rb_str_new2("C*"));
id = rb_class_new_instance(1, &oid, ObjectId);
2009-03-10 18:58:20 +00:00
*position += 12;
argv[0] = collection;
argv[1] = id;
2009-03-10 18:58:20 +00:00
value = rb_class_new_instance(2, argv, DBRef);
break;
}
2009-03-10 15:48:06 +00:00
case 14:
{
int value_length;
memcpy(&value_length, buffer + *position, 4);
value = ID2SYM(rb_intern(buffer + *position + 4));
2009-03-11 14:49:30 +00:00
*position += value_length + 4;
2009-03-10 15:48:06 +00:00
break;
}
2009-03-12 20:12:52 +00:00
case 15:
{
int code_length, scope_size;
VALUE code, scope, argv[2];
*position += 4;
code_length = *(int*)(buffer + *position) - 1;
*position += 4;
code = STR_NEW(buffer + *position, code_length);
2009-03-12 20:12:52 +00:00
*position += code_length + 1;
memcpy(&scope_size, buffer + *position, 4);
scope = elements_to_hash(buffer + *position + 4, scope_size - 5);
2009-03-12 20:12:52 +00:00
*position += scope_size;
argv[0] = code;
argv[1] = scope;
2009-03-12 20:12:52 +00:00
value = rb_class_new_instance(2, argv, Code);
break;
}
2009-03-10 14:23:42 +00:00
case 16:
{
int i;
memcpy(&i, buffer + *position, 4);
2009-03-11 14:49:30 +00:00
value = LL2NUM(i);
2009-03-10 14:23:42 +00:00
*position += 4;
break;
}
2009-04-22 17:49:56 +00:00
case 17:
{
int i;
int j;
memcpy(&i, buffer + *position, 4);
memcpy(&j, buffer + *position + 4, 4);
value = rb_ary_new3(2, LL2NUM(i), LL2NUM(j));
*position += 8;
break;
}
2009-08-04 14:39:25 +00:00
case 18:
{
long long ll;
memcpy(&ll, buffer + *position, 8);
value = LL2NUM(ll);
*position += 8;
break;
}
2010-01-20 23:50:07 +00:00
case 127:
{
value = rb_class_new_instance(0, NULL, MaxKey);
break;
}
2009-03-10 14:23:42 +00:00
default:
2009-03-10 17:19:18 +00:00
{
rb_raise(rb_eTypeError, "no c decoder for this type yet (%d)", type);
break;
}
2009-03-10 14:23:42 +00:00
}
return value;
}
static VALUE elements_to_hash(const char* buffer, int max) {
2009-03-10 19:41:28 +00:00
VALUE hash = rb_class_new_instance(0, NULL, OrderedHash);
2009-03-10 14:23:42 +00:00
int position = 0;
while (position < max) {
int type = (int)buffer[position++];
int name_length = strlen(buffer + position);
VALUE name = STR_NEW(buffer + position, name_length);
VALUE value;
2009-03-10 14:23:42 +00:00
position += name_length + 1;
value = get_value(buffer, &position, type);
2009-03-10 19:41:28 +00:00
rb_funcall(hash, rb_intern("[]="), 2, name, value);
2009-03-10 14:23:42 +00:00
}
return hash;
}
static VALUE method_deserialize(VALUE self, VALUE bson) {
const char* buffer = RSTRING_PTR(bson);
int remaining = RSTRING_LEN(bson);
2009-03-10 14:23:42 +00:00
// NOTE we just swallow the size and end byte here
buffer += 4;
remaining -= 5;
return elements_to_hash(buffer, remaining);
}
static VALUE fast_pack(VALUE self)
{
VALUE res;
long i;
char c;
res = rb_str_buf_new(0);
for (i = 0; i < RARRAY_LEN(self); i++) {
c = FIX2LONG(RARRAY_PTR(self)[i]);
rb_str_buf_cat(res, &c, sizeof(char));
}
return res;
}
static VALUE objectid_generate(VALUE self)
{
VALUE oid, digest;
char hostname[MAX_HOSTNAME_LENGTH];
unsigned char oid_bytes[12];
unsigned long t, inc;
unsigned short pid;
int i;
t = htonl(time(NULL));
MEMCPY(&oid_bytes, &t, unsigned char, 4);
2009-12-02 16:10:30 +00:00
if (gethostname(hostname, MAX_HOSTNAME_LENGTH) != 0) {
rb_raise(rb_eRuntimeError, "failed to get hostname");
}
digest = rb_funcall(DigestMD5, rb_intern("digest"), 1, rb_str_new2(hostname));
MEMCPY(&oid_bytes[4], RSTRING_PTR(digest), unsigned char, 3);
pid = htons(getpid());
MEMCPY(&oid_bytes[7], &pid, unsigned char, 2);
inc = htonl(FIX2ULONG(rb_funcall(self, rb_intern("get_inc"), 0)));
MEMCPY(&oid_bytes[9], ((unsigned char*)&inc + 1), unsigned char, 3);
oid = rb_ary_new2(12);
for(i = 0; i < 12; i++) {
rb_ary_store(oid, i, INT2FIX((unsigned int)oid_bytes[i]));
}
return oid;
}
2009-03-03 22:07:22 +00:00
void Init_cbson() {
2010-04-05 14:39:55 +00:00
VALUE bson, CBson, Digest, ext_version;
2009-03-10 17:19:18 +00:00
2010-04-05 14:39:55 +00:00
bson = rb_const_get(rb_cObject, rb_intern("BSON"));
rb_require("bson/types/binary");
Binary = rb_const_get(bson, rb_intern("Binary"));
rb_require("bson/types/objectid");
ObjectID = rb_const_get(bson, rb_intern("ObjectID"));
rb_require("bson/types/object_id");
ObjectId = rb_const_get(bson, rb_intern("ObjectId"));
2010-04-05 14:39:55 +00:00
rb_require("bson/types/dbref");
DBRef = rb_const_get(bson, rb_intern("DBRef"));
rb_require("bson/types/code");
Code = rb_const_get(bson, rb_intern("Code"));
rb_require("bson/types/min_max_keys");
MinKey = rb_const_get(bson, rb_intern("MinKey"));
MaxKey = rb_const_get(bson, rb_intern("MaxKey"));
Regexp = rb_const_get(rb_cObject, rb_intern("Regexp"));
2010-04-05 14:39:55 +00:00
rb_require("bson/exceptions");
2010-04-05 22:24:31 +00:00
InvalidKeyName = rb_const_get(bson, rb_intern("InvalidKeyName"));
2010-04-05 14:39:55 +00:00
InvalidStringEncoding = rb_const_get(bson, rb_intern("InvalidStringEncoding"));
InvalidDocument = rb_const_get(bson, rb_intern("InvalidDocument"));
rb_require("bson/ordered_hash");
2010-05-07 01:25:18 +00:00
OrderedHash = rb_const_get(bson, rb_intern("OrderedHash"));
2009-03-10 17:19:18 +00:00
CBson = rb_define_module("CBson");
2009-12-02 16:53:59 +00:00
ext_version = rb_str_new2(VERSION);
rb_define_const(CBson, "VERSION", ext_version);
2010-01-28 23:44:50 +00:00
rb_define_module_function(CBson, "serialize", method_serialize, 3);
2009-03-10 14:23:42 +00:00
rb_define_module_function(CBson, "deserialize", method_deserialize, 1);
rb_require("digest/md5");
Digest = rb_const_get(rb_cObject, rb_intern("Digest"));
DigestMD5 = rb_const_get(Digest, rb_intern("MD5"));
rb_define_method(ObjectID, "generate", objectid_generate, 0);
rb_define_method(ObjectId, "generate", objectid_generate, 0);
rb_define_method(rb_cArray, "fast_pack", fast_pack, 0);
2009-03-03 22:07:22 +00:00
}