static VALUE method_deserialize(VALUE self, VALUE bson) {
const char* buffer = RSTRING_PTR(bson);
int remaining = RSTRING_LENINT(bson);
// NOTE we just swallow the size and end byte here
buffer += 4;
remaining -= 5;
return elements_to_hash(buffer, remaining);
}
static VALUE method_deserialize(int argc, VALUE *argv, VALUE self) {
VALUE bson;
VALUE symbolize_keys = 0;
rb_scan_args(argc, argv, "11", &bson, &symbolize_keys);
const char* buffer = RSTRING_PTR(bson);
int remaining = RSTRING_LEN(bson);
// NOTE we just swallow the size and end byte here
buffer += 4;
remaining -= 5;
return elements_to_hash(buffer, remaining, symbolize_keys);
}
static VALUE method_deserialize(VALUE self, VALUE bson, VALUE opts) {
const char* buffer = RSTRING_PTR(bson);
int remaining = RSTRING_LENINT(bson);
struct deserialize_opts deserialize_opts;
deserialize_opts.compile_regex = 1;
if (rb_funcall(opts, rb_intern("has_key?"), 1, ID2SYM(rb_intern("compile_regex"))) == Qtrue &&
rb_hash_aref(opts, ID2SYM(rb_intern("compile_regex"))) == Qfalse) {
deserialize_opts.compile_regex = 0;
}
// NOTE we just swallow the size and end byte here
buffer += 4;
remaining -= 5;
return elements_to_hash(buffer, remaining, &deserialize_opts);
}
static VALUE get_value(const char* buffer, int* position, int type) {
VALUE value;
switch (type) {
case -1:
{
value = rb_class_new_instance(0, NULL, MinKey);
break;
}
case 1:
{
double d;
memcpy(&d, buffer + *position, 8);
value = rb_float_new(d);
*position += 8;
break;
}
case 2:
case 13:
{
int value_length;
value_length = *(int*)(buffer + *position) - 1;
*position += 4;
value = STR_NEW(buffer + *position, value_length);
*position += value_length + 1;
break;
}
case 3:
{
int size;
memcpy(&size, buffer + *position, 4);
if (strcmp(buffer + *position + 5, "$ref") == 0) { // DBRef
int offset = *position + 10;
VALUE argv[2];
int collection_length = *(int*)(buffer + offset) - 1;
char id_type;
offset += 4;
argv[0] = STR_NEW(buffer + offset, collection_length);
offset += collection_length + 1;
id_type = buffer[offset];
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 = (int)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;
}
case 5:
{
int length, subtype;
VALUE data, st;
VALUE argv[2];
memcpy(&length, buffer + *position, 4);
subtype = (unsigned char)buffer[*position + 4];
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;
value = rb_class_new_instance(2, argv, Binary);
*position += length + 5;
break;
}
case 6:
{
value = Qnil;
break;
}
case 7:
{
VALUE str = rb_str_new(buffer + *position, 12);
VALUE oid = rb_funcall(str, unpack_method, 1, rb_str_new2("C*"));
value = rb_class_new_instance(1, &oid, ObjectId);
*position += 12;
break;
}
case 8:
{
value = buffer[(*position)++] ? Qtrue : Qfalse;
break;
}
case 9:
{
long long millis;
memcpy(&millis, buffer + *position, 8);
value = rb_time_new(millis / 1000, (millis % 1000) * 1000);
value = rb_funcall(value, utc_method, 0);
*position += 8;
break;
}
case 10:
{
value = Qnil;
break;
}
case 11:
{
int pattern_length = (int)strlen(buffer + *position);
VALUE pattern = STR_NEW(buffer + *position, pattern_length);
int flags_length, flags = 0, i = 0;
VALUE argv[3];
*position += pattern_length + 1;
flags_length = (int)strlen(buffer + *position);
for (i = 0; i < flags_length; i++) {
char flag = buffer[*position + i];
if (flag == 'i') {
flags |= IGNORECASE;
}
else if (flag == 'm') {
flags |= MULTILINE;
}
else if (flag == 's') {
flags |= MULTILINE;
}
else if (flag == 'x') {
flags |= EXTENDED;
}
}
argv[0] = pattern;
argv[1] = INT2FIX(flags);
value = rb_class_new_instance(2, argv, Regexp);
*position += flags_length + 1;
break;
}
case 12:
{
int collection_length;
VALUE collection, str, oid, id, argv[2];
collection_length = *(int*)(buffer + *position) - 1;
*position += 4;
collection = STR_NEW(buffer + *position, collection_length);
*position += collection_length + 1;
str = rb_str_new(buffer + *position, 12);
oid = rb_funcall(str, unpack_method, 1, rb_str_new2("C*"));
id = rb_class_new_instance(1, &oid, ObjectId);
*position += 12;
argv[0] = collection;
argv[1] = id;
value = rb_class_new_instance(2, argv, DBRef);
break;
}
case 14:
{
int value_length;
memcpy(&value_length, buffer + *position, 4);
value = ID2SYM(rb_intern(buffer + *position + 4));
*position += value_length + 4;
break;
}
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);
*position += code_length + 1;
memcpy(&scope_size, buffer + *position, 4);
scope = elements_to_hash(buffer + *position + 4, scope_size - 5);
*position += scope_size;
argv[0] = code;
argv[1] = scope;
value = rb_class_new_instance(2, argv, Code);
break;
}
case 16:
{
int i;
memcpy(&i, buffer + *position, 4);
value = LL2NUM(i);
*position += 4;
break;
}
case 17:
{
int sec, inc;
VALUE argv[2];
memcpy(&inc, buffer + *position, 4);
memcpy(&sec, buffer + *position + 4, 4);
argv[0] = INT2FIX(sec);
argv[1] = INT2FIX(inc);
value = rb_class_new_instance(2, argv, Timestamp);
*position += 8;
break;
}
case 18:
{
long long ll;
memcpy(&ll, buffer + *position, 8);
value = LL2NUM(ll);
*position += 8;
break;
}
case 127:
{
value = rb_class_new_instance(0, NULL, MaxKey);
break;
}
default:
{
rb_raise(rb_eTypeError, "no c decoder for this type yet (%d)", type);
break;
}
}
return value;
}
static VALUE get_value(const char* buffer, int* position, int type) {
VALUE value;
switch (type) {
case 1:
{
double d;
memcpy(&d, buffer + *position, 8);
value = rb_float_new(d);
*position += 8;
break;
}
case 2:
case 13:
{
*position += 4;
int value_length = strlen(buffer + *position);
value = rb_str_new(buffer+ *position, value_length);
*position += value_length + 1;
break;
}
case 3:
{
int size;
memcpy(&size, buffer + *position, 4);
if (strcmp(buffer + *position + 5, "$ref") == 0) { // DBRef
int offset = *position + 14;
VALUE argv[2];
int collection_length = strlen(buffer + offset);
argv[0] = rb_str_new(buffer + offset, collection_length);
offset += collection_length + 1;
char id_type = buffer[offset];
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;
memcpy(&size, buffer + *position, 4);
int end = *position + size - 1;
*position += 4;
value = rb_ary_new();
while (*position < end) {
int type = (int)buffer[(*position)++];
int key_size = strlen(buffer + *position);
*position += key_size + 1; // just skip the key, they're in order.
VALUE to_append = get_value(buffer, position, type);
rb_ary_push(value, to_append);
}
(*position)++;
break;
}
case 5:
{
int length;
memcpy(&length, buffer + *position, 4);
int subtype = (unsigned char)buffer[*position + 4];
VALUE data;
if (subtype == 2) {
data = rb_str_new(buffer + *position + 9, length - 4);
} else {
data = rb_str_new(buffer + *position + 5, length);
}
VALUE st = INT2FIX(subtype);
VALUE argv[2] = {data, st};
value = rb_class_new_instance(2, argv, Binary);
*position += length + 5;
break;
}
case 6:
{
value = rb_class_new_instance(0, NULL, Undefined);
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);
*position += 12;
break;
}
case 8:
{
value = buffer[(*position)++] ? Qtrue : Qfalse;
break;
}
case 9:
{
long long millis;
memcpy(&millis, buffer + *position, 8);
VALUE seconds = INT2NUM(millis / 1000);
VALUE microseconds = INT2NUM((millis % 1000) * 1000);
value = rb_funcall(Time, rb_intern("at"), 2, seconds, microseconds);
*position += 8;
break;
}
case 10:
{
value = Qnil;
break;
}
case 11:
{
int pattern_length = strlen(buffer + *position);
VALUE pattern = rb_str_new(buffer + *position, pattern_length);
*position += pattern_length + 1;
int flags_length = strlen(buffer + *position);
int i = 0;
int flags = 0;
char extra[10];
extra[0] = 0;
for (i = 0; i < flags_length; i++) {
char flag = buffer[*position + i];
if (flag == 'i') {
flags |= IGNORECASE;
}
else if (flag == 'm') {
flags |= MULTILINE;
}
else if (flag == 'x') {
flags |= EXTENDED;
}
else if (strlen(extra) < 9) {
strncat(extra, &flag, 1);
}
}
VALUE argv[3] = {
pattern,
INT2FIX(flags),
rb_str_new2(extra)
};
value = rb_class_new_instance(3, argv, RegexpOfHolding);
*position += flags_length + 1;
break;
}
case 12:
{
*position += 4;
int collection_length = strlen(buffer + *position);
VALUE collection = rb_str_new(buffer + *position, collection_length);
*position += collection_length + 1;
VALUE str = rb_str_new(buffer + *position, 12);
VALUE oid = rb_funcall(str, rb_intern("unpack"), 1, rb_str_new2("C*"));
VALUE id = rb_class_new_instance(1, &oid, ObjectID);
*position += 12;
VALUE argv[2] = {collection, id};
value = rb_class_new_instance(2, argv, DBRef);
break;
}
case 14:
{
int value_length;
memcpy(&value_length, buffer + *position, 4);
value = ID2SYM(rb_intern(buffer + *position + 4));
*position += value_length + 4;
break;
}
case 15:
{
*position += 8;
int code_length = strlen(buffer + *position);
VALUE code = rb_str_new(buffer + *position, code_length);
*position += code_length + 1;
int scope_size;
memcpy(&scope_size, buffer + *position, 4);
VALUE scope = elements_to_hash(buffer + *position + 4, scope_size - 5);
*position += scope_size;
VALUE argv[2] = {code, scope};
value = rb_class_new_instance(2, argv, Code);
break;
}
case 16:
{
int i;
memcpy(&i, buffer + *position, 4);
value = LL2NUM(i);
*position += 4;
break;
}
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;
}
default:
{
rb_raise(rb_eTypeError, "no c decoder for this type yet (%d)", type);
break;
}
}
return value;
}
static VALUE get_value(const char* buffer, int* position,
unsigned char type, struct deserialize_opts * opts) {
VALUE value;
switch (type) {
case 255:
{
value = rb_class_new_instance(0, NULL, MinKey);
break;
}
case 1:
{
double d;
memcpy(&d, buffer + *position, 8);
value = rb_float_new(d);
*position += 8;
break;
}
case 2:
case 13:
{
int value_length;
value_length = *(int*)(buffer + *position) - 1;
*position += 4;
value = STR_NEW(buffer + *position, value_length);
*position += value_length + 1;
break;
}
case 3:
{
int size;
memcpy(&size, buffer + *position, 4);
if (strcmp(buffer + *position + 5, "$ref") == 0) { // DBRef
VALUE argv[2];
unsigned char id_type;
int offset = *position + 10;
int collection_length = *(int*)(buffer + offset) - 1;
offset += 4;
argv[0] = STR_NEW(buffer + offset, collection_length);
offset += collection_length + 1;
id_type = (unsigned char)buffer[offset];
offset += 5;
argv[1] = get_value(buffer, &offset, id_type, opts);
value = rb_class_new_instance(2, argv, DBRef);
} else {
value = elements_to_hash(buffer + *position + 4, size - 5, opts);
}
*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) {
VALUE to_append;
unsigned char type = (unsigned char)buffer[(*position)++];
int key_size = (int)strlen(buffer + *position);
*position += key_size + 1; // just skip the key, they're in order.
to_append = get_value(buffer, position, type, opts);
rb_ary_push(value, to_append);
}
(*position)++;
break;
}
case 5:
{
int length, subtype;
VALUE data, st;
VALUE argv[2];
memcpy(&length, buffer + *position, 4);
subtype = (unsigned char)buffer[*position + 4];
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;
value = rb_class_new_instance(2, argv, Binary);
*position += length + 5;
break;
}
case 6:
{
value = Qnil;
break;
}
case 7:
{
VALUE str = rb_str_new(buffer + *position, 12);
VALUE oid = rb_funcall(str, unpack_method, 1, rb_str_new2("C*"));
value = rb_class_new_instance(1, &oid, ObjectId);
*position += 12;
break;
}
case 8:
{
value = buffer[(*position)++] ? Qtrue : Qfalse;
break;
}
case 9:
{
int64_t millis;
memcpy(&millis, buffer + *position, 8);
// Support 64-bit time values in 32 bit environments in Ruby > 1.9
// Note: rb_time_num_new is not available pre Ruby 1.9
#if RUBY_API_VERSION_CODE >= 10900
#define add(x,y) (rb_funcall((x), '+', 1, (y)))
#define mul(x,y) (rb_funcall((x), '*', 1, (y)))
#define quo(x,y) (rb_funcall((x), rb_intern("quo"), 1, (y)))
VALUE d, timev;
d = LL2NUM(1000LL);
timev = add(LL2NUM(millis / 1000), quo(LL2NUM(millis % 1000), d));
value = rb_time_num_new(timev, Qnil);
#else
value = rb_time_new(millis / 1000, (millis % 1000) * 1000);
#endif
value = rb_funcall(value, utc_method, 0);
*position += 8;
break;
}
case 10:
{
value = Qnil;
break;
}
case 11:
{
int pattern_length = (int)strlen(buffer + *position);
VALUE pattern = STR_NEW(buffer + *position, pattern_length);
int flags_length;
VALUE argv[3], flags_str;
*position += pattern_length + 1;
flags_length = (int)strlen(buffer + *position);
flags_str = STR_NEW(buffer + *position, flags_length);
argv[0] = pattern;
argv[1] = flags_str;
value = rb_class_new_instance(2, argv, BSONRegex);
if (opts->compile_regex == 1) {
value = rb_funcall(value, rb_intern("try_compile"), 0);
}
*position += flags_length + 1;
break;
}
case 12:
{
int collection_length;
VALUE collection, str, oid, id, argv[2];
collection_length = *(int*)(buffer + *position) - 1;
*position += 4;
collection = STR_NEW(buffer + *position, collection_length);
*position += collection_length + 1;
str = rb_str_new(buffer + *position, 12);
oid = rb_funcall(str, unpack_method, 1, rb_str_new2("C*"));
id = rb_class_new_instance(1, &oid, ObjectId);
*position += 12;
argv[0] = collection;
argv[1] = id;
value = rb_class_new_instance(2, argv, DBRef);
break;
}
case 14:
{
int value_length;
memcpy(&value_length, buffer + *position, 4);
value = ID2SYM(rb_intern(buffer + *position + 4));
*position += value_length + 4;
break;
}
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);
*position += code_length + 1;
memcpy(&scope_size, buffer + *position, 4);
scope = elements_to_hash(buffer + *position + 4, scope_size - 5, opts);
*position += scope_size;
argv[0] = code;
argv[1] = scope;
value = rb_class_new_instance(2, argv, Code);
break;
}
case 16:
{
int i;
memcpy(&i, buffer + *position, 4);
value = LL2NUM(i);
*position += 4;
break;
}
case 17:
{
unsigned int sec, inc;
VALUE argv[2];
memcpy(&inc, buffer + *position, 4);
memcpy(&sec, buffer + *position + 4, 4);
argv[0] = UINT2NUM(sec);
argv[1] = UINT2NUM(inc);
value = rb_class_new_instance(2, argv, Timestamp);
*position += 8;
break;
}
case 18:
{
long long ll;
memcpy(&ll, buffer + *position, 8);
value = LL2NUM(ll);
*position += 8;
break;
}
case 127:
{
value = rb_class_new_instance(0, NULL, MaxKey);
break;
}
default:
{
rb_raise(rb_eTypeError, "no c decoder for this type yet (%d)", type);
break;
}
}
return value;
}
