animehunter-mongo_ext 0.3 → 0.4

data/ext/cbson/cbson.c

@@ -344,4 +344,424 @@ static int write_element_allow_id(VALUE key, VALUE value, VALUE extra, int allow
                 break;
             }
             if (strcmp(cls, "XGen::Mongo::Driver::ObjectID") == 0) {
-                VALUE as_array = rb_funcall(value, rb_intern("to_a%
+                VALUE as_array = rb_funcall(value, rb_intern("to_a"), 0);
+                int i;
+                write_name_and_type(buffer, key, 0x07);
+                for (i = 0; i < 12; i++) {
+                    char byte = (char)FIX2INT(RARRAY_PTR(as_array)[i]);
+                    buffer_write_bytes(buffer, &byte, 1);
+                }
+                break;
+            }
+            if (strcmp(cls, "XGen::Mongo::Driver::DBRef") == 0) {
+                int start_position, length_location, obj_length;
+                VALUE ns, oid;
+                write_name_and_type(buffer, key, 0x03);
+
+                start_position = buffer->position;
+
+                // save space for length
+                length_location = buffer_save_bytes(buffer, 4);
+
+                ns = rb_funcall(value, rb_intern("namespace"), 0);
+                write_element(rb_str_new2("$ref"), ns, pack_extra(buffer, Qfalse));
+                oid = rb_funcall(value, rb_intern("object_id"), 0);
+                write_element(rb_str_new2("$id"), oid, pack_extra(buffer, Qfalse));
+
+                // write null byte and fill in length
+                buffer_write_bytes(buffer, &zero, 1);
+                obj_length = buffer->position - start_position;
+                memcpy(buffer->buffer + length_location, &obj_length, 4);
+                break;
+            }
+            if (strcmp(cls, "XGen::Mongo::Driver::Undefined") == 0) {
+                write_name_and_type(buffer, key, 0x06);
+                break;
+            }
+        }
+    case T_DATA:
+        {
+            // TODO again, is this really the only way to do this?
+            const char* cls = rb_class2name(RBASIC(value)->klass);
+            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);
+                buffer_write_bytes(buffer, (const char*)&time_since_epoch, 8);
+                break;
+            }
+        }
+    case T_REGEXP:
+        {
+            int length = RREGEXP_SRC_LEN(value);
+            char* pattern = (char*)RREGEXP_SRC_PTR(value);
+            long flags = RREGEXP(value)->ptr->options;
+            VALUE has_extra;
+
+            write_name_and_type(buffer, key, 0x0B);
+
+            buffer_write_bytes(buffer, pattern, length);
+            buffer_write_bytes(buffer, &zero, 1);
+
+            if (flags & IGNORECASE) {
+                char ignorecase = 'i';
+                buffer_write_bytes(buffer, &ignorecase, 1);
+            }
+            if (flags & MULTILINE) {
+                char multiline = 'm';
+                buffer_write_bytes(buffer, &multiline, 1);
+            }
+            if (flags & EXTENDED) {
+                char extended = 'x';
+                buffer_write_bytes(buffer, &extended, 1);
+            }
+
+            has_extra = rb_funcall(value, rb_intern("respond_to?"), 1, rb_str_new2("extra_options_str"));
+            if (TYPE(has_extra) == T_TRUE) {
+                VALUE extra = rb_funcall(value, rb_intern("extra_options_str"), 0);
+                int old_position = buffer->position;
+                buffer_write_bytes(buffer, RSTRING_PTR(extra), RSTRING_LEN(extra));
+                qsort(buffer->buffer + old_position, RSTRING_LEN(extra), sizeof(char), cmp_char);
+            }
+            buffer_write_bytes(buffer, &zero, 1);
+
+            break;
+        }
+    default:
+        {
+            rb_raise(rb_eTypeError, "no c encoder for this type yet (%d)", TYPE(value));
+            break;
+        }
+    }
+    return ST_CONTINUE;
+}
+
+static int write_element(VALUE key, VALUE value, VALUE extra) {
+    return write_element_allow_id(key, value, extra, 0);
+}
+
+static void write_doc(bson_buffer* buffer, VALUE hash, VALUE check_keys) {
+    int start_position = buffer->position;
+    int length_location = buffer_save_bytes(buffer, 4);
+    int length;
+
+    VALUE key = rb_str_new2("_id");
+    if (rb_funcall(hash, rb_intern("has_key?"), 1, key) == Qtrue) {
+        VALUE id = rb_hash_aref(hash, key);
+        write_element_allow_id(key, id, pack_extra(buffer, check_keys), 1);
+    }
+    key = ID2SYM(rb_intern("_id"));
+    if (rb_funcall(hash, rb_intern("has_key?"), 1, key) == Qtrue) {
+        VALUE id = rb_hash_aref(hash, key);
+        write_element_allow_id(key, id, pack_extra(buffer, check_keys), 1);
+    }
+
+    // we have to check for an OrderedHash and handle that specially
+    if (strcmp(rb_class2name(RBASIC(hash)->klass), "OrderedHash") == 0) {
+        VALUE keys = rb_funcall(hash, rb_intern("keys"), 0);
+        int i;
+        for(i = 0; i < RARRAY_LEN(keys); i++) {
+            VALUE key = RARRAY_PTR(keys)[i];
+            VALUE value = rb_hash_aref(hash, key);
+
+            write_element(key, value, pack_extra(buffer, check_keys));
+        }
+    } else {
+        rb_hash_foreach(hash, write_element, pack_extra(buffer, check_keys));
+    }
+
+    // write null byte and fill in length
+    buffer_write_bytes(buffer, &zero, 1);
+    length = buffer->position - start_position;
+    memcpy(buffer->buffer + length_location, &length, 4);
+}
+
+static VALUE method_serialize(VALUE self, VALUE doc, VALUE check_keys) {
+    VALUE result;
+    bson_buffer* buffer = buffer_new();
+    assert(buffer);
+
+    write_doc(buffer, doc, check_keys);
+
+    result = rb_str_new(buffer->buffer, buffer->position);
+    buffer_free(buffer);
+    return result;
+}
+
+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:
+        {
+            int value_length;
+            *position += 4;
+            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);
+                char id_type;
+
+                argv[0] = rb_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 = 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];
+            data;
+            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 = 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;
+            VALUE seconds, microseconds;
+            memcpy(&millis, buffer + *position, 8);
+            seconds = INT2NUM(millis / 1000);
+            microseconds = INT2NUM((millis % 1000) * 1000);
+
+            value = rb_funcall(Time, rb_intern("at"), 2, seconds, microseconds);
+            value = rb_funcall(value, rb_intern("utc"), 0);
+            *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);
+            int flags_length, flags = 0, i = 0;
+            char extra[10];
+            VALUE argv[3];
+            *position += pattern_length + 1;
+
+            flags_length = strlen(buffer + *position);
+            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);
+                }
+            }
+            argv[0] = pattern;
+            argv[1] = INT2FIX(flags);
+            argv[2] = rb_str_new2(extra);
+            value = rb_class_new_instance(3, argv, RegexpOfHolding);
+            *position += flags_length + 1;
+            break;
+        }
+    case 12:
+        {
+            int collection_length;
+            VALUE collection, str, oid, id, argv[2];
+            *position += 4;
+            collection_length = strlen(buffer + *position);
+            collection = rb_str_new(buffer + *position, collection_length);
+            *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);
+            *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 += 8;
+            code_length = strlen(buffer + *position);
+            code = rb_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 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 elements_to_hash(const char* buffer, int max) {
+    VALUE hash = rb_class_new_instance(0, NULL, OrderedHash);
+    int position = 0;
+    while (position < max) {
+        int type = (int)buffer[position++];
+        int name_length = strlen(buffer + position);
+        VALUE name = rb_str_new(buffer + position, name_length);
+        VALUE value;
+        position += name_length + 1;
+        value = get_value(buffer, &position, type);
+        rb_funcall(hash, rb_intern("[]="), 2, name, value);
+    }
+    return hash;
+}
+
+static VALUE method_deserialize(VALUE self, VALUE bson) {
+    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);
+}
+
+void Init_cbson() {
+    VALUE driver, CBson;
+    Time = rb_const_get(rb_cObject, rb_intern("Time"));
+
+    driver = rb_const_get(rb_const_get(rb_const_get(rb_cObject,
+                                                    rb_intern("XGen")),
+                                       rb_intern("Mongo")),
+                          rb_intern("Driver"));
+    rb_require("mongo/types/binary");
+    Binary = rb_const_get(driver, rb_intern("Binary"));
+    rb_require("mongo/types/undefined");
+    Undefined = rb_const_get(driver, rb_intern("Undefined"));
+    rb_require("mongo/types/objectid");
+    ObjectID = rb_const_get(driver, rb_intern("ObjectID"));
+    rb_require("mongo/types/dbref");
+    DBRef = rb_const_get(driver, rb_intern("DBRef"));
+    rb_require("mongo/types/code");
+    Code = rb_const_get(driver, rb_intern("Code"));
+    rb_require("mongo/types/regexp_of_holding");
+    RegexpOfHolding = rb_const_get(driver, rb_intern("RegexpOfHolding"));
+    rb_require("mongo/util/ordered_hash");
+    OrderedHash = rb_const_get(rb_cObject, rb_intern("OrderedHash"));
+
+    CBson = rb_define_module("CBson");
+    rb_define_module_function(CBson, "serialize", method_serialize, 2);
+    rb_define_module_function(CBson, "deserialize", method_deserialize, 1);
+}

data/mongo-extensions.gemspec

@@ -7,7 +7,7 @@ TEST_FILES = []
 
 Gem::Specification.new do |s|
   s.name = 'mongo_ext'
-  s.version = '0.3'
+  s.version = '0.4'
   s.platform = Gem::Platform::RUBY
   s.summary = 'C extensions for the MongoDB Ruby driver'
   s.description = 'C extensions to accelerate the MondoDB Ruby driver. For more information about Mongo, see http://www.mongodb.org.'

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: animehunter-mongo_ext
 version: !ruby/object:Gem::Version 
-  version: "0.3"
+  version: "0.4"
 platform: ruby
 authors: 
 - Mike Dirolf