mongo_ext 0.15

data/Rakefile

@@ -0,0 +1,62 @@
+require 'rubygems'
+require 'rubygems/specification'
+require 'fileutils'
+require 'rake'
+require 'rake/testtask'
+require 'rake/gempackagetask'
+begin
+  require 'rake/contrib/rubyforgepublisher'
+rescue LoadError
+end
+require 'rbconfig'
+include Config
+
+gem_command = "gem"
+gem_command = "gem1.9" if $0.match(/1\.9$/) # use gem1.9 if we used rake1.9
+
+# NOTE: some of the tests assume Mongo is running
+Rake::TestTask.new do |t|
+  t.test_files = FileList['test/test*.rb']
+end
+
+desc "Generate documentation"
+task :rdoc do
+  version = eval(File.read("mongo-ruby-driver.gemspec")).version
+  out = File.join('html', version.to_s)
+  FileUtils.rm_rf('html')
+  system "rdoc --main README.rdoc --op #{out} --inline-source --quiet README.rdoc `find lib -name '*.rb'`"
+end
+
+desc "Publish documentation to mongo.rubyforge.org"
+task :publish => [:rdoc] do
+  # Assumes docs are in ./html
+  Rake::RubyForgePublisher.new(GEM, RUBYFORGE_USER).upload
+end
+
+namespace :gem do
+
+  desc "Install the gem locally"
+  task :install do
+    sh <<EOS
+#{gem_command} build mongo-ruby-driver.gemspec &&
+    sudo #{gem_command} install mongo-*.gem &&
+    rm mongo-*.gem
+EOS
+  end
+
+  desc "Install the optional c extensions"
+  task :install_extensions do
+    sh <<EOS
+#{gem_command} build mongo-extensions.gemspec &&
+    sudo #{gem_command} install mongo_ext-*.gem &&
+    rm mongo_ext-*.gem
+EOS
+  end
+
+end
+
+task :default => :list
+
+task :list do
+  system 'rake -T'
+end

data/ext/cbson/cbson.c

@@ -0,0 +1,764 @@
+/*
+ * Copyright 2009 10gen, Inc.
+ *
+ * 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.
+ */
+
+#include "ruby.h"
+
+#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 <assert.h>
+#include <math.h>
+
+#define INITIAL_BUFFER_SIZE 256
+
+static VALUE Binary;
+static VALUE Time;
+static VALUE ObjectID;
+static VALUE DBRef;
+static VALUE Code;
+static VALUE RegexpOfHolding;
+static VALUE OrderedHash;
+static VALUE InvalidName;
+
+#if HAVE_RUBY_ENCODING_H
+#include "ruby/encoding.h"
+#define STR_NEW(p,n) rb_enc_str_new((p), (n), rb_utf8_encoding())
+#else
+#define STR_NEW(p,n) rb_str_new((p), (n))
+#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. */
+#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) asprintf(buffer, "%d", i);
+#endif
+
+// this sucks too.
+#ifndef RREGEXP_SRC_PTR
+#define RREGEXP_SRC_PTR(r) RREGEXP(r)->str
+#define RREGEXP_SRC_LEN(r) RREGEXP(r)->len
+#endif
+
+typedef struct {
+    char* buffer;
+    int size;
+    int position;
+} bson_buffer;
+
+static char zero = 0;
+static char one = 1;
+
+static int cmp_char(const void* a, const void* b) {
+    return *(char*)a - *(char*)b;
+}
+
+static void write_doc(bson_buffer* buffer, VALUE hash, VALUE check_keys);
+static int write_element(VALUE key, VALUE value, VALUE extra);
+static VALUE elements_to_hash(const char* buffer, int max);
+
+static bson_buffer* buffer_new(void) {
+    bson_buffer* buffer;
+    buffer = ALLOC(bson_buffer);
+    assert(buffer);
+
+    buffer->size = INITIAL_BUFFER_SIZE;
+    buffer->position = 0;
+    buffer->buffer = ALLOC_N(char, INITIAL_BUFFER_SIZE);
+    assert(buffer->buffer);
+
+    return buffer;
+}
+
+static void buffer_free(bson_buffer* buffer) {
+    assert(buffer);
+    assert(buffer->buffer);
+
+    free(buffer->buffer);
+    free(buffer);
+}
+
+static void buffer_resize(bson_buffer* buffer, int min_length) {
+    int size = buffer->size;
+    if (size >= min_length) {
+        return;
+    }
+    while (size < min_length) {
+        size *= 2;
+    }
+    buffer->buffer = REALLOC_N(buffer->buffer, char, size);
+    assert(buffer->buffer);
+    buffer->size = size;
+}
+
+static void buffer_assure_space(bson_buffer* buffer, int size) {
+    if (buffer->position + size <= buffer->size) {
+        return;
+    }
+    buffer_resize(buffer, buffer->position + size);
+}
+
+/* returns offset for writing */
+static int buffer_save_bytes(bson_buffer* buffer, int size) {
+    int position = buffer->position;
+    buffer_assure_space(buffer, size);
+    buffer->position += size;
+    return position;
+}
+
+static void buffer_write_bytes(bson_buffer* buffer, const char* bytes, int size) {
+    buffer_assure_space(buffer, size);
+
+    memcpy(buffer->buffer + buffer->position, bytes, size);
+    buffer->position += size;
+}
+
+static VALUE pack_extra(bson_buffer* buffer, VALUE check_keys) {
+    return rb_ary_new3(2, LL2NUM((long long)buffer), check_keys);
+}
+
+static void write_name_and_type(bson_buffer* buffer, VALUE name, char type) {
+    buffer_write_bytes(buffer, &type, 1);
+    buffer_write_bytes(buffer, RSTRING_PTR(name), RSTRING_LEN(name));
+    buffer_write_bytes(buffer, &zero, 1);
+}
+
+static int write_element_allow_id(VALUE key, VALUE value, VALUE extra, int allow_id) {
+    bson_buffer* buffer = (bson_buffer*)NUM2LL(rb_ary_entry(extra, 0));
+    VALUE check_keys = rb_ary_entry(extra, 1);
+
+    if (TYPE(key) == T_SYMBOL) {
+        // TODO better way to do this... ?
+        key = rb_str_new2(rb_id2name(SYM2ID(key)));
+    }
+
+    if (TYPE(key) != T_STRING) {
+        rb_raise(rb_eTypeError, "keys must be strings or symbols");
+    }
+
+    if (!allow_id && strcmp("_id", RSTRING_PTR(key)) == 0) {
+        return ST_CONTINUE;
+    }
+
+    if (check_keys == Qtrue) {
+        int i;
+        if (RSTRING_LEN(key) > 0 && RSTRING_PTR(key)[0] == '$') {
+            rb_raise(InvalidName, "key must not start with '$'");
+        }
+        for (i = 0; i < RSTRING_LEN(key); i++) {
+            if (RSTRING_PTR(key)[i] == '.') {
+                rb_raise(InvalidName, "key must not contain '.'");
+            }
+        }
+    }
+
+    switch(TYPE(value)) {
+    case T_BIGNUM:
+    case T_FIXNUM:
+        {
+            if (rb_funcall(value, rb_intern(">"), 1, LL2NUM(9223372036854775807LL)) == Qtrue ||
+                rb_funcall(value, rb_intern("<"), 1, LL2NUM(-9223372036854775808LL)) == Qtrue) {
+                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);
+                buffer_write_bytes(buffer, (char*)&ll_value, 8);
+            } else {
+                int int_value;
+                write_name_and_type(buffer, key, 0x10);
+                int_value = NUM2LL(value);
+                buffer_write_bytes(buffer, (char*)&int_value, 4);
+            }
+            break;
+        }
+    case T_TRUE:
+        {
+            write_name_and_type(buffer, key, 0x08);
+            buffer_write_bytes(buffer, &one, 1);
+            break;
+        }
+    case T_FALSE:
+        {
+            write_name_and_type(buffer, key, 0x08);
+            buffer_write_bytes(buffer, &zero, 1);
+            break;
+        }
+    case T_FLOAT:
+        {
+            double d = NUM2DBL(value);
+            write_name_and_type(buffer, key, 0x01);
+            buffer_write_bytes(buffer, (char*)&d, 8);
+            break;
+        }
+    case T_NIL:
+        {
+            write_name_and_type(buffer, key, 0x0A);
+            break;
+        }
+    case T_HASH:
+        {
+            write_name_and_type(buffer, key, 0x03);
+            write_doc(buffer, value, check_keys);
+            break;
+        }
+    case T_ARRAY:
+        {
+            int start_position, length_location, items, i, obj_length;
+            VALUE* values;
+
+            write_name_and_type(buffer, key, 0x04);
+            start_position = buffer->position;
+
+            // save space for length
+            length_location = buffer_save_bytes(buffer, 4);
+
+            items = RARRAY_LEN(value);
+            values = RARRAY_PTR(value);
+            for(i = 0; i < items; i++) {
+                char* name;
+                VALUE key;
+                INT2STRING(&name, i);
+                key = rb_str_new2(name);
+                write_element(key, values[i], pack_extra(buffer, check_keys));
+                free(name);
+            }
+
+            // 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;
+        }
+    case T_STRING:
+        {
+            if (strcmp(rb_class2name(RBASIC(value)->klass),
+                       "Mongo::Code") == 0) {
+                int start_position, length_location, length, total_length;
+                write_name_and_type(buffer, key, 0x0F);
+
+                start_position = buffer->position;
+                length_location = buffer_save_bytes(buffer, 4);
+
+                length = RSTRING_LEN(value) + 1;
+                buffer_write_bytes(buffer, (char*)&length, 4);
+                buffer_write_bytes(buffer, RSTRING_PTR(value), length - 1);
+                buffer_write_bytes(buffer, &zero, 1);
+                write_doc(buffer, rb_funcall(value, rb_intern("scope"), 0), Qfalse);
+
+                total_length = buffer->position - start_position;
+                memcpy(buffer->buffer + length_location, &total_length, 4);
+
+                break;
+            } else {
+                int length = RSTRING_LEN(value) + 1;
+                write_name_and_type(buffer, key, 0x02);
+                buffer_write_bytes(buffer, (char*)&length, 4);
+                buffer_write_bytes(buffer, RSTRING_PTR(value), length - 1);
+                buffer_write_bytes(buffer, &zero, 1);
+                break;
+            }
+        }
+    case T_SYMBOL:
+        {
+            const char* str_value = rb_id2name(SYM2ID(value));
+            int length = strlen(str_value) + 1;
+            write_name_and_type(buffer, key, 0x0E);
+            buffer_write_bytes(buffer, (char*)&length, 4);
+            buffer_write_bytes(buffer, str_value, length);
+            break;
+        }
+    case T_OBJECT:
+        {
+            // TODO there has to be a better way to do these checks...
+            const char* cls = rb_class2name(RBASIC(value)->klass);
+            if (strcmp(cls, "Mongo::Binary") == 0 ||
+                strcmp(cls, "ByteBuffer") == 0) {
+                const char subtype = strcmp(cls, "ByteBuffer") ?
+                    (const char)FIX2INT(rb_funcall(value, rb_intern("subtype"), 0)) : 2;
+                VALUE string_data = rb_funcall(value, rb_intern("to_s"), 0);
+                int length = RSTRING_LEN(string_data);
+                write_name_and_type(buffer, key, 0x05);
+                if (subtype == 2) {
+                    const int other_length = length + 4;
+                    buffer_write_bytes(buffer, (const char*)&other_length, 4);
+                    buffer_write_bytes(buffer, &subtype, 1);
+                }
+                buffer_write_bytes(buffer, (const char*)&length, 4);
+                if (subtype != 2) {
+                    buffer_write_bytes(buffer, &subtype, 1);
+                }
+                buffer_write_bytes(buffer, RSTRING_PTR(string_data), length);
+                break;
+            }
+            if (strcmp(cls, "Mongo::ObjectID") == 0) {
+                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, "Mongo::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;
+            }
+        }
+    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 = 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] = 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];
+            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, 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 = 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 = 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 = 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;
+        }
+    case 18:
+        {
+            long long ll;
+            memcpy(&ll, buffer + *position, 8);
+            value = LL2NUM(ll);
+            *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 = 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 mongo, CBson;
+    Time = rb_const_get(rb_cObject, rb_intern("Time"));
+
+    mongo = rb_const_get(rb_cObject, rb_intern("Mongo"));
+    rb_require("mongo/types/binary");
+    Binary = rb_const_get(mongo, rb_intern("Binary"));
+    rb_require("mongo/types/objectid");
+    ObjectID = rb_const_get(mongo, rb_intern("ObjectID"));
+    rb_require("mongo/types/dbref");
+    DBRef = rb_const_get(mongo, rb_intern("DBRef"));
+    rb_require("mongo/types/code");
+    Code = rb_const_get(mongo, rb_intern("Code"));
+    rb_require("mongo/types/regexp_of_holding");
+    RegexpOfHolding = rb_const_get(mongo, rb_intern("RegexpOfHolding"));
+    rb_require("mongo/errors");
+    InvalidName = rb_const_get(mongo, rb_intern("InvalidName"));
+    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/ext/cbson/extconf.rb

@@ -0,0 +1,8 @@
+require 'mkmf'
+
+have_header("ruby/st.h") || have_header("st.h")
+have_header("ruby/regex.h") || have_header("regex.h")
+have_header("ruby/encoding.h")
+
+dir_config('cbson')
+create_makefile('mongo_ext/cbson')

data/mongo-extensions.gemspec

@@ -0,0 +1,28 @@
+# We need to list all of the included files because we aren't allowed to use
+# Dir[...] in the github sandbox.
+PACKAGE_FILES = ['Rakefile', 'mongo-extensions.gemspec',
+                 'ext/cbson/cbson.c',
+                 'ext/cbson/extconf.rb']
+TEST_FILES = []
+
+Gem::Specification.new do |s|
+  s.name = 'mongo_ext'
+
+  s.version = '0.15'
+  s.add_dependency('mongo', ['= 0.15'])
+
+  s.platform = Gem::Platform::RUBY
+  s.summary = 'C extensions for the MongoDB Ruby driver'
+  s.description = 'C extensions to accelerate the MongoDB Ruby driver. For more information about Mongo, see http://www.mongodb.org.'
+
+  s.require_paths = ['ext']
+  s.files = PACKAGE_FILES
+  s.test_files = TEST_FILES
+
+  s.has_rdoc = false
+  s.extensions << 'ext/cbson/extconf.rb'
+
+  s.author = 'Mike Dirolf'
+  s.email = 'mongodb-dev@googlegroups.com'
+  s.homepage = 'http://www.mongodb.org'
+end

metadata

@@ -0,0 +1,67 @@
+--- !ruby/object:Gem::Specification 
+name: mongo_ext
+version: !ruby/object:Gem::Version 
+  version: "0.15"
+platform: ruby
+authors: 
+- Mike Dirolf
+autorequire: 
+bindir: bin
+cert_chain: []
+
+date: 2009-10-05 00:00:00 -04:00
+default_executable: 
+dependencies: 
+- !ruby/object:Gem::Dependency 
+  name: mongo
+  type: :runtime
+  version_requirement: 
+  version_requirements: !ruby/object:Gem::Requirement 
+    requirements: 
+    - - "="
+      - !ruby/object:Gem::Version 
+        version: "0.15"
+    version: 
+description: C extensions to accelerate the MongoDB Ruby driver. For more information about Mongo, see http://www.mongodb.org.
+email: mongodb-dev@googlegroups.com
+executables: []
+
+extensions: 
+- ext/cbson/extconf.rb
+extra_rdoc_files: []
+
+files: 
+- Rakefile
+- mongo-extensions.gemspec
+- ext/cbson/cbson.c
+- ext/cbson/extconf.rb
+has_rdoc: true
+homepage: http://www.mongodb.org
+licenses: []
+
+post_install_message: 
+rdoc_options: []
+
+require_paths: 
+- ext
+required_ruby_version: !ruby/object:Gem::Requirement 
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      version: "0"
+  version: 
+required_rubygems_version: !ruby/object:Gem::Requirement 
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      version: "0"
+  version: 
+requirements: []
+
+rubyforge_project: 
+rubygems_version: 1.3.5
+signing_key: 
+specification_version: 3
+summary: C extensions for the MongoDB Ruby driver
+test_files: []
+