Optimized callback class to return Java objects

This commit is contained in:
Kyle Banker 2010-10-01 09:52:29 -04:00
parent 2daf86bb7b
commit 315e608cd4
11 changed files with 445 additions and 58 deletions

View File

@ -70,7 +70,7 @@ def profile(str)
end
end
def benchmark(str, n, coll_name, data, create_index=false)
def benchmark(str, n, coll_name, data, create_index=false, verbosity=true)
coll = @db.collection(coll_name)
coll.create_index('x') if create_index
profile(str) do
@ -79,7 +79,7 @@ def benchmark(str, n, coll_name, data, create_index=false)
td = tm.add do
n.times { |i| yield(coll, i) }
end
report(str, td.real, td.utime)
report(str, td.real, td.utime) if verbosity
end
end
@ -90,8 +90,8 @@ connection = Connection.new(host, port)
connection.drop_database("benchmark")
@db = connection.db('benchmark')
def benchmark_insert(desc, coll_name, data)
benchmark(desc, PER_TRIAL, coll_name, data) do |coll, i|
def benchmark_insert(desc, coll_name, data, verbosity=true)
benchmark(desc, PER_TRIAL, coll_name, data, verbosity) do |coll, i|
data['x'] = i
coll.insert(data)
data.delete(:_id)
@ -109,10 +109,18 @@ end
print_headings
if RUBY_PLATFORM =~ /java/
puts "***WARMUP***"
benchmark_insert('insert (small, no index)', 'small_none', SMALL, false)
benchmark_insert('insert (medium, no index)', 'medium_none', MEDIUM, false)
benchmark_insert('insert (large, no index)', 'large_none', LARGE, false)
puts "***WARMUP***"
end
benchmark_insert('insert (small, no index)', 'small_none', SMALL)
benchmark_insert('insert (medium, no index)', 'medium_none', MEDIUM)
benchmark_insert('insert (large, no index)', 'large_none', LARGE)
benchmark_insert_index('insert (small, index)', 'small_indexed', SMALL)
benchmark_insert_index('insert (medium, index)', 'medium_indexed', MEDIUM)
benchmark_insert_index('insert (large, index)', 'large_indexed', LARGE)
@ -138,6 +146,11 @@ end
benchmark_find_one('find_one (small, no index)', 'small_none', PER_TRIAL / 2)
benchmark_find_one('find_one (medium, no index)', 'medium_none', PER_TRIAL / 2)
benchmark_find_one('find_one (large, no index)', 'large_none', PER_TRIAL / 2)
benchmark_find_one('find_one (small, no index)', 'small_none', PER_TRIAL / 2)
benchmark_find_one('find_one (medium, no index)', 'medium_none', PER_TRIAL / 2)
benchmark_find_one('find_one (large, no index)', 'large_none', PER_TRIAL / 2)
benchmark_find_one('find_one (small, indexed)', 'small_indexed', PER_TRIAL / 2)
benchmark_find_one('find_one (medium, indexed)', 'medium_indexed', PER_TRIAL / 2)

Binary file not shown.

View File

@ -60,6 +60,7 @@ public class RubyBSONCallback implements BSONCallback {
_nameStack.clear();
}
// Note: this actually creates an OrderedHash.
public RubyHash createHash() {
RubyHash h = (RubyHash)JavaEmbedUtils.invokeMethod(_runtime, _rbclsOrderedHash, "new",
new Object[] { }, Object.class);
@ -108,6 +109,8 @@ public class RubyBSONCallback implements BSONCallback {
_stack.addLast( (RubyObject)hash );
}
// Note: we use []= because we're dealing with an OrderedHash, which in 1.8
// doesn't have an internal JRuby representation.
public void writeRubyHash(String key, RubyHash hash, IRubyObject obj) {
RubyString rkey = _runtime.newString(key);
JavaEmbedUtils.invokeMethod(_runtime, hash, "[]=",
@ -115,9 +118,8 @@ public class RubyBSONCallback implements BSONCallback {
}
public void writeRubyArray(String key, RubyArray array, IRubyObject obj) {
Long rkey = Long.parseLong(key);
RubyFixnum index = new RubyFixnum(_runtime, rkey);
array.aset((IRubyObject)index, obj);
Long index = Long.parseLong(key);
array.store(index, obj);
}
public void arrayStart(String key){
@ -281,7 +283,7 @@ public class RubyBSONCallback implements BSONCallback {
RubyArray result = RubyArray.newArray( _runtime, b.length );
for ( int i=0; i<b.length; i++ ) {
result.aset( RubyNumeric.dbl2num( _runtime, (double)i ), RubyNumeric.dbl2num( _runtime, (double)b[i] ) );
result.store( i, _runtime.newFixnum(b[i]) );
}
return result;
@ -313,8 +315,7 @@ public class RubyBSONCallback implements BSONCallback {
if(current instanceof RubyArray) {
RubyArray a = (RubyArray)current;
Long n = Long.parseLong(name);
RubyFixnum index = new RubyFixnum(_runtime, n);
a.aset((IRubyObject)index, (IRubyObject)o);
a.store(n, (IRubyObject)o);
}
else {
RubyString rkey = RubyString.newString(_runtime, name);
@ -322,11 +323,11 @@ public class RubyBSONCallback implements BSONCallback {
new Object[] { (IRubyObject)rkey, o }, Object.class);
}
}
protected RubyObject cur(){
return _stack.getLast();
}
public Object get(){
return _root;
}
@ -339,22 +340,6 @@ public class RubyBSONCallback implements BSONCallback {
return _stack.size() < 1;
}
// Helper method for checking whether a Ruby hash has a certain key.
private boolean _rbHashHasKey(RubyHash hash, String key) {
RubyBoolean b = hash.has_key_p( _runtime.newString( key ) );
return b == _runtime.getTrue();
}
// Helper method for getting a value from a Ruby hash.
private IRubyObject _rbHashGet(RubyHash hash, Object key) {
if (key instanceof String) {
return hash.op_aref( _runtime.getCurrentContext(), _runtime.newString((String)key) );
}
else {
return hash.op_aref( _runtime.getCurrentContext(), (RubyObject)key );
}
}
static final HashMap<String, Object> _getRuntimeCache(Ruby runtime) {
// each JRuby runtime may have different objects for these constants,
// so cache them separately for each runtime

View File

@ -19,6 +19,7 @@ import org.bson.BSONEncoder;
import org.jruby.javasupport.JavaEmbedUtils;
import org.jruby.javasupport.JavaUtil;
import org.jruby.java.proxies.JavaProxy;
import org.jruby.*;
import org.jruby.runtime.builtin.IRubyObject;
@ -225,20 +226,8 @@ public class RubyBSONEncoder extends BSONEncoder {
if ( val instanceof String )
putString(name, val.toString() );
// TODO: Clean up
else if ( val instanceof Number ) {
if ( val instanceof Double ) {
putNumber(name, (Number)val);
}
else {
long jval = ((Number)val).longValue();
if (jval > Integer.MIN_VALUE && jval < Integer.MAX_VALUE) {
putNumber(name, (int)jval );
}
else
putNumber(name, (Number)jval );
}
}
else if ( val instanceof Number )
putNumber(name, (Number)val);
else if ( val instanceof Boolean )
putBoolean(name, (Boolean)val);
@ -265,7 +254,6 @@ public class RubyBSONEncoder extends BSONEncoder {
putSymbol(name, new Symbol(val.toString()));
}
// TODO: Clean up
else if ( val instanceof RubyFixnum ) {
long jval = ((RubyFixnum)val).getLongValue();
if (jval >= Integer.MIN_VALUE && jval <= Integer.MAX_VALUE) {
@ -275,12 +263,22 @@ public class RubyBSONEncoder extends BSONEncoder {
putNumber(name, (Number)jval );
}
// TODO: Clean up
else if ( val instanceof RubyFloat ) {
double jval = ((RubyFloat)val).getValue();
putNumber(name, (Number)jval );
}
else if ( val instanceof JavaProxy ) {
Object obj = ((JavaProxy)val).getObject();
if ( obj instanceof ArrayList ) {
putIterable( name, ((ArrayList)obj));
}
else {
_rbRaise( (RubyClass)_rbclsInvalidDocument,
"Got a JavaProxy object which can't be serialized as a BSON type." );
}
}
else if ( val instanceof RubyNil )
putNull(name);
@ -316,6 +314,9 @@ public class RubyBSONEncoder extends BSONEncoder {
else if( klass.equals( "BSON::ObjectID" ) ) {
putRubyObjectId(name, (RubyObject)val );
}
else if( klass.equals( "Java::JavaUtil::ArrayList" ) ) {
putIterable(name, (Iterable)val );
}
else if ( klass.equals( "BSON::Code" ) ) {
putRubyCodeWScope(name, (RubyObject)val );
}

Binary file not shown.

View File

@ -0,0 +1,379 @@
// BSON Callback
// RubyBSONCallback.java
package org.jbson;
import org.jruby.*;
import org.jruby.util.ByteList;
import org.jruby.RubyString;
import org.jruby.runtime.builtin.IRubyObject;
import org.jruby.runtime.Block;
import org.jruby.runtime.CallType;
import org.jruby.runtime.callsite.CacheEntry;
import org.jruby.javasupport.JavaEmbedUtils;
import org.jruby.javasupport.JavaUtil;
import org.jruby.parser.ReOptions;
import org.jruby.RubyArray;
import java.io.*;
import java.util.*;
import java.util.regex.*;
import org.bson.*;
import org.bson.types.*;
public class RubyBSONJavaCallback implements BSONCallback {
private RubyHash _root;
private RubyModule _rbclsOrderedHash;
private RubyModule _rbclsObjectId;
private RubyModule _rbclsBinary;
private RubyModule _rbclsMinKey;
private RubyModule _rbclsMaxKey;
private RubyModule _rbclsDBRef;
private RubyModule _rbclsCode;
private final LinkedList<Object> _stack = new LinkedList<Object>();
private final LinkedList<String> _nameStack = new LinkedList<String>();
private Ruby _runtime;
static final HashMap<Ruby, HashMap> _runtimeCache = new HashMap<Ruby, HashMap>();
public RubyBSONJavaCallback(Ruby runtime) {
_runtime = runtime;
_rbclsOrderedHash = _lookupConstant( _runtime, "BSON::OrderedHash" );
_rbclsBinary = _lookupConstant( _runtime, "BSON::Binary" );
_rbclsDBRef = _lookupConstant( _runtime, "BSON::DBRef" );
_rbclsCode = _lookupConstant( _runtime, "BSON::Code" );
_rbclsMinKey = _lookupConstant( _runtime, "BSON::MinKey" );
_rbclsMaxKey = _lookupConstant( _runtime, "BSON::MaxKey" );
_rbclsObjectId = _lookupConstant( _runtime, "BSON::ObjectId");
}
public BSONCallback createBSONCallback(){
return new RubyBSONCallback(_runtime);
}
public void reset(){
_root = null;
_stack.clear();
_nameStack.clear();
}
public RubyHash createHash() {
RubyHash h = (RubyHash)JavaEmbedUtils.invokeMethod(_runtime, _rbclsOrderedHash, "new",
new Object[] { }, Object.class);
return h;
}
public Object create( boolean array , List<String> path ){
if ( array )
return new ArrayList<Object>();
return createHash();
}
public void objectStart(){
if ( _stack.size() > 0 ) {
throw new IllegalStateException( "something is wrong" );
}
_root = createHash();
_stack.add(_root);
}
public void objectStart(boolean f) {
objectStart();
}
public void objectStart(String key){
RubyHash hash = createHash();
_nameStack.addLast( key );
Object lastObject = _stack.getLast();
// Yes, this is a bit hacky.
if(lastObject instanceof RubyHash) {
writeRubyHash(key, (RubyHash)lastObject, (IRubyObject)hash);
}
else {
((ArrayList)lastObject).add(Integer.parseInt(key), hash);
//writeRubyArray(key, (RubyArray)lastObject, (IRubyObject)hash);
}
_stack.addLast( (RubyObject)hash );
}
public void writeRubyHash(String key, RubyHash hash, Object obj) {
RubyString rkey = _runtime.newString(key);
JavaEmbedUtils.invokeMethod(_runtime, hash, "[]=",
new Object[] { (IRubyObject)rkey, obj }, Object.class);
}
public void writeRubyArray(String key, RubyArray array, IRubyObject obj) {
Long rkey = Long.parseLong(key);
RubyFixnum index = new RubyFixnum(_runtime, rkey);
array.aset((IRubyObject)index, obj);
}
public void arrayStart(String key){
ArrayList<Object> array = new ArrayList<Object>();
Object lastObject = _stack.getLast();
_nameStack.addLast( key );
if(lastObject instanceof RubyHash) {
writeRubyHash(key, (RubyHash)lastObject, array);
}
else {
((ArrayList)lastObject).add(Integer.parseInt(key), array);
}
_stack.addLast( array );
}
public Object objectDone(){
Object o =_stack.removeLast();
if ( _nameStack.size() > 0 )
_nameStack.removeLast();
else if ( _stack.size() > 0 ) {
throw new IllegalStateException( "something is wrong" );
}
return o;
}
// Not used by Ruby decoder
public void arrayStart(){
}
public Object arrayDone(){
return objectDone();
}
public void gotNull( String name ){
_put(name, null);
}
// Undefined should be represented as a lack of key / value.
public void gotUndefined( String name ){
}
// TODO: Handle this
public void gotUUID( String name , long part1, long part2) {
//_put( name , new UUID(part1, part2) );
}
public void gotCode( String name , String code ){
Object rb_code_obj = JavaEmbedUtils.invokeMethod(_runtime, _rbclsCode,
"new", new Object[] { code }, Object.class);
_put( name , (RubyObject)rb_code_obj );
}
public void gotCodeWScope( String name , String code , Object scope ){
Object rb_code_obj = JavaEmbedUtils.invokeMethod(_runtime, _rbclsCode,
"new", new Object[] { code, (RubyHash)scope }, Object.class);
_put( name , (RubyObject)rb_code_obj );
}
public void gotMinKey( String name ){
Object minkey = JavaEmbedUtils.invokeMethod(_runtime, _rbclsMinKey, "new", new Object[] {}, Object.class);
_put( name, (RubyObject)minkey);
}
public void gotMaxKey( String name ){
Object maxkey = JavaEmbedUtils.invokeMethod(_runtime, _rbclsMaxKey, "new", new Object[] {}, Object.class);
_put( name, (RubyObject)maxkey);
}
public void gotBoolean( String name , boolean v ){
_put(name , v);
}
public void gotDouble( String name , double v ){
_put(name , v);
}
public void gotInt( String name , int v ){
_put(name , v);
}
public void gotLong( String name , long v ){
_put(name , v);
}
public void gotDate( String name , long millis ){
RubyTime time = RubyTime.newTime(_runtime, millis).gmtime();
_put( name , time );
}
public void gotRegex( String name , String pattern , String flags ){
int f = 0;
ByteList b = new ByteList(pattern.getBytes());
if(flags.contains("i")) {
f = f | ReOptions.RE_OPTION_IGNORECASE;
}
if(flags.contains("m")) {
f = f | ReOptions.RE_OPTION_MULTILINE;
}
if(flags.contains("x")) {
f = f | ReOptions.RE_OPTION_EXTENDED;
}
_put( name , RubyRegexp.newRegexp(_runtime, b, f) );
}
public void gotString( String name , String v ){
_put( name , v );
}
public void gotSymbol( String name , String v ){
ByteList bytes = new ByteList(v.getBytes());
RubySymbol symbol = _runtime.getSymbolTable().getSymbol(bytes);
_put( name , symbol );
}
// Timestamp is currently rendered in Ruby as a two-element array.
public void gotTimestamp( String name , int time , int inc ){
RubyFixnum rtime = RubyFixnum.newFixnum( _runtime, time );
RubyFixnum rinc = RubyFixnum.newFixnum( _runtime, inc );
RubyObject[] args = new RubyObject[2];
args[0] = rinc;
args[1] = rtime;
RubyArray result = RubyArray.newArray( _runtime, args );
_put ( name , result );
}
public void gotObjectId( String name , ObjectId id ){
IRubyObject arg = (IRubyObject)RubyString.newString(_runtime, id.toString());
// //System.out.println(id.toByteArray().length);
// byte[] b = id.toByteArray();
// RubyArray a = _runtime.newArray();
// for(int i=0; i < b.length; i++) {
// System.out.println(b[i]);
// a.append(_runtime.newFixnum(b[i]));
// }
Object[] args = new Object[] { arg };
Object result = JavaEmbedUtils.invokeMethod(_runtime, _rbclsObjectId, "from_string", args, Object.class);
_put( name, (RubyObject)result );
}
// TODO: Incredibly annoying to deserialize to a Ruby DBRef. Might just
// stop supporting this altogether in the driver.
public void gotDBRef( String name , String ns , ObjectId id ){
// _put( name , new BasicBSONObject( "$ns" , ns ).append( "$id" , id ) );
}
// TODO: I know that this is horrible. To be optimized.
private RubyArray ja2ra( byte[] b ) {
RubyArray result = RubyArray.newArray( _runtime, b.length );
for ( int i=0; i<b.length; i++ ) {
result.aset( RubyNumeric.dbl2num( _runtime, (double)i ), RubyNumeric.dbl2num( _runtime, (double)b[i] ) );
}
return result;
}
public void gotBinaryArray( String name , byte[] b ) {
RubyArray a = ja2ra( b );
Object[] args = new Object[] { a, 2 };
Object result = JavaEmbedUtils.invokeMethod(_runtime, _rbclsBinary, "new", args, Object.class);
_put( name, (RubyObject)result );
}
// TODO: fix abs stuff here. some kind of bad type issue
public void gotBinary( String name , byte type , byte[] data ){
RubyArray a = ja2ra( data );
Object[] args = new Object[] { a, RubyFixnum.newFixnum(_runtime, Math.abs( type )) };
Object result = JavaEmbedUtils.invokeMethod(_runtime, _rbclsBinary, "new", args, Object.class);
_put( name, (RubyObject)result );
}
protected void _put( String name , Object o ){
Object current = cur();
if(current instanceof ArrayList) {
((ArrayList)current).add(Integer.parseInt(name), (Object)o);
//RubyArray a = (RubyArray)current;
//Long n = Long.parseLong(name);
//RubyFixnum index = new RubyFixnum(_runtime, n);
//a.aset((IRubyObject)index, o);
}
else {
RubyString rkey = RubyString.newString(_runtime, name);
JavaEmbedUtils.invokeMethod(_runtime, current, "[]=",
new Object[] { (IRubyObject)rkey, o }, Object.class);
}
}
protected Object cur(){
return _stack.getLast();
}
public Object get(){
return _root;
}
protected void setRoot(RubyHash o) {
_root = o;
}
protected boolean isStackEmpty() {
return _stack.size() < 1;
}
// Helper method for checking whether a Ruby hash has a certain key.
private boolean _rbHashHasKey(RubyHash hash, String key) {
RubyBoolean b = hash.has_key_p( _runtime.newString( key ) );
return b == _runtime.getTrue();
}
// Helper method for getting a value from a Ruby hash.
private IRubyObject _rbHashGet(RubyHash hash, Object key) {
if (key instanceof String) {
return hash.op_aref( _runtime.getCurrentContext(), _runtime.newString((String)key) );
}
else {
return hash.op_aref( _runtime.getCurrentContext(), (RubyObject)key );
}
}
static final HashMap<String, Object> _getRuntimeCache(Ruby runtime) {
// each JRuby runtime may have different objects for these constants,
// so cache them separately for each runtime
@SuppressWarnings("unchecked") // aargh! Java!
HashMap<String, Object> cache = _runtimeCache.get( runtime );
if(cache == null) {
cache = new HashMap<String, Object>();
_runtimeCache.put( runtime, cache );
}
return cache;
}
static final RubyModule _lookupConstant(Ruby runtime, String name)
{
HashMap<String, Object> cache = _getRuntimeCache( runtime );
RubyModule module = (RubyModule) cache.get( name );
if(module == null && !cache.containsKey( name )) {
module = runtime.getClassFromPath( name );
cache.put( (String)name, (Object)module );
}
return module;
}
}

View File

@ -4,7 +4,7 @@ module BSON
def self.serialize(obj, check_keys=false, move_id=false)
raise InvalidDocument, "BSON_JAVA.serialize takes a Hash" unless obj.is_a?(Hash)
enc = get_encoder# Java::OrgJbson::RubyBSONEncoder.new(JRuby.runtime)
enc = Java::OrgJbson::RubyBSONEncoder.new(JRuby.runtime)
ByteBuffer.new(enc.encode(obj))
end
@ -17,11 +17,8 @@ module BSON
end
def self.deserialize(buf)
if buf.is_a? String
buf = ByteBuffer.new(buf) if buf
end
dec = get_decoder
callback = Java::OrgJbson::RubyBSONCallback.new(JRuby.runtime)
dec = Java::OrgBson::BSONDecoder.new
callback = Java::OrgJbson::RubyBSONJavaCallback.new(JRuby.runtime)
dec.decode(buf.to_s.to_java_bytes, callback)
callback.get
end

View File

@ -44,6 +44,8 @@ module BSON
@data = data || generate
end
attr_accessor :data
# Determine if the supplied string is legal. Legal strings will
# consist of 24 hexadecimal characters.
#
@ -132,6 +134,13 @@ module BSON
str
end
def to_e
@data.each do |i|
print i
print ' '
end
end
def inspect
"BSON::ObjectId('#{to_s}')"
end

View File

@ -23,7 +23,7 @@ class BSONTest < Test::Unit::TestCase
def setup
@encoder = BSON::BSON_JAVA
@decoder = BSON::BSON_RUBY#BSON::BSON_JAVA
@decoder = BSON::BSON_JAVA
end
def assert_doc_pass(doc, options={})
@ -338,13 +338,16 @@ class BSONTest < Test::Unit::TestCase
assert_doc_pass(doc)
end
def test_timestamp
val = {"test" => [4, 20]}
assert_equal val, @decoder.deserialize([0x13, 0x00, 0x00, 0x00,
0x11, 0x74, 0x65, 0x73,
0x74, 0x00, 0x04, 0x00,
0x00, 0x00, 0x14, 0x00,
0x00, 0x00, 0x00])
if !(RUBY_PLATFORM =~ /java/)
def test_timestamp
val = {"test" => [4, 20]}
assert_equal val, @decoder.deserialize([0x13, 0x00, 0x00, 0x00,
0x11, 0x74, 0x65, 0x73,
0x74, 0x00, 0x04, 0x00,
0x00, 0x00, 0x14, 0x00,
0x00, 0x00, 0x00])
end
end
def test_overflow