/**
* Serializes a <code>DBObject</code> into a string that can be sent to the database.
*
* <p>There is a pool of available encoders. Create a new one as follows:
*
* <blockquote>
*
* </pre>
*
* ByteEncoder encoder = ByteEncoder.get(); // try forever until an encoder is available </pre>
*
* </blockquote>
*
* <p>Each key/value pair in the <code>DBObject</code> is encoded in the following format:
*
* <blockquote>
*
* <i><type (</i><code>byte</code><i>)><name (</i><code>String</code></i>)><0 (</i>
* <code>byte</code><i>)><data (serialized data)></i>
*
* </blockquote>
*
* For example:
*
* <blockquote>
*
* <<code>NUMBER</code>><name>0<double> <code>// NUMBER = 1</code><br>
* <<code>STRING</code>><name>0<len><string>0 <code>// STRING = 2</code>
*
* </blockquote>
*/
public class ByteEncoder extends Bytes {
static final boolean DEBUG = Boolean.getBoolean("DEBUG.BE");
// things that won't get sent in the scope
static final Set<String> BAD_GLOBALS = new HashSet<String>();
static {
BAD_GLOBALS.add("db");
BAD_GLOBALS.add("local");
BAD_GLOBALS.add("core");
BAD_GLOBALS.add("args"); // TODO: should we get rid of this
BAD_GLOBALS.add("obj"); // TODO: get rid of this
}
/**
* Whether a given field is generated by the db for use by the db
*
* @param o an object to check
* @return whether the object is a db field
*/
public static boolean dbOnlyField(Object o) {
if (o == null) return false;
if (o instanceof String) return dbOnlyField(o.toString());
return false;
}
/**
* Whether a given field is generated by the db for use by the db
*
* @param s a string to check
* @return whether the string is a db field
*/
public static boolean dbOnlyField(String s) {
return s.equals("_ns") || s.equals("_save") || s.equals("_update");
}
/**
* Fetches a new <code>ByteEncoder</code> from the pool of available <code>ByteEncoder</code>s.
*
* @return a new <code>ByteEncoder</code>
*/
public static ByteEncoder get() {
return _pool.get();
}
/**
* Get the current position of this ByteEncoder
*
* @return
*/
public long getPosition() {
return _buf.position();
}
/**
* Get the remaining capacity of this ByteEncoder
*
* @return
*/
public long getRemaining() {
return _buf.remaining();
}
/** Resets and returns this encoder to the pool. */
protected void done() {
reset();
_pool.done(this);
}
static final SimplePool<ByteEncoder> _pool =
new SimplePool<ByteEncoder>("ByteEncoders", NUM_ENCODERS, NUM_ENCODERS * 2) {
protected ByteEncoder createNew() {
if (D) System.out.println("creating new ByteEncoder");
return new ByteEncoder();
}
protected long memSize(ByteEncoder d) {
return d._buf.capacity() + (2 * MAX_STRING) + 1024;
}
};
// ----
private ByteEncoder() {
_buf = ByteBuffer.allocateDirect(MAX_OBJECT_SIZE + 2048);
_buf.order(Bytes.ORDER);
}
/**
* Returns the bytes in the bytebuffer. Attempts to leave the bytebuffer in the same state. Note
* that mark, if set, is lost.
*
* @return array of bytes
*/
public byte[] getBytes() {
int pos = _buf.position();
int limit = _buf.limit();
flip();
byte[] arr = new byte[_buf.limit()];
_buf.get(arr);
flip();
_buf.position(pos);
_buf.limit(limit);
return arr;
}
/** Returns encoder to its starting state, ready to encode an object. */
protected void reset() {
_buf.position(0);
_buf.limit(_buf.capacity());
_flipped = false;
_dontRef.clear();
}
/** Switches the encoder from being write-only to being read-only. */
protected void flip() {
_buf.flip();
_flipped = true;
}
/**
* Encodes a <code>DBObject</code>. This is for the higher level api calls
*
* <p>If encoding an object fails, the buffer will be reset to the position prior to this put call
* and a BufferOverflowException will be thrown
*
* @param o the object to encode
* @return the number of characters in the encoding
*/
public int putObject(DBObject o) {
_buf.mark();
try {
return putObject(null, o);
} catch (BufferOverflowException bof) {
// reset to marked offset and wipe any written data
_buf.reset();
_buf.put(new byte[_buf.remaining()]);
_buf.reset();
throw new BufferOverflowException();
}
}
/** this is really for embedded objects */
private int putObject(String name, DBObject o) {
if (o == null) throw new NullPointerException("can't save a null object");
if (DEBUG)
System.out.println(
"putObject : " + name + " [" + o.getClass() + "]" + " # keys " + o.keySet().size());
if (_flipped) throw new IllegalStateException("already flipped");
final int start = _buf.position();
byte myType = OBJECT;
if (o instanceof List) myType = ARRAY;
if (_handleSpecialObjects(name, o)) return _buf.position() - start;
if (name != null) {
_put(myType, name);
}
final int sizePos = _buf.position();
_buf.putInt(0); // leaving space for this. set it at the end
List transientFields = null;
if (myType == OBJECT) {
if (o.containsField("_id")) _putObjectField("_id", o.get("_id"));
{
Object temp = o.get("_transientFields");
if (temp instanceof List) transientFields = (List) temp;
}
}
for (String s : o.keySet()) {
if (s.equals("_id")) continue;
if (transientFields != null && transientFields.contains(s)) continue;
Object val = o.get(s);
_putObjectField(s, val);
}
_buf.put(EOO);
_buf.putInt(sizePos, _buf.position() - sizePos);
return _buf.position() - start;
}
private void _putObjectField(String name, Object val) {
if (dbOnlyField(name) || name.equals("_transientFields")) return;
if (DEBUG) System.out.println("\t put thing : " + name);
if (name.equals("$where") && val instanceof String) {
_put(CODE, name);
_putValueString(val.toString());
return;
}
val = Bytes.applyEncodingHooks(val);
if (val == null) putNull(name);
else if (val instanceof Date) putDate(name, (Date) val);
else if (val instanceof Number) putNumber(name, (Number) val);
else if (val instanceof String) putString(name, val.toString());
else if (val instanceof ObjectId) putObjectId(name, (ObjectId) val);
else if (val instanceof DBObject) putObject(name, (DBObject) val);
else if (val instanceof Boolean) putBoolean(name, (Boolean) val);
else if (val instanceof Pattern) putPattern(name, (Pattern) val);
else if (val instanceof DBRegex) {
putDBRegex(name, (DBRegex) val);
} else if (val instanceof Map) putMap(name, (Map) val);
else if (val instanceof List) putList(name, (List) val);
else if (val instanceof byte[]) putBinary(name, (byte[]) val);
else if (val instanceof DBBinary) putBinary(name, (DBBinary) val);
else if (val.getClass().isArray()) putList(name, Arrays.asList((Object[]) val));
else if (val instanceof DBPointer) {
// temporary - there's the notion of "special object" , but for simple level 0...
DBPointer r = (DBPointer) val;
putDBPointer(name, r._ns, (ObjectId) r._id);
} else if (val instanceof DBRefBase) {
putDBRef(name, (DBRefBase) val);
} else if (val instanceof DBSymbol) {
putSymbol(name, (DBSymbol) val);
} else if (val instanceof DBUndefined) {
putUndefined(name);
} else if (val instanceof DBTimestamp) {
putTimestamp(name, (DBTimestamp) val);
} else throw new IllegalArgumentException("can't serialize " + val.getClass());
}
private void putList(String name, List l) {
_put(ARRAY, name);
final int sizePos = _buf.position();
_buf.putInt(0);
for (int i = 0; i < l.size(); i++) _putObjectField(String.valueOf(i), l.get(i));
_buf.put(EOO);
_buf.putInt(sizePos, _buf.position() - sizePos);
}
private void putMap(String name, Map m) {
_put(OBJECT, name);
final int sizePos = _buf.position();
_buf.putInt(0);
for (Object key : m.keySet()) _putObjectField(key.toString(), m.get(key));
_buf.put(EOO);
_buf.putInt(sizePos, _buf.position() - sizePos);
}
private boolean _handleSpecialObjects(String name, DBObject o) {
if (o == null) return false;
if (o instanceof DBCollection) {
DBCollection c = (DBCollection) o;
putDBPointer(name, c.getName(), Bytes.COLLECTION_REF_ID);
return true;
}
if (!_dontRefContains(o) && name != null && o instanceof DBPointer) {
DBPointer r = (DBPointer) o;
putDBPointer(name, r._ns, (ObjectId) r._id);
return true;
}
if (!(o instanceof List) && o.get(Bytes.NO_REF_HACK) != null) {
o.removeField(Bytes.NO_REF_HACK);
return false;
}
if (!_dontRefContains(o) && name != null && !(o instanceof List) && cameFromDB(o)) {
putDBPointer(name, o.get("_ns").toString(), (ObjectId) (o.get("_id")));
return true;
}
return false;
}
protected int putNull(String name) {
int start = _buf.position();
_put(NULL, name);
return _buf.position() - start;
}
protected int putUndefined(String name) {
int start = _buf.position();
_put(UNDEFINED, name);
return _buf.position() - start;
}
protected int putTimestamp(String name, DBTimestamp ts) {
int start = _buf.position();
_put(TIMESTAMP, name);
_buf.putInt(ts.getTime());
_buf.putInt(ts.getInc());
return _buf.position() - start;
}
protected int putBoolean(String name, Boolean b) {
int start = _buf.position();
_put(BOOLEAN, name);
_buf.put(b ? (byte) 0x1 : (byte) 0x0);
return _buf.position() - start;
}
protected int putDate(String name, Date d) {
int start = _buf.position();
_put(DATE, name);
_buf.putLong(d.getTime());
return _buf.position() - start;
}
protected int putNumber(String name, Number n) {
int start = _buf.position();
if (n instanceof Integer) {
_put(NUMBER_INT, name);
_buf.putInt(n.intValue());
} else if (n instanceof Long) {
_put(NUMBER_LONG, name);
_buf.putLong(n.longValue());
} else {
_put(NUMBER, name);
_buf.putDouble(n.doubleValue());
}
return _buf.position() - start;
}
protected void putBinary(String name, byte[] data) {
_put(BINARY, name);
_buf.putInt(4 + data.length);
_buf.put(B_BINARY);
_buf.putInt(data.length);
int before = _buf.position();
_buf.put(data);
int after = _buf.position();
com.mongodb.util.MyAsserts.assertEquals(after - before, data.length);
}
protected void putBinary(String name, DBBinary val) {
_put(BINARY, name);
_buf.putInt(val._data.length);
_buf.put(val._type);
_buf.put(val._data);
}
protected int putSymbol(String name, DBSymbol s) {
return _putString(name, s.getSymbol(), SYMBOL);
}
protected int putString(String name, String s) {
return _putString(name, s, STRING);
}
private int _putString(String name, String s, byte type) {
int start = _buf.position();
_put(type, name);
_putValueString(s);
return _buf.position() - start;
}
protected int putObjectId(String name, ObjectId oid) {
int start = _buf.position();
_put(OID, name);
_buf.putInt(oid._time);
_buf.putInt(oid._machine);
_buf.putInt(oid._inc);
return _buf.position() - start;
}
protected int putDBPointer(String name, String ns, ObjectId oid) {
int start = _buf.position();
_put(REF, name);
_putValueString(ns);
_buf.putInt(oid._time);
_buf.putInt(oid._machine);
_buf.putInt(oid._inc);
return _buf.position() - start;
}
protected void putDBRef(String name, DBRefBase ref) {
_put(OBJECT, name);
final int sizePos = _buf.position();
_buf.putInt(0);
_putObjectField("$ref", ref.getRef());
_putObjectField("$id", ref.getId());
_buf.put(EOO);
_buf.putInt(sizePos, _buf.position() - sizePos);
}
private int putDBRegex(String name, DBRegex regex) {
int start = _buf.position();
_put(REGEX, name);
_put(regex.getPattern());
String options = regex.getOptions();
TreeMap<Character, Character> sm = new TreeMap<Character, Character>();
for (int i = 0; i < options.length(); i++) {
sm.put(options.charAt(i), options.charAt(i));
}
StringBuffer sb = new StringBuffer();
for (char c : sm.keySet()) {
sb.append(c);
}
_put(sb.toString());
return _buf.position() - start;
}
private int putPattern(String name, Pattern p) {
int start = _buf.position();
_put(REGEX, name);
_put(p.pattern());
_put(patternFlags(p.flags()));
return _buf.position() - start;
}
// ----------------------------------------------
/** Encodes the type and key. */
private void _put(byte type, String name) {
_buf.put(type);
_put(name);
}
void _putValueString(String s) {
int lenPos = _buf.position();
_buf.putInt(0); // making space for size
int strLen = _put(s);
_buf.putInt(lenPos, strLen);
}
int _put(String name) {
_cbuf.position(0);
_cbuf.limit(_cbuf.capacity());
_cbuf.append(name);
_cbuf.flip();
final int start = _buf.position();
_encoder.encode(_cbuf, _buf, false);
_buf.put((byte) 0);
return _buf.position() - start;
}
boolean _dontRefContains(Object o) {
if (_dontRef.size() == 0) return false;
return _dontRef.peek().contains(o);
}
private final CharBuffer _cbuf = CharBuffer.allocate(MAX_STRING);
private final CharsetEncoder _encoder = _utf8.newEncoder();
private Stack<IdentitySet> _dontRef = new Stack<IdentitySet>();
private boolean _flipped = false;
final ByteBuffer _buf;
}
public static void main(String[] args) throws Throwable {
final int itrs = Integer.getInteger("iterations", 100000);
// final int itrs = Integer.getInteger("iterations", 12);
final int size = Integer.getInteger("size", 2048);
final int subsize = Integer.getInteger("subsize", 128);
final int maxchar = Integer.getInteger("maxchar", 128);
final String regex = System.getProperty("filter");
final Pattern filter = (regex == null) ? null : Pattern.compile(regex);
final boolean useSecurityManager = Boolean.getBoolean("SecurityManager");
if (useSecurityManager) System.setSecurityManager(new PermissiveSecurityManger());
final Random rnd = new Random();
String[] csns =
new String[] {
"Big5",
"Johab",
"EUC_CN",
"EUC_KR",
"MS932",
"MS936",
"MS949",
"MS950",
"GBK",
"Big5_HKSCS",
"Big5_HKSCS_2001",
"Big5_Solaris",
"MS950_HKSCS",
"MS950_HKSCS_XP",
"IBM1364",
"IBM1381",
"IBM1383",
"IBM930",
"IBM933",
"IBM935",
"IBM937",
"IBM939",
"IBM942",
"IBM943",
"IBM948",
"IBM949",
"IBM950",
"IBM970",
};
ArrayList<long[]> sum = new ArrayList<>();
for (final String csn : csns) {
final Charset cs = Charset.forName(csn);
List<Integer> cps = new ArrayList<>(0x4000);
int off = 0;
int cp = 0;
int n = 0;
CharsetEncoder enc = cs.newEncoder();
while (cp < 0x10000 && n < cps.size()) {
if (enc.canEncode((char) cp)) {
cps.add(cp);
n++;
}
cp++;
}
Collections.shuffle(cps);
char[] ca = new char[cps.size()];
for (int i = 0; i < cps.size(); i++) ca[i] = (char) (int) cps.get(i);
System.out.printf("%n--------%s---------%n", csn);
for (int sz = 8; sz <= 2048; sz *= 2) {
System.out.printf(" [len=%d]%n", sz);
final char[] chars = Arrays.copyOf(ca, sz);
final String str = new String(chars);
final byte[] bs = str.getBytes(cs);
Job[] jobs = {
new Job("String decode: csn") {
public void work() throws Throwable {
for (int i = 0; i < itrs; i++) new String(bs, csn);
}
},
new Job("String decode: cs") {
public void work() throws Throwable {
for (int i = 0; i < itrs; i++) new String(bs, cs);
}
},
new Job("String encode: csn") {
public void work() throws Throwable {
for (int i = 0; i < itrs; i++) str.getBytes(csn);
}
},
new Job("String encode: cs") {
public void work() throws Throwable {
for (int i = 0; i < itrs; i++) str.getBytes(cs);
}
},
};
sum.add(time(jobs));
}
}
}
/**
* Returns boolean value from system property or provided function.
*
* @param propName System property host.
* @param dflt Function that returns {@code Boolean}.
* @return {@code Boolean} value
*/
public static boolean boolValue(String propName, boolean dflt) {
String sysProp = getProperty(propName);
return (sysProp != null && !sysProp.isEmpty()) ? Boolean.getBoolean(sysProp) : dflt;
}