org.jredis.ri.alphazero.support

Class Convert

java.lang.Object

org.jredis.ri.alphazero.support.Convert

public class Convert
extends java.lang.Object

Perhaps a silly hack, but proven to speed things up. [TODO: document me!]

Since:

alpha.0

Version:

alpha.0, Apr 02, 2009

Field Summary

Fields

Modifier and Type	Field and Description
private static byte	BYTE_MINUS
private static byte	BYTE_NINE
private static byte	BYTE_PLUS
private static byte	BYTE_ZERO
private static byte[][]	i2b_65535
static int	INT_N_65535
static int	INT_P_65535
private static int	MAX_POSITIVE_32_BIT_DIGITS
private static int	MAX_POSITIVE_64_BIT_DIGITS

Constructor Summary

Constructors

Constructor and Description
Convert()

Method Summary

Modifier and Type	Method and Description
private static byte[]	getNegativeNumberBytes(byte[] unsigned)
static byte[]	toBytes(double dnum)
static byte[]	toBytes(int i) return the bytes of the string representation of the integer.
static byte[]	toBytes(long lnum) Will delegate to Convert#getBytes(int) if the 'long' number is actually within the range of our int cache, otherwise it will return the bytes using std JDK mechanisms.
static double	toDouble(byte[] stringRepOfDoublePrecisionBytes) TODO: optimize.
static int	toInt(byte[] potentiallySignedBytes)
static int	toInt(byte[] potentiallySignedAsciiBytes, int offsetin, int len) Converts the byte[]s of the ASCII representation of a decimal number to an int.
static long	toLong(byte[] potentiallySignedBytes)
static long	toLong(byte[] potentiallySignedAsciiBytes, int offsetin, int len) Its just like (really! :) toInt(byte[], int, int) but for Long values.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

INT_P_65535

public static final int INT_P_65535

See Also:

Constant Field Values

INT_N_65535

public static final int INT_N_65535

See Also:

Constant Field Values

i2b_65535

private static final byte[][] i2b_65535

BYTE_MINUS

private static final byte BYTE_MINUS

See Also:

Constant Field Values

BYTE_PLUS

private static final byte BYTE_PLUS

See Also:

Constant Field Values

BYTE_ZERO

private static final byte BYTE_ZERO

See Also:

Constant Field Values

BYTE_NINE

private static final byte BYTE_NINE

See Also:

Constant Field Values

MAX_POSITIVE_32_BIT_DIGITS

private static final int MAX_POSITIVE_32_BIT_DIGITS

See Also:

Constant Field Values

MAX_POSITIVE_64_BIT_DIGITS

private static final int MAX_POSITIVE_64_BIT_DIGITS

See Also:

Constant Field Values

Constructor Detail

Convert

public Convert()

Method Detail

toBytes

public static final byte[] toBytes(int i)

return the bytes of the string representation of the integer. Perhaps should be called getNumberBytes, or perhaps getHumanReadableBytes() Ex: 444 => "444".getBytes() => new byte[3]={52, 52, 52} If representation is not in our cache or too high (larger than Convert#INT_P_65535 ) then it will return whatever we would get from toBytes(int) which is using JDK libs for the same. The whole point here is to be faster since we need to convert numbers to the byte array of their string representation a lot in JRedis for the protocol and to cut out the unnecessary step of creating a new string simply because there apparently isn't any other way in JDK to go from a number to the bytes of its textual representation. (!)

Parameters:

i -

signed -

Returns:

toBytes

public static final byte[] toBytes(long lnum)

Will delegate to Convert#getBytes(int) if the 'long' number is actually within the range of our int cache, otherwise it will return the bytes using std JDK mechanisms.

Parameters:

lnum -

Returns:

toBytes

public static final byte[] toBytes(double dnum)

toInt

public static final int toInt(byte[] potentiallySignedAsciiBytes,
                              int offsetin,
                              int len)
                       throws java.lang.IllegalArgumentException

Converts the byte[]s of the ASCII representation of a decimal number to an int.

Expects a byte array of no more than MAX_POSITIVE_32_BIT_DIGITS bytes in length after accounting for potential leading byte indicating the sign of the number representation. 9 bytes (for a positive integer).

Parameters:

potentiallySignedAsciiBytes, - for example {49, 49, 52} ("114")

Returns:

Throws:

java.lang.IllegalArgumentException

toLong

public static final long toLong(byte[] potentiallySignedAsciiBytes,
                                int offsetin,
                                int len)
                         throws java.lang.IllegalArgumentException

Its just like (really! :) toInt(byte[], int, int) but for Long values. Max number of digits is now MAX_POSITIVE_64_BIT_DIGITS.

Parameters:

potentiallySignedAsciiBytes -

offset -

len -

Returns:

Throws:

java.lang.IllegalArgumentException

toInt

public static final int toInt(byte[] potentiallySignedBytes)
                       throws java.lang.IllegalArgumentException

Parameters:

potentiallySignedBytes -

Returns:

Throws:

java.lang.IllegalArgumentException

toLong

public static final long toLong(byte[] potentiallySignedBytes)
                         throws java.lang.IllegalArgumentException

Parameters:

potentiallySignedBytes -

Returns:

Throws:

java.lang.IllegalArgumentException

toDouble

public static final double toDouble(byte[] stringRepOfDoublePrecisionBytes)
                             throws java.lang.IllegalArgumentException

TODO: optimize.

Parameters:

potentiallySignedBytes -

Returns:

Throws:

java.lang.IllegalArgumentException

getNegativeNumberBytes

private static final byte[] getNegativeNumberBytes(byte[] unsigned)

Parameters:

unsigned -

Returns: