Package nom.tam.util

Class ByteFormatter

java.lang.Object
nom.tam.util.ByteFormatter

public final class ByteFormatter extends Object
This class provides mechanisms for efficiently formatting numbers and Strings. Data is appended to existing byte arrays. Note that the formatting of real or double values may differ slightly (in the last bit) from the standard Java packages since this routines are optimized for speed rather than accuracy.

The methods in this class create no objects.

If a number cannot fit into the requested space the truncateOnOverlow flag controls whether the formatter will attempt to append it using the available length in the output (a la C or Perl style formats). If this flag is set, or if the number cannot fit into space left in the buffer it is 'truncated' and the requested space is filled with a truncation fill character. A TruncationException may be thrown if the truncationThrow flag is set.

This class does not explicitly support separate methods for formatting reals in exponential notation. Real numbers near one are by default formatted in decimal notation while numbers with large (or very negative) exponents are formatted in exponential notation. By setting the limits at which these transitions take place the user can force either exponential or decimal notation.

  • Field Summary

    Fields
    Modifier and Type
    Field
    Description
    private static final double
    Maximum magnitude to print in non-scientific notation.
    private static final double
    Minimum magnitude to print in non-scientific notation.
    private static final byte[]
    Digits.
    private static final long
     
    private static final int
     
    private static final long
    The hidden bit in normalized double numbers.
    private static final int
    Used to check if double is normalized
    private static final int
     
    private static final int
     
    private static final long
     
    private static final int
     
    private static final int
     
    private static final int
    The hidden bit in normalized floating point numbers.
    private static final int
    Used to check if float is normalized
    private static final int
     
    private static final int
     
    private static final int
     
    private static final double
     
    static final String
     
    private static final long
     
    private static final int
     
    private static final int
    The maximum single digit integer.
    private static final int
    The maximum two digit integer.
    static final String
     
    static final String
     
    private static final int
    The underlying number base used in this class.
    private static final double[]
    Powers of 10.
    private final byte[]
    Internal buffers used in formatting fields
    private final byte[]
     
    private static final int
     
    private static final byte
    What do we use to fill when we cannot print the number?Default is often used in Fortran.
    private static final int
    What index of tenpow is 10^0
  • Constructor Summary

    Constructors
    Constructor
    Description
     
  • Method Summary

    Modifier and Type
    Method
    Description
    private int
    combineReal(double val, byte[] buf, int off, int len, byte[] mant, int lmant, int shift)
    This method formats a double given a decimal mantissa and exponent information.
    int
    format(boolean val, byte[] array)
    Format a boolean into an existing array.
    int
    format(boolean val, byte[] array, int off, int len)
    Format a boolean into an existing array
    int
    format(double val, byte[] array)
    Format a double into an array.
    int
    format(double val, byte[] buf, int off, int len)
    Format a double into an existing character array.
    int
    format(float val, byte[] array)
    Format a float into an array.
    int
    format(float val, byte[] buf, int off, int len)
    Format a float into an existing byteacter array.
    int
    format(int val, byte[] array)
    Format an int into an array.
    int
    format(int val, byte[] buf, int off, int len)
    Format an int into an existing array.
    int
    format(long val, byte[] array)
    Format a long into an array.
    int
    format(long val, byte[] buf, int off, int len)
    Format a long into an existing array.
    int
    format(String val, byte[] array)
    Insert a string at the beginning of an array.
    int
    format(String val, byte[] array, int off, int len)
    Insert a String into an existing character array.
    private int
    mantissa(byte[] mant, int lmant, int exp, boolean simple, byte[] buf, int off, int len)
    Write the mantissa of the number.
    private void
    truncationFiller(byte[] buffer, int offset, int length)
    Fill the buffer with truncation characters.

    Methods inherited from class java.lang.Object

    clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
  • Field Details

    • NOT_A_NUMBER

      public static final String NOT_A_NUMBER
      See Also:
    • INFINITY

      public static final String INFINITY
      See Also:
    • NEGATIVE_INFINITY

      public static final String NEGATIVE_INFINITY
      See Also:
    • DEFAULT_SIMPLE_MAX

      private static final double DEFAULT_SIMPLE_MAX
      Maximum magnitude to print in non-scientific notation.
      See Also:
    • DEFAULT_SIMPLE_MIN

      private static final double DEFAULT_SIMPLE_MIN
      Minimum magnitude to print in non-scientific notation.
      See Also:
    • DIGITS

      private static final byte[] DIGITS
      Digits. We could handle other bases by extending or truncating this list and changing the division by 10 (and it's factors) at various locations.
    • DOUBLE_EXPONENT_BIT_MASK

      private static final long DOUBLE_EXPONENT_BIT_MASK
      See Also:
    • DOUBLE_EXPONENT_EXCESS

      private static final int DOUBLE_EXPONENT_EXCESS
      See Also:
    • DOUBLE_EXPONENT_NORMALIZE_BIT

      private static final long DOUBLE_EXPONENT_NORMALIZE_BIT
      The hidden bit in normalized double numbers.
      See Also:
    • DOUBLE_MIN_EXPONENT

      private static final int DOUBLE_MIN_EXPONENT
      Used to check if double is normalized
      See Also:
    • DOUBLE_SHIFT_BASE

      private static final int DOUBLE_SHIFT_BASE
      See Also:
    • DOUBLE_SHIFT_LIMIT

      private static final int DOUBLE_SHIFT_LIMIT
      See Also:
    • DOUBLE_VALUE_BIT_MASK

      private static final long DOUBLE_VALUE_BIT_MASK
      See Also:
    • FLOAT_EXPONENT_BIT_MASK

      private static final int FLOAT_EXPONENT_BIT_MASK
      See Also:
    • FLOAT_EXPONENT_EXCESS

      private static final int FLOAT_EXPONENT_EXCESS
      See Also:
    • FLOAT_EXPONENT_NORMALIZE_BIT

      private static final int FLOAT_EXPONENT_NORMALIZE_BIT
      The hidden bit in normalized floating point numbers.
      See Also:
    • FLOAT_MIN_EXPONENT

      private static final int FLOAT_MIN_EXPONENT
      Used to check if float is normalized
      See Also:
    • FLOAT_SHIFT_BASE

      private static final int FLOAT_SHIFT_BASE
      See Also:
    • FLOAT_SHIFT_LIMIT

      private static final int FLOAT_SHIFT_LIMIT
      See Also:
    • FLOAT_VALUE_BIT_MASK

      private static final int FLOAT_VALUE_BIT_MASK
      See Also:
    • I_LOG_10

      private static final double I_LOG_10
    • LONG_TO_INT_MODULO

      private static final long LONG_TO_INT_MODULO
      See Also:
    • MAX_LONG_LENGTH

      private static final int MAX_LONG_LENGTH
      See Also:
    • MAXIMUM_SINGLE_DIGIT_INTEGER

      private static final int MAXIMUM_SINGLE_DIGIT_INTEGER
      The maximum single digit integer. Special case handling when rounding up and when incrementing the exponent.
      See Also:
    • MAXIMUM_TWO_DIGIT_INTEGER

      private static final int MAXIMUM_TWO_DIGIT_INTEGER
      The maximum two digit integer. When we increment an exponent of this value, the exponent gets longer. Don't need to worry about 999 since exponents can't be that big.
      See Also:
    • TEMP_BUFFER_SIZE

      private static final int TEMP_BUFFER_SIZE
      See Also:
    • NUMBER_BASE

      private static final int NUMBER_BASE
      The underlying number base used in this class.
      See Also:
    • NUMBER_BASE_POWERS

      private static final double[] NUMBER_BASE_POWERS
      Powers of 10. We over extend on both sides. These should perhaps be tabulated rather than computed though it may be faster to calculate them than to read in the extra bytes in the class file.
    • TRUNCATION_FILL

      private static final byte TRUNCATION_FILL
      What do we use to fill when we cannot print the number?Default is often used in Fortran.
      See Also:
    • ZERO_POW

      private static final int ZERO_POW
      What index of tenpow is 10^0
    • tbuf1

      private final byte[] tbuf1
      Internal buffers used in formatting fields
    • tbuf2

      private final byte[] tbuf2
  • Constructor Details

    • ByteFormatter

      public ByteFormatter()
  • Method Details

    • combineReal

      private int combineReal(double val, byte[] buf, int off, int len, byte[] mant, int lmant, int shift)
      This method formats a double given a decimal mantissa and exponent information.
      Parameters:
      val - The original number
      buf - Output buffer
      off - Offset into buffer
      len - Maximum number of characters to use in buffer.
      mant - A decimal mantissa for the number.
      lmant - The number of characters in the mantissa
      shift - The exponent of the power of 10 that we shifted val to get the given mantissa.
      Returns:
      Offset of next available character in buffer. @ * if the value was truncated
    • format

      public int format(boolean val, byte[] array)
      Format a boolean into an existing array.
      Parameters:
      val - value to write
      array - the array to fill
      Returns:
      Offset of next available character in buffer.
    • format

      public int format(boolean val, byte[] array, int off, int len)
      Format a boolean into an existing array
      Parameters:
      val - The boolean to be formatted
      array - The buffer in which to format the data.
      off - The starting offset within the buffer.
      len - The maximum number of characters to use use in formatting the number.
      Returns:
      Offset of next available character in buffer.
    • format

      public int format(double val, byte[] array)
      Format a double into an array.
      Parameters:
      val - The double to be formatted.
      array - The array in which to place the result.
      Returns:
      The number of characters used. @ * if the value was truncated
    • format

      public int format(double val, byte[] buf, int off, int len)
      Format a double into an existing character array.

      This is hard to do exactly right... The JDK code does stuff with rational arithmetic and so forth. We use a much simpler algorithm which may give an answer off in the lowest order bit. Since this is pure Java, it should still be consistent from machine to machine.

      Recall that the binary representation of the double is of the form d = 0.bbbbbbbb x 2n where there are up to 53 binary digits in the binary fraction (including the assumed leading 1 bit for normalized numbers). We find a value m such that 10m d is between 253 and 263. This product will be exactly convertible to a long with no loss of precision. Getting the decimal representation for that is trivial (see formatLong). This is a decimal mantissa and we have an exponent (-m). All we have to do is manipulate the decimal point to where we want to see it. Errors can arise due to roundoff in the scaling multiplication, but should be no more than a single bit.

      Parameters:
      val - Double to be formatted
      buf - Buffer in which result is to be stored
      off - Offset within buffer
      len - Maximum length of integer
      Returns:
      offset of next unused character in input buffer. @ * if the value was truncated
    • format

      public int format(float val, byte[] array)
      Format a float into an array.
      Parameters:
      val - The float to be formatted.
      array - The array in which to place the result.
      Returns:
      The number of characters used. @ * if the value was truncated
    • format

      public int format(float val, byte[] buf, int off, int len)
      Format a float into an existing byteacter array.

      This is hard to do exactly right... The JDK code does stuff with rational arithmetic and so forth. We use a much simpler algorithm which may give an answer off in the lowest order bit. Since this is pure Java, it should still be consistent from machine to machine.

      Recall that the binary representation of the float is of the form d = 0.bbbbbbbb x 2n where there are up to 24 binary digits in the binary fraction (including the assumed leading 1 bit for normalized numbers). We find a value m such that 10m d is between 224 and 232. This product will be exactly convertible to an int with no loss of precision. Getting the decimal representation for that is trivial (see formatInteger). This is a decimal mantissa and we have an exponent ( -m). All we have to do is manipulate the decimal point to where we want to see it. Errors can arise due to roundoff in the scaling multiplication, but should be very small.

      Parameters:
      val - Float to be formatted
      buf - Buffer in which result is to be stored
      off - Offset within buffer
      len - Maximum length of field
      Returns:
      Offset of next character in buffer. @ * if the value was truncated
    • format

      public int format(int val, byte[] array)
      Format an int into an array.
      Parameters:
      val - The int to be formatted.
      array - The array in which to place the result.
      Returns:
      The number of characters used. @ * if the value was truncated
    • format

      public int format(int val, byte[] buf, int off, int len)
      Format an int into an existing array.
      Parameters:
      val - Integer to be formatted
      buf - Buffer in which result is to be stored
      off - Offset within buffer
      len - Maximum length of integer
      Returns:
      offset of next unused character in input buffer.
    • format

      public int format(long val, byte[] array)
      Format a long into an array.
      Parameters:
      val - The long to be formatted.
      array - The array in which to place the result.
      Returns:
      The number of characters used. @ * if the value was truncated
    • format

      public int format(long val, byte[] buf, int off, int len)
      Format a long into an existing array.
      Parameters:
      val - Long to be formatted
      buf - Buffer in which result is to be stored
      off - Offset within buffer
      len - Maximum length of integer
      Returns:
      offset of next unused character in input buffer. @ * if the value was truncated
    • format

      public int format(String val, byte[] array)
      Insert a string at the beginning of an array. * @return Offset of next available character in buffer.
      Parameters:
      val - The string to be inserted. A null string will insert len spaces.
      array - The buffer in which to insert the string.
      Returns:
      Offset of next available character in buffer.
    • format

      public int format(String val, byte[] array, int off, int len)
      Insert a String into an existing character array. If the String is longer than len, then only the the initial len characters will be inserted.
      Parameters:
      val - The string to be inserted. A null string will insert len spaces.
      array - The buffer in which to insert the string.
      off - The starting offset to insert the string.
      len - The maximum number of characters to insert.
      Returns:
      Offset of next available character in buffer.
    • mantissa

      private int mantissa(byte[] mant, int lmant, int exp, boolean simple, byte[] buf, int off, int len)
      Write the mantissa of the number. This method addresses the subtleties involved in rounding numbers.
    • truncationFiller

      private void truncationFiller(byte[] buffer, int offset, int length)
      Fill the buffer with truncation characters. After filling the buffer, a TruncationException will be thrown if the appropriate flag is set.