FileDocCategorySizeDatePackage
DigitList.javaAPI DocJava SE 5 API22042Fri Aug 26 14:57:20 BST 2005java.text

DigitList

public final class DigitList extends Object implements Cloneable
Digit List. Private to DecimalFormat. Handles the transcoding between numeric values and strings of characters. Only handles non-negative numbers. The division of labor between DigitList and DecimalFormat is that DigitList handles the radix 10 representation issues; DecimalFormat handles the locale-specific issues such as positive/negative, grouping, decimal point, currency, and so on. A DigitList is really a representation of a floating point value. It may be an integer value; we assume that a double has sufficient precision to represent all digits of a long. The DigitList representation consists of a string of characters, which are the digits radix 10, from '0' to '9'. It also has a radix 10 exponent associated with it. The value represented by a DigitList object can be computed by mulitplying the fraction f, where 0 <= f < 1, derived by placing all the digits of the list to the right of the decimal point, by 10^exponent.
see
Locale
see
Format
see
NumberFormat
see
DecimalFormat
see
ChoiceFormat
see
MessageFormat
version
1.30 12/19/03
author
Mark Davis, Alan Liu

Fields Summary
public static final int
MAX_COUNT
The maximum number of significant digits in an IEEE 754 double, that is, in a Java double. This must not be increased, or garbage digits will be generated, and should not be decreased, or accuracy will be lost.
public int
decimalAt
These data members are intentionally public and can be set directly. The value represented is given by placing the decimal point before digits[decimalAt]. If decimalAt is < 0, then leading zeros between the decimal point and the first nonzero digit are implied. If decimalAt is > count, then trailing zeros between the digits[count-1] and the decimal point are implied. Equivalently, the represented value is given by f * 10^decimalAt. Here f is a value 0.1 <= f < 1 arrived at by placing the digits in Digits to the right of the decimal. DigitList is normalized, so if it is non-zero, figits[0] is non-zero. We don't allow denormalized numbers because our exponent is effectively of unlimited magnitude. The count value contains the number of significant digits present in digits[]. Zero is represented by any DigitList with count == 0 or with each digits[i] for all i <= count == '0'.
public int
count
public char[]
digits
private char[]
data
private static final char[]
LONG_MIN_REP
private StringBuffer
tempBuffer
Constructors Summary
Methods Summary
public voidappend(char digit)
Appends a digit to the list, extending the list when necessary.

        if (count == digits.length) {
            char[] data = new char[count + 100];
            System.arraycopy(digits, 0, data, 0, count);
            digits = data;
        }
        digits[count++] = digit;
    
public voidclear()
Clears out the digits. Use before appending them. Typically, you set a series of digits with append, then at the point you hit the decimal point, you set myDigitList.decimalAt = myDigitList.count; then go on appending digits.

        decimalAt = 0;
        count = 0;
    
public java.lang.Objectclone()
Creates a copy of this object.

return
a clone of this instance.

        try {
            DigitList other = (DigitList) super.clone();
            char[] newDigits = new char[digits.length];
            System.arraycopy(digits, 0, newDigits, 0, digits.length);
            other.digits = newDigits;
            return other;
        } catch (CloneNotSupportedException e) {
            throw new InternalError();
        }
    
public booleanequals(java.lang.Object obj)
equality test between two digit lists.

        if (this == obj)                      // quick check
            return true;
        if (!(obj instanceof DigitList))         // (1) same object?
            return false;
        DigitList other = (DigitList) obj;
        if (count != other.count ||
        decimalAt != other.decimalAt)
            return false;
        for (int i = 0; i < count; i++)
            if (digits[i] != other.digits[i])
                return false;
        return true;
    
private voidextendDigits(int len)

        if (len > digits.length) {
            digits = new char[len];
        }
    
booleanfitsIntoLong(boolean isPositive, boolean ignoreNegativeZero)
Return true if the number represented by this object can fit into a long.

param
isPositive true if this number should be regarded as positive
param
ignoreNegativeZero true if -0 should be regarded as identical to +0; otherwise they are considered distinct
return
true if this number fits into a Java long

        // Figure out if the result will fit in a long.  We have to
        // first look for nonzero digits after the decimal point;
        // then check the size.  If the digit count is 18 or less, then
        // the value can definitely be represented as a long.  If it is 19
        // then it may be too large.

        // Trim trailing zeros.  This does not change the represented value.
        while (count > 0 && digits[count - 1] == '0") {
            --count;
        }

        if (count == 0) {
            // Positive zero fits into a long, but negative zero can only
            // be represented as a double. - bug 4162852
            return isPositive || ignoreNegativeZero;
        }

        if (decimalAt < count || decimalAt > MAX_COUNT) {
            return false;
        }

        if (decimalAt < MAX_COUNT) return true;

        // At this point we have decimalAt == count, and count == MAX_COUNT.
        // The number will overflow if it is larger than 9223372036854775807
        // or smaller than -9223372036854775808.
        for (int i=0; i<count; ++i) {
            char dig = digits[i], max = LONG_MIN_REP[i];
            if (dig > max) return false;
            if (dig < max) return true;
        }

        // At this point the first count digits match.  If decimalAt is less
        // than count, then the remaining digits are zero, and we return true.
        if (count < decimalAt) return true;

        // Now we have a representation of Long.MIN_VALUE, without the leading
        // negative sign.  If this represents a positive value, then it does
        // not fit; otherwise it fits.
        return !isPositive;
    
public final java.math.BigDecimalgetBigDecimal()

        if (count == 0) {
            if (decimalAt == 0) {
                return BigDecimal.ZERO;
            } else {
                return new BigDecimal("0E" + decimalAt);
            }
        }

        StringBuffer temp = new StringBuffer(count + 12);
        temp.append('.");
        temp.append(digits, 0, count);
        temp.append('E");
        temp.append(decimalAt);
        return new BigDecimal(temp.toString());
    
private final char[]getDataChars(int length)

        if (data == null || data.length < length) {
            data = new char[length];
        }
        return data;
    
public final doublegetDouble()
Utility routine to get the value of the digit list If (count == 0) this throws a NumberFormatException, which mimics Long.parseLong().

        if (count == 0) {
            return 0.0;
        }

        StringBuffer temp = getStringBuffer();
        temp.append('.");
        temp.append(digits, 0, count);
        temp.append('E");
        temp.append(decimalAt);
        return Double.parseDouble(temp.toString());
    
public final longgetLong()
Utility routine to get the value of the digit list. If (count == 0) this returns 0, unlike Long.parseLong().

        // for now, simple implementation; later, do proper IEEE native stuff

        if (count == 0) {
            return 0;
        }

        // We have to check for this, because this is the one NEGATIVE value
        // we represent.  If we tried to just pass the digits off to parseLong,
        // we'd get a parse failure.
        if (isLongMIN_VALUE()) {
            return Long.MIN_VALUE;
        }

        StringBuffer temp = getStringBuffer();
        temp.append(digits, 0, count);
        for (int i = count; i < decimalAt; ++i) {
            temp.append('0");
        }
        return Long.parseLong(temp.toString());
    
private java.lang.StringBuffergetStringBuffer()

        if (tempBuffer == null) {
            tempBuffer = new StringBuffer(MAX_COUNT);
        } else {
            tempBuffer.setLength(0);
        }
        return tempBuffer;
    
public inthashCode()
Generates the hash code for the digit list.

        int hashcode = decimalAt;

        for (int i = 0; i < count; i++) {
            hashcode = hashcode * 37 + digits[i];
        }

        return hashcode;
    
private booleanisLongMIN_VALUE()
Returns true if this DigitList represents Long.MIN_VALUE; false, otherwise. This is required so that getLong() works.

        if (decimalAt != count || count != MAX_COUNT) {
            return false;
        }

        for (int i = 0; i < count; ++i) {
            if (digits[i] != LONG_MIN_REP[i]) return false;
        }

        return true;
    
booleanisZero()
Return true if the represented number is zero.


                 
      
        for (int i=0; i < count; ++i) {
            if (digits[i] != '0") {
                return false;
            }
        }
        return true;
    
private static final intparseInt(char[] str, int offset, int strLen)

        char c;
        boolean positive = true;
        if ((c = str[offset]) == '-") {
            positive = false;
            offset++;
        } else if (c == '+") {
            offset++;
        }

        int value = 0;
        while (offset < strLen) {
            c = str[offset++];
            if (c >= '0" && c <= '9") {
                value = value * 10 + (c - '0");
            } else {
                break;
            }
        }
        return positive ? value : -value;
    
private final voidround(int maximumDigits)
Round the representation to the given number of digits.

param
maximumDigits The maximum number of digits to be shown. Upon return, count will be less than or equal to maximumDigits.

        // Eliminate digits beyond maximum digits to be displayed.
        // Round up if appropriate.
        if (maximumDigits >= 0 && maximumDigits < count) {
            if (shouldRoundUp(maximumDigits)) {
                // Rounding up involved incrementing digits from LSD to MSD.
                // In most cases this is simple, but in a worst case situation
                // (9999..99) we have to adjust the decimalAt value.
                for (;;) {
                    --maximumDigits;
                    if (maximumDigits < 0) {
                        // We have all 9's, so we increment to a single digit
                        // of one and adjust the exponent.
                        digits[0] = '1";
                        ++decimalAt;
                        maximumDigits = 0; // Adjust the count
                        break;
                    }

                    ++digits[maximumDigits];
                    if (digits[maximumDigits] <= '9") break;
                    // digits[maximumDigits] = '0'; // Unnecessary since we'll truncate this
                }
                ++maximumDigits; // Increment for use as count
            }
            count = maximumDigits;

            // Eliminate trailing zeros.
            while (count > 1 && digits[count-1] == '0") {
                --count;
            }
        }
    
final voidset(java.lang.String s, int maximumDigits, boolean fixedPoint)
Generate a representation of the form DDDDD, DDDDD.DDDDD, or DDDDDE+/-DDDDD.

        int len = s.length();
        char[] source = getDataChars(len);
        s.getChars(0, len, source, 0);

        decimalAt = -1;
        count = 0;
        int exponent = 0;
        // Number of zeros between decimal point and first non-zero digit after
        // decimal point, for numbers < 1.
        int leadingZerosAfterDecimal = 0;
        boolean nonZeroDigitSeen = false;

        for (int i = 0; i < len; ) {
            char c = source[i++];
            if (c == '.") {
                decimalAt = count;
            } else if (c == 'e" || c == 'E") {
                exponent = parseInt(source, i, len);
                break;
            } else {
                if (!nonZeroDigitSeen) {
                    nonZeroDigitSeen = (c != '0");
                    if (!nonZeroDigitSeen && decimalAt != -1)
                        ++leadingZerosAfterDecimal;
                }
                if (nonZeroDigitSeen) {
                    digits[count++] = c;
                }
            }
        }
        if (decimalAt == -1) {
            decimalAt = count;
        }
        if (nonZeroDigitSeen) {
            decimalAt += exponent - leadingZerosAfterDecimal;
        }

        if (fixedPoint) {
            // The negative of the exponent represents the number of leading
            // zeros between the decimal and the first non-zero digit, for
            // a value < 0.1 (e.g., for 0.00123, -decimalAt == 2).  If this
            // is more than the maximum fraction digits, then we have an underflow
            // for the printed representation.
            if (-decimalAt > maximumDigits) {
                // Handle an underflow to zero when we round something like
                // 0.0009 to 2 fractional digits.
                count = 0;
                return;
            } else if (-decimalAt == maximumDigits) {
                // If we round 0.0009 to 3 fractional digits, then we have to
                // create a new one digit in the least significant location.
                if (shouldRoundUp(0)) {
                    count = 1;
                    ++decimalAt;
                    digits[0] = '1";
                } else {
                    count = 0;
                }
                return;
            }
            // else fall through
        }

        // Eliminate trailing zeros.
        while (count > 1 && digits[count - 1] == '0") {
            --count;
        }

        // Eliminate digits beyond maximum digits to be displayed.
        // Round up if appropriate.
        round(fixedPoint ? (maximumDigits + decimalAt) : maximumDigits);
    
public final voidset(long source)
Utility routine to set the value of the digit list from a long

        set(source, 0);
    
public final voidset(long source, int maximumDigits)
Set the digit list to a representation of the given long value.

param
source Value to be converted; must be >= 0 or == Long.MIN_VALUE.
param
maximumDigits The most digits which should be converted. If maximumDigits is lower than the number of significant digits in source, the representation will be rounded. Ignored if <= 0.

        // This method does not expect a negative number. However,
        // "source" can be a Long.MIN_VALUE (-9223372036854775808),
        // if the number being formatted is a Long.MIN_VALUE.  In that
        // case, it will be formatted as -Long.MIN_VALUE, a number
        // which is outside the legal range of a long, but which can
        // be represented by DigitList.
        if (source <= 0) {
            if (source == Long.MIN_VALUE) {
                decimalAt = count = MAX_COUNT;
                System.arraycopy(LONG_MIN_REP, 0, digits, 0, count);
            } else {
                decimalAt = count = 0; // Values <= 0 format as zero
            }
        } else {
            // Rewritten to improve performance.  I used to call
            // Long.toString(), which was about 4x slower than this code.
            int left = MAX_COUNT;
            int right;
            while (source > 0) {
                digits[--left] = (char)('0" + (source % 10));
                source /= 10;
            }
            decimalAt = MAX_COUNT - left;
            // Don't copy trailing zeros.  We are guaranteed that there is at
            // least one non-zero digit, so we don't have to check lower bounds.
            for (right = MAX_COUNT - 1; digits[right] == '0"; --right)
                ;
            count = right - left + 1;
            System.arraycopy(digits, left, digits, 0, count);
        }
        if (maximumDigits > 0) round(maximumDigits);
    
final voidset(java.math.BigDecimal source, int maximumDigits, boolean fixedPoint)
Set the digit list to a representation of the given BigDecimal value. This method supports both fixed-point and exponential notation.

param
source Value to be converted; must not be a value <= 0.
param
maximumDigits The most fractional or total digits which should be converted.
param
fixedPoint If true, then maximumDigits is the maximum fractional digits to be converted. If false, total digits.

        String s = source.toString();
        extendDigits(s.length());

        set(s, maximumDigits, fixedPoint);
    
final voidset(java.math.BigInteger source, int maximumDigits)
Set the digit list to a representation of the given BigInteger value.

param
source Value to be converted; must be >= 0.
param
maximumDigits The most digits which should be converted. If maximumDigits is lower than the number of significant digits in source, the representation will be rounded. Ignored if <= 0.

        String s = source.toString();
        int len = s.length();
        extendDigits(len);
        s.getChars(0, len, digits, 0);

        decimalAt = len;
        int right;
        for (right = len - 1; right >= 0 && digits[right] == '0"; --right)
            ;
        count = right + 1;

        if (maximumDigits > 0) {
            round(maximumDigits);
        }
    
public final voidset(double source, int maximumFractionDigits)
Set the digit list to a representation of the given double value. This method supports fixed-point notation.

param
source Value to be converted; must not be Inf, -Inf, Nan, or a value <= 0.
param
maximumFractionDigits The most fractional digits which should be converted.

        set(source, maximumFractionDigits, true);
    
final voidset(double source, int maximumDigits, boolean fixedPoint)
Set the digit list to a representation of the given double value. This method supports both fixed-point and exponential notation.

param
source Value to be converted; must not be Inf, -Inf, Nan, or a value <= 0.
param
maximumDigits The most fractional or total digits which should be converted.
param
fixedPoint If true, then maximumDigits is the maximum fractional digits to be converted. If false, total digits.

        set(Double.toString(source), maximumDigits, fixedPoint);
    
private booleanshouldRoundUp(int maximumDigits)
Return true if truncating the representation to the given number of digits will result in an increment to the last digit. This method implements half-even rounding, the default rounding mode. [bnf]

param
maximumDigits the number of digits to keep, from 0 to count-1. If 0, then all digits are rounded away, and this method returns true if a one should be generated (e.g., formatting 0.09 with "#.#").
return
true if digit maximumDigits-1 should be incremented

        boolean increment = false;
        // Implement IEEE half-even rounding
        if (maximumDigits < count) {
            if (digits[maximumDigits] > '5") {
                return true;
            } else if (digits[maximumDigits] == '5" ) {
                for (int i=maximumDigits+1; i<count; ++i) {
                    if (digits[i] != '0") {
                        return true;
                    }
                }
                return maximumDigits > 0 && (digits[maximumDigits-1] % 2 != 0);
            }
        }
        return false;
    
public java.lang.StringtoString()


       
        if (isZero()) {
            return "0";
        }
        StringBuffer buf = getStringBuffer();
        buf.append("0.");
        buf.append(digits, 0, count);
        buf.append("x10^");
        buf.append(decimalAt);
        return buf.toString();