public abstract Class FloatFormatter

Append the current value of exp in the format "e{:+02d}" (for example e+05, e-10, e+308 , etc.).

Parameters

exp:int

exponent value to append

Common code for g-format, none-format and r-format used when the exponent is such that an exponential presentation is chosen. Normally the method removes trailing digits so as to shorten the presentation without loss of significance. Although no minimum number of fractional digits is enforced in the exponential presentation, when minFracDigits<0 this signifies "no truncation" mode, in which trailing zeros generated in the conversion are not removed. This supports "%#g" format.

Parameters

digits:CharSequence

from converting the value at a given precision.

exp:int

would be the exponent (in e-format), used to position the decimal point.

Common code for g-format, none-format and r-format used when the exponent is such that a fixed-point presentation is chosen. Normally the method removes trailing digits so as to shorten the presentation without loss of significance. This method respects the minimum number of fractional digits (digits after the decimal point), in member minFracDigits, which is 0 for g-format and 1 for none-format and r-format. When minFracDigits<0 this signifies "no truncation" mode, in which trailing zeros generated in the conversion are not removed. This supports "%#g" format.

Parameters

digits:CharSequence

from converting the value at a given precision.

exp:int

would be the exponent (in e-format), used to position the decimal point.

precision:int

of conversion (number of significant digits).

Append the trailing fractional zeros, as required by certain formats, so that the total number of fractional digits is no less than specified. If there is no decimal point originally (and therefore no fractional part), the method will add a decimal point, even if it adds no zeros.

Parameters

n:int

smallest number of fractional digits on return

Append the trailing fractional zeros, as required by certain formats, so that the total number of fractional digits is no less than specified. If n<=0, the method leaves the result buffer unchanged.

Parameters

n:int

smallest number of fractional digits on return

Append a decimal point and trailing fractional zeros if necessary for 'e' and 'f' format. This should not be called if the result is not numeric ("inf" for example). This method deals with the following complexities: on return there will be at least the number of fractional digits specified in the argument n, and at least minFracDigits; further, if minFracDigits<0, signifying the "alternate mode" of certain formats, the method will ensure there is a decimal point, even if there are no fractional digits to follow.

Parameters

n:int

smallest number of fractional digits on return

The e-format helper function of format(double, String) that uses Java's BigDecimal to provide conversion and rounding. The converted number is appended to the result buffer, and start will be set to the index of its first character.

Parameters

value:double

to convert

positivePrefix:String

to use before positive values (e.g. "+") or null to default to ""

precision:int

precision (maximum number of fractional digits)

The f-format inner helper function of format(double, String) that uses Java's BigDecimal to provide conversion and rounding. The converted number is appended to the result buffer, and start will be set to the index of its first character.

Parameters

value:double

to convert

positivePrefix:String

to use before positive values (e.g. "+") or null to default to ""

precision:int

precision (maximum number of fractional digits)

Implementation of the variants of g-format, that uses Java's BigDecimal to provide conversion and rounding. These variants are g-format proper, alternate g-format (available for "%#g" formatting), n-format (as g but subsequently "internationalised"), and none-format (type code Spec.NONE).

None-format is the basis of float.__str__.

According to the Python documentation for g-format, the precise rules are as follows: suppose that the result formatted with presentation type 'e' and precision p-1 would have exponent exp. Then if -4 <= exp < p, the number is formatted with presentation type 'f' and precision p-1-exp. Otherwise, the number is formatted with presentation type 'e' and precision p-1. In both cases insignificant trailing zeros are removed from the significand, and the decimal point is also removed if there are no remaining digits following it.

The Python documentation says none-format is the same as g-format, but the observed behaviour differs from this, in that f-format is only used if -4 <= exp < p-1 (i.e. one less), and at least one digit to the right of the decimal point is preserved in the f-format (but not the e-format). That behaviour is controlled through the following arguments, with these recommended values:

Parameters

value:double

to convert

positivePrefix:String

to use before positive values (e.g. "+") or null to default to ""

precision:int

total number of significant digits (precision 0 behaves as 1)

expThresholdAdj:int

+precision = the exponent at which to resume using exponential notation

Implementation of r-format (float.__repr__) that uses Java's Double#toString(double) to provide conversion and rounding. That method gives us almost what we need, but not quite (sometimes it yields 18 digits): here we always round to 17 significant digits. Much of the formatting after conversion is shared with format_g(double, String, int, int, int). minFracDigits is consulted since while float.__repr__ truncates to one digit, within complex.__repr__ we truncate fully.

Parameters

value:double

to convert

positivePrefix:String

to use before positive values (e.g. "+") or null to default to ""

Return the index in result of the first letter. This is a helper for uppercase() and getExponent()

Remove trailing zeros from the fractional part, as required by certain formats, leaving at least the number of fractional digits specified. If the resultant number of fractional digits is zero, this method will also remove the trailing decimal point (if there is one).

Parameters

n:int

smallest number of fractional digits on return

Convert a double to digits and an exponent for use in float.__repr__ (or r-format). This method takes advantage of (or assumes) a close correspondence between Double#toString(double) and Python float.__repr__. The correspondence appears to be exact, insofar as the Java method produces the minimal non-zero digit string. It mostly chooses the same number of digits (and the same digits) as the CPython repr, but in a few cases Double.toString produces more digits. This method truncates to the number maxDigits, which in practice is always 17.

Parameters

value:double

to convert

maxDigits:int

maximum number of digits to return in buf.

buf:StringBuilder

for digits of result (recommend size be 20)

Returns:int

the exponent

Common code to deal with the sign, and the special cases "0", "-0", "nan, "inf", or "-inf". If the method returns false, we have started a non-zero number and the sign is already in result. The client need then only encode abs(value). If the method returns true, and lenMarker==0, the value was "0" or "-0": the caller may have to zero-extend this, and/or add an exponent, to match the requested format. If the method returns true, and lenMarker>0, the method has placed "nan, "inf" in the result buffer (preceded by a sign if necessary).

Parameters

value:double

to convert

positivePrefix:String

to use before positive values (e.g. "+") or null to default to ""

Returns:boolean

true if the value was one of "0", "-0", "nan, "inf", or "-inf".

Common code for g-format, none-format and r-format called when the conversion yields "inf", "nan" or zero. The method completes formatting of the zero, with the appropriate number of decimal places or (in particular circumstances) exponential; notation.

Parameters

precision:int

of conversion (number of significant digits).

expThreshold:int

if zero, causes choice of exponential notation for zero.