sdk/lib/core/double.dart

// Copyright (c) 2012, the Dart project authors.  Please see the AUTHORS file
// for details. All rights reserved. Use of this source code is governed by a
// BSD-style license that can be found in the LICENSE file.

part of dart.core;

// TODO: Convert this abstract class into a concrete class double
// that uses the patch class functionality to account for the
// different platform implementations.

/**
 * A double-precision floating point number.
 *
 * Representation of Dart doubles containing double specific constants
 * and operations and specializations of operations inherited from
 * [num]. Dart doubles are 64-bit floating-point numbers as specified in the
 * IEEE 754 standard.
 *
 * The [double] type is contagious. Operations on [double]s return
 * [double] results.
 *
 * It is a compile-time error for a class to attempt to extend or implement
 * double.
 */
abstract class double extends num {
  static const double NAN = 0.0 / 0.0;
  static const double INFINITY = 1.0 / 0.0;
  static const double NEGATIVE_INFINITY = -INFINITY;
  static const double MIN_POSITIVE = 5e-324;
  static const double MAX_FINITE = 1.7976931348623157e+308;

  double remainder(num other);

  /** Addition operator. */
  double operator +(num other);

  /** Subtraction operator. */
  double operator -(num other);

  /** Multiplication operator. */
  double operator *(num other);

  double operator %(num other);

  /** Division operator. */
  double operator /(num other);

  /**
   * Truncating division operator.
   *
   * The result of the truncating division [:a ~/ b:] is equivalent to
   * [:(a / b).truncate():].
   */
  int operator ~/(num other);

  /** Negate operator. */
  double operator -();

  /** Returns the absolute value of this [double]. */
  double abs();

  /**
   * Returns the integer closest to `this`.
   *
   * Rounds away from zero when there is no closest integer:
   *  [:(3.5).round() == 4:] and [:(-3.5).round() == -4:].
   *
   * If `this` is not finite (`NaN` or infinity), throws an [UnsupportedError].
   */
  int round();

  /**
   * Returns the greatest integer no greater than `this`.
   *
   * If `this` is not finite (`NaN` or infinity), throws an [UnsupportedError].
   */
  int floor();

  /**
   * Returns the least integer no smaller than `this`.
   *
   * If `this` is not finite (`NaN` or infinity), throws an [UnsupportedError].
   */
  int ceil();

  /**
   * Returns the integer obtained by discarding any fractional
   * digits from `this`.
   *
   * If `this` is not finite (`NaN` or infinity), throws an [UnsupportedError].
   */
  int truncate();

  /**
   * Returns the integer value, as a double, closest to `this`.
   *
   * Rounds away from zero when there is no closest integer:
   *  [:(3.5).round() == 4:] and [:(-3.5).round() == -4:].
   */
  double roundToDouble();

  /**
   * Returns the greatest integer value no greater than `this`.
   *
   * The result is a double.
   */
  double floorToDouble();

  /**
   * Returns the least integer value no smaller than `this`.
   *
   * The result is a double.
   */
  double ceilToDouble();

  /**
   * Returns the integer obtained by discarding any fractional
   * digits from `this`.
   *
   * The result is a double.
   */
  double truncateToDouble();

  /**
   * Provide a representation of this [double] value.
   *
   * The representation is a number literal such that the closest double value
   * to the representation's mathematical value is this [double].
   *
   * Returns "NaN" for the Not-a-Number value.
   * Returns "Infinity" and "-Infinity" for positive and negative Infinity.
   * Returns "-0.0" for negative zero.
   *
   * It should always be the case that if [:d:] is a [double], then
   * [:d == double.parse(d.toString()):].
   */
  String toString();

  /**
   * Parse [source] as an double literal and return its value.
   *
   * Accepts an optional sign (`+` or `-`) followed by either the characters
   * "Infinity", the characters "NaN" or a floating-point representation.
   * A floating-point representation is composed of a mantissa and an optional
   * exponent part. The mantissa is either a decimal point (`.`) followed by a
   * sequence of (decimal) digits, or a sequence of digits
   * optionally followed by a decimal point and optionally more digits. The
   * (optional) exponent part consists of the character "e" or "E", an optional
   * sign, and one or more digits.
   *
   * Leading and trailing whitespace is ignored.
   *
   * If the [source] is not a valid double literal, the [onError]
   * is called with the [source] as argument, and its return value is
   * used instead. If no `onError` is provided, a [FormatException]
   * is thrown instead.
   *
   * The [onError] function is only invoked if [source] is a [String] with an
   * invalid format. It is not invoked if the [source] is invalid for some
   * other reason, for example by being `null`.
   *
   * Examples of accepted strings:
   *
   *     "3.14"
   *     "  3.14 \xA0"
   *     "0."
   *     ".0"
   *     "-1.e3"
   *     "1234E+7"
   *     "+.12e-9"
   *     "-NaN"
   */
  external static double parse(String source,
                               [double onError(String source)]);
}