286

Given a double, I want to round it to a given number of points of precision after the decimal point, similar to PHP's round() function.

The closest thing I can find in the Dart docs is double.toStringAsPrecision(), but this is not quite what I need because it includes the digits before the decimal point in the total points of precision.

For example, using toStringAsPrecision(3):

0.123456789 rounds to 0.123  
9.123456789 rounds to 9.12  
98.123456789 rounds to 98.1  
987.123456789 rounds to 987  
9876.123456789 rounds to 9.88e+3

As the magnitude of the number increases, I correspondingly lose precision after the decimal place.

Lucas Meadows
  • 2,939
  • 2
  • 14
  • 4

28 Answers28

434

See the docs for num.toStringAsFixed().

String toStringAsFixed(int fractionDigits)

Returns a decimal-point string-representation of this.

Converts this to a double before computing the string representation.

  • If the absolute value of this is greater or equal to 10^21 then this methods returns an exponential representation computed by this.toStringAsExponential().

Examples:

1000000000000000000000.toStringAsExponential(3); // 1.000e+21
  • Otherwise the result is the closest string representation with exactly fractionDigits digits after the decimal point. If fractionDigits equals 0 then the decimal point is omitted.

The parameter fractionDigits must be an integer satisfying: 0 <= fractionDigits <= 20.

Examples:

1.toStringAsFixed(3);  // 1.000
(4321.12345678).toStringAsFixed(3);  // 4321.123
(4321.12345678).toStringAsFixed(5);  // 4321.12346
123456789012345678901.toStringAsFixed(3);  // 123456789012345683968.000
1000000000000000000000.toStringAsFixed(3); // 1e+21
5.25.toStringAsFixed(0); // 5
Vrushi Patel
  • 2,361
  • 1
  • 17
  • 30
Greg Lowe
  • 15,430
  • 2
  • 30
  • 33
  • Must be the accepted + Reference + Explanation + Example – Kohls Dec 10 '19 at 13:42
  • 5
    I argue is toStringAsFixed() the rounding method to use due to its inconsistency. For example, try `5.550.toStringAsFixed(1)` – brendan Jan 15 '20 at 06:56
  • Also watch out when rounding to 0 places as the expression returns an `int` instead of a `double` in that one case e.g. `num.parse(50.123.toStringAsFixed(0)) is int` - this can be fixed by adding `.toDouble()` to the end of it. – abulka Aug 22 '20 at 05:44
  • @brendan Given that a `double` that *attempts* to store 5.55 is actually 5.5499999999999998... then `toStringAsFixed(1)` returns the correct answer (5.5). – jamesdlin Feb 28 '22 at 04:02
  • Is there a good way to do this, but then also have it show zero decimals if the trailing decimals are all zeros? Ex: `12.000` would show as `12` but `12.011` would still show as `12.011`? – Rob Caraway Dec 08 '22 at 14:36
130

num.toStringAsFixed() rounds. This one turns you num (n) into a string with the number of decimals you want (2), and then parses it back to your num in one sweet line of code:

n = num.parse(n.toStringAsFixed(2));
MendelG
  • 14,885
  • 4
  • 25
  • 52
Yngvar Natland
  • 1,489
  • 1
  • 9
  • 4
  • 26
    That's the quickest way to do it. But it's dumb that dart doesn't have a direct way to do this – ThinkDigital Oct 12 '18 at 04:10
  • 3
    Extension method: `extension NumberRounding on num { num toPrecision(int precision) { return num.parse((this).toStringAsFixed(precision)); } }` – Vit Veres Apr 05 '20 at 19:40
  • @ThinkDigital Dart does have a direct way: `toStringAsFixed`. It does *not* have a direct way of rounding a `double` to a specified number of decimal digits *and keeping the result as a `double`* because that's a dumb thing to do. The decimal value you want very likely will not be exactly representable as a `double`. – jamesdlin Mar 01 '22 at 00:07
  • I think that this answer should clarify that what this code actually does is to return the *nearest* `double` to the rounded value. – jamesdlin Mar 01 '22 at 00:11
  • 2
    Had run some quick measurements. toString and parsing is roughly 50x slower than the math solution with POW for 10 million iterations. 25x slower for 1 million iterations and 10x slower for 100 000 iterations. In case someone wondered, like me. – Stefan Rein Oct 13 '22 at 14:01
100

Direct way:

double d = 2.3456789;
String inString = d.toStringAsFixed(2); // '2.35'
double inDouble = double.parse(inString); // 2.35 

Using an extension:

extension Ex on double {
  double toPrecision(int n) => double.parse(toStringAsFixed(n));
}

Usage:

void main() {
  double d = 2.3456789;
  double d1 = d.toPrecision(1); // 2.3
  double d2 = d.toPrecision(2); // 2.35
  double d3 = d.toPrecision(3); // 2.345
}
CopsOnRoad
  • 237,138
  • 77
  • 654
  • 440
70

Above solutions do not appropriately round numbers. I use:

double dp(double val, int places){ 
   num mod = pow(10.0, places); 
   return ((val * mod).round().toDouble() / mod); 
}
Ahmed Ashour
  • 5,179
  • 10
  • 35
  • 56
andyw
  • 3,505
  • 2
  • 30
  • 44
  • Confirmed, this rounds `dp(5.550, 1)` correctly to `5.6` - unlike the most popular answer which gives a slightly incorrect answer of `5.5` as @brendan pointed out. – abulka Aug 22 '20 at 05:49
  • 3
    Does not work for dp(73.4750, 2) It returns 73.47 instead of 73.48 https://www.calculatorsoup.com/calculators/math/roundingnumbers.php?x=73.4750&round=2&action=solve check out this link – Deep Shah Sep 15 '20 at 07:37
  • @DeepShah I believe this answer -> [ https://stackoverflow.com/a/66840734/3002719 ] addresses your concern about the " ***number 5 rounding rule*** " issue you identified here above. – SilSur Mar 28 '21 at 12:05
  • This is inherently problematic because rounding IEEE-754 floating-point numbers (which are binary) to decimal precision is nonsensical; many fractional decimal numbers can't be represented in binary. Also see [Is floating point math broken?](https://stackoverflow.com/q/588004/). – jamesdlin Aug 30 '21 at 07:29
  • can you please convert 4436.605 to 4436.61 (with 2 places) – Nazmus Saadat Jan 03 '22 at 09:12
  • @DeepShah To prevent floating-point errors use [this](https://pub.dev/packages/decimal) package and round with: ```Decimal.parse(this.toString()).round(scale: places).toDouble()```. Rounding is now correct. – Paul Feb 27 '22 at 10:40
  • @DeepShah, Paul - see my comment here [ https://stackoverflow.com/a/66840734/3002719 ] ... – SilSur Apr 29 '22 at 17:53
37
var price = 99.012334554;
price = price.toStringAsFixed(2);
print(price); // 99.01

That is the ref of dart. ref: https://api.dartlang.org/stable/2.3.0/dart-core/num/toStringAsFixed.html

MendelG
  • 14,885
  • 4
  • 25
  • 52
Rizwan Ansar
  • 690
  • 8
  • 17
  • This answer does not add anything to [Greg Lowe's much earlier answer](https://stackoverflow.com/a/28419456/). – jamesdlin Mar 01 '22 at 00:14
31
void main() {
  int decimals = 2;
  int fac = pow(10, decimals);
  double d = 1.234567889;
  d = (d * fac).round() / fac;
  print("d: $d");
}

Prints: 1.23

Robert
  • 5,484
  • 1
  • 22
  • 35
  • 3
    This is the best current way to do this. Rounding to a precision like 0.01 that isn't itself a double, isn't trivial. The result may not even be representable as a double at all. I highly recommend using integers for anything where decimal precision is important (like, say, money). – lrn Feb 10 '15 at 18:16
  • 1
    Also worth pointing out that if compiling to js, then Dart integers will also start to lose precision and you'll need to use a package like: https://pub.dartlang.org/packages/bignum – Greg Lowe Feb 10 '15 at 22:03
  • This method feels like the current best and most elegant way to do this. – gi097 Aug 20 '18 at 06:55
27

I used the toStringAsFixed() method, to round a number to specific numbers after the decimal point EX:

double num = 22.48132906

and when I rounded it to two numbers like this:

print(num.toStringAsFixed(2)) ;

It printed 22.48

and when I rounded to one number, it printed 22.5

MendelG
  • 14,885
  • 4
  • 25
  • 52
Sana'a Al-ahdal
  • 1,684
  • 1
  • 17
  • 33
  • 1
    This answer does not add anything to [Greg Lowe's much earlier answer](https://stackoverflow.com/a/28419456/). – jamesdlin Mar 01 '22 at 00:15
22

The modified answer of @andyw using Dart Extension methods:

extension Precision on double {
    double toPrecision(int fractionDigits) {
        double mod = pow(10, fractionDigits.toDouble());
        return ((this * mod).round().toDouble() / mod);
    }
}

Usage:

var latitude = 1.123456;
var latitudeWithFixedPrecision = latitude.toPrecision(3); // Outputs: 1.123
kubastick
  • 461
  • 6
  • 12
15

you can simply multiple the value in 100 and then round it and then divide it again into 100.

(number * 100).round() / 100.0;
Reza Taghizadeh
  • 347
  • 5
  • 11
14

To round a double in Dart to a given degree of precision AFTER the decimal point, you can use built-in solution in dart toStringAsFixed() method, but you have to convert it back to double

void main() {
  double step1 = 1/3;  
  print(step1); // 0.3333333333333333
  
  String step2 = step1.toStringAsFixed(2); 
  print(step2); // 0.33 
  
  double step3 = double.parse(step2);
  print(step3); // 0.33
}

Boken
  • 4,825
  • 10
  • 32
  • 42
sultanmyrza
  • 4,551
  • 1
  • 30
  • 24
  • 2
    This answer does not add anything to [Yngvar Natland's much earlier answer](https://stackoverflow.com/a/32205216/). Also, I think that this answer should clarify that what this code actually does is to return the *nearest* `double` to the rounded value. – jamesdlin Mar 01 '22 at 00:16
11
double value = 2.8032739273;
String formattedValue = value.toStringAsFixed(3);
tgdavies
  • 10,307
  • 4
  • 35
  • 40
Muhammad Umair Saqib
  • 1,287
  • 1
  • 9
  • 20
8

You can use toStringAsFixed in order to display the limited digits after decimal points. toStringAsFixed returns a decimal-point string-representation. toStringAsFixed accepts an argument called fraction Digits which is how many digits after decimal we want to display. Here is how to use it.

double pi = 3.1415926;
const val = pi.toStringAsFixed(2); // 3.14
Hadi Mir
  • 4,497
  • 2
  • 29
  • 31
6

I made this extension on double

import 'dart:math';

extension DoubleExtension on double {

  /// rounds the double to a specific decimal place
  double roundedPrecision(int places) {
    double mod = pow(10.0, places) as double;
    return ((this * mod).round().toDouble() / mod);
  }

  /// good for string output because it can remove trailing zeros
  /// and sometimes periods. Or optionally display the exact number of trailing
  /// zeros
  String roundedPrecisionToString(
    int places, {
    bool trailingZeros = false,
  }) {
    double mod = pow(10.0, places) as double;
    double round = ((this * mod).round().toDouble() / mod);
    String doubleToString =
        trailingZeros ? round.toStringAsFixed(places) : round.toString();
    if (!trailingZeros) {
      RegExp trailingZeros = RegExp(r'^[0-9]+.0+$');
      if (trailingZeros.hasMatch(doubleToString)) {
        doubleToString = doubleToString.split('.')[0];
      }
    }
    return doubleToString;
  }

  String toStringNoTrailingZeros() {
    String doubleToString = toString();
    RegExp trailingZeros = RegExp(r'^[0-9]+.0+$');
    if (trailingZeros.hasMatch(doubleToString)) {
      doubleToString = doubleToString.split('.')[0];
    }
    return doubleToString;
  }
}

Here are the passing tests.

import 'package:flutter_test/flutter_test.dart';
import 'package:project_name/utils/double_extension.dart';

void main() {
  group("rounded precision", () {
    test("rounding to 0 place results in an int", () {
      double num = 5.1234;
      double num2 = 5.8234;
      expect(num.roundedPrecision(0), 5);
      expect(num2.roundedPrecision(0), 6);
    });
    test("rounding to 1 place rounds correctly to 1 place", () {
      double num = 5.12;
      double num2 = 5.15;
      expect(num.roundedPrecision(1), 5.1);
      expect(num2.roundedPrecision(1), 5.2);
    });
    test(
        "rounding a number to a precision that is more accurate than the origional",
        () {
      double num = 5;
      expect(num.roundedPrecision(5), 5);
    });
  });

  group("rounded precision returns the correct string", () {
    test("rounding to 0 place results in an int", () {
      double num = 5.1234;
      double num2 = 5.8234;
      expect(num.roundedPrecisionToString(0), "5");
      expect(num2.roundedPrecisionToString(0), "6");
    });
    test("rounding to 1 place rounds correct", () {
      double num = 5.12;
      double num2 = 5.15;
      expect(num.roundedPrecisionToString(1), "5.1");
      expect(num2.roundedPrecisionToString(1), "5.2");
    });
    test("rounding to 2 places rounds correct", () {
      double num = 5.123;
      double num2 = 5.156;
      expect(num.roundedPrecisionToString(2), "5.12");
      expect(num2.roundedPrecisionToString(2), "5.16");
    });
    test("cut off all trailing zeros (and periods)", () {
      double num = 5;
      double num2 = 5.03000;
      expect(num.roundedPrecisionToString(5), "5");
      expect(num2.roundedPrecisionToString(5), "5.03");
    });
  });
}
Shawn Ashton
  • 357
  • 5
  • 10
5

Above solutions do not work for all cases. What worked for my problem was this solution that will round your number (0.5 to 1 or 0.49 to 0) and leave it without any decimals:

Input: 12.67

double myDouble = 12.67;
var myRoundedNumber; // Note the 'var' datatype

// Here I used 1 decimal. You can use another value in toStringAsFixed(x)
myRoundedNumber = double.parse((myDouble).toStringAsFixed(1));
myRoundedNumber = myRoundedNumber.round();

print(myRoundedNumber);

Output: 13

This link has other solutions too

Runsis
  • 831
  • 9
  • 19
5

You can create a reusable function that accept numberOfDecimal you want to format & utilizing toStringAsFixed() method to format the number and convert it back to double.

FYI, toStringAsFixed method does not round up number that ends with 5 (eg: toStringAsFixed round off 2.275 to 2.27 instead of 2.28). This is the default behaviour of dart toStringAsFixed method (similar to Javascript toFixed)

As a workaround, we can add 1 to the existing number after the last decimal number (eg: Add 0.0001 to 2.275 become 2.2751 & 2.2751 will round off correctly to 2.28)

double roundOffToXDecimal(double number, {int numberOfDecimal = 2}) {
  // To prevent number that ends with 5 not round up correctly in Dart (eg: 2.275 round off to 2.27 instead of 2.28)
  String numbersAfterDecimal = number.toString().split('.')[1];
  if (numbersAfterDecimal != '0') {
    int existingNumberOfDecimal = numbersAfterDecimal.length;
    double incrementValue = 1 / (10 * pow(10, existingNumberOfDecimal));
    if (number < 0) {
       number -= incrementValue;
    } else {
       number += incrementValue;
    }
  }

  return double.parse(number.toStringAsFixed(numberOfDecimal));
}

// Example of usage:
var price = roundOffToXDecimal(2.275, numberOfDecimal: 2)
print(price); // 2.28
luke77
  • 2,255
  • 2
  • 18
  • 30
  • 1
    Well done -> your solution works perfectly ( rounds + adheres to rounding rule ) and your code is way leaner than my solution. So I abandon my solution for yours - thanks for the effort man. [ `two thumbs up` ] – SilSur Apr 03 '23 at 08:02
  • `.toStringAsFixed()` *does* round up numbers that end with a 5; `0.5.toStringAsFixed(0)` is `'1'`, `0.25.toStringAsFixed(1)` is `'0.3'`, and `0.125.toStringAsFixed(2)` is `'0.13'`. As explained by the thread this answer links to, the reason why that doesn't happen in general is because `double`s are inherently imprecise when representing decimal numbers. For example, 2.275 as a `double` is actually 2.274999999999999911182158029987476766109466552734375. – jamesdlin Apr 03 '23 at 13:43
3

If you don't want any decimals when the resulting decimals are all zeroes, something like this would work:

String fixedDecimals(double d, int decimals, {bool removeZeroDecimals = true}){
  double mod = pow(10.0, decimals);
  double result = ((d * mod).round().toDouble() / mod);
  if( removeZeroDecimals && result - (result.truncate()) == 0.0 ) decimals = 0;
  return result.toStringAsFixed(decimals);
}

This will simply output 9 instead of 9.00 if the input is 9.004 and you want 2 decimals.

Magnus
  • 17,157
  • 19
  • 104
  • 189
  • I don't know why you guys are all giving truncating solutions when the questions clearly states -> `How do you round a double in Dart to a given degree of precision ...` – SilSur Apr 03 '23 at 07:39
3

Rounding a double, an IEEE-754 binary floating-point number, to a specific number of decimal digits is inherently problematic if you want a double result.

In the same way that fractions such as 1/3 can't be exactly represented with a finite number of decimal digits, many (well, infinitely many) decimal numbers can't be represented with a finite number of binary digits. For example, the decimal number 0.1 cannot be exactly represented in binary. While you could try to round 0.09999 to 0.1, as a double it would actually be "rounded" to 0.1000000000000000055511151231257827021181583404541015625. Most of the other answers that claim to round doubles with decimal precision actually return the nearest representable double. When you display those values to users by converting them to Strings, you might see more digits than you expect.

What you can do is to make the string representation look like a nice, rounded number when you ultimately show it to users, and that's what double.toStringAsFixed() does. That's also why when you print 0.100000000..., you might see 0.1 if the implementation is trying to pretty-print user-friendly values. However, don't be fooled: the double value would never actually be 0.1 exactly, and if you do repeated arithmetic with such inexact values, you can accumulate error (for example: 0.1 + 0.2).

Note that all of the above is fundamental to how binary floating-point numbers work and is not specific to Dart. Also see:

Bottom line: If you care about decimal precision, do NOT use binary floating-point types. This is particularly important if you're dealing with money.

You instead should use:

  • Integers. For example, if you are dealing with currency, instead of using double dollars = 1.23;, use int cents = 123;. Your calculations then always will be exact, and you can convert to the desired units only when displaying them to the user (and likewise can convert in the opposite direction when reading input from the user).
  • A type designed to represent decimal numbers with arbitrary precision. For example, package:decimal provides a Decimal type. With such a type, some of the other answers (such as multiplying by 100, rounding, and then dividing by 100) then would be appropriate. (But really you should use Decimal.round directly.)
jamesdlin
  • 81,374
  • 13
  • 159
  • 204
  • To round accurately with this package use: ```Decimal.parse(this.toString()).round(scale: places).toDouble()``` – Paul Feb 27 '22 at 10:37
  • @Paul You should get rid of the `toDouble()`. That defeats the point of using `package:decimal`. – jamesdlin Feb 27 '22 at 16:41
  • There might be cases where a double return is questioned, so I thought I should add it. Also: this rounds a double correctly (possibly only once). But yes, If you would need to do calculations with doubles/decimals you shouldn't do that obviously. – Paul Feb 27 '22 at 23:47
  • @Paul "Rounds a double correctly" is a nonsensical statement since you cannot exactly round a `double` in terms of base-10 digits. Fundamentally there are base-10 numbers that cannot be exactly represented as `double`s. No "rounding" can avoid that. – jamesdlin Feb 28 '22 at 00:27
2

Just write this extension on double

extension Round on double {
  double roundToPrecision(int n) {
    int fac = pow(10, n).toInt();
    return (this * fac).round() / fac;
  }
}
2

I think the accepted answer is not the perfect solution because it converts to string.

If you don't wanna convert to string and back to a double use double.toPrecision(decimalNumber) from GetX package.

If you don't wanna use GetX just for this (I highly recommend GetX, it will change your life with flutter) you can copy and paste this.

Remeber to import the file when you wanna use the extention.

import 'dart:math';

extension Precision on double {
  double toPrecision(int fractionDigits) {
    var mod = pow(10, fractionDigits.toDouble()).toDouble();
    return ((this * mod).round().toDouble() / mod);
  }
}
Mayb3Not
  • 441
  • 4
  • 10
2

if use dynamic type of data. You can use it.

 typeDecimal(data) => num.parse(data.toString()).toStringAsFixed(2);
2

also if you want to round the double value inside the Text.

Text('${carpetprice.toStringAsFixed(3)}',),

Abu Bakar
  • 21
  • 2
1

Never thought this was so complex in Dart but this is my solution:

double truncateDouble(double val, int decimals) {
    String valString = val.toString();
    int dotIndex = valString.indexOf('.');

    // not enough decimals
    int totalDecimals = valString.length - dotIndex - 1;
    if (totalDecimals < decimals) {
      decimals = totalDecimals;
    }

    valString = valString.substring(0, dotIndex + decimals + 1);

    return double.parse(valString);
  }

var val = truncateDouble(44.999, 2);
Dani
  • 3,128
  • 2
  • 43
  • 91
  • 1
    As with almost all of the other answers, this answer should clarify that what this code actually does is to return the *nearest* `double` to the truncated value, so when you convert it back to a `String`, you might end up with many more digits than you expect. It is inherently impossible to exactly represent most fractional decimal numbers in binary. – jamesdlin Feb 14 '23 at 18:38
  • @Dani - This solution also fails the basic rounding rule -> `truncateDouble(73.4750, 2)` returns `73.47` and not the correct value of `73.48` . @jamesdlin - alright then ... give us your solution (or atleast try to formulate a solution) instead of just shooting down others' attempts at trying to find a work-around to this "well-known" rounding problem. – SilSur Apr 03 '23 at 07:16
  • @SilSur I've already provided [my own answer](https://stackoverflow.com/a/68980598/). Furthermore, I'd hardly call suggesting a clarification to be "shooting down" an answer. – jamesdlin Apr 03 '23 at 13:32
0

ABANDONED !!! -> Please see @luke77 's code - it correctly solves this problem with much leaner code.

This DART rounding problem has been a long time coming (@LucasMeadows), since it's clear that it has not been adequately solved (as indicated by @DeepShah's observation) until now.

The well-known rounding rule (the unsolved problem):

" Rounding numbers that end with the number 5: round up if the result is an even number; round down if the result is an odd number. "

So here is the DART code solution:

double roundAccurately(double numToRound, int decimals) {

  // Step 1 - Prime IMPORTANT Function Parameters ...
  int iCutIndex = 0;
  String sDeciClipdNTR = "";
  num nMod = pow(10.0, decimals);
  String sNTR = numToRound.toString();
  int iLastDigitNTR = 0, i2ndLastDigitNTR = 0;
  debugPrint("Round => $numToRound to $decimals Decimal ${(decimals == 1) ? "Place" : "Places"} !!");   // Deactivate this 'print()' line in production code !!

  // Step 2 - Calculate Decimal Cut Index (i.e. string cut length) ...
  int iDeciPlaces = (decimals + 2);
  if (sNTR.contains('.')) {
    iCutIndex = sNTR.indexOf('.') + iDeciPlaces;
  } else {
    sNTR = sNTR + '.';
    iCutIndex = sNTR.indexOf('.') + iDeciPlaces;
  }

  // Step 3 - Cut input double to length of requested Decimal Places ...
  if (iCutIndex > sNTR.length) {                    // Check that decimal cutting is possible ...
    sNTR = sNTR + ("0" * iDeciPlaces);              // ... and fix (lengthen) the input double if it is too short.
    sDeciClipdNTR = sNTR.substring(0, iCutIndex);   // ... then cut string at indicated 'iCutIndex' !!
  } else {
    sDeciClipdNTR = sNTR.substring(0, iCutIndex);   // Cut string at indicated 'iCutIndex' !!
  }

  // Step 4 - Extract the Last and 2nd Last digits of the cut input double.
  int iLenSDCNTR = sDeciClipdNTR.length;
  iLastDigitNTR = int.parse(sDeciClipdNTR.substring(iLenSDCNTR - 1));   // Extract the last digit !!
  (decimals == 0)
    ? i2ndLastDigitNTR = int.parse(sDeciClipdNTR.substring(iLenSDCNTR - 3, iLenSDCNTR - 2))
    : i2ndLastDigitNTR = int.parse(sDeciClipdNTR.substring(iLenSDCNTR - 2, iLenSDCNTR - 1));

  // Step 5 - Execute the FINAL (Accurate) Rounding Process on the cut input double.
  double dAccuRound = 0;
  if (iLastDigitNTR == 5 && ((i2ndLastDigitNTR + 1) % 2 != 0)) {
    dAccuRound = double.parse(sDeciClipdNTR.substring(0, iLenSDCNTR - 1));
  } else {
    if (iLastDigitNTR < 5) {
      dAccuRound = double.parse(sDeciClipdNTR.substring(0, iLenSDCNTR - 1));
    } else {
      if (decimals == 0) {
        sDeciClipdNTR = sNTR.substring(0, iCutIndex - 2);
        dAccuRound = double.parse(sDeciClipdNTR) + 1;   // Finally - Round UP !!
      } else {
        double dModUnit = 1 / nMod;
        sDeciClipdNTR = sNTR.substring(0, iCutIndex - 1);
        dAccuRound = double.parse(sDeciClipdNTR) + dModUnit;   // Finally - Round UP !!
      }
    }
  }

  // Step 6 - Run final QUALITY CHECK !!
  double dResFin = double.parse(dAccuRound.toStringAsFixed(decimals));

  // Step 7 - Return result to function call ...
  debugPrint("Result (AccuRound) => $dResFin !!");   // Deactivate this 'print()' line in production code !!
  return dResFin;
}

It's a completely manual approach (and probably a bit of an overkill), but it works. Please test it (to exhaustion) and let me know if I've missed the mark.

ABANDONED !!! -> Please see @luke47 's code - it correctly solves this problem with much leaner code.

SilSur
  • 495
  • 1
  • 6
  • 21
  • Updated function (above): `roundAccurately(73.4750, 2)` returns `73.48` !! @deep-shah – SilSur Dec 16 '21 at 20:53
  • 2
    This is extremely complicated. Have a look at this answer and only use: ```Decimal.parse(this.toString()).round(scale: places).toDouble()``` – Paul Feb 27 '22 at 10:39
  • @Paul - so you want replace a standalone function with 63 lines of code with a 3rd Party Library [ https://github.com/a14n/dart-decimal/blob/master/lib/decimal.dart ] that has over 300 lines of code? Hmm... Also, have you tried quick trials with this library in Dartpad.dev? **( good luck with that !! )** ...but to each his own, I guess. – SilSur Apr 29 '22 at 17:43
  • there is no other solution than re-writing Dart from scratch?? – Dani Dec 06 '22 at 23:34
  • Actually I posted another solution – Dani Dec 07 '22 at 00:02
  • It is misleading to call this "a Dart problem". This is simply how binary floating-point numbers work. Binary floating-point numbers cannot exactly represent fractional decimal numbers in any language. – jamesdlin Feb 22 '23 at 17:28
0

If you want use special rounding. You can try this function (rounding).

void main(List<String> arguments) {
list.map((e) {
 log('list1');
 rounding(e, 0.05);
 rounding(e, 0.1);
 rounding(e, 0.2);
 rounding(e, 0.25);
 rounding(e, 0.5);
 rounding(e, 1);
 rounding(e, 10);
}).toList();
list2.map((e) {
 log('list2');
 rounding(e, 0.05);
 rounding(e, 0.1);
 rounding(e, 0.2);
 rounding(e, 0.25);
 rounding(e, 0.5);
 rounding(e, 1);
 rounding(e, 10);
}).toList();
}

const list = [1.11, 1.22, 1.33, 1.44, 1.55, 1.66, 1.77, 1.88, 1.99];

const list2 = [2.19, 3.28, 4.37, 5.46, 6.55, 7.64, 8.73, 9.82, 10.91];

void rounding(double price, double count) {
log('-----------------------');
log('price: $price, count: $count');
double _priceRemainder = price % count;
double _someDiff = count / _priceRemainder;
log('_price: ${_priceRemainder.toStringAsFixed(2)}');
log('_pricePlus: ${_someDiff.toStringAsFixed(2)}');
if (_someDiff.toStringAsFixed(2) == '1.00') {
 log('_someDiff = 1');
} else if (_someDiff > 1 && _someDiff <= 2 ||
   _someDiff.toStringAsFixed(2) == '2.00') {
 log('_someDiff > 1 && _someDiff <= 2 || _someDiff.toStringAsFixed(2) == 2.00');
 log('ceilToDouble: $price: ${(price + (count - _priceRemainder)).toStringAsFixed(2)}');
 log('floorToDouble: $price: ${(price - _priceRemainder).toStringAsFixed(2)}');
 log('roundToDouble: $price: ${(price + (count - _priceRemainder)).toStringAsFixed(2)}');
} else if (_someDiff > 2) {
 log('_someDiff > 2');
 log('ceilToDouble: $price: ${(price + (count - _priceRemainder)).toStringAsFixed(2)}');
 log('floorToDouble: $price: ${(price - _priceRemainder).toStringAsFixed(2)}');
 log('roundToDouble: $price: ${(price - _priceRemainder).toStringAsFixed(2)}');
}
log('-----------------------');
}

Debug console:


[log] price: 10.91, count: 0.05
[log] _price: 0.01
[log] _pricePlus: 5.00
[log] _someDiff > 2
[log] ceilToDouble: 10.91: 10.95
[log] floorToDouble: 10.91: 10.90
[log] roundToDouble: 10.91: 10.90
2
[log] -----------------------
[log] price: 10.91, count: 0.1
[log] _price: 0.01
[log] _pricePlus: 10.00
[log] _someDiff > 2
[log] ceilToDouble: 10.91: 11.00
[log] floorToDouble: 10.91: 10.90
[log] roundToDouble: 10.91: 10.90
2
[log] -----------------------
[log] price: 10.91, count: 0.2
[log] _price: 0.11
[log] _pricePlus: 1.82
[log] _someDiff > 1 && _someDiff <= 2 || _someDiff.toStringAsFixed(2) == 2.00
[log] ceilToDouble: 10.91: 11.00
[log] floorToDouble: 10.91: 10.80
[log] roundToDouble: 10.91: 11.00
2
[log] -----------------------
[log] price: 10.91, count: 0.25
[log] _price: 0.16
[log] _pricePlus: 1.56
[log] _someDiff > 1 && _someDiff <= 2 || _someDiff.toStringAsFixed(2) == 2.00
[log] ceilToDouble: 10.91: 11.00
[log] floorToDouble: 10.91: 10.75
[log] roundToDouble: 10.91: 11.00
2
[log] -----------------------
[log] price: 10.91, count: 0.5
[log] _price: 0.41
[log] _pricePlus: 1.22
[log] _someDiff > 1 && _someDiff <= 2 || _someDiff.toStringAsFixed(2) == 2.00
[log] ceilToDouble: 10.91: 11.00
[log] floorToDouble: 10.91: 10.50
[log] roundToDouble: 10.91: 11.00
2
[log] -----------------------
[log] price: 10.91, count: 1.0
[log] _price: 0.91
[log] _pricePlus: 1.10
[log] _someDiff > 1 && _someDiff <= 2 || _someDiff.toStringAsFixed(2) == 2.00
[log] ceilToDouble: 10.91: 11.00
[log] floorToDouble: 10.91: 10.00
[log] roundToDouble: 10.91: 11.00
2
[log] -----------------------
[log] price: 10.91, count: 10.0
[log] _price: 0.91
[log] _pricePlus: 10.99
[log] _someDiff > 2
[log] ceilToDouble: 10.91: 20.00
[log] floorToDouble: 10.91: 10.00
[log] roundToDouble: 10.91: 10.00
Nikolay
  • 603
  • 1
  • 6
  • 13
0

If you need proper rounding (up when first digit is 5) and you want to have trailing 0's you can use this method:

import 'dart:math';

String customRound(double val, int places) {
  num mod = pow(10.0, places);
  return ((val * mod).round().toDouble() / mod).toStringAsFixed(places);
}

customRound(2.345) // -> 2.35
customRound(2.500) // -> 2.50
Rens
  • 489
  • 5
  • 11
-1

This function you can call to get degree of precision in dark(flutter). double eval -> double that want to convert int i -> number of decimal point to return.

double doubleToPrecision(double eval, int i) {
double step1 = eval;//1/3
print(step1); // 0.3333333333333333

String step2 = step1.toStringAsFixed(2);
print(step2); // 0.33

double step3 = double.parse(step2);
print(step3); // 0.33
eval = step3;
return eval; }
-2

I prever converting my like this => `

num.tryParse("23.123456789")!.toDouble().roundToDouble()

`

Michael
  • 588
  • 11
  • 19
-5

This works pretty well

var price=99.012334554
price = price.roundTodouble();
print(price); // 99.01
David Buck
  • 3,752
  • 35
  • 31
  • 35