51

A thought struck me as I was writing a piece of JavaScript code that processed some floating point values. What is the decimal point symbol in JavaScript? Is it always .? Or is it culture-specific? And what about .toFixed() and .parseFloat()? If I'm processing a user input, it's likely to include the local culture-specific decimal separator symbol.

Ultimately I'd like to write code that supports both decimal points in user input - culture-specific and ., but I can't write such a code if I don't know what JavaScript expects.

Added: OK, Rubens Farias suggests to look at similar question which has a neat accepted answer:

function whatDecimalSeparator() {
    var n = 1.1;
    n = n.toLocaleString().substring(1, 2);
   return n;
}

That's nice, it lets me get the locale decimal point. A step towards the solution, no doubt.

Now, the remaining part would be to determine what the behavior of .parseFloat() is. Several answers point out that for floating point literals only . is valid. Does .parseFloat() act the same way? Or might it require the local decimal separator in some browser? Are there any different methods for parsing floating point numbers as well? Should I roll out my own just-to-be-sure?

Community
  • 1
  • 1
Vilx-
  • 104,512
  • 87
  • 279
  • 422
  • @Vilx- for what you're doing, I would consider using the unary (+) operator instead of `parseFloat()`. It will be much easier when validating input. See my updated answer. – Andy E Jan 18 '10 at 11:06

3 Answers3

24

According to the specification, a DecimalLiteral is defined as:

DecimalLiteral ::
    DecimalIntegerLiteral . DecimalDigitsopt ExponentPartopt 
    . DecimalDigits ExponentPartopt 
    DecimalIntegerLiteral ExponentPartopt

and for satisfying the parseFloat argument:

  1. Let inputString be ToString(string).
  2. Let trimmedString be a substring of inputString consisting of the leftmost character that is not a StrWhiteSpaceChar and all characters to the right of that character.(In other words, remove leading white space.)
  3. If neither trimmedString nor any prefix of trimmedString satisfies the syntax of a StrDecimalLiteral (see 9.3.1), return NaN.
  4. Let numberString be the longest prefix of trimmedString, which might be trimmedString itself, that satisfies the syntax of a StrDecimalLiteral.
  5. Return the Number value for the MV

So numberString becomes the longest prefix of trimmedString that satisfies the syntax of a StrDecimalLiteral, meaning the first parseable literal string number it finds in the input. Only the . can be used to specify a floating-point number. If you're accepting inputs from different locales, use a string replace:

function parseLocalNum(num) {
    return +(num.replace(",", "."));
}

The function uses the unary operator instead of parseFloat because it seems to me that you want to be strict about the input. parseFloat("1ABC") would be 1, whereas using the unary operator +"1ABC" returns NaN. This makes it MUCH easier to validate the input. Using parseFloat is just guessing that the input is in the correct format.

Andy E
  • 338,112
  • 86
  • 474
  • 445
  • That's for literals. Is `.parseFloat()` required to do the same? – Vilx- Jan 18 '10 at 10:38
  • 5
    Yes. `parseFloat("1,01")` in any locale will return `1`, not `1.01` – Andy E Jan 18 '10 at 10:41
  • 4
    Mozilla Developer Centre says: "parseFloat parses its argument, a string, and returns a floating point number. If it encounters a character other than a sign (+ or -), numeral (0-9), a decimal point, or an exponent, it returns the value up to that point and ignores that character and all succeeding characters. Leading and trailing spaces are allowed." https://developer.mozilla.org/en/Core_JavaScript_1.5_Reference/Global_Functions/parseFloat – Matt Sach Jan 18 '10 at 10:45
  • @Vilx- parseFloat looks for the first matching DecimalLiteral in the string that is passed as the argument. Anything else will return NaN. – Andy E Jan 18 '10 at 10:52
  • Except that it conveniently doesn't define "decimal point". XD – Vilx- Jan 18 '10 at 10:53
  • 1
    Because that's already defined in the ECMAScript 262 specs for DecimalLiteral :) – Matt Sach Jan 18 '10 at 10:55
  • @Vilx- It defines a DecimalLiteral, and the DecimalLiteral defines the decimal point :) – Andy E Jan 18 '10 at 10:56
  • OK, I'll take your word for it! :) – Vilx- Jan 18 '10 at 11:16
  • 1
    @Vilx- LOL it's not my word, it's ECMA International's word! – Andy E Jan 18 '10 at 11:17
  • 1
    I'll take your word because I'm too lazy to look up the spec myself. XD Thanks! :) – Vilx- Jan 18 '10 at 11:27
  • In case anyone wants to check the ECMA specs: https://www.ecma-international.org/ecma-262/10.0/index.html#prod-DecimalLiteral – ofri cofri Jun 03 '20 at 18:50
  • That last function doesn't cover all possible decimal separators - I've had issues before caused by people sending numbers with the Persian Decimal Separator ("momayyez") special unicode character, or a forward slash sometimes used in its place – deeBo Mar 03 '21 at 16:34
15

use:

theNumber.toLocaleString();

to get a properly formatted string with the right decimal and thousands separators

jspcal
  • 50,847
  • 7
  • 72
  • 76
  • 1
    OK. And how do I parse a string with these separators? – Vilx- Jan 18 '10 at 10:33
  • you'd need to strip out the seps, and make sure it conforms to something like: [\+\-]?([0-9]+|[0-9]*(\.([eE][\+\-]?[0-9]+))?) – jspcal Jan 18 '10 at 10:42
  • Fair enough, I can strip whatever is needed, and can replace the locale-specific separator with `.`. But is `.` the only thing that `.parseFloat()` expects? – Vilx- Jan 18 '10 at 10:45
  • See replies to Andy E's original answer (short answer: yes) – Matt Sach Jan 18 '10 at 10:49
  • I tried this in Google Chrome and do not see the "thousands separators" – jm. Aug 27 '12 at 18:32
2

As far as I'm aware, javascript itself only knows about the . separator for decimals. At least one person whose judgement I trust on JS things concurs:

http://www.merlyn.demon.co.uk/js-maths.htm#DTS

Matt Sach
  • 1,162
  • 16
  • 37
  • 2
    To add to this, see the old ECMAScript 262 specification on grammar wrt parsing numeric literals: http://bclary.com/2004/11/07/#a-7.8.3 – Anonymous Jan 18 '10 at 10:32
  • Ah yes, always good to have official documentation for this sort of thing, thanks :) – Matt Sach Jan 18 '10 at 10:35
  • Yes, that's for literals. What about parsing string input? – Vilx- Jan 18 '10 at 10:36
  • 1
    Take a fixed number such as 1234.56 and run it through the `.toLocaleString()` method mentioned by jspcal. Examine the result to work out what your decimal and thousand separators are. Run all numbers displayed to the user through a function which applies those separators (see the link I gave for examples of that). – Matt Sach Jan 18 '10 at 10:40