Skip to content


Subversion checkout URL

You can clone with
Download ZIP
An implementation of the current ECMAScript 6 String.format() proposal (
branch: master

ECMAScript 6 String.format()

This is an ECMAScript 5 based implementation of the current proposal for String.format() in ECMAScript 6. You can see the proposal at:

It is compatible with both broser and Node.js environments.


The basic concept is that you can include identifiers inside curly brackets that will be replaced by additional data that you pass in. The identifiers can be numbers (indicating which argument to replace with) or string identifiers (indicating which property of an object to replace with). Here's an example:

"The car is made by {0} in {1}.".format("Nissan", 2009);
// returns "Tom's father has two sons, Tom and Jerry."

"The car is made by {brand} in {year}.".format({brand: "Nissan", year: 2009});
// returns "The car is made by Nissan in 2009."

Identifiers can also use dot or square bracket notation to dig deeper:

"The car is made by {car.brand} in {car[year]}.".format({car: {brand: "Nissan", year: 2009}});
// returns "The car is made by Nissan in 2009."

You can also specify formatting. "Format specifiers" denote how a value should be formatted and are included after a colon inside the curly brackets:

"This number has at at least 4 digits after the decimal point: {0:.4}".format(5);
// "This number has at at least 4 digits after the decimal point: 5.0000"

You can also use a value as a format specifier:

"This number has at at least 4 digits after the decimal point: {value:{format}}".format({value: 5, format: ".4"});
// "This number has at at least 4 digits after the decimal point: 5.0000"

Many format specifiers are available. See the proposal for a full list. (Note: not all are implemented yet)

Finally, you can provide custom formatting for you own objects by implementing a toFormat(specifier) method on them:

var password = {
  toString: function() { return "nyah-nyah!"; },
  toFormat: function (specifier) {
    return (specifier === "secret") ? "***" : this.toString();

"The password is: {0:secret}".format(password);

There are more nitty gritty details; check the proposal or the tests to see them all.


This code was really just written on a lark on day, but please feel free to use and suggest changes. Using the proposed feature in real-world situations is the best way to determine whether it effectively serves developers' needs.


This code was written by Rob Brackett in 2012 and is licensed under an MIT-style license. See the full text in LICENSE.txt.

Please share and use! This library's best value is as feedback to the TC39 committee on how well the String.format() proposal fits your needs.


This implementation is probably not perfect; the proposal is vague on some parts (it appears to be more about style and syntax than about all the nitty gritty details) and intends to follow standard printf conventions, which I am not an expert on when it comes to edge cases and uncommon scenarios. Feel free to file issues related to correctness.

I am also interested in using this as a tool for developers to provide feedback on the usefulness of the proposal. Are there features that seem unnecessary or not sensible? Features that are missing? Please file issues for those, too.

Other Notes

In implementing this, I ran across some interesting issues:

  • Why does the e/E type for number formatting not behave the same as Number.toExponent()? Specifically, the proposal requires the exponent to be at least two digits, while Number.toExponent() does not do this. The proposal also specifies a space between the base and the "e", which toExponential() also does not do. One would assume the two should be harmonious.
  • The proposal is not clear on how the e/E type for number formatting should behave on NaN and Infinity.
  • It is not clear whether precision should apply to the g/G type.
  • The default type/behavior for Number.toFormat() isn't clear. Is it an error to not specify one? If not, what's the default? This implementation currently defaults to "s".
  • It is not clear how the f/F type should behave in the absence of a precision. Is it an error or does it default to some value?
  • The proposal mentions both "inf/INF" and "infinity/Infinity" for the f/F type. Not when each would be used.
  • Not clear on how g/G differs from the s type.
  • The "#" alternate flag seems a bit odd with exponential notation. (Note it's not even allowed in Python's format(), which the proposal takes cues from.)
  • The "#" alternate flag adds prefixes for octal and hexadecimal, but not for binary, which seems odd.
Something went wrong with that request. Please try again.