CODEWITHSUNDEEP
node.jsmongodbmongoosefloating-pointcurrency
Victor
Victor

Reputation: 8490

Mongoose price type

What the best way to store monetary values on Mongoose?

I know in Java there is the type BigDecimal to handle all monetary values to avoid this problem.

There is any type in mongo or javascript to avoid it?

Upvotes: 1

Views: 5486

Answers (3)

Félix Brunet
Félix Brunet

Reputation: 2173

Mongo data is stored as BSON (Binary JSON).

MongoDB adds some type on JSON. The complete list of these types can be found here, but there is no type specifically for money.

In short, for price, you can choose between storing it as a Number (a Double, to be precise) or as a String. For most use cases, Number should be more than enough.

If you need to deal with multiple currencies, I would recommend adding a separate field for storing the currency.

Upvotes: 0

Mikhail
Mikhail

Reputation: 900

Storing it as a Number is fine, the issue is when you need to retrieve these values and perform operations.

Given the imprecision of floating-point math, I use this BigNumber library (https://www.npmjs.com/package/bignumber.js) to safely perform operations with floats.

Upvotes: 1

Eric Postpischil
Eric Postpischil

Reputation: 224335

Unless compelled for some reason, you want a type that exactly represents the values you use. This means you should:

  • Avoid any binary floating-point format if you will be storing numbers that have decimal fractions, such as storing one dollar and twenty-three cents as 1.23.
  • Ensure the format can represent numbers as large as you need.

If those are satisfied, you can use whatever format suits your purposes (e.g., is small in size, is built into the language, or is easy to convert between BSON and the format you use it in). One way to use binary floating-point to represent currency exactly is to use whole numbers of a unit, such as integer numbers of pennies instead of fractional numbers of dollars.

Note that merely storing and retrieving amounts of currency is fairly simple, and it can be done with integer formats or floating-point formats as long as you follow the guidelines above. However, calculating with currency cannot be done easily with either integer or floating-point arithmetic except for very simple calculations such as additions, subtractions, and multiplications by integers. For example, neither integer arithmetic nor binary floating-point arithmetic can exactly represent the unit price of an item sold three-for-a-dollar, nor can they can they generally calculate interest or percentage discounts without rounding errors. To handle such calculations, selecting a format is not enough; you must design the operations carefully.

Upvotes: 1

Related Questions