Number and currency patterns
Number formats introduction
Pre-requisite topics to read:
Numbers are formatted using patterns, like "#,###.00". For example, the pattern "#,###.00" when used to format the number 12345.678 could result in "12'345,67" if the grouping separator for your locale use an apostrophe ('), and the decimal separator is a comma (,).
Whenever any of these symbols are in the English pattern, they must be retained in the pattern for your locale. The positions of some of them (%, ¤) may be changed, or spaces added or removed. The symbols will be replaced by the local equivalents, using the Number Symbols for your language. (See 💡 Helpful Tips below the table)
💡 Translation Tips
If your locale grouping separation is not by thousands, then move the comma separator in the format to the correct placement.
For example, if your language expects to see the separator in the first group of 3 and then 2s thereafter: 1,00,00,000.00, the pattern should reflect that: #,##,##0.###.
DO NOT "translate" the placeholder characters #,###.00. For example:
Do not change '.' to ',' even though in your language the decimal point is written with a comma.
Do not change the zero (0) or hash marks (#) after a decimal point (for example, #,##0.###); those will be reset by computer programs.
Even if your currency doesn't use any decimal points, do not remove the decimal indicators (for example, ¤#,##0.00).
Negative patterns can be handled two ways:
If the negative pattern is the same as the positive pattern with a minus sign in front (e.g. -1,234,00), then that negative pattern will be automatically generated.
You don't need to supply a negative pattern in that case, and if you do supply one it will be automatically removed.
If your locale uses a different way to express the negative number (other than a negative sign in front), then use the ";" to provide two patterns. The semicolon after the positive pattern indicates a separate format for the negative pattern.
For example, in an Accounting currency format, if your locale uses ( ) to express the negative number, specify the negative number with ";", such as #,##0.00¤;(#,##0.00¤). This example format indicates that the negative currencies appear like "(1'234,56£)" instead of "-1'234,56£".
For example, if the position of the negative sign is before the % symbol, you'd specify: #,##0%;#,##0-%.
Move the the currency symbol (¤) or percent sign (%), if it is used in a different position.
To deal with CLDR's default automatic space handling in place for currency symbol when using a currency code(e.g. USD):
Do NOT add a space ¤#,##0.### for result: $12 and USD 12.
ADD a manual space ¤ #,##0.### for result: $ 12 and USD 12.
Always verify with examples in the right information pane and see how the data from number symbols are used in formatting numbers. The final results of the number formatting will show the correct symbols for decimal, grouping, etc... from Number Symbols in your locale.
For bidi scripts (e.g. Arabic and Hebrew) you may need to add directionality markers (U+200E (<LRM> LEFT-TO-RIGHT MARK, U+200F <RLM> RIGHT-TO-LEFT MARK, U+061C <ALM> ARABIC LETTER MARK)
For number formats in bidi scripts, the Survey Tool shows examples in both a right-to-left context and a neutral context (with both positive and negative numeric values). In the future it may show examples in a left-to-right contex as well.
The hightest priority is to make the formats look correct in a right-to-left context; the next most important is to make them look correct in a a neutral cotext. If possible it is also a good idea to make them work in a left-to-right context.
To make them work in different contexts, you can add the directionality markers specified above. As a last resort you can also use the bidi isolate markers U+2066 <LRI> LEFT-TO-RIGHT ISOLATE, U+2067 <RLI> RIGHT-TO-LEFT ISOLATE, and U+2068 <FSI> FIRST STRONG ISOLATE, each of which must be paired with a terminating U+2069 <PDI> POP DIRECTIONAL ISOLATE. However, these bidi isolate markers can result in copy-paste issues if the PDI is not copied along with the other paired mark.
When adding directionaliry markers to formats, keep in mind that the minus and plus signs may also have directionality markers that affect layout of patterns with those signs.
Types of Number Patterns
There are four general-purpose number patterns supported in CLDR: decimal, currency, percent, and scientific.
Remember that the "." and "," are placeholders and do not mean the literal characters: computer programs that use CLDR will change them to the right number symbols for your language as explained in the section above.
The currency formats have two types (standard and accounting); each of these may have up to two additional alternate forms (-alphaNextToNumber, -noCurrency) as descibed in the table below.
Miscellaneous Number Patterns
Compact decimal formatting
There are also patterns for compact forms of numbers, such as the such as "1M" and "1 million". These patterns may be substantially different across languages, as shown by comparing the English patterns to the Japanese patterns in the table below. These patterns also have plural categories.
The compact currency formats can have an -alphaNextToNumber variant, as with the regular currency formats. Ther eis usually no need to provide a -noCurrency variant, since if that is not present the compact decimal formats will be used. However if those are not equivalent to what the noCurrency variant would be, then a separate -noCurrency variant of the ompact currency formats may be provided.
When computer programs use CLDR, the number of decimals can be changed by computer programs according to the task its designed for. For example, the pattern for 10,000 in the table below (00K for English, 0万 for Japanese) may be modified to have more or fewer decimals — it could be changed to have 3 digits of accuracy: 00.0K for English, 0.00万 for Japanese.
💡 Translation Tips
If it is impossible to abbreviate numbers in that way (e.g. 1M) in your language, then use a single 0 (zero digit). In Compact number formatting, "0" is special character indicating to use the standard decimal formatting/currency formatting. "0" should be used for decimal and currency when there's NO compact/abbreviated way to express the number. For example, in Basque, compact number expression starts with 1M; therefore, all compact numbers up to 1M are expressed as "0".
For example, the Japanese pattern for 1000 uses a zero digit because 1,000 is expressed with the character 千. This should be avoided wherever possible, however, since it may cause display issues on devices with small screens, such as mobile phones.
In some languages the text will inflect according to the number. For example, languages that write "1 million" but "2 millions".
The patterns do not allow for gender or case inflection, so use the most neutral form.
Remember that these are number patterns, so literal characters like a period (".") must occur within quotation marks if they are not the placeholder standing for the decimal period. See Pattern Characters.
Compact decimal patterns do not require a "0" when there is a single word or a symbol that is unique. For example, in Hebrew and Somali, 4-digit-short-one gets formatted to a single word or symbol, without an extra digit. In a Somali example,1000 would be formatted to "Kun" and not "0 Kun" (in Somali, you would express 1000 as "Kun" not "1 Kun").
However, this can only be done where the number is unique. For example, in French 4-digit-short-one (1000) is used to format not only 1K, but also 1.1K. If you supply a pattern with no digits, the user can't tell what the number is suppose to be. An error will occur if you try to do this.
There is a mechanism for special cases of exact matches. A special row can be added for your language for specific numbers, such as exactly 1, where the digits must be omitted. You would see a Code value of 4-digit-short-one, for example. If your language has such cases, please file a ticket to request adding them.
Different patterns are required for currency and non-currency. Please add both.
Plural Forms of Numbers
Some languages have multiple plural forms for numbers, and will have multiple plural categories for each number. See Plural Rules and Tool for numbers mapped to the various plural categories.