Attention

Version 3 is now the current version of MathJax. This document is for version 2.

Localization Strings

In MathJax v2.2 and later, the user interface can be localized to use languages other than English. This includes all information strings, menu items, warning messages, and so on. To make this possible, each string is given an ID that is used to obtain the localized version of the string. So the “File not found” message might have the ID NotFound. The localization data for each language associates the ID NotFound with the proper translation of the English phrase “File not found”.

Some of MathJax’s functions, like MathJax.Message.Set(), can accept localized strings as their parameters. To use a localized string in this case, use an array consisting of the ID followed by the English string (followed by any substitution arguments for the string, see below). For example,

MathJax.Message.Set(["NotFound","File not found"]);

would cause the “File not found” message to be displayed (in the currently selected language) in the MathJax message area at the bottom of left of the browser window. Note that MathJax.Message.Set() can still accept unlocalized strings, as it has traditionally:

MathJax.Message.Set("File not found");

Here the message will always be in English, regardless of the selected language.

The reason that the English string is also given (in addition to the ID), is because MathJax needs to have a fallback string to use in case the localization data doesn’t translate that ID, or if the localization data has failed to load. Providing the English string in addition to the ID guarantees that a fallback is available.

MathJax’s localization system is documented more fully in the Localization API documentation.

Parameter Substitution

Localized messages may need to include information, like file names, that are not known until the message is needed. In this case, the message string acts as a template and MathJax will insert the needed values into it at the appropriate places. To use such substitutions, you include the values in the localization string array following the English phrase, and use the special sequences %1, %2, etc., to refer to the parameters at the locations where they should appear in the message. For example,

MathJax.Message.Set(["NotFound","File %1 not found",filename]);

would cause the name stored in the variable filename to be inserted into the localized string at the location of the %1. Note that the localized string could use the parameters in a different order from how they appear in English, so MathJax can handle languages where the word order is different.

Although it would be rare to need more than 9 substitution parameters, you can use %10, %11, etc., to get the 10-th, 11-th, and so on. If you need a parameter to be followed directly by a number, use %{1}0 (rather than %10) to get the first parameter followed directly by a zero.

A % followed by anything other than a number or a { generates just the character following the percent sign, so %% would produce a single %, and %: would produce just :.

Plural forms

Some languages handle plural forms differently from how English does. In English, there are two forms: the one used for a single item, and the one used for everything else. For example, you would say “You have one new message” for a single message, but “You have three new messages” of there were three messages (or two, or zero, or anything other than one).

To handle plurals, you use a special “plural” directive within your message string. The format is

%{plural:%n|text1|text2}

where %n is the reference to the parameter (which should be a number) that controls which text to use. For English, there would be two texts, one (text1) for when the number is 1, and one (text2) for when it is anything else. Other languages may have more forms (e.g., Welsh has six different plural forms, a different one for 0 (zero), 1 (one), 2 (two), 3 (few), 6 (many), and anything else, so Welsh translation plural forms would have six different texts). The details of how to map the numeric value to the text strings is handled by the translation data for the selected language.

As an example, you might use

MathJax.Message.Set(["NewMessages","You have %1 new %{plural:%1|message|messages}",n]);

where n is a variable holding the number of new messages.

Alternatively,

MathJax.Message.Set(["NewMessages","You have %{plural:%1|a new message|%1 new messages}",n]);

shows how you can do substitution within the plural texts themselves.

Note that the translation string may contain such constructs even if the original English one doesn’t. For example

MathJax.Message.Set(["alone","We are %1 in this family but alone in this World.",n]);

could be translated into French by

"Nous sommes %1 dans cette famille mais %{plural:%1|seul|seuls} en ce monde."

Note that if one of the options for the plural forms requires a literal close brace, it can be quoted with a percent. For instance,

%{plural:%1|One {only%}|Two {or more%}}

would produce One {only} when the first argument is 1, and Two {or more} otherwise.

If a message needs to include a literal string that looks like one of these selectors, the original % can be quoted. So %%{plural:%%1|A|B} would be the literal string %{plural:%1|A|B}.

Number forms

Decimal numbers are represented differently in different languages. For example, 3.14159 is an English representation of an approximation to the mathematical constant pi, while in European countries, it would be written 3,14159. MathJax will convert a number to the proper format before inserting it into a localized string. For example

MathJax.Message.Set(["pi","The value of pi is approximately %1",3.14159]);

would show the value as 3.14159 in English, but 3,14159 if French is the selected language.

ID’s and Domains

Because MathJax consists of a number of separate components and can be extended by third party code, it is possible for two different components to want to use the same ID value for a string, leading to an ID name collision. To help avoid this, MathJax allows identifier domains that are used to isolate collections of identifiers for one component from those for another component. For example, each input jax has its own domain, as do many of the MathJax extensions. This means you only have to worry about collisions within your own domain, and so can more easily manage the uniqueness if the ID’s in use.

To use a domain with your ID, replace the ID with an array consisting of the domain and the ID. For example, the TeX input jax uses the domain TeX, so

MathJax.Message.Set([["TeX","MissingBrace"],"Missing Close Brace"]);

would set the message to the translation associated with the ID MissingBrace in the TeX domain.

Some functions that take localization strings automatically prepend the domain to the ID (if one isn’t already given). For example, the Error() function of the TeX input jax uses the TeX domain if one isn’t supplied, so

TEX.Error(["MissingBrace","Missing Close Brace"]);

will generate the MissingBrace error from the TeX domain without having to specify the TeX domain explicitly.

HTML Snippets

MathJax provides a means of specifiying HTML code in javascript called HTML snippets. These frequently include text that needs to be localized, so you can include localization strings (like those described above) within an HTML snippet in any location where you would normally have a regular text string. For example, the snippet

[
  "Follow this link: ",
  ["a",{href:"http://www.mathjax.org"},[
    ["img",{src:"external.gif"}]
  ]]
]

includes the text “Follow this link:” which should be localized. You can change it to a localization string to cause it to be translated to the selected langauge:

[
  ["FollowLink","Follow this link"],": ",
  ["a",{href:"http://www.mathjax.org"},[
    ["img",{src:"external.gif"}]
  ]]
]

(Here we use the ID FollowLink to obtain the translation). Note that you can include substitution parameters as usual:

[
  ["ClickMessages","Click for %1 new %{plural:%1|messsage|messages}",n],": ",
  ["a",{href:"messages.html"},[
    ["img",{src:"external.gif"}]
  ]]
]

It is even possible to substitute HTML snippets into a localized string (when it is within an HTML snippet):

[
  ["MathJaxLink","This is documented at the %1 website",[
    ["a",{href:"http://docs.mathjax.org"},["MathJax]]
  ]]
]

Note, however, that a better approach to this exampe is given in the next section.

Since an HTML snippet might contain several strings that need to be localized, you may want to be able to specify the domain to use for all the strings within the snippet. Within a snippet, you can use an entry of the form [domain,snippet] to force the snippet to be processed with default domain domain. E.g.

[
  ["TeX",[
    ["ul",{},[
      ["li",{},[["MissingBrace","Missing close brace"]]],
      ["li",{},[["ExtraBrace","Extra close brace"]]],
    ]]
  ]],
  ["MathML",[
    ["ul",{},[
      ["li",{},[["UnknownNode","Unknown node type: %1",type]]],
      ["li",{},[["BadAttribute","Illegal attribute: %1",attr]]],
    ]]
  ]
]

would create two undordered lists, one with translations from the TeX domain, and one from the MathML domain.

To summarize the format of an HTML snippet, it is an array with each entry being one of the following:

  • A text string, which becomes text in the resulting HTML; this is untranslated.
  • An array of the form ["tag"], ["tag",{properties}], or ["tag",{properties},HTML-snippet], which becomes the given HTML tag, with the given properties, containing the given HTML-snippet as its children.
  • An array of the form [id,message] or [id,message,parameters], which is first translated, then parameter substitution performed, and the result added to the HTML (either as text or as HTML tags if the message included Markdown syntax). Note that the id can be either an id or an array [domain,id], and that the parameters could be HTML snippets themselves.
  • An array of the form [domain,HTML-snippet], which becomes the HTML-snippet with its localizations done from the given domain.

Markdown Notation

HTML snippets allow you to create styled markup, like bold or italics, but this requires breaking the text up into smaller strings that fall in between HTML tags. That makes it hard to translate, since the strings are not full phrases. To make the creation of strings with bold, italics, and hyperlinks easier to localize, MathJax allows the strings within HTML snippets to be written in a limited Markdown syntax (very limited). You can use *bold*, **italics**, ***bold-italics***, [link-text](url), and `code` to obtain bold, italics, bold-italics, hyperlinks, and code blocks. For instance, the link example above could be more easily handled via

[
  ["MathJaxLink","This is documented at the [MathJax](%1) website",
                 "http://docs.mathjax.org"]
]

while

[
  ["Renderer","*Renderer*: lets you select the output renderer"]
]

will produce the equivalent of <b>Renderer</b>: lets you select the output render in the appropriate language.