LanguageDetector: availability() static method

options

An object specifying configuration options for the LanguageDetector. Possible values include:

expectedInputLanguages

An array of strings specifying the expected languages of the input text to have its language detected. These should be valid BCP 47 language tags (as specified in RFC 5646). Defaults to ["en"]

A Promise that fulfills with an enumerated value indicating whether support is available (or will be available) for a given LanguageDetector configuration, or null if support could not be determined.

Possible values include:

available

The browser supports the given configuration and it can be used immediately.

downloadable

The browser supports the given configuration, but it first needs to download an AI model, or some fune-tuning data for the model.

downloading

The browser supports the given configuration, but it has to finish an ongoing download before it can proceed.

unavailable

The browser does not support the given configuration.

InvalidStateError DOMException

Thrown if the page's Document is not yet active.

OperationError DOMException

Thrown if initialization of the AI model failed for any reason.

UnknownError DOMException

Thrown if the availability() call failed for any other reason, or a reason the user agent did not wish to disclose.

If usage of the method is blocked by a language-detector Permissions-Policy, the promise rejects with a value of unavailable.

In the following snippet, we start by checking the availability of the model for detecting a couple of languages using the availability() method:

• If it returns unavailable, we print an appropriate error message to the console.
• If it returns available, we create a language detector using the create() method, passing it the expectedInputLanguages. The required AI model is available, so we can use it immediately.
• If it returns a different value (that is, downloadable or downloading), we run the same create() method call, but this time we include a monitor that logs the percentage of the model downloaded each time the downloadprogress event fires.

const availability = await LanguageDetector.availability({
  expectedInputLanguages: ["en-US", "ja"],
});
let detector;

if (availability === "unavailable") {
  console.log(`Detection not supported; try a different set of languages.`);
  return;
} else if (availability === "available") {
  detector = await LanguageDetector.create({
    expectedInputLanguages: ["en-US", "zh"],
  });
} else {
  detector = await LanguageDetector.create({
    expectedInputLanguages: ["en-US", "zh"],
    monitor: (monitor) => {
      monitor.addEventListener("downloadprogress", (e) => {
        console.log(`Downloaded ${Math.floor(e.loaded * 100)}%`);
      });
    },
  });
}

• Using the Translator and Language Detector APIs