Enabling your ad network to work with translation-related Google Search features

Google Search offers severaltranslation-related features that enable users to access translated content. If you run an ad network and your ads aren't working properly on translated web pages, you'll need to follow the steps in this guide to make sure your ads render or attribute correctly.

Our approach

When users access translated content provided byGoogle Translatefrom within search results, Google retrieves the page from the publisher, rewrites the source URL, and translates the web page after the user clicks the translated result.

Convert the Google Translate URL to the original URL

If you run an ad network that relies on the publisher’s source URL, you'll need to convert the Google Translate URL to make sure your ads are working properly. Follow these steps to decode the publisher’s hostname:

1. Extract the domain prefix from the hostname, by removing the.translate.googsuffix.
2. Split the_x_tr_encparameter by the,(comma) character and save it asencoding_list.
3. Prepend the value of the_x_tr_hpparameter to the domain prefix, if it exists.
4. Ifencoding_listcontains1and the output begins with1-,remove the1-prefix from the output of step 2.
5. Ifencoding_listcontains0and the output begins with0-,remove the0-prefix from the output of step 3. If you removed the prefix, setis_idntotrue.Otherwise, setis_idntofalse.
6. Replace/\b-\b/(regex) with the.(dot) character.
7. Replace the--(double hyphen) character with the-(hyphen) character.
8. Ifis_idnis set totrue,add the punycode prefixxn--.
9. Optional:Convert to Unicode.

Sample JavaScript code for decoding the hostname from a Google Translate URL

function decodeHostname(proxyUrl) {
const parsedProxyUrl = new URL(proxyUrl);
const fullHost = parsedProxyUrl.hostname;
// 1. Extract the domain prefix from the hostname, by removing the
".translate.goog" suffix
let domainPrefix = fullHost.substring(0, fullHost.indexOf('.'));

// 2. Split _x_tr_enc parameter by "," (comma), save as encodingList
const encodingList = parsedProxyUrl.searchParams.has('_x_tr_enc')?
parsedProxyUrl.searchParams.get('_x_tr_enc').split(','):
[];

// 3. Prepend value of _x_tr_hp parameter to the domain prefix, if it exists
if (parsedProxyUrl.searchParams.has('_x_tr_hp')) {
domainPrefix = parsedProxyUrl.searchParams.get('_x_tr_hp') + domainPrefix;
}

// 4. Remove '1-' prefix from the output of step 2 if encodingList contains
// '1' and the output begins with '1-'.
if (encodingList.includes('1') && domainPrefix.startsWith('1-')) {
domainPrefix = domainPrefix.substring(2);
}

// 5. Remove '0-' prefix from the output of step 3 if encodingList contains
// '0' and the output begins with '0-'.
// Set isIdn to true if removed, false otherwise.
let isIdn = false;
if (encodingList.includes('0') && domainPrefix.startsWith('0-')) {
isIdn = true;
domainPrefix = domainPrefix.substring(2);
}

// 6. Replace /\b-\b/ (regex) with '.' (dot) character.
// 7. Replace '--' (double hyphen) with '-' (hyphen).
let decodedSegment =
domainPrefix.replaceAll(/\b-\b/g, '.').replaceAll('--', '-');

// 8. If isIdn equals true, add the punycode prefix 'xn--'.
if (isIdn) {
decodedSegment = 'xn--' + decodedSegment;
}
return decodedSegment;
}

Reconstruct the URL

1. Using the original page URL, replace the hostname with the decoded hostname.
2. Remove all_x_tr_*parameters.

Test your code

You can create unit tests for your code using the following table. Given aproxyUrl, thedecodeHostnamemust match the expected value.

The following table can only be used to test the hostname decoding. You’ll need to ensure that the path, fragment, and original parameters of the URL are preserved as is.