Text ID based translations

The text ID translation mechanism is an industrial strength system for internationalization and localization. Each text in the application has a unique identifier (text ID) that you use in the source code instead of text. This makes it much easier to manage large numbers of translated texts.

Internationalizing with text IDs

When using text IDs instead of plain text, the general method of internationalizing an application is the same but the details are a bit different:

1. The functions and macros for the text-ID-based translation system are different from the plain-text system. You use the qsTrId() function instead of qsTr(), the QT_TRID_NOOP() macro instead of QT_TR_NOOP(), and QT_TRID_N_NOOP() macro instead of QT_TR_N_NOOP()).
2. Use text IDs as user interface strings rather than plain text strings. For example, qsTrId("id-back-not-front")
3. You cannot specify a context parameter with a text ID, and therefore identically spelled words with different meanings need separate text IDs. For example, qsTrId("id-back-backstep") differentiates the back-step Back from the id-back-not-front Back.
4. Since context names are not allowed for text-ID-based translations, Qt Linguist lists the IDs in a file without context names.
5. The engineering English text that you see in the user interface for development builds is indicated with a //% comment. If you do not include this, the text ID is shown in the user interface. This is especially important when you have texts with parameters. The //% comment needs to include the parameters indicators in the string. For example, //% "Number of files: %1"
6. The //: comments that provide extra information to the translator are optional in the plain-text system. However, with the text-ID-based system, this extra information becomes essential because without it you only have the text ID and the translator might not be able to make a sensible translation from that without further context. You can use long descriptive text IDs and no comments, but comments are often easier to understand.

The side-by-side code snippets below show a comparison of text-ID -based and plain-text-based translations:

Text {
    id: backTxt;
    //: The back of the object, not the front
    //% "Back"
    //~ Context Not related to back-stepping
    text: qsTrId("id-back-not-front");
}

Text {
    id: backTxt;
    //: The back of the object, not the front
    //~ Context Not related to back-stepping
    text: qsTr("Back","Not front")
}

Localizing with text IDs

Localizing with text IDs follows much the same process as for plain text.

You use the lupdate tool to generate the TS files where you add the translations. The source values in the translation files will be text IDs rather than plain text, and therefore you need either descriptive text IDs or good additional comments, or both to ensure that the translations are accurate.

The example text-ID-based user interface text from above results in the following content in the .ts file:

<message id="id-back-not-front">
    <source>Back</source>
    <extracomment>The back of the object, not the front</extracomment>
    <translation type="unfinished"></translation>
    <extra-Context>Not related to back-stepping</extra-Context>
</message>

When using lrelease, you need to specify that the keys for translated texts are based on text IDs, rather than plain text. If you use qsTr() to mark the strings as translatable in the code, the id attribute is not set and lrelease ignores them.

However, if there is no translation available for a given text (which is generally the case until late in development), the text ID will be shown in the user interface rather than a proper text. In order to make the application more usable for testing, you can make lrelease use the Engineering English source text (from the //% comments) as the translated text and mark it with some indicator, such as an exclamation mark (!), so you can see texts that are not yet translated.

CMake configuration

When building with CMake, use the prefix qml_ for .ts files. For example, qml_en.ts. In the CMakeLists.txt file, add the qt_add_translations function, where you list the *.ts files as values of TS_FILES, set the value of RESOURCE_PREFIX to the URI of the main.qml file for the project followed by /i18n, and set the value of LRELEASE_OPTIONS to -idbased:

qt_add_translations(${CMAKE_PROJECT_NAME}
    TS_FILES i18n/qml_de_DE.ts i18n/qml_en_US.ts
    RESOURCE_PREFIX Main/i18n
    LRELEASE_OPTIONS -idbased
)

Advanced use with qmake

For projects that target a large number of locales, you can remove the TRANSLATIONS info from the .pro file and, instead, manage the translations with a separate script. The script can call lrelease and lupdate for each of the desired targets.

The updates could be scripted something like this:

lupdate -recursive <project-dir> -ts <project-dir>/i18n/myapp-text_en_GB.ts
lupdate -recursive <project-dir> -ts <project-dir>/i18n/myapp-text_en_US.ts
...

The generation of the final .qm files could be scripted something like this:

lrelease -idbased <project-dir>/i18n/myapp-text_en_GB.ts
lrelease -idbased <project-dir>/i18n/myapp-text_en_US.ts
...