Translation Memory

Pootle provides suggested translations to the current string. Translator can use these suggestions as their translation or to aid their translation.

Suggestions are based on previous translations of similar strings. These Translation Memory (TM) matches mean that you can speed up your translation and ensure consistency across your work.

Using Translation Memory

Translation Memory suggestions are automatically retrieved when you enter a new translation unit. These are displayed below the editing widget. You can insert a TM suggestion by clicking on the suggestion row.

The differences between the current string and the suggested string are highlighted, this allows you to see how the two differ and helps you make changes to the suggestion to make it work as the current translation.

Configuring Translation Memory

Translation Memory will work out of the box with a default Pootle installation. There are two methods of getting Translation Memory.

  1. Amagama - for remote Translation Memory
  2. Elasticsearch - for local Translation Memory

Amagama based remote TM

By default Pootle will query Translate’s Amagama Translation Memory server, which hosts translations of an extensive collection of Opensource software.

If you want to setup and connect to your own TM server then the AMAGAMA_URL will allow you to point to a private TM server.

Elasticsearch based local TM

New in version 2.7.

By default this is configured to operate if local TM is available. To use it you will need to install Elasticsearch and populate the initial translation memory using the update_tmserver command:

(env) $ pootle update_tmserver --rebuild

Once populated Pootle will keep Local TM up-to-date.


Note that Elasticsearch depends on Java.

Local TM settings can be adjusted in POOTLE_TM_SERVER.

You may want to disable Amagama by setting AMAGAMA_URL to '' if you are using Elasticsearch local TM, though the two will operate together.