Working with languages

Delivering content in a way that users can understand is vital in providing a good UX. Since a bot could focus on an audience that speaks a particular language, botogram provides an i18n platform that allows you to translate its default messages, such as /help‘s one, to another language.

Available translations

The following languages are currently included in the botogram package:

  • br (Brazilian Portuguese)
  • en (English)
  • it (Italian)

Setting your bot’s language

Botogram bots currently only support using a single language for their messages. This also means that it is not possible at the moment to translate messages on a per-user basis.

While bots will use the English translation by default, it is possible to change the language in use by changing the bot’s botogram.Bot.lang property to the target language’s code. Please note that the selected language must be supported by the botogram version you are using.

bot.lang = "it"

After doing this, the bot will start using the translated messages included in the package. If a message hasn’t been translated to the selected language, the bot will fall back on the English default.

Overriding default messages

New in version 0.5.

As described in Translating botogram to a new language, new and updated translations’ availability is limited to new botogram releases, meaning that it could take some time for them to reach end users.

Packaged translations may also not always fit a specific use case, making it necessary for you to edit some of the messages. While you could use a custom build of the package with a modified translation, it is also possible to programmatically override the translation of single messages through botogram.Bot.override_i18n, a dictionary that works basically the same way as a .po file, associating msgid‘s to msgstr‘s:

bot.override_i18n = {
     "Use /help to get a list of all the commands.": \
     "Utilizza /help per ottenere la lista di tutti i comandi."
 }

We’ll go more in depth on the translation format in the next section.

Translating botogram to a new language

If your language isn’t yet supported by botogram, you can contribute your own translation by forking the project’s git repository and opening a new pull request. See Living on the edge for instructions on how to clone the repository and install the required dependencies. It is recommended that you don’t install this bleeding edge clone as a global package: in fact, you can completely avoid installing it, while building and testing it in a local virtual environment may be useful in order to catch errors.

Botogram handles i18n via GNU gettext, which stores translations in plain-text .po files that are then compiled while installing the package. You can find the all the translations that are currently included in the package in the i18n/ directory.

You can generate a new language file with the following command:

$ invoke i18n-new <code>

where code is the ISO 639-1 code assigned the language you are translating to. This will create a new language file located at i18n/langs/<code>.po. The first few line will look like this:

msgid ""
msgstr ""
"Project-Id-Version: botogram 1.0.dev0\n"
"Report-Msgid-Bugs-To: https://github.com/pietroalbini/botogram/issues\n"
"POT-Creation-Date: 2017-10-06 19:21+0200\n"
"PO-Revision-Date: 2017-10-11 15:02+0200\n"
"Last-Translator: FULL NAME <EMAIL@ADDRESS>\n"
"Language: de\n"
"Language-Team: de <LL@li.org>\n"
"Plural-Forms: nplurals=2; plural=(n != 1)\n"
"MIME-Version: 1.0\n"
"Content-Type: text/plain; charset=utf-8\n"
"Content-Transfer-Encoding: 8bit\n"
"Generated-By: Babel 2.3.4\n"

The first thing you should do is to fill out the Last-Translator field with your contact information. You may also want to do the same with the copyright notice at the top of the document.

The remainder of the file is were translations are actually defined:

#: botogram/defaults.py:46
msgid "Use /help to get a list of all the commands."
msgstr ""

Each message is assigned a msgid string which identifies it across translations: in botogram it is the English translation for that message. msgstr fields are instead specific to each translation and define that message’s translation for the file’s language: this is where you need to enter your translation. If a msgstr is empty (as they are by default) botogram will default to the English translation.

Some messages could contain HTML formatting or Python string interpolation: your translation should reflect these as closely as possible. If you need context on the usage of a message, you can refer to its usages in the source code included in the comment line above each string.

Just to be sure your syntax is correct, you can ensure your translation will compile correctly by invoking

$ invoke i18n-compile

If the command succeeds there’s good chance you didn’t mess up anything.

Once you’re done, you can commit and push your changes to your fork and propose them to be merged into the upstream repository to be included in the next botogram release.

Updating a translation

As botogram evolves, more message will probably be added to the codebase, and it is also possible for currently included translations to contain mistakes.

The workflow for updating a translation is basically the same as the one described in Translating botogram to a new language, but you may also need to use

$ invoke i18n-extract

to extract new messages from the codebase. The command also ensures references in comments are up-to-date with their current location.

Running the command will always result in the .pot file and .po files being updated, at least for what concerns the POT-Creation-Date header. You should check your diff and avoid committing any change that doesn’t impact the actual translation and the source code references in comments.

While trivial, it would be nice if you also changed the PO-Revision-Date header to reflect your changes.