Translating mnemonic files

From Apertium
Revision as of 12:16, 15 June 2018 by Bech (talk | contribs) (update : é allowed formats for the right part of mnemonic files)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

En français

Mnemonic files are an alternative to .po files to generate multilingual user interfaces.

Except for the possible comments (starting with #) at the beginning of a file, the mnemonic files is made of lines with the following format:

Mnemonic             "formatted_string_for_printf"
              or
Mnemonic             character_string

In these lines, the mnemonic (starting on the first column) must be kept unchanged from a language to another, while the string farther to the right is translated.

However, a string to be translated can contain the sequences \r or \n or other similar ones which, not only should not be translated, but must not disturb the translation of the following word when it touches the letter having a special meaning.

Similarly, indications of format as %s, %3d, %4.2f ... are kept unchanged.

Mnemonic files format is not directly carried out by Apertium package. It is necessary to recover the package apertium-c-formatters to have a deformatter which takes into account specificities of this format.

Getting and installing apertium-c-formatters package

If you did not already recover apertium-c-formatters package for translating man pages, let start by making the following operations:

Download apertium-c-formatters from GIT:

git clone --depth 1 https://github.com/apertium/apertium-c-formatters

Compile the source files:

make

Yes, compilation is very simple, and fast!

Install the tools:

make install

The above command assumes you have write access to /usr/local/bin and /usr/local/share/man. If not, enter the command:

sudo make install

By defaut, as for lttoolbox, apertium, and the language pairs, the installation is done in /usr/local/bin and /usr/local/share. If you wish to change the installation directory, you will have to change the first line of the makefile.

install_dir=/usr/local

will be replaced by the parent directory where the executable commands will be. For example, if you put:

install_dir=/usr

tools will be installed in /usr/bin and man pages in /usr/share/man .

Using

Available tools

Available tools to translate the mnemonic files are the following:

  • desmnemo : deformatter for mnemonic files,
  • remnemo : reformatter for mnemonic files,
  • apertium-mnemo : a shell intended to make the translation of mnemonic files easier.

Translation by calling the various tools

Assuming we are in the directory samples-mnemo of the package apertium-c-formatter and we want to translate into Spanish the file mess-libremail.en We can type:

cat mess-libremail.en | desmnemo | apertium -f none en-es | remnemo > mess-libremail.es-utf

If the source language was not English, we could first notice that we translate a mnemonic file using UTF-8 charset. If the file format is encoded ISO-8859-n (or any other character set), we need to convert the data into UTF-8 to make them compatible with Apertium, which will also provide a result UTF-8 encoded.

In addition, the command apertium is executed with the -f none option, which is necessary for the deformatter apertium-destxt not to be used.

We can also give file names as parameters of the deformatter and the reformatter:

desmnemo mess-libremail.en | apertium -f none en-es | remnemo - mess-libremail.es-utf

We can notice the dash - as the first parameter of remnemo . Reformatters usually use the standard input (in this case the result produced by the command apertium) whereas it can be useful to preserve the result of reformatting in a file. But the problem is the general syntax of apertium reformatters require to specify the output file as the 2nd parameter. The - as the first parameter permit to overcome this problem.

More simple with the command apertium-mnemo

When a data format is integrated directly into the apertium command, there is the -f option to translate data produced in this format without having to call "by hand" a deformatter and a reformatter. The command apertium-mnemo is a shell which permits to do in the same way.

For example, to translate into Esperanto the file mess-libremail.en , we can type:

cat mess-libremail.en | apertium-mnemo en-eo > mess-libremail.es-utf

or even better:

apertium-mnemo en-eo mess-libremail.en mess-libremail.es-utf

The options -u and -d datadir of the command apertium are caried out by apertium-mnemo. You just have place them before the parameter indicating the translation direction (as for the command apertium).