My Problems with OmegaT

From B-Ob8ungen
Jump to navigation Jump to search

In the beginning I was not aware that OmegaT is installed with the interface your system is running with. In my case windows xp had been installed with German as the default language, based in Germany. Since there was a German translation I automatically got a German interface. Later I discovered that parts of the interface remained in English.

View should be "Ansicht" in German
All elements in "View" remain in English
Expressions partly misleading

This did not bother me or better, I would have preferred to have the interface completely in English. My first idea was to create another user on my computer and place this person to Great Britain using British English. Again later I discovered that the same effect can be achieved in an easier way:

Change the OmegaT.l4J.ini file

You can edit the OmegaT.l4J.ini file with a simple text editor (Notepad or even PhpDesigner2005 will do). Simply remove the '#' before the '-' and state for instance

# Language 
-Duser.language=EN 

and the same for

# Country 
-Duser.country=GB

Segmentation Rules

The first documents I translated with OmegaT had Turkish as the source language and English or German as the target language. The source texts had ordinal numbers spelled with a dot at the end and names only given by the initials, followed by a full stop. Accordingly not whole sentences, but parts of sentences (until the first full stop) were segmented. In the Help section (User's Manual) I was told

In order to create an empty set of rules for a new language pattern, click Add in the upper half of the dialog. An empty line will appear at the bottom of the upper table.

When the window opened I saw the upper pane with default languages and a few other rules. The only button to click on said OK not ADD.

Window for segmentation rules in German interface
Add a new language
Add a new rule
click on image to enlarge click on image to enlarge click on image to enlarge

I never considered to click OK believing that the window would close. It took me about one hour to solve the problem. I searched the messages of the Yahoo mail group and googled for it, but nobody seemed to have encountered my "problem". I installed the Beta version of OmegaT instead hoping that the ADD button would be in place. This was, however, not the case. Meanwhile I know that I should have clicked on OK. I have also managed to create some rules for Turkish according to what I had found in the rules for German. If you do not want a segment for a dot after ordinal numbers the rule would be:

[0-9]\. (before line break) and \s (after line break)

Having changed from Windows to Ubuntu I thought there must be a way of importing/exporting segmentation rules. On the newly installed OmegaT (better to say made ready to run) I found the file segmentation.conf, but in the programs running under Window I did not find such a file and, therefore, could not export or import anything. Since I created only two rules for Turkish it should not be difficult to do that again (easier than searching for it over hours).

Getting rid of orphan segments

I took the description from here

To remove orphan segments:

  1. Create translated documents.
  2. Make a copy of omegat/project_save.tmx (just in case).
  3. Replace omegat/project_save.tmx with either <name of your> project>-omegat.tmx or <name of your project>-level2.tmx.

New translation to German

My problems with the segmentation rules made me consider to make a new translation. Most of the things you need to know for that is explained on a HowTo page of Marc Prior. First of all I wanted to know what kind of translations had been done. That can be seen in a list only visible for subscribers to the Yahoo News-Group found on this page.

I discovered that earlier versions of OmegaT had been translated and it seemed as if a complete translation existed of version 1.7 (at the time of writing version 1.8 is the latest stable version and the beta version has gone up to 2.0.3). The person who had done that translation was so kind to send me his project (source and target file together with the translation memory) and the coordinator for localization told me that I should translate the most recent version to be found on a SourceForge page.

Meanwhile I have done the translation of the minimal package and wanted to test it. This is fine with the so called instant start guide (actually a html-file). I replaced the old file with my new translation and just had to insert two images in the sub folder to doc/de/images and that's it. The test for the readme.utf8 can only be done by directly opening it in an simple text file editor. I realized that the main directory of OmgeaT contained readme-files for other languages and saved my translated file (readme.utf8) accordingly as readme_de.txt. Since the readme file is displayed for users of Windows after installation I deinstalled and installed OmegaT. Although my newly created file was in the right location (as I believe) again just the English readme.txt file showed up. Meanwhile Didier Briel informed me that the German readme file has to be in the installation package to "perform" correctly (and that only in the Windows version, others can look at the readme file in their own language, if a translation exists).

Test the user interface

Marc Prior's HowTo page says the following:

Launch OmegaT from the command line (MS-DOS window) with the following commands (if you are unsure how to do this, refer to the OmegaT user manual)
    cd [full path to OmegaT]
    java -jar OmegaT.jar resource-bundle=[name of bundle with full path]
where [full path to OmegaT] is the full path to the folder containing the OmegaT program, and [name of bundle with full path] is the name of your translation of the resource bundle (bundle_en.properties, unless you have changed the name) including the full path to its location.

In my home edition of win xp there is the possibility to run commands (in German it says something like "execute"). The small window that opens allows for just one line to be inserted. The command cd = change directory was not recognized. Likewise I could not execute the second line. Meanwhile Didier Briel told me that I should have written just "command" into that line in order that a proper DOS windows opens.

In addition, the user's manual provides this information:

...the two commands described above are contained in a file (a "script", also called a ".bat file" on Windows systems). When this file is executed, the commands within it are automatically carried out. Consequently, to make changes to the launch command, it is sufficient to modify the file. When OmegaT is launched from the file, the new commands are then executed.

Easily said, but who would tell me, how to make such a batch file. I just tried my luck and wrote the two lines into an empty file with Notepad and saved it under "all files", "encoding" utf8 by the name of "omegat.bat". When I double clicked it OmegaT started with the correct(ed) interface.

Images of German interface

Here you can see the test results (German interface). If you like, compare to the initial screenshots above (to enlarge click on image).

View now reads "Ansicht"
Elements in "Ansicht" are in German
"Editing Behaviour" is "Editierverhalten"

Some more screenshots:

"Add" now reads "Hinzufügen"
Option -> Editing Behaviour before
Option -> Editing Behaviour after
Choose source and target language

Further remarks and files for download

In April 2013 OmegaT3.0 (beta) was released. I was asked whether I could translate the user interface (UI), a Windows specific file, the tutorial and "readme". I was also advised to use the beta version for my translation. However I was unable to install OmegaT 3.0 under Ubuntu 11.10 (see next chapter).

As of 10 July 2009 I have finished the translation of the full version, meaning that the user manual is now available in an updated German version. The German files (including images) from the target folder can be downloaded at http://ob.nubati.net/w/docs/target.zip (633kB) and the whole project including the translation memory and the source files plus the beginning of a glossary can be downloaded at http://ob.nubati.net/w/docs/OT_2.0.zip (almost 2MB).

My translation(s) are far from ideal or perfect. One of the points for discussion could be as to whether or not some expressions should be kept in the original English form. When you explain CAT as computer assisted translation, the German Computer unterstützte Übersetzung does not correspond to CAT and when you speak of TM as translation memory the German Übersetzungsspeicher does not correspond either. The term fuzzy matches might also be better than unscharfe Treffer. If someone is interested in discussing these questions, please drop me a line at the e-mail address for hoberdiek in the Yahoo Group.

If you want to work on the files, in http://ob.nubati.net/w/docs/Omegat2.0.zip you'll find the files

  • Bundel_de.properties, readme_de.txt and omegat.bat in the main folder (if you want to use the comegat.bat file you have to customize the paths and possibly file name to your needs/settings)
  • my translation memories TMs in the sub folder "tm" and
  • the Instant Start Guide in the sub folder docs/de, plus the corresponding images in docs/de/images

Useful hint: After finishing my translation I had many matches for which I had used the prefix [Hit] instead of [Fuzzy]. It is, of course, easy to remove the prefixes from the target file with "find" and "replace", but the secret how to do that in OmegaT is also simply:

go to the sub folder omegat; find the file project_save.tmx; open it in a simple editor and remove the prefixes.

Installing OmegaT under Ubuntu

Main article: Running OmegaT 3.0 in Ubuntu

I have been (willingly) missing several updates of Ubuntu and in May 2013 I am still working with version 11.10 (Oneiric Ocelot). The Software Center of Ubuntu offers version 2.3.3_4 of OmegaT. Because of some Tokenizers I had also installed version 2.5 and 2.6. In both cases all I had to do was to make the OmegaT.jar file executable and these versions would run either by double-clicking on them or by hitting the short cuts that I had created on the desktop.

On the page How to install OmegaT under Linux several ways are explained:

  • The Linux installation script, linux-install.sh.

Not said was that you launch the script in the folder by typing ./linux-install.sh. I did so in the terminal, was prompted for my password and after some time the process stopped. However, all that had been achieved was that the program was now "doubled" in /opt/omegat/OmegaT_3.0.0

  • The kaptain script

This is said to enable you to select launch and configuration options from a user-friendly GUI. For the kaptain script to work, however, you must first install kaptain. If it is not in your distribution's repository, you can download it from kaptain.sourceforge.net.

I did not find the program in the software center. Therefore, I downloaded it, but was unable to install it (could not find suitable instructions). Only the "download now" button on this page http://linuxappfinder.com/package/kaptain made it possible to install kaptain using the software center.

  • Launching OmegaT from the command line

To launch from the command line, I opened the terminal and navigated to the folder containing the file OmegaT.jar, in the first place this was the folder I had created for the program with JRE

cd .omegat3.0

entered the command

java -jar OmegaT.jar

and got

15680: Info: Java: Sun Microsystems Inc. ver. 1.6.0_27, executed from
'/usr/lib/jvm/java-6-openjdk/jre' (LOG_STARTUP_INFO)
Exception in thread "main" java.lang.NoClassDefFoundError:
org/omegat/core/matching/ITokenizer
at java.lang.ClassLoader.defineClass1(Native Method)
...
org.omegat.filters2.master.PluginUtils.loadFromManifest(PluginUtils.java:258)
at org.omegat.filters2.master.PluginUtils.loadPlugins(PluginUtils.java:110)
at org.omegat.Main.main(Main.java:172)
Caused by: java.lang.ClassNotFoundException:
org.omegat.core.matching.ITokenizer

The list was much longer and ended with saying that there were another 23 messages.

I asked two friends and all that could be assumed was that this problem had something to do with the Tokenizers. First, I deleted version 2.5 and 2.6 permanently, since the /plugins/ folders had a large number of Tokenizers. Nothing happened after that, i.e. I could not launch OmegaT. I found that version 2.3.3_4 also had a folder /plugins/ including *.jar files such as lucene-analyzers-2.9.0.jar, OmegaT-tokenizers.jar ect. Only after I had deleted all these files, opened the terminal, went to the folder with the script

cd /opt/omegat/OmegaT_3.0.0

and entered the command

java -jar OmegaT.jar

OmegaT 3.0 was launched.