Re: [l10n-dev] Translatable contents extraction ?
Op Maandag 2008-02-25 skryf Jean-Christophe Helary: > Is it possible to only have the PO parts that need translation/ > updating and not the whole set ? > > All the already translated parts are irrelevant to the translation > itself (except when used as translation memories). > > Is there a simple tool that can extract the translation data and later > merge the translated data ? pofilter and pomerge will help you do this. In fact, if you send your translations right back to Pootle, you can just upload the translated subsets when you upload (as long as you don't choose "overwrite" when you download. The default behaviour should be "merge", which is what you want. http://translate.sourceforge.net/wiki/toolkit/pofilter http://translate.sourceforge.net/wiki/toolkit/pomerge You can download a ZIP file of all the PO files in the project/directory where you want to do this. You are interested in "pofilter --test=untranslated", but the page above will give more information on the command line use. Friedel - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
[l10n-dev] Translatable contents extraction ?
Is it possible to only have the PO parts that need translation/ updating and not the whole set ? All the already translated parts are irrelevant to the translation itself (except when used as translation memories). Is there a simple tool that can extract the translation data and later merge the translated data ? Jean-Christophe Helary http://mac4translators.blogspot.com/ - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
[l10n-dev] Re: [ja-translate] Re: [l10n-dev] How can we review with Pootle ?
Reiko, If the number of fuzzy is small, we can work on Pootle directly, can't we ? That is what I would suggest. If the update volume is high, such as in HC, we can use [TM] mark to be inserted to the leverage from TMX, right ? Is there any way to do the opposite, taht is, mark "new" translation ? It is possible to insert it manually. But what I propose is an automatic insertion when OmegaT recognizes a 100% match. If there's any mark put on the new translation, we can search that segment with that mark. I understand. It would be indeed very convenient :) Especially since searches in OmegaT cover both source and target without distinction ... The ideal is that we get only the segments to translate or update, not the whole package. That is a waste of resources and requires useless roundtrip manipulations... I suggest we extract all the non translated segments before starting the translations. That would make all the manipulations above irrelevant. JC Even if there's no such a way, your workaround will be a big help. Thank you again for your help! Regards, -Reiko Jean-Christophe Helary wrote: On 18 févr. 08, at 18:48, Jean-Christophe Helary wrote: Let me confirm. I understand the new/fuzzy is identified on OmegaT, but once the translator did the translation and put the translated string to the "untranslated" segment, how the reviewer can recognize which one is the strings to review ? Reiko, PO is not exactly the strong point of OmegaT :) I'll check tonight with a PO from Pootle and will get back to you later. Maybe on the ja list ? Reiko, I have just tried OmegaT with Localization.po from javainstaller2. The file is translated at 97% and contains only 2 fuzzies to check. The conclusion is that OmegaT is useless for files that mostly contain translated and fuzzy strings. Ideally, a source file should not contain such strings and all the reference should be stored in a TMX. The fuzzies should be left empty for normal translation. If you work with a file that is mostly untranslated and where the reference parts are clearly separated from the source, it is trivial to set OmegaT to insert the TM reference with a prefix to distinguish it from the Translator's input. Just set OmegaT to automatically insert 100% matches with a [TM] prefix, or anything you want. The translator will still be in control of the process and will be able to do modifications to the input if necessary. When the reviewer checks the file, only the parts that are not marked with [TM] will have to be checked. I understand that this is not an ideal workflow though... Jean-Christophe Helary http://mac4translators.blogspot.com/ - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] -- Reiko Saito Japanese Language Lead Translation Language and Information Services (TLIS) Globalization Services Sun Microsystems, Inc. Email: [EMAIL PROTECTED] Phone: +81 3 5962 4912 Blog: http://blogs.sun.com/reiko - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] Jean-Christophe Helary http://mac4translators.blogspot.com/ - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
[l10n-dev] Re: [ja-translate] Re: [l10n-dev] How can we review with Pootle ?
Hi JC, Thanks very much for your time. If the number of fuzzy is small, we can work on Pootle directly, can't we ? If the update volume is high, such as in HC, we can use [TM] mark to be inserted to the leverage from TMX, right ? Is there any way to do the opposite, taht is, mark "new" translation ? If there's any mark put on the new translation, we can search that segment with that mark. Even if there's no such a way, your workaround will be a big help. Thank you again for your help! Regards, -Reiko Jean-Christophe Helary wrote: On 18 févr. 08, at 18:48, Jean-Christophe Helary wrote: Let me confirm. I understand the new/fuzzy is identified on OmegaT, but once the translator did the translation and put the translated string to the "untranslated" segment, how the reviewer can recognize which one is the strings to review ? Reiko, PO is not exactly the strong point of OmegaT :) I'll check tonight with a PO from Pootle and will get back to you later. Maybe on the ja list ? Reiko, I have just tried OmegaT with Localization.po from javainstaller2. The file is translated at 97% and contains only 2 fuzzies to check. The conclusion is that OmegaT is useless for files that mostly contain translated and fuzzy strings. Ideally, a source file should not contain such strings and all the reference should be stored in a TMX. The fuzzies should be left empty for normal translation. If you work with a file that is mostly untranslated and where the reference parts are clearly separated from the source, it is trivial to set OmegaT to insert the TM reference with a prefix to distinguish it from the Translator's input. Just set OmegaT to automatically insert 100% matches with a [TM] prefix, or anything you want. The translator will still be in control of the process and will be able to do modifications to the input if necessary. When the reviewer checks the file, only the parts that are not marked with [TM] will have to be checked. I understand that this is not an ideal workflow though... Jean-Christophe Helary http://mac4translators.blogspot.com/ - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] -- Reiko Saito Japanese Language Lead Translation Language and Information Services (TLIS) Globalization Services Sun Microsystems, Inc. Email: [EMAIL PROTECTED] Phone: +81 3 5962 4912 Blog: http://blogs.sun.com/reiko - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]