Problem translating sdlxliff file with tikal with okf_xliff-sdl filter

Issue #674 new
Leon Tiberiu Cristian created an issue

tikal -ms msconfig.txt -sl en -tl ro -fc okf_xliff-sdl -t "WA38 0-7_SPECIFICATIONS.mif.sdlxliff" -oe utf-8 -opt 30 -maketmx mpretrans.tmx -trace


Okapi Tikal - Localization Toolset Version: 2.0.35


Trace: 1 class net.sf.okapi.applications.tikal.Main Trace: 2 ProtectionDomain (file:/F:/!Ahk/okapi-apps_win32-x86_64_0.35/lib/tikal.jar <no signer cert ificates>) sun.misc.Launcher$AppClassLoader@55f96302 <no principals> java.security.Permissions@1cbf6e72 ( ("java.io.FilePermission" "\F:!Ahk\okapi-apps_win32-x86_64_0.35\lib\tikal.jar" "read") ("java.lang.RuntimePermission" "exitVM") )

Trace: 3 (file:/F:/!Ahk/okapi-apps_win32-x86_64_0.35/lib/tikal.jar <no signer certificates>) Trace: 4 file:/F:/!Ahk/okapi-apps_win32-x86_64_0.35/lib/tikal.jar Trace: 5 /F:/!Ahk/okapi-apps_win32-x86_64_0.35/lib/tikal.jar Translation Source language: en Target language: ro Default input encoding: windows-1250 Output encoding: utf-8 Filter configuration: okf_xliff-sdl Output: /F:/!Ahk/okapi-apps_win32-x86_64_0.35/WA380-7_SPECIFICATIONS.mif.out.sdlxliff Input: /F:/!Ahk/okapi-apps_win32-x86_64_0.35/WA380-7_SPECIFICATIONS.mif.sdlxliff Trace: XMLInputFactory: com.ctc.wstx.stax.WstxInputFactory Leveraging settings: Microsoft-Translator Service: http://api.microsofttranslator.com/v2/Http.svc Processeed segments = 0 Best matches that are exact = 0 Best matches that are fuzzy = 0

Totals: Processed segments = 0 Best matches that are exact = 0 Best matches that are fuzzy = 0 Done in 2.824s

I see that the general filter "okf_xliff" works F:!Ahk\okapi-apps_win32-x86_64_0.35>tikal -ms msconfig.txt -sl en -tl ro -fc okf_xliff -t "WA380-7_SPECIFICATIONS.mif.sdlxliff" -oe utf-8 -opt 30 -maketmx mpretrans.tmx


Okapi Tikal - Localization Toolset Version: 2.0.35


Translation Source language: en Target language: ro Default input encoding: windows-1250 Output encoding: utf-8 Filter configuration: okf_xliff Output: /F:/!Ahk/okapi-apps_win32-x86_64_0.35/WA380-7_SPECIFICATIONS.mif.out.sdlxliff Input: /F:/!Ahk/okapi-apps_win32-x86_64_0.35/WA380-7_SPECIFICATIONS.mif.sdlxliff Leveraging settings: Microsoft-Translator Service: http://api.microsofttranslator.com/v2/Http.svc Processeed segments = 192 Best matches that are exact = 0 Best matches that are fuzzy = 161

Totals: Processed segments = 192 Best matches that are exact = 0 Best matches that are fuzzy = 161 Done in 50.856s

Is the okf_xliff-sdl filter not yet implemented for translating or other tasks? Also small typo in "Processeed segments" Thank you!

Comments (6)

  1. ysavourel

    The okf_xliff-sdl should work for any action: the filters are separated from the other steps. The difference between both filter configuration is just configuration, so I'm not sure why the behavior is so different and looks like a security issue.

    Maybe the file itself has something the SDL variant of the filter has trouble with. Could you post an example file that cause the issue?

  2. ysavourel

    Thanks for the file. I cannot reproduce the error. But I do get a problem: When using the default okf_xliff-sdl configuration I do not get an extracted segment.

    That issue is related to the SDLXLIFF file itself: It does not have segment markers in the source.

    If you open it is a text editor you will see that you have <seg-source> elements but no <mrk> in the <source>. In SDLXLIFF such segment are usually non-translatable segment when you do have a target. There is an option (You can see it in Rainbow, if you open the SDLXLIFF configuration via "Tools > Filters Configurations") that is "Skip seg-sources with no marked segments" which is on. If you create a custom configuration and check off that option, and use that configuration you get an extraction. And for me it creates the output and the TMX with Romanian.

    I'll attached the custom configuration, but you could create it yourself from Rainbow too.

    Maybe this will make your security-related issue go away. It seems to be unrelated problems though.

  3. Leon Tiberiu Cristian reporter

    Thank you for your time, I will look into it and test it, I do not know why the file is not properly tagged, I could open it in SDL Trados without problems.

  4. Log in to comment