topic Re: Problem with POI on Linux in Alfresco Archive

Problem with POI on Linux

orybak — Tue, 13 Nov 2007 17:16:35 GMT

Hello, everyone!I have encountered such a problem with metadata extraction: poi library extracts metadata from office files as unreadable symbols.In some files I have metadata written in English. It is extracted correctly. However, when metadata (for example document title or description) is written

Re: Problem with POI on Linux

orybak — Wed, 14 Nov 2007 08:35:31 GMT

OS: Windows 2000 Server (English)
Installed Locales: English (US), Russian.

Found some interesting effects:
1) If the default locale in Russian then metadata is extracted successfully from ms office documents BUT :!: CIFS does not allow to enter any folder with cyrillic characters in it.
2) When the default locale is English(US) CIFS is working, BUT :!: metadata is not extracted correctly

Did enyone observe such effects? Could anyone propose some solution/workaround for this?

Thanks in advance.

Re: Problem with POI on Linux

orybak — Tue, 11 Dec 2007 09:19:18 GMT

I have found a solution to this problem. Both POI and CIFS are working correctly when we explicitly defined application language and file encoding. So all we need is to add the following parameters to java executable:

-Duser.language=ru -Duser.country=RU -Duser.region=RU -Duser.variant=RU -Dfile.encoding=CP1251‍

After it is done, metadata is extracted successfully from office documents and CIFS is working correctly with folders containing cyrillic characters