Open Source Document Management System | OpenKM

Server Debian Lenny (LANG=en_US.UTF-8)
In OpenKM 4.1, created folder ÉM-MS (with E acute to begin)
Exported the repository to folder openkmexport41 in the file system
Export successful and folder ÉM-MS appears (with contents) in openkmexport41 with correct name
Many other files and folders, some with accented characters, exported also.
Fresh install of 5.0, repository empty
In OpenKM 5.0, ask to import from openkmexport41 to the repository
Operation aborts with message /home/dominiquem/okmexport41/��M-MS (No such file or directory)
Nothing is copied
Same problem with all sub-folders even with lower case accented characters.

Note the fact that ÉM becomes ��M with two unknown characters for the one É. It seems that the import function cannot read mutibyte UTF-8 in folder names.

I'm unable to reproduce the issue under my Ubuntu box. Created some folders and documents with non-ascii chars (like Cyrillic and Greek) and were imported without problems.

I have no problem uploading folders and files with special characters in the OpenKM repository, or renaming them in the OpenKM window. This works also through WebDAV.

My first post was only about importing with the administrators’s import function: login as OkmAdmin > Administration (screen OpenKM Administration) > icon with a green arrow « Repository import ». I could do a « Repository export » with special characters from OpenKm 4.x, but the « Repository import » in 5.0 fails on special characters.

More precisely, /home/dominiquem/okm is the result of the « Repository export » to the filesystem, with /home/dominiquem/okm/ÉM-MS as first subfolder (besides many others). When I try a « Repository import » in 5.0 from /home/dominiquem/okm, it blocks immediately on « /home/dominiquem/okm/��M-MS (No such file or directory) » and nothing is imported. The fact that there are two �� for one « É » character seems to indicate that the code of « Repository import » does not handle UTF-8 correctly.

But on a distant computer, the same content is copied without problem in a WebDAV mount of the Taxonomy and it then appears in the Taxonomy with correct accents. WebDAV provides thus a workaround of the bug in « Repository import ».

I know what you mean, but I can't reproduce this behaviour on my Ubuntu box. The files were copied from a Windows system?

Well, I have a problem with accents but I don't know if it's related.

I used the Uploader to import a directory tree. Some directory names contain accents. These directories were created fine in OpenKM. But the uploader didn't manage to import the files they contained, arguing that the files were not found.

Any hint on this issue?

Thanks!

Yes, I know that uploader applet has issues with non-ascii filenames. We have created n issue to track this bug: 0001532: Upload applet has problems with non-ascii filenames.

We've founded some problem on importation when users incorrectly mounted network resources. If it's your case ensure you've mount with utf-8 charset, browser in your unit after be mounted and ensure special characters are right while browsing it.

I've encountered this problem too. I'm not using mounted network anything, but I did copy all the files from a Windows system into a Linux system (Ubuntu). Is there a way to fix this without having to change all the filenames that I am uploading?

Which OpenKM version?

Sorry, OKM 5.1.9 on Ubuntu 10.10

Well, if you copy files with no-ascii chars in the name from Windows to Linux, you can get the name garbled because of different encoding in both Operating Systems. Also it depends on the Linux and Windows configuration, of course.