On Fri, Jun 26, 2015 at 08:13:05PM +0300, Guy Rutenberg wrote: > Hi, > > I've attached patches for two bugs I've encountered in the lyxpak script: > > The first bug is related to how the common path prefix for the included > files is calculated. If the common string prefix (the byte-by-byte prefix) > of the included files is also a file, the script would get the wrong > path-prefix and fail. The first patch fixes this issue. > > The second bug is related to handling non-ascii filenames when exporting to > zip file. Zip files created by Python's zipfile module by default store > filenames as 8-bit cp-437 encoding. This is problematic when filenames use > characters from any other encdoing. To solve this issue I explicitly pass a > unicode object as 'arcname' which causes the zipfile module to store the > file as utf-8 and set the approperiate flag bits for the file in the > archive. > > I've tested both patches in Linux, but I guess some testing on Windows is > needed as well. There is one point which I expect trouble with is in the > conversion to the unicode object. The lyx file itself is utf-8 so the > conversion to unicode is straight-forward for any filename extracted from > it. But the filename of the lyxfile is in the file-system encoding, which > might differ, hence causing problems.
Thanks, I'll have a look. -- Enrico