While tackling issue #529, I stumbled across the odd behaviour that
re-compressing the archive with 7-Zip broke what should be a valid
Manuskript project.
After investigation it turned out that the code that loads the texts
sensibly expects there to only be files tracked in the files dictionary.
It is completely valid for a zip file to contain entries describing the
contained directories. The logical fix is to simply avoid adding these
directory entries to our files dictionary in the first place.
See issue #281.
When loading a project that has the setting **Save to one single
file** disabled, Manuskript tries to read all directories and files
under the project directory.
Manuskript expects all files to contain valid unicode characters.
However if a file containing non-unicode characters is read then
Manuskript will crash.
The error message displayed on the console is similar to the
following:
----- begin snippet -----
Traceback (most recent call last):
File "/home/gedakc/workspace/manuskript.olivierkes/bin/../manuskript/ui/welcome.py", line 134, in loadRecentFile
self.mw.loadProject(act.data())
File "/home/gedakc/workspace/manuskript.olivierkes/bin/../manuskript/mainWindow.py", line 566, in loadProject
self.loadDatas(project)
File "/home/gedakc/workspace/manuskript.olivierkes/bin/../manuskript/mainWindow.py", line 793, in loadDatas
errors = loadSave.loadProject(project)
File "/home/gedakc/workspace/manuskript.olivierkes/bin/../manuskript/loadSave.py", line 66, in loadProject
v1.loadProject(project, zip=isZip)
File "/home/gedakc/workspace/manuskript.olivierkes/bin/../manuskript/load_save/version_1.py", line 657, in loadProject
files[os.path.join(p, f)] = fo.read()
File "/usr/lib/python3.5/codecs.py", line 321, in decode
(result, consumed) = self._buffer_decode(data, self.errors, final)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x80 in position 3131: invalid start byte
----- end snippet -----
There are at least two known situations in which files with
non-unicode characters can arise:
A. The project is on Mac OS X and the operating system automatically
creates a .DS_Store file.
B. The project is under git version control and contains a .git
subdirectory.
This enhancement prevents the Manuskript crash on project load by
ignoring all directory and file names that start with a period.
