Correctly support filenames with special characters

[rzimmer]

Currently, it seems that filenames like Alex' file.txt get encoded to Alex' file.txt in the TOC.xml when creating an xar file. When extracting the archive, this encoded filename gets used as final filename (thus creating a file called Alex' file.txt), because there is no decoding happening. This should be investigated more and fixed.

[rzimmer]

Couldn't reproduce this issue anymore. Closing this for now.

[rzimmer]

The official xar tool produces invalid XML for the ToC: It does not escape chars like ' to '. This Java port does and thus produces valid XML (currently there seems to be no way to turn this behaviour off).

It seems that 7-Zip has problems with this. Creating an archive using xar for Java with the following files

├── test's.txt
└── xar.patch

0 directories, 2 files

leads to the following file listing for the created archive:

$ 7z l test.xar 

7-Zip [64] 16.02 : Copyright (c) 1999-2016 Igor Pavlov : 2016-05-21
p7zip Version 16.02 (locale=de_DE.UTF-8,Utf16=on,HugeFiles=on,64 bits,8 CPUs Intel(R) Core(TM) i7-6820HQ CPU @ 2.70GHz (506E3),ASM,AES-NI)

Scanning the drive for archives:
1 file, 1007 bytes (1 KiB)

Listing archive: test.xar

--
Path = test.xar
Type = Xar
Physical Size = 1007
Headers Size = 295

   Date      Time    Attr         Size   Compressed  Name
------------------- ----- ------------ ------------  ------------------------
                    .....           10           16  test's.txt
                    .....          676          676  xar.patch
                    .....          741          741  [TOC].xml
------------------- ----- ------------ ------------  ------------------------
                                  1427         1433  3 files

$ xar -t -f test.xar 
xar.patch
test's.txt

$ java -jar xar-cli-0.9.5-SNAPSHOT-standalone.jar -t -f test.xar
INFO xar.patch
INFO test's.txt

While the official xar implementation (presumably using libxml) as well as the Java one and even file-roller (GNOME's archive manager) correctly de-escape the file name from the TOC.xml, 7-Zip doesn't and thus while create wrongly named files when extracting.

I consider this rather a bug in 7-Zip than in xar for Java, although xar (official) behaves somehow buggy, too, when not creating valid XML...

The currently used XML implementation doesn't allow disabling escaping, thus this won't be fixed in the near future. Leaving this open for reference, though...