I am creating a zip archive containing two identical files at different paths. Does the zip archive format support something akin to the Unix concept of hard links? By this I mean the ability to store the file only once (saving space), but to index that data blob to two different paths within the zip archive.
If the file format does support this, how could I go about creating such an archive using free tools in Ubuntu?
On
As @Greg said, ZIP doesn't support hardlinks.
But if I understand right, your purpose is to reduce size of the compressed archive. So let's think of an alternative solution.
Lets run simple tests to check compression ratio of different archive libraries. I've created two identical binary files and compressed them using ZIP, BZ2, RAR and 7z.
8641969 test.bin
8641969 test2.bin
First time only one file was compressed. Second time two files were compressed:
ZIP:
$zip -9 test1.zip test.bin
$zip -9 test2.zip test.bin test2.bin
8636837 test1.zip
17273654 test2.zip
BZIP2:
$export BZIP=--fast
$tar cjf test1.tbz test.bin
$tar cjf test2.tbz test.bin test2.bin
8694997 test1.tbz
17389167 test2.tbz
7z:
$7z a -mx=9 test1.7z test.bin
$7z a -mx=9 test2.7z test.bin test2.bin
8705285 test1.7z
8707054 test2.7z
RAR:
$rar a -m5 test1.rar test.bin
$rar a -m5 test2.rar test.bin test2.bin
8649970 test1.rar
17299916 test2.rar
Conclusion: It seems that only 7z does the job good. Consider using it in your application.
Of course you will need to do more tests at your environment with your files to see if it really what you need. Also you can play with options to see on which level of compression you get the best compression ratio/speed balance.
No, the Zip file format does not support this. This is because the Local File Header contains information about the file, including its name, followed immediately by the compressed data for the file. It is not possible for two different Local File Headers to point to the same compressed data.