#18 dictunzip fails on multiples of chunkLength

[Ivo Danihelka]

I noticed a bug in dictzip uncompressing.
It fails when the size of the original non-compressed file is a multiple of chunkLength (58315 bytes).

Examples of produced errors:
{{{
$ dictunzip one_chunk.txt.dz
dictunzip (dict_data_read_): Internal error
h->chunks[1] = 2520048 >= 65535 (OUT_BUFFER_SIZE)
Aborting dictunzip...
Aborted
$ dictunzip two_chunks.txt.dz
dictunzip (dict_data_read_): inflate: (null)
}}}

The suspicious code is in dict_data_read_().
The lastChunk should be computed as: (start + size - 1) / chunkLength
See the attached patch.

A workaround is to use gzip to extract the files.

[Ivo Danihelka]

Patch to fix lastChunk computation.

[Aleksey Cheusov]

I've applied your patch and created regression tests for dictzip.
Thank you for your report!