It is not a schizophrenic zip file to have inline headers that are not referenced in the TOC. the TOC is the only source of truth in a zip file. It was designed this way in purpose so that you can add new versions of files on your 20 disk zip without having to re-write all 20 disks. Pkzip would read the TOC from disk 20, append your new file to disk 20 (or 21 if there was not enough space) and then write a new TOC at the end that does not reference the old file still in the zip. That is by design. Reading anything other than the files in the TOC is an invalid zip reader
While that is true (even if some people really want to stream read zips and will thus trigger tar-style conflicts) that is not what the article is about. The article is really more about software using the "size of central directory" field to find the central directory (a sin Python's standard library apparently commits) versus the "offset of central directory".
A commenter on an other post of that article noted that technically there's no reason for the central directory to be right before the EOCD so seeking backwards from the EOCD by the size of CD is just incorrect. In fact zip was designed such that the central directory could be split across multiple disks (and later files), so it was not possible to guarantee a simple backwards jump from the EOCD to the start of CD.
It's talking about how the EOCD contains both the size of the central directory and the offset of the start of the central directory, which is redundant. So we end up with some tools honoring the offset, while some subtract the size from the EOCD.
It's an interesting hole that the test cases don't cover any of Microsoft Office, Windows Explorer, PowerShell's various cmdlets, or the several major .NET ZIP archive libraries. It would seem that the author just does not use Microsoft Windows.
There's a whole extra level of archive file format tooling gotchas that one misses out on when one assumes "UNIX" for everything, and does not account for "FAT", "NTFS", "HPFS", and even "OpenVMS".
The trick depends upon different implementations doing different things. Not likely for Word (though I suppose it is -possible- across different versions or different OSes).
To respond to Grandfather comment, modern Office files are really just ZIPs with different extensions, they even have the magic string "PK" at the very beginning of the file.
I do wonder, since a lot of tools outside of the MS ecosystem can read Office files (e.g. LibreOffice and Google Docs as well as plenty of other online tools), if indeed the hack as described by the article is possible. One would just need to figure out the ZIP stacks used by said tools.
You can even just rename a docx file to use the zip extension and then manually unzip it for those curious. If I remember correctly, the contents are XML files with structure encoding the formatting around the content.
The described exploit seems theoretical. In order to create the schizophrenic ZIP, the attacker would have to figure out what ZIP stacks are being used and ensure they act differently - if the 2 departments use the same stack, then the exploit can't work, can it?
There was a time when passing ZIP files around was a very popular method of software distribution, and things like this were gotchas that had to be watched for. It was widely known, at least amongst sysops, that the varied toolsets that handled ZIP archives were functionally different. And there were scanners and sanity checkers, and bugfixes to PKUNZIP, that dealt in this stuff for uploaded files and FREQ responses.
Did people exploit the differences? Yes. Although it was mainly on the level of creating prank ZIP files on non-Microsoft operating systems with 8.3 filenames such as "PRN" or "CLOCK$".
However, the truly terrible idea of self-extracting archives was popular, which meant that archives with "interesting" arrangements of the archive within the overall file were widespread. ZIP comments were also liberally applied and altered by pretty much every BBS that passed an archive along. And the Unix people wanted to be able to use pipes, something that the MS-DOS original never had to cater for.
Also, there were people who exploited the fact that different tools took different things as gospel. Even within the past decade one can find people still being caught out by the fact that there's a header field that instructs what the pathname separator character(s) used are; and that ZIP tools that expect non-seekable streams operate differently to ZIP tools that expect seekable regular files.
A more realistic attack would be something like, slipping a malicious payload past a scanner by emailing a zip file that appears innocent when unpacked with the scanner’s zip implementation but produces malware when unpacked with the email client’s implementation. There’s a decent chance they’ll be different, and it wouldn’t be too hard to guess which ones a target might be using.
The author explains in the article that they previously gave a presentation outlining various techniques to achieve a "schizophrenic" zip file. The blog post discusses an additional technique that was not present in their previous presentation.
A commenter on an other post of that article noted that technically there's no reason for the central directory to be right before the EOCD so seeking backwards from the EOCD by the size of CD is just incorrect. In fact zip was designed such that the central directory could be split across multiple disks (and later files), so it was not possible to guarantee a simple backwards jump from the EOCD to the start of CD.
It's talking about how the EOCD contains both the size of the central directory and the offset of the start of the central directory, which is redundant. So we end up with some tools honoring the offset, while some subtract the size from the EOCD.
There's a whole extra level of archive file format tooling gotchas that one misses out on when one assumes "UNIX" for everything, and does not account for "FAT", "NTFS", "HPFS", and even "OpenVMS".
Or ZIP64. (-:
* https://github.com/dotnet/runtime/blob/main/src/libraries/Sy...
* https://github.com/mihula/ProDotNetZip/blob/main/src/Zip/Zip...
Then you have people on Linux or macOS who might also use LibreOffice, Apples Office suite, or something else entirely.
And given MS Office is the de facto standard, you’ll often see people open OOXML documents within non-MS office suites.
After all, OOXML is an open standard (sarcasm).
ODF (the document formats favoured by most other office suites) is also ZIP-based XML. So they too could be vulnerable.
I do wonder, since a lot of tools outside of the MS ecosystem can read Office files (e.g. LibreOffice and Google Docs as well as plenty of other online tools), if indeed the hack as described by the article is possible. One would just need to figure out the ZIP stacks used by said tools.
There was a time when passing ZIP files around was a very popular method of software distribution, and things like this were gotchas that had to be watched for. It was widely known, at least amongst sysops, that the varied toolsets that handled ZIP archives were functionally different. And there were scanners and sanity checkers, and bugfixes to PKUNZIP, that dealt in this stuff for uploaded files and FREQ responses.
Did people exploit the differences? Yes. Although it was mainly on the level of creating prank ZIP files on non-Microsoft operating systems with 8.3 filenames such as "PRN" or "CLOCK$".
* https://groups.google.com/g/alt.comp.virus/c/zLV-Y2a71gs/m/U...
However, the truly terrible idea of self-extracting archives was popular, which meant that archives with "interesting" arrangements of the archive within the overall file were widespread. ZIP comments were also liberally applied and altered by pretty much every BBS that passed an archive along. And the Unix people wanted to be able to use pipes, something that the MS-DOS original never had to cater for.
Also, there were people who exploited the fact that different tools took different things as gospel. Even within the past decade one can find people still being caught out by the fact that there's a header field that instructs what the pathname separator character(s) used are; and that ZIP tools that expect non-seekable streams operate differently to ZIP tools that expect seekable regular files.