0

MSI database contains set of tables, and I can successfully enumerate File table, which has all deployable file' meta-deta. What I need to extract is the actual contents of those files. msiexec, lessmsi, 7-zip all can do it, but I couldn't find any source/API to do it.

What I've discovered it that all other (resource) files are in Binary table, and Data field can be used to get content of those files (like icons, custom DLL etc).

Further, I found and know that Media table contains information about the .CAB file (MSI has all content embedded with <MediaTemplate EmbedCab="yes"/>. This simply means the CAB file contains the actual content. I probably need to read contents from "Structured Storage" of the .msi file.

How to extract the contents of CAB/MSI file, using native C Msi* functions?

Ajay
  • 18,086
  • 12
  • 59
  • 105
  • You asked for native C Msi* functions. But if it's sufficient to do it interactively you can look at the dark.exe from the WiX toolset. There is also source code available. – harper Feb 21 '18 at 06:19
  • I will see the source code (assume it is native). I cannot use external exe, as I need to read something in my custom DLL. – Ajay Feb 21 '18 at 06:22
  • My answer below explains how you can query all the tables and extract assemble the directory tree of files. If your willing to look at .NET WiX/DTF has a InstallPackage::ExtractFiles() method that does all this for you. The source code is here: https://github.com/wixtoolset/Dtf/blob/master/src/WixToolset.Dtf.WindowsInstaller.Package/InstallPackage.cs – Christopher Painter Feb 21 '18 at 20:54
  • The DTF code should explain the process and every class it calls to interact with MSI has doco that says what the underlying Win32 API is that you would use in c++. – Christopher Painter Feb 21 '18 at 20:55
  • Hmm, just saw this perfectly legit question down-voted. What on earth is that for? Let's upvote - this is a good question. You got good answers already, I added one with some info on the **Win32 functions** you can use below. Also, consider searching for `msi.chm` in your WiX installation folder or the Windows SDK - it is the MSI bible - the MSI SDK documentation itself. Much quicker to look things up in than the online MSDN version. And you get a sense of scope as well - how much content there really is to review. – Stein Åsmul Feb 26 '18 at 17:00

3 Answers3

3

The most straightforward to extract all the files to some location is to install the product in "advertised" mode. If you do a:

msiexec /a [path to msi] TARGETDIR=[some folder]

you'll see what happens.

In C++ call MsiInstallProduct () with that command line.

PhilDW
  • 20,260
  • 1
  • 18
  • 28
  • 1
    This is definitely the right first method to recommend. But as a point of nomenclature, this (/a) is performing an administrative installation instead of an advertised install (/j); see [Command-Line Options](https://msdn.microsoft.com/en-us/library/windows/desktop/aa367988). – Michael Urman Feb 23 '18 at 19:00
  • Quite right. Between Phil, Michael, Stein and myself you aren't going to get any wrong answers in this question. Your probably looking at 100 years of setup development expertise right there. – Christopher Painter Feb 23 '18 at 19:54
3

Phil has given you the easy/simple answer but I thought I might give you a little more information since you've done some research. Checkout:

https://msdn.microsoft.com/en-us/library/windows/desktop/aa372919(v=vs.85).aspx

This is where the structured storage is. You'll see something like Disk1.cab as the Name (PK) and binary data. The data is a CAB file with the file entry in the cab matching the File.File column. From there you can use the File.FileName column to get the short name and long name (you'll want the long name no doubt) and do a joint to the Component table to get the directory table ID.

You'll also need to recurse the directory table to build the tree of directories and know where to put the files.

Fun stuff. There's some libraries in C# that make this WAY simpler. Or just call msiexec /a as Phil says. :)

Christopher Painter
  • 54,556
  • 6
  • 63
  • 100
  • I've found out that virtual table `_Streams` holds the deployable files. I can also use structured-storage API directly on the MSI to get the files/CAB. Thanks! – Ajay Feb 23 '18 at 06:30
  • 1
    It took me 6 months back in 2003 but I think the entire MSI SDK documentation is indexed in my brain.:) – Christopher Painter Feb 23 '18 at 08:17
2

You have gotten many good answers already, including the use of dark.exe from the WiX toolkit. By downloading the WiX source code you should be able to get the code you need ready-made from there. I assume you may already have done this.

Chris has already linked to the DTF code you can check, but here is a link directly to dark.exe as well: https://github.com/wixtoolset/wix3/tree/develop/src/tools/dark. I would try both. This is C#, you seem to want native.

UPDATE: Before I get to the Win32 features you can use, check out this little summary of the C# DTF features: How to programmatically read the properties inside an MSI file?

Native Win32 functions: The database functions to deal with an MSI file can be found on MSDN (this is to deal with the MSI file as a database). There are also MSI Installer Functions (used to deal with the MSI file as an actual installer).

You can certainly find good examples of native code for this with a good Google search. Have fun!

BTW: It would help with a description of the actual problem you are trying to solve as well as what you need technically. There could - as always - be less involved ways to achieve what you need. Unless you are writing a security software or malware scanner or something super-involved.

And so it is clear: WiX's dark.exe fully decompiles MSI files into WiX source files and the resource files used to build them - you can then text and binary compare the various types of content (text compare for tables, binary compare for binaries, etc...). The process to do so via command line is described in the following answer: How can I compare the content of two (or more) MSI files? (this is about comparing MSI files, but one option to do so is to decompile them - see section on dark.exe - just for reference for others who find your question).

I like to link things together so we can find content easily at a later point in time. Strictly speaking it doesn't seem necessary here, you have what you need I think but others could perhaps benefit from some further links. Here are some related links:

Stein Åsmul
  • 39,960
  • 25
  • 91
  • 164