about summary refs log tree commit diff
path: root/pkgs/santander
Commit message (Collapse)AuthorAgeFilesLines
* pkgs: Move all of my packages into pkgs/aszligaszlig2017-04-234-471/+0
| | | | | | | | | | | | | This already has started in e0abe1dbbda54c1f048b5d38df05e1a3289216a6 with @Profpatsch putting his packages into its own namespace, so let's continue on that and move my crap into my own namespace as well. The only difference in my approach is that I'm now also using a new function called callPackageScope, which declutters pkgs/default.nix a bit and moves the individual callPackage invocations into aszlig/default.nix. Signed-off-by: aszlig <aszlig@redmoonstudios.org>
* pkgs/santander: Whitelist webkitgtk2 for dwbaszlig2017-04-191-9/+15
| | | | | | | | | | | | | | | | | | | | | | | | | | | WebKitGTK+ has a bunch of security vulnerabilities: * WSA-2017-0002 * WSA-2017-0001 * WSA-2016-0006 * WSA-2016-0005 * WSA-2016-0004 We're using dwb in conjunction with this insecure version only for *one* specific application with *one* specific web site, so the attack surface is relatively low. The main issue with this is that the plugin required for the Santander page is based on NPAPI and most of the browsers out there do not support NPAPI, except a few ones: https://en.wikipedia.org/wiki/NPAPI#Browser_support I've tried to embed the plugin within Midori but I couldn't get it to work so far (Midori just says "No plugin available"), so I'm sticking with dwb for now, because the other browsers listed here either have the same problem (uzbl) or are simply too heavyweight. Signed-off-by: aszlig <aszlig@redmoonstudios.org>
* pkgs/santander: Switch to 32bit buildaszlig2016-10-191-3/+3
| | | | | | | | | | Right now the linking process with wineg++ fails on 64bit and I currently don't have time to properly look into why this happens. Another workaround would be to just use patchelf to fix the errors afterwards, but in the end everything except dwb has to be 32bit anyway. Signed-off-by: aszlig <aszlig@redmoonstudios.org>
* pkgs/santander: Trim down closure sizeaszlig2016-10-191-2/+4
| | | | | | | | | | | | | | | Since the version bump of wineStaging, stuff like mono and gecko are included when creating a new wineprefix. We can avoid this by overriding DLLs using "mscoree,mshtml=". In addition to that, we now use wineMinimal instead of the wineStaging attribute from nixpkgs so that we really have as few options compiled in as possible, especially because we really don't need them for the plugin. This trims down the closure size by a few hundred megabytes. Signed-off-by: aszlig <aszlig@redmoonstudios.org>
* pkgs/santander: Suppress substitution failureaszlig2016-08-291-0/+3
| | | | | | | | | | | | | Since NixOS/nixpkgs@b6c204f substituting a file which doesn't exist will fail and the preFixup phase tries to substitute something in the "install-dependency" script, which doesn't exist here in vuizvui (we're using a later version directly from upstream Git). I'm suppressing this simply by null'ing out the preFixup phase instead of fixing this properly because I'm not going to stay at this bank for a very long time. Signed-off-by: aszlig <aszlig@redmoonstudios.org>
* pkgs: Add a wrapped browser for Santander HBCIaszlig2016-06-064-0/+460
I didn't get the starcoscard to run with aqbanking so far and the bank itself is very uncooperative if it comes to giving specific details about their implementation of FinTS 3.00, so in the end I'm going to move away from the bank. But during transition this will work much better than running a Windows VM (which I didn't have access to in the meantime, so I *had* to get this running somehow), especially because we can wrap this plugin in *any* browser that supports NPAPI. Also, there seems to be some work implementing PPAPI support for pipelight, but the branch is stale since quite a while: https://bitbucket.org/mmueller2012/pipelight/branch/ppapi Going back to the pesky Santander plugin: In order to support PC/SC-Lite, we need to patch Wine to get support for the winscard API. We also patch out unixfs, so while there definitely are better sandboxing options this should suffice so that the plugin doesn't write garbage on any location of the system (basically it works entirely read-only). So in the end we get a nice and small dwb browser, which directly opens up the login page along with the plugin. The browser is wrapped so that it only writes to a temporary location, so as soon as it is closed all the cruft is cleaned up afterwards. Signed-off-by: aszlig <aszlig@redmoonstudios.org>