Skip to content

Latest commit

 

History

History
239 lines (165 loc) · 10.7 KB

Compile.rst

File metadata and controls

239 lines (165 loc) · 10.7 KB
Raw

Building DFHACK

Linux

On Linux, DFHack acts as a library that shadows parts of the SDL API using LD_PRELOAD.

How to get the code

DFHack doesn't have any kind of system of code snapshots in place, so you will have to get code from the github repository using git. Having a 'git' package installed is the minimal requirement, but some sort of git gui or git integration for your favorite text editor/IDE will certainly help.

The code resides here: https://github.com/peterix/dfhack

If you just want to compile DFHack or work on it by contributing patches, it's quite enough to clone from the read-only address:

git clone git://github.com/peterix/dfhack.git
cd dfhack
git submodule init
git submodule update

If you want to get really involved with the development, create an account on github, make a clone there and then use that as your remote repository instead. Detailed instructions are beyond the scope of this document. If you need help, join us on IRC (#dfhack channel on freenode).

Dependencies

DFHack is meant to be installed into an existing DF folder, so get one ready.

For building, you need a 32-bit version of GCC. For example, to build DFHack on a 64-bit distribution like Arch, you'll need the multilib development tools and libraries.

Before you can build anything, you'll also need cmake. It is advisable to also get ccmake on distributions that split the cmake package into multiple parts.

For the code generation parts, you need perl and the XML::LibXML and XML::LibXSLT perl packages. You should be able to find them in your distro repositories (on Arch linux 'perl-xml-libxml' and 'perl-xml-libxslt').

Build

Building is fairly straightforward. Enter the build folder and start the build like this:

cd build
cmake .. -DCMAKE_BUILD_TYPE:string=Release -DCMAKE_INSTALL_PREFIX=/home/user/DF
make install

Obviously, replace the install path with path to your DF. This will build the library along with the normal set of plugins and install them into your DF folder.

Alternatively, you can use ccmake instead of cmake:

cd build
ccmake ..
make install

This will show a curses-based interface that lets you set all of the extra options.

You can also use a cmake-friendly IDE like KDevelop 4 or the cmake-gui program.

Windows

On Windows, DFHack replaces the SDL library distributed with DF.

How to get the code

DFHack doesn't have any kind of system of code snapshots in place, so you will have to get code from the github repository using git. You will need some sort of Windows port of git, or a GUI. Some examples:

The code resides here: https://github.com/peterix/dfhack

If you just want to compile DFHack or work on it by contributing patches, it's quite enough to clone from the read-only address:

git clone git://github.com/peterix/dfhack.git
cd dfhack
git submodule init
git submodule update

The tortoisegit GUI should have the equivalent options included.

If you want to get really involved with the development, create an account on github, make a clone there and then use that as your remote repository instead. Detailed instructions are beyond the scope of this document. If you need help, join us on IRC (#dfhack channel on freenode).

Dependencies

First, you need cmake. Get the win32 installer version from the official site: http://www.cmake.org/cmake/resources/software.html

It has the usual installer wizard. Make sure you let it add its binary folder to your binary search PATH so the tool can be later run from anywhere.

You'll need a copy of Microsoft Visual C++ 2010. The Express version is sufficient. Grab it from Microsoft's site.

For the code generation parts, you'll need perl and XML::LibXML. You can install them like this:

  • download and install strawberry perl from http://strawberryperl.com/
  • reboot so that the system can pick up the new binary path
  • open a cmd.exe window and run "cpan XML::LibXML" (obviously without the quotes). This can take a while to complete.
  • Same with "cpan XML::LibXSLT".

If you already have a different version of perl (for example the one from cygwin), you can run into some trouble. Either remove the other perl install from PATH, or install libxml and libxslt for it instead. Strawberry perl works though and has all the required packages.

Build

There are several different batch files in the build folder along with a script that's used for picking the DF path.

First, run set_df_path.vbs and point the dialog that pops up at your DF folder that you want to use for development. Next, run one of the scripts with generate prefix. These create the MSVC solution file(s):

  • all will create a solution with everything enabled (and the kitchen sink).
  • gui will pop up the cmake gui and let you pick and choose what to build. This is probably what you want most of the time. Set the options you are interested in, then hit configure, then generate. More options can appear after the configure step.
  • minimal will create a minimal solution with just the bare necessities - the main library and standard plugins.

Then you can either open the solution with MSVC or use one of the msbuild scripts:

  • Scripts with build prefix will only build.
  • Scripts with install prefix will build DFHack and install it to the previously selected DF path.
  • Scripts with package prefix will build and create a .zip package of DFHack.

When you open the solution in MSVC, make sure you never use the Debug builds. Those aren't binary-compatible with DF. If you try to use a debug build with DF, you'll only get crashes. So pick either Release or RelWithDebInfo build and build the INSTALL target.

The debug scripts actually do RelWithDebInfo builds.

Build types

cmake allows you to pick a build type by changing this variable: CMAKE_BUILD_TYPE

cmake .. -DCMAKE_BUILD_TYPE:string=BUILD_TYPE

Without specifying a build type or 'None', cmake uses the CMAKE_CXX_FLAGS variable for building.

Valid and useful build types include 'Release', 'Debug' and 'RelWithDebInfo'. 'Debug' is not available on Windows.

Using the library as a developer

Currently, the most direct way to use the library is to write a plugin that can be loaded by it. All the plugins can be found in the 'plugins' folder. There's no in-depth documentation on how to write one yet, but it should be easy enough to copy one and just follow the pattern.

Other than through plugins, it is possible to use DFHack via remote access interface, or by writing Lua scripts.

The most important parts of DFHack are the Core, Console, Modules and Plugins.

  • Core acts as the centerpiece of DFHack - it acts as a filter between DF and SDL and synchronizes the various plugins with DF.
  • Console is a thread-safe console that can be used to invoke commands exported by Plugins.
  • Modules actually describe the way to access information in DF's memory. You can get them from the Core. Most modules are split into two parts: high-level and low-level. Higl-level is mostly method calls, low-level publicly visible pointers to DF's data structures.
  • Plugins are the tools that use all the other stuff to make things happen. A plugin can have a list of commands that it exports and an onupdate function that will be called each DF game tick.

Rudimentary API documentation can be built using doxygen (see build options with ccmake or cmake-gui).

DFHack consists of variously licensed code, but invariably weak copyleft. The main license is zlib/libpng, some bits are MIT licensed, and some are BSD licensed.

Feel free to add your own extensions and plugins. Contributing back to the dfhack repository is welcome and the right thing to do :)

DF data structure definitions

DFHack uses information about the game data structures, represented via xml files in the library/xml/ submodule.

Data structure layouts are described in files following the df.*.xml name pattern. This information is transformed by a perl script into C++ headers describing the structures, and associated metadata for the Lua wrapper. These headers and data are then compiled into the DFHack libraries, thus necessitating a compatibility break every time layouts change; in return it significantly boosts the efficiency and capabilities of DFHack code.

Global object addresses are stored in symbols.xml, which is copied to the dfhack release package and loaded as data at runtime.

Remote access interface

DFHack supports remote access by exchanging Google protobuf messages via a TCP socket. Both the core and plugins can define remotely accessible methods. The dfhack-run command uses this interface to invoke ordinary console commands.

Currently the supported set of requests is limited, because the developers don't know what exactly is most useful.

Protocol client implementations exist for Java and C#.

Contributing to DFHack

Several things should be kept in mind when contributing to DFHack.

Coding style

DFhack uses ANSI formatting and four spaces as indentation. Line endings are UNIX. The files use UTF-8 encoding. Code not following this won't make me happy, because I'll have to fix it. There's a good chance I'll make you fix it ;)

How to get new code into DFHack

You can send patches or make a clone of the github repo and ask me on the IRC channel to pull your code in. I'll review it and see if there are any problems. I'll fix them if they are minor.

Fixes are higher in priority. If you want to work on something, but don't know what, check out http://github.com/peterix/dfhack/issues --this is also a good place to dump new ideas and/or bugs that need fixing.

Memory research

If you want to do memory research, you'll need some tools and some knowledge. In general, you'll need a good memory viewer and optionally something to look at machine code without getting crazy :)

Good windows tools include:

  • Cheat Engine
  • IDA Pro (the free version)

Good linux tools:

  • angavrilov's df-structures gui (visit us on IRC for details).
  • edb (Evan's Debugger)
  • IDA Pro running under wine.
  • Some of the tools residing in the legacy dfhack branch.

Using publicly known information and analyzing the game's data is preferred.