Static code analysis of any autotools project with oclint
The following is a HowTo describing the setup of OCLint for any C/C++ project using autotools.
1. OCLint Setup
First step is downloading OCLint, as there are no package so far, it's just extracting the tarball somewhere in $HOME. Check out the latest release link on http://archives.oclint.org/releases/.cd wget "http://archives.oclint.org/releases/0.8/oclint-0.8.1-x86_64-linux-3.13.0-35-generic.tar.gz" tar zxvf oclint-0.8.1-x86_64-linux-3.13.0-35-generic.tar.gzThis should leave you with a copy of OCLint in ~/oclint-0.8.1
2. Bear Setup
As project usually consist of a lot of source files in different subdirectories it is hard for a linter to know where to look for files. While "cmake" has support for dumping a list of source files it processes during a run "make" doesn't. This is where the "Bear" wrapper comes to play: instead ofmakeyou run
bear makeso "bear" can track all files being compiled. It will dump a JSON file "compile_commands.json" which OCLint can use to do analysis of all files. To setup Bear do the following
cd git clone https://github.com/rizsotto/Bear.git cd Bear cmake . make
3. Analyzing Code
Now we have all the tools we need. Let's download some autotools project like Liferea. Before doing code analysis it should be downloaded and build at least once:git clone https://github.com/lwindolf/liferea.git cd liferea sh autogen.sh makeNow we collect all code file compilation instructions with bear:
make clean bear makeAnd if this succeed we can start a complete analysis with
~/oclint-0.8.1/bin/oclint-json-compilation-databasewhich will run OCLint with the input from "compile_commands.json" produced by "bear". Don't call "oclint" directly as you'd need to pass all compile flags manually. If all went well you could see code analysis lines like those:
[...] conf.c:263:9: useless parentheses P3 conf.c:274:9: useless parentheses P3 conf.c:284:9: useless parentheses P3 conf.c:46:1: high cyclomatic complexity P2 Cyclomatic Complexity Number 33 exceeds limit of 10 conf.c:157:1: high cyclomatic complexity P2 Cyclomatic Complexity Number 12 exceeds limit of 10 conf.c:229:1: high cyclomatic complexity P2 Cyclomatic Complexity Number 30 exceeds limit of 10 conf.c:78:1: long method P3 Method with 55 lines exceeds limit of 50 conf.c:50:2: short variable name P3 Variable name with 2 characters is shorter than the threshold of 3 conf.c:52:2: short variable name P3 Variable name with 1 characters is shorter than the threshold of 3 [...]