PC-lint (plus) invoker from JSON compilation databases

Big picture

To check existing codebases with PC-lint (plus) one has to:

• (according to manual) prepare a local PC-lint setup
  • creating a lin.bat or lin.sh that invokes PC-lint (plus)
  • create a std.lnt that includes co.lnt and options.lnt
  • create a co.lnt that sets up the compiler being used
  • create an options.lnt that sets up the output format and
  • create a suppressions.lnt for project-specific suppressions
• invoke lint with all include paths and defines per compilation-unit.

In order to integrate with existing build chains this tool can be used to execute PC-lint using the build job's JSON compilation database.

Usage

Invoke lint4jsondb.py

W:\> lint4jsondb.py --compilation-db commands.json --lint-path D:\pclp --lint-binary pclp64.exe --jobs 6 -- std.lnt

where

• --compilation-db points to your build system's JSON compilation database
• --lint-path points to you PC-lint root path (that contains the binaries and the lnt directory)
• --lint-binary names the PC-lint binary you want to execute (either pclp32, pclp64 or lint-nt)
• --jobs is the number of parallel PC-lint jobs to spawn (defaults to number of CPU core)
• -- and everything after it will be passed to the PC-lint binary; use it to point to your std.lnt

Optionally you can control which files of the overall JSON compilation database SHALL be processed using

• --include-only <regexp> will include only those files whose full file-path matches <regexp>
• --exclude-all <regexp> will additionally exclude all those files whose full file-path matches <regexp>

Additionally you can control how PC-lint is invoked via --exec-mode

• either once per file using each
• or once per JSON compilation database using all

It is recommended to use all (the default) as this allows PC-lint to further track member usage inter-dependencies.

Further notes

• might work with FlexeLint as well (I don't have one to test with)

Prerequisites

• Python >= 2.7
• Package ijson (for parsing JSON compilation databases on the fly)
• Package mock (mocking calls, since Python 3.3 shipped with it)

JSON compilation database

The format specification for JSON compilation databases is available from LLVM.

...with CMake

Set CMAKE_EXPORT_COMPILE_COMMANDS to ON - or just pass -DCMAKE_EXPORT_COMPILE_COMMANDS=ON when invoking cmake.

For more information refer to the CMake manual.

...with make-based projects

Use Bear, see Bear on github.

...with qbs

Run qbs in generate mode and use the clangdb generator

$ > qbs generate -g clangdb

Refer to qbs manual on generators