Skip to content

Tool to generate compile_commands.json from the Bazel build system

License

Notifications You must be signed in to change notification settings

specto-dev/bazel-compilation-database

 
 

Repository files navigation

Compilation database with Bazel Tests Migration

If you use Bazel and want to use libclang based editors and tools, you can now generate JSON compilation database easily without using build intercept hooks. The advantage is that you can generate the database even if your source code does not compile, and the generation process is much faster.

For more information on compilation database, Guillaume Papin has an excellent article.

How to Use

Entire repo

Running generate.sh script from this project with current directory somewhere in your bazel workspace will generate a compile_commands.json file in the top-level directory of your workspace. You can even symlink the script to somewhere in your PATH.

For example,

INSTALL_DIR="/usr/local/bin"
VERSION="0.4.5"

# Download and symlink.
(
  cd "${INSTALL_DIR}" \
  && curl -L "https://github.com/grailbio/bazel-compilation-database/archive/${VERSION}.tar.gz" | tar -xz \
  && ln -f -s "${INSTALL_DIR}/bazel-compilation-database-${VERSION}/generate.sh" bazel-compdb
)

bazel-compdb # This will generate compile_commands.json in your workspace root.

# You can tweak some behavior with flags:
# 1. To use the source dir instead of bazel-execroot for directory in which clang commands are run.
bazel-compdb -s

Selected targets

You can define a target of rule type compilation_database with the attribute targets as a list of top-level cc_.* labels which you want to include in your compilation database. You do not need to include targets that are dependencies of your top-level targets. So these will mostly be targets of type cc_binary and cc_test.

For example,

In your WORKSPACE file:

# Change master to the git tag you want.
http_archive(
    name = "com_grail_bazel_compdb",
    strip_prefix = "bazel-compilation-database-master",
    urls = ["https://github.com/grailbio/bazel-compilation-database/archive/master.tar.gz"],
)

In your BUILD file located in any package:

## Replace workspace_name and dir_path as per your setup.
load("@com_grail_bazel_compdb//:aspects.bzl", "compilation_database")

compilation_database(
    name = "example_compdb",
    targets = [
        "//a_cc_binary_label",
        "//a_cc_library_label",
    ],
    # [Optional]
    # If your exec root (value returned by `bazel info execution_root`)
    # is constant across your users, then you can supply the value here.
    # Otherwise, the default is `__EXEC_ROOT__` which you can replace in
    # the output file using `sed` or similar tool (see below).
    exec_root = "/path/to/bazel/exec_root",
)

Then, in your terminal (you can wrap this in a shell script and check it in your repo):

# Command to generate the compilation database file.
bazel build //path/to/pkg/dir:example_compdb

# Location of the compilation database file.
outfile="$(bazel info bazel-bin)/path/to/pkg/dir/compile_commands.json"

# Command to replace the marker for exec_root in the file.
execroot=$(bazel info execution_root)
sed -i.bak "s@__EXEC_ROOT__@${execroot}@" "${outfile}"

# The compilation database is now ready to use at this location.
echo "Compilation Database: ${outfile}"

YouCompleteMe

If you want to use this project solely for semantic auto completion using ycmd (YouCompleteMe) based editor plugins, then the easiest approach is to install this project as a vim plugin with your favourite plugin manager. The plugin will set g:ycm_global_ycm_extra_conf and instrument bazel with the correct paths. e.g. Using Plugged add the following to your vimrc.

Plug 'grailbio/bazel-compilation-database'

An alternative approach is to follow the instructions as above for making the files available in this repo somewhere in the workspace, and then configure vim to use the .ycm_extra_conf.py script that you just extracted. One way is to make a symlink to the py script from the top of your workspace root. Another way is to set the ycm_global_ycm_extra_conf variable in vim.

With both of these approaches, you don't have to maintain a separate compile_commands.json file through a script and/or a compilation_database target. Compile commands are fetched from bazel as the files are opened in your editor.

Contributing

Contributions are most welcome. Please submit a pull request giving the owners of this github repo access to your branch for minor style related edits, etc.

Known Issues

Please check open issues at the github repo.

We have tested only for C and C++ code, and with tools like YouCompleteMe, rtags, and the woboq code browser.

Alternatives

  1. Kythe: uses Bazel action listeners
  2. Bear: uses build intercept hooks

These approaches could be more accurate than the approach of this tool in some rare cases, but need a more complicated setup and a full build every time you refresh the database.

About

Tool to generate compile_commands.json from the Bazel build system

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Starlark 47.7%
  • Python 25.1%
  • Shell 20.4%
  • C++ 6.6%
  • Vim Script 0.2%