First ensure you have Bazel installed, for example with

brew install bazelisk

then from the ql directory run

bazel run //swift:install

If you are running on macOS and you encounter errors mentioning XXX is unavailable: introduced in macOS YY.ZZ, you will need to run this from the root of your codeql checkout:

echo common --macos_sdk_version=$(sw_vers --productVersion) >> local.bazelrc

which will install swift/extractor-pack.

Notice you can run bazel run :create-extractor-pack if you already are in the swift directory.

Using codeql ... --search-path=swift/extractor-pack will then pick up the Swift extractor. You can also use --search-path=., as the extractor pack is mentioned in the root codeql-workspace.yml. Alternatively, you can set up the search path in the per-user CodeQL configuration file .

Run

bazel run //swift/codegen

to update generated files. This can be shortened to bazel run codegen if you are in the swift directory.

You can also run ../misc/codegen/codegen.py, as long as you are beneath the swift directory.

A log file is produced for each run under CODEQL_EXTRACTOR_SWIFT_LOG_DIR (the usual DB log directory).

You can use the environment variable CODEQL_EXTRACTOR_SWIFT_LOG_LEVELS to configure levels for loggers and outputs. This must have the form of a comma separated spec:min_level list, where spec is either a glob pattern (made up of alphanumeric, /, * and . characters) for matching logger names or one of out:binary, out:text, out:console or out:diagnostics, and min_level is one of trace, debug, info, warning, error, critical or no_logs to turn logs completely off.

Current output default levels are no binary logs, info logs or higher in the text file and warning logs or higher on standard error. By default, all loggers are configured with the lowest logging level of all outputs (info by default). Logger names are visible in the textual logs between [...]. Examples are extractor/dispatcher or extractor/<source filename>.trap. An example of CODEQL_EXTRACTOR_SWIFT_LOG_LEVELS usage is the following:

export CODEQL_EXTRACTOR_SWIFT_LOG_LEVELS=out:console:trace,out:text:no_logs,*:warning,*.trap:trace

This will turn off generation of a text log file, redirecting all logs to standard error, but will make all loggers only write warnings or above, except for trap emission logs which will output all logs.

You can use CLion with the official IntelliJ Bazel plugin, creating the project from scratch with default options. This is known to have issues on non-Linux platforms.

The CMakeLists.txt file allows to load the Swift extractor as a CMake project, which allows integration into a wider variety of IDEs. Building with CMake also creates a compile_commands.json compilation database that can be picked up by even more IDEs. In particular, opening the swift directory in VSCode should work.

If you want to debug a specific run of the extractor within an integration test or a complex codeql database create invocation, you can do so using gdbserver or lldb-server.

For example with gdbserver, you can

export CODEQL_EXTRACTOR_SWIFT_RUN_UNDER="gdbserver :1234"
export CODEQL_EXTRACTOR_SWIFT_RUN_UNDER_FILTER="SomeSwiftSource\.swift"  # can be any regex matching extractor args

before starting the database extraction, and when that source is encountered the extractor will be run under a gdbserver instance listening on port 1234. You can then attach to the running debugging server from gdb or your IDE. Please refer to your IDE's instructions for how to set up remote debugging.

In particular for breakpoints to work you might need to setup the following remote path mapping:

The extractor is single-threaded, and there was no effort to make anything in it thread-safe.

This can only be done with access to the internal repository at the moment. Some (incomplete) instructions are found here.