gen_compile_commands.py

#!/usr/bin/env python
import json
import subprocess
import os
import sys


def resolve_depset_to_paths(artifacts, path_fragments, depset):
    all_inputs = []
    for artifact_id in depset["directArtifactIds"]:
        artifact = artifacts[artifact_id]

        path = []
        path_fragment_id = artifact.get("pathFragmentId")
        while path_fragment_id is not None:
            path_fragment = path_fragments[path_fragment_id]
            path.append(path_fragment["label"])
            path_fragment_id = path_fragment.get("parentId")

        path = os.path.join(*reversed(path))
        all_inputs.append(path)

    return all_inputs


def main():

    ROOT = os.path.realpath(
        os.path.join(os.path.dirname(os.path.realpath(__file__)), "..")
    )

    # TODO do we need CppLink CppLTOIndexing ?
    cmd = [
        "bazel",
        "aquery",
        'mnemonic("CppCompile", //...)',
        "--output",
        "jsonproto",
    ]
    ret = subprocess.check_output(cmd)
    ret = ret.decode("utf-8")

    in_data = json.loads(ret)
    dep_set_of_files = {d["id"]: d for d in in_data["depSetOfFiles"]}
    artifacts = {d["id"]: d for d in in_data["artifacts"]}
    path_fragments = {d["id"]: d for d in in_data["pathFragments"]}

    outdata = []
    for action in in_data["actions"]:

        # resolve input paths
        all_inputs = []
        for ds_id in action["inputDepSetIds"]:
            depset = dep_set_of_files[ds_id]
            all_inputs.extend(
                resolve_depset_to_paths(artifacts, path_fragments, depset)
            )

            for ds_id in depset.get("transitiveDepSetIds", []):
                all_inputs.extend(
                    resolve_depset_to_paths(
                        artifacts, path_fragments, dep_set_of_files[ds_id]
                    )
                )

        # resolve output path
        artifact = artifacts[action["primaryOutputId"]]
        path_fragment_id = artifact["pathFragmentId"]
        path_fragment = path_fragments[path_fragment_id]

        output_path = []
        while "parentId" in path_fragment:
            output_path.append(path_fragment["label"])
            path_fragment = path_fragments[path_fragment["parentId"]]
        output_path = os.path.join(*reversed(output_path))

        cc_inputs = list(
            filter(
                lambda x: x.endswith(".c")
                or x.endswith(".cxx")
                or x.endswith(".cc")
                or x.endswith(".cpp"),
                all_inputs,
            )
        )

        assert len(cc_inputs) == 1
        command = action["arguments"]
        cc_inputs[0] in command, "Input should be on the command line"
        outdata.append(
            {
                "command": command,
                "directory": ROOT,
                "file": cc_inputs[0],
                "output": output_path,
            }
        )

    outfile = os.path.join(ROOT, "compile_commands.json")
    with open(outfile, "w") as f:
        json.dump(outdata, f)


if __name__ == "__main__":
    sys.exit(main())