modules/llvm.py

import os
import re

from linaropy.git.worktree import Worktree


class LLVMSubproject(object):
    """
    Class that keeps track of everything related to an LLVM subproject (clang,
    lld, compiler-rt etc): repo URL, location preferred by CMake,
    CMake variable for adding or removing it from the build etc.
    """

    def __init__(self, cmake_path, cmake_var):
        """Create an LLVMSubproject with the provided info.

        Parameters
        ----------
        cmake_path
            Path relative to the LLVM source root where this subproject should
            live so that CMake can automatically pick it up during the build.
        cmake_var
            The name of the CMake variable that can be used to enable/disable
            building this subproject.
        """
        self.cmake_path = cmake_path
        self.cmake_var = cmake_var

    def get_cmake_path(self, llvm_source_directory):
        """
        Get the path where this subproject should live in the given LLVM tree so
        that CMake can pick it up by default.
        """
        return os.path.join(llvm_source_directory, self.cmake_path)

    @classmethod
    def get_all_subprojects(cls):
        """
        Return a dictionary of all the LLVM subprojects.
        At the moment, llvm itself is not part of the subprojects, because it
        always needs to be there and everything is relative to it.
        """
        return {
            "clang":
                cls(os.path.join("tools", "clang"), "LLVM_TOOL_CLANG_BUILD"),
            "compiler-rt":
                cls(os.path.join("projects", "compiler-rt"),
                    "LLVM_TOOL_COMPILER_RT_BUILD"),
            "libcxx":
                cls(os.path.join("projects", "libcxx"),
                    "LLVM_TOOL_LIBCXX_BUILD"),
            "libcxxabi":
                cls(os.path.join("projects", "libcxxabi"),
                    "LLVM_TOOL_LIBCXXABI_BUILD"),
            "libunwind":
                cls(os.path.join("projects", "libunwind"),
                    "LLVM_TOOL_LIBUNWIND_BUILD"),
            "lld":
                cls(os.path.join("tools", "lld"), "LLVM_TOOL_LLD_BUILD"),
            "lldb":
                cls(os.path.join("tools", "lldb"), "LLVM_TOOL_LLDB_BUILD"),
            "test-suite":
                cls(os.path.join("projects", "test-suite"), None),
        }


class LLVMSourceConfig(object):
    """Class for managing an LLVM source tree.

    It keeps track of which subprojects are enabled in a given tree and provides
    functionality for adding / removing them.
    """

    def __init__(self, proj, sourcePath,
                 subprojs=LLVMSubproject.get_all_subprojects()):
        """ Create a source configuration.

        Parameters
        ----------
        proj : Proj
            Temporary project directory (used mostly for logging).
        sourcePath
            Must point to a valid LLVM source tree.
        subprojs : dictionary
            Dictionary containing a number of LLVMSubproject objects.
            By default, this contains all the LLVM subprojects as returned by
            LLVMSubproject.get_all_subprojects(), but any subset would work just
            as well.
            The keys will be used to identify the subprojects in any of this
            class's methods. It is an error to invoke them with a subproj that
            does not exist in this dictionary.
        """
        sourcePath = str(sourcePath)
        if not os.path.isdir(sourcePath):
            raise EnvironmentError("Invalid path to LLVM source tree")

        self.proj = proj
        self.llvmSourceTree = Worktree(proj, sourcePath)
        self.subprojs = subprojs

    def get_enabled_subprojects(self):
        """Get a list of the subprojects enabled in this configuration."""
        enabled = []
        for subproj in self.subprojs:
            if self.__is_enabled(subproj):
                enabled.append(subproj)
        return enabled

    def update(self, add={}, remove=[]):
        """Update the configuration by adding/removing subprojects.

        Parameters
        ----------
        add : dictionary
            A map of (subproject name, repo) to add to the config. The order in
            which the subprojects are added is unspecified. Subprojects that are
            already enabled in the config are ignored.
        remove: list
            A list of subproject names to remove from the config. The order in
            which the subprojects are removed is unspecified. Duplicates and
            subprojects that aren't enabled in the config are ignored. A
            subproject may be removed even if it is in an invalid state.

        For both add and remove, the subproject name must exist in the
        dictionary that the config was initialized with and the repo must be a
        valid GitRepo object.

        TODO: If adding/removing a project fails, this function should try to
        recover.
        """

        # Make the inputs friendly
        if add is None:
            add = {}

        if remove is None:
            remove = []

        for subproj in list(add.keys()):
            if subproj in remove:
                raise ValueError("Can't add and remove {} at the same time"
                                 .format(subproj))

        for (subproj, repo) in list(add.items()):
            self.__add_subproject(subproj, repo)

        for subproj in remove:
            self.__remove_subproject(subproj)

    def for_each_enabled(self, action):
        """Perform the given action for each enabled subproject including LLVM.

        The action must be a callable object receiving the path to the
        subproject's directory as its only parameter.

        If the action throws an exception, it will be rethrown as a
        RuntimeError. Note that this does not have transactional behaviour.
        """
        for subproj in self.get_enabled_subprojects():
            try:
                action(self.__get_subproj_cmake_path(subproj))
            except Exception as exc:
                raise RuntimeError("Error while processing {}".format(subproj)) from exc

        # Visit LLVM last, in case getting the enabled subprojects errors out.
        action(self.llvmSourceTree.repodir)

    def __get_subproj_object(self, subprojName):
        """Get the LLVMSubproject object corresponding to subprojName."""
        if not subprojName in list(self.subprojs.keys()):
            raise ValueError("Unknown llvm subproject {0}".format(subprojName))
        return self.subprojs[subprojName]

    def __get_subproj_cmake_path(self, subprojName):
        """Get the full path to subprojName in this source tree."""
        subproj = self.__get_subproj_object(subprojName)
        return subproj.get_cmake_path(self.llvmSourceTree.repodir)

    def __is_enabled(self, subprojName):
        """
        Check if subproj is enabled in this configuration. A subproj is
        considered to be enabled if it is a worktree on the correct branch. If
        a directory for the subproject exists but does not satisfy those
        conditions, an EnvironmentError is thrown.
        """
        subprojPath = self.__get_subproj_cmake_path(subprojName)

        if not os.path.isdir(subprojPath):
            return False

        existing = Worktree(self.proj, subprojPath)

        if existing.getbranch() != self.llvmSourceTree.getbranch():
            raise EnvironmentError("{} is on branch {}, but should be on {}".format(
                subprojName, existing.getbranch(), self.llvmSourceTree.getbranch()))

        return True

    # TODO: add_subproject, remove_subproject and is_enabled should live in
    # another object (AddRemoveStrategy?) that would know what we want to add
    # (worktrees, links, whatever)
    def __add_subproject(self, subprojName, subprojRepo):
        """Add a given subproject to this configuration.

        This will make sure the subproject's sources are visible in the proper
        place in the LLVM source tree that the configuration was created with.

        We currently achieve this by creating a worktree for the subproject, but
        if more flexibility is needed we can add a config option. The branch
        that the worktree will be created with is the same branch that the
        existing LLVM source tree is on, and it will be tracking a branch
        corresponding to the one that the LLVM branch was forked from.
        """
        path = self.__get_subproj_cmake_path(subprojName)

        if self.__is_enabled(subprojName):
            # Subproject has already been added, nothing to do.
            return

        if os.path.exists(path):
            raise EnvironmentError(
                "{} already exists but is not a valid subproject directory."
                    .format(path) +
                "Please make sure it is a worktree on the same branch as LLVM.")

        # Create a new worktree
        branch = self.llvmSourceTree.getbranch()
        if subprojRepo.branch_exists(branch):
            Worktree.create(self.proj, subprojRepo, path, branch)
        else:
            trackedBranch = "master"  # TODO: track proper branch
            Worktree.create(
                self.proj,
                subprojRepo,
                path,
                branch,
                trackedBranch)

    def __remove_subproject(self, subprojName):
        """Remove a given subproject from this build configuration."""
        path = self.__get_subproj_cmake_path(subprojName)

        if not os.path.isdir(path):
            return

        worktree = Worktree(self.proj, path)
        worktree.clean(False)


# FIXME: repo.pushToBranch doesn't work, because it doesn't handle remote
# branches properly. Furthermore, there's no support for getting the remote URL,
# so we need to resort to raw git commands. We may also consider moving the
# functionality for parsing info out of the remote URL into the repo object.
from sh import git
from linaropy.cd import cd


def get_user_from_remote(url):
    """Get the username used as part of the remote URL, or None.

    The remote URLs that we expect to see look like $protocol://$user@$location.
    If they look any different, we won't be able to parse them.
    """
    pattern = re.compile("(.*://)?(?P<user>.*)@(.*)\n?")
    match = pattern.match(str(url))
    if match is None:
        return None
    return match.group('user')


def push_current_branch(proj, pathToRepo):
    """Push the current branch to origin.

    It will be pushed into linaro-local/$user/$branch, where $branch is the
    current branch and $user is the username used as part of the remote URL.
    """
    repo = Worktree(proj, pathToRepo)

    with cd(repo.repodir):
        remote = git("remote", "get-url", "--push", "origin").strip()
    user = get_user_from_remote(remote)
    if not user:
        raise EnvironmentError("Couldn't parse user from {}.".format(remote))

    local_branch = repo.getbranch()
    remote_branch = "linaro-local/{}/{}".format(user, local_branch)
    if not repo.is_valid_branch_name(remote_branch):
        raise EnvironmentError(
            "{} is not a valid branch name.".format(remote_branch))

    with cd(repo.repodir):
        git("push", "-u", "origin", "+{}:{}".format(local_branch, remote_branch))