GNOME.org
 

[Notes] [Git][BuildGrid/buildgrid][mablanch/61-bazel-support] 4 commits: utils.py: New gRPC OutputFile maker helper

Title: GitLab

Martin Blanchard pushed to branch mablanch/61-bazel-support at BuildGrid / buildgrid

Commits:

2 changed files:

Changes:

  • buildgrid/_app/bots/temp_directory.py
    ... ... @@ -19,7 +19,7 @@ import tempfile
    19 19
    20 20 
     from google.protobuf import any_pb2
    21 21
    22 
    -from buildgrid.utils import read_file, create_digest, write_fetch_directory, parse_to_pb2_from_fetch
    22 
    +from buildgrid.utils import output_file_maker, write_fetch_directory, parse_to_pb2_from_fetch
    23 23 
     from buildgrid._protos.build.bazel.remote.execution.v2 import remote_execution_pb2, remote_execution_pb2_grpc
    24 24 
     from buildgrid._protos.google.bytestream import bytestream_pb2_grpc
    25 25
    ... ... @@ -35,54 +35,81 @@ def work_temp_directory(context, lease):
    35 35 
         action_digest = remote_execution_pb2.Digest()
    36 36 
         lease.payload.Unpack(action_digest)
    37 37
    38 
    -    action = remote_execution_pb2.Action()
    38 
    +    action = parse_to_pb2_from_fetch(remote_execution_pb2.Action(),
    39 
    +                                     stub_bytestream, action_digest, instance_name)
    39 40
    40 
    -    action = parse_to_pb2_from_fetch(action, stub_bytestream, action_digest, instance_name)
    41 
    +    with tempfile.TemporaryDirectory() as temp_directory:
    41 42
    42 
    -    with tempfile.TemporaryDirectory() as temp_dir:
    43 
    +        command = parse_to_pb2_from_fetch(remote_execution_pb2.Command(),
    44 
    +                                          stub_bytestream, action.command_digest, instance_name)
    43 45
    44 
    -        command = remote_execution_pb2.Command()
    45 
    -        command = parse_to_pb2_from_fetch(command, stub_bytestream, action.command_digest, instance_name)
    46 
    +        write_fetch_directory(temp_directory, stub_bytestream,
    47 
    +                              action.input_root_digest, instance_name)
    46 48
    47 
    -        arguments = "cd {} &&".format(temp_dir)
    49 
    +        execution_envionment = os.environ.copy()
    50 
    +        for variable in command.environment_variables:
    51 
    +            if variable.name not in ['PATH', 'PWD']:
    52 
    +                execution_envionment[variable.name] = variable.value
    48 53
    54 
    +        command_arguments = list()
    49 55 
             for argument in command.arguments:
    50 
    -            arguments += " {}".format(argument)
    51 
    -
    52 
    -        context.logger.info(arguments)
    53 
    -
    54 
    -        write_fetch_directory(temp_dir, stub_bytestream, action.input_root_digest, instance_name)
    55 
    -
    56 
    -        proc = subprocess.Popen(arguments,
    57 
    -                                shell=True,
    58 
    -                                stdin=subprocess.PIPE,
    59 
    -                                stdout=subprocess.PIPE)
    60 
    -
    61 
    -        # TODO: Should return the std_out to the user
    62 
    -        proc.communicate()
    63 
    -
    64 
    -        result = remote_execution_pb2.ActionResult()
    65 
    -        requests = []
    66 
    -        for output_file in command.output_files:
    67 
    -            path = os.path.join(temp_dir, output_file)
    68 
    -            chunk = read_file(path)
    69 
    -
    70 
    -            digest = create_digest(chunk)
    71 
    -
    72 
    -            result.output_files.extend([remote_execution_pb2.OutputFile(path=output_file,
    73 
    -                                                                        digest=digest)])
    74 
    -
    75 
    -            requests.append(remote_execution_pb2.BatchUpdateBlobsRequest.Request(
    76 
    -                digest=digest, data=chunk))
    77 
    -
    78 
    -        request = remote_execution_pb2.BatchUpdateBlobsRequest(instance_name=instance_name,
    79 
    -                                                               requests=requests)
    56 
    +            command_arguments.append(argument.strip())
    57 
    +
    58 
    +        working_directory = None
    59 
    +        if command.working_directory:
    60 
    +            working_directory = os.path.join(temp_directory,
    61 
    +                                             command.working_directory)
    62 
    +            os.makedirs(working_directory, exist_ok=True)
    63 
    +        else:
    64 
    +            working_directory = temp_directory
    65 
    +
    66 
    +        # Ensure that output files structure exists:
    67 
    +        for output_path in command.output_files:
    68 
    +            directory_path = os.path.join(working_directory,
    69 
    +                                          os.path.dirname(output_path))
    70 
    +            os.makedirs(directory_path, exist_ok=True)
    71 
    +
    72 
    +        process = subprocess.Popen(command_arguments,
    73 
    +                                   cwd=working_directory,
    74 
    +                                   universal_newlines=True,
    75 
    +                                   env=execution_envionment,
    76 
    +                                   stdin=subprocess.PIPE,
    77 
    +                                   stdout=subprocess.PIPE)
    78 
    +        # TODO: Should return the stdout and stderr to the user.
    79 
    +        process.communicate()
    80 
    +
    81 
    +        update_requests = remote_execution_pb2.BatchUpdateBlobsRequest(instance_name=instance_name)
    82 
    +        action_result = remote_execution_pb2.ActionResult()
    83 
    +
    84 
    +        for output_path in command.output_files:
    85 
    +            file_path = os.path.join(working_directory, output_path)
    86 
    +            # Missing outputs should simply be omitted in ActionResult:
    87 
    +            if not os.path.isfile(file_path):
    88 
    +                continue
    89 
    +
    90 
    +            # OutputFile.path should be relative to the working direcory:
    91 
    +            output_file, update_request = output_file_maker(file_path, working_directory)
    92 
    +
    93 
    +            action_result.output_files.extend([output_file])
    94 
    +            update_requests.requests.extend([update_request])
    95 
    +
    96 
    +        for output_path in command.output_directories:
    97 
    +            directory_path = os.path.join(working_directory, output_path)
    98 
    +            # Missing outputs should simply be omitted in ActionResult:
    99 
    +            if not os.path.isdir(directory_path):
    100 
    +                continue
    101 
    +
    102 
    +            # OutputDirectory.path should be relative to the working direcory:
    103 
    +            output_directory, update_request = output_directory_maker(directory_path, working_directory)
    104 
    +
    105 
    +            action_result.output_directories.extend([output_directory])
    106 
    +            update_requests.requests.extend(update_request)
    80 107
    81 108 
             stub_cas = remote_execution_pb2_grpc.ContentAddressableStorageStub(context.channel)
    82 
    -        stub_cas.BatchUpdateBlobs(request)
    109 
    +        stub_cas.BatchUpdateBlobs(update_requests)
    83 110
    84 111 
             result_any = any_pb2.Any()
    85 
    -        result_any.Pack(result)
    112 
    +        result_any.Pack(action_result)
    86 113
    87 114 
             lease.result.CopyFrom(result_any)
    88 115

  • buildgrid/utils.py
    ... ... @@ -99,7 +99,15 @@ def parse_to_pb2_from_fetch(pb2, stub, digest, instance_name=""):
    99 99
    100 100
    101 101 
     def create_digest(bytes_to_digest):
    102 
    -    """ Creates a hash based on the hex digest and returns the digest
    102 
    +    """Computes the :obj:`Digest` of a piece of data.
    103 
    +
    104 
    +    The :obj:`Digest` of a data is a function of its hash **and** size.
    105 
    +
    106 
    +    Args:
    107 
    +        bytes_to_digest (bytes): byte data to digest.
    108 
    +
    109 
    +    Returns:
    110 
    +        :obj:`Digest`: The gRPC :obj:`Digest` for the given byte data.
    103 111 
         """
    104 112 
         return remote_execution_pb2.Digest(hash=HASH(bytes_to_digest).hexdigest(),
    105 113 
                                            size_bytes=len(bytes_to_digest))
    ... ... @@ -136,6 +144,187 @@ def file_maker(file_path, file_digest):
    136 144 
                                              is_executable=os.access(file_path, os.X_OK))
    137 145
    138 146
    139 
    -def read_file(read):
    140 
    -    with open(read, 'rb') as f:
    141 
    -        return f.read()
    147 
    +def directory_maker(directory_path):
    148 
    +    """
    149 
    +    """
    150 
    +    if not os.path.isabs(directory_path):
    151 
    +        directory_path = os.path.abspath(directory_path)
    152 
    +
    153 
    +    child_directories = list()
    154 
    +    update_requests = list()
    155 
    +
    156 
    +    files, directories, symlinks = list(), list(), list()
    157 
    +    for directory_entry in os.scandir(directory_path):
    158 
    +        # Create a FileNode and corresponding BatchUpdateBlobsRequest:
    159 
    +        if directory_entry.is_file(follow_symlinks=False):
    160 
    +            node_blob = read_file(directory_entry.path)
    161 
    +            node_digest = create_digest(node_blob)
    162 
    +
    163 
    +            node = remote_execution_pb2.FileNode()
    164 
    +            node.name = directory_entry.name
    165 
    +            node.digest = node_digest
    166 
    +            node.is_executable = os.access(directory_entry.path, os.X_OK)
    167 
    +
    168 
    +            node_request = remote_execution_pb2.BatchUpdateBlobsRequest.Request(digest=node_digest)
    169 
    +            node_request.data = node_blob
    170 
    +
    171 
    +            update_requests.append(node_request)
    172 
    +            files.append(node)
    173 
    +
    174 
    +        # Create a DirectoryNode and corresponding BatchUpdateBlobsRequest:
    175 
    +        elif directory_entry.is_dir(follow_symlinks=False):
    176 
    +            node_directory, _, node_requests = directory_maker(directory_entry.path)
    177 
    +
    178 
    +            node = remote_execution_pb2.DirectoryNode()
    179 
    +            node.name = directory_entry.name
    180 
    +            node.digest = node_requests[-1].digest
    181 
    +
    182 
    +            child_directories.append(node_directory)
    183 
    +            update_requests.expend(node_requests)
    184 
    +            directories.append(node)
    185 
    +
    186 
    +        # Create a SymlinkNode if necessary;
    187 
    +        elif os.path.islink(directory_entry.path):
    188 
    +            node_target = os.readlink(directory_entry.path)
    189 
    +
    190 
    +            node = remote_execution_pb2.SymlinkNode()
    191 
    +            node.name = directory_entry.name
    192 
    +            node.target = node_target
    193 
    +
    194 
    +            symlinks.append(node)
    195 
    +
    196 
    +    directory = remote_execution_pb2.Directory()
    197 
    +    directory.files.extend(files.sort())
    198 
    +    directory.directories.extend(directories.sort())
    199 
    +    directory.symlinks.extend(symlinks.sort())
    200 
    +
    201 
    +    directory_blob = directory.SerializeToString()
    202 
    +    directory_digest = create_digest(directory_blob)
    203 
    +
    204 
    +    update_request = remote_execution_pb2.BatchUpdateBlobsRequest.Request(digest=directory_digest)
    205 
    +    update_request.data = directory_blob
    206 
    +
    207 
    +    update_requests.append(update_request)
    208 
    +
    209 
    +    return directory, child_directories, update_requests
    210 
    +
    211 
    +
    212 
    +def read_file(file_path):
    213 
    +    """Loads raw file content in memory.
    214 
    +
    215 
    +    Returns:
    216 
    +        bytes: Raw file's content until EOF.
    217 
    +
    218 
    +    Raises:
    219 
    +        OSError: If `file_path` does not exist or is not readable.
    220 
    +    """
    221 
    +    with open(file_path, 'rb') as byte_file:
    222 
    +        return byte_file.read()
    223 
    +
    224 
    +
    225 
    +def output_file_maker(file_path, input_path):
    226 
    +    """Creates an :obj:`OutputFile` from a local file.
    227 
    +
    228 
    +    `file_path` **must** point inside or be relative to `input_path`.
    229 
    +
    230 
    +    Args:
    231 
    +        file_path (str): absolute or relative path to a local file.
    232 
    +        input_path (str): absolute or relative path to the input root directory.
    233 
    +
    234 
    +    Returns:
    235 
    +        :obj:`OutputFile`, :obj:`BatchUpdateBlobsRequest`: Tuple of a new gRPC
    236 
    +        :obj:`OutputFile` object for the file pointed by `file_path` and the
    237 
    +        corresponding :obj:`BatchUpdateBlobsRequest` for CAS upload.
    238 
    +    """
    239 
    +    if not os.path.isabs(file_path):
    240 
    +        file_path = os.path.abspath(file_path)
    241 
    +    if not os.path.isabs(input_path):
    242 
    +        input_path = os.path.abspath(input_path)
    243 
    +
    244 
    +    file_blob = read_file(file_path)
    245 
    +    file_digest = create_digest(file_blob)
    246 
    +
    247 
    +    output_file = remote_execution_pb2.OutputFile(digest=file_digest)
    248 
    +    output_file.path = os.path.relpath(file_path, start=input_path)
    249 
    +    output_file.is_executable = os.access(file_path, os.X_OK)
    250 
    +
    251 
    +    update_request = remote_execution_pb2.BatchUpdateBlobsRequest.Request(digest=file_digest)
    252 
    +    update_request.data = file_blob
    253 
    +
    254 
    +    return output_file, update_request
    255 
    +
    256 
    +
    257 
    +def output_directory_maker(directory_path, working_path):
    258 
    +    """Creates a gRPC :obj:`OutputDirectory` from a local directory.
    259 
    +
    260 
    +    `directory_path` **must** point inside or be relative to `input_path`.
    261 
    +
    262 
    +    Args:
    263 
    +        directory_path (str): absolute or relative path to a local directory.
    264 
    +        working_path (str): absolute or relative path to the working directory.
    265 
    +
    266 
    +    Returns:
    267 
    +        :obj:`OutputDirectory`, :obj:`BatchUpdateBlobsRequest`: Tuple of a new
    268 
    +        gRPC :obj:`OutputDirectory` for the directory pointed by
    269 
    +        `directory_path` and the corresponding list of
    270 
    +        :obj:`BatchUpdateBlobsRequest` for CAS upload.
    271 
    +    """
    272 
    +    if not os.path.isabs(directory_path):
    273 
    +        directory_path = os.path.abspath(directory_path)
    274 
    +    if not os.path.isabs(working_path):
    275 
    +        working_path = os.path.abspath(working_path)
    276 
    +
    277 
    +    tree, update_requests = tree_maker(directory_path)
    278 
    +
    279 
    +    output_directory = remote_execution_pb2.OutputDirectory()
    280 
    +    output_directory.tree_digest = update_requests[-1].digest
    281 
    +    output_directory.path = os.path.relpath(directory_path, start=working_path)
    282 
    +
    283 
    +    output_directory_blob = output_directory.SerializeToString()
    284 
    +    output_directory_digest = create_digest(output_directory_blob)
    285 
    +
    286 
    +    update_request = remote_execution_pb2.BatchUpdateBlobsRequest.Request(digest=output_directory_digest)
    287 
    +    update_request.data = output_directory_blob
    288 
    +
    289 
    +    update_requests.append(update_request)
    290 
    +
    291 
    +    return output_directory, update_requests
    292 
    +
    293 
    +
    294 
    +def tree_maker(directory_path):
    295 
    +    """Creates a gRPC :obj:`Tree` from a local directory.
    296 
    +
    297 
    +    Args:
    298 
    +        directory_path (str): absolute or relative path to a local directory.
    299 
    +
    300 
    +    Returns:
    301 
    +        :obj:`Tree`, :obj:`BatchUpdateBlobsRequest`: Tuple of a new
    302 
    +        gRPC :obj:`Tree` for the directory pointed by `directory_path` and the
    303 
    +        corresponding list of :obj:`BatchUpdateBlobsRequest` for CAS upload.
    304 
    +
    305 
    +        The :obj:`BatchUpdateBlobsRequest` list may come in any order. However,
    306 
    +        its last element is guaranteed to be the :obj:`Tree`'s request.
    307 
    +    """
    308 
    +    if not os.path.isabs(directory_path):
    309 
    +        directory_path = os.path.abspath(directory_path)
    310 
    +
    311 
    +    update_requests = list()
    312 
    +    directory, update_request directory_maker(directory_path)
    313 
    +
    314 
    +
    315 
    +        update_requests.expend(update_request)
    316 
    +
    317 
    +
    318 
    +    tree = remote_execution_pb2.Tree()
    319 
    +    tree.children.expend()
    320 
    +    tree.root =
    321 
    +
    322 
    +    tree_blob = tree.SerializeToString()
    323 
    +    tree_digest = create_digest(file_blob)
    324 
    +
    325 
    +    update_request = remote_execution_pb2.BatchUpdateBlobsRequest.Request(digest=tree_digest)
    326 
    +    update_request.data = tree_blob
    327 
    +
    328 
    +    update_requests.append(update_request)
    329 
    +
    330 
    +    return tree, update_requests

    • —
    View it on GitLab.
    You're receiving this email because of your account on gitlab.com. If you'd like to receive fewer emails, you can adjust your notification settings.

    [Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]