Bazel：你如何获得生成文件的路径？

Question

在 Bazel 中，给定构建目标，脚本（在 Bazel 之外运行ning）如何获取生成文件的路径？

场景：我正在使用 Bazel 进行构建，然后在完成后，我想将结果复制到服务器。我只需要知道要复制哪些文件。我可以对文件列表进行硬编码，但我不想那样做。

一个简单的例子：这个 Bazel 脚本：

genrule(
    name = "main",
    srcs = ["main.in"],
    outs = ["main.out"],
    cmd = "cp $< $@",
)

如果您随后创建一个名为 main.in 的文件，然后创建运行 bazel build :main，bazel 报告：

INFO: Found 1 target...
Target //:main up-to-date:
  bazel-genfiles/main.out
INFO: Elapsed time: 6.427s, Critical Path: 0.40s

所以有：bazel-genfiles/main.out。但是我可以使用什么机器可读技术来获得该路径？（我可以解析 bazel build 的输出，但我们不鼓励这样做。）

我发现最接近的是使用 bazel query --output=xml :main，它以 XML 格式转储有关 :main 的信息。输出包括这一行：

<rule-output name="//:main.out"/>

这太接近我想要的了。但是 name 是 Bazel 的标签格式；我不知道如何将其作为路径获取。

我可以在那个 name 字段上做一些字符串替换，把它变成 bazel-genfiles/main.out；但即使那样也不可靠。如果我的 genrule 包含 output_to_bindir = 1，那么输出将是 bazel-bin/main.out。

此外，并非所有规则在 XML 输出中都有 <rule-output> 字段。例如，如果我的 BUILD 文件有此代码来制作 C 库：

cc_library(
    name = "mylib",
    srcs = glob(["*.c"])
)

bazel query --output=xml :mylib 的输出不包含 <rule-output> 或任何其他有用的内容：

<?xml version="1.1" encoding="UTF-8" standalone="no"?>
<query version="2">
  <rule class="cc_library" location="/Users/mikemorearty/src/bazel/test1/BUILD:8:1" name="//:mylib">
    <string name="name" value="mylib"/>
    <list name="srcs">
      <label value="//:foo.c"/>
    </list>
    <rule-input name="//:foo.c"/>
    <rule-input name="//tools/defaults:crosstool"/>
    <rule-input name="@bazel_tools//tools/cpp:stl"/>
  </rule>
</query>

Answer 1

在 bazel 的两个运行之间，输出路径应该相同。也就是说，如果您构建 //path/to:target 然后 bazel clean 并再次构建，它应该会生成相同的文件。由于此输出文件是常量，您可以运行

ls "$(bazel info bazel-genfiles)/main.out"

我相信这会给你一个参考，一旦构建发生（它不会为你构建它）。

如果您希望从 target 转到将依赖于 rules_* 的文件名，您运行ning。例如在他们的项目 rules_go, the output path depends on the arguments to the go_library target. The rules_go team has recently documented this behavior 中。

一般来说，二进制输出路径在不同版本之间应该是稳定的，您可以相信它们不会有太大差异。但是，根据我的经验，这个问题通常表明您应该考虑将流程中以前的外部部分作为 genrule 或自定义规则移动到 Bazel 中。例如，我以前使用这个技巧来 assemble NPM 包，但现在我做了整个事情 in Bazel 并且有一个 target 生成 .tar我有兴趣上传到 NPM。也许你可以跟进一些关于你感兴趣的事情的细节，我们也许能够通过一个不依赖于外部系统理解 Bazel 构建路径的解决方案。

Answer 2

您可以使用bazel aquery查询得到这些信息动作图。

这是一个稍微丰富一点的例子，一个文件有两个输出文件种族规则：

$ ls
BUILD  main.in  WORKSPACE
$ cat WORKSPACE
$ cat BUILD
genrule(
    name = "main",
    srcs = ["main.in"],
    outs = ["main.o1", "main.o2"],
    cmd = "cp $< $(location main.o1); cp $< $(location main.o2)",
)
$ cat main.in
hello

使用bazel aquery //:main --output=textproto查询具有机器可读输出的动作图（原型为analysis.ActionGraphContainer）：

$ bazel aquery //:main --output=textproto >aquery_result 2>/dev/null
$ cat aquery_result
artifacts {
  id: "0"
  exec_path: "main.in"
}
artifacts {
  id: "1"
  exec_path: "external/bazel_tools/tools/genrule/genrule-setup.sh"
}
artifacts {
  id: "2"
  exec_path: "bazel-out/k8-fastbuild/genfiles/main.o1"
}
artifacts {
  id: "3"
  exec_path: "bazel-out/k8-fastbuild/genfiles/main.o2"
}
actions {
  target_id: "0"
  action_key: "dd7fd759bbecce118a399c6ce7b0c4aa"
  mnemonic: "Genrule"
  configuration_id: "0"
  arguments: "/bin/bash"
  arguments: "-c"
  arguments: "source external/bazel_tools/tools/genrule/genrule-setup.sh; cp main.in bazel-out/k8-fastbuild/genfiles/main.o1; cp main.in bazel-out/k8-fastbuild/genfiles/main.o2"
  input_dep_set_ids: "0"
  output_ids: "2"
  output_ids: "3"
}
targets {
  id: "0"
  label: "//:main"
  rule_class_id: "0"
}
dep_set_of_files {
  id: "0"
  direct_artifact_ids: "0"
  direct_artifact_ids: "1"
}
configuration {
  id: "0"
  mnemonic: "k8-fastbuild"
  platform_name: "k8"
}
rule_classes {
  id: "0"
  name: "genrule"
}

数据并不完全集中在一处，但请注意：

ID 为 2 和 3 的工件对应于我们想要的两个输出文件，并将这些工件的输出位置列为磁盘上相对于工作区根目录的文件路径；
目标 ID 0 的 artifacts 条目与工件相关联 ID 2 和 3；和
ID 为 "0" 的 targets 条目与 //:main 关联标签。

鉴于这个简单的结构，我们可以轻松地将脚本组合在一起以列出与提供的标签对应的所有输出文件。我找不到直接依赖 Bazel 对 analysis.proto 或其定义的方法来自外部存储库的语言绑定，因此您可以修补以下脚本进入 bazelbuild/bazel 存储库本身：

tools/list_outputs/list_outputs.py

# Copyright 2019 The Bazel Authors. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#    http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
r"""Parse an `aquery` result to list outputs created for a target.

Use this binary in conjunction with `bazel aquery` to determine the
paths on disk to output files of a target.

Example usage: first, query the action graph for the target that you
want to analyze:

    bazel aquery //path/to:target --output=textproto >/tmp/aquery_result

Then, from the Bazel repository:

    bazel run //tools/list_outputs -- \
        --aquery_result /tmp/aquery_result \
        --label //path/to:target \
        ;

This will print a list of zero or more output files emitted by the given
target, like:

    bazel-out/k8-fastbuild/foo.genfile
    bazel-out/k8-fastbuild/bar.genfile

If the provided label does not appear in the output graph, an error will
be raised.
"""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import sys

from absl import app
from absl import flags
from google.protobuf import text_format
from src.main.protobuf import analysis_pb2


flags.DEFINE_string(
    "aquery_result",
    None,
    "Path to file containing result of `bazel aquery ... --output=textproto`.",
)
flags.DEFINE_string(
    "label",
    None,
    "Label whose outputs to print.",
)


def die(message):
  sys.stderr.write("fatal: %s\n" % (message,))
  sys.exit(1)


def main(unused_argv):
  if flags.FLAGS.aquery_result is None:
    raise app.UsageError("Missing `--aquery_result` argument.")
  if flags.FLAGS.label is None:
    raise app.UsageError("Missing `--label` argument.")

  if flags.FLAGS.aquery_result == "-":
    aquery_result = sys.stdin.read()
  else:
    with open(flags.FLAGS.aquery_result) as infile:
      aquery_result = infile.read()
  label = flags.FLAGS.label

  action_graph_container = analysis_pb2.ActionGraphContainer()
  text_format.Merge(aquery_result, action_graph_container)

  matching_targets = [
      t for t in action_graph_container.targets
      if t.label == label
  ]
  if len(matching_targets) != 1:
    die(
        "expected exactly one target with label %r; found: %s"
        % (label, sorted(t.label for t in matching_targets))
    )
  target = matching_targets[0]

  all_artifact_ids = frozenset(
      artifact_id
      for action in action_graph_container.actions
      if action.target_id == target.id
      for artifact_id in action.output_ids
  )
  for artifact in action_graph_container.artifacts:
    if artifact.id in all_artifact_ids:
      print(artifact.exec_path)


if __name__ == "__main__":
  app.run(main)

tools/list_outputs/BUILD

# Copyright 2019 The Bazel Authors. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#    http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

package(default_visibility = ["//visibility:public"])

licenses(["notice"])  # Apache 2.0

filegroup(
    name = "srcs",
    srcs = glob(["**"]),
)

py_binary(
    name = "list_outputs",
    srcs = ["list_outputs.py"],
    srcs_version = "PY2AND3",
    deps = [
        "//third_party/py/abseil",
        "//src/main/protobuf:analysis_py_proto",
    ],
)

作为 Git 补丁，为方便起见： https://gist.github.com/wchargin/5e6a43a203d6c95454aae2886c5b54e4

请注意，此代码尚未经过审查或验证正确性；我提供它主要是作为示例。如果对你有用你，然后也许这个周末我可以为它写一些测试并 PR 针对 Bazel 本身。

Bazel：你如何获得生成文件的路径？

Bazel: How do you get the path to a generated file?

bazel