在Bazel中，给定构建目标，脚本(在Bazel外部运行)如何获取生成文件的路径?
In Bazel, given a build target, how would a script (which is running outside of Bazel) get the path to the generated file?
方案:我正在使用Bazel进行构建，然后在完成后将结果复制到服务器.我只需要知道要复制什么文件即可.我可以对文件列表进行硬编码，但我不想这样做.
Scenario: I'm using Bazel to do the build, and then when it's done, I want to copy the result to a server. I just need to know what files to copy. I could hard-code the list of files, but I would prefer not to do that.
一个简单的例子:这个Bazel脚本:
A simple example: This Bazel script:
genrule( name = "main", srcs = ["main.in"], outs = ["main.out"], cmd = "cp $< $@", )如果然后创建一个名为main.in的文件，然后运行bazel build :main，则bazel报告:
If you then make a file named main.in and then run bazel build :main, bazel reports:
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，我可以使用哪种机器可读技术来获得该路径? (我可以解析bazel build的输出，但是我们不建议这样做.)
So there is is: bazel-genfiles/main.out. But what machine-readable technique can I use to get that path? (I could parse the output of bazel build, but we are discouraged from doing that.)
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.输出包括以下行:
The closest I have found is to use bazel query --output=xml :main, which dumps information about :main in XML format. The output includes this line:
<rule-output name="//:main.out"/>那太接近我想要的了.但是name是Bazel的标签格式.我不知道如何获得它.
That is so close to what I want. But the name is in Bazel's label format; I don't see how to get it as a path.
我可以在该name字段上进行某种字符串替换，以将其转换为bazel-genfiles/main.out；但即使那样也不可靠.如果我的genrule包含output_to_bindir = 1，则输出将改为bazel-bin/main.out.
I could do some kind of string replacement on that name field, to turn it into bazel-genfiles/main.out; but even that isn't reliable. If my genrule had included output_to_bindir = 1, then the output would have been bazel-bin/main.out instead.
此外，并非所有规则在XML输出中都有一个<rule-output>字段.例如，如果我的BUILD文件具有用于创建C库的以下代码:
Furthermore, not all rules have a <rule-output> field in the XML output. For example, if my BUILD file has this code to make a C library:
cc_library( name = "mylib", srcs = glob(["*.c"]) )bazel query --output=xml :mylib的输出不包含<rule-output>或其他有用的内容:
The output of bazel query --output=xml :mylib does not contain a <rule-output> or anything else helpful:
<?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>推荐答案
您可以使用 bazel aquery 进行查询 动作图.
You can get this information by using bazel aquery to query the action graph.
这是一个稍微丰富一点的示例，其中有一个来自两个输出文件 风格:
Here’s a slightly richer example, with two output files from a single genrule:
$ 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来查询具有机器可读输出的动作图(原型为 analysis.ActionGraphContainer ):
Use bazel aquery //:main --output=textproto to query the action graph with machine-readable output (the proto is 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" }数据并非完全集中在一处，但请注意:
The data isn’t exactly all in one place, but note that:
- ID为2和3的工件与我们两个所需的工件相对应 输出文件，并将这些工件的输出位置列出为 相对于工作空间根目录的磁盘上文件的路径；
- 目标ID为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相关联 标签.
- the artifacts with IDs 2 and 3 correspond to our two desired output files, and list the output locations of those artifacts as paths to files on disk relative to your workspace root;
- the artifacts entry with target ID 0 is associated with artifact IDs 2 and 3; and
- the targets entry with ID "0" is associated with the //:main label.
鉴于这种简单的结构，我们可以轻松地将脚本打包在一起 列出与提供的标签相对应的所有输出文件.我找不到 直接取决于Bazel对analysis.proto或其定义的定义 来自外部存储库的语言绑定，因此您可以修补 将以下脚本放入bazelbuild/bazel存储库本身:
Given this simple structure, we can easily whip together a script to list all output files corresponding to a provided label. I can’t find a way to depend directly on Bazel’s definition of analysis.proto or its language bindings from an external repository, so you can patch the following script into the bazelbuild/bazel repository itself:
工具/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 # # www.apache/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)工具/列表输出/已构建
# 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 # # www.apache/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，为了您的方便: gist.github/wchargin/5e6a43a203d6c95454aae2886c5b54e4
As a Git patch, for your convenience: gist.github/wchargin/5e6a43a203d6c95454aae2886c5b54e4
请注意，此代码尚未经过审核或验证 正确性我主要以它为例.如果有用的话 您，那么也许这个周末我可以为此编写一些测试并进行公关 针对Bazel本身.
Please note that this code hasn’t been reviewed or verified for correctness; I provide it primarily as an example. If it’s useful to you, then maybe this weekend I can write some tests for it and PR it against Bazel itself.
更多推荐
Bazel:如何获取生成文件的路径?
发布评论