site/docs/cquery.html

---
layout: documentation
title: Cquery (configurable query)
---
<h1> Configurable Query (cquery)</h1>


<p>
  <code>cquery</code> is a variant
  of <a href="query.html"><code>query</code></a> that correctly handles
  <a href="configurable-attributes.md"><code>select()</code></a> and build
  options' effects on the build graph.
</p>

<p>
  It achieves this by running over the results of Bazel's
  <a href="https://docs.bazel.build/versions/main/skylark/concepts.html#evaluation-model">analysis
    phase</a>, which integrates these effects. <code>query</code>, by constrast,
  runs over the results of Bazel's loading phase, before options are evaluated.
</p>

<p>
  For example:
</p>

<pre>
$ cat > tree/BUILD &lt;&lt;EOF
sh_library(
    name = "ash",
    deps = select({
        ":excelsior": [":manna-ash"],
        ":americana": [":white-ash"],
        "//conditions:default": [":common-ash"],
    }),
)
sh_library(name = "manna-ash")
sh_library(name = "white-ash")
sh_library(name = "common-ash")
config_setting(
    name = "excelsior",
    values = {"define": "species=excelsior"},
)
config_setting(
    name = "americana",
    values = {"define": "species=americana"},
)
EOF
</pre>

<pre>
# Traditional query: query doesn't know which select() branch will be chosen
# so it conservatively lists all of them.
$ bazel query "deps(//tree:ash)" --noimplicit_deps
//tree:ash
//tree:white-ash
//tree:manna-ash
//tree:common-ash

# cquery: cquery lets you set build options at the command line and chooses
# the exact dependencies that implies.
$ bazel cquery "deps(//tree:ash)" --define species=excelsior --noimplicit_deps
//tree:ash (9f87702)
//tree:manna-ash (9f87702)
</pre>

<p>
  Each result includes a <a href="#configurations">unique identifier</a> <code>(9f87702)</code> of
  the <a href="https://docs.bazel.build/glossary.html#configuration">configuration</a> the target is
  built with.
</p>

<p>
  Since <code>cquery</code> runs over the configured target graph. it doesn't
  have insight into artifacts like build actions nor access to
  <code><a href="be/general.html#test_suite">test_suite</a></code>
  rules as they are not configured targets. For the former,
  see <code><a href="aquery.html">aquery</a></code>.
</p>

<h2 id='basic-syntax'>Basic syntax</h2>

<p>A simple <code>cquery</code> call looks like:</p>

<p><code>bazel cquery "function(//target)"</code></p>

<p>The query expression <code>"function(//target)"</code> consists of the following:

<ul>
  <li>
    <b><code>function(...)</code></b> is the function to run on the target. <code>cquery</code>
    supports most
    of <code>query</code>'s <a href="query.html#functions">functions</a>, plus a
    few new ones.
  </li>
  <li><b><code>//target</code></b> is the expression fed to the function. In this example, the
    expression is a simple target. But the query language also allows nesting of functions.
    See the <a href="query-how-to.html">Query How-To</a> for examples.
   </li>
</ul>

<p>
  <code>cquery</code> requires a target to run through the
  <a href="https://docs.bazel.build/versions/main/skylark/concepts.html#evaluation-model">loading
  and analysis</a>
  phases. Unless otherwise specified, <code>cquery</code> parses the target(s)
  listed in the query
  expression. See <a href="#universe_scope-comma-separated-list"><code>--universe_scope</code></a>
  for querying dependencies of top-level build targets.
</p>

<h2 id='configurations'>Configurations</h2>

<p>
  The line:
</p>

<pre>
//tree:ash (9f87702)
</pre>

<p>
  means <code>//tree:ash</code> was built in a configuration with ID <code>9f87702</code>. For most
  targets, this is an opaque hash of the build option values defining the configuration.
</p>

<p>
  To see the configuration's complete contents, run:
</p>

<pre>
$ bazel config 9f87702
</pre>

<p>
  The host configuration uses the special ID <code>(HOST)</code>. Non-generated source files, like
  those commonly found in <code>srcs</code>, use the special ID <code>(null)</code> (because they
  don't need to be configured).
</p>

<p>
  <code>9f87702</code> is a prefix of the complete ID. This is because complete IDs are
  SHA-256 hashes, which are long and hard to follow. <code>cquery</code> understands any valid
   prefix of a complete ID, similar to
  <a href="https://git-scm.com/book/en/v2/Git-Tools-Revision-Selection#_revision_selection">Git short hashes</a>.
  To see complete IDs, run <code>$ bazel config</code>.
</p>

<h2 id='target-pattern-evaluation'>Target pattern evaluation</h2>

<p>
  <code>//foo</code> has a different meaning for <code>cquery</code> than
  for <code>query</code>. This is because <code>cquery</code>
  evaluates <i>configured</i> targets and the build graph may have multiple
  configured versions of <code>//foo</code>.
</p>

<p>
  For <code>cquery</code>, a target pattern in the query expression evaluates
  to every configured target with a label that matches that pattern. Output is
  deterministic, but <code>cquery</code> makes no ordering guarantee beyond the
  <a href="query.html#graph-order">core query ordering contract</a>.
</p>

<p>
  This produces subtler results for query expressions than with <code>query</code>.
  For example, the following can produce multiple results:

  <pre>
# Analyzes //foo in the target configuration, but also analyzes
# //genrule_with_foo_as_tool which depends on a host-configured
# //foo. So there are two configured target instances of //foo in
# the build graph.
$ bazel cquery //foo --universe_scope=//foo,//genrule_with_foo_as_tool
//foo (9f87702)
//foo (HOST)
  </pre>

</p>

<p>
  If you want to precisely declare which instance to query over, use
  the <a href="#config"><code>config</code></a> function.
</p>


<p>
  See <code>query</code>'s <a href="query.html#target-patterns">target pattern
  documentation</a> for more information on target patterns.
</p>

<h2 id='functions'>Functions</h2>

<p>
  Of the <a href="query.html#functions" title="list of query functions">set of functions</a>
  supported by <code>query</code>, <code>cquery</code> supports
  all
  but <a href="query.html#visible"><code>visible</code></a>,
  <a href="query.html#siblings"><code>siblings</code></a>,
  <a href="query.html#buildfiles"><code>buildfiles</code></a>,
  and <a href="query.html#tests"><code>tests</code></a>.
</p>

<p>
  <code>cquery</code> also introduces the following new functions:
</p>

<h3 id="config">config</h3>

<p><code>expr ::= config(expr, word)</code></p>

<p>
  The <code>config</code> operator attempts to find the configured target for
  the label denoted by the first argument and configuration specified by the
  second argument.
</p>

<p>
  Valid values for the second argument
  are <code>target</code>, <code>host</code>, <code>null</code>, or a
  <a href="#configurations">custom configuration hash</a>. Hashes can be retrieved from <code>$
  bazel config</code> or a prevous <code>cquery</code>'s output.
</p>

<p>
  Examples:
</p>

<pre>
$ bazel cquery "config(//bar, host)" --universe_scope=//foo
</pre>

<pre>
$ bazel cquery "deps(//foo)"
//bar (HOST)
//baz (3732cc8)

$ bazel cquery "config(//baz, 3732cc8)"
</pre>

<p>
  If not all results of the first argument can be found in the specified
  configuration, only those that can be found are returned. If no results
  can be found in the specified configuration, the query fails.
</p>

<h2 id='options'>Options</h2>

<h3>Build options</h3>

<p>
  <code>cquery</code> runs over a regular Bazel build and thus inherits the
  set of

  <a href="command-line-reference.html#build-options">options</a>
  available during a build.
</p>

<h3> Using cquery options</h3>

<h4 id="universe_scope-comma-separated-list"><code>--universe_scope</code> (comma-separated list)</h4>

<p>
  Often, the dependencies of configured targets go through
  <a href="https://docs.bazel.build/versions/main/skylark/rules.html#configurations">transitions</a>,
  which causes their configuration to differ from their dependent. This flag allows you to query
  a target as if it were built as a dependency or a transitive dependency of another target. For
  example:
</p>

<pre>
# x/BUILD
genrule(
     name = "my_gen",
     srcs = ["x.in"],
     outs = ["x.cc"],
     cmd = "$(locations :tool) $&lt; >$@",
     tools = [":tool"],
)
cc_library(
    name = "tool",
)
</pre>

<p>
  Genrules configure their tools in the
  <a href="https://docs.bazel.build/versions/main/skylark/rules.html#configurations">host configuration</a>
  so the following queries would produce the following outputs:
</p>

<table class="table table-condensed table-bordered table-params">
  <thead>
    <tr>
      <th>Query</th>
      <th>Target Built</th>
      <th>Output</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>bazel cquery "//x:tool"</td>
      <td>//x:tool</td>
      <td>//x:tool(targetconfig)</td>
    </tr>
    <tr>
      <td>bazel cquery "//x:tool" --universe_scope="//x:my_gen"</td>
      <td>//x:my_gen</td>
      <td>//x:tool(hostconfig)</td>
    </tr>
  </tbody>
</table>

<p>
  If this flag is set, its contents are built. <em>If it's not set, all targets
  mentioned in the query expression are built</em> instead. The transitive closure of the
  built targets are used as the universe of the query. Either way, the targets to
  be built must be buildable at the top level (that is, compatible with top-level
  options). <code>cquery</code> returns results in the transitive closure of these
  top-level targets.
</p>

<p>
  Even if it's possible to build all targets in a query expression at the top
  level, it may be beneficial to not do so. For example, explicitly setting
  <code>--universe_scope</code> could prevent building targets multiple times in
  configurations you don't care about. It could also help specify which configuration version of a
  target you're looking for (since it's not currently possible
  to fully specify this any other way). You should set this flag
  if your query expression is more complex than <code>deps(//foo)</code>.
</p>

<h4><code>--implicit_deps</code> (boolean, default=True)</h4>

<p>
  Setting this flag to false filters out all results that aren't explicitly set in
  the BUILD file and instead set elsewhere by Bazel. This includes filtering resolved
  toolchains.
</p>

<h4><code>--tool_deps</code> (boolean, default=True)</h4>

<p>
  Setting this flag to false filters out all configured targets for which the
  path from the queried target to them crosses a transition between the target
  configuration and the
  <a href="https://docs.bazel.build/versions/main/skylark/rules.html#configurations">non-target configurations</a>.
  If the queried target is in the target configuration, setting <code>--notool_deps</code> will
  only return targets that also are in the target configuration. If the queried
  target is in a non-target configuration, setting <code>--notool_deps</code> will only return
  targets also in non-target configurations. This setting generally does not affect filtering
  of resolved toolchains.
</p>

<h4><code>--include_aspects</code> (boolean, default=True)</h4>

<p>
  <a href="https://docs.bazel.build/versions/main/skylark/aspects.html">Aspects</a> can add
  additional dependencies to a build. By default, <code>cquery</code> doesn't follow aspects because
  they make the queryable graph bigger, which uses more memory. But following them produces more
  accurate results.
</p>

<p>
  If you're not worried about the memory impact of large queries, enable this flag by default in
  your bazelrc.
</p>

<p>
  If you query with aspects disabled, you can experience a problem where target X fails while
  building target Y but <code>cquery somepath(Y, X)</code> and <code>cquery deps(Y) | grep 'X'
  </code> return no results because the dependency occurs through an aspect.
</p>

<h2 id='output-formats'>Output formats</h2>

<p> By default, cquery outputs results in a dependency-ordered list of label and configuration pairs.
There are other options for exposing the results as well. </p>

<h3> Transitions </h3>

<pre>
--transitions=lite
--transitions=full
</pre>

<p>
  Configuration <a href="https://docs.bazel.build/versions/main/skylark/rules.html#configurations">transitions</a>
  are used to build targets underneath the top level targets in different configurations than the top
  level targets.
</p>

<p>
  For example, a target might impose a transition to the host configuration on all
  dependencies in its <code>tools</code> attribute. These are known as attribute
  transitions. Rules can also impose transitions on their own configurations,
  known as rule class transitions. This output format outputs information about
  these transitions such as what type they are and the effect they have on build
  options.
</p>

<p>This output format is triggered by the <code>--transitions</code> flag which by default is set to
  <code>NONE</code>. It can be set to <code>FULL</code> or <code>LITE</code> mode. <code>FULL</code>
  mode outputs information about rule class transitions and attribute transitions including a detailed
  diff of the options before and after the transition. <code>LITE</code> mode outputs the same information
  without the options diff.
</p>

<h3>Protocol message output</h3>

<pre>
--output=proto
</pre>

<p>
  This option causes the resulting targets to be printed in a binary protocol
  buffer form. The definition of the protocol buffer can be found at

  <a href="https://github.com/bazelbuild/bazel/blob/master/src/main/protobuf/analysis.proto">src/main/protobuf/analysis.proto</a>.
<h4>--[no]proto:include_configurations</h4>

<p>
  By default, cquery results return configuration information as part of each
  configured target. If you'd like to omit this information and get proto output
  that is formatted exactly like query's proto output, set this flag to false

 .


<p>
  See <a href="query.html#output-proto">query's proto output documentation</a>
  for more proto output-related options.
</p>

<p>
  <b>Note</b>: while selects are resolved both at the top level of returned
  targets and within attributes, all possible inputs for selects are still
  included as <code>rule_input</code> fields.
</p>

<h3>Graph output</h3>

<pre>
--output=graph
</pre>

<p>
  This option generates output as a Graphviz-compatible .dot file. See <code>query</code>'s
  <a href="query.html#output-graph">graph output documentation</a> for details. <code>cquery</code>
  also supports

  <a href="query.html#graph-node_limit-n"><code>--graph:node_limit</code></a> and
  <a href="query.html#no-graph-factored"><code>--graph:factored</code></a>.

</p>

<h3>Defining the output format using Starlark</h3>

<pre>
--output=starlark
</pre>

<p>
  This output format calls a
  <a href="https://docs.bazel.build/versions/main/skylark/language.html">Starlark</a>
  function for each configured target in the query result, and prints the value returned by the
  call. The <code>--starlark:file</code> flag specifies the location of a Starlark file that
  defines a function named <code>format</code> with a single parameter, <code>target</code>. This
  function is called for each
  <a href="https://docs.bazel.build/versions/main/skylark/lib/Target.html">Target</a> in the
  query result. Alternatively, for convenience, you may specify just the body of a function
  declared as <code>def format(target): return expr</code> by using the
  <code>--starlark:expr</code> flag.

<h4>'cquery' Starlark dialect</h4>
<p>
  The cquery Starlark environment differs from a BUILD or .bzl file. It includes all core Starlark
  <a href="https://github.com/bazelbuild/starlark/blob/master/spec.md#built-in-constants-and-functions">
  built-in constants and functions</a>, plus a few cquery-specific ones described below,
  but not (for example) <code>glob</code>, <code>native</code>, or <code>rule</code>, and it does
  not support load statements.
</p>
<h5 id='build_options'>build_options(target)</h5>
<p>
  <code>build_options(target)</code> returns a map whose keys are build option identifiers (see
  <a href="https://docs.bazel.build/versions/main/skylark/config.html">Configurations</a>)
  and whose values are their Starlark values. Build options whose values are not legal Starlark
  values are omitted from this map.
</p><p>
  If the target is an input file, <code>build_options(target)</code> returns None, as input file
  targets have a null configuration.
</p>
<h5 id='providers'>providers(target)</h5>
<p>
  <code>providers(target)</code> returns a map whose keys are names of
  <a href="https://docs.bazel.build/versions/main/skylark/rules.html#providers">providers</a>
  (for example, <code>"DefaultInfo"</code>) and whose values are their Starlark values. Providers
  whose values are not legal Starlark values are omitted from this map.
</p>

<h4>Examples</h4>

<p>
Print a space-separated list of the base names of all files produced by <code>//foo</code>:
</p>
<pre>
  bazel cquery //foo --output=starlark \
    --starlark:expr="' '.join([f.basename for f in target.files.to_list()])"
</pre>
<p>
Print a space-separated list of the paths of all files produced by <b>rule</b> targets in
<code>//bar</code> and its subpackages:
</p>
<pre>
  bazel cquery 'kind(rule, //bar/...)' --output=starlark \
    --starlark:expr="' '.join([f.path for f in target.files.to_list()])"
</pre>
<p>
Print a list of the mnemonics of all actions registered by <code>//foo</code>.
</p>
<pre>
  bazel cquery //foo --output=starlark \
    --starlark:expr="[a.mnemonic for a in target.actions]"
</pre>
<p>
Print a list of compilation outputs registered by a <code>cc_library</code> <code>//baz</code>.
</p>
<pre>
  bazel cquery //baz --output=starlark \
    --starlark:expr="[f.path for f in target.output_groups.compilation_outputs.to_list()]"
</pre>
<p>
Print the value of the command line option <code>--javacopt</code> when building <code>//foo</code>.
</p>
<pre>
  bazel cquery //foo --output=starlark \
    --starlark:expr="build_options(target)['//command_line_option:javacopt']"
</pre>
<p>
  Print the label of each target with exactly one output. This example uses Starlark functions
  defined in a file.
</p>
<pre>
  $ cat example.cquery

  def has_one_output(target):
    return len(target.files.to_list()) == 1

  def format(target):
    if has_one_output(target):
      return target.label
    else:
      return ""


  $ bazel cquery //baz --output=starlark --starlark:file=example.cquery
</pre>
<p>
  Print the label of each target which is strictly Python 3. This example uses Starlark functions
  defined in a file.
</p>
<pre>
  $ cat example.cquery

  def format(target):
    p = providers(target)
    py_info = p.get("PyInfo")
    if py_info and py_info.has_py3_only_sources:
      return target.label
    else:
      return ""


  $ bazel cquery //baz --output=starlark --starlark:file=example.cquery
</pre>

<p>
  Extract a value from a user defined Provider.
</p>
<pre>
  $ cat some_package/my_rule.bzl

  MyRuleInfo = provider(fields={"color": "the name of a color"})

  def _my_rule_impl(ctx):
      ...
      return [MyRuleInfo(color="red")]

  my_rule = rule(
      implementation = _my_rule_impl,
      attrs = {...},
  )

  $ cat example.cquery

  def format(target):
    p = providers(target)
    my_rule_info = p.get("//some_package:my_rule.bzl%MyRuleInfo'")
    if my_rule_info:
      return my_rule_info.color
    return ""


  $ bazel cquery //baz --output=starlark --starlark:file=example.cquery
</pre>


<h2 id='compare'>cquery vs. query</h2>

<p>
  <code>cquery</code> and <code>query</code> complement each other and excel in
  different niches. Consider the following to decide which is right for you:

  <ul>
    <li>
      <code>cquery</code> follows specific <code>select()</code> branches to
      model the exact graph you build. <code>query</code> doesn't know which
      branch the build chooses, so overapproximates by including all branches.
    </li>
   <li>
      <code>cquery</code>'s precision requires building more of the graph than
      <code>query</code> does. Specifically, <code>cquery</code>
      evaluates <i>configured targets</i> while <code>query</code> only
      evaluates <i>targets</i>. This takes more time and uses more memory.
   </li>
   <li>
      <code>cquery</code>'s intepretation of
      the <a href="query.html#concepts">query language</a> introduces ambiguity
      that <code>query</code> avoids. For example,
      if <code>"//foo"</code> exists in two configurations, which one
      should <code>cquery "deps(//foo)"</code> use?
      The <code><a href="#config">config</a></code></code> function can help with
      this.
   </li>
   <li>
      As a newer tool, <code>cquery</code> lacks support for certain use
      cases. See <a href="#known-issues">Known issues</a> for details.
   </li>
  </ul>
</p>

<h2 id='known-issues'>Known issues</h2>

<ul>
  <li>
    <strong>All targets that <code>cquery</code> "builds" must have the same
    configuration.</strong>

    <p>
      Before evaluating queries, <code>cquery</code> triggers a build up to just
      before the point where build actions would execute. The targets it
      "builds" are by default selected from all labels that appear in the query
      expression (this can be overridden
      with <a href="#universe_scope-comma-separated-list"><code>--universe_scope</code></a>). These
      must have the same configuration.
    </p>

    <p>
      While these generally share the top-level "target" configuration,
      rules can change their own configuration with

      <a href="skylark/config.html#incoming-edge-transitions">incoming edge transitions</a>.
      This is where </code>cquery</code> falls short.
    </p>

    <p>
      Workaround: If possible, set <code>--universe_scope</code> to a stricter
      scope. For example:
    </p>

    <pre>
# This command attempts to build the transitive closures of both //foo and
# //bar. //bar uses an incoming edge transition to change its --cpu flag.
$ bazel cquery 'somepath(//foo, //bar)'
ERROR: Error doing post analysis query: Top-level targets //foo and //bar
have different configurations (top-level targets with different
configurations is not supported)

# This command only builds the transitive closure of //foo, under which
# //bar should exist in the correct configuration.
$ bazel cquery 'somepath(//foo, //bar)' --universe_scope=//foo
   </pre>
  </li>

  <li>
    <strong>No support
    for <a href="query.html#output-xml"><code>--output=xml</code></a>.</strong>
  </li>

  <li>
    <strong>Non-deterministic output.</strong>

    <p>
      <code>cquery</code> does not automatically wipe the build graph from
      previous commands and is therefore prone to picking up results from past
      queries. For example, <code>genquery</code> exerts a host transition on
      its <code>tools</code> attribute - that is, it configures its tools in the
      <a href="https://docs.bazel.build/versions/main/skylark/rules.html#configurations">host
      configuration</a>.
    </p>

    <p>
      You can see the lingering effects of that transition below.
    </p>

<pre>
$ cat > foo/BUILD &lt;&lt;&lt;EOF
genrule(
    name = "my_gen",
    srcs = ["x.in"],
    outs = ["x.cc"],
    cmd = "$(locations :tool) $&lt; >$@",
    tools = [":tool"],
)
cc_library(
    name = "tool",
)
EOF

$ bazel cquery "//foo:tool"
tool(target_config)

$ bazel cquery "deps(//foo:my_gen)"
my_gen (target_config)
tool (host_config)
...

$ bazel cquery "//foo:tool"
tool(host_config)
</pre>

<p>
  Workaround: change any startup option to force re-analysis of configured targets. For example,
  add <code>--test_arg=&lt;whatever&gt;</code> to your build command.
</p>

  </li>
</ul>

<h2 id='troubleshooting'>Troubleshooting</h2>

<h3>Recursive target patterns (<code>/...</code>)</h3>

<p>
  If you encounter:
</p>

<pre>
$ bazel cquery --universe_scope=//foo:app "somepath(//foo:app, //foo/...)"
ERROR: Error doing post analysis query: Evaluation failed: Unable to load package '[foo]'
because package is not in scope. Check that all target patterns in query expression are within the
--universe_scope of this query.
</pre>

<p>
  this incorrectly suggests package <code>//foo</code> isn't in scope even though
  <code>--universe_scope=//foo:app</code> includes it. This is due to design limitations in
  <code>cquery</code>. As a workaround, explicitly include <code>//foo/...</code> in the universe
  scope:
</p>

<pre>
$ bazel cquery --universe_scope=//foo:app,//foo/... "somepath(//foo:app, //foo/...)"
</pre>

<p>
  If that doesn't work (for example, because some target in <code>//foo/...</code> can't build
  with the chosen build flags), manually unwrap the pattern into its constituent packages with a
  pre-processing query:
</p>

<pre>
# Replace "//foo/..." with a subshell query call (not cquery!) outputting each package, piped into
# a sed call converting "&lt;pkg&gt;" to "//&lt;pkg&gt;:*", piped into a "+"-delimited line merge.
# Output looks like "//foo:*+//foo/bar:*+//foo/baz".
#
$  bazel cquery --universe_scope=//foo:app "somepath(//foo:app, $(bazel query //foo/...
--output=package | sed -e 's/^/\/\//' -e 's/$/:*/' | paste -sd "+" -))"
</pre>