Bazel Rules Testing¶

rules_testing is a collection of utilities, libraries, and frameworks to make testing Starlark and Bazel rules easy and pleasant.

version |version|

Installation¶

To use rules_testing, you need to modify WORKSPACE or MODULE.bazel to depend on rules_testing. We recommend using bzlmod because it’s simpler.

For bzlmod, add this to your MODULE.bazel:

bazel_dep(name = "rules_testing", version = "<VERSION>", dev_dependency=True)

See the GitHub releases page for available versions.

For WORKSPACE, see the GitHub releases page for the necessary config to copy and paste.

Analysis tests¶

Analysis testing means testing something during the analysis phase of Bazel execution – this is when rule logic is run.

See Analysis tests for how to write analysis tests.

Fluent asserts¶

Included in rules_testing is a fluent, truth-style asserts library.

See Truth docs for how to use it.

Analysis Tests¶

Analysis tests are the typical way to test rule behavior. They allow observing behavior about a rule that isn’t visible to a regular test as well as modifying Bazel configuration state to test rule behavior for e.g. different platforms.

If you’ve ever wanted to verify…

A certain combination of flags
Building for another OS
That certain providers are returned
That aspects behaved a certain way

Or other observable information, then an analysis test does that.

Quick start¶

For a quick copy/paste start, create a .bzl file with your test code, and a BUILD.bazel file to load your tests and declare them. Here’s a skeleton:

# BUILD
load(":my_tests.bzl", "my_test_suite")

my_test_suite(name="my_test_suite")

# my_tests.bzl

load("@rules_testing//lib:analysis_test.bzl", "test_suite", "analysis_test")
load("@rules_testing//lib:util.bzl", "util")

def _test_hello(name):
    util.helper_target(
        native.filegroup,
        name = name + "_subject",
        srcs = ["hello_world.txt"],
    )
    analysis_test(
        name = name,
        impl = _test_hello_impl,
        target = name + "_subject"
    )

def _test_hello_impl(env, target):
    env.expect.that_target(target).default_outputs().contains(
        "hello_world.txt"
    )

def my_test_suite(name):
    test_suite(
        name = name,
        tests = [
            _test_hello,
        ]
    )

Arranging the test¶

The arrange part of a test defines a target using the rule under test and sets up its dependencies. This is done by writing a macro, which runs during the loading phase, that instantiates the target under test and dependencies. All the targets taking part in the arrangement should be tagged with manual so that they are ignored by common build patterns (e.g. //... or foo:all).

Example:

load("@rules_proto/defs:proto_library.bzl", "proto_library")


def _test_basic(name):
  """Verifies basic behavior of a proto_library rule."""
  # (1) Arrange
  proto_library(name=name + '_foo', srcs=["foo.proto"], deps=[name + "_bar"], tags=["manual"])
  proto_library(name=name + '_bar', srcs=["bar.proto"], tags=["manual"])

  # (2) Act
  ...

TIP: Source source files aren’t required to exist. This is because the analysis phase only records the path to source files; they aren’t read until after the analysis phase. The macro function should be named after the behaviour being tested (e.g. _test_frob_compiler_passed_qux_flag). The setup targets should follow the macro naming conventions, that is all targets should include the name argument as a prefix – this helps tests avoid creating conflicting names.

Limitations¶

Bazel limits the number of transitive dependencies that can be used in the setup. The limit is controlled by --analysis_testing_deps_limit flag.

Mocking toolchains (adding a toolchain used only in the test) is not possible at the moment.

Running the analysis phase¶

The act part runs the analysis phase for a specific target and calls a user supplied function. All of the work is done by Bazel and the framework. Use analysis_test macro to pass in the target to analyse and a function that will be called with the analysis results:

load("@rules_testing//lib:analysis_test.bzl", "analysis_test")


def _test_basic(name):
  ...

  # (2) Act
  analysis_test(name, target=name + "_foo", impl=_test_basic)

Assertions¶

The assert function (in example _test_basic) gets env and target as parameters, where…

env is information about the overall build and test
target is the target under test (as specified in the target attribute during the arrange step).

The env.expect attribute provides a truth.Expect object, which allows writing fluent asserts:

def _test_basic(env, target):
  env.expect.assert_that(target).runfiles().contains_at_least("foo.txt")
  env.expect.assert_that(target).action_generating("foo.txt").contains_flag_values("--a")

Note that you aren’t required to use env.expect. If you want to perform asserts another way, then env.fail() can be called to register any failures.

Collecting the tests together¶

Use the test_suite function to collect all tests together:

load("@rules_testing//lib:analysis_test.bzl", "test_suite")


def proto_library_test_suite(name):
  test_suite(
      name=name,
      tests=[
          _test_basic,
          _test_advanced,
      ]
  )

In your BUILD file instantiate the suite:

load("//path/to/your/package:proto_library_tests.bzl", "proto_library_test_suite")
proto_library_test_suite(name = "proto_library_test_suite")

The function instantiates all test macros and wraps them into a single target. This removes the need to load and call each test separately in the BUILD file.

Advanced test collection, reuse, and parameterizing¶

If you have many tests and rules and need to re-use them between each other, then there are a couple tricks to make it easy:

Tests aren’t required to all be in the same file. So long as you can load the arrange function and pass it to test_suite, then you can split tests into multiple files for reuse.
Similarly, arrange functions themselves aren’t required to take only a name argument – only the functions passed to test_suite.test require this.

By using lists and lambdas, we can define collections of tests and have multiple rules reuse them:

# base_tests.bzl

_base_tests = []

def _test_common(name, rule_under_test):
  rule_under_test(...)
  analysis_test(...)

def _test_common_impl(env, target):
  env.expect.that_target(target).contains(...)

_base_tests.append(_test_common)

def create_base_tests(rule_under_test):
    return [
        lambda name: test(name=name, rule_under_test=rule_under_test)
        for test in _base_tests
    ]

# my_binary_tests.bzl
load("//my/my_binary.bzl", "my_binary")
load(":base_tests.bzl", "create_base_tests")
load("@rules_testing//lib:analysis_test.bzl", "test_suite")

def my_binary_suite(name):
    test_suite(
        name = name,
        tests = create_base_tests(my_binary)
    )

# my_test_tests.bzl
load("//my/my_test.bzl", "my_test")
load(":base_tests.bzl", "base_tests")
load("@rules_testing//lib:analysis_test.bzl", "test_suite")

def my_test_suite(name):
    test_suite(
        name = name,
        tests = create_base_tests(my_test)
    )

Tips and best practices¶

Use private names for your tests, def _test_foo. This allows buildifier to detect when you’ve forgotten to put a test in the tests attribute. The framework will strip leading underscores from the test name
Tag the arranged inputs of your tests with tags=["manual"]; the util.helper_target function helps with this. This prevents common build patterns (e.g. bazel test //... or bazel test :all) from trying to build them.
Put each rule’s tests into their own directory with their own BUILD file. This allows better isolation between the rules’ test suites in several ways:
- When reusing tests, target names are less likely to collide.
- During the edit-run cycle, modifications to verify one rule that would break another rule can be ignored until you’re ready to test the other rule.

Guides¶

Analysis tests¶

Analysis testing means testing something during the analysis phase of Bazel execution – this is when rule logic is run.

See Analysis testing for how to write analysis tests.

Fluent asserts¶

Included in rules_testing is a fluent, truth-style asserts library.

See Truth docs for how to use it.

Truth Guide¶

Also see: Truth API reference

What is Truth?¶

Truth is a style of doing asserts that makes it easy to perform complex assertions that are easy to understand and give actionable error messages.

The basic way it works is wrapping a value in a type-specific object that provides type-specific assertion methods. This style provides several benefits:

A fluent API that more directly expresses the assertion
More egonomic assert functions
Error messages with more informative context
Promotes code reuses at the type-level.

Example Usage¶

Note that all examples assume usage of the rules_testing analysis_test framework, but truth itself does not require it.

def test_foo(env, target):
    subject = env.expect.that_target(target)
    subject.runfiles().contains_at_least(["foo.txt"])
    subject.executable().equals("bar.exe")

    subject = env.expect.that_action(...)
    subject.contains_at_least_args(...)

Subjects¶

Subjects are wrappers around a value that provide ways to assert on the value, access sub-values of it, or otherwise augment interacting with the wrapped value. For example, TargetSubject wraps Bazel Target objects and RunfilesSubject wraps Bazel runfiles objects. Normally accessing a target’s runfiles and verifying the runfiles contents would require the verbose target[DefaultInfo].default_runfiles, plus additional code to convert a runfiles object’s files, symlinks, root_symlinks, and empty_filenames into a single list to verify. With subject classes, however, it can be concisely expressed as expect.that_target(target).runfiles().contains(path).

The Truth library provides subjects for types that are built into Bazel, but custom subjects can be implemented to handle custom providers or other objects.

Predicates¶

Because Starlark’s data model doesn’t allow customizing equality checking, some subjects allow matching values by using a predicate function. This makes it easier to, for example, ignore a platform-specific file extension.

This is implemented using the structural Matcher “interface”. This is a struct that contains the predicate function and a description of what the function does, which allows for more intelligible error messages.

A variety of matchers are in truth.bzl#matching, but custom matches can be implemented using matching.custom_matcher

Writing a new Subject¶

Writing a new Subject involves two basic pieces:

Creating a constructor function, e.g. _foo_subject_new, that takes the actual value and an ExpectMeta object (see _expect_meta_new()).
Adding a method to expect or another Subject class to pass along state and instantiate the new subject; both may be modified if the actual object can be independenly created or obtained through another subject.

For top-level subjects, a method named that_foo() should be added to the expect class.

For child-subjects, an appropriately named method should be added to the parent subject, and the parent subject should call ExpectMeta.derive() to create a new set of meta data for the child subject.

The assert methods a subject provides are up to the subject, but try to follow the naming scheme of other subjects. The purpose of a custom subject is to make it easier to write tests that are correct and informative. It’s common to have a combination of ergonomic asserts for common cases, and delegating to child-subjects for other cases.

Adding asserts to a subject¶

Fundamentally, an assert method calls ExpectMeta.add_failure() to record when there is a failure. That method will wire together any surrounding context with the provided error message information. Otherwise an assertion is free to implement checks how it pleases.

The naming of functions should mostly read naturally, but doesn’t need to be perfect grammatically. Be aware of ambiguous words like “contains” or “matches”. For example, contains_flag("--foo") – does this check that “–flag” was specified at all (ignoring value), or that it was specified and has no value?

Assertion functions can make use of a variety of helper methods in processing values, comparing them, and generating error messages. Helpers of particular note are:

_check_*: These functions implement comparison, error formatting, and error reporting.
_compare_*: These functions implements comparison for different cases and take care of various edge cases.
_format_failure_*: These functions create human-friendly messages describing both the observed values and the problem with them.
_format_problem_*: These functions format only the problem identified.
_format_actual_*: These functions format only the observed values.

API Reference¶

ActionSubject¶

ActionSubject.new¶

ActionSubject.new(action, meta)

Creates an “ActionSubject” struct.

Method: ActionSubject.new

Example usage:

expect(env).that_action(action).not_contains_arg("foo")

PARAMETERS ¶

action¶:: (Action) value to check against.
meta¶:: (ExpectMeta) of call chain information.

RETURNS ¶: ActionSubject object.

ActionSubject.parse_flags¶

ActionSubject.parse_flags(argv)

PARAMETERS ¶

argv¶:: undocumented

ActionSubject.argv¶

ActionSubject.argv()

Returns a CollectionSubject for the action’s argv.

Method: ActionSubject.argv

RETURNS ¶: CollectionSubject object.

ActionSubject.contains_at_least_args¶

ActionSubject.contains_at_least_args(args)

Assert that an action contains at least the provided args.

Method: ActionSubject.contains_at_least_args

Example usage: expect(env).that_action(action).contains_at_least_args([“foo”, “bar”]).

PARAMETERS ¶

args¶:: (list of str) all the args must be in the argv exactly as provided. Multiplicity is respected.

RETURNS ¶: Ordered (see _ordered_incorrectly_new).

ActionSubject.not_contains_arg¶

ActionSubject.not_contains_arg(arg)

Assert that an action does not contain an arg.

Example usage: expect(env).that_action(action).not_contains_arg(“should-not-exist”)

PARAMETERS ¶

arg¶:: (str) the arg that cannot be present in the argv.

ActionSubject.substitutions¶

ActionSubject.substitutions()

Creates a DictSubject to assert on the substitutions dict.

Method: ActionSubject.substitutions.

RETURNS ¶: DictSubject struct.

ActionSubject.has_flags_specified¶

ActionSubject.has_flags_specified(flags)

Assert that an action has the given flags present (but ignore any value).

Method: ActionSubject.has_flags_specified

This parses the argv, assuming the typical formats (--flag=value, --flag value, and --flag). Any of the formats will be matched.

Example usage, given argv = ["--a", "--b=1", "--c", "2"]: expect(env).that_action(action).has_flags_specified([ “–a”, “–b”, “–c”])

PARAMETERS ¶

flags¶:: (list of str) The flags to check for. Include the leading “–“. Multiplicity is respected. A flag is considered present if any of these forms are detected: --flag=value, --flag value, or a lone --flag.

RETURNS ¶: Ordered (see _ordered_incorrectly_new).

ActionSubject.mnemonic¶

ActionSubject.mnemonic()

Returns a StrSubject for the action’s mnemonic.

Method: ActionSubject.mnemonic

RETURNS ¶: StrSubject object.

ActionSubject.inputs¶

ActionSubject.inputs()

Returns a DepsetFileSubject for the action’s inputs.

Method: ActionSubject.inputs

RETURNS ¶: DepsetFileSubject of the action’s inputs.

ActionSubject.contains_flag_values¶

ActionSubject.contains_flag_values(flag_values)

Assert that an action’s argv has the given (”–flag”, “value”) entries.

Method: ActionSubject.contains_flag_values

This parses the argv, assuming the typical formats (--flag=value, --flag value, and --flag). Note, however, that for the --flag value and --flag forms, the parsing can’t know how many args, if any, a flag actually consumes, so it simply takes the first following arg, if any, as the matching value.

NOTE: This function can give misleading results checking flags that don’t consume any args (e.g. boolean flags). Use has_flags_specified() to test for such flags. Such cases will either show the subsequent arg as the value, or None if the flag was the last arg in argv.

Example usage, given argv = ["--b=1", "--c", "2"]: expect(env).that_action(action).contains_flag_values([ (”–b”, “1”), (”–c”, “2”) ])

PARAMETERS ¶

flag_values¶:: (list of (str name, str) tuples) Include the leading “–” in the flag name. Order and duplicates aren’t checked. Flags without a value found use None as their value.

ActionSubject.contains_none_of_flag_values¶

ActionSubject.contains_none_of_flag_values(flag_values)

Assert that an action’s argv has none of the given (”–flag”, “value”) entries.

Method: ActionSubject.contains_none_of_flag_values

This parses the argv, assuming the typical formats (--flag=value, --flag value, and --flag). Note, however, that for the --flag value and --flag forms, the parsing can’t know how many args, if any, a flag actually consumes, so it simply takes the first following arg, if any, as the matching value.

NOTE: This function can give misleading results checking flags that don’t consume any args (e.g. boolean flags). Use has_flags_specified() to test for such flags.

PARAMETERS ¶

flag_values¶:: (list of (str name, str value) tuples) Include the leading “–” in the flag name. Order and duplicates aren’t checked.

ActionSubject.contains_at_least_inputs¶

ActionSubject.contains_at_least_inputs(inputs)

Assert the action’s inputs contains at least all of inputs.

Method: ActionSubject.contains_at_least_inputs

Example usage: expect(env).that_action(action).contains_at_least_inputs([])

PARAMETERS ¶

inputs¶:: (collection of File) All must be present. Multiplicity is respected.

RETURNS ¶: Ordered (see _ordered_incorrectly_new).

ActionSubject.content¶

ActionSubject.content()

Returns a StrSubject for Action.content.

Method: ActionSubject.content

RETURNS ¶: StrSubject object.

ActionSubject.env¶

ActionSubject.env()

Returns a DictSubject for Action.env.

Method: ActionSubject.env

Analysis test¶

Support for testing analysis phase logic, such as rules.

analysis_test¶

analysis_test(name, target, impl, expect_failure=False, attrs={}, fragments=[], config_settings={}, extra_target_under_test_aspects=[], collect_actions_recursively=False)

Creates an analysis test from its implementation function.

An analysis test verifies the behavior of a “real” rule target by examining and asserting on the providers given by the real target.

Each analysis test is defined in an implementation function. This function handles the boilerplate to create and return a test target and captures the implementation function’s name so that it can be printed in test feedback.

An example of an analysis test:

def basic_test(name):
    my_rule(name = name + "_subject", ...)

    analysistest(name = name, target = name + "_subject", impl = _your_test)

def _your_test(env, target, actions):
    env.assert_that(target).runfiles().contains_at_least("foo.txt")
    env.assert_that(find_action(actions, generating="foo.txt")).argv().contains("--a")

PARAMETERS ¶

name¶:: Name of the target. It should be a Starlark identifier, matching pattern ‘[A-Za-z_][A-Za-z0-9_]*’.
target¶:: The target to test.
impl¶:: The implementation function of the unit test.
expect_failure¶:: (default False) If true, the analysis test will expect the target to fail. Assertions can be made on the underlying failure using truth.expect_failure
attrs¶:: (default {}) An optional dictionary to supplement the attrs passed to the unit test’s rule() constructor.
fragments¶:: (default []) An optional list of fragment names that can be used to give rules access to language-specific parts of configuration.
config_settings¶:: (default {}) A dictionary of configuration settings to change for the target under test and its dependencies. This may be used to essentially change ‘build flags’ for the target under test, and may thus be utilized to test multiple targets with different flags in a single build
extra_target_under_test_aspects¶:: (default []) An optional list of aspects to apply to the target_under_test in addition to those set up by default for the test harness itself.
collect_actions_recursively¶:: (default False) If true, runs testing_aspect over all attributes, otherwise it is only applied to the target under test.

RETURNS ¶: (None)

test_suite¶

test_suite(name, tests, test_kwargs={})

Instantiates given test macros and gathers their main targets into a test_suite.

Use this function to wrap all tests into a single target.

def simple_test_suite(name):
  test_suite(
      name = name,
      tests = [
          your_test,
          your_other_test,
      ]
  )

Then, in your BUILD file, simply load the macro and invoke it to have all of the targets created:

load("//path/to/your/package:tests.bzl", "simple_test_suite")
simple_test_suite(name = "simple_test_suite")

PARAMETERS ¶

name¶:: The name of the test_suite target.
tests¶:: A list of test macros, each taking name as a parameter, which will be passed the computed name of the test.
test_kwargs¶:: (default {}) Additional kwargs to pass onto each test function call.

BoolSubject¶

BoolSubject.new¶

BoolSubject.new(value, meta)

Creates a “BoolSubject” struct.

Method: BoolSubject.new

PARAMETERS ¶

value¶:: (bool) the value to assert against.
meta¶:: (ExpectMeta) the metadata about the call chain.

RETURNS ¶: A BoolSubject.

BoolSubject.equals¶

BoolSubject.equals(expected)

Assert that the bool is equal to expected.

Method: BoolSubject.equals

PARAMETERS ¶

expected¶:: (bool) the expected value.

BoolSubject.not_equals¶

BoolSubject.not_equals(unexpected)

Assert that the bool is not equal to unexpected.

Method: BoolSubject.not_equals

PARAMETERS ¶

unexpected¶:: (bool) the value actual cannot equal.

CollectionSubject¶

CollectionSubject.new¶

CollectionSubject.new(values, meta, container_name=”values”, sortable=True, element_plural_name=”elements”)

Creates a “CollectionSubject” struct.

Method: CollectionSubject.new

Public Attributes:

actual: The wrapped collection.

PARAMETERS ¶

values¶:: ([collection]) the values to assert against.
meta¶:: (ExpectMeta) the metadata about the call chain.
container_name¶:: (default "values") (str) conceptual name of the container.
sortable¶:: (default True) (bool) True if output should be sorted for display, False if not.
element_plural_name¶:: (default "elements") (str) the plural word for the values in the container.

RETURNS ¶: CollectionSubject.

CollectionSubject.has_size¶

CollectionSubject.has_size(expected)

Asserts that expected is the size of the collection.

Method: CollectionSubject.has_size

PARAMETERS ¶

expected¶:: ([int]) the expected size of the collection.

CollectionSubject.contains¶

CollectionSubject.contains(expected)

Asserts that expected is within the collection.

Method: CollectionSubject.contains

PARAMETERS ¶

expected¶:: (str) the value that must be present.

CollectionSubject.contains_exactly¶

CollectionSubject.contains_exactly(expected)

Check that a collection contains exactly the given elements.

Method: CollectionSubject.contains_exactly

Multiplicity is respected.
The collection must contain all the values, no more or less.
Checking that the order of matches is the same as the passed-in matchers order can be done by call in_order().

The collection must contain all the values and no more. Multiplicity of values is respected. Checking that the order of matches is the same as the passed-in matchers order can done by calling in_order().

PARAMETERS ¶

expected¶:: (list) values that must exist.

RETURNS ¶: Ordered (see _ordered_incorrectly_new).

CollectionSubject.contains_exactly_predicates¶

CollectionSubject.contains_exactly_predicates(expected)

Check that the values correspond 1:1 to the predicates.

Method: CollectionSubject.contains_exactly_predicates

There must be a 1:1 correspondence between the container values and the predicates.
Multiplicity is respected (i.e., if the same predicate occurs twice, then two distinct elements must match).
Matching occurs in first-seen order. That is, a predicate will “consume” the first value in actual_container it matches.
The collection must match all the predicates, no more or less.
Checking that the order of matches is the same as the passed-in matchers order can be done by call in_order().

Note that confusing results may occur if predicates with overlapping match conditions are used. For example, given: actual=[“a”, “ab”, “abc”], predicates=[, , ]

Then the result will be they aren’t equal: the first two predicates consume “a” and “ab”, leaving only “abc” for the predicate to match against, which fails.

PARAMETERS ¶

expected¶:: (list of [Matcher]) that must match.

RETURNS ¶: Ordered (see _ordered_incorrectly_new).

CollectionSubject.contains_none_of¶

CollectionSubject.contains_none_of(values)

Asserts the collection contains none of values.

Method: CollectionSubject.contains_none_of

PARAMETERS ¶

values¶:: ([collection]) values of which none of are allowed to exist.

CollectionSubject.contains_predicate¶

CollectionSubject.contains_predicate(matcher)

Asserts that matcher matches at least one value.

Method: CollectionSubject.contains_predicate

PARAMETERS ¶

matcher¶:: ([Matcher]) (see matchers struct).

CollectionSubject.contains_at_least¶

CollectionSubject.contains_at_least(expect_contains)

Assert that the collection is a subset of the given predicates.

Method: CollectionSubject.contains_at_least

The collection must contain all the values. It can contain extra elements. The multiplicity of values is respected. Checking that the relative order of matches is the same as the passed-in expected values order can done by calling in_order().

PARAMETERS ¶

expect_contains¶:: (list) values that must be in the collection.

RETURNS ¶: Ordered (see _ordered_incorrectly_new).

CollectionSubject.contains_at_least_predicates¶

CollectionSubject.contains_at_least_predicates(matchers)

Assert that the collection is a subset of the given predicates.

Method: CollectionSubject.contains_at_least_predicates

The collection must match all the predicates. It can contain extra elements. The multiplicity of matchers is respected. Checking that the relative order of matches is the same as the passed-in matchers order can done by calling in_order().

PARAMETERS ¶

matchers¶:: (list of [Matcher]) (see matchers struct).

RETURNS ¶: Ordered (see _ordered_incorrectly_new).

CollectionSubject.not_contains_predicate¶

CollectionSubject.not_contains_predicate(matcher)

Asserts that matcher matches no values in the collection.

Method: CollectionSubject.not_contains_predicate

PARAMETERS ¶

matcher¶:: [Matcher] object (see matchers struct).

DepsetFileSubject¶

DepsetFileSubject.new¶

DepsetFileSubject.new(files, meta, container_name=”depset”, element_plural_name=”files”)

Creates a DepsetFileSubject asserting on files.

Method: DepsetFileSubject.new

PARAMETERS ¶

files¶:: (depset of File) the values to assert on.
meta¶:: (ExpectMeta) of call chain information.
container_name¶:: (default "depset") (str) conceptual name of the container.
element_plural_name¶:: (default "files") (str) the plural word for the values in the container.

RETURNS ¶: DepsetFileSubject object.

DepsetFileSubject.contains¶

DepsetFileSubject.contains(expected)

Asserts that the depset of files contains the provided path/file.

Method: DepsetFileSubject.contains

PARAMETERS ¶

expected¶:: (str | File) If a string path is provided, it is compared to the short path of the files and are formatted using [ExpectMeta.format_str] and its current contextual keywords. Note that, when using File objects, two files’ configurations must be the same for them to be considered equal.

DepsetFileSubject.contains_at_least¶

DepsetFileSubject.contains_at_least(expected)

Asserts that the depset of files contains at least the provided paths.

Method: DepsetFileSubject.contains_at_least

PARAMETERS ¶

expected¶:: ([collection] of str | collection of File) multiplicity is respected. If string paths are provided, they are compared to the short path of the files and are formatted using ExpectMeta.format_str and its current contextual keywords. Note that, when using File objects, two files’ configurations must be the same for them to be considered equal.

RETURNS ¶: Ordered (see _ordered_incorrectly_new).

DepsetFileSubject.contains_any_in¶

DepsetFileSubject.contains_any_in(expected)

Asserts that any of the values in expected exist.

Method: DepsetFileSubject.contains_any_in

PARAMETERS ¶

expected¶:: ([collection] of str paths | [collection] of File) at least one of the values must exist. Note that, when using File objects, two files’ configurations must be the same for them to be considered equal. When string paths are provided, they are compared to File.short_path.

DepsetFileSubject.contains_at_least_predicates¶

DepsetFileSubject.contains_at_least_predicates(matchers)

Assert that the depset is a subset of the given predicates.

Method: DepsetFileSubject.contains_at_least_predicates

The depset must match all the predicates. It can contain extra elements. The multiplicity of matchers is respected. Checking that the relative order of matches is the same as the passed-in matchers order can done by calling in_order().

PARAMETERS ¶

matchers¶:: (list of [Matcher]) (see matchers struct) that accept File objects.

RETURNS ¶: Ordered (see _ordered_incorrectly_new).

DepsetFileSubject.contains_predicate¶

DepsetFileSubject.contains_predicate(matcher)

Asserts that matcher matches at least one value.

Method: DepsetFileSubject.contains_predicate

PARAMETERS ¶

matcher¶:: [Matcher] (see matching struct) that accepts File objects.

DepsetFileSubject.contains_exactly¶

DepsetFileSubject.contains_exactly(paths)

Asserts the depset of files contains exactly the given paths.

Method: DepsetFileSubject.contains_exactly

PARAMETERS ¶

paths¶:: ([collection] of str) the paths that must exist. These are compared to the short_path values of the files in the depset. All the paths, and no more, must exist.

DepsetFileSubject.not_contains¶

DepsetFileSubject.not_contains(short_path)

Asserts that short_path is not in the depset.

Method: DepsetFileSubject.not_contains_predicate

PARAMETERS ¶

short_path¶:: (str) the short path that should not be present.

DepsetFileSubject.not_contains_predicate¶

DepsetFileSubject.not_contains_predicate(matcher)

Asserts that nothing in the depset matches matcher.

Method: DepsetFileSubject.not_contains_predicate

PARAMETERS ¶

matcher¶:: ([Matcher]) that must match. It operates on File objects.

DictSubject¶

DictSubject.new¶

DictSubject.new(actual, meta, container_name=”dict”, key_plural_name=”keys”)

Creates a new DictSubject.

Method: DictSubject.new

PARAMETERS ¶

actual¶:: (dict) the dict to assert against.
meta¶:: (ExpectMeta) of call chain information.
container_name¶:: (default "dict") (str) conceptual name of the dict.
key_plural_name¶:: (default "keys") (str) the plural word for the keys of the dict.

RETURNS ¶: New DictSubject struct.

DictSubject.contains_at_least¶

DictSubject.contains_at_least(at_least)

Assert the dict has at least the entries from at_least.

Method: DictSubject.contains_at_least

PARAMETERS ¶

at_least¶:: (dict) the subset of keys/values that must exist. Extra keys are allowed. Order is not checked.

DictSubject.contains_exactly¶

DictSubject.contains_exactly(expected)

Assert the dict has exactly the provided values.

Method: DictSubject.contains_exactly

PARAMETERS ¶

expected¶:: (dict) the values that must exist. Missing values or extra values are not allowed. Order is not checked.

DictSubject.contains_none_of¶

DictSubject.contains_none_of(none_of)

Assert the dict contains none of none_of keys/values.

Method: DictSubject.contains_none_of

PARAMETERS ¶

none_of¶:: (dict) the keys/values that must not exist. Order is not checked.

DictSubject.keys¶

DictSubject.keys()

Returns a CollectionSubject for the dict’s keys.

Method: DictSubject.keys

RETURNS ¶: CollectionSubject of the keys.

ExecutionInfoSubject¶

ExecutionInfoSubject.new¶

ExecutionInfoSubject.new(info, meta)

Create a new ExecutionInfoSubject

Method: ExecutionInfoSubject.new

PARAMETERS ¶

info¶:: ([testing.ExecutionInfo]) provider instance.
meta¶:: (ExpectMeta) of call chain information.

RETURNS ¶: ExecutionInfoSubject struct.

ExecutionInfoSubject.requirements¶

ExecutionInfoSubject.requirements()

Create a DictSubject for the requirements values.

Method: ExecutionInfoSubject.requirements

RETURNS ¶: DictSubject of the requirements.

ExecutionInfoSubject.exec_group¶

ExecutionInfoSubject.exec_group()

Create a StrSubject for the exec_group value.

Method: ExecutionInfoSubject.exec_group

RETURNS ¶: A StrSubject for the exec group.

Expect¶

Expect.new_from_env¶

Expect.new_from_env(env)

Wrapper around env.

This is the entry point to the Truth-style assertions. Example usage: expect = expect(env) expect.that_action(action).contains_at_least_args(…)

The passed in env object allows optional attributes to be set to customize behavior. Usually this is helpful for testing. See _fake_env() in truth_tests.bzl for examples.

fail: callable that takes a failure message. If present, it will be called instead of the regular Expect.add_failure logic.
get_provider: callable that takes 2 positional args (target and provider) and returns the found provider or fails.
has_provider: callable that takes 2 positional args (a Target and a [provider]) and returns bool (True if present, False otherwise) or fails.

PARAMETERS ¶

env¶:: unittest env struct, or some approximation. There are several attributes that override regular behavior; see above doc.

RETURNS ¶: Expect object

Expect.new¶

Expect.new(env, meta)

Creates a new Expect object.

Internal; only other Expect methods should be calling this.

PARAMETERS ¶

env¶:: unittest env struct or some approximation.
meta¶:: (ExpectMeta) metadata about call chain and state.

RETURNS ¶: Expect object

Expect.that_action¶

Expect.that_action(action)

Creates a subject for asserting Actions.

PARAMETERS ¶

action¶:: (Action) the action to check.

RETURNS ¶: ActionSubject object.

Expect.that_bool¶

Expect.that_bool(value, expr=”boolean”)

Creates a subject for asserting a boolean.

PARAMETERS ¶

value¶:: (bool) the bool to check.
expr¶:: (default "boolean") (str) the starting “value of” expression to report in errors.

RETURNS ¶: BoolSubject object.

Expect.that_collection¶

Expect.that_collection(collection, expr=”collection”)

Creates a subject for asserting collections.

PARAMETERS ¶

collection¶:: The collection (list or depset) to assert.
expr¶:: (default "collection") (str) the starting “value of” expression to report in errors.

RETURNS ¶: CollectionSubject object.

Expect.that_depset_of_files¶

Expect.that_depset_of_files(depset_files)

Creates a subject for asserting a depset of files.

Method: Expect.that_depset_of_files

PARAMETERS ¶

depset_files¶:: (depset of File) the values to assert on.

RETURNS ¶: DepsetFileSubject object.

Expect.that_dict¶

Expect.that_dict(mapping, meta=None)

Creates a subject for asserting a dict.

Method: Expect.that_dict

PARAMETERS ¶

mapping¶:: (dict) the values to assert on
meta¶:: (default None) (ExpectMeta) optional custom call chain information to use instead

RETURNS ¶: DictSubject object.

Expect.that_file¶

Expect.that_file(file, meta=None)

Creates a subject for asserting a file.

Method: Expect.that_file

PARAMETERS ¶

file¶:: (File) the value to assert.
meta¶:: (default None) (ExpectMeta) optional custom call chain information to use instead

RETURNS ¶: FileSubject object.

Expect.that_int¶

Expect.that_int(value, expr=”integer”)

Creates a subject for asserting an int.

Method: Expect.that_int

PARAMETERS ¶

value¶:: ([int]) the value to check against.
expr¶:: (default "integer") (str) the starting “value of” expression to report in errors.

RETURNS ¶: IntSubject object.

Expect.that_str¶

Expect.that_str(value)

Creates a subject for asserting a str.

PARAMETERS ¶

value¶:: (str) the value to check against.

RETURNS ¶: StrSubject object.

Expect.that_target¶

Expect.that_target(target)

Creates a subject for asserting a Target.

This adds the following parameters to ExpectMeta.format_str: {package}: The target’s package, e.g. “foo/bar” from “//foo/bar:baz” {name}: The target’s base name, e.g., “baz” from “//foo/bar:baz”

PARAMETERS ¶

target¶:: (Target) subject target to check against.

RETURNS ¶: TargetSubject object.

Expect.where¶

Expect.where(details)

Add additional information about the assertion.

This is useful for attaching information that isn’t part of the call chain or some reason. Example usage:

expect(env).where(platform=ctx.attr.platform).that_str(...)

Would include “platform: {ctx.attr.platform}” in failure messages.

PARAMETERS ¶

details¶:: (dict of str to value) Each named arg is added to the metadata details with the provided string, which is printed as part of displaying any failures.

RETURNS ¶: Expect object with separate metadata derived from the original self.

ExpectMeta¶

ExpectMeta object implementation.

ExpectMeta.new¶

ExpectMeta.new(env, exprs=[], details=[], format_str_kwargs=None)

Creates a new “ExpectMeta” struct”.

Method: ExpectMeta.new

ExpectMeta objects are internal helpers for the Expect object and Subject objects. They are used for Subjects to store and communicate state through a series of call chains and asserts.

This constructor should only be directly called by Expect objects. When a parent Subject is creating a child-Subject, then [derive()] should be used.

Env objects¶

The env object basically provides a way to interact with things outside of the truth assertions framework. This allows easier testing of the framework itself and decouples it from a particular test framework (which makes it usuable by by rules_testing’s analysis_test and skylib’s analysistest)

The env object requires the following attribute:

ctx: The test’s ctx.

The env object allows the following attributes to customize behavior:

fail: A callable that accepts a single string, which is the failure message. Its return value is ignored. This is called when an assertion fails. It’s generally expected that it records a failure instead of immediately failing.
has_provider: (callable) it accepts two positional args, target and provider and returns bool. This is used to implement Provider in target operations.
get_provider: (callable) it accepts two positional args, target and provider and returns the provder value. This is used to implement target[Provider].

PARAMETERS ¶

env¶:: unittest env struct or some approximation.
exprs¶:: (default []) (list of str) the expression strings of the call chain for the subject.
details¶:: (default []) (list of str) additional details to print on error. These are usually informative details of the objects under test.
format_str_kwargs¶:: (default None) optional dict of format() kwargs. These kwargs are propagated through derive() calls and used when ExpectMeta.format_str() is called.

RETURNS ¶: ExpectMeta object.

ExpectMeta.derive¶

ExpectMeta.derive(expr=None, details=None, format_str_kwargs={})

Create a derivation of the current meta object for a child-Subject.

Method: ExpectMeta.derive

When a Subject needs to create a child-Subject, it derives a new meta object to pass to the child. This separates the parent’s state from the child’s state and allows any failures generated by the child to include the context of the parent creator.

Example usage:

def _foo_subject_action_named(self, name):
    meta = self.meta.derive("action_named({})".format(name),
                            "action: {}".format(...))
    return ActionSubject(..., meta)
def _foo_subject_name(self):
    # No extra detail to include)
    meta self.meta.derive("name()", None)

PARAMETERS ¶

expr¶:: (default None) (str) human-friendly description of the call chain expression. e.g., if foo_subject.bar_named("baz") returns a child-subject, then “bar_named(“bar”)” would be the expression.
details¶:: (default None) (optional list of str) human-friendly descriptions of additional detail to include in errors. This is usually additional information the child Subject wouldn’t include itself. e.g. if foo.first_action_argv().contains(1), returned a ListSubject, then including “first action: Action FooCompile” helps add context to the error message. If there is no additional detail to include, pass None.
format_str_kwargs¶:: (default {}) (dict of format()-kwargs) additional kwargs to make available to format_str calls.

RETURNS ¶: ExpectMeta object.

ExpectMeta.format_str¶

ExpectMeta.format_str(template)

Interpolate contextual keywords into a string.

This uses the normal format() style (i.e. using {}). Keywords refer to parts of the call chain.

The particular keywords supported depend on the call chain. The following are always present: {workspace}: The name of the workspace, e.g. “rules_proto”. {test_name}: The base name of the current test.

PARAMETERS ¶

template¶:: (str) the format template string to use.

RETURNS ¶: str; the template with parameters replaced.

ExpectMeta.get_provider¶

ExpectMeta.get_provider(target, provider)

Get a provider from a target.

This is equivalent to target[provider]; the extra level of indirection is to aid testing.

PARAMETERS ¶

target¶:: (Target) the target to get the provider from.
provider¶:: The provider type to get.

RETURNS ¶: The found provider, or fails if not present.

ExpectMeta.has_provider¶

ExpectMeta.has_provider(target, provider)

Tells if a target has a provider.

This is equivalent to provider in target; the extra level of indirection is to aid testing.

PARAMETERS ¶

target¶:: (Target) the target to check for the provider.
provider¶:: the provider type to check for.

RETURNS ¶: True if the target has the provider, False if not.

ExpectMeta.add_failure¶

ExpectMeta.add_failure(problem, actual)

Adds a failure with context.

Method: ExpectMeta.add_failure

Adds the given error message. Context from the subject and prior call chains is automatically added.

PARAMETERS ¶

problem¶:: (str) a string describing the expected value or problem detected, and the expected values that weren’t satisfied. A colon should be used to separate the description from the values. The description should be brief and include the word “expected”, e.g. “expected: foo”, or “expected values missing: ”, the key point being the reader can easily take the values shown and look for it in the actual values displayed below it.
actual¶:: (str) a string describing the values observed. A colon should be used to separate the description from the observed values. The description should be brief and include the word “actual”, e.g., “actual: bar”. The values should include the actual, observed, values and pertinent information about them.

ExpectMeta.call_fail¶

ExpectMeta.call_fail(msg)

Adds a failure to the test run.

PARAMETERS ¶

msg¶:: (str) the failure message.

FileSubject¶

FileSubject.new¶

FileSubject.new(file, meta)

Creates a FileSubject asserting against the given file.

Method: FileSubject.new

PARAMETERS ¶

file¶:: (File) the file to assert against.
meta¶:: (ExpectMeta)

RETURNS ¶: FileSubject object.

FileSubject.equals¶

FileSubject.equals(expected)

Asserts that expected references the same file as self.

This uses Bazel’s notion of File equality, which usually includes the configuration, owning action, internal hash, etc of a File. The particulars of comparison depend on the actual Java type implementing the File object (some ignore owner, for example).

NOTE: This does not compare file content. Starlark cannot read files.

NOTE: Same files generated by different owners are likely considered not equal to each other. The alternative for this is to assert the File.path paths are equal using [FileSubject.path()]

Method: FileSubject.equals

PARAMETERS ¶

expected¶:: undocumented

FileSubject.path¶

FileSubject.path()

Returns a StrSubject asserting on the files path value.

Method: FileSubject.path

RETURNS ¶: StrSubject object.

FileSubject.short_path_equals¶

FileSubject.short_path_equals(path)

Asserts the file’s short path is equal to the given path.

Method: FileSubject.short_path_equals

PARAMETERS ¶

path¶:: (str) the value the file’s short_path must be equal to.

InstrumentedFilesInfoSubject¶

InstrumentedFilesInfoSubject.new¶

InstrumentedFilesInfoSubject.new(info, meta)

Creates a subject to assert on InstrumentedFilesInfo providers.

Method: InstrumentedFilesInfoSubject.new

PARAMETERS ¶

info¶:: ([InstrumentedFilesInfo]) provider instance.
meta¶:: (ExpectMeta) the meta data about the call chain.

RETURNS ¶: An InstrumentedFilesInfoSubject struct.

InstrumentedFilesInfoSubject.instrumented_files¶

InstrumentedFilesInfoSubject.instrumented_files()

Returns a DesetFileSubject of the instrumented files.

Method: InstrumentedFilesInfoSubject.instrumented_files

InstrumentedFilesInfoSubject.metadata_files¶

InstrumentedFilesInfoSubject.metadata_files()

Returns a DesetFileSubject of the metadata files.

Method: InstrumentedFilesInfoSubject.metadata_files

IntSubject¶

IntSubject.new¶

IntSubject.new(value, meta)

Create an “IntSubject” struct.

Method: IntSubject.new

PARAMETERS ¶

value¶:: (optional [int]) the value to perform asserts against may be None.
meta¶:: (ExpectMeta) the meta data about the call chain.

RETURNS ¶: IntSubject.

IntSubject.equals¶

IntSubject.equals(other)

Assert that the subject is equal to the given value.

Method: IntSubject.equals

PARAMETERS ¶

other¶:: ([int]) value the subject must be equal to.

IntSubject.is_greater_than¶

IntSubject.is_greater_than(other)

Asserts that the subject is greater than the given value.

Method: IntSubject.is_greater_than

PARAMETERS ¶

other¶:: ([int]) value the subject must be greater than.

IntSubject.not_equals¶

IntSubject.not_equals(unexpected)

Assert that the int is not equal to unexpected.

Method: IntSubject.not_equals

PARAMETERS ¶

unexpected¶:: ([int]) the value actual cannot equal.

LabelSubject¶

LabelSubject.new¶

LabelSubject.new(label, meta)

Creates a new LabelSubject for asserting Label objects.

Method: LabelSubject.new

PARAMETERS ¶

label¶:: (Label) the label to check against.
meta¶:: (ExpectMeta) the metadata about the call chain.

RETURNS ¶: LabelSubject.

LabelSubject.equals¶

LabelSubject.equals(other)

Asserts the label is equal to other.

Method: LabelSubject.equals

PARAMETERS ¶

other¶:: (Label | str) the expected value. If a str is passed, it will be converted to a Label using the Label function.

LabelSubject.is_in¶

LabelSubject.is_in(any_of)

Asserts that the label is any of the provided values.

PARAMETERS ¶

any_of¶:: ([collection] of (Label | str)) If strings are provided, they must be parsable by Label.

Ordered¶

IN_ORDER.in_order¶

IN_ORDER.in_order()

Checks that the values were in order.

OrderedIncorrectly.new¶

OrderedIncorrectly.new(format_problem, format_actual, meta)

Creates a new Ordered object that fails due to incorrectly ordered values.

This creates an Ordered object that always fails. If order is correct, use the _IN_ORDER constant.

PARAMETERS ¶

format_problem¶:: (callable) accepts no args and returns string (the reported problem description).
format_actual¶:: (callable) accepts not args and returns tring (the reported actual description).
meta¶:: (ExpectMeta) used to report the failure.

RETURNS ¶: Ordered object.

OrderedIncorrectly.in_order¶

OrderedIncorrectly.in_order()

Checks that the values were in order.

RunEnvironmentInfoSubject¶

RunEnvironmentInfoSubject.new¶

RunEnvironmentInfoSubject.new(info, meta)

Creates a new RunEnvironmentInfoSubject

Method: RunEnvironmentInfoSubject.new

PARAMETERS ¶

info¶:: ([RunEnvironmentInfo]) provider instance.
meta¶:: (ExpectMeta) of call chain information.

RunEnvironmentInfoSubject.environment¶

RunEnvironmentInfoSubject.environment()

Creates a DictSubject to assert on the environment dict.

Method: RunEnvironmentInfoSubject.environment

RETURNS ¶: DictSubject of the str->str environment map.

RunEnvironmentInfoSubject.inherited_environment¶

RunEnvironmentInfoSubject.inherited_environment()

Creates a CollectionSubject to assert on the inherited_environment list.

Method: RunEnvironmentInfoSubject.inherited_environment

RETURNS ¶: CollectionSubject of str; from the [RunEnvironmentInfo.inherited_environment] list.

RunfilesSubject¶

RunfilesSubject.new¶

RunfilesSubject.new(runfiles, meta, kind=None)

Creates a “RunfilesSubject” struct.

Method: RunfilesSubject.new

PARAMETERS ¶

runfiles¶:: ([runfiles]) the runfiles to check against.
meta¶:: (ExpectMeta) the metadata about the call chain.
kind¶:: (default None) (optional str) what type of runfiles they are, usually “data” or “default”. If not known or not applicable, use None.

RETURNS ¶: RunfilesSubject object.

RunfilesSubject.contains¶

RunfilesSubject.contains(expected)

Assert that the runfiles contains the provided path.

Method: RunfilesSubject.contains

PARAMETERS ¶

expected¶:: (str) the path to check is present. This will be formatted using ExpectMeta.format_str and its current contextual keywords. Note that paths are runfiles-root relative (i.e. you likely need to include the workspace name.)

RunfilesSubject.contains_at_least¶

RunfilesSubject.contains_at_least(paths)

Assert that the runfiles contains at least all of the provided paths.

Method: RunfilesSubject.contains_at_least

All the paths must exist, but extra paths are allowed. Order is not checked. Multiplicity is respected.

PARAMETERS ¶

paths¶:: ((collection of str) | [runfiles]) the paths that must exist. If a collection of strings is provided, they will be formatted using [ExpectMeta.format_str], so its template keywords can be directly passed. If a runfiles object is passed, it is converted to a set of path strings.

RunfilesSubject.contains_predicate¶

RunfilesSubject.contains_predicate(matcher)

Asserts that matcher matches at least one value.

Method: RunfilesSubject.contains_predicate

PARAMETERS ¶

matcher¶:: callable that takes 1 positional arg (str path) and returns boolean.

RunfilesSubject.contains_exactly¶

RunfilesSubject.contains_exactly(paths)

Asserts that the runfiles contains_exactly the set of paths

Method: RunfilesSubject.contains_exactly

PARAMETERS ¶

paths¶:: ([collection] of str) the paths to check. These will be formatted using meta.format_str, so its template keywords can be directly passed. All the paths must exist in the runfiles exactly as provided, and no extra paths may exist.

RunfilesSubject.contains_none_of¶

RunfilesSubject.contains_none_of(paths, require_workspace_prefix=True)

Asserts the runfiles contain none of paths.

Method: RunfilesSubject.contains_none_of

PARAMETERS ¶

paths¶:: ([collection] of str) the paths that should not exist. They should be runfiles root-relative paths (not workspace relative). The value is formatted using ExpectMeta.format_str and the current contextual keywords.
require_workspace_prefix¶:: (default True) (bool) True to check that the path includes the workspace prefix. This is to guard against accidentallly passing a workspace relative path, which will (almost) never exist, and cause the test to always pass. Specify False if the file being checked for is actually a runfiles-root relative path that isn’t under the workspace itself.

RunfilesSubject.not_contains¶

RunfilesSubject.not_contains(path, require_workspace_prefix=True)

Assert that the runfiles does not contain the given path.

Method: RunfilesSubject.not_contains

PARAMETERS ¶

path¶:: (str) the path that should not exist. It should be a runfiles root-relative path (not workspace relative). The value is formatted using format_str, so its template keywords can be directly passed.
require_workspace_prefix¶:: (default True) (bool) True to check that the path includes the workspace prefix. This is to guard against accidentallly passing a workspace relative path, which will (almost) never exist, and cause the test to always pass. Specify False if the file being checked for is actually a runfiles-root relative path that isn’t under the workspace itself.

RunfilesSubject.not_contains_predicate¶

RunfilesSubject.not_contains_predicate(matcher)

Asserts that none of the runfiles match matcher.

Method: RunfilesSubject.not_contains_predicate

PARAMETERS ¶

matcher¶:: [Matcher] that accepts a string (runfiles root-relative path).

RunfilesSubject.check_workspace_prefix¶

RunfilesSubject.check_workspace_prefix(path)

PARAMETERS ¶

path¶:: undocumented

StrSubject¶

StrSubject.new¶

StrSubject.new(actual, meta)

Creates a subject for asserting strings.

Method: StrSubject.new

PARAMETERS ¶

actual¶:: (str) the string to check against.
meta¶:: (ExpectMeta) of call chain information.

RETURNS ¶: StrSubject object.

StrSubject.contains¶

StrSubject.contains(substr)

Assert that the subject contains the substring substr.

Method: StrSubject.contains

PARAMETERS ¶

substr¶:: (str) the substring to check for.

StrSubject.equals¶

StrSubject.equals(other)

Assert that the subject string equals the other string.

Method: StrSubject.equals

PARAMETERS ¶

other¶:: (str) the expected value it should equal.

StrSubject.not_equals¶

StrSubject.not_equals(unexpected)

Assert that the string is not equal to unexpected.

Method: BoolSubject.not_equals

PARAMETERS ¶

unexpected¶:: (str) the value actual cannot equal.

StrSubject.split¶

StrSubject.split(sep)

Return a CollectionSubject for the actual string split by sep.

Method: StrSubject.split

PARAMETERS ¶

sep¶:: undocumented

TargetSubject¶

TargetSubject wraps a Target object and provides method for asserting its state.

TargetSubject.new¶

TargetSubject.new(target, meta)

Creates a subject for asserting Targets.

Method: TargetSubject.new

Public attributes:

actual: The wrapped Target object.

PARAMETERS ¶

target¶:: (Target) the target to check against.
meta¶:: (ExpectMeta) metadata about the call chain.

RETURNS ¶: TargetSubject object

TargetSubject.runfiles¶

TargetSubject.runfiles()

Creates a subject asserting on the target’s default runfiles.

Method: TargetSubject.runfiles

RETURNS ¶: RunfilesSubject object.

TargetSubject.tags¶

TargetSubject.tags()

Gets the target’s tags as a CollectionSubject

Method: TargetSubject.tags

RETURNS ¶: CollectionSubject asserting the target’s tags.

TargetSubject.get_attr¶

TargetSubject.get_attr(name)

PARAMETERS ¶

name¶:: undocumented

TargetSubject.data_runfiles¶

TargetSubject.data_runfiles()

Creates a subject asserting on the target’s data runfiles.

Method: TargetSubject.data_runfiles

RETURNS ¶: RunfilesSubject object

TargetSubject.default_outputs¶

TargetSubject.default_outputs()

Creates a subject asserting on the target’s default outputs.

Method: TargetSubject.default_outputs

RETURNS ¶: DepsetFileSubject object.

TargetSubject.executable¶

TargetSubject.executable()

Creates a subject asesrting on the target’s executable File.

Method: TargetSubject.executable

RETURNS ¶: FileSubject object.

TargetSubject.failures¶

TargetSubject.failures()

Creates a subject asserting on the target’s failure message strings.

Method: TargetSubject.failures

RETURNS ¶: CollectionSubject of str.

TargetSubject.has_provider¶

TargetSubject.has_provider(provider)

Asserts that the target as provider provider.

Method: TargetSubject.has_provider

PARAMETERS ¶

provider¶:: The provider object to check for.

TargetSubject.label¶

TargetSubject.label()

Returns a LabelSubject for the target’s label value.

Method: TargetSubject.label

TargetSubject.output_group¶

TargetSubject.output_group(name)

Returns a DepsetFileSubject of the files in the named output group.

Method: TargetSubject.output_group

PARAMETERS ¶

name¶:: (str) an output group name. If it isn’t present, an error is raised.

RETURNS ¶: DepsetFileSubject of the named output group.

TargetSubject.provider¶

TargetSubject.provider(provider_key, factory=None)

Returns a subject for a provider in the target.

Method: TargetSubject.provider

PARAMETERS ¶

provider_key¶:: The provider key to create a subject for
factory¶:: (default None) optional callable. The factory function to use to create the subject for the found provider. Required if the provider key is not an inherently supported provider. It must have the following signature: def factory(value, /, *, meta).

RETURNS ¶: A subject wrapper of the provider value.

TargetSubject.action_generating¶

TargetSubject.action_generating(short_path)

Get the single action generating the given path.

Method: TargetSubject.action_generating

NOTE: in order to use this method, the target must have the TestingAspectInfo provider (added by the testing_aspect aspect.)

PARAMETERS ¶

short_path¶:: (str) the output’s short_path to match. The value is formatted using format_str, so its template keywords can be directly passed.

RETURNS ¶: ActionSubject for the matching action. If no action is found, or more than one action matches, then an error is raised.

TargetSubject.action_named¶

TargetSubject.action_named(mnemonic)

Get the single action with the matching mnemonic.

Method: TargetSubject.action_named

NOTE: in order to use this method, the target must have the [TestingAspectInfo] provider (added by the [testing_aspect] aspect.)

PARAMETERS ¶

mnemonic¶:: (str) the mnemonic to match

RETURNS ¶: ActionSubject. If no action matches, or more than one action matches, an error is raised.

TargetSubject.attr¶

TargetSubject.attr(name, factory=None)

Gets a subject-wrapped value for the named attribute.

Method: TargetSubject.attr

NOTE: in order to use this method, the target must have the TestingAspectInfo provider (added by the testing_aspect aspect.)

PARAMETERS ¶

name¶:: (str) the attribute to get. If it’s an unsupported attribute, and no explicit factory was provided, an error will be raised.
factory¶:: (default None) (callable) function to create the returned subject based on the attribute value. If specified, it takes precedence over the attributes that are inherently understood. It must have the following signature: def factory(value, *, meta), where value is the value of the attribute, and meta is the call chain metadata.

RETURNS ¶: A Subject-like object for the given attribute. The particular subject type returned depends on attribute and factory arg. If it isn’t know what type of subject to use for the attribute, an error is raised.

Truth¶

Truth-style asserts for Bazel’s Starlark.

These asserts follow the Truth-style way of performing assertions. This basically means the actual value is wrapped in a type-specific object that provides type-specific assertion methods. This style provides several benefits: * A fluent API that more directly expresses the assertion * More egonomic assert functions * Error messages with more informative context * Promotes code reuses at the type-level.

For more detailed documentation, see the docs on GitHub.

Basic usage¶

NOTE: This example assumes usage of [rules_testing]’s [analysis_test] framework, but that framework is not required.

def foo_test(env, target):
    subject = env.expect.that_target(target)
    subject.runfiles().contains_at_least(["foo.txt"])
    subject.executable().equals("bar.exe")

    subject = env.expect.that_action(...)
    subject.contains_at_least_args(...)

matching.contains¶

matching.contains(contained)

Match that contained is within the to-be-matched value.

This is equivalent to: contained in to_be_matched. See _match_is_in for the reversed operation.

PARAMETERS ¶

contained¶:: the value that to-be-matched value must contain.

RETURNS ¶: [Matcher] (see _match_custom).

matching.custom¶

matching.custom(desc, func)

Wrap an arbitrary function up as a Matcher.

Method: Matcher.new

Matcher struct attributes:

desc: (str) a human-friendly description
match: (callable) accepts 1 positional arg (the value to match) and returns bool (True if it matched, False if not).

PARAMETERS ¶

desc¶:: (str) a human-friendly string describing what is matched.
func¶:: (callable) accepts 1 positional arg (the value to match) and returns bool (True if it matched, False if not).

RETURNS ¶: [Matcher] (see above).

matching.equals_wrapper¶

matching.equals_wrapper(value)

Match that a value equals value, but use value as the desc.

This is a helper so that simple equality comparisons can re-use predicate based APIs.

PARAMETERS ¶

value¶:: object, the value that must be equal to.

RETURNS ¶: [Matcher] (see _match_custom()), whose description is value.

matching.file_basename_contains¶

matching.file_basename_contains(substr)

Match that a a File.basename string contains a substring.

PARAMETERS ¶

substr¶:: (str) the substring to match.

RETURNS ¶: [Matcher] (see _match_custom()).

matching.file_path_matches¶

matching.file_path_matches(pattern)

Match that a File.path string matches a glob-style pattern.

PARAMETERS ¶

pattern¶:: (str) the pattern to match. “*” can be used to denote “match anything”.

RETURNS ¶: [Matcher] (see _match_custom).

matching.is_in¶

matching.is_in(values)

Match that the to-be-matched value is in a collection of other values.

This is equivalent to: to_be_matched in values. See _match_contains for the reversed operation.

PARAMETERS ¶

values¶:: The collection that the value must be within.

RETURNS ¶: [Matcher] (see _match_custom()).

matching.never¶

matching.never(desc)

A matcher that never matches.

This is mostly useful for testing, as it allows preventing any match while providing a custom description.

PARAMETERS ¶

desc¶:: (str) human-friendly string.

RETURNS ¶: [Matcher] (see _match_custom).

matching.str_endswith¶

matching.str_endswith(suffix)

Match that a string contains another string.

PARAMETERS ¶

suffix¶:: (str) the suffix that must be present

RETURNS ¶: [Matcher] (see _match_custom).

matching.str_matches¶

matching.str_matches(pattern)

Match that a string matches a glob-style pattern.

PARAMETERS ¶

pattern¶:: (str) the pattern to match. * can be used to denote “match anything”. There is an implicit * at the start and end of the pattern.

RETURNS ¶: [Matcher] object.

matching.str_startswith¶

matching.str_startswith(prefix)

Match that a string contains another string.

PARAMETERS ¶

prefix¶:: (str) the prefix that must be present

RETURNS ¶: [Matcher] (see _match_custom).

subjects.bool¶

subjects.bool(value, meta)

Creates a “BoolSubject” struct.

Method: BoolSubject.new

PARAMETERS ¶

value¶:: (bool) the value to assert against.
meta¶:: (ExpectMeta) the metadata about the call chain.

RETURNS ¶: A BoolSubject.

subjects.collection¶

subjects.collection(values, meta, container_name=”values”, sortable=True, element_plural_name=”elements”)

Creates a “CollectionSubject” struct.

Method: CollectionSubject.new

Public Attributes:

actual: The wrapped collection.

PARAMETERS ¶

values¶:: ([collection]) the values to assert against.
meta¶:: (ExpectMeta) the metadata about the call chain.
container_name¶:: (default "values") (str) conceptual name of the container.
sortable¶:: (default True) (bool) True if output should be sorted for display, False if not.
element_plural_name¶:: (default "elements") (str) the plural word for the values in the container.

RETURNS ¶: CollectionSubject.

subjects.depset_file¶

subjects.depset_file(files, meta, container_name=”depset”, element_plural_name=”files”)

Creates a DepsetFileSubject asserting on files.

Method: DepsetFileSubject.new

PARAMETERS ¶

files¶:: (depset of File) the values to assert on.
meta¶:: (ExpectMeta) of call chain information.
container_name¶:: (default "depset") (str) conceptual name of the container.
element_plural_name¶:: (default "files") (str) the plural word for the values in the container.

RETURNS ¶: DepsetFileSubject object.

subjects.int¶

subjects.int(value, meta)

Create an “IntSubject” struct.

Method: IntSubject.new

PARAMETERS ¶

value¶:: (optional [int]) the value to perform asserts against may be None.
meta¶:: (ExpectMeta) the meta data about the call chain.

RETURNS ¶: IntSubject.

subjects.label¶

subjects.label(label, meta)

Creates a new LabelSubject for asserting Label objects.

Method: LabelSubject.new

PARAMETERS ¶

label¶:: (Label) the label to check against.
meta¶:: (ExpectMeta) the metadata about the call chain.

RETURNS ¶: LabelSubject.

truth.expect¶

truth.expect(env)

Wrapper around env.

This is the entry point to the Truth-style assertions. Example usage: expect = expect(env) expect.that_action(action).contains_at_least_args(…)

The passed in env object allows optional attributes to be set to customize behavior. Usually this is helpful for testing. See _fake_env() in truth_tests.bzl for examples.

fail: callable that takes a failure message. If present, it will be called instead of the regular Expect.add_failure logic.
get_provider: callable that takes 2 positional args (target and provider) and returns the found provider or fails.
has_provider: callable that takes 2 positional args (a Target and a [provider]) and returns bool (True if present, False otherwise) or fails.

PARAMETERS ¶

env¶:: unittest env struct, or some approximation. There are several attributes that override regular behavior; see above doc.

RETURNS ¶: Expect object

Util¶

Various utilities to aid with testing.

force_exec_config¶

force_exec_config(name, tools=[])

Rule to force arbitrary targets to cfg=exec so they can be tested when used as tools.

ATTRIBUTES ¶

name¶:: (required Name) A unique name for this target.
tools¶:: (optional list of labels, default []) A list of tools to force into the exec config

TestingAspectInfo¶

TestingAspectInfo(attrs, actions, vars, bin_path)

Details about a target-under-test useful for testing.

FIELDS ¶

attrs¶:: The raw attributes of the target under test.
actions¶:: The actions registered for the target under test.
vars¶:: The var dict (ctx.var) for the target under text.
bin_path¶:: str; the ctx.bin_dir.path value (aka execroot).

empty_file¶

empty_file(name)

Generates an empty file and returns the target name for it.

PARAMETERS ¶

name¶:: str, name of the generated output file.

RETURNS ¶: str, the name of the generated output.

get_target_actions¶

get_target_actions(env)

PARAMETERS ¶

env¶:: undocumented

get_target_attrs¶

get_target_attrs(env)

PARAMETERS ¶

env¶:: undocumented

helper_target¶

helper_target(rule, kwargs)

Define a target only used as a Starlark test input.

This is useful for e.g. analysis tests, which have to setup a small graph of targets that should only be built via the test (e.g. they may require config settings the test sets). Tags are added to hide the target from :all, /..., TAP, etc.

PARAMETERS ¶

rule¶:: rule-like function.
kwargs¶:: Any kwargs to pass to rule. Additional tags will be added to hide the target.

is_file¶

is_file(obj)

Tells if an object is a File object.

PARAMETERS ¶

obj¶:: undocumented

is_runfiles¶

is_runfiles(obj)

Tells if an object is a runfiles object.

PARAMETERS ¶

obj¶:: undocumented

merge_kwargs¶

merge_kwargs(kwargs)

Merges multiple dicts of kwargs.

This is similar to dict.update except: * If a key’s value is a list, it’ll be concatenated to any existing value. * An error is raised when the same non-list key occurs more than once.

PARAMETERS ¶

kwargs¶:: kwarg arg dicts to merge

RETURNS ¶: dict of the merged kwarg dics.

runfiles_map¶

runfiles_map(workspace_name, runfiles)

Convert runfiles to a path->file mapping.

This approximates how Bazel materializes the runfiles on the file system.

PARAMETERS ¶

workspace_name¶:: str; the workspace the runfiles belong to.
runfiles¶:: runfiles; the runfiles to convert to a map.

RETURNS ¶: dict[str, optional File] that maps the path under the runfiles root to it’s backing file. The file may be None if the path came from runfiles.empty_filenames.

runfiles_paths¶

runfiles_paths(workspace_name, runfiles)

Returns the root-relative short paths for the files in runfiles.

PARAMETERS ¶

workspace_name¶:: str, the workspace name (ctx.workspace_name).
runfiles¶:: runfiles, the runfiles to convert to short paths.

RETURNS ¶: list of short paths but runfiles root-relative. e.g. ‘myworkspace/foo/bar.py’.

short_paths¶

short_paths(files_depset)

Returns the short_path paths for a depset of files.

PARAMETERS ¶

files_depset¶:: undocumented

skip_test¶

skip_test(name)

Defines a test target that is always skipped.

This is useful for tests that should be skipped if some condition, determinable during the loading phase, isn’t met. The resulting target will show up as “SKIPPED” in the output.

If possible, prefer to use target_compatible_with to mark tests as incompatible. This avoids confusing behavior where the type of a target varies depending on loading-phase behavior.

PARAMETERS ¶

name¶:: The name of the target.

util.empty_file¶

util.empty_file(name)

Generates an empty file and returns the target name for it.

PARAMETERS ¶

name¶:: str, name of the generated output file.

RETURNS ¶: str, the name of the generated output.

util.helper_target¶

util.helper_target(rule, kwargs)

Define a target only used as a Starlark test input.

This is useful for e.g. analysis tests, which have to setup a small graph of targets that should only be built via the test (e.g. they may require config settings the test sets). Tags are added to hide the target from :all, /..., TAP, etc.

PARAMETERS ¶

rule¶:: rule-like function.
kwargs¶:: Any kwargs to pass to rule. Additional tags will be added to hide the target.

util.merge_kwargs¶

util.merge_kwargs(kwargs)

Merges multiple dicts of kwargs.

This is similar to dict.update except: * If a key’s value is a list, it’ll be concatenated to any existing value. * An error is raised when the same non-list key occurs more than once.

PARAMETERS ¶

kwargs¶:: kwarg arg dicts to merge

RETURNS ¶: dict of the merged kwarg dics.

util.runfiles_map¶

util.runfiles_map(workspace_name, runfiles)

Convert runfiles to a path->file mapping.

This approximates how Bazel materializes the runfiles on the file system.

PARAMETERS ¶

workspace_name¶:: str; the workspace the runfiles belong to.
runfiles¶:: runfiles; the runfiles to convert to a map.

RETURNS ¶: dict[str, optional File] that maps the path under the runfiles root to it’s backing file. The file may be None if the path came from runfiles.empty_filenames.

util.runfiles_paths¶

util.runfiles_paths(workspace_name, runfiles)

Returns the root-relative short paths for the files in runfiles.

PARAMETERS ¶

workspace_name¶:: str, the workspace name (ctx.workspace_name).
runfiles¶:: runfiles, the runfiles to convert to short paths.

RETURNS ¶: list of short paths but runfiles root-relative. e.g. ‘myworkspace/foo/bar.py’.

util.short_paths¶

util.short_paths(files_depset)

Returns the short_path paths for a depset of files.

PARAMETERS ¶

files_depset¶:: undocumented

util.skip_test¶

util.skip_test(name)

Defines a test target that is always skipped.

This is useful for tests that should be skipped if some condition, determinable during the loading phase, isn’t met. The resulting target will show up as “SKIPPED” in the output.

If possible, prefer to use target_compatible_with to mark tests as incompatible. This avoids confusing behavior where the type of a target varies depending on loading-phase behavior.

PARAMETERS ¶

name¶:: The name of the target.

recursive_testing_aspect¶

recursive_testing_aspect(name)

ASPECT ATTRIBUTES

Name	Type
*	String

ATTRIBUTES

Name	Description	Type	Mandatory	Default
name	A unique name for this target.	Name	required

testing_aspect¶

testing_aspect(name)

ASPECT ATTRIBUTES

ATTRIBUTES

Name	Description	Type	Mandatory	Default
name	A unique name for this target.	Name	required