View Source ExUnit.CaptureLog (ExUnit v1.18.0-dev)

Functionality to capture logs for testing.

Examples

defmodule AssertionTest do
  use ExUnit.Case

  import ExUnit.CaptureLog
  require Logger

  test "example" do
    {result, log} =
      with_log(fn ->
        Logger.error("log msg")
        2 + 2
      end)

    assert result == 4
    assert log =~ "log msg"
  end

  test "check multiple captures concurrently" do
    fun = fn ->
      for msg <- ["hello", "hi"] do
        assert capture_log(fn -> Logger.error(msg) end) =~ msg
      end

      Logger.debug("testing")
    end

    assert capture_log(fun) =~ "hello"
    assert capture_log(fun) =~ "testing"
  end
end

Summary

Functions

Captures Logger messages generated when evaluating fun.

Invokes the given fun and returns the result and captured log.

Functions

capture_log(opts \\ [], fun)

@spec capture_log(
  keyword(),
  (-> any())
) :: String.t()

Captures Logger messages generated when evaluating fun.

Returns the binary which is the captured output. The captured log messages will be formatted using Logger.default_formatter/1. Any option, besides the :level, will be forwarded as an override to the default formatter.

This function mutes the default logger handler and captures any log messages sent to Logger from the calling processes. It is possible to ensure explicit log messages from other processes are captured by waiting for their exit or monitor signal.

Note that when the async is set to true on use ExUnit.Case, messages from other tests might be captured. This is OK as long you consider such cases in your assertions, typically by using the =~/2 operator to perform partial matches.

It is possible to configure the level to capture with :level, which will set the capturing level for the duration of the capture, for instance, if the log level is set to :error, then any message with the lower level will be ignored. The default level is nil, which will capture all messages. Note this setting does not override the overall Logger.level/0 value. Therefore, if Logger.level/0 is set to a higher level than the one configured in this function, no message will be captured. The behaviour is undetermined if async tests change Logger level.

To get the result of the evaluation along with the captured log, use with_log/2.

with_log(opts \\ [], fun)

(since 1.13.0)
@spec with_log(
  keyword(),
  (-> result)
) :: {result, String.t()}
when result: any()

Invokes the given fun and returns the result and captured log.

It accepts the same arguments and options as capture_log/2.

Examples

{result, log} =
  with_log(fn ->
    Logger.error("log msg")
    2 + 2
  end)

assert result == 4
assert log =~ "log msg"