Interactive approval testing for your bash scripts.

approvals.bash is a lightweight interactive testing framework for your bash scripts.

With it, your tests will compare the output of a command with an expected output stored in the approvals folder.

• When the approval file does not exist, the actual output will be displayed, and you will be prompted to approve (and save) it.
• When the approval file exists, but contains different data, the diff will be displayed, and you will be prompted to approve (and save) it.
• When the approval file exists and contains matching data, the approval test will pass.
• When a new/updated approval is rejected, your tests will exit immediately with exit code 1.
• When running in a CI environment (CI variable exists), or on GitHub Actions (GITHUB_ACTIONS variable exists), your tests will run in non-interactive mode - tests will fail automatically if they do not match.

Download approvals.bash to your test folder. You can also download it by running this:

wget get.dannyb.co/approvals.bash

In your test script, add approve commands, using this syntax:

approve <command> [<approval file name>]

For example:

#!/usr/bin/env bash
source approvals.bash

approve "ls -s"
approve "ls -s" "ls_size"
# ... more tests

If your apptovals test files become too long or complex, you may the describe, context or it commands to annotate your tests. These commands are purely decorative and will output the provided strings as captions.

describe "ls"
  it "shows the list of files"
    approve "ls"

context "when in a non-empty directory"
  cd ./tmp
  describe "ls -s"
    approve "ls -s"
    approve "ls -s" "ls_size"
  
  cd ../

You can use any of the annotation commands, in any order. The indentation is optional.

You can test the exit code of the last approved command by using the expect_exit_code <code> command:

For example:

approve "some-command --that --fails"
expect_exit_code 1

In case the command under test outputs slightly different values each run, you can use the allow_diff <regex> command. This regex will be replaced with an asterisk (*) in the actual output in the following approve call (once only).

allow_diff "[0-9]\+"
approve "date"

You can use the fail <message> command to trigger custom failures:

approve "some-command-that-creates some-dir"
[[ -d some-dir ]] || fail "Expected directory some-dir to exist"

By default, all approvales are stored in the ./approvals directory (relative to the currently running test).

If you wish to store approvals in a different directory, set the APPROVALS_DIR environment variable before running your tests:

# test/test.sh
export APPROVALS_DIR=$PWD/my-approvals

approve "..."

Setting AUTO_APPROVE=1 prior to running your tests will automatically approve all tests, as if the user has pressed a to approve each changed or new test.

approvals.bash was tested in bash (version 4.0 or higher) and zsh, but might work in other shells as well.

• The approve file in this repository
• Approvals for fuzzy-cd, the fuzzy search cd patch
• Approvals for respec, the RSpec convenience wrapper
• Approvals for git-changelog, the changelog generator
• Approvals for rush, the personal shell-based package manager

• Approvals for opcode, the local alias manager
• Approvals for alf, the global bash alias manager

If you experience any issue, have a question or a suggestion, or if you wish to contribute, feel free to open an issue.