This directory holds automated test scripts for git-publish.

Running the test cases

Just run:

$ ./testing/run_tests.sh

Optionally, the path to the git-publish binary can be provided as argument:

$ ./testing/run_tests.sh /path/to/git-publish

Adding new test cases

To add a new test case, just create a new executable file in this directory whose name starts with a digit.

Useful environment variables

• $GIT_PUBLISH: path to git-publish script being tested
• $TEST_DIR: path to temporary test log directory. Test cases are allowed (and encouraged) to create debug and log files inside this directory.
• $FAKE_GIT_COMMAND_LOG: command log generated by fake_git

Fake git command

When running test cases, a fake git binary will be in $PATH, which will only accept a subset of non-destructive git commands. If your test case requires additional git commands to work, see the PASSTHROUGH_COMMANDS and SPECIAL_COMMANDS variables in the fake_git script.

Writing new test cases using Bash

Test cases using bash can include functions.sh for useful utility functions.

$PATH will contain the git-publish script being tested, so you can just run git-publish directly. e.g.:

Example:

source "$TESTS_DIR/functions.sh"
git-publish --some-arguments
grep -q 'do-something' "$FAKE_GIT_COMMAND_LOG" || \
   abort "git-publish didn't run `git do-something`"

Writing new test cases using Python

Python test cases can import utility functions from the test_utils module. Example:

from test_utils import *
git_publish('--some-arguments')
assert last_git_command() == ['do-something', 'xxx', 'yyy'], \
    "git-publish didn't run `git do-something` as expected"