At least Clang 7.0 and Clant-tidy 7.0 is required to run the tests.
Before running the tests the CodeChecker package needs to be built!
# Build the package make package
Every test target has an
*in_env version (Eg.
make test_in_env) which
automatically creates and sources a virtualenv for the tests.
||run all tests (unit and functional for the analyzer and the web)|
||run all tests (unit and functional), automatically setup and source a virtualenv|
||unittests for the analyzer and the web|
||functional tests for the analyzer and the web (SQLite and PostgreSQL)|
||functional test for the web part (SQLite)|
||run only a specific web test (SQLite)|
||run only a specific analyzer test|
Change test workspace root
CC_TEST_WORKSPACE environment variable the root directory for the
tests can be changed. With this environment variable multiple test types
(sqlite, postgresql) can be run parallel. Every test should create the
temporary directories under the given root directory.
Sqlite tests with a changed workspace root can be run like this:
CC_TEST_WORKSPACE=/tmp/sqlite_test_workspace make -C web test_sqlite
Clean test workspace
Run tests with PostgreSQL
At least one of the database drivers needs to be available to use the
PostgreSQL database backend.
psycopg2 is used by default if not found
pg8000 is used.
.noserc configuration file in the repository root is used to configure running the tests:
Further configuration options can be found here
Virtual environment to run tests
Test running virtual environment is automatically created and sourced by the
make test* commands.
Create a python virtualenv for development:
Create new unit test
Use the add_new_unit_test.py script to generate new test files.
Add new tests for the created
tests/unit/test_module_name.py file and run
the unit tests:
Create new functional test
Use the add_new_func_test.py script to generate new test files.
From the repository root run the new test template (should pass):
make TEST="tests/functional/mynewfeature" run_test
Create c/cpp project and generate test reports
If it is possible generate and add the report files used during the test to the repository. It will speed up the tests because no analysis is needed, during each test and the results will not depend on the analyzer version in the test environment.
Make sure to generate plist reports where the source file names do not contain the source file path only the source file name because the reports will be moved during the tests to temporary directories.
A simple test c/cpp should look like this:
Makefile # build and analyze targets for the test source files main.cpp # cpp file with errors in it reports/base/*.plist # first report set used for the tests reports/new/*.plist # second report set used for the tests
In each functional test
mynewfeature/__init__.py is doing the functional test
- exporting test related data to the test directories
- analyzing a test project (multiple times if needed)
- start a server connected to a specific database (should be stopped in teardown)
- prepare and cleanup the test workspace
Add new functional testcase
The actual test cases go to
The setup part of these test read up the generated test configuration file by
Test cases should only use test related configuration values
ONLY from the configuration file
generated by the
The tests should
ONLY modify the files in the workspace provided for them.
Test cases can:
- rerun the analysis
- connect to the server started by the
- run command line commands
- modify configuration files