osrm-backend/third_party/protozero/test/README.md

53 lines
1.7 KiB
Markdown
Raw Normal View History

# Tests
Tests are using the [Catch Unit Test Framework](https://github.com/philsquared/Catch).
## Organization of the unit tests
Unit tests test low-level functions of the library. They are in the `unit`
directory.
## Organization of the reader/writer test cases
The hart of the tests are the reader/writer tests checking all aspects of
decoding and encoding protobuf files.
Each test case is in its own directory under the `t` directory. Each directory
contains (some of) the following files:
* `reader_test_cases.cpp`: The C++ source code that runs the reader tests.
* `writer_test_cases.cpp`: The C++ source code that runs the writer tests.
* `data-*.pbf`: PBF data files used by the tests.
* `testcase.proto`: Protobuf file describing the format of the data files.
* `testcase.cpp`: C++ file for creating the data files.
### Reader tests
The CMake config finds all the `reader_test_cases.cpp` files and compiles them.
Together with the `reader_tests.cpp` file they make up the `reader_tests`
executable which can be called to execute all the reader tests.
### Extra writer tests
The CMake config finds all the `writer_test_cases.cpp` files and compiles them.
Together with the `writer_tests.cpp` file they make up the `writer_tests`
executable which can be called to execute all the writer tests.
The writer tests need the Google protobuf library to work.
## Creating test data from scratch
Most tests use test data stored in PBF format in their directory. The files
have the suffix `.pbf`. Most of those files have been generated from the
provided `testcase.proto` and `testcase.cpp` files.
Usually you do not have to do this, but if you want to re-generate the PBF
data files, you can do so:
cd test
./create_pbf_test_data.sh