1. Get yourself a sample file¶
Its a very good idea to start developing by getting a sample source file for two reasons:
- It will get picked up by the The Testbed so you will be able to get instant feedback as you work on your handler.
- You will get a lot of tests for free. These tests will parse the sample file into a template and stringset, compile them back in your source file and check whether the template matches the expected one and that the resulted file matches the source. It will also try to translate the strings based on some common ones found in a dictionary and check that it can compile a language file that matches the expected one.
Put your sample file in
example, our sample SRT file goes to
2. Generate expected template and language files¶
In order to generate the expected template and language files mentioned above, you can use the bin/create_files.py script once you have a working handler:
In order to get the tests we mentioned for free, make sure your test class inherits from the:
Define a set of tests to be run by every file format.
The class that inherits from this must define the following:
HANDLER_CLASS, eg: PlaintextHandler
TESTFILE_BASE, eg: openformats/tests/formats/plaintext/files
You might have noticed that by using a working handler to make the expected sample files and then testing against them seems pointless. Well, you’re right, they are, initially. The point of for them to serve as regression tests, as you later make changes to your handler.
3. Add your own tests¶
Testing that a handler works correctly against a valid source file is good, but you will want to also test more things, like:
- The hashes produced take the correct information into account
- The metadata of the extracted strings is what you want
- ParseErrors are raised when appropriate and they produce a helpful error message
- Sections of the compiled files are removed when the relevant strings are missing from the stringset given as input
- Anything to get your coverage higher
This is to help you write multilingual strings as test inputs in your tests without screwing up your code’s syntax. Eg:
''' 1 00:01:28.797 --> 00:01:30.297 X:240 Y:480 Hello world '''
will be converted to:
'\n1\n00:01:28.797 --> 00:01:30.297 X:240 Y:480\nHello world\n'
_test_parse_error(source, error_msg, parse_kwargs=None)¶
Test that trying to parse ‘source’ raises an error with a message exactly like ‘error_msg’
5. Run the test suite¶
python setup.py test