This unit test for the `nvs::Page` C++ class uses CMock to abstract its dependency to the `spi_flash` component. Its `CMakeLists.txt` builds mocks instead of the actual `spi_flash` component by setting the component property `USE_MOCK` for `spi_flash`. It also needs some stubs for CRC32 since the ROM component can not be mocked yet. These are retrieved from the `mock` directory inside `nvs_flash`.
The mocked version of `spi_flash` receives all function calls to the `esp_partition` API. E.g., have a look at the first unit test in the [NVS page test](main/nvs_page_test.cpp), the test function `test_Page_load_reading_header_fails()`. It simply checks whether an error is propagated correctly from the `esp_partition` API to upper layers during a call to `nvs::Page::load`. The relevant mocking code here is:
It tells the generated mocks to expect a call to `esp_partition` API's `esp_partition_read_raw()`, accept any arguments and return `ESP_ERR_INVALID_ARG`. When `nvs::Page::load` is called later, the mock will "replay" this action and the test will check that the return value is correctly propagated with:
For more information on how to control the mock's behavior, take a look at the [CMock README](../../../cmock/CMock/docs/CMock_Summary.md). The rest of `main/nvs_page_test.cpp` also provides many more examples on how to control the mocks.
## Requirements
* The usual IDF requirements
* The host's gcc/g++
*`ruby`
This application has been tested on Ubuntu 20.04 with `gcc` version *9.3.0*.
## Build
First, make sure that the target is set to Linux. Run `idf.py --preview set-target linux` if you are not sure. Then do a normal IDF build: `idf.py build`.
## Run
IDF monitor doesn't work yet for Linux. You have to run the app manually:
```bash
./build/host_nvs_page_test.elf
```
## Coverage
To generate the coverage, run: `idf.py coverage`. Afterwards, you can view the coverage by opening `build/coverage_report/index.html` with your browser. Note that you need to run the application at least once before generating the coverage information. If you run it multiple times, the coverage information adds up.
## Example Output
Ideally, all tests pass, which is indicated by the last two log lines after the dashed line:
