Babelfish has introduced a new extension named babelfishpg_unit which enables us to run unit tests. Please follow the build instructions to build and install the babelfishpg_unit extension.

To maintain a well-organized structure, the directory layout will be as follows:

• babelfish_extensions
  • contrib
    • babelfishpg_common
    • babelfishpg_money
    • babelfishpg_tsql
    • babelfishpg_tds
    • babelfishpg_unit
      • Makefile
      • babelfishpg_unit.control
      • babelfishpg_unit--1.0.0.sql
      • babelfishpg_unit.c
      • babelfishpg_unit.h
      • test_1.c
      • test_2.c

• To add a new test to the existing framework, within the framework's directory, create a new .c file and name it according to the test we want to add (e.g., test_1.c). Add the function in the file which is to be tested.

  • Declare the function you want to test as extern in babelfishpg_unit.h. Eg:

    extern TestResult* test_int4_fixeddecimal_ge(void)
    

  • Metadata for each test function is added as a row in the tests array, containing the necessary information for the test. This tests array is located in babelfishpg_unit.c. Eg:

    TestInfo tests[]=
    {
        {&test_int4_fixeddecimal_ge, true, "GreaterThanOrEqualToCheck_INT4_FIXEDDECIMAL", "babelfish_money_datatype"},
    };
    

  • Return type of all functions which are to be tested must be TestInfo*.

  • Every function should have some expected and obtained output. One should use TEST_ASSERT_TESTCASE(expected == obtained, testResult) first and then TEST_ASSERT(expected == obtained, testResult) to obtain the status of a test. In the TEST_ASSERT_TESTCASE marco, we only set the result field of TestResult struct to true/false based on result. In TEST_ASSERT, we decide whether all the tests have passed or not. Eg:

    TestResult* test_int4_fixeddecimal_ge(void)
        {
            /*
            * This function checks whether val1 is greater than or equal to val2 or not.
            */  
    
            int val1[] = {1522, -100, 5, 0, -856, 0};
            int val2[] = {982, 200, 0, -24, -567, 0};
    
            int numValues = sizeof(val1) / sizeof(val1[0]);
    
            TestResult* testResult = palloc0(sizeof(TestResult));
            testResult->result = true;
    
            for (int i = 0; i < numValues; i++) 
            {
                bool expected = (val1[i] >= val2[i]);
                Datum temp = DirectFunctionCall2(int4_fixeddecimal_ge, Int32GetDatum(val1[i]), Int32GetDatum(val2[i]));
                bool obtained = DatumGetBool(temp);  
    
                TEST_ASSERT_TESTCASE(expected == obtained, testResult);
            }
    
            TEST_ASSERT(expected == obtained, testResult);
    
            return testResult;
        }
    

  • If you want to put any custom message for any test in the output table, use :

    • snprintf(testResult->message, MAX_TEST_MESSAGE_LENGTH, "%s", errorData->message);
    • This will fill the message column with expected message. When the tests are run, if the condition is met, then you will observe the same message corresponding to that text.

  • If there is any ASSERTION error and TEST_ASSERT_TESTCASE and TEST_ASSERT macros are used to check the condition, then message column will be filled with the error message providing the line where the assertion failed.