I have implemented the new test framework to test the Reverse Engineering SQL. I have integrated it as a part of API/Regression test suite. It will work when we run all the test cases or module wise test case.
How it works: Attached patch contains the generic framework to read all the JSON files from the tests->version based (example 9.6_plus, 10_plus or default) folder. Run all the test scenarios present in the JSON file in sequential order.
Format of the JSON file is mentioned in "<path_of_source>web/pgadmin/browser/server_groups/servers/databases/casts/tests/default/test.json"
For expected SQL we will have following two options:
Provide the expected sql in scenario itself as parameter "expected_sql" : "<SQL>".
Create a output file with any name in the same directory where the JSON file resides and specify the parameter "expected_sql_file": "<name of the file>"
Attached patch contains both the above mentioned examples.
Please review it.
Nice!
A few comments:
- The scenario name should be "Reverse Engineered SQL Test Cases"
- After the scenario name is output, can we output a \n so the next line isn't appended to the name?
Will fix the above.
- How do we run only the re_sql tests? I tried the obvious ways (e.g. python runtests.py --pkg regression.re_sql.tests.test_resql.ReverseEngineeringSQLTestCase) but got errors. Please add an example to web/regression/README.
It is not a pgadmin module and we have kept it in regression folder, so will have to change the existing code. I have tried but facing issues when run only "regression.re_sql.tests", will continue working on this.
- Once we have a way to run these tests only, please add a "make check-resql" target to the Makefile.
- Can the expected output be formatted in the JSON such that it doesn't use \n, but uses regular line breaks? That would make it easier to copy/paste.
I have tried that during implementation, but JSON does not allow line-breaks.
