Raise error if malformed test metadata is given

[idorax]

Fix: #1881

[idorax]

With the patch, the test result looks like:

$ cat main.fmf 
summary: Concise summary describing what the test does
test: false 
framework: beakerlib

$ tmt run -avvv provision -h local
/var/tmp/tmt/run-001
Found 1 plan.

/plans/foo
summary: Basic smoke test
    discover
        how: fmf
        order: 50
        directory: /home/huanli/dev/1881/foo
                warn: /tests/foo:test - False is not of type 'string'
    finish
        prune: Prune plan workdir '/var/tmp/tmt/run-001/plans/foo'.
        summary: 0 tasks completed
Oops, False is not of type 'string'.

Hi @lukaszachy and @happz, please help to review, thanks!

[idorax]

Hi @happz, with your PR #1869,

• https://github.com/teemtee/tmt/blob/main/tmt/base.py#L912

   912	    test: Optional[ShellScript] = field(
   913	        default=None,
   914	        normalize=lambda key_address, raw_value, logger:
   915	            None if raw_value is None else ShellScript(raw_value))

I don't know how to check raw_value is a str in a lambda function, so I added the check to ShellScript.__init__(). Please help to review and correct me if it's incorrect, thanks!

[happz]

Hi @happz, with your PR #1869,

• https://github.com/teemtee/tmt/blob/main/tmt/base.py#L912

   912	    test: Optional[ShellScript] = field(
   913	        default=None,
   914	        normalize=lambda key_address, raw_value, logger:
   915	            None if raw_value is None else ShellScript(raw_value))

I don't know how to check raw_value is a str in a lambda function, so I added the check to ShellScript.__init__(). Please help to review and correct me if it's incorrect, thanks!

lambdas are a bit too clumsy in Python, once there are more than just a few actions to take. How about adding tmt.utils.normalize_shell_script()? There's already normalize_shell_script_list(), we can have one for a single script test allow, and make it nice and readable.

[idorax]

lambdas are a bit too clumsy in Python, once there are more than just a few actions to take. How about adding tmt.utils.normalize_shell_script()? There's already normalize_shell_script_list(), we can have one for a single script test allow, and make it nice and readable.

Hi @happz, function tmt.utils.normalize_shell_script() is added. Now the test result looks like:

$ cat main.fmf 
summary: Concise summary describing what the test does
test: false 
framework: beakerlib
$ tmt run -avvv provision -h local
/var/tmp/tmt/run-056
Found 1 plan.

/plans/foo
summary: Basic smoke test
    discover
        how: fmf
        order: 50
        directory: /home/huanli/dev/1881/foo
                warn: /tests/foo:test - False is not of type 'string'
    finish
    
    Prune '/plans/foo' plan workdir '/var/tmp/tmt/run-056/plans/foo'.
        summary: 0 tasks completed

Field '/tests/foo:test' must be a string, 'bool' found.

Please help to review again, thanks!

[idorax]

Hi @happz and @lukaszachy , a simple test is added, please help to review, thanks!

[psss]

Thanks for fixing this!

[psss]

The failing test is irrelevant, known issue #2063.