Initial validation

[edublancas]

We should validate the input notebook:

• [ ] if there are no markdown cells, show an error
• [ ] if there are only H1 headings

[idomic]

On the second bullet. @edublancas who defines what's a few? Also what's invalid in it that we need an error? Imo the user and maybe it's a valid notebook for them? I assume you meant H2s since we don't support H1s currently. Unless it's a 300 lines notebook and there's no or maybe 1-2 headers and then it make sense to output something to the user, but then it's pretty similar to https://github.com/ploomber/soorgeon/issues/47

[edublancas]

I'd say:

• Only H1 headings (regardless of the number, since we don't support them): throw an error
• Only one H2 heading, this is the same as #47

I updated the original description

[94rain]

I am working on implementing the following logic. Just to confirm if my understanding is correct.

• No H2 headers:
  • Only H1 headers: Only H1 headings are found. At this time, only H2 heading are supported.
  • No H1 and H2 headers: Expected notebook to have at least one markdown H2 heading.
• Only one H2 header: Warning: refactoring successful but only one H2 heading detected...
• Otherwise: run smoothly

I will create _get_h1_header method, similar to _get_h2_header. But I think I will need to change the regex in _get_h2_header to ^\s*##\s+(.+) (add ^). Otherwise \s*##\s+(.+) will match ###.

And the regex in the new method _get_h1_header will be ^\s*#\s+(.+).

I will start to write tests for test_get_h2_header and test_get_h1_header.

[idomic]

Sounds about right, Only one H2 header will be dealt in a different PR. The function sounds good, just validate the regex actually matches to H1s (I tried here on python and seems it doesn't https://regex101.com/).

[94rain]

What's the input that fails the match? I tried the following cases and they all passed

@pytest.mark.parametrize('md, expected', [
    ['# Header', 'header'],
    ['# H1\n## H2', 'h1'],
    [' \t #   H1', 'h1'],
    ['  ##   H2', None],
    ['something', None],
    ['# Something\nignore me', 'something'],
])
def test_get_h1_header(md, expected):
    assert split._get_h1_header(md) == expected

[idomic]

^\s*##\s+(.+)

Is it working with this pattern? Also I tried it online and not locally with a py console, that might be failing due to the parser.

[94rain]

Yes, it is.

What's the input? I tried ## 1 as the input online and it worked.

[idomic]

Small comment @94rain did we cover the case where no markdowns are found at all? Didn't see any in the test cases.

[94rain]

Does test_find_breaks_error_if_no_h1_and_h2_headers count? Expected notebook to have at least one markdown H2 heading. (No H1 and H2 headers)

This does not check whether or not there are other markdown texts.

[idomic]

Not really since there's still markdown in the string you're checking, try having a sample with a json or plain-text

[94rain]

Not really since there's still markdown in the string you're checking, try having a sample with a json or plain-text

OK. Just additional tests? Or do we expect the error message to be different to Expected notebook to have at least one markdown H2 heading.?

[idomic]

Yes exactly!