Fix markdown formatting

[gar1t]

Includes fix to typo in function name.

Ref: #447

[svlandeg]

Thanks for the PR!

Before I review the actual code edits, I want to take a step back and better understand what the desired output would be. In the linked issue #447, the author uses this example:

@app.command(name="tester-cmd", help="""
    Header

    Line 1
    Line 2
    Line 3
""")

which they want to get parsed (expected output) as

Header                                                                                                                                                                     
Line 1 
Line 2
Line 3   

However, with this PR, the output would actually be

 Header

 Line 1 Line 2 Line 3

(at least on my Windows system)

While currently on master it's

 Header
 Line 1 Line 2 Line 3

With mode rich, the output on master is

 Header
 Line 1
 Line 2
 Line 3

and this PR changes that to

 Header

 Line 1
 Line 2
 Line 3

Is that what we want?

Basically, I suggest that we first extend the test suite to cover various versions of the test (different number of newlines and different modes) and record which is the expected output for each. Then, we can look into how to get that fixed for all cases.

[gar1t]

The source:

Header

Line 1
Line 2
Line 3

I think ought to have a space after Header (as per the PR). That's my thought based on how Markdown is typically formatted in HTML.

E.g. while this is just one HTML formatter, it renders space between Header and the subsequent text block.

https://markdowntohtml.com/

I suspect it will be very hard to find any Markdown-to-HTML formatter that doesn't do this. I'd be curious to see it!

On this basis, I don't think the PR is controversial.

I would re-engage the OP for #447 and get his or her thoughts.

[svlandeg]

The source:

Header

Line 1
Line 2
Line 3

I think ought to have a space after Header (as per the PR).

Can you elaborate? With this PR, that input gives me

 Header

 Line 1 Line 2 Line 3

There's a space after Header, but the lines still aren't separated. Is that what you'd expect? I'm asking because the current unit test doesn't cover this case.

[gar1t]

I agree, whatever the agreed upon formatting behavior, the test case should include that source to put the matter to rest per the OP issue.

I should correct my comment above: not all markdown-to-html formatters treat text blocks the same. There are two ways to treat text block that I'm seeing:

1. A single line-ending to terminate the block
2. Two line-endings terminate the block

The GitHub formatter, e.g. uses method 1. You can try this by pasting the source and previewing it. The lines appear on separate lines.

This formatter uses method 2. If you paste the source in you'll see the text block is treated as one block. The lines don't appear separately.

The same is true for Python's markdown module:

> echo "Header 1
::: 
::: Line 1
::: Line 2
::: Line 2" | python -m markdown
<p>Header 1</p>
<p>Line 1
Line 2
Line 2</p>

I would still argue that typer should implement method 2. I think this is typical for tools and source code (where editors help enforce max line lengths with formatting), whereas the GitHub formatter is suited for human editing in simple text areas (like this one!)

I'm happy to add that test case.

[svlandeg]

I'm happy to add that test case.

That would be great, yes! Looking at some detailed unit tests will help us decide what the final behaviour is we want/need.