# Main Headline

Normal text that is wrapped should become
one long line in the output.

### <i>Skipped</i> header level

H3s should be preceded by at least one H2.  
H2 should be preceded by at least one H1.

Should die on error unless you have a flag saying that it's OK.

## Inline **markup**

We need \*escape characters\* to work \`well\`.

**This** _text_ *should* become ***really* good** `plain` <cite><span>text</span></cite>.
Change <del>strike-through</del> to ^W style syntax.
Delete <del>these three words</del> and call it a day.

### Unordered Lists

* How to make an unordered list
    * And what to do about nesting
    * Turn it into a sequence of flat lists?
        - Or some other solution?
        - Not sure
* Just flatten them for now but print a warning.

### Ordered Lists

1. Hardwrapped lists is apparently a thing in lowdown
   that we need to deal with.
2. Just turn them into normal hardwrapped lines is the plan.

#### Header Level 4

Should just become: "Header Level 4." with a period. Print a warning on H4, H5, H6 on but don't die.

##### Header Level 5.

Should just become: "Header Level 4." with one period, not two.

###### Ordered lists

1. Ordered lists should renumber their numbers
    1. And hardwrap
    1. Same nesting problem
1. as unordered lists have.
2. Flatten, and then renumber, and print a warning.

<h3>Blockquote joining</h3>

> These lines
> should *be* all on
> one line
>
> This should be  
> two separate lines.

### Blockquote Nesting

>> This is an example of a nested quote.
>
> It should go through. Good luck Gemini clients.

This isn't to extend the client necessarily.
Maybe I should die with error and make this an option instead.

>>> Here is an even deeper
>>> nested quote.
>>
>> I just pass them through.
>
> List flattened, quotes not.

### Preformatted text


	We need a <a>untouched</a> *pre-formatted* mode
	that respects line breaks

	and empty lines.

We want to replace horizontal rules with one extra newline.

-----

So there should be two newlines between these paragraphs. Other than
that, we want to collapse blank lines down to one, outside of
preformat toggled lines.

Images and Links
----------------

### Images

![Images are treated like any other links](https://idiomdrottning.org/idiomdrottning_dark_thistle.png "Hover titles are ditched")

### Links


Some text with [a link](gemini://url.to/url) in the middle of it.

Maybe some [longer
text][lt] with several [links
inside] is a good
thing to test. And [a non-link bracket].

[lt]: gemini://so.boring/and/long "My Favorite Longer Text"
[links inside]: gemini://some.url

The correct placement for all accumulated links that have been
"extracted" in such a way, is which ever comes first out of:

- Just before the next non-"extracted" link (i.e. a link from a list or stand-alone `<p>`, see below)
- Just before the next header
- Just before the end of the document

Some examples of non-extracted links:

- another plain list entry
- some [link](gemini://some.url) in a list
- [some other link in a list](gemini://some.url)
- a non-linked list entry

for lists. Both `<p><a></p>` and `<li><a></li>` should be
turned into link lines without blank lines in between.

A [very short p with a link](gemini://some.url)

[An all-link p](gemini://some.url)

---

A p that's [shorter](gemini://some.url) than sixty characters counts as short

A p that's [longer](gemini://some.url) than sixty characters counts as long and needs to be extracted.

Unless [the link is more than half longer of that p regardless of length then it counts](gemini://some.url) as non-extracted.

----

[My friend](gemini://some.url) wrote:

> The perfect quote intro.
> Classic idomatic gemini.

## Tables

<table>
  <thead>
    <tr>
      <th class="jam">col a</th>
      <th>col b</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td>a</td>
      <td>html</td>
    </tr>
    <tr>
      <td>table</td>
      <td>also</td>
    </tr>
    <tr>
      <td>without</td>
      <td>any colspan</td>
    </tr>
  </tbody>
</table>

## Extensions

It doesn't need to handle LaTeX math. On the longer roadmap, make a
separate preprocessor to turn LaTeX math into AsciiMath.

I do want it to support liquid style includes like {% include_absolute 'foo' %}