Both unordered and ordered list don't work when nested of the same type:

Two nested ordered lists

HTML:

<ol>
    <li>Top level 1</li>
    <li>Top level 2
        <ol>
            <li>A</li>
            <li>B</li>
            <li>C</li>
        </ol>
    </li>
    <li>Top level 3</li>
</ol>

Output:

1. Top level 1
 2. Top level 2
        
 1. A
 2. B
 3. C
 3. Top level 3

Two nested unordered lists

HTML:

<ul>
    <li>Top level 1</li>
    <li>Top level 2
        <ul>
            <li>A</li>
            <li>B</li>
            <li>C</li>
        </ul>
    </li>
    <li>Top level 3</li>
</ul>

Output:

- Top level 1
- Top level 2
        
- A
- B
- C
- Top level 3

Beep boop. Your images are optimized!

Your image file size has been reduced by 39% 🎉

📝 docs | :octocat: repo | 🙋🏾 issues | 🏪 marketplace

~Imgbot - Part of Optimole family

I was running in to an issue where list items using tags where getting indented when they shouldn't of been.

Example:

<ol>
    <li>A</li>
    <li>B</li>
    <li><b>C</b></li>
</ol>

Output:

1. A
2. B
3.     **C**

I added a test for this case as well. When doing the roundtrip style test, this indentation got lost, so I made the test compare the markdown output.

While Unmarkd currently supports tables, it spits out the html it was given. It would be nice if it supported tables:

| Syntax      | Description |
| ----------- | ----------- |
| Header      | Title       |
| Paragraph   | Text        |

When the following HTML block is parsed:

<ul>
    <li>Unordered lists, and:
        <ol>
            <li>One</li>
            <li>Two</li>
            <li>Three</li>
        </ol>
    </li>
    <li>More</li>
</ul>

The output is incorrect:

 * Unordered lists, and:
 0. One
 1. Two
 2. Three
 * More

Apply this patch:

diff --git a/tests/test_roundtrip.py b/tests/test_roundtrip.py
index a836024..5c1e097 100644
--- a/tests/test_roundtrip.py
+++ b/tests/test_roundtrip.py
@@ -1,10 +1,9 @@
 import unicodedata
 
 import markdown_it
-from hypothesis import assume, example, given
-from hypothesis import strategies as st
-
 import unmarkd
+from hypothesis import assume, example, given, reproduce_failure
+from hypothesis import strategies as st
 
 md = markdown_it.MarkdownIt()
 
@@ -17,6 +16,7 @@ def helper(text: str, func=unmarkd.unmark) -> None:
 
 
 @given(text=st.text(st.characters(blacklist_categories=("Cc", "Cf", "Cs", "Co", "Cn"))))
+@reproduce_failure("6.10.1", b"AAEADgEADgEADgA=")
 def test_roundtrip_commonmark_unmark(text):
     assume(unicodedata.normalize("NFKC", text) == text)
     helper(text)

Or add an example with text=">>>". Tests will fail

1. html2text is fast but not very configurable (there's only so any options)
2. Tomd sucks
3. Add an unmarker (with html2text-style configuration) to prove that unmarkd's configurability is at least equal to html2text

Instead of using commonmark, maybe https://github.com/executablebooks/markdown-it-py, https://github.com/trentm/python-markdown2, https://github.com/lepture/mistune, or https://github.com/Python-Markdown/markdown.

_{Also, I found tomd which might render this project useless 😬}

When encountering input like <em>Italic and <strong>bold and italic</strong></em>, the output is wrong, usually shadowed by the outer tag (in this case, <em>)

I've noticed that unmarkers.BaseUnmarker been documented as an "abstract base class" when we're actually using it otherwise.

Also, there's some dead code and we should actually sprinkle @staticmethod on some of them.

Here's my idea:

• Move all the tag_* methods in BaseUnmarker ➡️ BasicUnmarker
• Rename: BaseUnmarker ➡️ AbstractUnmarker
• Alias: BaseUnmarker ➡️ BasicMarker
• Run shed on the whole codebase (with --refactor)

Version bump: minor

1. Parse any css files or style tags found. Save it
2. When a class attribute is found, try to resolve it to the css
3. Add the resolved to the style attribute: convert to inline css