fix: parse empty list item with trailing space (#3984)

A bare list marker followed only by a space (e.g. `- ` or `1. `) was not
recognized as a list and fell through to a paragraph, so `- ` rendered as
`<p>- </p>` instead of an empty list item. A marker with no trailing space
(`-`, `1.`) and an empty item that continues an existing list were already
handled correctly, making the behavior inconsistent.

The block `list` rule required content after the marker's whitespace
(`[ \t][^\n]+?`), so a marker followed only by whitespace failed to match.
Relax it to `[ \t][^\n]*?` so a marker with trailing whitespace and no
content still starts a list, matching the CommonMark reference and
markdown-it.

To keep CommonMark's rule that an empty list item cannot interrupt a
paragraph, the paragraph list-interrupt patterns now require a non-blank
character after the marker, so `foo\n+ ` stays a single paragraph while
`foo\n- bar` still interrupts.

Author: sarathfrancis90

src/rules.ts

@@ -136,7 +136,7 @@ const def = edit(/^ {0,3}\[(label)\]: *(?:\n[ \t]*)?([^<\s][^\s]*|<.*?>)(?:(?: +
   .replace('title', /(?:"(?:\\"?|[^"\\])*"|'[^'\n]*(?:\n[^'\n]+)*\n?'|\([^()]*\))/)
   .getRegex();
 
-const list = edit(/^(bull)([ \t][^\n]+?)?(?:\n|$)/)
+const list = edit(/^(bull)([ \t][^\n]*?)?(?:\n|$)/)
   .replace(/bull/g, bullet)
   .getRegex();
 
@@ -170,7 +170,7 @@ const paragraph = edit(_paragraph)
   .replace('|table', '')
   .replace('blockquote', ' {0,3}>')
   .replace('fences', ' {0,3}(?:`{3,}(?=[^`\\n]*\\n)|~{3,})[^\\n]*\\n')
-  .replace('list', ' {0,3}(?:[*+-]|1[.)])[ \\t]') // only lists starting from 1 can interrupt
+  .replace('list', ' {0,3}(?:[*+-]|1[.)])[ \\t]+[^ \\t\\n]') // only non-empty lists starting from 1 can interrupt
   .replace('html', '</?(?:tag)(?: +|\\n|/?>)|<(?:script|pre|style|textarea|!--)')
   .replace('tag', _tag) // pars can be interrupted by type (6) html blocks
   .getRegex();
@@ -214,7 +214,7 @@ const gfmTable = edit(
   .replace('blockquote', ' {0,3}>')
   .replace('code', '(?: {4}| {0,3}\t)[^\\n]')
   .replace('fences', ' {0,3}(?:`{3,}(?=[^`\\n]*\\n)|~{3,})[^\\n]*\\n')
-  .replace('list', ' {0,3}(?:[*+-]|1[.)])[ \\t]') // only lists starting from 1 can interrupt
+  .replace('list', ' {0,3}(?:[*+-]|1[.)])[ \\t]') // any bullet ends the table rows
   .replace('html', '</?(?:tag)(?: +|\\n|/?>)|<(?:script|pre|style|textarea|!--)')
   .replace('tag', _tag) // tables can be interrupted by type (6) html blocks
   .getRegex();
@@ -230,7 +230,7 @@ const blockGfm: Record<BlockKeys, RegExp> = {
     .replace('table', gfmTable) // interrupt paragraphs with table
     .replace('blockquote', ' {0,3}>')
     .replace('fences', ' {0,3}(?:`{3,}(?=[^`\\n]*\\n)|~{3,})[^\\n]*\\n')
-    .replace('list', ' {0,3}(?:[*+-]|1[.)])[ \\t]') // only lists starting from 1 can interrupt
+    .replace('list', ' {0,3}(?:[*+-]|1[.)])[ \\t]+[^ \\t\\n]') // only non-empty lists starting from 1 can interrupt
     .replace('html', '</?(?:tag)(?: +|\\n|/?>)|<(?:script|pre|style|textarea|!--)')
     .replace('tag', _tag) // pars can be interrupted by type (6) html blocks
     .getRegex(),

test/specs/new/list_item_empty_trailing_space.html

@@ -0,0 +1,16 @@
+<ul>
+<li></li>
+</ul>
+<ul>
+<li></li>
+</ul>
+<ol>
+<li></li>
+</ol>
+<ol>
+<li></li>
+</ol>
+<p>foo
++</p>
+<p>bar
+1.</p>

test/specs/new/list_item_empty_trailing_space.md

@@ -0,0 +1,13 @@
+- 
+
+* 
+
+1. 
+
+1) 
+
+foo
++ 
+
+bar
+1.