Merge pull request #730 from mikamikasuki/fix/table-multibacktick-code-spans

Fix table boundaries around multi-backtick code spans

Author: nicholasserra

CHANGES.md

@@ -2,6 +2,7 @@
 
 ## python-markdown2 2.5.6 (not yet released)
 
+- [pull #730] Fix `tables` extra splitting multi-backtick code spans at a leading pipe and merging cells after a code span containing literal backticks.
 - [pull #729] Fix `tables` extra merging cells when a pipe directly follows a code span, as in compact rows like `|`-v`|verbose|`.
 - [pull #725] Fix `tables` extra dropping escaped pipes at the end of header and body rows.
 - [pull #687] Fix AssertionError hashing HTML blocks spread over multiple lines (#686)

README.md

@@ -103,6 +103,10 @@ as a script:
 The `footnotes` extra keeps the same number when a footnote is referenced more
 than once.
 
+The `tables` extra supports code spans delimited by single or multiple backticks,
+including an operator cell such as ```` ``|`` ````. A pipe immediately after the
+closing delimiter separates the next cell.
+
 There are a number of currently implemented extras for tables, footnotes,
 syntax coloring of `<pre>`-blocks, auto-linking patterns, table of contents,
 Smarty Pants (for fancy quotes, dashes, etc.) and more. See the [Extras

lib/markdown2.py

@@ -4260,12 +4260,19 @@ def _split_row(row: str) -> list[str]:
         # An escaped final pipe belongs to the last cell, not the table border.
         if not row.endswith(r'\|'):
             row = row.removesuffix('|')
+        code_spans = list(Markdown._code_span_re.finditer(row))
+        code_starts = {m.start(2) for m in code_spans}
+        code_ends = {m.end() for m in code_spans}
         cells, start = [], 0
         for m in re.finditer(r'(?<!\\)\|', row):
-            # A pipe right after the backtick that opens a code span (`|`) stays in
-            # the cell; one right after a closing backtick is a cell border.
-            if m.start() and row[m.start() - 1] == '`' and row.count('`', 0, m.start()) % 2:
-                continue
+            pos = m.start()
+            # A pipe after an opening backtick delimiter stays in the cell.
+            # A pipe after the matching closing delimiter is a cell border.
+            if pos and row[pos - 1] == '`':
+                if pos in code_starts or (
+                    pos not in code_ends and row.count('`', 0, pos) % 2
+                ):
+                    continue
             cells.append(row[start:m.start()])
             start = m.end()
         cells.append(row[start:])

test/tm-cases/tables_multibacktick_code_spans.html

@@ -0,0 +1,34 @@
+<table>
+<thead>
+<tr>
+  <th>Operator</th>
+  <th>Meaning</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+  <td><code>|</code></td>
+  <td>bitwise or</td>
+</tr>
+<tr>
+  <td><code>|</code></td>
+  <td>bitwise or</td>
+</tr>
+<tr>
+  <td><code>|</code></td>
+  <td>bitwise or</td>
+</tr>
+<tr>
+  <td><code>x`y</code></td>
+  <td>literal backtick</td>
+</tr>
+<tr>
+  <td><code>x``y</code></td>
+  <td>two literal backticks</td>
+</tr>
+<tr>
+  <td><code>|</code></td>
+  <td><code>or</code></td>
+</tr>
+</tbody>
+</table>

test/tm-cases/tables_multibacktick_code_spans.opts

@@ -0,0 +1 @@
+{"extras": ["tables"]}

test/tm-cases/tables_multibacktick_code_spans.text

@@ -0,0 +1,8 @@
+| Operator | Meaning |
+| --- | --- |
+| `|` | bitwise or |
+| ``|`` | bitwise or |
+| ````|```` | bitwise or |
+|``x`y``|literal backtick|
+|````x``y````|two literal backticks|
+|``|``|`or`|