allow tables with headers in first (or any) column

markdownify/__init__.py

@@ -281,20 +281,19 @@ class MarkdownConverter(object):
         text_data = []
         rendered_header = False
         for row in rows:
-            headers = row.find_all('th')
-            columns = row.find_all('td')
-            if not rendered_header and len(headers) > 0:
-                headers = [head.text.strip() for head in headers]
-                text_data.append('| ' + ' | '.join(headers) + ' |')
-                text_data.append('| ' + ' | '.join(['---'] * len(headers)) + ' |')
+            cells = row.find_all(['td', 'th'])
+            is_headrow = all([cell.name == 'th' for cell in cells])
+            texts = [cell.text.strip() for cell in cells]
+            if not rendered_header and is_headrow:
+                text_data.append('| ' + ' | '.join(texts) + ' |')
+                text_data.append('| ' + ' | '.join(['---'] * len(cells)) + ' |')
                 rendered_header = True
-            elif len(columns) > 0:
+            elif len(cells) > 0:
                 if not rendered_header:
-                    text_data.append('| ' + ' | '.join([''] * len(columns)) + ' |')
-                    text_data.append('| ' + ' | '.join(['---'] * len(columns)) + ' |')
+                    text_data.append('| ' + ' | '.join([''] * len(cells)) + ' |')
+                    text_data.append('| ' + ' | '.join(['---'] * len(cells)) + ' |')
                     rendered_header = True
-                columns = [colm.text.strip() for colm in columns]
-                text_data.append('| ' + ' | '.join(columns) + ' |')
+                text_data.append('| ' + ' | '.join(texts) + ' |')
             else:
                 continue
         return '\n'.join(text_data)

tests/test_conversions.py

@@ -60,6 +60,25 @@ table = """<table>
 </table>"""
 
 
+table_with_header_column = """<table>
+    <tr>
+        <th>Firstname</th>
+        <th>Lastname</th>
+        <th>Age</th>
+    </tr>
+    <tr>
+        <th>Jill</th>
+        <td>Smith</td>
+        <td>50</td>
+    </tr>
+    <tr>
+        <th>Eve</th>
+        <td>Jackson</td>
+        <td>94</td>
+    </tr>
+</table>"""
+
+
 table_head_body = """<table>
     <thead>
         <tr>
@@ -335,6 +354,7 @@ def test_div():
 
 def test_table():
     assert md(table) == '| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |'
+    assert md(table_with_header_column) == '| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |'
     assert md(table_head_body) == '| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |'
     assert md(table_missing_text) == '|  | Lastname | Age |\n| --- | --- | --- |\n| Jill |  | 50 |\n| Eve | Jackson | 94 |'
     assert md(table_missing_head) == '|  |  |  |\n| --- | --- | --- |\n| Firstname | Lastname | Age |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |'