Cyclic0007/python-markdownify
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Generalize handling of colspan in case where colspan is in first row but header row is missing (#203)

Browse Source
This commit is contained in:
Stephen V. Brown
2025-03-04 20:01:16 -05:00
committed by GitHub
parent 618747c18c
commit 7908f1492a
2 changed files with 27 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
markdownify/__init__.py
View File

@@ -701,6 +701,12 @@ class MarkdownConverter(object):
)
overline = ''
underline = ''
full_colspan = 0
for cell in cells:
if 'colspan' in cell.attrs and cell['colspan'].isdigit():
full_colspan += int(cell["colspan"])
else:
full_colspan += 1
if ((is_headrow
or (is_head_row_missing
and self.options['table_infer_header']))
@@ -709,12 +715,6 @@ class MarkdownConverter(object):
# - is headline or
# - headline is missing and header inference is enabled
# print headline underline
full_colspan = 0
for cell in cells:
if 'colspan' in cell.attrs and cell['colspan'].isdigit():
full_colspan += int(cell["colspan"])
else:
full_colspan += 1
underline += '| ' + ' | '.join(['---'] * full_colspan) + ' |' + '\n'
elif ((is_head_row_missing
and not self.options['table_infer_header'])
@@ -727,8 +727,8 @@ class MarkdownConverter(object):
# - the parent is table or
# - the parent is tbody at the beginning of a table.
# print empty headline above this row
overline += '| ' + ' | '.join([''] * len(cells)) + ' |' + '\n'
overline += '| ' + ' | '.join(['---'] * len(cells)) + ' |' + '\n'
overline += '| ' + ' | '.join([''] * full_colspan) + ' |' + '\n'
overline += '| ' + ' | '.join(['---'] * full_colspan) + ' |' + '\n'
return overline + '|' + text + '\n' + underline

19
tests/test_tables.py
View File

@@ -267,6 +267,23 @@ table_with_undefined_colspan = """<table>
</tr>
</table>"""
table_with_colspan_missing_head = """<table>
<tr>
<td colspan="2">Name</td>
<td>Age</td>
</tr>
<tr>
<td>Jill</td>
<td>Smith</td>
<td>50</td>
</tr>
<tr>
<td>Eve</td>
<td>Jackson</td>
<td>94</td>
</tr>
</table>"""
def test_table():
assert md(table) == '\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
@@ -283,6 +300,7 @@ def test_table():
assert md(table_with_caption) == 'TEXT\n\nCaption\n\n| | | |\n| --- | --- | --- |\n| Firstname | Lastname | Age |\n\n'
assert md(table_with_colspan) == '\n\n| Name | | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
assert md(table_with_undefined_colspan) == '\n\n| Name | Age |\n| --- | --- |\n| Jill | Smith |\n\n'
assert md(table_with_colspan_missing_head) == '\n\n| | | |\n| --- | --- | --- |\n| Name | | Age |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
def test_table_infer_header():
@@ -300,3 +318,4 @@ def test_table_infer_header():
assert md(table_with_caption, table_infer_header=True) == 'TEXT\n\nCaption\n\n| Firstname | Lastname | Age |\n| --- | --- | --- |\n\n'
assert md(table_with_colspan, table_infer_header=True) == '\n\n| Name | | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'
assert md(table_with_undefined_colspan, table_infer_header=True) == '\n\n| Name | Age |\n| --- | --- |\n| Jill | Smith |\n\n'
assert md(table_with_colspan_missing_head, table_infer_header=True) == '\n\n| Name | | Age |\n| --- | --- | --- |\n| Jill | Smith | 50 |\n| Eve | Jackson | 94 |\n\n'