def test_fix_tables():
cases = [
(
'add a table row',
'''
<table>
<tr><td>A</td></tr>
<ins><tr><td>B</td></tr></ins>
</table>
''',
'''
<table>
<tr><td>A</td></tr>
<tr><td><ins>B</ins></td></tr>
</table>
'''
),
(
'remove ins and del tags at the wrong level of the table',
'''
<table>
<ins> </ins><del> </del>
<thead>
<ins> </ins><del> </del>
</thead>
<tfoot>
<ins> </ins><del> </del>
</tfoot>
<tbody>
<ins> </ins><del> </del>
<tr>
<ins> </ins><del> </del>
<td><ins>A</ins></td>
</tr>
</tbody>
</table>
''',
'''
<table>
<thead></thead>
<tfoot></tfoot>
<tbody>
<tr>
<td><ins>A</ins></td>
</tr>
</tbody>
</table>
''',
),
]
for test_name, changes, fixed_changes in cases:
changes = collapse(changes)
fixed_changes = collapse(fixed_changes)
def test():
changes_dom = parse_minidom(changes, strict_xml=True)
fix_tables(changes_dom)
assert_html_equal(minidom_tostring(changes_dom), fixed_changes)
test.description = 'test_fix_tables - %s' % test_name
yield test
def test_fix_tables():
cases = [
(
'add a table row',
'''
<table>
<tr><td>A</td></tr>
<ins><tr><td>B</td></tr></ins>
</table>
''',
'''
<table>
<tr><td>A</td></tr>
<tr><td><ins>B</ins></td></tr>
</table>
'''
),
(
'remove ins and del tags at the wrong level of the table',
'''
<table>
<ins> </ins><del> </del>
<thead>
<ins> </ins><del> </del>
</thead>
<tfoot>
<ins> </ins><del> </del>
</tfoot>
<tbody>
<ins> </ins><del> </del>
<tr>
<ins> </ins><del> </del>
<td><ins>A</ins></td>
</tr>
</tbody>
</table>
''',
'''
<table>
<thead></thead>
<tfoot></tfoot>
<tbody>
<tr>
<td><ins>A</ins></td>
</tr>
</tbody>
</table>
''',
),
]
for test_name, changes, fixed_changes in cases:
changes = collapse(changes)
fixed_changes = collapse(fixed_changes)
def test():
changes_dom = parse_minidom(changes, strict_xml=True)
fix_tables(changes_dom)
assert_html_equal(minidom_tostring(changes_dom), fixed_changes)
test.description = 'test_fix_tables - %s' % test_name
yield test
test_cases = [(
'nested list items',
collapse('''
<ul>
<li>Monday
<ul>
<li>2pm - 3pm</li>
</ul>
</li>
<li>Wednesday
<ul>
<li>11am - Noon</li>
<li>3pm - 5pm</li>
</ul>
</li>
<li>Thursday
<ul>
<li>11am - Noon</li>
</ul>
</li>
<li>Friday
<ul>
<li>Noon - 1pm</li>
</ul>
</li>
</ul>
'''),
collapse('''
<ul>
<li>Tuesday
def test_fix_lists():
cases = [
(
"simple list item insert",
"""
<ol>
<li>one</li>
<ins><li>two</li></ins>
</ol>
""",
"""
<ol>
<li>one</li>
<li><ins>two</ins></li>
</ol>
""",
),
(
"multiple list item insert",
"""
<ol>
<li>one</li>
<ins>
<li>two</li>
<li>three</li>
</ins>
</ol>
""",
"""
<ol>
<li>one</li>
<li><ins>two</ins></li>
<li><ins>three</ins></li>
</ol>
""",
),
(
"simple list item delete afterward",
"""
<ol>
<li>one</li>
<del><li>one and a half</li></del>
</ol>
""",
"""
<ol>
<li>one</li>
<li class="del-li"><del>one and a half</del></li>
</ol>
""",
),
(
"simple list item delete first",
"""
<ol>
<del><li>one half</li></del>
<li>one</li>
</ol>
""",
"""
<ol>
<li class="del-li"><del>one half</del></li>
<li>one</li>
</ol>
""",
),
(
"multiple list item delete first",
"""
<ol>
<del>
<li>one third</li>
<li>two thirds</li>
</del>
<li>one</li>
</ol>
""",
"""
<ol>
<li class="del-li"><del>one third</del></li>
<li class="del-li"><del>two thirds</del></li>
<li>one</li>
</ol>
""",
),
(
"insert and delete separately",
"""
<ol>
<li>one</li>
<ins><li>two</li></ins>
<li>three</li>
<del><li>three point five</li></del>
<li>four</li>
</ol>
""",
"""
<ol>
<li>one</li>
<li><ins>two</ins></li>
<li>three</del>
<li class="del-li"><del>three point five</del></li>
<li>four</li>
</ol>
""",
),
(
"multiple list item delete",
"""
<ol>
<li>one</li>
<del>
<li>two</li>
<li>three</li>
</del>
</ol>
""",
"""
<ol>
<li>one</li>
<li class="del-li"><del>two</del></li>
<li class="del-li"><del>three</del></li>
</ol>
""",
),
(
"delete only list item",
"""
<ol>
<del>
<li>one</li>
</del>
</ol>
""",
"""
<ol>
<li class="del-li"><del>one</del></li>
</ol>
""",
),
]
for test_name, changes, fixed_changes in cases:
changes = collapse(changes)
fixed_changes = collapse(fixed_changes)
def test():
changes_dom = parse_minidom(changes)
fix_lists(changes_dom)
assert_html_equal(minidom_tostring(changes_dom), fixed_changes)
test.description = "test_fix_lists - %s" % test_name
yield test
def test_fix_lists():
cases = [
('simple list item insert', '''
<ol>
<li>one</li>
<ins><li>two</li></ins>
</ol>
''', '''
<ol>
<li>one</li>
<li><ins>two</ins></li>
</ol>
'''),
('multiple list item insert', '''
<ol>
<li>one</li>
<ins>
<li>two</li>
<li>three</li>
</ins>
</ol>
''', '''
<ol>
<li>one</li>
<li><ins>two</ins></li>
<li><ins>three</ins></li>
</ol>
'''),
('simple list item delete afterward', '''
<ol>
<li>one</li>
<del><li>one and a half</li></del>
</ol>
''', '''
<ol>
<li>one</li>
<li class="del-li"><del>one and a half</del></li>
</ol>
'''),
('simple list item delete first', '''
<ol>
<del><li>one half</li></del>
<li>one</li>
</ol>
''', '''
<ol>
<li class="del-li"><del>one half</del></li>
<li>one</li>
</ol>
'''),
('multiple list item delete first', '''
<ol>
<del>
<li>one third</li>
<li>two thirds</li>
</del>
<li>one</li>
</ol>
''', '''
<ol>
<li class="del-li"><del>one third</del></li>
<li class="del-li"><del>two thirds</del></li>
<li>one</li>
</ol>
'''),
('insert and delete separately', '''
<ol>
<li>one</li>
<ins><li>two</li></ins>
<li>three</li>
<del><li>three point five</li></del>
<li>four</li>
</ol>
''', '''
<ol>
<li>one</li>
<li><ins>two</ins></li>
<li>three</del>
<li class="del-li"><del>three point five</del></li>
<li>four</li>
</ol>
'''),
('multiple list item delete', '''
<ol>
<li>one</li>
<del>
<li>two</li>
<li>three</li>
</del>
</ol>
''', '''
<ol>
<li>one</li>
<li class="del-li"><del>two</del></li>
<li class="del-li"><del>three</del></li>
</ol>
'''),
('delete only list item', '''
<ol>
<del>
<li>one</li>
</del>
</ol>
''', '''
<ol>
<li class="del-li"><del>one</del></li>
</ol>
'''),
('LI full content change does not add another LI', '''
<ol>
<del>
<li>AAA</li>
</del>
<ins>
<li>BBB</li>
</ins>
</ol>
''', '''
<ol>
<li><del>AAA</del><ins>BBB</ins></li>
</ol>
'''),
('LI full content change keeps attrs', '''
<ol>
<del>
<li class="old" id="foo">AAA</li>
</del>
<ins>
<li class="new">BBB</li>
</ins>
</ol>
''', '''
<ol>
<li class="new"><del>AAA</del><ins>BBB</ins></li>
</ol>
'''),
(
'LI changes markup internalization fix not done if next tag is not an insert', # noqa
'''
<ol>
<del>
<li>AAA</li>
</del>
<li><strong>BBB</strong></li>
<ins>
<li>CCC</li>
</ins>
</ol>
''',
'''
<ol>
<li class="del-li">
<del>AAA</del>
</li>
<li><strong>BBB</strong></li>
<li><ins>CCC</ins></li>
</ol>
''',
),
(
'LI changes markup internalization fix not done if next tag is not an insert', # noqa
'''
<ol>
<del>
<li>AAA</li>
</del>
<li><strong>BBB</strong></li>
<ins>
<li>CCC</li>
</ins>
</ol>
''',
'''
<ol>
<li class="del-li">
<del>AAA</del>
</li>
<li><strong>BBB</strong></li>
<li><ins>CCC</ins></li>
</ol>
''',
),
(
'LI after del must be ins',
'''
<ol>
<del>
<li>AAA</li>
</del>
<del>
<li>BBB</li>
</del>
<ins>
<li>CCC</li>
</ins>
</ol>
''',
'''
<ol>
<li class="del-li">
<del>AAA</del>
</li>
<li><del>BBB</del><ins>CCC</ins></li>
</ol>
''',
),
(
'LI changes markup internalization fix not performed if next tags child is not li', # noqa
'''
<ol>
<del>
<li>AAA</li>
</del>
<ins>
<foo>BBB</foo>
</ins>
</ol>
''',
'''
<ol>
<li class="del-li">
<del>AAA</del>
</li>
<ins>
<foo>BBB</foo>
</ins>
</ol>
''',
),
(
'LI changes markup internalization fix not performed if next tags is text', # noqa
'''
<ol>
<del>
<li>AAA</li>
</del>
<ins>
BBB
</ins>
</ol>
''',
'''
<ol>
<li class="del-li">
<del>AAA</del>
</li>
<ins>
BBB
</ins>
</ol>
''',
),
]
for test_name, changes, fixed_changes in cases:
changes = collapse(changes)
fixed_changes = collapse(fixed_changes)
def test():
changes_dom = parse_minidom(changes)
fix_lists(changes_dom)
assert_html_equal(minidom_tostring(changes_dom), fixed_changes)
test.description = 'test_fix_lists - %s' % test_name
yield test
'TD content change does not show TD removal',
'''
<table>
<tbody>
<tr>
<td>AAA</td>
<td>BBB</td>
</tr>
</tbody>
</table>
''',
collapse('''
<table>
<tbody>
<tr>
<td>ZZZ</td>
<td>BBB</td>
</tr>
</tbody>
</table>
'''),
collapse('''
<table>
<tbody>
<tr>
<td><del>AAA</del><ins>ZZZ</ins></td>
<td>BBB</td>
</tr>
</tbody>
</table>
'''),
[ # The result of this will be fixed by fix_tables
def test_fix_lists():
cases = [
('simple list item insert', '''
<ol>
<li>one</li>
<ins><li>two</li></ins>
</ol>
''', '''
<ol>
<li>one</li>
<li><ins>two</ins></li>
</ol>
'''),
('multiple list item insert', '''
<ol>
<li>one</li>
<ins>
<li>two</li>
<li>three</li>
</ins>
</ol>
''', '''
<ol>
<li>one</li>
<li><ins>two</ins></li>
<li><ins>three</ins></li>
</ol>
'''),
('simple list item delete afterward', '''
<ol>
<li>one</li>
<del><li>one and a half</li></del>
</ol>
''', '''
<ol>
<li>one</li>
<li class="del-li"><del>one and a half</del></li>
</ol>
'''),
('simple list item delete first', '''
<ol>
<del><li>one half</li></del>
<li>one</li>
</ol>
''', '''
<ol>
<li class="del-li"><del>one half</del></li>
<li>one</li>
</ol>
'''),
('multiple list item delete first', '''
<ol>
<del>
<li>one third</li>
<li>two thirds</li>
</del>
<li>one</li>
</ol>
''', '''
<ol>
<li class="del-li"><del>one third</del></li>
<li class="del-li"><del>two thirds</del></li>
<li>one</li>
</ol>
'''),
('insert and delete separately', '''
<ol>
<li>one</li>
<ins><li>two</li></ins>
<li>three</li>
<del><li>three point five</li></del>
<li>four</li>
</ol>
''', '''
<ol>
<li>one</li>
<li><ins>two</ins></li>
<li>three</del>
<li class="del-li"><del>three point five</del></li>
<li>four</li>
</ol>
'''),
('multiple list item delete', '''
<ol>
<li>one</li>
<del>
<li>two</li>
<li>three</li>
</del>
</ol>
''', '''
<ol>
<li>one</li>
<li class="del-li"><del>two</del></li>
<li class="del-li"><del>three</del></li>
</ol>
'''),
('delete only list item', '''
<ol>
<del>
<li>one</li>
</del>
</ol>
''', '''
<ol>
<li class="del-li"><del>one</del></li>
</ol>
'''),
]
for test_name, changes, fixed_changes in cases:
changes = collapse(changes)
fixed_changes = collapse(fixed_changes)
def test():
changes_dom = parse_minidom(changes)
fix_lists(changes_dom)
assert_html_equal(minidom_tostring(changes_dom), fixed_changes)
test.description = 'test_fix_lists - %s' % test_name
yield test
test_cases = [
(
'nested list items',
collapse('''
<ul>
<li>Monday
<ul>
<li>2pm - 3pm</li>
</ul>
</li>
<li>Wednesday
<ul>
<li>11am - Noon</li>
<li>3pm - 5pm</li>
</ul>
</li>
<li>Thursday
<ul>
<li>11am - Noon</li>
</ul>
</li>
<li>Friday
<ul>
<li>Noon - 1pm</li>
</ul>
</li>
</ul>
'''),
collapse('''
<ul>
<li>Tuesday
<table class="table_class"><tbody>
<tr>
<td colspan="2">top across</td>
</tr>
<tr>
<td>bottom left</td>
<td>bottom right</td>
</tr>
</tbody></table>
''',
collapse('''
<table class="table_class"><tbody>
<tr>
<td colspan="2">top across</td>
</tr>
<tr>
<td>bottom left</td>
<td>bottom right</td>
</tr>
</tbody></table>
'''),
),
(
'whitespace changes in a table with nbsp entity',
'''
<table>
<tbody>
<tr>
<td> </td>
<td> </td>
<td> </td>
def test_fix_lists():
cases = [
(
'simple list item insert',
'''
<ol>
<li>one</li>
<ins><li>two</li></ins>
</ol>
''',
'''
<ol>
<li>one</li>
<li><ins>two</ins></li>
</ol>
'''
),
(
'multiple list item insert',
'''
<ol>
<li>one</li>
<ins>
<li>two</li>
<li>three</li>
</ins>
</ol>
''',
'''
<ol>
<li>one</li>
<li><ins>two</ins></li>
<li><ins>three</ins></li>
</ol>
'''
),
(
'simple list item delete afterward',
'''
<ol>
<li>one</li>
<del><li>one and a half</li></del>
</ol>
''',
'''
<ol>
<li>one</li>
<li class="del-li"><del>one and a half</del></li>
</ol>
'''
),
(
'simple list item delete first',
'''
<ol>
<del><li>one half</li></del>
<li>one</li>
</ol>
''',
'''
<ol>
<li class="del-li"><del>one half</del></li>
<li>one</li>
</ol>
'''
),
(
'multiple list item delete first',
'''
<ol>
<del>
<li>one third</li>
<li>two thirds</li>
</del>
<li>one</li>
</ol>
''',
'''
<ol>
<li class="del-li"><del>one third</del></li>
<li class="del-li"><del>two thirds</del></li>
<li>one</li>
</ol>
'''
),
(
'insert and delete separately',
'''
<ol>
<li>one</li>
<ins><li>two</li></ins>
<li>three</li>
<del><li>three point five</li></del>
<li>four</li>
</ol>
''',
'''
<ol>
<li>one</li>
<li><ins>two</ins></li>
<li>three</del>
<li class="del-li"><del>three point five</del></li>
<li>four</li>
</ol>
'''
),
(
'multiple list item delete',
'''
<ol>
<li>one</li>
<del>
<li>two</li>
<li>three</li>
</del>
</ol>
''',
'''
<ol>
<li>one</li>
<li class="del-li"><del>two</del></li>
<li class="del-li"><del>three</del></li>
</ol>
'''
),
(
'delete only list item',
'''
<ol>
<del>
<li>one</li>
</del>
</ol>
''',
'''
<ol>
<li class="del-li"><del>one</del></li>
</ol>
'''
),
(
'LI full content change does not add another LI',
'''
<ol>
<del>
<li>AAA</li>
</del>
<ins>
<li>BBB</li>
</ins>
</ol>
''',
'''
<ol>
<li><del>AAA</del><ins>BBB</ins></li>
</ol>
'''
),
(
'LI full content change keeps attrs',
'''
<ol>
<del>
<li class="old" id="foo">AAA</li>
</del>
<ins>
<li class="new">BBB</li>
</ins>
</ol>
''',
'''
<ol>
<li class="new"><del>AAA</del><ins>BBB</ins></li>
</ol>
'''
),
(
'LI changes markup internalization fix not done if next tag is not an insert', # noqa
'''
<ol>
<del>
<li>AAA</li>
</del>
<li><strong>BBB</strong></li>
<ins>
<li>CCC</li>
</ins>
</ol>
''',
'''
<ol>
<li class="del-li">
<del>AAA</del>
</li>
<li><strong>BBB</strong></li>
<li><ins>CCC</ins></li>
</ol>
''',
),
(
'LI changes markup internalization fix not done if next tag is not an insert', # noqa
'''
<ol>
<del>
<li>AAA</li>
</del>
<li><strong>BBB</strong></li>
<ins>
<li>CCC</li>
</ins>
</ol>
''',
'''
<ol>
<li class="del-li">
<del>AAA</del>
</li>
<li><strong>BBB</strong></li>
<li><ins>CCC</ins></li>
</ol>
''',
),
(
'LI after del must be ins',
'''
<ol>
<del>
<li>AAA</li>
</del>
<del>
<li>BBB</li>
</del>
<ins>
<li>CCC</li>
</ins>
</ol>
''',
'''
<ol>
<li class="del-li">
<del>AAA</del>
</li>
<li><del>BBB</del><ins>CCC</ins></li>
</ol>
''',
),
(
'LI changes markup internalization fix not performed if next tags child is not li', # noqa
'''
<ol>
<del>
<li>AAA</li>
</del>
<ins>
<foo>BBB</foo>
</ins>
</ol>
''',
'''
<ol>
<li class="del-li">
<del>AAA</del>
</li>
<ins>
<foo>BBB</foo>
</ins>
</ol>
''',
),
(
'LI changes markup internalization fix not performed if next tags is text', # noqa
'''
<ol>
<del>
<li>AAA</li>
</del>
<ins>
BBB
</ins>
</ol>
''',
'''
<ol>
<li class="del-li">
<del>AAA</del>
</li>
<ins>
BBB
</ins>
</ol>
''',
),
]
for test_name, changes, fixed_changes in cases:
changes = collapse(changes)
fixed_changes = collapse(fixed_changes)
def test():
changes_dom = parse_minidom(changes)
fix_lists(changes_dom)
assert_html_equal(minidom_tostring(changes_dom), fixed_changes)
test.description = 'test_fix_lists - %s' % test_name
yield test
