def write_mediawiki(pages, path, template):
"""Write a markup file for each page recovered from the XML."""
makedirs(path, exist_ok=True)
with open(template) as tmpl_file:
tmpl = tmpl_file.read()
for page_id, page_data in pages.items():
body = page_data.get('body', '')
body_text = "\n".join(to_mw(body))
filename = page_data.get('title', '') + '_' + str(page_id) + '.txt'
filepath = join(path, filename)
data = {
'body': body_text,
'title': filename,
'version': page_data.get('version'),
'current': page_data.get('contentStatus'),
'created': page_data.get('creationDate'),
'latest_mod': page_data.get('lastModificationDate'),
'position': page_data.get('position'),
}
filecontent = tmpl.format(**data)
with open(filepath, 'w') as outfile:
outfile.write(filecontent)
def test_unordered_list_b(self):
"""There are two versions of unordered list items in
confluence markup, this tests the `-` version.
"""
page_body = 'Some text.\n- first item.\n- second item.\n'
expected = 'Some text.\n* first item.\n* second item.\n'
result = '\n'.join(to_mw(page_body))
self.assertEqual(result, expected)
def test_no_inline_unordered_lists(self):
"""Unordered lists must start at the beginning of a line."""
# Note: we don't test for unordered lists with `*`, because these
# would indicate emphasis when not used at the beginning of the line.
page_body = 'Some text with - a pseudo-list\nSome more -- fake lists.\n'
# Should be a no-op.
result = '\n'.join(to_mw(page_body))
self.assertEqual(result, page_body)
def test_no_inline_headings(self):
"""Most markup which is currently implemented must be at the
beginning of lines, i.e. a `h1. ` in the middle of a line
should not be interpreted as heading markup.
"""
page_body = 'h1. A Heading\n\nA line with h1. in it.\nA line with h3. in it.\n'
expected = '= A Heading =\n\nA line with h1. in it.\nA line with h3. in it.\n'
result = '\n'.join(to_mw(page_body))
self.assertEqual(result, expected)
def test_internal_link(self):
"""An internal link without special label text."""
lines = [
('A line with a [PageLink] in it.\n',
'A line with a [[PageLink]] in it.\n'),
('This line ends with [PageLink].\n',
'This line ends with [[PageLink]].\n'),
('[PageLink] at the start.\n', '[[PageLink]] at the start.\n'),
]
for confluence, mediawiki in lines:
result = '\n'.join(to_mw(confluence))
self.assertEqual(result, mediawiki)
def test_internal_link_with_label(self):
"""An internal link with custom label text."""
lines = [
('A line with a [My Label|PageLink#anchor] in it.\n',
'A line with a [[PageLink#anchor|My Label]] in it.\n'),
('This line ends with [My Label|PageLink#anchor].\n',
'This line ends with [[PageLink#anchor|My Label]].\n'),
('[My Label|PageLink#anchor] at the start.\n',
'[[PageLink#anchor|My Label]] at the start.\n'),
]
for confluence, mediawiki in lines:
result = '\n'.join(to_mw(confluence))
self.assertEqual(result, mediawiki)
def test_strong_emphasis(self):
"""This type of emphasis is normally rendered as bold text."""
known_items = [
("text with *inline emphasis* and more text",
"text with '''inline emphasis''' and more text"),
("*emphasis at the beginning* and more text",
"'''emphasis at the beginning''' and more text"),
("emphasis with following *non-white-space*, and text",
"emphasis with following '''non-white-space''', and text"),
]
for page_body, expected in known_items:
result = '\n'.join(to_mw(page_body))
self.assertEqual(result, expected)
def test_external_link(self):
"""An external link. In this specific case, `to_mw()` should
be a no-op.
"""
lines = [
('A line with a [https://example.com/page#anchor] in it.\n',
'A line with a [https://example.com/page#anchor] in it.\n'),
('This line ends with [https://example.com/page#anchor].\n',
'This line ends with [https://example.com/page#anchor].\n'),
('[https://example.com/page#anchor] at the start.\n',
'[https://example.com/page#anchor] at the start.\n'),
]
for confluence, mediawiki in lines:
result = '\n'.join(to_mw(confluence))
self.assertEqual(result, mediawiki)
def test_external_link_with_label(self):
"""An external link with custom label text. Contrary to the version without label,
the syntax of mediawiki and confluence differs slightly.
"""
lines = [
('A line with a [My Label|https://example.com/page#anchor] in it.\n',
'A line with a [https://example.com/page#anchor My Label] in it.\n'
),
('This line ends with [My Label|https://example.com/page#anchor].\n',
'This line ends with [https://example.com/page#anchor My Label].\n'
),
('[My Label|https://example.com/page#anchor] at the start.\n',
'[https://example.com/page#anchor My Label] at the start.\n'),
]
for confluence, mediawiki in lines:
result = '\n'.join(to_mw(confluence))
self.assertEqual(result, mediawiki)
def test_nested_ordered_lists(self):
"""Test nested ordered lists, but not mixed ones."""
page_body = 'Some text.\n# first, first\n## second, first\n## second, second\n# first, second\n'
expected = 'Some text.\n# first, first\n## second, first\n## second, second\n# first, second\n'
result = '\n'.join(to_mw(page_body))
self.assertEqual(result, expected)
def test_ordered_list(self):
"""Test if ordered list items are transformed correctly."""
page_body = 'Some text.\n# first item\n# second item\n'
expected = 'Some text.\n# first item\n# second item\n'
result = '\n'.join(to_mw(page_body))
self.assertEqual(result, expected)
def test_nested_unordered_lists_b(self):
"""Test nested unordered lists of type `-`, but not mixed ones."""
page_body = 'Some text.\n- first level, first item\n-- second level, first item\n-- second level, second item\n- first level, second item\n'
expected = 'Some text.\n* first level, first item\n** second level, first item\n** second level, second item\n* first level, second item\n'
result = '\n'.join(to_mw(page_body))
self.assertEqual(result, expected)
def test_various_headings(self):
"""Test transformation of heading markup."""
page_body = 'h1. H1\nh2. H2\nh3. H3\nh4. H4\n'
expected = '= H1 =\n== H2 ==\n=== H3 ===\n==== H4 ====\n'
result = '\n'.join(to_mw(page_body))
self.assertEqual(result, expected)
def test_empty_page(self):
"""A completely empty body should yield an empty generator."""
page_body = ''
expected = ''
result = '\n'.join(to_mw(page_body))
self.assertEqual(result, expected)
def test_no_inline_ordered_lists(self):
"""Ordered lists must start at the beginning of a line."""
page_body = 'Some text with # a pseudo list\nSome more ## fake lists\n'
expected = 'Some text with # a pseudo list\nSome more ## fake lists\n'
result = '\n'.join(to_mw(page_body))
self.assertEqual(result, expected)
def test_mixed_lists(self):
"""Ordered and unordered lists can be mixed."""
page_body = '# first\n#* mixed\n#* list\n# for us\n'
expected = '# first\n#* mixed\n#* list\n# for us\n'
result = '\n'.join(to_mw(page_body))
self.assertEqual(result, expected)