def test_convert_div(self):
html = """<div>Blah</div>"""
expected_html = """<table><tr><td>Blah</td></tr></table>"""
self.assertTrue(is_html_equal(mediawiki.process_html(html, "Test convert div"), expected_html))
html = """<div class="adr">123 Main Street</div>"""
expected_html = """<span class="adr">123 Main Street</span>"""
self.assertTrue(is_html_equal(mediawiki.process_html(html, "Test convert special div"), expected_html))
def test_convert_div(self):
html = """<div>Blah</div>"""
expected_html = """<table><tr><td>Blah</td></tr></table>"""
self.assertTrue(
is_html_equal(mediawiki.process_html(html, "Test convert div"),
expected_html))
html = """<div class="adr">123 Main Street</div>"""
expected_html = """<span class="adr">123 Main Street</span>"""
self.assertTrue(
is_html_equal(
mediawiki.process_html(html, "Test convert special div"),
expected_html))
def test_external_links(self):
# Make sure we preserve external links.
html = """<p>Some text here</p>
<p>And now a link: <a href="http://example.org/testing/?hi=1&what=there">Waverly Road</a> woo!</p>""" % self.env
expected_html = """<p>Some text here</p>
<p>And now a link: <a href="http://example.org/testing/?hi=1&what=there">Waverly Road</a> woo!</p>""" % self.env
self.assertTrue(is_html_equal(mediawiki.process_html(html), expected_html))
html = """<p><a href="http://news.google.com/newspapers?id=eCVbAAAAIBAJ&sjid=ZU8NAAAAIBAJ&pg=3926%2C4841530" class="external text">Ann Arbor Argus</a>, Feb. 27, 1891.</p>"""
expected_html = """<p><a href="http://news.google.com/newspapers?id=eCVbAAAAIBAJ&sjid=ZU8NAAAAIBAJ&pg=3926%2C4841530" class="external text">Ann Arbor Argus</a>, Feb. 27, 1891.</p>"""
self.assertTrue(is_html_equal(mediawiki.process_html(html), expected_html))
def test_external_links(self):
# Make sure we preserve external links.
html = """<p>Some text here</p>
<p>And now a link: <a href="http://example.org/testing/?hi=1&what=there">Waverly Road</a> woo!</p>""" % self.env
expected_html = """<p>Some text here</p>
<p>And now a link: <a href="http://example.org/testing/?hi=1&what=there">Waverly Road</a> woo!</p>""" % self.env
self.assertTrue(
is_html_equal(mediawiki.process_html(html), expected_html))
html = """<p><a href="http://news.google.com/newspapers?id=eCVbAAAAIBAJ&sjid=ZU8NAAAAIBAJ&pg=3926%2C4841530" class="external text">Ann Arbor Argus</a>, Feb. 27, 1891.</p>"""
expected_html = """<p><a href="http://news.google.com/newspapers?id=eCVbAAAAIBAJ&sjid=ZU8NAAAAIBAJ&pg=3926%2C4841530" class="external text">Ann Arbor Argus</a>, Feb. 27, 1891.</p>"""
self.assertTrue(
is_html_equal(mediawiki.process_html(html), expected_html))
def test_internal_links(self):
# Make sure we turn mediawiki internal links into our-style
# internal wiki links.
# A link to a page that doesn't exist.
html = """<p>Some text here</p>
<p>And now a link: <a href="%(SCRIPT_PATH)s?title=Waverly_Road&action=edit&redlink=1" class="new" title="Waverly Road (page does not exist)">Waverly Road</a> woo!</p>""" % self.env
expected_html = """<p>Some text here</p>
<p>And now a link: <a href="Waverly%20Road">Waverly Road</a> woo!</p>"""
self.assertTrue(is_html_equal(mediawiki.process_html(html), expected_html))
# A link to a page that does exist.
html = """<p>Some text here</p>
<p>And now a link: <a href="%(SCRIPT_PATH)s/Ann_Arbor" title="Ann Arbor">Ann Arbor</a> woo!</p>""" % self.env
expected_html = """<p>Some text here</p>
<p>And now a link: <a href="Ann%20Arbor">Ann Arbor</a> woo!</p>"""
self.assertTrue(is_html_equal(mediawiki.process_html(html), expected_html))
# A link to a redirect in MW.
html = """<a href="%(SCRIPT_PATH)s/Ypsilanti" title="Ypsilanti" class="mw-redirect">Ypsilanti</a>""" % self.env
expected_html = """<a href="Ypsilanti">Ypsilanti</a>"""
def test_internal_links(self):
# Make sure we turn mediawiki internal links into our-style
# internal wiki links.
# A link to a page that doesn't exist.
html = """<p>Some text here</p>
<p>And now a link: <a href="%(SCRIPT_PATH)s?title=Waverly_Road&action=edit&redlink=1" class="new" title="Waverly Road (page does not exist)">Waverly Road</a> woo!</p>""" % self.env
expected_html = """<p>Some text here</p>
<p>And now a link: <a href="Waverly%20Road">Waverly Road</a> woo!</p>"""
self.assertTrue(
is_html_equal(mediawiki.process_html(html), expected_html))
# A link to a page that does exist.
html = """<p>Some text here</p>
<p>And now a link: <a href="%(SCRIPT_PATH)s/Ann_Arbor" title="Ann Arbor">Ann Arbor</a> woo!</p>""" % self.env
expected_html = """<p>Some text here</p>
<p>And now a link: <a href="Ann%20Arbor">Ann Arbor</a> woo!</p>"""
self.assertTrue(
is_html_equal(mediawiki.process_html(html), expected_html))
# A link to a redirect in MW.
html = """<a href="%(SCRIPT_PATH)s/Ypsilanti" title="Ypsilanti" class="mw-redirect">Ypsilanti</a>""" % self.env
expected_html = """<a href="Ypsilanti">Ypsilanti</a>"""
def test_fix_embed(self):
html = """<p><object width="320" height="245"><param name="movie" value="http://www.archive.org/flow/FlowPlayerLight.swf?config=%7Bembedded%3Atrue%2CshowFullScreenButton%3Atrue%2CshowMuteVolumeButton%3Atrue%2CshowMenu%3Atrue%2CautoBuffering%3Atrue%2CautoPlay%3Afalse%2CinitialScale%3A%27fit%27%2CmenuItems%3A%5Bfalse%2Cfalse%2Cfalse%2Cfalse%2Ctrue%2Ctrue%2Cfalse%5D%2CusePlayOverlay%3Afalse%2CshowPlayListButtons%3Atrue%2CplayList%3A%5B%7Burl%3A%27ssfGNSTRIK1%2FssfGNSTRIK1%5F512kb%2Emp4%27%7D%5D%2CcontrolBarGloss%3A%27high%27%2CshowVolumeSlider%3Atrue%2CbaseURL%3A%27http%3A%2F%2Fwww%2Earchive%2Eorg%2Fdownload%2F%27%2Cloop%3Afalse%2CcontrolBarBackgroundColor%3A%270x000000%27%7D"/><param name="wmode" value="transparent"/><embed height="245" width="320" wmode="transparent" type="application/x-shockwave-flash" src="http://www.archive.org/flow/FlowPlayerLight.swf?config=%7Bembedded%3Atrue%2CshowFullScreenButton%3Atrue%2CshowMuteVolumeButton%3Atrue%2CshowMenu%3Atrue%2CautoBuffering%3Atrue%2CautoPlay%3Afalse%2CinitialScale%3A%27fit%27%2CmenuItems%3A%5Bfalse%2Cfalse%2Cfalse%2Cfalse%2Ctrue%2Ctrue%2Cfalse%5D%2CusePlayOverlay%3Afalse%2CshowPlayListButtons%3Atrue%2CplayList%3A%5B%7Burl%3A%27ssfGNSTRIK1%2FssfGNSTRIK1%5F512kb%2Emp4%27%7D%5D%2CcontrolBarGloss%3A%27high%27%2CshowVolumeSlider%3Atrue%2CbaseURL%3A%27http%3A%2F%2Fwww%2Earchive%2Eorg%2Fdownload%2F%27%2Cloop%3Afalse%2CcontrolBarBackgroundColor%3A%270x000000%27%7D"/></object></p>"""
expected_html = '<p><span class="plugin embed"><iframe width="320" height="245" src="http://www.archive.org/embed/ssfGNSTRIK1"/></span></p>'
self.assertEqual(mediawiki.process_html(html, "Test fix embeds"), expected_html)
def test_double_table(self):
html = """<table>
<tbody>
<tr>
<td>
<table style="width: 30em;">
<tbody>
<tr>
<th colspan="2" style="background-color: #ccccff; text-align: center;">
<strong>Lube Cruise</strong></th>
</tr>
<tr>
<th>
Business Name</th>
<td>
Lube Cruise</td>
</tr>
<tr>
<th>
Address</th>
<td>
5794 Fake</td>
</tr>
<tr>
<th>
City</th>
<td>
Fake</td>
</tr>
</tbody>
</table>
</td>
</tr>
</tbody>
</table>"""
expected_html = """<table style="width: 30em;">
<tbody>
<tr>
<th colspan="2" style="background-color: #ccccff; text-align: center;">
<strong>Lube Cruise</strong></th>
</tr>
<tr>
<th>
Business Name</th>
<td>
Lube Cruise</td>
</tr>
<tr>
<th>
Address</th>
<td>
5794 Fake</td>
</tr>
<tr>
<th>
City</th>
<td>
Fake</td>
</tr>
</tbody>
</table>
"""
self.assertTrue(is_html_equal(mediawiki.process_html(html, "Test double table"), expected_html))
def test_remove_spans(self):
html = """<p><span class="fn something">hi there <a href="http://example.com/">dude</a></span></p>"""
expected_html = """<p>hi there <a href="http://example.com/">dude</a></p>"""
self.assertTrue(
is_html_equal(mediawiki.process_html(html, "Test remove span"),
expected_html))
def test_skip_small_tag(self):
html = """<p>this is some <small>small text</small> here.</p>"""
expected_html = """<p>this is some small text here.</p>"""
self.assertTrue(is_html_equal(mediawiki.process_html(html), expected_html))
def test_remove_headline_labels(self):
html = """<h2><span class="mw-headline" id="Water"> Water </span></h2>"""
expected_html = """<h2>Water</h2>"""
self.assertTrue(is_html_equal(mediawiki.process_html(html), expected_html))
def test_remove_headline_labels(self):
html = """<h2><span class="mw-headline" id="Water"> Water </span></h2>"""
expected_html = """<h2>Water</h2>"""
self.assertTrue(
is_html_equal(mediawiki.process_html(html), expected_html))
def test_fix_i_b_tags(self):
html = """<p>Some <i>text <b>here</b></i></p><p>and <i>then</i> <b>some</b> more</p>"""
expected_html = """<p>Some <em>text <strong>here</strong></em></p><p>and <em>then</em> <strong>some</strong> more</p>"""
self.assertTrue(
is_html_equal(mediawiki.process_html(html), expected_html))
def test_fix_embed(self):
html = """<p><object width="320" height="245"><param name="movie" value="http://www.archive.org/flow/FlowPlayerLight.swf?config=%7Bembedded%3Atrue%2CshowFullScreenButton%3Atrue%2CshowMuteVolumeButton%3Atrue%2CshowMenu%3Atrue%2CautoBuffering%3Atrue%2CautoPlay%3Afalse%2CinitialScale%3A%27fit%27%2CmenuItems%3A%5Bfalse%2Cfalse%2Cfalse%2Cfalse%2Ctrue%2Ctrue%2Cfalse%5D%2CusePlayOverlay%3Afalse%2CshowPlayListButtons%3Atrue%2CplayList%3A%5B%7Burl%3A%27ssfGNSTRIK1%2FssfGNSTRIK1%5F512kb%2Emp4%27%7D%5D%2CcontrolBarGloss%3A%27high%27%2CshowVolumeSlider%3Atrue%2CbaseURL%3A%27http%3A%2F%2Fwww%2Earchive%2Eorg%2Fdownload%2F%27%2Cloop%3Afalse%2CcontrolBarBackgroundColor%3A%270x000000%27%7D"/><param name="wmode" value="transparent"/><embed height="245" width="320" wmode="transparent" type="application/x-shockwave-flash" src="http://www.archive.org/flow/FlowPlayerLight.swf?config=%7Bembedded%3Atrue%2CshowFullScreenButton%3Atrue%2CshowMuteVolumeButton%3Atrue%2CshowMenu%3Atrue%2CautoBuffering%3Atrue%2CautoPlay%3Afalse%2CinitialScale%3A%27fit%27%2CmenuItems%3A%5Bfalse%2Cfalse%2Cfalse%2Cfalse%2Ctrue%2Ctrue%2Cfalse%5D%2CusePlayOverlay%3Afalse%2CshowPlayListButtons%3Atrue%2CplayList%3A%5B%7Burl%3A%27ssfGNSTRIK1%2FssfGNSTRIK1%5F512kb%2Emp4%27%7D%5D%2CcontrolBarGloss%3A%27high%27%2CshowVolumeSlider%3Atrue%2CbaseURL%3A%27http%3A%2F%2Fwww%2Earchive%2Eorg%2Fdownload%2F%27%2Cloop%3Afalse%2CcontrolBarBackgroundColor%3A%270x000000%27%7D"/></object></p>"""
expected_html = '<p><span class="plugin embed"><iframe width="320" height="245" src="http://www.archive.org/embed/ssfGNSTRIK1"/></span></p>'
self.assertEqual(mediawiki.process_html(html, "Test fix embeds"),
expected_html)
def test_double_table(self):
html = """<table>
<tbody>
<tr>
<td>
<table style="width: 30em;">
<tbody>
<tr>
<th colspan="2" style="background-color: #ccccff; text-align: center;">
<strong>Lube Cruise</strong></th>
</tr>
<tr>
<th>
Business Name</th>
<td>
Lube Cruise</td>
</tr>
<tr>
<th>
Address</th>
<td>
5794 Fake</td>
</tr>
<tr>
<th>
City</th>
<td>
Fake</td>
</tr>
</tbody>
</table>
</td>
</tr>
</tbody>
</table>"""
expected_html = """<table style="width: 30em;">
<tbody>
<tr>
<th colspan="2" style="background-color: #ccccff; text-align: center;">
<strong>Lube Cruise</strong></th>
</tr>
<tr>
<th>
Business Name</th>
<td>
Lube Cruise</td>
</tr>
<tr>
<th>
Address</th>
<td>
5794 Fake</td>
</tr>
<tr>
<th>
City</th>
<td>
Fake</td>
</tr>
</tbody>
</table>
"""
self.assertTrue(
is_html_equal(mediawiki.process_html(html, "Test double table"),
expected_html))
def test_remove_edit_labels(self):
html = """<h2><span class="editsection">[<a href="%(SCRIPT_PATH)s?title=After-hours_emergency&action=edit&section=2" title="Edit section: Water">edit</a>]</span> <span class="mw-headline" id="Water"> Water </span></h2>""" % self.env
expected_html = """<h2>Water</h2>"""
self.assertTrue(
is_html_equal(mediawiki.process_html(html), expected_html))
def test_fix_i_b_tags(self):
html = """<p>Some <i>text <b>here</b></i></p><p>and <i>then</i> <b>some</b> more</p>"""
expected_html = """<p>Some <em>text <strong>here</strong></em></p><p>and <em>then</em> <strong>some</strong> more</p>"""
self.assertTrue(is_html_equal(mediawiki.process_html(html), expected_html))
def test_skip_small_tag(self):
html = """<p>this is some <small>small text</small> here.</p>"""
expected_html = """<p>this is some small text here.</p>"""
self.assertTrue(
is_html_equal(mediawiki.process_html(html), expected_html))
def test_remove_edit_labels(self):
html = """<h2><span class="editsection">[<a href="%(SCRIPT_PATH)s?title=After-hours_emergency&action=edit&section=2" title="Edit section: Water">edit</a>]</span> <span class="mw-headline" id="Water"> Water </span></h2>""" % self.env
expected_html = """<h2>Water</h2>"""
self.assertTrue(is_html_equal(mediawiki.process_html(html), expected_html))
def test_remove_spans(self):
html = """<p><span class="fn something">hi there <a href="http://example.com/">dude</a></span></p>"""
expected_html = """<p>hi there <a href="http://example.com/">dude</a></p>"""
self.assertTrue(is_html_equal(mediawiki.process_html(html, "Test remove span"), expected_html))
def test_google_maps(self):
html = """<p>stuff</p><googlemap lat="42.243338" lon="-83.616152" zoom="19" scale="yes" overview="yes"> </googlemap>"""
expected_html = """<p>stuff</p>"""
self.assertTrue(is_html_equal(mediawiki.process_html(html, "Test pagename"), expected_html))
def test_google_maps(self):
html = """<p>stuff</p><googlemap lat="42.243338" lon="-83.616152" zoom="19" scale="yes" overview="yes"> </googlemap>"""
expected_html = """<p>stuff</p>"""
self.assertTrue(
is_html_equal(mediawiki.process_html(html, "Test pagename"),
expected_html))
