def testMatchingAttr(self):
html = HTMLReader().readString(self._html)
self.assertEqual(
html.tagWithMatchingAttr('lang', 'en').name(), 'body')
self.assertEqual(
html.tagWithMatchingAttr('id', 'dataTable').name(), 'table')
self.assertEqual(html.tagWithId('dataTable').name(), 'table')
def checkMatchingAttr(self):
html = HTMLReader().readString(self._html)
self.assertEqual(
html.tagWithMatchingAttr('lang', 'en').name(), 'body')
self.assertEqual(
html.tagWithMatchingAttr('id', 'dataTable').name(), 'table')
self.assertEqual(html.tagWithId('dataTable').name(), 'table')
def checkBasics(self):
reader = HTMLReader()
tag = reader.readString('<html> </html>')
self.assertEqual(tag.name(), 'html')
self.assertEqual(reader.rootTag(), tag)
self.assertTrue(reader.filename() is None)
out = StringIO()
tag.pprint(out)
self.assertEqual(out.getvalue(), '<html>\n</html>\n')
def checkBasics(self):
reader = HTMLReader()
tag = reader.readString('<html> </html>')
self.assertEqual(tag.name(), 'html')
self.assertEqual(reader.rootTag(), tag)
self.assertTrue(reader.filename() is None)
out = StringIO()
tag.pprint(out)
self.assertEqual(out.getvalue(), '<html>\n</html>\n')
def checkBasics(self):
reader = HTMLReader()
tag = reader.readString('<html> </html>')
assert tag.name() == 'html'
assert reader.rootTag() == tag
assert reader.filename() == None
out = StringIO()
tag.pprint(out)
assert out.getvalue() == '<html>\n</html>\n'
def checkReuseReader(self):
reader = HTMLReader()
reader.readString('<html> </html>')
tag = reader.readString('<html> <body> </body> </html>')
self.assertFalse(reader.rootTag() is None)
self.assertEqual(reader.rootTag(), tag)
tag = reader.readString('<html> </html>', retainRootTag=0)
self.assertFalse(tag is None)
self.assertTrue(reader.rootTag() is None)
def checkReuseReader(self):
reader = HTMLReader()
reader.readString('<html> </html>')
tag = reader.readString('<html> <body> </body> </html>')
assert reader.rootTag() is not None
assert reader.rootTag() == tag
tag = reader.readString('<html> </html>', retainRootTag=0)
assert tag is not None
assert reader.rootTag() is None
def checkAccess(self):
html = HTMLReader().readString(self._html)
# Name
self.assertEqual(html.name(), 'html')
# Attrs
self.assertEqual(html.numAttrs(), 0)
self.assertFalse(html.hasAttr('foo'))
self.assertRaises(KeyError, html.attr, 'foo')
self.assertTrue(html.attr('foo', None) is None)
# Children and subtags, when both are the same.
for numFoos, fooAt, foos in [[
html.numChildren, html.childAt, html.children
], [html.numSubtags, html.subtagAt, html.subtags]]:
self.assertEqual(numFoos(), 2)
self.assertEqual(len(foos()), 2)
self.assertEqual(fooAt(0).name(), 'head')
self.assertEqual(fooAt(1).name(), 'body')
# Children and subtags when they're different
body = html.subtagAt(1)
p = body.subtagAt(0)
self.assertEqual(p.name(), 'p')
self.assertEqual(p.numChildren(), 3)
self.assertEqual(p.numSubtags(), 1)
def checkAccess(self):
html = HTMLReader().readString(self._html)
# Name
assert html.name() == 'html'
# Attrs
assert html.numAttrs() == 0
assert not html.hasAttr('foo')
self.assertRaises(KeyError, html.attr, 'foo')
assert html.attr('foo', None) is None
# Children and subtags, when both are the same.
for numFoos, fooAt, foos in [[
html.numChildren, html.childAt, html.children
], [html.numSubtags, html.subtagAt, html.subtags]]:
assert numFoos() == 2
assert len(foos()) == 2
assert fooAt(0).name() == 'head'
assert fooAt(1).name() == 'body'
# Children and subtags when they're different
body = html.subtagAt(1)
p = body.subtagAt(0)
assert p.name() == 'p'
assert p.numChildren() == 3
assert p.numSubtags() == 1
def checkInvalidHTML(self):
from WebUtils.HTMLTag import HTMLTagUnbalancedError, HTMLTagIncompleteError
reader = HTMLReader()
html = '<html> <body> <table> </body> </html>'
self.assertRaises(HTMLTagUnbalancedError, reader.readString, html)
html = '<html> <body>'
self.assertRaises(HTMLTagIncompleteError, reader.readString, html)
def checkReuseReader(self):
reader = HTMLReader()
reader.readString('<html> </html>')
tag = reader.readString('<html> <body> </body> </html>')
self.assertFalse(reader.rootTag() is None)
self.assertEqual(reader.rootTag(), tag)
tag = reader.readString('<html> </html>', retainRootTag=0)
self.assertFalse(tag is None)
self.assertTrue(reader.rootTag() is None)
def checkAccess(self):
html = HTMLReader().readString(self._html)
# Name
self.assertEqual(html.name(), 'html')
# Attrs
self.assertEqual(html.numAttrs(), 0)
self.assertFalse(html.hasAttr('foo'))
self.assertRaises(KeyError, html.attr, 'foo')
self.assertTrue(html.attr('foo', None) is None)
# Children and subtags, when both are the same.
for numFoos, fooAt, foos in [
[html.numChildren, html.childAt, html.children],
[html.numSubtags, html.subtagAt, html.subtags]]:
self.assertEqual(numFoos(), 2)
self.assertEqual(len(foos()), 2)
self.assertEqual(fooAt(0).name(), 'head')
self.assertEqual(fooAt(1).name(), 'body')
# Children and subtags when they're different
body = html.subtagAt(1)
p = body.subtagAt(0)
self.assertEqual(p.name(), 'p')
self.assertEqual(p.numChildren(), 3)
self.assertEqual(p.numSubtags(), 1)
def checkMatchingAttr(self):
html = HTMLReader().readString(self._html)
assert html.tagWithMatchingAttr('color', 'white').name() == 'body'
assert html.tagWithMatchingAttr('id', 'dataTable').name() == 'table'
assert html.tagWithId('dataTable').name() == 'table'