def test_callback(self):
html = '<span>1</span><span>2</span>'
self.assertListEqual(String(css='span').parse(html), ['1', '2'])
self.assertListEqual(
String(css='span', callback=int).parse(html), [1, 2])
self.assertEqual(
String(css='span:first-child', callback=int, quant=1).parse(html),
1)
self.assertListEqual(String(css='div', callback=int).parse(html), [])
def test_basic(self):
# css
val = Prefix(css='li',
children=[String(name='name', css='span',
quant=2)]).parse(self.html)
self.assertDictEqual(val, {'name': ['Mike', 'John']})
# xpath
val = Prefix(xpath='//li',
children=[String(name='name', xpath='span',
quant=2)]).parse(self.html)
self.assertDictEqual(val, {'name': ['Mike', 'John']})
def test_callback(self):
val = Prefix(xpath='//li',
callback=lambda d: d['name'],
children=[
String(name='name', css='span', quant=2),
]).parse(self.html)
self.assertListEqual(val, ['Mike', 'John'])
def test_callback(self):
val = Group(css='li',
quant=2,
callback=lambda d: d['name'],
children=[
String(name='name', css='span', quant=1),
]).parse(self.html)
self.assertListEqual(val, ['Mike', 'John'])
def test_basic(self):
html = '<span data-val="rocks">Hello <b>world</b>!</span>'
# by default extract _text
self.assertEqual(
String(name='val', css='span', quant=1).parse(html)['val'],
'Hello !')
self.assertEqual(
String(name='val', css='span', quant=1,
attr='_text').parse(html)['val'], 'Hello !')
self.assertEqual(
String(name='val', css='span', quant=1,
attr='_all_text').parse(html)['val'], 'Hello world!')
self.assertEqual(
String(name='val', css='span', quant=1,
attr='data-val').parse(html)['val'], 'rocks')
self.assertEqual(
String(name='val', css='span', quant=1,
attr='data-invalid').parse(html)['val'], '')
def test_build(self):
# missing children for Group / Prefix parsers
self.assertRaisesRegexp(
ParserError, r'You must specify "children" for Prefix parser',
Prefix)
self.assertRaisesRegexp(
ParserError, r'You must specify "children" for Group parser',
Group)
# missing name of children elements
self.assertRaisesRegexp(
ParserError, r'Children elements inherited from BaseNamedParser',
lambda: Prefix(children=[String()]))
self.assertRaisesRegexp(
ParserError, r'Children elements inherited from BaseNamedParser',
lambda: Group(children=[String()]))
self.assertRaisesRegexp(
ParserError, r'Children elements inherited from BaseNamedParser',
lambda: Prefix(children=[Prefix(children=[String()])]))
self.assertRaisesRegexp(
ParserError, r'Children elements inherited from BaseNamedParser',
lambda: Prefix(children=[Group(name='x', children=[String()])]))
def test_basic(self):
extracted = {
'val': [{
'name': 'Mike',
'link': None
}, {
'name': 'John',
'link': 'http://example.com/test'
}]
}
# css
val = Group(name='val',
css='li',
quant=2,
children=[
String(name='name', css='span', quant=1),
Url(name='link', css='a', quant='?')
]).parse(self.html, url='http://example.com/')
self.assertDictEqual(val, extracted)
# xpath
val = Group(name='val',
css='li',
quant=2,
children=[
String(name='name', xpath='span', quant=1),
Url(name='link', xpath='a', quant='?')
]).parse(self.html, url='http://example.com/')
self.assertDictEqual(val, extracted)
val = Group(name='val',
css='li',
quant=2,
children=[
String(name='name', xpath='descendant::span', quant=1),
Url(name='link', xpath='descendant::a', quant='?')
]).parse(self.html, url='http://example.com/')
self.assertDictEqual(val, extracted)