class TestPathRuler(TestRuler):
def setUp(self):
super(TestPathRuler, self).setUp()
self.ruler = PathRuler()
def test_get_element_attrs(self):
attrs = self.ruler._get_element_attrs(self.element01)
self.failUnless(len(attrs) == 1)
self.failUnless(attrs['class'] == 'mediumb-text')
def test_is_unique(self):
description01 = self.ruler._get_element_description(self.element01)
self.failUnless(self.ruler._is_unique(self.soup01, description01))
description02 = self.ruler._get_element_description(self.element02)
self.failIf(self.ruler._is_unique(self.soup01, description02))
def test_get_sibling_number(self):
number = self.ruler._get_sibling_number(self.element03)
self.failUnless(number == 2)
pass
def test_get_element_path(self):
path = self.ruler._get_element_path(self.soup01, self.element02)
self.failUnless(len(path) == 3)
def test_rule_element(self):
elements = self.example01.content.findAll('span',
{u'class':u'small-text'})
rule = self.ruler._rule_element(self.example01, elements[1])
expected = [{u'table of contents': 1},
[u'td', {u'colspan': u'2', u'class': u'small-text'}, 1]]
self.failUnless(rule.pattern == expected)
def test_should_merge(self):
rule01 = Rule([[u'a', {u'a01':u'x'}, 0], [u'b', {}, 1]])
rule02 = Rule([[u'a', {}, 1], [u'b', {u'b02':'x'}, 2]])
should_merge = self.ruler._should_merge(rule01, rule02)
self.failUnless(should_merge == False)
rule01 = Rule([[u'a', {u'a01':u'x'}, 0], [u'b', {}, 1], [u'c', {}, 2]])
rule02 = Rule([[u'a', {}, 1], [u'b', {u'b02':'x'}, 2]])
should_merge = self.ruler._should_merge(rule01, rule02)
self.failUnless(should_merge == False)
rule01 = Rule([[u'a', {u'a01':u'x'}, 0], [u'b', {u'b02':'x'}, 1]])
rule02 = Rule([[u'a', {u'a01':u'x'}, 1], [u'b', {u'b02':'x'}, 2]])
should_merge = self.ruler._should_merge(rule01, rule02)
self.failUnless(should_merge == True)
def test_rule_example(self):
rules = self.ruler._rule_example(self.example01)
pattern01 = [{}, [u'td', {u'colspan': u'2', u'class': u'small-text'}, 1],
[u'span', {u'class': u'small-text'}, 2]]
pattern02 = [{}, [u'td', {u'colspan': u'2', u'class': u'small-text'}, 1],
[u'div', {u'class': u'small-text'}, 7]]
self.failUnless(len(rules) == 2)
self.failUnless(rules[0].pattern == pattern01)
self.failUnless(rules[1].pattern == pattern02)
def test_merge_patterns(self):
general = [{}, [u'td', {u'colspan': u'2', u'class': u'small-text'}, 1],
[u'span', {u'class': u'small-text'}, 5]]
# Merge patterns with different length paths
pattern = [{}, [u'span', {u'class': u'big-text'}, 5]]
self.failUnlessRaises(ValueError, self.ruler._merge_patterns, general,
pattern)
# Merge patterns with different attributes list
pattern = [{}, [u'td', {u'class': u'small-text'}, 1],
[u'span', {u'class': u'small-text'}, 5]]
result = self.ruler._merge_patterns(general, pattern)
expected = [{}, [u'td', {u'class': u'small-text'}, 1],
[u'span', {u'class': u'small-text'}, 5]]
self.failUnless(result == expected, "Different attributes")
# Merge patters with different attribute values
pattern = [{}, [u'td', {u'class': u'small-text'}, 1],
[u'span', {u'class': u'big-text'}, 5]]
result = self.ruler._merge_patterns(general, pattern)
expected = [{}, ['td', {u'class': u'small-text'}, 1],
[u'span', {}, 5]]
self.failUnless(result == expected, "Different attribute values")
def test_rule(self):
rules = self.ruler.rule(set([self.example01, self.example03]))
expected = [PathRule(['.*', {}, [u'td', {u'colspan': u'2',
u'class': u'small-text'}, 1],
[u'span', {u'class': u'small-text'}, 2]]),
PathRule(['.*', {}, [u'td', {u'colspan': u'2',
u'class': u'small-text'}, 1],
[u'div', {u'class': u'small-text'}, 7]])]
self.failUnless(rules == expected)
def test_get_content_elements(self):
elements = self.ruler._get_content_elements(self.example01.value,
self.example01.content)
expected = [u' Volume 70 , Issue 16-18 (October 2007)',
u' Year of Publication: 2007 ']
self.failUnless(len(elements) == 2)
self.failUnless(elements == expected)
elements = self.ruler._get_content_elements(self.example05.value,
self.example05.content)
expected = [u'149-154']
self.failUnless(len(elements) == 1)
self.failUnless(elements == expected)
def test_get_invalid_content_element(self):
example = Example(value='random text', content=BeautifulSoup(''))
elements = self.ruler._get_content_elements(example.value,
example.content)
self.failIf(elements)
def test_apply(self):
rule = PathRule(['.*', {}, [u'td', {u'colspan': u'2',
u'class': u'small-text'}, 1],
[u'span', {u'class': u'small-text'}, 5]
])
result = rule.apply(self.example01.content)
self.failUnless(result[0] == u' Volume 70 , Issue 16-18 (October '
'2007)')
def setUp(self):
super(TestPathRuler, self).setUp()
self.ruler = PathRuler()
class TestPathRuler(TestRuler):
def setUp(self):
super(TestPathRuler, self).setUp()
self.ruler = PathRuler()
def test_get_element_attrs(self):
attrs = self.ruler._get_element_attrs(self.element01)
self.failUnless(len(attrs) == 1)
self.failUnless(attrs['class'] == 'mediumb-text')
def test_is_unique(self):
description01 = self.ruler._get_element_description(self.element01)
self.failUnless(self.ruler._is_unique(self.soup01, description01))
description02 = self.ruler._get_element_description(self.element02)
self.failIf(self.ruler._is_unique(self.soup01, description02))
def test_get_sibling_number(self):
number = self.ruler._get_sibling_number(self.element03)
self.failUnless(number == 2)
pass
def test_get_element_path(self):
path = self.ruler._get_element_path(self.soup01, self.element02)
self.failUnless(len(path) == 3)
def test_rule_element(self):
elements = self.example01.content.findAll('span',
{u'class': u'small-text'})
rule = self.ruler._rule_element(self.example01, elements[1])
expected = [{
u'table of contents': 1
}, [u'td', {
u'colspan': u'2',
u'class': u'small-text'
}, 1]]
self.failUnless(rule.pattern == expected)
def test_should_merge(self):
rule01 = Rule([[u'a', {u'a01': u'x'}, 0], [u'b', {}, 1]])
rule02 = Rule([[u'a', {}, 1], [u'b', {u'b02': 'x'}, 2]])
should_merge = self.ruler._should_merge(rule01, rule02)
self.failUnless(should_merge == False)
rule01 = Rule([[u'a', {
u'a01': u'x'
}, 0], [u'b', {}, 1], [u'c', {}, 2]])
rule02 = Rule([[u'a', {}, 1], [u'b', {u'b02': 'x'}, 2]])
should_merge = self.ruler._should_merge(rule01, rule02)
self.failUnless(should_merge == False)
rule01 = Rule([[u'a', {u'a01': u'x'}, 0], [u'b', {u'b02': 'x'}, 1]])
rule02 = Rule([[u'a', {u'a01': u'x'}, 1], [u'b', {u'b02': 'x'}, 2]])
should_merge = self.ruler._should_merge(rule01, rule02)
self.failUnless(should_merge == True)
def test_rule_example(self):
rules = self.ruler._rule_example(self.example01)
pattern01 = [{},
[u'td', {
u'colspan': u'2',
u'class': u'small-text'
}, 1], [u'span', {
u'class': u'small-text'
}, 2]]
pattern02 = [{},
[u'td', {
u'colspan': u'2',
u'class': u'small-text'
}, 1], [u'div', {
u'class': u'small-text'
}, 7]]
self.failUnless(len(rules) == 2)
self.failUnless(rules[0].pattern == pattern01)
self.failUnless(rules[1].pattern == pattern02)
def test_merge_patterns(self):
general = [{}, [u'td', {
u'colspan': u'2',
u'class': u'small-text'
}, 1], [u'span', {
u'class': u'small-text'
}, 5]]
# Merge patterns with different length paths
pattern = [{}, [u'span', {u'class': u'big-text'}, 5]]
self.failUnlessRaises(ValueError, self.ruler._merge_patterns, general,
pattern)
# Merge patterns with different attributes list
pattern = [{}, [u'td', {
u'class': u'small-text'
}, 1], [u'span', {
u'class': u'small-text'
}, 5]]
result = self.ruler._merge_patterns(general, pattern)
expected = [{}, [u'td', {
u'class': u'small-text'
}, 1], [u'span', {
u'class': u'small-text'
}, 5]]
self.failUnless(result == expected, "Different attributes")
# Merge patters with different attribute values
pattern = [{}, [u'td', {
u'class': u'small-text'
}, 1], [u'span', {
u'class': u'big-text'
}, 5]]
result = self.ruler._merge_patterns(general, pattern)
expected = [{}, ['td', {u'class': u'small-text'}, 1], [u'span', {}, 5]]
self.failUnless(result == expected, "Different attribute values")
def test_rule(self):
rules = self.ruler.rule(set([self.example01, self.example03]))
expected = [
PathRule([
'.*', {},
[u'td', {
u'colspan': u'2',
u'class': u'small-text'
}, 1], [u'span', {
u'class': u'small-text'
}, 2]
]),
PathRule([
'.*', {},
[u'td', {
u'colspan': u'2',
u'class': u'small-text'
}, 1], [u'div', {
u'class': u'small-text'
}, 7]
])
]
self.failUnless(rules == expected)
def test_get_content_elements(self):
elements = self.ruler._get_content_elements(self.example01.value,
self.example01.content)
expected = [
u' Volume 70 , Issue 16-18 (October 2007)',
u' Year of Publication: 2007 '
]
self.failUnless(len(elements) == 2)
self.failUnless(elements == expected)
elements = self.ruler._get_content_elements(self.example05.value,
self.example05.content)
expected = [u'149-154']
self.failUnless(len(elements) == 1)
self.failUnless(elements == expected)
def test_get_invalid_content_element(self):
example = Example(value='random text', content=BeautifulSoup(''))
elements = self.ruler._get_content_elements(example.value,
example.content)
self.failIf(elements)
def test_apply(self):
rule = PathRule([
'.*', {}, [u'td', {
u'colspan': u'2',
u'class': u'small-text'
}, 1], [u'span', {
u'class': u'small-text'
}, 5]
])
result = rule.apply(self.example01.content)
self.failUnless(result[0] == u' Volume 70 , Issue 16-18 (October '
'2007)')
