def generate_wrappers(self, url):
wrapper_manager = WrapperGateway()
example_manager = ExampleGateway(max_examples=self.max_examples,
max_examples_from_db=
self.max_examples_from_db,
seconds_between_requests=
self.secs_between_reqs)
example_sets = example_manager.get_examples(self.wrapper_gen_examples,
url, self.min_validity)
rulers = []
for set in example_sets:
log.info('Starting wrapper training for set "%s"' % set) #@UndefinedVariable
if set == 'author' or set == 'editor':
rulers = [MultiValuePathRuler(),
SeparatorsRegexRuler(),
ElementsRegexRuler(),
PersonRuler()]
else:
try:
value_guide = self.value_guides[set]
pass
except KeyError:
value_guide = '.*'
rulers = [PathRuler(value_guide), RegexRuler()]
trainer = WrapperTrainer(rulers, self.wrapper_gen_examples)
try:
wrappers = trainer.train(example_sets[set])
wrappers = self._prune_wrappers(wrappers)
wrapper_manager.persist_wrappers(url, set, wrappers)
log.info('Trainer generated %d wrappers' % len(wrappers)) #@UndefinedVariable
except Exception, e:
log.error('Error training wrapper for set "%s": %s' % (set, e)) #@UndefinedVariable
def setUp(self):
self.wt = WrapperTrainer([MockRuler()], num_examples=3)
self.nsets = 5
self.example_sets = MockExampleManager()._get_examples(
self.nsets, 5, 10)
self.example_set = [
Example('v01', 'c01'),
Example('v02', 'c02'),
Example('v03', 'c03')
]
def setUp(self):
self.wt = WrapperTrainer([MockRuler()], num_examples=3)
self.nsets = 5
self.example_sets = MockExampleManager()._get_examples(self.nsets,
5, 10)
self.example_set = [Example('v01', 'c01'),
Example('v02', 'c02'),
Example('v03', 'c03')]
def generate_wrappers(self, url):
wrapper_manager = WrapperGateway()
example_manager = ExampleGateway(
max_examples=self.max_examples,
max_examples_from_db=self.max_examples_from_db,
seconds_between_requests=self.secs_between_reqs)
example_sets = example_manager.get_examples(self.wrapper_gen_examples,
url, self.min_validity)
rulers = []
for set in example_sets:
log.info('Starting wrapper training for set "%s"' %
set) #@UndefinedVariable
if set == 'author' or set == 'editor':
rulers = [
MultiValuePathRuler(),
SeparatorsRegexRuler(),
ElementsRegexRuler(),
PersonRuler()
]
else:
try:
value_guide = self.value_guides[set]
pass
except KeyError:
value_guide = '.*'
rulers = [PathRuler(value_guide), RegexRuler()]
trainer = WrapperTrainer(rulers, self.wrapper_gen_examples)
try:
wrappers = trainer.train(example_sets[set])
wrappers = self._prune_wrappers(wrappers)
wrapper_manager.persist_wrappers(url, set, wrappers)
log.info('Trainer generated %d wrappers' %
len(wrappers)) #@UndefinedVariable
except Exception, e:
log.error('Error training wrapper for set "%s": %s' %
(set, e)) #@UndefinedVariable
class TestWrapperTrainer(unittest.TestCase):
def setUp(self):
self.wt = WrapperTrainer([MockRuler()], num_examples=3)
self.nsets = 5
self.example_sets = MockExampleManager()._get_examples(
self.nsets, 5, 10)
self.example_set = [
Example('v01', 'c01'),
Example('v02', 'c02'),
Example('v03', 'c03')
]
def test_train_too_few_examples(self):
self.wt.set_num_examples(20)
result = self.wt.train(self.example_sets)
self.failUnless(len(result) == 0)
def test_train(self):
self.wt.rulers = [MockRuler(), MockRuler()]
wrappers = self.wt.train(self.example_sets['a'])
self.failUnless(len(wrappers) == 4)
self.failUnless(len(wrappers[0].rules) == 2)
self.failUnless(wrappers[0].downvotes > 0)
def test_get_new_example_set(self):
example_set = self.wt._get_new_example_set(MockRule('xx'),
self.example_set)
self.failUnless(len(example_set) == 3)
self.failUnless(example_set[0].content == 'c01_rxx')
def test_get_rule_sets(self):
rules = self.wt._get_rule_sets([MockRuler(), MockRuler()],
self.example_set)
self.failUnless(len(rules) == 4)
length_bools = map(lambda x: len(x) == 2, rules)
self.failUnless(length_bools.count(False) == 0)
def test_evaluate_single_value_wrapper(self):
info = 'In Germany'
value = 'Germany'
result = self.wt._evaluate_single_value_wrapper(info, value)
self.failUnless(result == True)
info = 'Germany'
value = 'In Germany'
result = self.wt._evaluate_single_value_wrapper(info, value)
self.failUnless(result == False)
def test_evaluate_multi_value_wrapper(self):
info_list = ['Alberto Del Angel', 'Pierre Geurts', 'Damien Ernst']
values = [
'(Alberto.*Angel|Angel.*Alberto)',
'(Geurts.*Pierre|Pierre.*Geurts)', '(Damien.*Ernst|Ernst.*Damien)'
]
result = self.wt._evaluate_multi_value_wrapper(info_list, values)
self.failUnless(result == True)
info_list = ['Alberto Del Ange']
result = self.wt._evaluate_multi_value_wrapper(info_list, values)
self.failUnless(result == False)
def test_evaluate_wrapper(self):
self.example_set = [
Example('c01_r22_r33', 'c01'),
Example('c02_r22_r33', 'c02'),
Example('incorrect_value', 'c03')
]
wrapper = Wrapper(rules=[MockRule(22), MockRule(33)])
self.wt._evaluate_wrapper(wrapper, self.example_set)
self.failUnless(wrapper.upvotes == 2)
self.failUnless(wrapper.downvotes == 1)
class TestWrapperTrainer(unittest.TestCase):
def setUp(self):
self.wt = WrapperTrainer([MockRuler()], num_examples=3)
self.nsets = 5
self.example_sets = MockExampleManager()._get_examples(self.nsets,
5, 10)
self.example_set = [Example('v01', 'c01'),
Example('v02', 'c02'),
Example('v03', 'c03')]
def test_train_too_few_examples(self):
self.wt.set_num_examples(20)
result = self.wt.train(self.example_sets)
self.failUnless(len(result) == 0)
def test_train(self):
self.wt.rulers = [MockRuler(), MockRuler()]
wrappers = self.wt.train(self.example_sets['a'])
self.failUnless(len(wrappers) == 4)
self.failUnless(len(wrappers[0].rules) == 2)
self.failUnless(wrappers[0].downvotes > 0)
def test_get_new_example_set(self):
example_set = self.wt._get_new_example_set(MockRule('xx'),
self.example_set)
self.failUnless(len(example_set) == 3)
self.failUnless(example_set[0].content == 'c01_rxx')
def test_get_rule_sets(self):
rules = self.wt._get_rule_sets([MockRuler(), MockRuler()],
self.example_set)
self.failUnless(len(rules) == 4)
length_bools = map(lambda x: len(x) == 2, rules)
self.failUnless(length_bools.count(False) == 0)
def test_evaluate_single_value_wrapper(self):
info = 'In Germany'
value = 'Germany'
result = self.wt._evaluate_single_value_wrapper(info, value)
self.failUnless(result == True)
info = 'Germany'
value = 'In Germany'
result = self.wt._evaluate_single_value_wrapper(info, value)
self.failUnless(result == False)
def test_evaluate_multi_value_wrapper(self):
info_list = ['Alberto Del Angel',
'Pierre Geurts',
'Damien Ernst']
values = ['(Alberto.*Angel|Angel.*Alberto)',
'(Geurts.*Pierre|Pierre.*Geurts)',
'(Damien.*Ernst|Ernst.*Damien)']
result = self.wt._evaluate_multi_value_wrapper(info_list, values)
self.failUnless(result == True)
info_list = ['Alberto Del Ange']
result = self.wt._evaluate_multi_value_wrapper(info_list, values)
self.failUnless(result == False)
def test_evaluate_wrapper(self):
self.example_set = [Example('c01_r22_r33', 'c01'),
Example('c02_r22_r33', 'c02'),
Example('incorrect_value', 'c03')]
wrapper = Wrapper(rules=[MockRule(22), MockRule(33)])
self.wt._evaluate_wrapper(wrapper, self.example_set)
self.failUnless(wrapper.upvotes == 2)
self.failUnless(wrapper.downvotes == 1)
