class TestElementsRegexRuler(object):#(TestRuler):
    def setUp(self):
        self.ruler = ElementsRegexRuler()
        super(TestElementsRegexRuler, self).setUp()
        
        self.example06 = Example(['(Alberto.*Angel|Angel.*Alberto)',
                                  '(Geurts.*Pierre|Pierre.*Geurts)',
                                  '(Damien.*Ernst|Ernst.*Damien)'],
                                 ['The author Alberto Del Angel',
                                  'Pierre Geurts The author',
                                  'Damien Ernst'])

        self.example07 = Example(['(Alberto.*Angel|Angel.*Alberto)',
                                  '(Geurts.*Pierre|Pierre.*Geurts)',
                                  '(Damien.*Ernst|Ernst.*Damien)'],
                                 ['Alberto Del Angel',
                                  'Pierre Geurts',
                                  'Damien Ernst'])
        
    def test_rule_example(self):
        result = self.ruler._rule_example(self.example06)
        self.failUnless(result[0].pattern == u'The\\ author\\ (.*)')

    def test_rule(self):
        rules = self.ruler.rule([self.example06, self.example07])
        self.failUnless(len(rules) == 2)
class TestElementsRegexRuler(object):  #(TestRuler):
    def setUp(self):
        self.ruler = ElementsRegexRuler()
        super(TestElementsRegexRuler, self).setUp()

        self.example06 = Example([
            '(Alberto.*Angel|Angel.*Alberto)',
            '(Geurts.*Pierre|Pierre.*Geurts)', '(Damien.*Ernst|Ernst.*Damien)'
        ], [
            'The author Alberto Del Angel', 'Pierre Geurts The author',
            'Damien Ernst'
        ])

        self.example07 = Example([
            '(Alberto.*Angel|Angel.*Alberto)',
            '(Geurts.*Pierre|Pierre.*Geurts)', '(Damien.*Ernst|Ernst.*Damien)'
        ], ['Alberto Del Angel', 'Pierre Geurts', 'Damien Ernst'])

    def test_rule_example(self):
        result = self.ruler._rule_example(self.example06)
        self.failUnless(result[0].pattern == u'The\\ author\\ (.*)')

    def test_rule(self):
        rules = self.ruler.rule([self.example06, self.example07])
        self.failUnless(len(rules) == 2)
    def setUp(self):
        self.ruler = ElementsRegexRuler()
        super(TestElementsRegexRuler, self).setUp()

        self.example06 = Example([
            '(Alberto.*Angel|Angel.*Alberto)',
            '(Geurts.*Pierre|Pierre.*Geurts)', '(Damien.*Ernst|Ernst.*Damien)'
        ], [
            'The author Alberto Del Angel', 'Pierre Geurts The author',
            'Damien Ernst'
        ])

        self.example07 = Example([
            '(Alberto.*Angel|Angel.*Alberto)',
            '(Geurts.*Pierre|Pierre.*Geurts)', '(Damien.*Ernst|Ernst.*Damien)'
        ], ['Alberto Del Angel', 'Pierre Geurts', 'Damien Ernst'])
    def setUp(self):
        self.ruler = ElementsRegexRuler()
        super(TestElementsRegexRuler, self).setUp()
        
        self.example06 = Example(['(Alberto.*Angel|Angel.*Alberto)',
                                  '(Geurts.*Pierre|Pierre.*Geurts)',
                                  '(Damien.*Ernst|Ernst.*Damien)'],
                                 ['The author Alberto Del Angel',
                                  'Pierre Geurts The author',
                                  'Damien Ernst'])

        self.example07 = Example(['(Alberto.*Angel|Angel.*Alberto)',
                                  '(Geurts.*Pierre|Pierre.*Geurts)',
                                  '(Damien.*Ernst|Ernst.*Damien)'],
                                 ['Alberto Del Angel',
                                  'Pierre Geurts',
                                  'Damien Ernst'])
Exemple #5
0
    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