예제 #1
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
예제 #2
0
class TestExampleGateway(unittest.TestCase):
    def setUp(self):
        self.eg = ExampleGateway()

    def tearDown(self):
        pass

    def xtestGetExamples(self):
        examples = self.eg.get_examples(2, "http://zetcode.com", 0.5) #@UnusedVariable
        self.eg.session.flush()
        self.eg.session.close()
예제 #3
0
class TestExampleGateway(unittest.TestCase):
    def setUp(self):
        self.eg = ExampleGateway()

    def tearDown(self):
        pass

    def xtestGetExamples(self):
        examples = self.eg.get_examples(2, "http://zetcode.com",
                                        0.5)  #@UnusedVariable
        self.eg.session.flush()
        self.eg.session.close()
예제 #4
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
예제 #5
0
 def setUp(self):
     self.eg = ExampleGateway()
예제 #6
0
 def setUp(self):
     self.eg = ExampleGateway()