class ReferenceEntryFormatter(object):
def __init__(self, format=ReferenceFormat.BIBTEX):
self.reference_gw = ReferenceGateway()
self.format = format
self.util_factory = UtilFactory()
def format_reference(self, reference_id):
log.debug(
'Retrieving reference from the database') #@UndefinedVariable
reference = self.reference_gw.find_reference_by_id(reference_id)
if not reference:
log.error(
'Reference with id %d could not be retrieved' #@UndefinedVariable
% reference_id)
return None
formatter = ReferenceFormatter()
try:
generator = self.util_factory.create_generator(self.format)
except UtilCreationError as e:
log.error('Could not create a formatter for %s: %s'
% #@UndefinedVariable
(self.format, e.args))
return None
log.debug('Starting to format') #@UndefinedVariable
formatter.format_reference(reference, generator)
return reference.entry
class ReferenceEntryFormatter(object):
def __init__(self, format=ReferenceFormat.BIBTEX):
self.reference_gw = ReferenceGateway()
self.format = format
self.util_factory = UtilFactory()
def format_reference(self, reference_id):
log.debug('Retrieving reference from the database') #@UndefinedVariable
reference = self.reference_gw.find_reference_by_id(reference_id)
if not reference:
log.error('Reference with id %d could not be retrieved' #@UndefinedVariable
% reference_id)
return None
formatter = ReferenceFormatter()
try:
generator = self.util_factory.create_generator(self.format)
except UtilCreationError as e:
log.error('Could not create a formatter for %s: %s' % #@UndefinedVariable
(self.format, e.args))
return None
log.debug('Starting to format') #@UndefinedVariable
formatter.format_reference(reference, generator)
return reference.entry
def __init__(self, format=ReferenceFormat.BIBTEX):
super(ReferenceImporter, self).__init__()
self.name = 'Importer'
self.format = format
self.util_factory = UtilFactory()
self.ref_controller = ReferencesController(self.util_factory,
self.format)
self.path = ''
def __init__(self):
self.factory = UtilFactory()
self.rce = controllers.RCEController(self.factory, 10, 10)
self.ir = controllers.IRController(self.factory)
self.results_cache_filename = 'customHeaders_cache_file.txt'
self.results_cache = {}
self.browser = Browser()
self.basepath = '/home/rxuriguera/benchmark/pdfsets/pageHeader/'
self.basename = 'pageHeader'
self.len_range = range(4, 15)
class TestUtilFactory(unittest.TestCase):
def setUp(self):
self.uf = UtilFactory()
def tearDown(self):
pass
def test_create_extractor(self):
extractor = self.uf.create_extractor(FileFormat.PDF, FileFormat.TXT) #@UnusedVariable
self.assertRaises(UtilCreationError,
self.uf.create_extractor, FileFormat.TXT, FileFormat.PDF)
def __init__(self):
self.info = {}
self.nexamples = 4
self.base_path = '/home/rxuriguera/benchmark/pages/'
self.fields = ['addres', 'author', 'isbn', 'issn', 'journal', 'number', 'pages', 'publisher', 'title', 'volume', 'year']
self.libraries = ['informaworld']#['acm', 'citeulike', 'computerorg', 'econpapers', 'ideas', 'informaworld', 'sciencedirect', 'scientificcommons', 'springer']
self.factory = UtilFactory()
self.iec = IEController(self.factory, secs_between_reqs=0,
wrapper_gen_examples=self.nexamples)
self.rec = ReferencesController(self.factory)
def setUp(self):
factory = UtilFactory()
self.iec = IEController(factory, ReferenceFormat.BIBTEX)
self.top_results = [
SearchResult(
'result01',
'http://portal.acm.org/citation.cfm?id=507338.507355'),
SearchResult(
'result01',
'http://www.springerlink.com/index/D7X7KX6772HQ2135.pdf')
]
self.empty_page = BeautifulSoup("<html><head/><body/></html>")
self.page = self._get_soup('acm01.html')
self.text = 'ss'
def setUp(self):
self.uf = UtilFactory()
def setUp(self):
factory = UtilFactory()
self.rec = ReferencesController(factory, ReferenceFormat.BIBTEX)
self.path = normpath(
join(dirname(__file__), ('../../../../tests/'
'fixtures/references/bibtex/import.bib')))
def __init__(self, format=ReferenceFormat.BIBTEX):
self.reference_gw = ReferenceGateway()
self.format = format
self.util_factory = UtilFactory()
def __init__(self, url):
super(WrapperGenerator, self).__init__()
self.name = 'WrapTrainer'
self.url = url
self.factory = UtilFactory()
self.ie_controller = IEController(self.factory)
def __init__(self, format=ReferenceFormat.BIBTEX):
self.reference_gw = ReferenceGateway()
self.format = format
self.util_factory = UtilFactory()
def setUp(self):
factory = UtilFactory()
self.rcec = RCEController(factory)
self.pdf = normpath(join(dirname(__file__), ('../../../../tests/'
'fixtures/extraction/article.pdf')))
def __init__(self):
self.factory = UtilFactory()
