def derive_company_name(cls,
ant: CopyrightAnnotation,
phrase: str) -> None:
if ant.company:
ant.company = ant.company.strip(' ,;-(:')
if cls.reg_valid_company_name.search(ant.company):
return
ant.company = ''
possible_names = [n.group(0) for n in cls.reg_company_name.finditer(ant.name)]
if not possible_names:
possible_names = [n.group(0) for n in cls.reg_company_name.finditer(phrase)]
if possible_names:
ant.company = cls.take_best_company_name(possible_names)
ant.company = ant.company.strip(' ,;-(:')
def make_annotation_from_pattrn(self, locale: str, ptrn: PatternFound,
phrase: LineOrPhrase) -> TextAnnotation:
ant = CopyrightAnnotation(name=ptrn.name,
coords=(ptrn.start, ptrn.end),
text=phrase.text[ptrn.start:ptrn.end],
locale=locale)
ant.company = ptrn.company # pattern in in fact CopyrightPatternFound
ant.year_start = ptrn.start_year
ant.year_end = ptrn.end_year
return ant
def test_format_copyright_annotation(self):
cp = CopyrightAnnotation(name='Siemens',
coords=(0, 100),
text='text text',
locale='locale')
cp.company = 'Siemens'
cp.year_start = 1996
s = cp.get_cite() # '/copyright/Siemens/1996'
self.assertGreater(s.find('copyright'), -1)
self.assertGreater(s.find('Siemens'), -1)
self.assertGreater(s.find('1996'), -1)
cp.year_end = 2019
cp.locale = 'en'
s = cp.get_cite() # '/en/copyright/Siemens/1996/2019'
self.assertGreater(s.find('copyright'), -1)
self.assertGreater(s.find('Siemens'), -1)
self.assertGreater(s.find('1996'), -1)
self.assertGreater(s.find('2019'), -1)