def parse(self, log: ProcessLogger, text, text_unit_id, _text_unit_lang, **kwargs) -> ParseResults:
# TODO: what's the logic behind [:200] ... < 100 ?
found = list(copyright.get_copyright_annotations(text, return_sources=True))
if found:
unique = set(found)
return ParseResults({CopyrightUsage: [CopyrightUsage(text_unit_id=text_unit_id,
year=item.date,
name=item.name[:200],
copyright_str=item.text[:200],
count=found.count(item)
) for item in unique if len(item.name) < 100]})
def parse(self, text, text_unit_id, _text_unit_lang,
**kwargs) -> ParseResults:
found = list(copyright.get_copyright(text, return_sources=True))
if found:
unique = set(found)
return ParseResults({
CopyrightUsage: [
CopyrightUsage(text_unit_id=text_unit_id,
year=item[1],
name=item[2][:200],
copyright_str=item[3][:200],
count=found.count(item)) for item in unique
if len(item[2]) < 100
]
})