def accepts(self):
return (InputFormat(self,
format_id='filtered',
extension='.filtered.json'),
InputComponent(self,
FilterTweets,
format_json=self.format_json))
def accepts(self):
return (InputFormat(self,
format_id='events',
extension='.deduplicated'),
InputComponent(self,
DeduplicateEvents,
similarity_threshold=self.similarity_threshold))
def accepts(self):
return (InputFormat(self, format_id='tokenized', extension='tok.json'),
InputComponent(self,
Tokenize,
config=self.config,
strip_punctuation=self.strip_punctuation,
lowercase=self.to_lowercase))
def accepts(self):
"""Returns a tuple of all the initial inputs and other workflows this component accepts as input (a disjunction, only one will be selected)"""
return (InputFormat(self,
format_id='tiffdir',
extension='tiffdir',
directory=True),
InputComponent(self, ExtractPages))
def accepts(self):
return (InputFormat(self,
format_id='enhanced_events',
extension='.enhanced'),
InputComponent(self,
EnhanceEvents,
similarity_threshold=self.similarity_threshold))
def accepts(self):
"""Returns a tuple of all the initial inputs and other workflows this component accepts as input (a disjunction, only one will be selected)"""
return (
InputFormat(self, format_id='folia', extension='folia.xml'),
InputFormat(self, format_id='txt', extension='txt'),
InputComponent(self, ConvertToFoLiA)
)
def accepts(self):
return (InputFormat(self,
format_id='merged_events',
extension='.merged'),
InputComponent(self,
MergeEvents,
overlap_threshold=self.overlap_threshold,
similarity_threshold=self.similarity_threshold))
def accepts(self):
return (InputFormat(self, format_id='cityref', extension='.json'),
InputComponent(self,
ExtractCityref,
config=self.config,
strip_punctuation=self.strip_punctuation,
to_lowercase=self.to_lowercase,
citylist=self.citylist,
skip_date=self.skip_date,
skip_month=self.skip_month,
skip_timeunit=self.skip_timeunit,
skip_day=self.skip_day))
def accepts(self):
return (InputFormat(self,
format_id='dateref',
extension='.dateref.json'),
InputComponent(self,
ExtractDateref,
config=self.config,
strip_punctuation=self.strip_punctuation,
to_lowercase=self.to_lowercase,
skip_datematch=self.skip_date,
skip_monthmatch=self.skip_month,
skip_timeunitmatch=self.skip_timeunit,
skip_daymatch=self.skip_day))
def accepts(self):
return (InputFormat(self, format_id='txt',
extension='txt'), InputComponent(self, Lowercaser))
def accepts(self):
"""Returns a tuple of all the initial inputs and other workflows this component accepts as input (a disjunction, only one will be selected)"""
return InputComponent(self, OCR_document)