Exemplo n.º 1
0
 def accepts(self):
     return (
         InputFormat(self, format_id='tei', extension='tei.xml'),
         InputFormat(self, format_id='docx', extension='docx'),
         InputFormat(self, format_id='rst', extension='rst'),
         InputFormat(self, format_id='alpinodocdir', extension='alpinodocdir',directory=True),
     )
Exemplo n.º 2
0
 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)
     )
Exemplo n.º 3
0
 def accepts(self):
     return [(InputFormat(self,
                          format_id='events',
                          extension='.events.integrated',
                          inputparameter='events'),
              InputFormat(self,
                          format_id='archivedir',
                          extension='.archive',
                          inputparameter='archivedir'))]
Exemplo n.º 4
0
 def accepts(self):
     return InputFormat(self, format_id='events',
                        extension='.enhanced'), InputFormat(
                            self,
                            format_id='events',
                            extension='.integrated'), InputFormat(
                                self,
                                format_id='events',
                                extension='.types')
Exemplo n.º 5
0
 def accepts(self):
     #Note: tuple in a list, the outer list corresponds to options, while the inner tuples are conjunctions
     return [(InputFormat(self,
                          format_id='train',
                          extension='train',
                          inputparameter='trainfile'),
              InputFormat(self,
                          format_id='test',
                          extension='test',
                          inputparameter='testfile'))]
Exemplo n.º 6
0
 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='txtdir',
                         extension='txtdir',
                         directory=True),
             InputFormat(self,
                         format_id='foliadir',
                         extension='foliadir',
                         directory=True))
Exemplo n.º 7
0
 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))
Exemplo n.º 8
0
 def accepts(self):
     return (InputFormat(self,
                         format_id='events',
                         extension='.deduplicated'),
             InputComponent(self,
                            DeduplicateEvents,
                            similarity_threshold=self.similarity_threshold))
Exemplo n.º 9
0
 def accepts(self):
     return (InputFormat(self,
                         format_id='enhanced_events',
                         extension='.enhanced'),
             InputComponent(self,
                            EnhanceEvents,
                            similarity_threshold=self.similarity_threshold))
Exemplo n.º 10
0
 def accepts(self):
     return (InputFormat(self,
                         format_id='filtered',
                         extension='.filtered.json'),
             InputComponent(self,
                            FilterTweets,
                            format_json=self.format_json))
Exemplo n.º 11
0
 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))
Exemplo n.º 12
0
 def accepts(self):
     return [(InputFormat(self,
                          format_id='entity_counts',
                          extension='.counts.npz',
                          inputparameter='entity_counts'),
              InputFormat(self,
                          format_id='dates',
                          extension='.counts_dates',
                          inputparameter='dates'),
              InputFormat(self,
                          format_id='vocabulary',
                          extension='.counts_vocabulary',
                          inputparameter='vocabulary'),
              InputFormat(self,
                          format_id='events',
                          extension='.events',
                          inputparameter='events'))]
Exemplo n.º 13
0
 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))
Exemplo n.º 14
0
 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))
Exemplo n.º 15
0
 def accepts(self):
     return InputFormat(self,
                        format_id='txtdir',
                        extension='txtdir',
                        directory=True)
Exemplo n.º 16
0
 def accepts(self):
     return InputFormat(self, format_id='txt', extension='txt')
Exemplo n.º 17
0
 def accepts(self):
     return InputFormat(self,
                        format_id='tiff',
                        extension=self.tiff_extension,
                        directory=True),
Exemplo n.º 18
0
 def accepts(self):
     return [ ( InputFormat(self,format_id='tweetdir',extension='.tweets',inputparameter='tweetdir'), InputFormat(self,format_id='events',extension='.events',inputparameter='events'), InputFormat(self,format_id='entity_burstiness',extension='.burstiness.txt',inputparameter='entity_burstiness') ) ]
Exemplo n.º 19
0
 def accepts(self):
     return (InputFormat(self, format_id='tweetdir', extension='.tweets'), )
Exemplo n.º 20
0
 def accepts(self):
     return InputFormat(self,format_id='events',extension='.events.integrated')
Exemplo n.º 21
0
 def accepts(self):
     return (InputFormat(self, format_id='txt',
                         extension='txt'), InputComponent(self, Lowercaser))
Exemplo n.º 22
0
 def accepts(self):
     return InputFormat(self, format_id='eventdir', extension='.events')
Exemplo n.º 23
0
 def accepts(self):
     return [(InputFormat(self,
                          format_id='events',
                          extension='.events',
                          inputparameter='events'))]
Exemplo n.º 24
0
 def accepts(self):
     return [ ( InputFormat(self,format_id='predictiondir',extension='.instances',inputparameter='predictiondir'), InputFormat(self,format_id='events',extension='.events.integrated',inputparameter='events') ) ]
Exemplo n.º 25
0
 def accepts(self):
     return InputFormat(self, format_id='tweets', extension='.gz')
Exemplo n.º 26
0
 def accepts(self):
     return (
         InputFormat(self, format_id='folia', extension=self.folia_extension),
         InputFormat(self, format_id='foliadir', extension='foliadir'))
Exemplo n.º 27
0
 def accepts(self):
     return [ ( InputFormat(self,format_id='entity_counts_dir',extension='.timeseries',inputparameter='entity_counts_dir') ) ]