66
77class Delete (Block ):
88
9- def __init__ (self ,empty = False ,** kwargs ):
9+ def __init__ (self ,coref = True ,empty = False ,misc = False ,** kwargs ):
10+ """Args:
11+ coref: delete coreference attributes in MISC, i.e (Entity|Bridge|SplitAnte)
12+ empty: delete all empty nodes and references to them (from DEPS and MISC[Functor])
13+ misc: delete all attributes in MISC except for SpaceAfter
14+ """
1015super ().__init__ (** kwargs )
16+ self .coref = coref
1117self .empty = empty
18+ self .misc = misc
1219
1320def is_root_reachable_by_deps (self ,node ,parents_to_ignore = None ):
1421""" Check if the root node is reachable from node, possibly after deleting the parents_to_ignore nodes.
@@ -54,7 +61,8 @@ def _deps_ignore_nodes(self, node, parents_to_ignore):
5461
5562def process_document (self ,doc ):
5663# This block should work both with coreference loaded (deserialized) and not.
57- doc ._eid_to_entity = None
64+ if self .coref :
65+ doc ._eid_to_entity = None
5866for root in doc .trees :
5967if self .empty :
6068for node in root .descendants :
@@ -74,7 +82,12 @@ def process_document(self, doc):
7482del node .misc ['Functor' ]
7583root .empty_nodes = []
7684
77- for node in root .descendants + root .empty_nodes :
78- node ._mentions = []
79- for attr in ('Entity' ,'Bridge' ,'SplitAnte' ):
80- del node .misc [attr ]
85+ if self .coref or self .misc :
86+ for node in root .descendants + root .empty_nodes :
87+ if self .misc :
88+ node .misc = 'SpaceAfter=No' if node .no_space_after else None
89+ if self .coref :
90+ node ._mentions = []
91+ if not self .misc :
92+ for attr in ('Entity' ,'Bridge' ,'SplitAnte' ):
93+ del node .misc [attr ]