class Token(dr.Ann):
raw = dr.Text()
norm = dr.Text()
pos = dr.Field()
lemma = dr.Field()
span = dr.Slice()
tidx = dr.Field()
class Token(dr.Ann):
span = dr.Slice()
raw = dr.Field()
norm = dr.Field()
def __repr__(self):
return 'Token(norm={0!r}, span={1}:{2})'.format(
self.norm, self.span.start, self.span.stop)
class WikiDoc(dr.Doc):
id = dr.Field()
name = dr.Field()
tokens = dr.Store(Token)
links = dr.Store(Link)
sections = dr.Store(Section)
sentences = dr.Store(Sentence)
paragraphs = dr.Store(Paragraph)
class MyDoc(dr.Doc):
b = dr.Field()
a = dr.Field()
_c = dr.Field(serial='c')
sl = dr.Slice()
anns = dr.Store(MyAnn)
more_anns = dr.Store(MyAnn)
specialised_repr_anns = dr.Store(SpecialisedReprAnn)
class Entity(dr.Ann):
token_span = dr.Slice(Token)
type = dr.Field()
def __repr__(self):
return 'Entity(type={0!r}, token_span={1}:{2})'.format(
self.type, self.token_span.start, self.token_span.stop)
class Node(dr.Ann):
label = dr.Field()
parent = dr.SelfPointer()
other = dr.Pointer('Node', store='nodes2')
def __repr__(self):
return 'Node({0})'.format(self.label)
class Meta:
name = 'Node'
class SliceAnnot(dr.Ann):
span = dr.Slice('MyAnnot')
name = dr.Field()
def __repr__(self):
return '{}(span={}, name={})'.format(self.__class__.__name__,
self.span, self.name)
class Meta:
name = 'SliceAnnot'
class MyAnnot(dr.Ann):
field = dr.Field()
children = dr.Pointers('MyAnnot')
child = dr.Pointer('MyAnnot')
def __repr__(self):
return '{}(field={}, children={}, child={})'.format(
self.__class__.__name__, self.field, self.children, self.child)
class Meta:
name = 'MyAnnot'
def __call__(self):
attr = self.args.field_name
evaluator = self.evaluator
reader, writer = self.stream_reader_writer
for i, doc in enumerate(reader):
if attr not in doc._dr_s2p:
# TODO: externalise reflection methods
doc._dr_s2p[attr] = attr
doc._dr_fields[attr] = dr.Field(serial=attr)
setattr(doc, attr, evaluator(doc, i))
writer.write(doc)
class DocWithField(dr.Doc):
name = dr.Field()
class Meta:
serial = 'writer.DocWithField'
class Node(dr.Ann): label = dr.Field()
class X(dr.Ann):
foo = dr.Field(serial='chicken')
bar = dr.Field()
class Section(dr.Ann):
title = dr.Field()
span = dr.Slice(Token)
class NamedEntity(dr.Ann):
span = dr.Slice(Token)
label = dr.Field()
class Sent(dr.Ann):
span = dr.Slice(Token)
number = dr.Field()
class Z(dr.Ann):
p = dr.Pointer(A, serial='zp')
value = dr.Field()
class Meta:
serial = 'writer.Z'
class Foo(dr.Ann):
val = dr.Field()
class Doc1(dr.Doc):
name = dr.Field(serial='filename')
xs = dr.Store(X)
class Doc2(dr.Doc):
filename = dr.Field()
xs = dr.Store(X)
class Event(dr.Ann):
name = dr.Field()
date = dr.DateTime()
class SchwaDoc(dr.Doc):
doc_id = dr.Field()
tokens = dr.Store(Token)
named_entities = dr.Store(NamedEntity)
class DocWithFieldWithSerial(dr.Doc):
name = dr.Field(serial='filename')
class Meta:
serial = 'writer.DocWithFieldWithSerial'
class A(dr.Ann):
value = dr.Field()
class Meta:
serial = 'writer.A'
class Node(dr.Ann):
label = dr.Field()
children = dr.Store('Node')
class Meta:
name = 'Node'
class Token(dr.Ann): raw = dr.Field()
class Link(dr.Ann):
span = dr.Slice(Token)
anchor = dr.Field()
target = dr.Field()
class Token(dr.Ann):
span = dr.Slice()
raw = dr.Field()
class X(dr.Ann):
name = dr.Field()
class Bar(dr.Ann):
val = dr.Field()
class MyAnn(dr.Ann):
foo = dr.Field()
others = dr.SelfPointers()
prev = dr.SelfPointer()
