spaCy/website/docs/api/spangroup.md

title	tag	source	new
SpanGroup	class	spacy/tokens/span_group.pyx	3

A group of arbitrary, potentially overlapping Span objects that all belong to the same Doc object. The group can be named, and you can attach additional attributes to it. Span groups are generally accessed via the Doc.spans attribute, which will convert lists of spans into a SpanGroup object for you automatically on assignment. SpanGroup objects behave similar to lists, so you can append Span objects to them or access a member at a given index.

SpanGroup.__init__

Create a SpanGroup.

Example

doc = nlp("Their goi ng home")
spans = [doc[0:1], doc[2:4]]

# Construction 1
from spacy.tokens import SpanGroup

group = SpanGroup(doc, name="errors", spans=spans, attrs={"annotator": "matt"})
doc.spans["errors"] = group

# Construction 2
doc.spans["errors"] = spans
assert isinstance(doc.spans["errors"], SpanGroup)

Name	Description
doc	The document the span group belongs to. Doc
keyword-only
name	The name of the span group. If the span group is created automatically on assignment to doc.spans, the key name is used. Defaults to "". str
attrs	Optional JSON-serializable attributes to attach to the span group. Dict[str, Any]
spans	The spans to add to the span group. Iterable[Span]

SpanGroup.doc

The Doc object the span group is referring to.

Example

doc = nlp("Their goi ng home")
doc.spans["errors"] = [doc[0:1], doc[2:4]]
assert doc.spans["errors"].doc == doc

Name	Description
RETURNS	The reference document. Doc

SpanGroup.has_overlap

Check whether the span group contains overlapping spans.

Example

doc = nlp("Their goi ng home")
doc.spans["errors"] = [doc[0:1], doc[2:4]]
assert not doc.spans["errors"].has_overlap
doc.spans["errors"].append(doc[1:2])
assert doc.spans["errors"].has_overlap

Name	Description
RETURNS	Whether the span group contains overlaps. bool

SpanGroup.__len__

Get the number of spans in the group.

Example

doc = nlp("Their goi ng home")
doc.spans["errors"] = [doc[0:1], doc[2:4]]
assert len(doc.spans["errors"]) == 2

Name	Description
RETURNS	The number of spans in the group. int

SpanGroup.__getitem__

Get a span from the group.

Example

doc = nlp("Their goi ng home")
doc.spans["errors"] = [doc[0:1], doc[2:4]]
span = doc.spans["errors"][1]
assert span.text == "goi ng"

Name	Description
i	The item index. int
RETURNS	The span at the given index. Span

SpanGroup.append

Add a Span object to the group. The span must refer to the same Doc object as the span group.

Example

doc = nlp("Their goi ng home")
doc.spans["errors"] = [doc[0:1]]
doc.spans["errors"].append(doc[2:4])
assert len(doc.spans["errors"]) == 2

Name	Description
span	The span to append. Span

SpanGroup.extend

Add multiple Span objects to the group. All spans must refer to the same Doc object as the span group.

Example

doc = nlp("Their goi ng home")
doc.spans["errors"] = []
doc.spans["errors"].extend([doc[2:4], doc[0:1]])
assert len(doc.spans["errors"]) == 2

Name	Description
spans	The spans to add. Iterable[Span]

SpanGroup.to_bytes

Serialize the span group to a bytestring.

Example

doc = nlp("Their goi ng home")
doc.spans["errors"] = [doc[0:1], doc[2:4]]
group_bytes = doc.spans["errors"].to_bytes()

Name	Description
RETURNS	The serialized SpanGroup. bytes

SpanGroup.from_bytes

Load the span group from a bytestring. Modifies the object in place and returns it.

Example

from spacy.tokens import SpanGroup

doc = nlp("Their goi ng home")
doc.spans["errors"] = [doc[0:1], doc[2:4]]
group_bytes = doc.spans["errors"].to_bytes()
new_group = SpanGroup()
new_group.from_bytes(group_bytes)

Name	Description
bytes_data	The data to load from. bytes
RETURNS	The SpanGroup object. SpanGroup