Problem with "Span.as_doc()" method in Spacy - spacy

Question

I am working on extraction of dative and direct object using Spacy. Noun.chunks already have already dependency tagging for their roots like dative and dobj, and what I am trying to do is to get Span and save it as Doc to apply further analysis.
I have the following code:
import spacy
nlp = spacy.load("en_core_web_lg")
doc = nlp(open("/-textfile").read())
so far so good, next I got Span objects;
datives = []
for dat in doc.noun_chunks:
if dat.root.dep_ == "dative" and dat.root.head.pos_ == "VERB":
dative.append(dat.sent)
Now I have all the sentences with noun.chunks of which roots are dative and head is a VERB
However, I would to like get token data like from the datives []
dativesent = datives.as_doc()
But the problem is as datives [] is already a list, I cannot convert it to a DOC.
How can I save the sentences with dative-noun.chunks as a DOC?

Answer 1

You can iterate over a sentence (which is a Span) just like a Doc to access the tokens:
import spacy
nlp = spacy.load("en_core_web_sm")
doc = nlp("She gave the dog a bone. He read a book. They gave her a book.")
dative_sents = []
for nc in doc.noun_chunks:
if nc.root.dep_ == "dative" and nc.root.head.pos_ == "VERB":
dative_sents.append(nc.sent)
for dative_sent in dative_sents:
print("Sentence with dative:", dative_sent.text)
for token in dative_sent:
print(token.text, token.pos_, token.dep_)
print()
Output:
Sentence with dative: She gave the dog a bone.
She PRON nsubj
gave VERB ROOT
the DET det
dog NOUN dative
a DET det
bone NOUN dobj
. PUNCT punct
Sentence with dative: They gave her a book.
They PRON nsubj
gave VERB ROOT
her PRON dative
a DET det
book NOUN dobj
. PUNCT punct