pre-annotation-spacy.py

import spacy 

nlp = spacy.load('en_core_web_md')

def annotate_text(doc):
  ls = []
  for ent in doc.ents:
    
    entry = dict()
    entry['text'] = ent.text
    entry['label'] = ent.label_ 
    entry['start'] = ent.start_char
    entry['end'] = ent.end_char
    entry['propertiesList'] = []
    ls.append(entry)

  return ls

doc = nlp(corpus[0])
annotate_text(doc)

def annotate_corpus(corpus,nlp):
  res = []
  for doc in nlp.pipe(corpus):
    entry = dict()
    entry['document'] = str(doc)
    entry['annotation'] = annotate_text(doc)
    res.append(entry)

  return res

data = annotate_corpus(corpus,nlp)
data[0]