def parse_doc(s):
data = s
paragraphs = data.split("\n\n")
current_doc = Document()
current_sentence = Sentence()
for paragraph in paragraphs:
if len(paragraph.strip()) == 0:
continue
# initial comment
if paragraph.startswith("# parser"):
break
# new doc identifier
if paragraph.strip().startswith("url ="):
current_doc = Document()
current_doc.meta = "# newdoc " + paragraph
continue
# sentence
current_sentence = Sentence()
words = paragraph.split("\n")
for word in words:
if word.startswith("# sent_id"):
current_sentence.sent_id = word
elif word.startswith("# text"):
current_sentence.text = word
else:
current_sentence.add_word_conll(word)
current_doc.add_sentence(current_sentence)
return current_doc
