OUTPUT Output xml file with tokenized text
"""
import xml.sax
import sys
from collections import namedtuple, defaultdict
from Tokenizer import PTBTokenizer
# data types for annotation
aspect_term = namedtuple("aspect_term", "term polarity start end")
aspect_category = namedtuple("aspect_category", "category polarity")
# create Penn Treebank tokenizer
tokenizer = PTBTokenizer()
class AnnotationHandler(xml.sax.ContentHandler):
def __init__(self):
self.sentences = {}
self.aspect_terms = defaultdict(list)
self.aspect_categories = defaultdict(list)
self.text = ""
def startElement(self, name, attrs):
if name == "sentence":
self.id = int(attrs['id'])
elif name == "text":
self.text = ""
elif name == "aspectTerm":
