class AnnotatedTweetReader(object):
"""Reads in tweets annotated for LID and codeswitching."""
def __init__(self, infile):
self._csv = UnicodeDictReader(infile, delimiter='\t')
def __iter__(self):
return self
def next(self):
"""Return relevant fields from the next row of the CSV."""
row = self._csv.next()
goldlid = GOLDLID_ABBREVIATIONS[row[FIELD_GOLDLID]] if row[FIELD_GOLDLID] else None
tokens = row[FIELD_TWEET].split()
lidcs = row[FIELD_PREDLID]
return tokens, lidcs, goldlid
def __init__(self, infile):
self._csv = UnicodeDictReader(infile, delimiter='\t')