def fieldStart(self, field):
DefaultConllCallback.fieldStart(self, field.lower())
if (self.cc_field == 'title'):
self._title = []
self._cat = self._gold_map.get(self.cc_title, None)
else:
self.first = True
self.mode = NERTrainingCallback.NO_LINK
def fileEnd(self):
"""Closes the output file."""
if self._sent.num_train > 0:
sys.stderr.write("Written {0} sentences out of {1}, with avg length = {2} for file {3}.\n".format(
self._sent.num_train, self._sent.num_sentences,
float(self._sent.num_words) / self._sent.num_train,
self.cc_file))
self._sent.clear_statistics()
DefaultConllCallback.fileEnd(self)
self._out.close()
self._out = None
def __init__(self, out_dir):
"""Initializes the callback. The training sentences are written to
C{outs}.
@param outs the output stream to write the data to."""
if not ensure_dir(out_dir):
raise ValueError
DefaultConllCallback.__init__(self)
self._sent = SentenceData()
self._out_dir = out_dir
self._title = []
self._cat = None
self._gold_map = {}
def fieldStart(self, field):
DefaultConllCallback.fieldStart(self, field)
self._sentence = (field.lower() == 'body' and not self.cc_redirect)
def documentStart(self, title):
DefaultConllCallback.documentStart(self, title)
self._words = []
def fileStart(self, file_name):
"""Opens the output file."""
DefaultConllCallback.fileStart(self, file_name)
self._out = FileWriter(
os.path.join(self._out_dir, os.path.basename(file_name))).open()
def __init__(self, out_dir):
"""@param out_dir the directory the output files will be put."""
if not ensure_dir(out_dir):
raise ValueError
DefaultConllCallback.__init__(self)
self._out_dir = out_dir
def fileEnd(self):
"""Closes the output file."""
DefaultConllCallback.fileEnd(self)
self._out.close()
self._out = None