Example #1
0
def postag_texts(path,procn=cpu_count()):
  # filling the queue of jobs (POS-dictionaries and output filenames)
  jobs = Queue()
  for fname in os.listdir(path):
    if os.path.splitext(fname)[-1].lower() != '.par':
      continue
    jobs.put(os.path.join(path,fname))
  lock = Lock()
  # executing the POS tagging in parallel
  util.parex(jobs,processor_pos,lock,(),procn=procn,store_results=False)
Example #2
0
def extract_cooc(path,procn=cpu_count(),add_verbs=True):
  # filling the queue of jobs (POS-dictionaries and output filenames)
  jobs = Queue()
  for fname in os.listdir(path):
    if os.path.splitext(fname)[-1].lower() != '.pos':
      continue
    dct = parse_pos(os.path.join(path,fname))
    filename = os.path.join(path,os.path.splitext(fname)[0]+'.t2s')
    jobs.put((dct,filename))
  lock = Lock()
  # executing the COOC extraction in parallel
  args = (add_verbs,)
  util.parex(jobs,processor_cooc,lock,args,procn=procn,store_results=False)