def get_trigram(mode, use_raw=False, corpi=CORPI, name_stem=''):
if use_raw:
tg = load_corpi(mode, corpi)
tg.save_trigrams(os.path.join(TRIGRAM_MODEL_DIR, '%s%s.txt' % (name_stem, mode)))
else:
tg = Trigram(mode=mode)
tg.load_trigrams(os.path.join(TRIGRAM_MODEL_DIR, '%s%s.txt' % (name_stem, mode)))
return tg
def load_corpi(mode, corpi=CORPI):
if isinstance(mode, Trigram):
tg = mode
mode = tg.mode
else:
tg = Trigram(mode=mode)
for base, gz, precooked in corpi:
if precooked:
tg.load_trigrams(cooked_corpi_path(base, mode, gz))
else:
fn = raw_corpi_path(base, gz)
tg.import_text(fn)
return tg