# -*- coding: utf-8 -*-
import sys
sys.path[0:0] = ["build/lib.linux-i686-2.6/"]
import cmmseg
#cmmseg.init('F:\\deps\\mmseg\\src\\win32')
cmmseg.init('/usr/local/coreseek/dict/')
rs = cmmseg.segment((u'中文分词').encode('utf-8'))
for i in rs:
print i.decode('utf-8')
def init_cmmseg_dict(dict_path):
logger.info("init cmmseg %s" % dict_path)
try:
cmmseg.instance(dict_path)
except:
cmmseg.init(dict_path)
def __init__(self, dict_etc_path="/usr/local/coreseek/dict"):
cmmseg.init(dict_etc_path)
Tokenizer.__init__(self, cmmseg.segment)
# -*- coding: utf-8 -*-
import cmmseg
cmmseg.init('F:\\deps\\mmseg\\src\\win32')
rs = cmmseg.segment((u'中文分词').encode('utf-8'))
for i in rs:
print i.decode('utf-8')
def init_cmmseg_dict(dict_path):
logger.info("init cmmseg %s" % dict_path)
try:
cmmseg.instance(dict_path)
except:
cmmseg.init(dict_path)
def __init__(self, dict_etc_path="/usr/local/coreseek/dict"):
cmmseg.init(dict_etc_path)
Tokenizer.__init__(self, cmmseg.segment)