示例#1
0
    def setUp(self):
        cwd = os.getcwd()
        if sys.platform == 'win32':
            self.textfile = os.path.join(cwd, 'tests', 'test_sjis.txt')
        else:
            self.textfile = os.path.join(cwd, 'tests', 'test_utf8.txt')

        yamlfile = os.path.join(cwd, 'tests', 'test_utf8.yml')
        self.env = env.MeCabEnv()

        self.b2s, self.s2b = support.string_support(self.env.charset)

        self.testrc = os.path.join(cwd, 'tests', 'testmecabrc')

        with codecs.open(self.textfile, 'r') as f:
            self.text = f.readlines()[0].strip()

        with codecs.open(yamlfile, 'r', encoding='utf-8') as f:
            self.yaml = yaml.load(f)

        cmd = ['mecab', '-P']
        mout = Popen(cmd, stdout=PIPE).communicate()
        res = self.b2s(mout[0])
        m = re.search('(?<=dicdir:\s).*', res)
        ipadic = path.abspath(m.group(0).strip())
        with open(path.join(os.getcwd(), 'tests', 'mecabrc.tmp'), 'r') as fin:
            tmpl = Template(fin.read())

            tmpl = tmpl.substitute({'ipadic': ipadic})

            with open(self.testrc, 'w') as fout:
                fout.write(tmpl)
示例#2
0
    def setUp(self):
        self.env = env.MeCabEnv()
        enc = self.env.charset

        self.bytes2str, self.str2bytes = support.string_support(enc)
        self.split_pattern, self.split_features = support.splitter_support(enc)

        cwd = os.getcwd()
        yamlfile = os.path.join(cwd, 'tests', 'test_utf8.yml')

        with codecs.open(yamlfile, 'r', encoding='utf-8') as f:
            self.yaml = yaml.load(f)
示例#3
0
    def setUp(self):
        self.env = env.MeCabEnv()
        enc = self.env.charset

        self.bytes2str, self.str2bytes = support.string_support(enc)
        self.split_pattern, self.split_features = support.splitter_support(enc)

        cwd = os.getcwd()
        yamlfile = os.path.join(cwd, 'tests', 'test_utf8.yml')

        with codecs.open(yamlfile, 'r', encoding='utf-8') as f:
            self.yaml = yaml.load(f)