示例#1
0
def check(args):
    """
    clpa check <STRING>
    """
    if len(args.args) != 1:
        raise ParserError('only one argument allowed')
    check = check_string(args.args[0], load_whitelist())
    print('\t'.join(args.args[0].split(' ')))
    print('\t'.join(check))
示例#2
0
文件: cli.py 项目: LinguList/clpa
def check(args):
    """
    clpa check <STRING>
    """
    if len(args.args) != 1:
        raise ParserError('only one argument allowed')
    check = check_string(args.args[0], load_whitelist())
    print('\t'.join(args.args[0].split(' ')))
    print('\t'.join(check))
示例#3
0
    def test_find_token(self):
        from pyclpa.util import find_token, load_whitelist, load_alias

        wl = load_whitelist()
        patterns = load_alias('patterns.tsv')
        assert not find_token('t', {}, {}, {}, {}, [])
        assert find_token('t', wl, {}, {}, {}, []) == 't'
        assert find_token('th', wl, {'h': 'ʰ'}, {}, {}, []) == 'tʰ'
        assert find_token('th', wl, {}, {'th': 'x'}, {}, []) == 'x'
        with self.assertRaises(ValueError):
            find_token('th', wl, {}, {'th': 'X'}, {}, [])
        assert find_token('th', wl, {}, {}, patterns, []) == 'tʰ'
        assert find_token('th', wl, {}, {}, {}, ['h']) == 't'
示例#4
0
文件: base.py 项目: glottobank/clpa
 def __init__(self,
              whitelist=None,
              alias=None,
              delete=None,
              explicit=None,
              patterns=None,
              accents=None,
              rules=None):
     self.whitelist = whitelist or load_whitelist()
     self.alias = alias or load_alias('alias.tsv')
     self.delete = delete or ['\u0361', '\u035c', '\u0301']
     self.explicit = explicit or load_alias('explicit.tsv')
     self.patterns = patterns or load_alias('patterns.tsv')
     self.accents = accents or "ˈˌ'"
     self.rules = rules or []
示例#5
0
文件: base.py 项目: LinguList/clpa
 def __init__(self,
              whitelist=None,
              alias=None,
              delete=None,
              explicit=None,
              patterns=None,
              accents=None,
              rules=None,
              normalized=None):
     self.whitelist = whitelist or load_whitelist()
     self.alias = alias or load_alias('alias.tsv')
     self.delete = delete or ['\u0361', '\u035c', '\u0301']
     self.explicit = explicit or load_alias('explicit.tsv')
     self.patterns = patterns or load_alias('patterns.tsv')
     self.accents = accents or "ˈˌ'"
     self.rules = rules or []
     self.normalized = normalized or load_normalized('normalized.tsv')
示例#6
0
    def test_check_string(self):
        from pyclpa.util import check_string, load_whitelist

        check = check_string('m a tt i s', load_whitelist())
        assert check[2] == '?'
示例#7
0
    def test_load_whitelist(self):
        from pyclpa.util import load_whitelist

        assert load_whitelist()['t']['ID'] == 'c118'