func TokenizerConstructor(config map[string]interface{}, cache *registry.Cache) (analysis.Tokenizer, error) { remainingTokenizer, err := cache.TokenizerNamed(unicode.Name) if err != nil { return nil, err } return exception.NewExceptionsTokenizer(exceptionsRegexp, remainingTokenizer), nil }
func analyzeFn(cfg *Config) error { reExc, err := regexp.Compile(`(?i)c\+\+`) if err != nil { return err } uni := unicode.NewUnicodeTokenizer() tokenizer := exception.NewExceptionsTokenizer(reExc, uni) tokens := tokenizer.Tokenize([]byte(*analyzeArg)) for _, t := range tokens { fmt.Println(t) } return nil }