UDF_Max/src/main/java/com/winhc/bigdata/udf/string/EfficaciousStr.java

package com.winhc.bigdata.udf.string;

import com.aliyun.odps.udf.UDF;
import com.aliyun.odps.utils.StringUtils;

import java.util.regex.Pattern;

/**
 * @author: XuJiakai
 * 2021/8/23 15:17
 */
public class EfficaciousStr extends UDF {
    private static final Pattern pattern = Pattern.compile("^[\\u4e00-\\u9fa50-9a-zA-Z ]+$");
    private static final Pattern num_pat = Pattern.compile("^[0-9 ]+$");
    private static final Pattern person_pat = Pattern.compile("^[\\u4e00-\\u9fa5a-zA-Z ]+$");
    private static final Pattern eng_pat = Pattern.compile("^[a-zA-Z ]+$");

    public static final String NAME = "公司名";
    public static final String PERSON = "人名";
    public static final String APP = "品牌";

    public Boolean evaluate(String label, String str) {
        if (StringUtils.isEmpty(str)) {
            return false;
        }
        if ((NAME.equals(label) || PERSON.equals(label)) && num_pat.matcher(str).matches()) {
            return false;
        }
        switch (label) {
            case NAME:
                return person_pat.matcher(str).matches();
            case PERSON:
                if(eng_pat.matcher(str).matches()){
                    return false;
                }
                return person_pat.matcher(str).matches();
            case APP:
                return pattern.matcher(str).matches();
            default:
                throw new RuntimeException("label is not define," + label);
        }
    }
}