OtherDatasave/src/main/java/com/zyzs/otherdatasave/service/QueryPaper.java

package com.zyzs.otherdatasave.service;

import com.alibaba.fastjson.JSONObject;
import com.bfd.crawler.elasti.ElastiProducerHigh;
import com.bfd.crawler.utils.JsonUtils;
import com.zyzs.otherdatasave.bean.Clini;
import com.zyzs.otherdatasave.bean.Paper;
import com.zyzs.otherdatasave.cache.Constants;
import com.zyzs.otherdatasave.util.DataCheckUtil;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.springframework.stereotype.Service;

import javax.annotation.PostConstruct;
import java.time.LocalDateTime;
import java.time.format.DateTimeFormatter;
import java.util.*;

import static com.zyzs.otherdatasave.util.MfMD5Util.GetMD5Code;
@Service
public class QueryPaper {

    private static final Logger LOGGER = LoggerFactory.getLogger(QueryPaper.class);
    @PostConstruct
    public void init() {

    }

    public void query() {
        String inputMessage = Constants.getPaper().poll();// poll -->若队列为空，返回null
        if (Objects.isNull(inputMessage)) {
            return;
        }
        Map<String, Object> messageMap = new HashMap<>();
        if (inputMessage.length()>10)  {
            try  {
                try {
                    messageMap = JsonUtils.parseObject(inputMessage);
                } catch (Exception e) {
                    e.printStackTrace();
                }
                Paper paper=new Paper();
                paper.setTitle((String) messageMap.get("title"));
                paper.setCrawlUrl((String) messageMap.get("crawlUrl"));
                paper.setContent((String) messageMap.get("content"));

//
//                List<String> authorList = new ArrayList<>();
//                String author=(String) messageMap.get("author");
//                if(author.contains(";")){
//                    // 如果包含分号，则按分号分割
//                    authorList = Arrays.asList(author.split(";"));
//                }else if (author.contains(",")){
//                    // 如果包含分号，则按分号分割
//                    authorList = Arrays.asList(author.split(","));
//                }
//                if(authorList.size()>0){
//                  List authors=new ArrayList<>();
//                    for (String name:authorList){
//                        Map ma=new HashMap();
//                        ma.put("id","");
//                        ma.put("name",name);
//    //                  组织信息
//                        Map or =new HashMap();
//                        or.put("id","");
//                        or.put("name","");
//                        List organs=new ArrayList<>();
//                        organs.add(or);
//                        ma.put("organs",organs);
//                        authors.add(ma);
//                    }
//                    paper.setAuthors(authors);
//                }

                if (messageMap.containsKey("filePath")){
                    List<String> file = (List) messageMap.get("filePath");
                    List<String> cleanedList = new ArrayList<>();
                    for (String url : file) {
                        // 使用replaceAll方法移除匹配的<url>标签
                        String cleanedUrl = url.replaceAll("http://192.168.0.41:8081", "");
                        cleanedList.add(cleanedUrl);
                    }
                    paper.setFilePath(cleanedList);
                    paper.setFilePathSize((List) messageMap.get("filePathSize"));
                }

                List<Map> authors= (List<Map>) messageMap.get("authors");
                if(authors.size()>0){
                    List authorsall=new ArrayList<>();
                    for (Map<String,Object>keyValueMap : authors){
                        for (Map.Entry<String, Object> entry : keyValueMap.entrySet()) {
                            String key = entry.getKey(); // 获取键
                            Object value = entry.getValue(); // 获取值
                            Map ma=new HashMap();
                            ma.put("id","");
                            ma.put("name",key);
                            //                  组织信息
                            Map or =new HashMap();
                            or.put("id","");
                            or.put("name",value);
                            List organs=new ArrayList<>();
                            organs.add(or);
                            ma.put("organs",organs);
                            authorsall.add(ma);
                            paper.setAuthors(authorsall);
                        }
                    }
                }


    //来源信息
                Map source=new HashMap();
                source.put("id", "");
                source.put("name", messageMap.get("crawlUrl"));
                source.put("type", "");
                List sourcelist=new ArrayList<>();
                sourcelist.add(source);
                paper.setSource(sourcelist);

                String inputDate = (String) messageMap.get("pubDate");
                String outputFormat = "yyyy-MM-dd";
                // 定义输入格式
                DateTimeFormatter inputFormatter = DateTimeFormatter.ofPattern("yyyy-MM-dd HH:mm:ss");
                // 解析输入字符串为 LocalDateTime 对象
                LocalDateTime dateTime = LocalDateTime.parse(inputDate, inputFormatter);
                // 格式化为所需的输出格式
                String formattedDate = dateTime.format(DateTimeFormatter.ofPattern(outputFormat));
                paper.setPubDate(formattedDate);
                paper.setDoi((String) messageMap.get("doi"));
                paper.setClassify((String) messageMap.get("classify"));
                paper.setKeywords((String) messageMap.get("keywords"));
                paper.setSummary((String) messageMap.get("summary"));
                paper.setTopics((String) messageMap.get("topics"));
                paper.setFieldsSubject((String) messageMap.get("fieldsSubject"));
                paper.setReferences((String) messageMap.get("references"));
                String docid = GetMD5Code((String) messageMap.get("doi"));
                paper.setDocId(docid);
                paper.setDataId(docid);
                paper.set_id_(docid);
                paper.setCountry((String) messageMap.get("country"));

                paper.setTranslatetitle("");
                paper.setTranslatekeywords ("");
                paper.setTranslatesummary("");
                paper.setIsshow("20250520");
                long dateTimenow = System.currentTimeMillis();
                paper.setCreateTime(dateTimenow);
                String  createTimeStr=DataCheckUtil.getCurrentTime(dateTimenow);
                paper.setCreateTimeStr(createTimeStr);
                LOGGER.info("Parse Paper={}", JSONObject.toJSON(paper));
                ElastiProducerHigh elastiProducer = ElastiProducerHigh.getInstance(1, 3, "cl_special_1.0_paper_csci","_doc" );
                elastiProducer.sendMessageToEs(JsonUtils.toJSONString(paper));
            }
         catch (Exception e) {
            LOGGER.info("Parse PaperError={}", JSONObject.toJSON(inputMessage));
            e.printStackTrace();
         }
        }
    }
}