使用solrj进行DIH操作

背景说明：在一个项目中需要将Mongodb中的数据导入到solr中完成搜索。在solr中Mysql数据库有对应的DIH包，可以通过配置sql语句完成数据的导入。Mongodb下也有开源的工具用来实现数据的导入。看了下文档，感觉这个工具对数据的定制性不高，并且是python的，不是很满足项目需求。最后决定使用solrj来完成数据的导入。

一、 遇到的问题

1. 全量数据很大，在全量或者增量时无法一次性将数据全部获取： 对数据进行分页获取。（关于两种分页获取数据的性能问题，后面会单独介绍）

2. 全量在更新数据时，需要将之前的老数据clean掉，增量则不需要： clean其实就是删除所有数据。

3. 由于使用了分页获取数据，全量的clean操作必须是在全量开始之前完成，并且为了保证在做全量过程中，之前的老数据不会丢失，删除全部数据的操作对应的commit==false，且在整个全量过程中commit==false, 在最后完成全量后，再进行commit。

4.增量操作和全量操作是通过配置不同的trigger完成的，比如增量每隔五分钟执行一次，全量则一天执行一次。如果没有对任务进行控制，可能会造成 全量和增量同时在做。刚才说了，全量操作的整个过程，commit==false, 所以对增量和全量的任务必须加互斥锁。

二、相关的实现

package com.meizu.galaxy2.solr;

import org.apache.log4j.Logger;
import org.apache.solr.client.solrj.SolrQuery;
import org.apache.solr.client.solrj.SolrServerException;
import org.apache.solr.client.solrj.impl.CloudSolrClient;
import org.apache.solr.client.solrj.request.UpdateRequest;
import org.apache.solr.common.SolrInputDocument;
import org.apache.solr.common.params.ModifiableSolrParams;

import java.io.IOException;
import java.lang.reflect.Field;
import java.util.List;

/**
 * Created by ltao on 2015-7-16.
 */
public class CloudSolrDIHClient {

    private static final Logger logger = Logger.getLogger(CloudSolrDIHClient.class);
    private static final String ID_FILED_NAME = "id";
    private CloudSolrClient solrClient;

    private static final int BATCH_SIZE=500;

    private String defaultCollection;


    public CloudSolrDIHClient(String zkHost, String zkNodePath, int zkClientTimeout, int zkConnectTimeout, String defaultCollection) {

        if (!zkHost.startsWith("zookeeper")) {
            logger.error("zk host must start with zookeeper://");
            return;
        } else {

            String hosts = zkHost.substring(12);
            hosts = hosts + zkNodePath;
            solrClient = new org.apache.solr.client.solrj.impl.CloudSolrClient(hosts);
            solrClient.setZkClientTimeout(zkClientTimeout);
            solrClient.setZkConnectTimeout(zkConnectTimeout);
            this.defaultCollection = defaultCollection;
        }
    }


    public void connect() throws Exception {
        solrClient.connect();
    }

    public void addDoc(SolrInputDocument doc) {
        if (this.defaultCollection != null) {
            this.addDoc(defaultCollection, doc);
        } else {
            logger.error("default collection should not be null");
        }

    }

    public void addDoc(String collection, SolrInputDocument doc) {
        try {
            solrClient.add(collection, doc);
        } catch (Exception e) {
            logger.error("add Doc occurs an error,collection:" + collection + ",doc_id:" + doc.getFieldValue(ID_FILED_NAME), e);
        }
    }


    public void addDocs(List<SolrInputDocument> docs) {
        if (this.defaultCollection != null) {
            this.addDocs(defaultCollection, docs);
        } else {
            logger.error("default collection should not be null");
        }
    }


    public void addDocs(String collection, List<SolrInputDocument> docs) {
        if(docs!=null && docs.size()>0) {
            int size=docs.size();
            if(size<=BATCH_SIZE) {
                try {
                    solrClient.add(collection, docs);
                } catch (Exception e) {
                    logger.error("add Docs occurs an error,collection:" + collection, e);
                }
            }
            else
            {
                int end=size>BATCH_SIZE? BATCH_SIZE:size;
                int start=0;
                while(true)
                {
                    List<SolrInputDocument> subList=docs.subList(start,end);
                    try {
                        solrClient.add(collection, subList);
                    } catch (Exception e) {
                        logger.error("add Docs occurs an error,collection:" + collection, e);
                    }
                    if(end==size)
                    {
                        break;
                    }
                    start=start+BATCH_SIZE;
                    end=(end+BATCH_SIZE);
                   if(end>size)
                   {
                       end=size;
                   }
                }

            }
        }

    }

    public void deleteDocByIds(List<String> ids) {
        if (this.defaultCollection != null) {
            this.deleteDocByIds(defaultCollection, ids);
        }
    }

    public void deleteDocByIds(String collection, List<String> ids) {

        try {
            solrClient.deleteById(collection,ids);
        } catch (Exception e) {
            logger.error("delete Docs occurs an error,collection:" + collection ,e);
        }

    }


    public void deleteDocById(String collection, String id) {
        try {
            solrClient.deleteById(collection, id);
        } catch (Exception e) {
            logger.error("delete Doc occurs an error,collection:" + collection + ",doc_id:" + id, e);
        }
    }

    public void deleteDocById(String id) {
        if (this.defaultCollection != null) {
            this.deleteDocById(defaultCollection, id);

        } else {
            logger.error("default collection should not be null");
        }
    }


    public void addBean(String collection, Object obj) {
        try {
            solrClient.addBean(collection, obj);
        } catch (Exception e) {
            String id = null;
            try {
                Field idFiled = obj.getClass().getDeclaredField(ID_FILED_NAME);
                if (idFiled != null) {
                    idFiled.setAccessible(true);
                    Object idFiledValue = idFiled.get(obj);
                    id = idFiledValue != null ? idFiledValue.toString() : "";
                }
            } catch (Exception e1) {
                logger.error("get id field occurs an error", e1);
            }
            logger.error("add bean occurs an error,collection:" + collection + ",bean_id:" + id, e);
        }
    }

    public void addBean(Object obj) throws SolrServerException, IOException {
        if (this.defaultCollection != null) {
            this.addBean(defaultCollection, obj);
        } else {
            logger.error("default collection should not be null");
        }

    }

    public void addBeans(List<Object> objs) throws SolrServerException, IOException {
        if (this.defaultCollection != null) {
            this.addBean(defaultCollection, objs);
        } else {
            logger.error("default collection should not be null");
        }
    }


    public void addBeans(String collection, List<Object> objs) {
        if(objs!=null && objs.size()>0) {
            int size=objs.size();
            if(size<=BATCH_SIZE) {
                try {
                    solrClient.addBeans(collection, objs);
                } catch (Exception e) {
                    logger.error("addBeans occurs an error,collection:" + collection, e);
                }
            }
            else
            {
                int end=size>BATCH_SIZE? BATCH_SIZE:size;
                int start=0;
                while(true)
                {
                    List<Object> subList=objs.subList(start,end);
                    try {
                        solrClient.addBeans(collection, subList);
                    } catch (Exception e) {
                        logger.error("addBeans occurs an error,collection:" + collection, e);
                    }
                    if(end==size)
                    {
                        break;
                    }
                    start=start+BATCH_SIZE;
                    end=(end+BATCH_SIZE);
                    if(end>size)
                    {
                        end=size;
                    }
                }

            }
        }
    }


    public void commit() throws SolrServerException, IOException {
        this.commit(defaultCollection);
    }

    public void commit(String collection) throws SolrServerException, IOException {
        solrClient.commit(collection);
    }

    public void clean(Boolean clean) throws SolrServerException, IOException {
        this.clean(defaultCollection, clean);
    }

    public void clean(String collection, Boolean clean) throws SolrServerException, IOException {
        UpdateRequest updateRequest = new UpdateRequest();
        updateRequest.setParam("stream.body", "<delete><query>*:*</query></delete>");
        updateRequest.setParam("commit", Boolean.toString(clean));
        solrClient.request(updateRequest, collection);
    }


    public void close() throws IOException {
        if (solrClient != null) {
            solrClient.close();
        }
    }


}

上面是对solrclient的简单封装：增加、删除、clean

全量和增量进行加锁互斥

  private static Lock lock = new ReentrantLock();

    public void importDelta() {
        boolean hasGetLock = false;
        try {
            hasGetLock = lock.tryLock();
            if (hasGetLock) {
                logger.info("start import delta hotel data ");
                long start = System.currentTimeMillis();
                hotelService.importDelta();
                long end = System.currentTimeMillis();
                logger.info("finish import delta hotel data ,spend " + (end - start) + " ms");
            }
        } finally {
            if (hasGetLock) {
                lock.unlock();
            }
        }

    }

    public void importAll() {

        try {
            lock.lock();
            logger.info("start import all hotel data ");
            long start = System.currentTimeMillis();
            hotelService.importAll();
            long end = System.currentTimeMillis();
            logger.info("finish import all hotel data ,spend " + (end - start) + " ms");
        } finally {
            lock.unlock();
        }
    }

    public DataImportService getHotelService() {
        return hotelService;
    }

这里用了Lock的tryLock,tryLock()会尝试获取锁，如果当前锁已被使用，则放弃该次获取锁操作。lock()则会阻塞，直到获取到锁。这样可以较大概率的保证全量一定能够执行。(如果增量一直都在运行,可能会造成全量一直阻塞，在实际运行中不会遇到这种情况;或者在某种机缘巧合下，增量一个接一个的获取到了锁，全量则一直阻塞，个人觉得应该可以使用公平锁解决刚才的这个问题，不过其实没必要)。