elasticsearch客户端_一线架构师，ElasticSearch-2.0.0集群安装配置与API使用实践下篇...

前言

本篇是承接上一篇的内容，因为头条发文有限制，不得不作为第二篇内容给大家介绍，希望大家能够理解和支持！！！

Java客户端

如果熟悉Java语言，而不想使用脚本等其他方式操作ElasticSearch搜索集群，则可以使用ElasticSearch提供的Java客户端API来编码实现，能够更加灵活地控制。ElasticSearch提供的Java客户端支持全部常用操作，如更新索引、索引文档、搜索文档、删除索引等等操作，而且还支持其他一些功能，如同步异步模式、explain查询等，下面我们通过代码来了解一下。如果使用Maven管理Java代码，可以在pom.xml文件中加入如下依赖：

     org.elasticsearch     elasticsearch     2.0.0

创建一个ElasticSearch客户端，代码如下所示：

// create & configure clientSettings settings = Settings.settingsBuilder()          .put("cluster.name", "dw_search_engine")          .put("client.transport.sniff", true)          .build();final Client client = TransportClient.builder().settings(settings).build()          .addTransportAddress(newAddress("es-01", 9300))          .addTransportAddress(newAddress("es-02", 9300));

可以将你的ElasticSearch集群的节点通过上面的addTransportAddress方法，都与Client对象关联起来，这样在操作ElasticSearch集群中的索引/更新/删除/搜索文档的时候，就能够自动感知。上面newAddress方法如下：

private static InetSocketTransportAddress newAddress(String host, int port) throws UnknownHostException {     return new InetSocketTransportAddress(InetAddress.getByName(host), port);}

另外，也可以通过在配置文件elasticsearch.yml中指定相关配置，例如：

cluster.name: dw_search_engineclient.transport.sniff: trueclient.transport.ping_timeout: 10sclient.transport.nodes_sampler_interval: 10s

那么，创建客户端需要从配置文件中读取配置内容，具体可以查看官方文档。

• 准备工作

索引的时候，我们是从一个本地文件中读取数据，并构建索引文档需要的格式，然后请求ElasticSearch集群执行索引操作，下面代码是一些基本准备工作：

final String index = "basis_device_info";final String type = "user"; // index documentsString f = "C:甥敳獲yanjunDesktopbasis_device_info.txt";File in = new File(f);

从文件中，每次读取一行记录，然后构建一个JSON格式字符串，通过XContentBuilder来表示，代码如下所示：

protected static XContentBuilder createSource(String[] a) throws IOException {     return jsonBuilder()             .startObject()                 .field("installid", a[0])                    .field("appid", a[1])                    .field("udid", a[2])                    .field("channel", a[3])                    .field("version", a[4])                    .field("osversion", a[5])                    .field("device_name", a[6])                    .field("producer", a[7])                    .field("device_type", a[8])                    .field("resolution", a[9])                    .field("screen_size", a[10])                    .field("mac", a[11])                    .field("idfa", a[12])                    .field("idfv", a[13])                    .field("imei", a[14])                    .field("create_time", a[15])             .endObject();}

下面我们从API的功能入手，分别详细说明，并附加代码展示用法。

• 创建索引

可以直接通过Java客户端库来创建索引，代码如下所示：

protected static void createIndex(final Client client, String index) {     Map indexSettings = Maps.newHashMap();     indexSettings.put("number_of_shards", "4");     indexSettings.put("number_of_replicas", "1");     CreateIndexRequest createIndexRequest = new CreateIndexRequest(               index, Settings.settingsBuilder().put(indexSettings).build());     CreateIndexResponse createIndexResponse = client.admin().indices().create(createIndexRequest).actionGet();     System.out.println(createIndexResponse);}

• 创建Mappings

通过Java客户端创建Mappings，相对比较复杂一点，需要拼接对应的JSON字符串，实现代码如下所示：

protected static void createMappings(final Client client, String index) throws IOException, InterruptedException, ExecutionException {     XContentBuilder basisInfoMapping = jsonBuilder()               .startObject()                    .startObject("_all")                         .field("enabled", "false")                    .endObject()                    .startObject("properties")                         .startObject("id")                              .field("type", "string")                         .endObject()                         .startObject("name")                              .field("type", "string")                              .field("index", "analyzed")                         .endObject()                         .startObject("age")                              .field("type", "int")                         .endObject()                         .startObject("birthday")                              .field("type", "date")                              .field("format", "yyyy-MM-dd HH:mm:ss")                              .field("index", "not_analyzed")                         .endObject()                    .endObject()               .endObject();          XContentBuilder deviceInfoMapping = jsonBuilder()               .startObject()                    .startObject("_all")                         .field("enabled", "false")                    .endObject()                    .startObject("properties")                         .startObject("udid")                              .field("type", "string")                         .endObject()                         .startObject("device_name")                              .field("type", "string")                              .field("index", "analyzed")                         .endObject()                         .startObject("privoder")                              .field("type", "string")                              .field("index", "analyzed")                         .endObject()                         .startObject("os_version")                              .field("type", "string")                         .endObject()                    .endObject()               .endObject();          PutMappingRequest putMappingRequest = Requests.putMappingRequest(index)          .type("basic_info")          .source(basisInfoMapping)         .type("device_info")         .source(deviceInfoMapping);          System.out.println(putMappingRequest.indicesOptions());          PutMappingResponse putMappingResponse = client.admin().indices().putMapping(putMappingRequest).get();     System.out.println(putMappingResponse);}

上面代码创建了一个名称为app_user_info的索引，该索引具有basic_info和device_info这2个type，可以通过elasticsearch_head插件，在Web管理页面上查看对应的索引信息。

• 索引单个文档

从文件中读取数据，一条记录构造一个文档，然后执行索引，代码如下所示：

protected static void indexDocs(final Client client, final String index, final String type, File in) {     BufferedReader reader = null;     try {          reader = new BufferedReader(new FileReader(in.getAbsoluteFile()));          String line = null;          while((line = reader.readLine()) != null) {               String[] a = line.split("", -1);               if(a.length == 16) {                    String udid = a[2];                    IndexResponse response =                              client                              .prepareIndex(index, type, udid)                              .setSource(createSource(a))                              .get();                    System.out.println(response.toString());               }          }               } catch (Exception e) {          throw Throwables.propagate(e);     } finally {          closeQuietly(reader);     }}

• 批量索引

批量索引有多种方式，首先，通过Bulk API进行索引，我们自己控制每一个batch的大小，代码如下所示：

protected static void indexBulk(final Client client, final String index, final String type, File in) {     BulkRequestBuilder bulkRequest = client.prepareBulk();     final int batchSize = 100;     int counter = 0;     BufferedReader reader = null;     try {          reader = new BufferedReader(new FileReader(in.getAbsoluteFile()));          String line = null;          while((line = reader.readLine()) != null) {               String[] a = line.split("", -1);               if(a.length == 16) {                    String udid = a[2];                    IndexRequestBuilder indexRequestBuilder =                              client                              .prepareIndex(index, type, udid)                              .setSource(createSource(a));                    bulkRequest.add(indexRequestBuilder);                    if(++counter >= batchSize) {                         System.out.println(!bulkRequest.get().hasFailures());                         counter = 0;                         bulkRequest = client.prepareBulk();                    }               }          }               } catch (Exception e) {          throw Throwables.propagate(e);     } finally {          System.out.println(!bulkRequest.get().hasFailures());          closeQuietly(reader);     }}

另一种方式，是根据ElasticSearch提供的Bulk Processor来实现，只需要设置相关参数，就可以实现批量索引，这种方式更加灵活，示例如下所示：

protected static void indexUsingBulkProcessor(final Client client, final String index, final String type, File in) throws InterruptedException {     String name = "device_info_processor";     int bulkActions = 1000;     ByteSizeValue bulkSize = new ByteSizeValue(100, ByteSizeUnit.MB);     TimeValue flushInterval = TimeValue.timeValueSeconds(60);     int concurrentRequests = 12;          // create bulk processor     final BulkProcessor bulkProcessor = BulkProcessor.builder(client, new BulkProcessor.Listener() {           public void afterBulk(long id, BulkRequest req, BulkResponse resp) {               System.out.println("id=" + id + ", resp=" + resp);          }           public void afterBulk(long id, BulkRequest req, Throwable cause) {               System.out.println("id=" + id + ", req=" + req + ", cause=" + cause);                       }           public void beforeBulk(long id, BulkRequest req) {               System.out.println("id=" + id + ", req=" + req);                       }               })     .setName(name)     .setBulkActions(bulkActions)     .setBulkSize(bulkSize)     .setFlushInterval(flushInterval)     .setConcurrentRequests(concurrentRequests)     .build();          // index documents     BufferedReader reader = null;     try {          reader = new BufferedReader(new FileReader(in.getAbsoluteFile()));          String line = null;          while((line = reader.readLine()) != null) {               String[] a = line.split("", -1);               if(a.length == 16) {                    String udid = a[2];                    bulkProcessor.add(new IndexRequest(index, type, udid).source(createSource(a)));               }          }               } catch (Exception e) {          throw Throwables.propagate(e);     } finally {          closeQuietly(reader);                    // close bulk processor          bulkProcessor.awaitClose(60, TimeUnit.SECONDS);     }}

可以通过实现自定义的BulkProcessor.Listener，它提供了Hook的功能，比如，索引某个文档失败的话，可以在Hook方法中增加处理，实现重试的功能；再比如，如果索引成功，给其他系统服务一个回调，等等。

• 更新文档

更新文档中的某些字段，需要指定id的值，以及需要更新的字段的值，代码如下所示：

protected static void updateDoc(final Client client, final String index, final String type) throws IOException, InterruptedException, ExecutionException {     String id = "60e90ddcb1a61622028b8d92112a646c";     UpdateRequest updateRequest = new UpdateRequest(index, type, id);     updateRequest.doc(jsonBuilder()               .startObject()                 .field("channel", "h-google")                 .field("appid", "1")             .endObject());     UpdateResponse response = client.update(updateRequest).get();     System.out.println(response);}

如果更新文档的时候，文档不存在，则需要先执行索引操作，再进行更新操作，将这两个操作合并到一起，使用upsert操作，代码如下所示：

protected static void upsertDoc(final Client client, final String index, final String type) throws IOException, InterruptedException, ExecutionException {     String id = "fdd5ff7f56b613f0acb2c20a1ebc35e4";     IndexRequest indexRequest = new IndexRequest(index, type, id).source(jsonBuilder()                 .startObject()                     .field("installid", "00000BSe")                     .field("appid", "0")                     .field("udid", "fdd5ff7f56b613f0acb2c20a1ebc35e4")                     .field("channel", "A-wandoujia")                     .field("version", "3.1.1")                     .field("resolution", "960*540")                     .field("mac", "00:08:22:be:1b:b7")                     .field("device_type", "0")                     .field("device_name", "HTC")                     .field("producer", "alps")                     .field("create_time", "2015-01-17 17:15:36")                 .endObject());          UpdateRequest updateRequest = new UpdateRequest(index, type, id).doc(jsonBuilder()                 .startObject()                     .field("resolution", "540*960")                     .field("channel", "h-baidu")                     .field("version", "3.1.1")                     .field("imei", "861622010000056")                 .endObject())             .upsert(indexRequest);                 UpdateResponse response = client.update(updateRequest).get();     System.out.println(response);}

• 删除文档

删除文档，需要指定文档的id的值，代码如下所示：

protected static void deleteDoc(final Client client, final String index, final String type) {     String id = "60e90ddcb1a61622028b8d92112a646c";     DeleteResponse response = client.prepareDelete(index, type, id).get();     System.out.println(response);}

• 搜索文档

搜索文档，可以根据需要构造指定的查询(Query)，可以设置过滤器等等，然后提交搜索，示例代码如下所示：

protected static void searchDocs(final Client client, final String index, final String type) {     SearchResponse response = client          .prepareSearch(index)          .setTypes(type)          .setQuery(QueryBuilders.termQuery("device_name", "xiaomi"))          .setPostFilter(QueryBuilders.rangeQuery("create_time").from("2015-01-16 00:00:00").to("2015-01-16 23:59:59"))          .setFrom(30).setSize(10).setExplain(true)          .execute()          .actionGet();     System.out.println(response);}

查询(Query)的构造有很多的方式，比如构造布尔查询，指定与、或、非关系，然后提交搜索。执行搜索，可以设置搜索文档的起始偏移位置以及每次取多少个结果文档，这便能实现分页功能。

(原创：时延军(包含链接：http://shiyanjun.cn)

可以转发关注小编，每天更新技术好文~~~感谢大家支持！！！！