Azure Cosmos DB 是 Azure 提供的完全托管的 NoSQL 数据库服务。 它允许轻松构建多区域分布式且高度可缩放的应用程序。 此指南将指导你完成创建 Java 应用程序的过程，该应用程序使用 Azure Cosmos DB for NoSQL 数据库，并实现更改源提取器以进行实时数据处理。 Java 应用程序使用 Azure Cosmos DB Java SDK v4 与 Azure Cosmos DB for NoSQL 通信。
重要
本教程仅适用于 Azure Cosmos DB Java SDK v4。 有关详细信息，请查看 Azure Cosmos DB Java SDK v4 发行说明、 Maven 存储库、 Azure Cosmos DB 中的更改源处理器和 Azure Cosmos DB Java SDK v4 故障排除指南 。 如果你当前使用的是早于 v4 的版本，请参阅迁移到 Azure Cosmos DB Java SDK v4 指南，获取升级到 v4 的相关帮助。
先决条件
Azure Cosmos DB 帐户：也可以从 Azure 门户 创建它，也可以使用 Azure Cosmos DB 模拟器 。
Java 开发环境：确保计算机上安装了至少 8 个版本的 Java 开发工具包（JDK）。
Azure Cosmos DB Java SDK V4：提供与 Azure Cosmos DB 交互所需的功能。
背景
Azure Cosmos DB 更改源提供事件驱动的接口，用于触发对文档插入进行响应的操作，并且具有多种用途。
管理更改源事件的工作在很大程度上由 SDK 中内置的更改源处理器库负责。 如果需要，该库足够强大，可以在多个工作节点之间分发变更馈送事件。 只需提供更改源库回调即可。
此 Java 应用程序的简单示例演示了使用 Azure Cosmos DB 和更改源处理器进行实时数据处理。 应用程序将示例文档插入“源容器”以模拟数据流。 “更改源处理器”绑定到源容器，处理传入的更改并记录文档内容。 处理器会自动管理并行处理的租约。
源代码
可以克隆 SDK 示例存储库，并在以下位置 SampleChangeFeedProcessor.java找到此示例：
git clone https://github.com/Azure-Samples/azure-cosmos-java-sql-api-samples.git
cd azure-cosmos-java-sql-api-sample/src/main/java/com/azure/cosmos/examples/changefeed/
Walkthrough
- 在 Java 应用程序中配置
ChangeFeedProcessorOptions，使用 Azure Cosmos DB 和 Azure Cosmos DB Java SDK V4。ChangeFeedProcessorOptions提供必要设置以控制更改馈送处理器在数据处理期间的行为。
// Copyright (c) Microsoft Corporation. All rights reserved.
// Licensed under the MIT License.
package com.azure.cosmos.examples.changefeed;
import com.azure.cosmos.ChangeFeedProcessor;
import com.azure.cosmos.ChangeFeedProcessorBuilder;
import com.azure.cosmos.ConsistencyLevel;
import com.azure.cosmos.CosmosAsyncClient;
import com.azure.cosmos.CosmosAsyncContainer;
import com.azure.cosmos.CosmosAsyncDatabase;
import com.azure.cosmos.CosmosClientBuilder;
import com.azure.cosmos.CosmosException;
import com.azure.cosmos.examples.common.AccountSettings;
import com.azure.cosmos.examples.common.CustomPOJO2;
import com.azure.cosmos.implementation.Utils;
import com.azure.cosmos.models.ChangeFeedProcessorOptions;
import com.azure.cosmos.models.CosmosContainerProperties;
import com.azure.cosmos.models.CosmosContainerRequestOptions;
import com.azure.cosmos.models.CosmosContainerResponse;
import com.azure.cosmos.models.CosmosDatabaseResponse;
import com.azure.cosmos.models.ThroughputProperties;
import com.fasterxml.jackson.core.JsonProcessingException;
import com.fasterxml.jackson.databind.JsonNode;
import com.fasterxml.jackson.databind.ObjectMapper;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import reactor.core.scheduler.Schedulers;
import java.time.Duration;
import java.util.List;
import java.util.UUID;
import java.util.function.Consumer;
/**
* Sample for Change Feed Processor.
* This sample models an application where documents are being inserted into one container (the "feed container"),
* and meanwhile another worker thread or worker application is pulling inserted documents from the feed container's Change Feed
* and operating on them in some way. For one or more workers to process the Change Feed of a container, the workers must first contact the server
* and "lease" access to monitor one or more partitions of the feed container. The Change Feed Processor Library
* handles leasing automatically for you, however you must create a separate "lease container" where the Change Feed
* Processor Library can store and track leases container partitions.
*/
public class SampleChangeFeedProcessor {
public static int WAIT_FOR_WORK = 60000;
public static final String DATABASE_NAME = "db_" + UUID.randomUUID();
public static final String COLLECTION_NAME = "coll_" + UUID.randomUUID();
private static final ObjectMapper OBJECT_MAPPER = Utils.getSimpleObjectMapper();
protected static Logger logger = LoggerFactory.getLogger(SampleChangeFeedProcessor.class);
private static boolean isWorkCompleted = false;
private static ChangeFeedProcessorOptions options;
public static void main(String[] args) {
logger.info("Begin Sample");
try {
// <ChangeFeedProcessorOptions>
options = new ChangeFeedProcessorOptions();
options.setStartFromBeginning(false);
options.setLeasePrefix("myChangeFeedDeploymentUnit");
// </ChangeFeedProcessorOptions>
//Summary of the next four commands:
//-Create an asynchronous Azure Cosmos DB client and database so that we can issue async requests to the DB
//-Create a "feed container" and a "lease container" in the DB
logger.info("Create CosmosClient");
CosmosAsyncClient client = getCosmosClient();
logger.info("Create sample's database: " + DATABASE_NAME);
CosmosAsyncDatabase cosmosDatabase = createNewDatabase(client, DATABASE_NAME);
logger.info("Create container for documents: " + COLLECTION_NAME);
CosmosAsyncContainer feedContainer = createNewCollection(client, DATABASE_NAME, COLLECTION_NAME);
logger.info("Create container for lease: " + COLLECTION_NAME + "-leases");
CosmosAsyncContainer leaseContainer = createNewLeaseCollection(client, DATABASE_NAME, COLLECTION_NAME + "-leases");
//Model of a worker thread or application which leases access to monitor one or more feed container
//partitions via the Change Feed. In a real-world application you might deploy this code in an Azure function.
//The next line causes the worker to create and start an instance of the Change Feed Processor. See the implementation of getChangeFeedProcessor() for guidance
//on creating a handler for Change Feed events. In this stream, we also trigger the insertion of 10 documents on a separate
//thread.
// <StartChangeFeedProcessor>
logger.info("Start Change Feed Processor on worker (handles changes asynchronously)");
ChangeFeedProcessor changeFeedProcessorInstance = new ChangeFeedProcessorBuilder()
.hostName("SampleHost_1")
.feedContainer(feedContainer)
.leaseContainer(leaseContainer)
.handleChanges(handleChanges())
.buildChangeFeedProcessor();
changeFeedProcessorInstance.start()
.subscribeOn(Schedulers.boundedElastic())
.subscribe();
// </StartChangeFeedProcessor>
//These two lines model an application which is inserting ten documents into the feed container
logger.info("Start application that inserts documents into feed container");
createNewDocumentsCustomPOJO(feedContainer, 10, Duration.ofSeconds(3));
//This loop models the Worker main loop, which spins while its Change Feed Processor instance asynchronously
//handles incoming Change Feed events from the feed container. Of course in this sample, polling
//isWorkCompleted is unnecessary because items are being added to the feed container on the same thread, and you
//can see just above isWorkCompleted is set to true.
//But conceptually the worker is part of a different thread or application than the one which is inserting
//into the feed container; so this code illustrates the worker waiting and listening for changes to the feed container
long remainingWork = WAIT_FOR_WORK;
while (!isWorkCompleted && remainingWork > 0) {
Thread.sleep(100);
remainingWork -= 100;
}
//When all documents have been processed, clean up
if (isWorkCompleted) {
changeFeedProcessorInstance.stop().subscribe();
} else {
throw new RuntimeException("The change feed processor initialization and automatic create document feeding process did not complete in the expected time");
}
logger.info("Delete sample's database: " + DATABASE_NAME);
//deleteDatabase(cosmosDatabase);
Thread.sleep(500);
} catch (Exception e) {
e.printStackTrace();
}
logger.info("End Sample");
}
// <Delegate>
private static Consumer<List<JsonNode>> handleChanges() {
return (List<JsonNode> docs) -> {
logger.info("Start handleChanges()");
for (JsonNode document : docs) {
try {
//Change Feed hands the document to you in the form of a JsonNode
//As a developer you have two options for handling the JsonNode document provided to you by Change Feed
//One option is to operate on the document in the form of a JsonNode, as shown below. This is great
//especially if you do not have a single uniform data model for all documents.
logger.info("Document received: " + OBJECT_MAPPER.writerWithDefaultPrettyPrinter()
.writeValueAsString(document));
//You can also transform the JsonNode to a POJO having the same structure as the JsonNode,
//as shown below. Then you can operate on the POJO.
CustomPOJO2 pojo_doc = OBJECT_MAPPER.treeToValue(document, CustomPOJO2.class);
logger.info("id: " + pojo_doc.getId());
} catch (JsonProcessingException e) {
e.printStackTrace();
}
}
isWorkCompleted = true;
logger.info("End handleChanges()");
};
}
// </Delegate>
public static CosmosAsyncClient getCosmosClient() {
return new CosmosClientBuilder()
.endpoint(AccountSettings.HOST)
.key(AccountSettings.MASTER_KEY)
.contentResponseOnWriteEnabled(true)
.consistencyLevel(ConsistencyLevel.SESSION)
.buildAsyncClient();
}
public static CosmosAsyncDatabase createNewDatabase(CosmosAsyncClient client, String databaseName) {
CosmosDatabaseResponse databaseResponse = client.createDatabaseIfNotExists(databaseName).block();
return client.getDatabase(databaseResponse.getProperties().getId());
}
public static void deleteDatabase(CosmosAsyncDatabase cosmosDatabase) {
cosmosDatabase.delete().block();
}
public static CosmosAsyncContainer createNewCollection(CosmosAsyncClient client, String databaseName, String collectionName) {
CosmosAsyncDatabase databaseLink = client.getDatabase(databaseName);
CosmosAsyncContainer collectionLink = databaseLink.getContainer(collectionName);
CosmosContainerResponse containerResponse = null;
try {
containerResponse = collectionLink.read().block();
if (containerResponse != null) {
throw new IllegalArgumentException(String.format("Collection %s already exists in database %s.", collectionName, databaseName));
}
} catch (RuntimeException ex) {
if (ex instanceof CosmosException) {
CosmosException CosmosException = (CosmosException) ex;
if (CosmosException.getStatusCode() != 404) {
throw ex;
}
} else {
throw ex;
}
}
CosmosContainerProperties containerSettings = new CosmosContainerProperties(collectionName, "/pk");
CosmosContainerRequestOptions requestOptions = new CosmosContainerRequestOptions();
ThroughputProperties throughputProperties = ThroughputProperties.createManualThroughput(10000);
containerResponse = databaseLink.createContainer(containerSettings, throughputProperties, requestOptions).block();
if (containerResponse == null) {
throw new RuntimeException(String.format("Failed to create collection %s in database %s.", collectionName, databaseName));
}
return databaseLink.getContainer(containerResponse.getProperties().getId());
}
public static CosmosAsyncContainer createNewLeaseCollection(CosmosAsyncClient client, String databaseName, String leaseCollectionName) {
CosmosAsyncDatabase databaseLink = client.getDatabase(databaseName);
CosmosAsyncContainer leaseCollectionLink = databaseLink.getContainer(leaseCollectionName);
CosmosContainerResponse leaseContainerResponse = null;
try {
leaseContainerResponse = leaseCollectionLink.read().block();
if (leaseContainerResponse != null) {
leaseCollectionLink.delete().block();
try {
Thread.sleep(1000);
} catch (InterruptedException ex) {
ex.printStackTrace();
}
}
} catch (RuntimeException ex) {
if (ex instanceof CosmosException) {
CosmosException CosmosException = (CosmosException) ex;
if (CosmosException.getStatusCode() != 404) {
throw ex;
}
} else {
throw ex;
}
}
CosmosContainerProperties containerSettings = new CosmosContainerProperties(leaseCollectionName, "/id");
CosmosContainerRequestOptions requestOptions = new CosmosContainerRequestOptions();
ThroughputProperties throughputProperties = ThroughputProperties.createManualThroughput(400);
leaseContainerResponse = databaseLink.createContainer(containerSettings, throughputProperties, requestOptions).block();
if (leaseContainerResponse == null) {
throw new RuntimeException(String.format("Failed to create collection %s in database %s.", leaseCollectionName, databaseName));
}
return databaseLink.getContainer(leaseContainerResponse.getProperties().getId());
}
public static void createNewDocumentsCustomPOJO(CosmosAsyncContainer containerClient, int count, Duration delay) {
String suffix = UUID.randomUUID().toString();
for (int i = 0; i <= count; i++) {
CustomPOJO2 document = new CustomPOJO2();
document.setId(String.format("0%d-%s", i, suffix));
document.setPk(document.getId()); // This is a very simple example, so we'll just have a partition key (/pk) field that we set equal to id
containerClient.createItem(document).subscribe(doc -> {
logger.info("Document write: " + doc);
});
long remainingWork = delay.toMillis();
try {
while (remainingWork > 0) {
Thread.sleep(100);
remainingWork -= 100;
}
} catch (InterruptedException iex) {
// exception caught
break;
}
}
}
}
- 使用相关配置进行初始化
ChangeFeedProcessor，包括主机名、源容器、租用容器和数据处理逻辑。 该方法start()启动数据处理，启用来自源容器的传入数据更改的并发和实时处理。
// Copyright (c) Microsoft Corporation. All rights reserved.
// Licensed under the MIT License.
package com.azure.cosmos.examples.changefeed;
import com.azure.cosmos.ChangeFeedProcessor;
import com.azure.cosmos.ChangeFeedProcessorBuilder;
import com.azure.cosmos.ConsistencyLevel;
import com.azure.cosmos.CosmosAsyncClient;
import com.azure.cosmos.CosmosAsyncContainer;
import com.azure.cosmos.CosmosAsyncDatabase;
import com.azure.cosmos.CosmosClientBuilder;
import com.azure.cosmos.CosmosException;
import com.azure.cosmos.examples.common.AccountSettings;
import com.azure.cosmos.examples.common.CustomPOJO2;
import com.azure.cosmos.implementation.Utils;
import com.azure.cosmos.models.ChangeFeedProcessorOptions;
import com.azure.cosmos.models.CosmosContainerProperties;
import com.azure.cosmos.models.CosmosContainerRequestOptions;
import com.azure.cosmos.models.CosmosContainerResponse;
import com.azure.cosmos.models.CosmosDatabaseResponse;
import com.azure.cosmos.models.ThroughputProperties;
import com.fasterxml.jackson.core.JsonProcessingException;
import com.fasterxml.jackson.databind.JsonNode;
import com.fasterxml.jackson.databind.ObjectMapper;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import reactor.core.scheduler.Schedulers;
import java.time.Duration;
import java.util.List;
import java.util.UUID;
import java.util.function.Consumer;
/**
* Sample for Change Feed Processor.
* This sample models an application where documents are being inserted into one container (the "feed container"),
* and meanwhile another worker thread or worker application is pulling inserted documents from the feed container's Change Feed
* and operating on them in some way. For one or more workers to process the Change Feed of a container, the workers must first contact the server
* and "lease" access to monitor one or more partitions of the feed container. The Change Feed Processor Library
* handles leasing automatically for you, however you must create a separate "lease container" where the Change Feed
* Processor Library can store and track leases container partitions.
*/
public class SampleChangeFeedProcessor {
public static int WAIT_FOR_WORK = 60000;
public static final String DATABASE_NAME = "db_" + UUID.randomUUID();
public static final String COLLECTION_NAME = "coll_" + UUID.randomUUID();
private static final ObjectMapper OBJECT_MAPPER = Utils.getSimpleObjectMapper();
protected static Logger logger = LoggerFactory.getLogger(SampleChangeFeedProcessor.class);
private static boolean isWorkCompleted = false;
private static ChangeFeedProcessorOptions options;
public static void main(String[] args) {
logger.info("Begin Sample");
try {
// <ChangeFeedProcessorOptions>
options = new ChangeFeedProcessorOptions();
options.setStartFromBeginning(false);
options.setLeasePrefix("myChangeFeedDeploymentUnit");
// </ChangeFeedProcessorOptions>
//Summary of the next four commands:
//-Create an asynchronous Azure Cosmos DB client and database so that we can issue async requests to the DB
//-Create a "feed container" and a "lease container" in the DB
logger.info("Create CosmosClient");
CosmosAsyncClient client = getCosmosClient();
logger.info("Create sample's database: " + DATABASE_NAME);
CosmosAsyncDatabase cosmosDatabase = createNewDatabase(client, DATABASE_NAME);
logger.info("Create container for documents: " + COLLECTION_NAME);
CosmosAsyncContainer feedContainer = createNewCollection(client, DATABASE_NAME, COLLECTION_NAME);
logger.info("Create container for lease: " + COLLECTION_NAME + "-leases");
CosmosAsyncContainer leaseContainer = createNewLeaseCollection(client, DATABASE_NAME, COLLECTION_NAME + "-leases");
//Model of a worker thread or application which leases access to monitor one or more feed container
//partitions via the Change Feed. In a real-world application you might deploy this code in an Azure function.
//The next line causes the worker to create and start an instance of the Change Feed Processor. See the implementation of getChangeFeedProcessor() for guidance
//on creating a handler for Change Feed events. In this stream, we also trigger the insertion of 10 documents on a separate
//thread.
// <StartChangeFeedProcessor>
logger.info("Start Change Feed Processor on worker (handles changes asynchronously)");
ChangeFeedProcessor changeFeedProcessorInstance = new ChangeFeedProcessorBuilder()
.hostName("SampleHost_1")
.feedContainer(feedContainer)
.leaseContainer(leaseContainer)
.handleChanges(handleChanges())
.buildChangeFeedProcessor();
changeFeedProcessorInstance.start()
.subscribeOn(Schedulers.boundedElastic())
.subscribe();
// </StartChangeFeedProcessor>
//These two lines model an application which is inserting ten documents into the feed container
logger.info("Start application that inserts documents into feed container");
createNewDocumentsCustomPOJO(feedContainer, 10, Duration.ofSeconds(3));
//This loop models the Worker main loop, which spins while its Change Feed Processor instance asynchronously
//handles incoming Change Feed events from the feed container. Of course in this sample, polling
//isWorkCompleted is unnecessary because items are being added to the feed container on the same thread, and you
//can see just above isWorkCompleted is set to true.
//But conceptually the worker is part of a different thread or application than the one which is inserting
//into the feed container; so this code illustrates the worker waiting and listening for changes to the feed container
long remainingWork = WAIT_FOR_WORK;
while (!isWorkCompleted && remainingWork > 0) {
Thread.sleep(100);
remainingWork -= 100;
}
//When all documents have been processed, clean up
if (isWorkCompleted) {
changeFeedProcessorInstance.stop().subscribe();
} else {
throw new RuntimeException("The change feed processor initialization and automatic create document feeding process did not complete in the expected time");
}
logger.info("Delete sample's database: " + DATABASE_NAME);
//deleteDatabase(cosmosDatabase);
Thread.sleep(500);
} catch (Exception e) {
e.printStackTrace();
}
logger.info("End Sample");
}
// <Delegate>
private static Consumer<List<JsonNode>> handleChanges() {
return (List<JsonNode> docs) -> {
logger.info("Start handleChanges()");
for (JsonNode document : docs) {
try {
//Change Feed hands the document to you in the form of a JsonNode
//As a developer you have two options for handling the JsonNode document provided to you by Change Feed
//One option is to operate on the document in the form of a JsonNode, as shown below. This is great
//especially if you do not have a single uniform data model for all documents.
logger.info("Document received: " + OBJECT_MAPPER.writerWithDefaultPrettyPrinter()
.writeValueAsString(document));
//You can also transform the JsonNode to a POJO having the same structure as the JsonNode,
//as shown below. Then you can operate on the POJO.
CustomPOJO2 pojo_doc = OBJECT_MAPPER.treeToValue(document, CustomPOJO2.class);
logger.info("id: " + pojo_doc.getId());
} catch (JsonProcessingException e) {
e.printStackTrace();
}
}
isWorkCompleted = true;
logger.info("End handleChanges()");
};
}
// </Delegate>
public static CosmosAsyncClient getCosmosClient() {
return new CosmosClientBuilder()
.endpoint(AccountSettings.HOST)
.key(AccountSettings.MASTER_KEY)
.contentResponseOnWriteEnabled(true)
.consistencyLevel(ConsistencyLevel.SESSION)
.buildAsyncClient();
}
public static CosmosAsyncDatabase createNewDatabase(CosmosAsyncClient client, String databaseName) {
CosmosDatabaseResponse databaseResponse = client.createDatabaseIfNotExists(databaseName).block();
return client.getDatabase(databaseResponse.getProperties().getId());
}
public static void deleteDatabase(CosmosAsyncDatabase cosmosDatabase) {
cosmosDatabase.delete().block();
}
public static CosmosAsyncContainer createNewCollection(CosmosAsyncClient client, String databaseName, String collectionName) {
CosmosAsyncDatabase databaseLink = client.getDatabase(databaseName);
CosmosAsyncContainer collectionLink = databaseLink.getContainer(collectionName);
CosmosContainerResponse containerResponse = null;
try {
containerResponse = collectionLink.read().block();
if (containerResponse != null) {
throw new IllegalArgumentException(String.format("Collection %s already exists in database %s.", collectionName, databaseName));
}
} catch (RuntimeException ex) {
if (ex instanceof CosmosException) {
CosmosException CosmosException = (CosmosException) ex;
if (CosmosException.getStatusCode() != 404) {
throw ex;
}
} else {
throw ex;
}
}
CosmosContainerProperties containerSettings = new CosmosContainerProperties(collectionName, "/pk");
CosmosContainerRequestOptions requestOptions = new CosmosContainerRequestOptions();
ThroughputProperties throughputProperties = ThroughputProperties.createManualThroughput(10000);
containerResponse = databaseLink.createContainer(containerSettings, throughputProperties, requestOptions).block();
if (containerResponse == null) {
throw new RuntimeException(String.format("Failed to create collection %s in database %s.", collectionName, databaseName));
}
return databaseLink.getContainer(containerResponse.getProperties().getId());
}
public static CosmosAsyncContainer createNewLeaseCollection(CosmosAsyncClient client, String databaseName, String leaseCollectionName) {
CosmosAsyncDatabase databaseLink = client.getDatabase(databaseName);
CosmosAsyncContainer leaseCollectionLink = databaseLink.getContainer(leaseCollectionName);
CosmosContainerResponse leaseContainerResponse = null;
try {
leaseContainerResponse = leaseCollectionLink.read().block();
if (leaseContainerResponse != null) {
leaseCollectionLink.delete().block();
try {
Thread.sleep(1000);
} catch (InterruptedException ex) {
ex.printStackTrace();
}
}
} catch (RuntimeException ex) {
if (ex instanceof CosmosException) {
CosmosException CosmosException = (CosmosException) ex;
if (CosmosException.getStatusCode() != 404) {
throw ex;
}
} else {
throw ex;
}
}
CosmosContainerProperties containerSettings = new CosmosContainerProperties(leaseCollectionName, "/id");
CosmosContainerRequestOptions requestOptions = new CosmosContainerRequestOptions();
ThroughputProperties throughputProperties = ThroughputProperties.createManualThroughput(400);
leaseContainerResponse = databaseLink.createContainer(containerSettings, throughputProperties, requestOptions).block();
if (leaseContainerResponse == null) {
throw new RuntimeException(String.format("Failed to create collection %s in database %s.", leaseCollectionName, databaseName));
}
return databaseLink.getContainer(leaseContainerResponse.getProperties().getId());
}
public static void createNewDocumentsCustomPOJO(CosmosAsyncContainer containerClient, int count, Duration delay) {
String suffix = UUID.randomUUID().toString();
for (int i = 0; i <= count; i++) {
CustomPOJO2 document = new CustomPOJO2();
document.setId(String.format("0%d-%s", i, suffix));
document.setPk(document.getId()); // This is a very simple example, so we'll just have a partition key (/pk) field that we set equal to id
containerClient.createItem(document).subscribe(doc -> {
logger.info("Document write: " + doc);
});
long remainingWork = delay.toMillis();
try {
while (remainingWork > 0) {
Thread.sleep(100);
remainingWork -= 100;
}
} catch (InterruptedException iex) {
// exception caught
break;
}
}
}
}
- 指定委托通过
handleChanges()方法处理传入的数据变更。 该方法处理从 Change Feed 收到的 JsonNode 文档。 作为开发人员，您有两个选项来处理由更改提要提供的 JsonNode 文档。 一个选项是以 JsonNode 的形式对文档进行操作。 这非常出色，尤其是在没有针对所有文档的单个统一数据模型的情况下。 第二个选项 - 将 JsonNode 转换为与 JsonNode 具有相同结构的 POJO。 然后，可以对POJO进行操作。
// Copyright (c) Microsoft Corporation. All rights reserved.
// Licensed under the MIT License.
package com.azure.cosmos.examples.changefeed;
import com.azure.cosmos.ChangeFeedProcessor;
import com.azure.cosmos.ChangeFeedProcessorBuilder;
import com.azure.cosmos.ConsistencyLevel;
import com.azure.cosmos.CosmosAsyncClient;
import com.azure.cosmos.CosmosAsyncContainer;
import com.azure.cosmos.CosmosAsyncDatabase;
import com.azure.cosmos.CosmosClientBuilder;
import com.azure.cosmos.CosmosException;
import com.azure.cosmos.examples.common.AccountSettings;
import com.azure.cosmos.examples.common.CustomPOJO2;
import com.azure.cosmos.implementation.Utils;
import com.azure.cosmos.models.ChangeFeedProcessorOptions;
import com.azure.cosmos.models.CosmosContainerProperties;
import com.azure.cosmos.models.CosmosContainerRequestOptions;
import com.azure.cosmos.models.CosmosContainerResponse;
import com.azure.cosmos.models.CosmosDatabaseResponse;
import com.azure.cosmos.models.ThroughputProperties;
import com.fasterxml.jackson.core.JsonProcessingException;
import com.fasterxml.jackson.databind.JsonNode;
import com.fasterxml.jackson.databind.ObjectMapper;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import reactor.core.scheduler.Schedulers;
import java.time.Duration;
import java.util.List;
import java.util.UUID;
import java.util.function.Consumer;
/**
* Sample for Change Feed Processor.
* This sample models an application where documents are being inserted into one container (the "feed container"),
* and meanwhile another worker thread or worker application is pulling inserted documents from the feed container's Change Feed
* and operating on them in some way. For one or more workers to process the Change Feed of a container, the workers must first contact the server
* and "lease" access to monitor one or more partitions of the feed container. The Change Feed Processor Library
* handles leasing automatically for you, however you must create a separate "lease container" where the Change Feed
* Processor Library can store and track leases container partitions.
*/
public class SampleChangeFeedProcessor {
public static int WAIT_FOR_WORK = 60000;
public static final String DATABASE_NAME = "db_" + UUID.randomUUID();
public static final String COLLECTION_NAME = "coll_" + UUID.randomUUID();
private static final ObjectMapper OBJECT_MAPPER = Utils.getSimpleObjectMapper();
protected static Logger logger = LoggerFactory.getLogger(SampleChangeFeedProcessor.class);
private static boolean isWorkCompleted = false;
private static ChangeFeedProcessorOptions options;
public static void main(String[] args) {
logger.info("Begin Sample");
try {
// <ChangeFeedProcessorOptions>
options = new ChangeFeedProcessorOptions();
options.setStartFromBeginning(false);
options.setLeasePrefix("myChangeFeedDeploymentUnit");
// </ChangeFeedProcessorOptions>
//Summary of the next four commands:
//-Create an asynchronous Azure Cosmos DB client and database so that we can issue async requests to the DB
//-Create a "feed container" and a "lease container" in the DB
logger.info("Create CosmosClient");
CosmosAsyncClient client = getCosmosClient();
logger.info("Create sample's database: " + DATABASE_NAME);
CosmosAsyncDatabase cosmosDatabase = createNewDatabase(client, DATABASE_NAME);
logger.info("Create container for documents: " + COLLECTION_NAME);
CosmosAsyncContainer feedContainer = createNewCollection(client, DATABASE_NAME, COLLECTION_NAME);
logger.info("Create container for lease: " + COLLECTION_NAME + "-leases");
CosmosAsyncContainer leaseContainer = createNewLeaseCollection(client, DATABASE_NAME, COLLECTION_NAME + "-leases");
//Model of a worker thread or application which leases access to monitor one or more feed container
//partitions via the Change Feed. In a real-world application you might deploy this code in an Azure function.
//The next line causes the worker to create and start an instance of the Change Feed Processor. See the implementation of getChangeFeedProcessor() for guidance
//on creating a handler for Change Feed events. In this stream, we also trigger the insertion of 10 documents on a separate
//thread.
// <StartChangeFeedProcessor>
logger.info("Start Change Feed Processor on worker (handles changes asynchronously)");
ChangeFeedProcessor changeFeedProcessorInstance = new ChangeFeedProcessorBuilder()
.hostName("SampleHost_1")
.feedContainer(feedContainer)
.leaseContainer(leaseContainer)
.handleChanges(handleChanges())
.buildChangeFeedProcessor();
changeFeedProcessorInstance.start()
.subscribeOn(Schedulers.boundedElastic())
.subscribe();
// </StartChangeFeedProcessor>
//These two lines model an application which is inserting ten documents into the feed container
logger.info("Start application that inserts documents into feed container");
createNewDocumentsCustomPOJO(feedContainer, 10, Duration.ofSeconds(3));
//This loop models the Worker main loop, which spins while its Change Feed Processor instance asynchronously
//handles incoming Change Feed events from the feed container. Of course in this sample, polling
//isWorkCompleted is unnecessary because items are being added to the feed container on the same thread, and you
//can see just above isWorkCompleted is set to true.
//But conceptually the worker is part of a different thread or application than the one which is inserting
//into the feed container; so this code illustrates the worker waiting and listening for changes to the feed container
long remainingWork = WAIT_FOR_WORK;
while (!isWorkCompleted && remainingWork > 0) {
Thread.sleep(100);
remainingWork -= 100;
}
//When all documents have been processed, clean up
if (isWorkCompleted) {
changeFeedProcessorInstance.stop().subscribe();
} else {
throw new RuntimeException("The change feed processor initialization and automatic create document feeding process did not complete in the expected time");
}
logger.info("Delete sample's database: " + DATABASE_NAME);
//deleteDatabase(cosmosDatabase);
Thread.sleep(500);
} catch (Exception e) {
e.printStackTrace();
}
logger.info("End Sample");
}
// <Delegate>
private static Consumer<List<JsonNode>> handleChanges() {
return (List<JsonNode> docs) -> {
logger.info("Start handleChanges()");
for (JsonNode document : docs) {
try {
//Change Feed hands the document to you in the form of a JsonNode
//As a developer you have two options for handling the JsonNode document provided to you by Change Feed
//One option is to operate on the document in the form of a JsonNode, as shown below. This is great
//especially if you do not have a single uniform data model for all documents.
logger.info("Document received: " + OBJECT_MAPPER.writerWithDefaultPrettyPrinter()
.writeValueAsString(document));
//You can also transform the JsonNode to a POJO having the same structure as the JsonNode,
//as shown below. Then you can operate on the POJO.
CustomPOJO2 pojo_doc = OBJECT_MAPPER.treeToValue(document, CustomPOJO2.class);
logger.info("id: " + pojo_doc.getId());
} catch (JsonProcessingException e) {
e.printStackTrace();
}
}
isWorkCompleted = true;
logger.info("End handleChanges()");
};
}
// </Delegate>
public static CosmosAsyncClient getCosmosClient() {
return new CosmosClientBuilder()
.endpoint(AccountSettings.HOST)
.key(AccountSettings.MASTER_KEY)
.contentResponseOnWriteEnabled(true)
.consistencyLevel(ConsistencyLevel.SESSION)
.buildAsyncClient();
}
public static CosmosAsyncDatabase createNewDatabase(CosmosAsyncClient client, String databaseName) {
CosmosDatabaseResponse databaseResponse = client.createDatabaseIfNotExists(databaseName).block();
return client.getDatabase(databaseResponse.getProperties().getId());
}
public static void deleteDatabase(CosmosAsyncDatabase cosmosDatabase) {
cosmosDatabase.delete().block();
}
public static CosmosAsyncContainer createNewCollection(CosmosAsyncClient client, String databaseName, String collectionName) {
CosmosAsyncDatabase databaseLink = client.getDatabase(databaseName);
CosmosAsyncContainer collectionLink = databaseLink.getContainer(collectionName);
CosmosContainerResponse containerResponse = null;
try {
containerResponse = collectionLink.read().block();
if (containerResponse != null) {
throw new IllegalArgumentException(String.format("Collection %s already exists in database %s.", collectionName, databaseName));
}
} catch (RuntimeException ex) {
if (ex instanceof CosmosException) {
CosmosException CosmosException = (CosmosException) ex;
if (CosmosException.getStatusCode() != 404) {
throw ex;
}
} else {
throw ex;
}
}
CosmosContainerProperties containerSettings = new CosmosContainerProperties(collectionName, "/pk");
CosmosContainerRequestOptions requestOptions = new CosmosContainerRequestOptions();
ThroughputProperties throughputProperties = ThroughputProperties.createManualThroughput(10000);
containerResponse = databaseLink.createContainer(containerSettings, throughputProperties, requestOptions).block();
if (containerResponse == null) {
throw new RuntimeException(String.format("Failed to create collection %s in database %s.", collectionName, databaseName));
}
return databaseLink.getContainer(containerResponse.getProperties().getId());
}
public static CosmosAsyncContainer createNewLeaseCollection(CosmosAsyncClient client, String databaseName, String leaseCollectionName) {
CosmosAsyncDatabase databaseLink = client.getDatabase(databaseName);
CosmosAsyncContainer leaseCollectionLink = databaseLink.getContainer(leaseCollectionName);
CosmosContainerResponse leaseContainerResponse = null;
try {
leaseContainerResponse = leaseCollectionLink.read().block();
if (leaseContainerResponse != null) {
leaseCollectionLink.delete().block();
try {
Thread.sleep(1000);
} catch (InterruptedException ex) {
ex.printStackTrace();
}
}
} catch (RuntimeException ex) {
if (ex instanceof CosmosException) {
CosmosException CosmosException = (CosmosException) ex;
if (CosmosException.getStatusCode() != 404) {
throw ex;
}
} else {
throw ex;
}
}
CosmosContainerProperties containerSettings = new CosmosContainerProperties(leaseCollectionName, "/id");
CosmosContainerRequestOptions requestOptions = new CosmosContainerRequestOptions();
ThroughputProperties throughputProperties = ThroughputProperties.createManualThroughput(400);
leaseContainerResponse = databaseLink.createContainer(containerSettings, throughputProperties, requestOptions).block();
if (leaseContainerResponse == null) {
throw new RuntimeException(String.format("Failed to create collection %s in database %s.", leaseCollectionName, databaseName));
}
return databaseLink.getContainer(leaseContainerResponse.getProperties().getId());
}
public static void createNewDocumentsCustomPOJO(CosmosAsyncContainer containerClient, int count, Duration delay) {
String suffix = UUID.randomUUID().toString();
for (int i = 0; i <= count; i++) {
CustomPOJO2 document = new CustomPOJO2();
document.setId(String.format("0%d-%s", i, suffix));
document.setPk(document.getId()); // This is a very simple example, so we'll just have a partition key (/pk) field that we set equal to id
containerClient.createItem(document).subscribe(doc -> {
logger.info("Document write: " + doc);
});
long remainingWork = delay.toMillis();
try {
while (remainingWork > 0) {
Thread.sleep(100);
remainingWork -= 100;
}
} catch (InterruptedException iex) {
// exception caught
break;
}
}
}
}
- 生成并运行 Java 应用程序。 应用程序启动更改源处理器，将示例文档插入源容器，并处理传入的更改。
结论
在本指南中，您将学习如何使用 Azure Cosmos DB Java SDK V4 创建一个 Java 应用程序，该应用程序使用 Azure Cosmos DB for NoSQL 数据库，并利用更改源处理器进行实时数据处理。 可以扩展此应用程序来处理更复杂的用例，并使用 Azure Cosmos DB 构建可靠、可缩放和多区域分布式应用程序。
其他资源
后续步骤
现在，您可以通过以下文章进一步了解有关更改馈送估算器的信息：