共用方式為

Azure OpenAI client library for Java - version 1.0.0-beta.13

  • 發行項

Azure OpenAI is a managed service that allows developers to deploy, tune, and generate content from OpenAI models on Azure resources.

The Azure OpenAI client library for Java is an adaptation of OpenAI's REST APIs that provides an idiomatic interface and rich integration with the rest of the Azure SDK ecosystem.

Use the client library for Azure OpenAI to:

For concrete examples you can have a look at the following links. Some of the more common scenarios are covered:

If you want to see the full code for these snippets check out our samples folder.

Source code | API reference documentation | Product Documentation | Samples | Troubleshooting

Getting started

Prerequisites

Adding the package to your product

<dependency>
    <groupId>com.azure</groupId>
    <artifactId>azure-ai-openai</artifactId>
    <version>1.0.0-beta.12</version>
</dependency>

Authentication

In order to interact with the Azure OpenAI Service you'll need to create an instance of client class, OpenAIAsyncClient or OpenAIClient by using OpenAIClientBuilder. To configure a client for use with Azure OpenAI, provide a valid endpoint URI to an Azure OpenAI resource along with a corresponding key credential, token credential, or Azure Identity credential that's authorized to use the Azure OpenAI resource.

Create an Azure OpenAI client with key credential

Get Azure OpenAI key credential from the Azure Portal.

OpenAIClient client = new OpenAIClientBuilder()
    .credential(new AzureKeyCredential("{key}"))
    .endpoint("{endpoint}")
    .buildClient();

or

OpenAIAsyncClient client = new OpenAIClientBuilder()
    .credential(new AzureKeyCredential("{key}"))
    .endpoint("{endpoint}")
    .buildAsyncClient();

Support for non-Azure OpenAI

The SDK also supports operating against the public non-Azure OpenAI. The response models remain the same, only the setup of the OpenAIClient is slightly different. First, get Non-Azure OpenAI API key from Open AI authentication API keys. Then setup your OpenAIClient as follows:

OpenAIClient client = new OpenAIClientBuilder()
    .credential(new KeyCredential("{openai-secret-key}"))
    .buildClient();

or

OpenAIAsyncClient client = new OpenAIClientBuilder()
    .credential(new KeyCredential("{openai-secret-key}"))
    .buildAsyncClient();

Create an Azure OpenAI client with Azure Entra ID credential

Azure SDK for Java supports an Azure Identity package, making it easy to get credentials from Microsoft identity platform.

Authentication with Entra ID requires some initial setup:

  • Add the Azure Identity package
<dependency>
    <groupId>com.azure</groupId>
    <artifactId>azure-identity</artifactId>
    <version>1.14.2</version>
</dependency>

After setup, you can choose which type of credential from azure.identity to use. As an example, DefaultAzureCredential can be used to authenticate the client: Set the values of the client ID, tenant ID, and client secret of the Azure EntraID application as environment variables: AZURE_CLIENT_ID, AZURE_TENANT_ID, AZURE_CLIENT_SECRET.

Authorization is easiest using DefaultAzureCredential. It finds the best credential to use in its running environment. For more information about using Azure Entra ID authorization with OpenAI service, please refer to the associated documentation.

TokenCredential defaultCredential = new DefaultAzureCredentialBuilder().build();
OpenAIClient client = new OpenAIClientBuilder()
    .credential(defaultCredential)
    .endpoint("{endpoint}")
    .buildClient();

Create a client with proxy options

Create an OpenAI client with proxy options.

// Proxy options
final String hostname = "{your-host-name}";
final int port = 447; // your port number

ProxyOptions proxyOptions = new ProxyOptions(ProxyOptions.Type.HTTP, new InetSocketAddress(hostname, port))
    .setCredentials("{username}", "{password}");

OpenAIClient client = new OpenAIClientBuilder()
    .credential(new AzureKeyCredential("{key}"))
    .endpoint("{endpoint}")
    .clientOptions(new HttpClientOptions().setProxyOptions(proxyOptions))
    .buildClient();

Key concepts

Examples

The following sections provide several code snippets covering some of the most common OpenAI service tasks, including:

Legacy completions

It is generally preferable to use Chat Completions instead. However, Completions are still supported:

List<String> prompt = new ArrayList<>();
prompt.add("Say this is a test");

Completions completions = client.getCompletions("{deploymentOrModelName}", new CompletionsOptions(prompt));

System.out.printf("Model ID=%s is created at %s.%n", completions.getId(), completions.getCreatedAt());
for (Choice choice : completions.getChoices()) {
    System.out.printf("Index: %d, Text: %s.%n", choice.getIndex(), choice.getText());
}

For a complete sample example, see sample Text Completions.

Streaming legacy completions

List<String> prompt = new ArrayList<>();
prompt.add("How to bake a cake?");

IterableStream<Completions> completionsStream = client
    .getCompletionsStream("{deploymentOrModelName}", new CompletionsOptions(prompt));

completionsStream
    .stream()
    .forEach(completions -> System.out.print(completions.getChoices().get(0).getText()));

For a complete sample example, see sample Streaming Text Completions.

Chat completions

List<ChatRequestMessage> chatMessages = new ArrayList<>();
chatMessages.add(new ChatRequestSystemMessage("You are a helpful assistant. You will talk like a pirate."));
chatMessages.add(new ChatRequestUserMessage("Can you help me?"));
chatMessages.add(new ChatRequestAssistantMessage("Of course, me hearty! What can I do for ye?"));
chatMessages.add(new ChatRequestUserMessage("What's the best way to train a parrot?"));

ChatCompletions chatCompletions = client.getChatCompletions("{deploymentOrModelName}",
    new ChatCompletionsOptions(chatMessages));

System.out.printf("Model ID=%s is created at %s.%n", chatCompletions.getId(), chatCompletions.getCreatedAt());
for (ChatChoice choice : chatCompletions.getChoices()) {
    ChatResponseMessage message = choice.getMessage();
    System.out.printf("Index: %d, Chat Role: %s.%n", choice.getIndex(), message.getRole());
    System.out.println("Message:");
    System.out.println(message.getContent());
}

For a complete sample example, see sample Chat Completions.

For function call sample, see function call. However, they are considered a legacy feature. Using tools is the preferred way. For more details see sample tool calls.

For Bring Your Own Data sample, see Bring Your Own Data.

Please refer to the service documentation for a conceptual discussion of text completion.

Streaming chat completions

List<ChatRequestMessage> chatMessages = new ArrayList<>();
chatMessages.add(new ChatRequestSystemMessage("You are a helpful assistant. You will talk like a pirate."));
chatMessages.add(new ChatRequestUserMessage("Can you help me?"));
chatMessages.add(new ChatRequestAssistantMessage("Of course, me hearty! What can I do for ye?"));
chatMessages.add(new ChatRequestUserMessage("What's the best way to train a parrot?"));

client.getChatCompletionsStream("{deploymentOrModelName}", new ChatCompletionsOptions(chatMessages))
        .forEach(chatCompletions -> {
            if (CoreUtils.isNullOrEmpty(chatCompletions.getChoices())) {
                return;
            }
            ChatResponseMessage delta = chatCompletions.getChoices().get(0).getDelta();
            if (delta.getRole() != null) {
                System.out.println("Role = " + delta.getRole());
            }
            if (delta.getContent() != null) {
                String content = delta.getContent();
                System.out.print(content);
            }
        });

To compute tokens in streaming chat completions, see sample Streaming Chat Completions.

Text embeddings

EmbeddingsOptions embeddingsOptions = new EmbeddingsOptions(
    Arrays.asList("Your text string goes here"));

Embeddings embeddings = client.getEmbeddings("{deploymentOrModelName}", embeddingsOptions);

for (EmbeddingItem item : embeddings.getData()) {
    System.out.printf("Index: %d.%n", item.getPromptIndex());
    for (Float embedding : item.getEmbedding()) {
        System.out.printf("%f;", embedding);
    }
}

For a complete sample example, see sample Embedding.

Please refer to the service documentation for a conceptual discussion of openAI embedding.

Image Generation

ImageGenerationOptions imageGenerationOptions = new ImageGenerationOptions(
    "A drawing of the Seattle skyline in the style of Van Gogh");
ImageGenerations images = client.getImageGenerations("{deploymentOrModelName}", imageGenerationOptions);

for (ImageGenerationData imageGenerationData : images.getData()) {
    System.out.printf(
        "Image location URL that provides temporary access to download the generated image is %s.%n",
        imageGenerationData.getUrl());
}

For a complete sample example, see sample Image Generation.

Audio Transcription

The OpenAI service starts supporting audio transcription with the introduction of Whisper models. The following code snippet shows how to use the service to transcribe audio.

String fileName = "{your-file-name}";
Path filePath = Paths.get("{your-file-path}" + fileName);

byte[] file = BinaryData.fromFile(filePath).toBytes();
AudioTranscriptionOptions transcriptionOptions = new AudioTranscriptionOptions(file)
    .setResponseFormat(AudioTranscriptionFormat.JSON);

AudioTranscription transcription = client.getAudioTranscription("{deploymentOrModelName}", fileName, transcriptionOptions);

System.out.println("Transcription: " + transcription.getText());

For a complete sample example, see sample Audio Transcription. Please refer to the service documentation for a conceptual discussion of Whisper.

Audio Translation

The OpenAI service starts supporting audio translation with the introduction of Whisper models. The following code snippet shows how to use the service to translate audio.

String fileName = "{your-file-name}";
Path filePath = Paths.get("{your-file-path}" + fileName);

byte[] file = BinaryData.fromFile(filePath).toBytes();
AudioTranslationOptions translationOptions = new AudioTranslationOptions(file)
    .setResponseFormat(AudioTranslationFormat.JSON);

AudioTranslation translation = client.getAudioTranslation("{deploymentOrModelName}", fileName, translationOptions);

System.out.println("Translation: " + translation.getText());

For a complete sample example, see sample Audio Translation. Please refer to the service documentation for a conceptual discussion of Whisper.

Text completions with images

Currently, only available in OpenAI. For more details please visit the OpenAI vision documentation page. By providing image URLs, it is possible to use images along with prompts. For more details see the usage samples for chat completions with images.

List<ChatRequestMessage> chatMessages = new ArrayList<>();
chatMessages.add(new ChatRequestSystemMessage("You are a helpful assistant that describes images"));
chatMessages.add(new ChatRequestUserMessage(Arrays.asList(
        new ChatMessageTextContentItem("Please describe this image"),
        new ChatMessageImageContentItem(
                new ChatMessageImageUrl("https://upload.wikimedia.org/wikipedia/commons/thumb/4/44/Microsoft_logo.svg/512px-Microsoft_logo.svg.png"))
)));

ChatCompletionsOptions chatCompletionsOptions = new ChatCompletionsOptions(chatMessages);
ChatCompletions chatCompletions = client.getChatCompletions("{deploymentOrModelName}", chatCompletionsOptions);

System.out.println("Chat completion: " + chatCompletions.getChoices().get(0).getMessage().getContent());

Tool calls

Tools extend chat completions by allowing an assistant to invoke defined functions and other capabilities in the process of fulfilling a chat completions request. To use chat tools, start by defining a function tool:

List<ChatRequestMessage> chatMessages = Arrays.asList(
        new ChatRequestSystemMessage("You are a helpful assistant."),
        new ChatRequestUserMessage("What sort of clothing should I wear today in Berlin?")
);
ChatCompletionsToolDefinition toolDefinition = new ChatCompletionsFunctionToolDefinition(
        new ChatCompletionsFunctionToolDefinitionFunction("MyFunctionName"));

ChatCompletionsOptions chatCompletionsOptions = new ChatCompletionsOptions(chatMessages);
chatCompletionsOptions.setTools(Arrays.asList(toolDefinition));

ChatCompletions chatCompletions = client.getChatCompletions("{deploymentOrModelName}", chatCompletionsOptions);

ChatChoice choice = chatCompletions.getChoices().get(0);
// The LLM is requesting the calling of the function we defined in the original request
if (choice.getFinishReason() == CompletionsFinishReason.TOOL_CALLS) {
    ChatCompletionsFunctionToolCall toolCall = (ChatCompletionsFunctionToolCall) choice.getMessage().getToolCalls().get(0);
    String functionArguments = toolCall.getFunction().getArguments();

    // As an additional step, you may want to deserialize the parameters, so you can call your function
    MyFunctionCallArguments parameters = BinaryData.fromString(functionArguments).toObject(MyFunctionCallArguments.class);

    String functionCallResult = "{the-result-of-my-function}"; // myFunction(parameters...);

    ChatRequestAssistantMessage assistantMessage = new ChatRequestAssistantMessage("");
    assistantMessage.setToolCalls(choice.getMessage().getToolCalls());

    // We include:
    // - The past 2 messages from the original request
    // - A new ChatRequestAssistantMessage with the tool calls from the original request
    // - A new ChatRequestToolMessage with the result of our function call
    List<ChatRequestMessage> followUpMessages = Arrays.asList(
            chatMessages.get(0),
            chatMessages.get(1),
            assistantMessage,
            new ChatRequestToolMessage(functionCallResult, toolCall.getId())
    );

    ChatCompletionsOptions followUpChatCompletionsOptions = new ChatCompletionsOptions(followUpMessages);

    ChatCompletions followUpChatCompletions = client.getChatCompletions("{deploymentOrModelName}", followUpChatCompletionsOptions);

    // This time the finish reason is STOPPED
    ChatChoice followUpChoice = followUpChatCompletions.getChoices().get(0);
    if (followUpChoice.getFinishReason() == CompletionsFinishReason.STOPPED) {
        System.out.println("Chat Completions Result: " + followUpChoice.getMessage().getContent());
    }
}

Text To Speech

The OpenAI service starts supporting text to speech with the introduction of tts models. The following code snippet shows how to use the service to convert text to speech.

String deploymentOrModelId = "{azure-open-ai-deployment-model-id}";
SpeechGenerationOptions options = new SpeechGenerationOptions(
        "Today is a wonderful day to build something people love!",
        SpeechVoice.ALLOY);
BinaryData speech = client.generateSpeechFromText(deploymentOrModelId, options);
// Checkout your generated speech in the file system.
Path path = Paths.get("{your-local-file-path}/speech.wav");
Files.write(path, speech.toBytes());

See sample Text to Speech for a complete sample. Please refer to the service documentation for a conceptual discussion of Text to Speech.

File operations

The OpenAI service supports upload, get, list, and delete operations for interacting File APIs with OpenAI service.

// Upload a file
FileDetails fileDetails = new FileDetails(
    BinaryData.fromFile(Paths.get("{your-local-file-path}/batch_tasks.jsonl")),
    "batch_tasks.jsonl");
OpenAIFile file = client.uploadFile(fileDetails, FilePurpose.BATCH);
String fileId = file.getId();
// Get single file
OpenAIFile fileFromBackend = client.getFile(fileId);
// List files
List<OpenAIFile> files = client.listFiles(FilePurpose.ASSISTANTS);
// Delete file
FileDeletionStatus deletionStatus = client.deleteFile(fileId);

For a complete sample example, see sample File Operations.

Batch operations

The OpenAI service supports create, get, list, and delete operations for interacting Batch APIs with OpenAI service. Use Batch API to send asynchronous groups of requests with 50% lower costs, a separate pool of significantly higher rate limits, and a clear 24-hour turnaround time. The service is ideal for processing jobs that don't require immediate responses.

String fileId = "{fileId-from-service-side}";
// Create a batch
Batch batch = client.createBatch(new BatchCreateRequest("/chat/completions", fileId, "24h"));
// Get single file
byte[] fileContent = client.getFileContent(batch.getOutputFileId());
// List batches
PageableList<Batch> batchPageableList = client.listBatches();
// Cancel a batch
Batch cancelledBatch = client.cancelBatch(batch.getId());

For a complete sample example, see sample Batch Operations.

Structured Outputs

Structured Outputs can be enabled by setting the parameter strict: true in an API call with either a defined response format or function definitions.

ChatCompletionsOptions chatCompletionsOptions = new ChatCompletionsOptions(Arrays.asList(new ChatRequestUserMessage("What is the weather in Seattle?")))
    // Previously, the response_format parameter was only available to specify that the model should return a valid JSON.
    // In addition to this, we are introducing a new way of specifying which JSON schema to follow.
    .setResponseFormat(new ChatCompletionsJsonSchemaResponseFormat(
        new ChatCompletionsJsonSchemaResponseFormatJsonSchema("get_weather")
            .setStrict(true)
            .setDescription("Fetches the weather in the given location")
            .setSchema(BinaryData.fromObject(new Parameters()))));

For a full sample, see Structured Output: Response Format. For more details see the OpenAI structured output documentation.

Upload large files in multiple parts

uploads allows you to upload large files in multiple parts.

CreateUploadRequest createUploadRequest = new CreateUploadRequest("{fileNameToCreate}", CreateUploadRequestPurpose.ASSISTANTS,
    totalFilesSize, "text/plain");
Upload upload = client.createUpload(createUploadRequest);
String uploadId = upload.getId();

UploadPart uploadPartAdded = client.addUploadPart(uploadId,
    new AddUploadPartRequest(new DataFileDetails(BinaryData.fromFile(path)).setFilename("{fileName}")));
String uploadPartAddedId = uploadPartAdded.getId();
System.out.println("Upload part added, upload part ID = " + uploadPartAddedId);

UploadPart uploadPartAdded2 = client.addUploadPart(uploadId,
    new AddUploadPartRequest(new DataFileDetails(BinaryData.fromFile(path2)).setFilename("{fileName2}")));
String uploadPartAddedId2 = uploadPartAdded2.getId();
System.out.println("Upload part 2 added, upload part ID = " + uploadPartAddedId2);

CompleteUploadRequest completeUploadRequest = new CompleteUploadRequest(Arrays.asList(uploadPartAddedId, uploadPartAddedId2));
Upload completeUpload = client.completeUpload(uploadId, completeUploadRequest);
System.out.println("Upload completed, upload ID = " + completeUpload.getId());

For a full sample, see Upload large files in multiple parts. For more details see the OpenAI uploads documentation.

Troubleshooting

Enable client logging

You can set the AZURE_LOG_LEVEL environment variable to view logging statements made in the client library. For example, setting AZURE_LOG_LEVEL=2 would show all informational, warning, and error log messages. The log levels can be found here: log levels.

Default HTTP Client

All client libraries by default use the Netty HTTP client. Adding the above dependency will automatically configure the client library to use the Netty HTTP client. Configuring or changing the HTTP client is detailed in the HTTP clients wiki.

Default SSL library

All client libraries, by default, use the Tomcat-native Boring SSL library to enable native-level performance for SSL operations. The Boring SSL library is an uber jar containing native libraries for Linux / macOS / Windows, and provides better performance compared to the default SSL implementation within the JDK. For more information, including how to reduce the dependency size, refer to the performance tuning section of the wiki.

For more details, see TROUBLESHOOTING guideline.

Next steps

  • Samples are explained in detail here.

Contributing

For details on contributing to this repository, see the contributing guide.

  1. Fork it
  2. Create your feature branch (git checkout -b my-new-feature)
  3. Commit your changes (git commit -am 'Add some feature')
  4. Push to the branch (git push origin my-new-feature)
  5. Create new Pull Request