Visitar URL original
FirebaseAILogic Framework Reference
Blog Go to console

FirebaseAILogic Framework Reference

TemplateChat

public final class TemplateChat : Sendable

[Public Preview] An ongoing conversation with a model backed by a server prompt template.

Warning

This API is a public preview and may be subject to change.

A TemplateChat manages conversation history on the client while utilizing the prompt, system instructions, and configurations defined in a server prompt template. The template's configurations apply to every turn automatically.

Function Calling in Chat

If the template defines function declarations in its frontmatter tools, the model may return a FunctionCallPart in its response candidates. When this occurs, execute the local function logic and supply the result as a FunctionResponsePart in a new message using one of the sendMessage(_:) or sendMessageStream(_:) methods to continue the conversation.

For more details, see Multi-turn interactions with server prompt templates.

  • The previous content from the chat that has been successfully sent and received from the model.

    This history is provided to the model for each subsequent message sent, and is injected into the server prompt template at the position marked by the {{history}} tag.

    Declaration

    Swift

    public var history: [ModelContent] { get set }

  • [Public Preview] Sends a message to the model and returns the response.

    Warning

    This API is a public preview and may be subject to change.

    Throws

    A GenerateContentError if the request failed, or an EncodingError if inputs cannot be encoded.

    Declaration

    Swift

    public func sendMessage(_ content: [ModelContent]) async throws -> GenerateContentResponse

    Parameters

    content

    The message content to send to the model, including any FunctionResponseParts responding to previous function calls; pass an empty array to send an empty turn.

    Return Value

    The content generated by the model.

  • [Public Preview] Sends a message to the model and returns the response.

    Warning

    This API is a public preview and may be subject to change.

    Calling this method with no arguments sends an empty turn, which runs the template's own prompt against the current history without adding any new content. This is typically how the first turn of a conversation is started when the template supplies the prompt itself:

    // Template content: "What was the weather in {{city}} on {{date}}?\n{{history}}"
let chat = model.startChat(templateID: "weather", inputs: ["city": "Boston", ...])
let response = try await chat.sendMessage()

    Throws

    A GenerateContentError if the request failed, or an EncodingError if inputs cannot be encoded.

    Declaration

    Swift

    public func sendMessage(_ parts: any PartsRepresentable...) async throws
  -> GenerateContentResponse

    Parameters

    parts

    The new content to send as a single chat message; pass no arguments to send an empty turn.

    Return Value

    The content generated by the model.

  • [Public Preview] Sends a message to the model and returns the response as a stream of GenerateContentResponses.

    Warning

    This API is a public preview and may be subject to change.

    Throws

    A GenerateContentError if the request failed, or an EncodingError if inputs cannot be encoded.

    Declaration

    Swift

    @available(macOS 12.0, watchOS 8.0, *)
public func sendMessageStream(_ content: [ModelContent]) throws
  -> AsyncThrowingStream<GenerateContentResponse, Error>

    Parameters

    content

    The message to send to the model, including any FunctionResponseParts responding to previous function calls; pass an empty array to send an empty turn.

    Return Value

    An AsyncThrowingStream that yields GenerateContentResponse objects.

  • [Public Preview] Sends a message to the model and returns the response as a stream of GenerateContentResponses.

    Warning

    This API is a public preview and may be subject to change.

    Calling this method with no arguments sends an empty turn, which runs the template's own prompt against the current history without adding any new content.

    Throws

    A GenerateContentError if the request failed, or an EncodingError if inputs cannot be encoded.

    Declaration

    Swift

    @available(macOS 12.0, watchOS 8.0, *)
public func sendMessageStream(_ parts: any PartsRepresentable...) throws
  -> AsyncThrowingStream<GenerateContentResponse, Error>

    Parameters

    parts

    The new content to send as a single chat message; pass no arguments to send an empty turn.

    Return Value

    An AsyncThrowingStream that yields GenerateContentResponse objects.