types.go

// Copyright 2024 Google LLC
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//      http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

// Code generated by the Google Gen AI SDK generator DO NOT EDIT.

package genai

import (
	"cloud.google.com/go/civil"
	"encoding/json"
	"errors"
	"fmt"
	"log"
	"strconv"
	"strings"
	"time"
)

// Outcome of the code execution.
type Outcome string

const (
	// Unspecified status. This value should not be used.
	OutcomeUnspecified Outcome = "OUTCOME_UNSPECIFIED"
	// Code execution completed successfully.
	OutcomeOK Outcome = "OUTCOME_OK"
	// Code execution finished but with a failure. `stderr` should contain the reason.
	OutcomeFailed Outcome = "OUTCOME_FAILED"
	// Code execution ran for too long, and was cancelled. There may or may not be a partial
	// output present.
	OutcomeDeadlineExceeded Outcome = "OUTCOME_DEADLINE_EXCEEDED"
)

// Programming language of the `code`.
type Language string

const (
	// Unspecified language. This value should not be used.
	LanguageUnspecified Language = "LANGUAGE_UNSPECIFIED"
	// Python >= 3.10, with numpy and simpy available.
	LanguagePython Language = "PYTHON"
)

// A basic data type.
type Type string

const (
	TypeUnspecified Type = "TYPE_UNSPECIFIED"
	TypeString      Type = "STRING"
	TypeNumber      Type = "NUMBER"
	TypeInteger     Type = "INTEGER"
	TypeBoolean     Type = "BOOLEAN"
	TypeArray       Type = "ARRAY"
	TypeObject      Type = "OBJECT"
)

// Harm category.
type HarmCategory string

const (
	// The harm category is unspecified.
	HarmCategoryUnspecified HarmCategory = "HARM_CATEGORY_UNSPECIFIED"
	// The harm category is hate speech.
	HarmCategoryHateSpeech HarmCategory = "HARM_CATEGORY_HATE_SPEECH"
	// The harm category is dangerous content.
	HarmCategoryDangerousContent HarmCategory = "HARM_CATEGORY_DANGEROUS_CONTENT"
	// The harm category is harassment.
	HarmCategoryHarassment HarmCategory = "HARM_CATEGORY_HARASSMENT"
	// The harm category is sexually explicit content.
	HarmCategorySexuallyExplicit HarmCategory = "HARM_CATEGORY_SEXUALLY_EXPLICIT"
	// The harm category is civic integrity.
	HarmCategoryCivicIntegrity HarmCategory = "HARM_CATEGORY_CIVIC_INTEGRITY"
)

// Specify if the threshold is used for probability or severity score. If not specified,
// the threshold is used for probability score.
type HarmBlockMethod string

const (
	// The harm block method is unspecified.
	HarmBlockMethodUnspecified HarmBlockMethod = "HARM_BLOCK_METHOD_UNSPECIFIED"
	// The harm block method uses both probability and severity scores.
	HarmBlockMethodSeverity HarmBlockMethod = "SEVERITY"
	// The harm block method uses the probability score.
	HarmBlockMethodProbability HarmBlockMethod = "PROBABILITY"
)

// The harm block threshold.
type HarmBlockThreshold string

const (
	// Unspecified harm block threshold.
	HarmBlockThresholdUnspecified HarmBlockThreshold = "HARM_BLOCK_THRESHOLD_UNSPECIFIED"
	// Block low threshold and above (i.e. block more).
	HarmBlockThresholdBlockLowAndAbove HarmBlockThreshold = "BLOCK_LOW_AND_ABOVE"
	// Block medium threshold and above.
	HarmBlockThresholdBlockMediumAndAbove HarmBlockThreshold = "BLOCK_MEDIUM_AND_ABOVE"
	// Block only high threshold (i.e. block less).
	HarmBlockThresholdBlockOnlyHigh HarmBlockThreshold = "BLOCK_ONLY_HIGH"
	// Block none.
	HarmBlockThresholdBlockNone HarmBlockThreshold = "BLOCK_NONE"
	// Turn off the safety filter.
	HarmBlockThresholdOff HarmBlockThreshold = "OFF"
)

// The mode of the predictor to be used in dynamic retrieval.
type Mode string

const (
	// Always trigger retrieval.
	ModeUnspecified Mode = "MODE_UNSPECIFIED"
	// Run retrieval only when system decides it is necessary.
	ModeDynamic Mode = "MODE_DYNAMIC"
)

// The reason why the model stopped generating tokens. If empty, the model has not stopped
// generating the tokens.
type FinishReason string

const (
	// The finish reason is unspecified.
	FinishReasonUnspecified FinishReason = "FINISH_REASON_UNSPECIFIED"
	// Token generation reached a natural stopping point or a configured stop sequence.
	FinishReasonStop FinishReason = "STOP"
	// Token generation reached the configured maximum output tokens.
	FinishReasonMaxTokens FinishReason = "MAX_TOKENS"
	// Token generation stopped because the content potentially contains safety violations.
	// NOTE: When streaming, content is empty if content filters blocks the output.
	FinishReasonSafety FinishReason = "SAFETY"
	// The token generation stopped because of potential recitation.
	FinishReasonRecitation FinishReason = "RECITATION"
	// All other reasons that stopped the token generation.
	FinishReasonOther FinishReason = "OTHER"
	// Token generation stopped because the content contains forbidden terms.
	FinishReasonBlocklist FinishReason = "BLOCKLIST"
	// Token generation stopped for potentially containing prohibited content.
	FinishReasonProhibitedContent FinishReason = "PROHIBITED_CONTENT"
	// Token generation stopped because the content potentially contains Sensitive Personally
	// Identifiable Information (SPII).
	FinishReasonSPII FinishReason = "SPII"
	// The function call generated by the model is invalid.
	FinishReasonMalformedFunctionCall FinishReason = "MALFORMED_FUNCTION_CALL"
)

// Harm probability levels in the content.
type HarmProbability string

const (
	// Harm probability unspecified.
	HarmProbabilityUnspecified HarmProbability = "HARM_PROBABILITY_UNSPECIFIED"
	// Negligible level of harm.
	HarmProbabilityNegligible HarmProbability = "NEGLIGIBLE"
	// Low level of harm.
	HarmProbabilityLow HarmProbability = "LOW"
	// Medium level of harm.
	HarmProbabilityMedium HarmProbability = "MEDIUM"
	// High level of harm.
	HarmProbabilityHigh HarmProbability = "HIGH"
)

// Harm severity levels in the content.
type HarmSeverity string

const (
	// Harm severity unspecified.
	HarmSeverityUnspecified HarmSeverity = "HARM_SEVERITY_UNSPECIFIED"
	// Negligible level of harm severity.
	HarmSeverityNegligible HarmSeverity = "HARM_SEVERITY_NEGLIGIBLE"
	// Low level of harm severity.
	HarmSeverityLow HarmSeverity = "HARM_SEVERITY_LOW"
	// Medium level of harm severity.
	HarmSeverityMedium HarmSeverity = "HARM_SEVERITY_MEDIUM"
	// High level of harm severity.
	HarmSeverityHigh HarmSeverity = "HARM_SEVERITY_HIGH"
)

// Blocked reason.
type BlockedReason string

const (
	// Unspecified blocked reason.
	BlockedReasonUnspecified BlockedReason = "BLOCKED_REASON_UNSPECIFIED"
	// Candidates blocked due to safety.
	BlockedReasonSafety BlockedReason = "SAFETY"
	// Candidates blocked due to other reason.
	BlockedReasonOther BlockedReason = "OTHER"
	// Candidates blocked due to the terms which are included from the terminology blocklist.
	BlockedReasonBlocklist BlockedReason = "BLOCKLIST"
	// Candidates blocked due to prohibited content.
	BlockedReasonProhibitedContent BlockedReason = "PROHIBITED_CONTENT"
)

type DeploymentResourcesType string

const (
	// Should not be used.
	DeploymentResourcesTypeUnspecified DeploymentResourcesType = "DEPLOYMENT_RESOURCES_TYPE_UNSPECIFIED"
	// Resources that are dedicated to the DeployedModel, and that need a higher degree
	// of manual configuration.
	DeploymentResourcesTypeDedicatedResources DeploymentResourcesType = "DEDICATED_RESOURCES"
	// Resources that to large degree are decided by Vertex AI, and require only a modest
	// additional configuration.
	DeploymentResourcesTypeAutomaticResources DeploymentResourcesType = "AUTOMATIC_RESOURCES"
	// Resources that can be shared by multiple DeployedModels. A pre-configured DeploymentResourcePool
	// is required.
	DeploymentResourcesTypeSharedResources DeploymentResourcesType = "SHARED_RESOURCES"
)

// Config for the dynamic retrieval config mode.
type DynamicRetrievalConfigMode string

const (
	// Always trigger retrieval.
	DynamicRetrievalConfigModeUnspecified DynamicRetrievalConfigMode = "MODE_UNSPECIFIED"
	// Run retrieval only when system decides it is necessary.
	DynamicRetrievalConfigModeDynamic DynamicRetrievalConfigMode = "MODE_DYNAMIC"
)

// Config for the function calling config mode.
type FunctionCallingConfigMode string

const (
	// The function calling config mode is unspecified. Should not be used.
	FunctionCallingConfigModeUnspecified FunctionCallingConfigMode = "MODE_UNSPECIFIED"
	// Default model behavior, model decides to predict either function calls or natural
	// language response.
	FunctionCallingConfigModeAuto FunctionCallingConfigMode = "AUTO"
	// Model is constrained to always predicting function calls only. If "allowed_function_names"
	// are set, the predicted function calls will be limited to any one of "allowed_function_names",
	// else the predicted function calls will be any one of the provided "function_declarations".
	FunctionCallingConfigModeAny FunctionCallingConfigMode = "ANY"
	// Model will not predict any function calls. Model behavior is same as when not passing
	// any function declarations.
	FunctionCallingConfigModeNone FunctionCallingConfigMode = "NONE"
)

// The media resolution to use.
type MediaResolution string

const (
	// Media resolution has not been set
	MediaResolutionUnspecified MediaResolution = "MEDIA_RESOLUTION_UNSPECIFIED"
	// Media resolution set to low (64 tokens).
	MediaResolutionLow MediaResolution = "MEDIA_RESOLUTION_LOW"
	// Media resolution set to medium (256 tokens).
	MediaResolutionMedium MediaResolution = "MEDIA_RESOLUTION_MEDIUM"
	// Media resolution set to high (zoomed reframing with 256 tokens).
	MediaResolutionHigh MediaResolution = "MEDIA_RESOLUTION_HIGH"
)

// Enum that controls the safety filter level for objectionable content.
type SafetyFilterLevel string

const (
	SafetyFilterLevelBlockLowAndAbove    SafetyFilterLevel = "BLOCK_LOW_AND_ABOVE"
	SafetyFilterLevelBlockMediumAndAbove SafetyFilterLevel = "BLOCK_MEDIUM_AND_ABOVE"
	SafetyFilterLevelBlockOnlyHigh       SafetyFilterLevel = "BLOCK_ONLY_HIGH"
	SafetyFilterLevelBlockNone           SafetyFilterLevel = "BLOCK_NONE"
)

// Enum that controls the generation of people.
type PersonGeneration string

const (
	PersonGenerationDontAllow  PersonGeneration = "DONT_ALLOW"
	PersonGenerationAllowAdult PersonGeneration = "ALLOW_ADULT"
	PersonGenerationAllowAll   PersonGeneration = "ALLOW_ALL"
)

// Enum that specifies the language of the text in the prompt.
type ImagePromptLanguage string

const (
	ImagePromptLanguageAuto ImagePromptLanguage = "auto"
	ImagePromptLanguageEn   ImagePromptLanguage = "en"
	ImagePromptLanguageJa   ImagePromptLanguage = "ja"
	ImagePromptLanguageKo   ImagePromptLanguage = "ko"
	ImagePromptLanguageHi   ImagePromptLanguage = "hi"
)

// Enum representing the mask mode of a mask reference image.
type MaskReferenceMode string

const (
	MaskReferenceModeMaskModeDefault      MaskReferenceMode = "MASK_MODE_DEFAULT"
	MaskReferenceModeMaskModeUserProvided MaskReferenceMode = "MASK_MODE_USER_PROVIDED"
	MaskReferenceModeMaskModeBackground   MaskReferenceMode = "MASK_MODE_BACKGROUND"
	MaskReferenceModeMaskModeForeground   MaskReferenceMode = "MASK_MODE_FOREGROUND"
	MaskReferenceModeMaskModeSemantic     MaskReferenceMode = "MASK_MODE_SEMANTIC"
)

// Enum representing the control type of a control reference image.
type ControlReferenceType string

const (
	ControlReferenceTypeDefault  ControlReferenceType = "CONTROL_TYPE_DEFAULT"
	ControlReferenceTypeCanny    ControlReferenceType = "CONTROL_TYPE_CANNY"
	ControlReferenceTypeScribble ControlReferenceType = "CONTROL_TYPE_SCRIBBLE"
	ControlReferenceTypeFaceMesh ControlReferenceType = "CONTROL_TYPE_FACE_MESH"
)

// Enum representing the subject type of a subject reference image.
type SubjectReferenceType string

const (
	SubjectReferenceTypeSubjectTypeDefault SubjectReferenceType = "SUBJECT_TYPE_DEFAULT"
	SubjectReferenceTypeSubjectTypePerson  SubjectReferenceType = "SUBJECT_TYPE_PERSON"
	SubjectReferenceTypeSubjectTypeAnimal  SubjectReferenceType = "SUBJECT_TYPE_ANIMAL"
	SubjectReferenceTypeSubjectTypeProduct SubjectReferenceType = "SUBJECT_TYPE_PRODUCT"
)

// Server content modalities.
type Modality string

const (
	// The modality is unspecified.
	ModalityUnspecified Modality = "MODALITY_UNSPECIFIED"
	// Indicates the model should return text
	ModalityText Modality = "TEXT"
	// Indicates the model should return images.
	ModalityImage Modality = "IMAGE"
	// Indicates the model should return images.
	ModalityAudio Modality = "AUDIO"
)

// Metadata describes the input video content.
type VideoMetadata struct {
	// Optional. The end offset of the video.
	EndOffset string `json:"endOffset,omitempty"`
	// Optional. The start offset of the video.
	StartOffset string `json:"startOffset,omitempty"`
}

// Result of executing the [ExecutableCode]. Always follows a `part` containing the
// [ExecutableCode].
type CodeExecutionResult struct {
	// Required. Outcome of the code execution.
	Outcome Outcome `json:"outcome,omitempty"`
	// Optional. Contains stdout when code execution is successful, stderr or other description
	// otherwise.
	Output string `json:"output,omitempty"`
}

// Code generated by the model that is meant to be executed, and the result returned
// to the model. Generated when using the [FunctionDeclaration] tool and [FunctionCallingConfig]
// mode is set to [Mode.CODE].
type ExecutableCode struct {
	// Required. The code to be executed.
	Code string `json:"code,omitempty"`
	// Required. Programming language of the `code`.
	Language Language `json:"language,omitempty"`
}

// URI based data.
type FileData struct {
	// Required. URI.
	FileURI string `json:"fileUri,omitempty"`
	// Required. The IANA standard MIME type of the source data.
	MIMEType string `json:"mimeType,omitempty"`
}

// A function call.
type FunctionCall struct {
	// The unique ID of the function call. If populated, the client to execute the
	// `function_call` and return the response with the matching `id`.
	ID string `json:"id,omitempty"`
	// Optional. Required. The function parameters and values in JSON object format. See
	// [FunctionDeclaration.parameters] for parameter details.
	Args map[string]any `json:"args,omitempty"`
	// Required. The name of the function to call. Matches [FunctionDeclaration.name].
	Name string `json:"name,omitempty"`
}

// A function response.
type FunctionResponse struct {
	// The ID of the function call this response is for. Populated by the client
	// to match the corresponding function call `id`.
	ID string `json:"id,omitempty"`
	// Required. The name of the function to call. Matches [FunctionDeclaration.name] and
	// [FunctionCall.name].
	Name string `json:"name,omitempty"`
	// Required. The function response in JSON object format. Use "output" key to specify
	// function output and "error" key to specify error details (if any). If "output" and
	// "error" keys are not specified, then whole "response" is treated as function output.
	Response map[string]any `json:"response,omitempty"`
}

// Content blob.
type Blob struct {
	// Required. Raw bytes.
	Data []byte `json:"data,omitempty"`
	// Required. The IANA standard MIME type of the source data.
	MIMEType string `json:"mimeType,omitempty"`
}

// A datatype containing media content.
// Exactly one field within a Part should be set, representing the specific type
// of content being conveyed. Using multiple fields within the same `Part`
// instance is considered invalid.
type Part struct {
	// Metadata for a given video.
	VideoMetadata *VideoMetadata `json:"videoMetadata,omitempty"`
	// Indicates if the part is thought from the model.
	Thought bool `json:"thought,omitempty"`
	// Optional. Result of executing the [ExecutableCode].
	CodeExecutionResult *CodeExecutionResult `json:"codeExecutionResult,omitempty"`
	// Optional. Code generated by the model that is meant to be executed.
	ExecutableCode *ExecutableCode `json:"executableCode,omitempty"`
	// Optional. URI based data.
	FileData *FileData `json:"fileData,omitempty"`
	// Optional. A predicted [FunctionCall] returned from the model that contains a string
	// representing the [FunctionDeclaration.name] with the parameters and their values.
	FunctionCall *FunctionCall `json:"functionCall,omitempty"`
	// Optional. The result output of a [FunctionCall] that contains a string representing
	// the [FunctionDeclaration.name] and a structured JSON object containing any output
	// from the function call. It is used as context to the model.
	FunctionResponse *FunctionResponse `json:"functionResponse,omitempty"`
	// Optional. Inlined bytes data.
	InlineData *Blob `json:"inlineData,omitempty"`
	// Optional. Text part (can be code).
	Text string `json:"text,omitempty"`
}

// NewPartFromURI builds a Part from a given file URI and mime type.
func NewPartFromURI(fileURI, mimeType string) *Part {
	return &Part{
		FileData: &FileData{
			FileURI:  fileURI,
			MIMEType: mimeType,
		},
	}
}

// NewPartFromText builds a Part from a given text.
func NewPartFromText(text string) *Part {
	return &Part{
		Text: text,
	}
}

// NewPartFromBytes builds a Part from a given byte array and mime type.
func NewPartFromBytes(data []byte, mimeType string) *Part {
	return &Part{
		InlineData: &Blob{
			Data:     data,
			MIMEType: mimeType,
		},
	}
}

// NewPartFromFunctionCall builds a Part from a given function call.
func NewPartFromFunctionCall(name string, args map[string]any) *Part {
	return &Part{
		FunctionCall: &FunctionCall{
			Name: name,
			Args: args,
		},
	}
}

// NewPartFromFunctionResponse builds a Part from a given function response.
func NewPartFromFunctionResponse(name string, response map[string]any) *Part {
	return &Part{
		FunctionResponse: &FunctionResponse{
			Name:     name,
			Response: response,
		},
	}
}

// NewPartFromVideoMetadata builds a Part from a given end offset and start offset.
func NewPartFromVideoMetadata(endOffset, startOffset string) *Part {
	return &Part{
		VideoMetadata: &VideoMetadata{
			EndOffset:   endOffset,
			StartOffset: startOffset,
		},
	}
}

// NewPartFromExecutableCode builds a Part from a given executable code and language.
func NewPartFromExecutableCode(code string, language Language) *Part {
	return &Part{
		ExecutableCode: &ExecutableCode{
			Code:     code,
			Language: language,
		},
	}
}

// NewPartFromCodeExecutionResult builds a Part from a given outcome and output.
func NewPartFromCodeExecutionResult(outcome Outcome, output string) *Part {
	return &Part{
		CodeExecutionResult: &CodeExecutionResult{
			Outcome: outcome,
			Output:  output,
		},
	}
}

// Contains the multi-part content of a message.
type Content struct {
	// List of parts that constitute a single message. Each part may have
	// a different IANA MIME type.
	Parts []*Part `json:"parts,omitempty"`
	// Optional. The producer of the content. Must be either 'user' or
	// 'model'. Useful to set for multi-turn conversations, otherwise can be
	// left blank or unset. If role is not specified, SDK will determine the role.
	Role string `json:"role,omitempty"`
}

// NewUserContent builds a Content with a "user" role from a list of parts.
func NewUserContentFromParts(parts []*Part) *Content {
	return &Content{
		Parts: parts,
		Role:  "user",
	}
}

// NewUserContentFromText builds a Content with a "user" role from a single text string.
func NewUserContentFromText(text string) *Content {
	return &Content{
		Parts: []*Part{
			NewPartFromText(text),
		},
		Role: "user",
	}
}

// NewUserContentFromBytes builds a Content with a "user" role from a single byte array.
func NewUserContentFromBytes(data []byte, mimeType string) *Content {
	return &Content{
		Parts: []*Part{
			NewPartFromBytes(data, mimeType),
		},
		Role: "user",
	}
}

// NewUserContentFromURI builds a Content with a "user" role from a single file URI.
func NewUserContentFromURI(fileURI, mimeType string) *Content {
	return &Content{
		Parts: []*Part{
			NewPartFromURI(fileURI, mimeType),
		},
		Role: "user",
	}
}

// NewUserContentFromFunctionResponse builds a Content with a "user" role from a single function response.
func NewUserContentFromFunctionResponse(name string, response map[string]any) *Content {
	return &Content{
		Parts: []*Part{
			NewPartFromFunctionResponse(name, response),
		},
		Role: "user",
	}
}

// NewUserContentFromExecutableCode builds a Content with a "user" role from a single executable code.
func NewUserContentFromExecutableCode(code string, language Language) *Content {
	return &Content{
		Parts: []*Part{
			NewPartFromExecutableCode(code, language),
		},
		Role: "user",
	}
}

// NewUserContentFromCodeExecutionResult builds a Content with a "user" role from a single code execution result.
func NewUserContentFromCodeExecutionResult(outcome Outcome, output string) *Content {
	return &Content{
		Parts: []*Part{
			NewPartFromCodeExecutionResult(outcome, output),
		},
		Role: "user",
	}
}

// NewModelContent builds a Content with a "model" role from a list of parts.
func NewModelContentFromParts(parts []*Part) *Content {
	return &Content{
		Parts: parts,
		Role:  "model",
	}
}

// NewModelContentFromText builds a Content with a "model" role from a single text string.
func NewModelContentFromText(text string) *Content {
	return &Content{
		Parts: []*Part{
			NewPartFromText(text),
		},
		Role: "model",
	}
}

// NewModelContentFromBytes builds a Content with a "model" role from a single byte array.
func NewModelContentFromBytes(data []byte, mimeType string) *Content {
	return &Content{
		Parts: []*Part{
			NewPartFromBytes(data, mimeType),
		},
		Role: "model",
	}
}

// NewModelContentFromURI builds a Content with a "model" role from a single file URI.
func NewModelContentFromURI(fileURI, mimeType string) *Content {
	return &Content{
		Parts: []*Part{
			NewPartFromURI(fileURI, mimeType),
		},
		Role: "model",
	}
}

// NewModelContentFromFunctionCall builds a Content with a "model" role from a single function call.
func NewModelContentFromFunctionCall(name string, args map[string]any) *Content {
	return &Content{
		Parts: []*Part{
			NewPartFromFunctionCall(name, args),
		},
		Role: "model",
	}
}

// NewModelContentFromExecutableCode builds a Content with a "model" role from a single executable code.
func NewModelContentFromExecutableCode(code string, language Language) *Content {
	return &Content{
		Parts: []*Part{
			NewPartFromExecutableCode(code, language),
		},
		Role: "model",
	}
}

// NewModelContentFromCodeExecutionResult builds a Content with a "model" role from a single code execution result.
func NewModelContentFromCodeExecutionResult(outcome Outcome, output string) *Content {
	return &Content{
		Parts: []*Part{
			NewPartFromCodeExecutionResult(outcome, output),
		},
		Role: "model",
	}
}

// HTTP options to be used in each of the requests.
type HTTPOptions struct {
	// BaseURL specifies the base URL for the API endpoint. If unset, defaults to "https://generativelanguage.googleapis.com/"
	// for the Gemini API backend, and location-specific Vertex AI endpoint (e.g., "https://us-central1-aiplatform.googleapis.com/
	BaseURL string `json:"baseUrl,omitempty"`
	// APIVersion specifies the version of the API to use.
	APIVersion string `json:"apiVersion,omitempty"`
	// Timeout sets the timeout for HTTP requests in milliseconds. If unset, defaults to
	// "v1beta" for the Gemini API, and "v1beta1" for the Vertex AI.
	Timeout int64 `json:"timeout,omitempty"`
}

// Schema that defines the format of input and output data. Represents a select subset
// of an OpenAPI 3.0 schema object. You can find more details and examples at https://spec.openapis.org/oas/v3.0.3.html#schema-object
type Schema struct {
	// Optional. Minimum number of the elements for Type.ARRAY.
	MinItems *int64 `json:"minItems,omitempty"`
	// Optional. Example of the object. Will only populated when the object is the root.
	Example any `json:"example,omitempty"`
	// Optional. The order of the properties. Not a standard field in open API spec. Only
	// used to support the order of the properties.
	PropertyOrdering []string `json:"propertyOrdering,omitempty"`
	// Optional. Pattern of the Type.STRING to restrict a string to a regular expression.
	Pattern string `json:"pattern,omitempty"`
	// Optional. SCHEMA FIELDS FOR TYPE INTEGER and NUMBER Minimum value of the Type.INTEGER
	// and Type.NUMBER
	Minimum *float64 `json:"minimum,omitempty"`
	// Optional. Default value of the data.
	Default any `json:"default,omitempty"`
	// Optional. The value should be validated against any (one or more) of the subschemas
	// in the list.
	AnyOf []*Schema `json:"anyOf,omitempty"`
	// Optional. Maximum length of the Type.STRING
	MaxLength *int64 `json:"maxLength,omitempty"`
	// Optional. The title of the Schema.
	Title string `json:"title,omitempty"`
	// Optional. SCHEMA FIELDS FOR TYPE STRING Minimum length of the Type.STRING
	MinLength *int64 `json:"minLength,omitempty"`
	// Optional. Minimum number of the properties for Type.OBJECT.
	MinProperties *int64 `json:"minProperties,omitempty"`
	// Optional. Maximum number of the elements for Type.ARRAY.
	MaxItems *int64 `json:"maxItems,omitempty"`
	// Optional. Maximum value of the Type.INTEGER and Type.NUMBER
	Maximum *float64 `json:"maximum,omitempty"`
	// Optional. Indicates if the value may be null.
	Nullable bool `json:"nullable,omitempty"`
	// Optional. Maximum number of the properties for Type.OBJECT.
	MaxProperties *int64 `json:"maxProperties,omitempty"`
	// Optional. The type of the data.
	Type Type `json:"type,omitempty"`
	// Optional. The description of the data.
	Description string `json:"description,omitempty"`
	// Optional. Possible values of the element of primitive type with enum format. Examples:
	// 1. We can define direction as : {type:STRING, format:enum, enum:["EAST", NORTH",
	// "SOUTH", "WEST"]} 2. We can define apartment number as : {type:INTEGER, format:enum,
	// enum:["101", "201", "301"]}
	Enum []string `json:"enum,omitempty"`
	// Optional. The format of the data. Supported formats: for NUMBER type: "float", "double"
	// for INTEGER type: "int32", "int64" for STRING type: "email", "byte", etc
	Format string `json:"format,omitempty"`
	// Optional. SCHEMA FIELDS FOR TYPE ARRAY Schema of the elements of Type.ARRAY.
	Items *Schema `json:"items,omitempty"`
	// Optional. SCHEMA FIELDS FOR TYPE OBJECT Properties of Type.OBJECT.
	Properties map[string]*Schema `json:"properties,omitempty"`
	// Optional. Required properties of Type.OBJECT.
	Required []string `json:"required,omitempty"`
}

// Safety settings.
type SafetySetting struct {
	// Determines if the harm block method uses probability or probability
	// and severity scores.
	Method HarmBlockMethod `json:"method,omitempty"`
	// Required. Harm category.
	Category HarmCategory `json:"category,omitempty"`
	// Required. The harm block threshold.
	Threshold HarmBlockThreshold `json:"threshold,omitempty"`
}

// Defines a function that the model can generate JSON inputs for.
// The inputs are based on `OpenAPI 3.0 specifications
// <https://spec.openapis.org/oas/v3.0.3>`_.
type FunctionDeclaration struct {
	// Describes the output from the function in the OpenAPI JSON Schema
	// Object format.
	Response *Schema `json:"response,omitempty"`
	// Optional. Description and purpose of the function. Model uses it to decide how and
	// whether to call the function.
	Description string `json:"description,omitempty"`
	// Required. The name of the function to call. Must start with a letter or an underscore.
	// Must be a-z, A-Z, 0-9, or contain underscores, dots and dashes, with a maximum length
	// of 64.
	Name string `json:"name,omitempty"`
	// Optional. Describes the parameters to this function in JSON Schema Object format.
	// Reflects the Open API 3.03 Parameter Object. string Key: the name of the parameter.
	// Parameter names are case sensitive. Schema Value: the Schema defining the type used
	// for the parameter. For function with no parameters, this can be left unset. Parameter
	// names must start with a letter or an underscore and must only contain chars a-z,
	// A-Z, 0-9, or underscores with a maximum length of 64. Example with 1 required and
	// 1 optional parameter: type: OBJECT properties: param1: type: STRING param2: type:
	// INTEGER required: - param1
	Parameters *Schema `json:"parameters,omitempty"`
}

// Tool to support Google Search in Model. Powered by Google.
type GoogleSearch struct {
}

// Describes the options to customize dynamic retrieval.
type DynamicRetrievalConfig struct {
	// The mode of the predictor to be used in dynamic retrieval.
	Mode DynamicRetrievalConfigMode `json:"mode,omitempty"`
	// Optional. The threshold to be used in dynamic retrieval. If not set, a system default
	// value is used.
	DynamicThreshold *float64 `json:"dynamicThreshold,omitempty"`
}

// Tool to retrieve public web data for grounding, powered by Google.
type GoogleSearchRetrieval struct {
	// Specifies the dynamic retrieval configuration for the given source.
	DynamicRetrievalConfig *DynamicRetrievalConfig `json:"dynamicRetrievalConfig,omitempty"`
}

// Retrieve from Vertex AI Search datastore for grounding. See https://cloud.google.com/products/agent-builder
type VertexAISearch struct {
	// Required. Fully-qualified Vertex AI Search data store resource ID. Format: `projects/{project}/locations/{location}/collections/{collection}/dataStores/{dataStore}`
	Datastore string `json:"datastore,omitempty"`
}

// The definition of the RAG resource.
type VertexRAGStoreRAGResource struct {
	// Optional. RAGCorpora resource name. Format: `projects/{project}/locations/{location}/ragCorpora/{rag_corpus}`
	RAGCorpus string `json:"ragCorpus,omitempty"`
	// Optional. rag_file_id. The files should be in the same rag_corpus set in rag_corpus
	// field.
	RAGFileIDs []string `json:"ragFileIds,omitempty"`
}

// Retrieve from Vertex RAG Store for grounding. You can find API default values and
// more details at https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/rag-api-v1#parameters-list
type VertexRAGStore struct {
	// Optional. Deprecated. Please use rag_resources instead.
	RAGCorpora []string `json:"ragCorpora,omitempty"`
	// Optional. The representation of the RAG source. It can be used to specify corpus
	// only or ragfiles. Currently only support one corpus or multiple files from one corpus.
	// In the future we may open up multiple corpora support.
	RAGResources []*VertexRAGStoreRAGResource `json:"ragResources,omitempty"`
	// Optional. Number of top k results to return from the selected corpora.
	SimilarityTopK *int64 `json:"similarityTopK,omitempty"`
	// Optional. Only return results with vector distance smaller than the threshold.
	VectorDistanceThreshold *float64 `json:"vectorDistanceThreshold,omitempty"`
}

// Defines a retrieval tool that model can call to access external knowledge.
type Retrieval struct {
	// Optional. Deprecated. This option is no longer supported.
	DisableAttribution bool `json:"disableAttribution,omitempty"`
	// Set to use data source powered by Vertex AI Search.
	VertexAISearch *VertexAISearch `json:"vertexAiSearch,omitempty"`
	// Set to use data source powered by Vertex RAG store. User data is uploaded via the
	// VertexRAGDataService.
	VertexRAGStore *VertexRAGStore `json:"vertexRagStore,omitempty"`
}

// Tool that executes code generated by the model, and automatically returns the result
// to the model. See also [ExecutableCode]and [CodeExecutionResult] which are input
// and output to this tool.
type ToolCodeExecution struct {
}

// Tool details of a tool that the model may use to generate a response.
type Tool struct {
	// List of function declarations that the tool supports.
	FunctionDeclarations []*FunctionDeclaration `json:"functionDeclarations,omitempty"`
	// Optional. Retrieval tool type. System will always execute the provided retrieval
	// tool(s) to get external knowledge to answer the prompt. Retrieval results are presented
	// to the model for generation.
	Retrieval *Retrieval `json:"retrieval,omitempty"`
	// Optional. Google Search tool type. Specialized retrieval tool
	// that is powered by Google Search.
	GoogleSearch *GoogleSearch `json:"googleSearch,omitempty"`
	// Optional. GoogleSearchRetrieval tool type. Specialized retrieval tool that is powered
	// by Google search.
	GoogleSearchRetrieval *GoogleSearchRetrieval `json:"googleSearchRetrieval,omitempty"`
	// Optional. CodeExecution tool type. Enables the model to execute code as part of generation.
	// This field is only used by the Gemini Developer API services.
	CodeExecution *ToolCodeExecution `json:"codeExecution,omitempty"`
}

// Function calling config.
type FunctionCallingConfig struct {
	// Optional. Function calling mode.
	Mode FunctionCallingConfigMode `json:"mode,omitempty"`
	// Optional. Function names to call. Only set when the Mode is ANY. Function names should
	// match [FunctionDeclaration.name]. With mode set to ANY, model will predict a function
	// call from the set of function names provided.
	AllowedFunctionNames []string `json:"allowedFunctionNames,omitempty"`
}

// Tool config.
// This config is shared for all tools provided in the request.
type ToolConfig struct {
	// Optional. Function calling config.
	FunctionCallingConfig *FunctionCallingConfig `json:"functionCallingConfig,omitempty"`
}

// The configuration for the prebuilt speaker to use.
type PrebuiltVoiceConfig struct {
	// The name of the prebuilt voice to use.
	VoiceName string `json:"voiceName,omitempty"`
}

// The configuration for the voice to use.
type VoiceConfig struct {
	// The configuration for the speaker to use.
	PrebuiltVoiceConfig *PrebuiltVoiceConfig `json:"prebuiltVoiceConfig,omitempty"`
}

// The speech generation configuration.
type SpeechConfig struct {
	// The configuration for the speaker to use.
	VoiceConfig *VoiceConfig `json:"voiceConfig,omitempty"`
}

// The thinking features configuration.
type ThinkingConfig struct {
	// Indicates whether to include thoughts in the response. If true, thoughts are returned
	// only if the model supports thought and thoughts are available.
	IncludeThoughts bool `json:"includeThoughts,omitempty"`
}

// When automated routing is specified, the routing will be determined by the pretrained
// routing model and customer provided model routing preference.
type GenerationConfigRoutingConfigAutoRoutingMode struct {
	// The model routing preference.
	ModelRoutingPreference string `json:"modelRoutingPreference,omitempty"`
}

// When manual routing is set, the specified model will be used directly.
type GenerationConfigRoutingConfigManualRoutingMode struct {
	// The model name to use. Only the public LLM models are accepted. e.g. 'gemini-1.5-pro-001'.
	ModelName string `json:"modelName,omitempty"`
}

// The configuration for routing the request to a specific model.
type GenerationConfigRoutingConfig struct {
	// Automated routing.
	AutoMode *GenerationConfigRoutingConfigAutoRoutingMode `json:"autoMode,omitempty"`
	// Manual routing.
	ManualMode *GenerationConfigRoutingConfigManualRoutingMode `json:"manualMode,omitempty"`
}

// Optional configuration for the GenerateContent. You can find API default values and
// more details at https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/inference#generationconfig
// and https://cloud.google.com/vertex-ai/generative-ai/docs/multimodal/content-generation-parameters.
type GenerateContentConfig struct {
	// Instructions for the model to steer it toward better performance.
	// For example, "Answer as concisely as possible" or "Don't use technical
	// terms in your response".
	SystemInstruction *Content `json:"systemInstruction,omitempty"`
	// Value that controls the degree of randomness in token selection.
	// Lower temperatures are good for prompts that require a less open-ended or
	// creative response, while higher temperatures can lead to more diverse or
	// creative results.
	Temperature *float64 `json:"temperature,omitempty"`
	// Tokens are selected from the most to least probable until the sum
	// of their probabilities equals this value. Use a lower value for less
	// random responses and a higher value for more random responses.
	TopP *float64 `json:"topP,omitempty"`
	// For each token selection step, the ``top_k`` tokens with the
	// highest probabilities are sampled. Then tokens are further filtered based
	// on ``top_p`` with the final token selected using temperature sampling. Use
	// a lower number for less random responses and a higher number for more
	// random responses.
	TopK *float64 `json:"topK,omitempty"`
	// Number of response variations to return.
	CandidateCount *int64 `json:"candidateCount,omitempty"`
	// Maximum number of tokens that can be generated in the response.
	MaxOutputTokens *int64 `json:"maxOutputTokens,omitempty"`
	// List of strings that tells the model to stop generating text if one
	// of the strings is encountered in the response.
	StopSequences []string `json:"stopSequences,omitempty"`
	// Whether to return the log probabilities of the tokens that were
	// chosen by the model at each step.
	ResponseLogprobs bool `json:"responseLogprobs,omitempty"`
	// Number of top candidate tokens to return the log probabilities for
	// at each generation step.
	Logprobs *int64 `json:"logprobs,omitempty"`
	// Positive values penalize tokens that already appear in the
	// generated text, increasing the probability of generating more diverse
	// content.
	PresencePenalty *float64 `json:"presencePenalty,omitempty"`
	// Positive values penalize tokens that repeatedly appear in the
	// generated text, increasing the probability of generating more diverse
	// content.
	FrequencyPenalty *float64 `json:"frequencyPenalty,omitempty"`
	// When ``seed`` is fixed to a specific number, the model makes a best
	// effort to provide the same response for repeated requests. By default, a
	// random number is used.
	Seed *int64 `json:"seed,omitempty"`
	// Output response media type of the generated candidate text.
	ResponseMIMEType string `json:"responseMimeType,omitempty"`
	// Schema that the generated candidate text must adhere to.
	ResponseSchema *Schema `json:"responseSchema,omitempty"`
	// Configuration for model router requests.
	RoutingConfig *GenerationConfigRoutingConfig `json:"routingConfig,omitempty"`
	// Safety settings in the request to block unsafe content in the
	// response.
	SafetySettings []*SafetySetting `json:"safetySettings,omitempty"`
	// Code that enables the system to interact with external systems to
	// perform an action outside of the knowledge and scope of the model.
	Tools []*Tool `json:"tools,omitempty"`
	// Associates model output to a specific function call.
	ToolConfig *ToolConfig `json:"toolConfig,omitempty"`
	// Labels with user-defined metadata to break down billed charges.
	Labels map[string]string `json:"labels,omitempty"`
	// Resource name of a context cache that can be used in subsequent
	// requests.
	CachedContent string `json:"cachedContent,omitempty"`
	// The requested modalities of the response. Represents the set of
	// modalities that the model can return.
	ResponseModalities []string `json:"responseModalities,omitempty"`
	// If specified, the media resolution specified will be used.
	MediaResolution MediaResolution `json:"mediaResolution,omitempty"`
	// The speech generation configuration.
	SpeechConfig *SpeechConfig `json:"speechConfig,omitempty"`
	// If enabled, audio timestamp will be included in the request to the
	// model.
	AudioTimestamp bool `json:"audioTimestamp,omitempty"`
	// The thinking features configuration.
	ThinkingConfig *ThinkingConfig `json:"thinkingConfig,omitempty"`
}

// Config for models.generate_content parameters.
type GenerateContentParameters struct {
	// ID of the model to use. For a list of models, see `Google models
	// <https://cloud.google.com/vertex-ai/generative-ai/docs/learn/models>`_.
	Model string `json:"model,omitempty"`
	// Content of the request.
	Contents []*Content `json:"contents,omitempty"`
	// Configuration that contains optional model parameters.
	Config *GenerateContentConfig `json:"config,omitempty"`
}

// Source attributions for content.
type Citation struct {
	// Output only. End index into the content.
	EndIndex int64 `json:"endIndex,omitempty"`
	// Output only. License of the attribution.
	License string `json:"license,omitempty"`
	// Output only. Publication date of the attribution.
	PublicationDate *civil.Date `json:"publicationDate,omitempty"`
	// Output only. Start index into the content.
	StartIndex int64 `json:"startIndex,omitempty"`
	// Output only. Title of the attribution.
	Title string `json:"title,omitempty"`
	// Output only. URL reference of the attribution.
	URI string `json:"uri,omitempty"`
}

// UnmarshalJSON custom unmarshalling to handle PublicationDate as a map containing year, month, and day.
func (c *Citation) UnmarshalJSON(data []byte) error {
	type Alias Citation
	aux := &struct {
		PublicationDate map[string]int `json:"publicationDate"`
		*Alias
	}{
		Alias: (*Alias)(c),
	}

	if err := json.Unmarshal(data, &aux); err != nil {
		return err
	}

	if aux.PublicationDate != nil {
		var year, month, day int
		var ok bool
		if year, ok = aux.PublicationDate["year"]; !ok {
			return fmt.Errorf("key %q not found", "year")
		}
		if month, ok = aux.PublicationDate["month"]; !ok {
			return fmt.Errorf("key %q not found", "month")
		}
		if day, ok = aux.PublicationDate["day"]; !ok {
			return fmt.Errorf("key %q not found", "day")
		}

		c.PublicationDate = &civil.Date{Year: year, Month: time.Month(month), Day: day}
	}

	return nil
}

// Citation information when the model quotes another source.
type CitationMetadata struct {
	// Contains citation information when the model directly quotes, at
	// length, from another source. Can include traditional websites and code
	// repositories.
	Citations []*Citation `json:"citations,omitempty"`
}

// Chunk from context retrieved by the retrieval tools.
type GroundingChunkRetrievedContext struct {
	// Text of the attribution.
	Text string `json:"text,omitempty"`
	// Title of the attribution.
	Title string `json:"title,omitempty"`
	// URI reference of the attribution.
	URI string `json:"uri,omitempty"`
}

// Chunk from the web.
type GroundingChunkWeb struct {
	// Title of the chunk.
	Title string `json:"title,omitempty"`
	// URI reference of the chunk.
	URI string `json:"uri,omitempty"`
}

// Grounding chunk.
type GroundingChunk struct {
	// Grounding chunk from context retrieved by the retrieval tools.
	RetrievedContext *GroundingChunkRetrievedContext `json:"retrievedContext,omitempty"`
	// Grounding chunk from the web.
	Web *GroundingChunkWeb `json:"web,omitempty"`
}

// Segment of the content.
type Segment struct {
	// Output only. End index in the given Part, measured in bytes. Offset from the start
	// of the Part, exclusive, starting at zero.
	EndIndex int64 `json:"endIndex,omitempty"`
	// Output only. The index of a Part object within its parent Content object.
	PartIndex int64 `json:"partIndex,omitempty"`
	// Output only. Start index in the given Part, measured in bytes. Offset from the start
	// of the Part, inclusive, starting at zero.
	StartIndex int64 `json:"startIndex,omitempty"`
	// Output only. The text corresponding to the segment from the response.
	Text string `json:"text,omitempty"`
}

// Grounding support.
type GroundingSupport struct {
	// Confidence score of the support references. Ranges from 0 to 1. 1 is the most confident.
	// This list must have the same size as the grounding_chunk_indices.
	ConfidenceScores []float64 `json:"confidenceScores,omitempty"`
	// A list of indices (into 'grounding_chunk') specifying the citations associated with
	// the claim. For instance [1,3,4] means that grounding_chunk[1], grounding_chunk[3],
	// grounding_chunk[4] are the retrieved content attributed to the claim.
	GroundingChunkIndices []int64 `json:"groundingChunkIndices,omitempty"`
	// Segment of the content this support belongs to.
	Segment *Segment `json:"segment,omitempty"`
}

// Metadata related to retrieval in the grounding flow.
type RetrievalMetadata struct {
	// Optional. Score indicating how likely information from Google Search could help answer
	// the prompt. The score is in the range `[0, 1]`, where 0 is the least likely and 1
	// is the most likely. This score is only populated when Google Search grounding and
	// dynamic retrieval is enabled. It will be compared to the threshold to determine whether
	// to trigger Google Search. If nil, then API will determine the default value.
	GoogleSearchDynamicRetrievalScore *float64 `json:"googleSearchDynamicRetrievalScore,omitempty"`
}

// Google search entry point.
type SearchEntryPoint struct {
	// Optional. Web content snippet that can be embedded in a web page or an app webview.
	RenderedContent string `json:"renderedContent,omitempty"`
	// Optional. Base64 encoded JSON representing array of tuple.
	SDKBlob []byte `json:"sdkBlob,omitempty"`
}

// Metadata returned to client when grounding is enabled.
type GroundingMetadata struct {
	// List of supporting references retrieved from specified grounding source.
	GroundingChunks []*GroundingChunk `json:"groundingChunks,omitempty"`
	// Optional. List of grounding support.
	GroundingSupports []*GroundingSupport `json:"groundingSupports,omitempty"`
	// Optional. Output only. Retrieval metadata.
	RetrievalMetadata *RetrievalMetadata `json:"retrievalMetadata,omitempty"`
	// Optional. Queries executed by the retrieval tools.
	RetrievalQueries []string `json:"retrievalQueries,omitempty"`
	// Optional. Google search entry for the following-up web searches.
	SearchEntryPoint *SearchEntryPoint `json:"searchEntryPoint,omitempty"`
	// Optional. Web search queries for the following-up web search.
	WebSearchQueries []string `json:"webSearchQueries,omitempty"`
}

// Candidate for the logprobs token and score.
type LogprobsResultCandidate struct {
	// The candidate's log probability. If nil, then no LogProbability is returned by the
	// API.
	LogProbability *float64 `json:"logProbability,omitempty"`
	// The candidate's token string value.
	Token string `json:"token,omitempty"`
	// The candidate's token ID value. If nil, then no TokenID is returned by the API.
	TokenID *int64 `json:"tokenId,omitempty"`
}

// Candidates with top log probabilities at each decoding step.
type LogprobsResultTopCandidates struct {
	// Sorted by log probability in descending order.
	Candidates []*LogprobsResultCandidate `json:"candidates,omitempty"`
}

// Logprobs Result
type LogprobsResult struct {
	// Length = total number of decoding steps. The chosen candidates may or may not be
	// in top_candidates.
	ChosenCandidates []*LogprobsResultCandidate `json:"chosenCandidates,omitempty"`
	// Length = total number of decoding steps.
	TopCandidates []*LogprobsResultTopCandidates `json:"topCandidates,omitempty"`
}

// Safety rating corresponding to the generated content.
type SafetyRating struct {
	// Output only. Indicates whether the content was filtered out because of this rating.
	Blocked bool `json:"blocked,omitempty"`
	// Output only. Harm category.
	Category HarmCategory `json:"category,omitempty"`
	// Output only. Harm probability levels in the content.
	Probability HarmProbability `json:"probability,omitempty"`
	// Output only. Harm probability score. If nil, then no ProbabilityScore is returned
	// by the API.
	ProbabilityScore *float64 `json:"probabilityScore,omitempty"`
	// Output only. Harm severity levels in the content.
	Severity HarmSeverity `json:"severity,omitempty"`
	// Output only. Harm severity score. If nil, then no ProbabilityScore is returned by
	// the API.
	SeverityScore *float64 `json:"severityScore,omitempty"`
}

// A response candidate generated from the model.
type Candidate struct {
	// Contains the multi-part content of the response.
	Content *Content `json:"content,omitempty"`
	// Source attribution of the generated content.
	CitationMetadata *CitationMetadata `json:"citationMetadata,omitempty"`
	// Describes the reason the model stopped generating tokens.
	FinishMessage string `json:"finishMessage,omitempty"`
	// Number of tokens for this candidate. If nil, then no TokenCount is returned by the
	// API.
	TokenCount *int64 `json:"tokenCount,omitempty"`
	// Output only. Average log probability score of the candidate. If nil, then no AvgLogprobs
	// is returned by the API.
	AvgLogprobs *float64 `json:"avgLogprobs,omitempty"`
	// Output only. The reason why the model stopped generating tokens. If empty, the model
	// has not stopped generating the tokens.
	FinishReason FinishReason `json:"finishReason,omitempty"`
	// Output only. Metadata specifies sources used to ground generated content.
	GroundingMetadata *GroundingMetadata `json:"groundingMetadata,omitempty"`
	// Output only. Index of the candidate. If nil, then no Index is returned by the API.
	Index *int64 `json:"index,omitempty"`
	// Output only. Log-likelihood scores for the response tokens and top tokens
	LogprobsResult *LogprobsResult `json:"logprobsResult,omitempty"`
	// Output only. List of ratings for the safety of a response candidate. There is at
	// most one rating per category.
	SafetyRatings []*SafetyRating `json:"safetyRatings,omitempty"`
}

// Content filter results for a prompt sent in the request.
type GenerateContentResponsePromptFeedback struct {
	// Output only. Blocked reason.
	BlockReason BlockedReason `json:"blockReason,omitempty"`
	// Output only. A readable block reason message.
	BlockReasonMessage string `json:"blockReasonMessage,omitempty"`
	// Output only. Safety ratings.
	SafetyRatings []*SafetyRating `json:"safetyRatings,omitempty"`
}

// Usage metadata about response(s).
type GenerateContentResponseUsageMetadata struct {
	// Output only. Number of tokens in the cached part in the input (the cached content).
	// If nil, then no CachedContentTokenCount is returned by the API.
	CachedContentTokenCount *int64 `json:"cachedContentTokenCount,omitempty"`
	// Number of tokens in the response(all the generated response candidates). If nil,
	// then no CandidatesTokenCount is returned by the API.
	CandidatesTokenCount *int64 `json:"candidatesTokenCount,omitempty"`
	// Number of tokens in the prompt. When cached_content is set, this is still the total
	// effective prompt size meaning this includes the number of tokens in the cached content.
	// If nil, then no PromptTokenCount is returned by the API.
	PromptTokenCount *int64 `json:"promptTokenCount,omitempty"`
	// Total token count for prompt and response candidates.
	TotalTokenCount int64 `json:"totalTokenCount,omitempty"`
}

// Response message for PredictionService.GenerateContent.
type GenerateContentResponse struct {
	// Response variations returned by the model.
	Candidates []*Candidate `json:"candidates,omitempty"`
	// Output only. The model version used to generate the response.
	ModelVersion string `json:"modelVersion,omitempty"`
	// Output only. Content filter results for a prompt sent in the request. Note: Sent
	// only in the first stream chunk. Only happens when no candidates were generated due
	// to content violations.
	PromptFeedback *GenerateContentResponsePromptFeedback `json:"promptFeedback,omitempty"`
	// Usage metadata about the response(s).
	UsageMetadata *GenerateContentResponseUsageMetadata `json:"usageMetadata,omitempty"`
}

// Text concatenates all the text parts in the GenerateContentResponse.
func (r *GenerateContentResponse) Text() (string, error) {
	if len(r.Candidates) == 0 || r.Candidates[0].Content == nil || len(r.Candidates[0].Content.Parts) == 0 {
		return "", nil
	}

	if len(r.Candidates) > 1 {
		log.Printf("Warning: there are multiple candidates in the response, returning text from the first one.")
	}

	var texts []string
	for _, part := range r.Candidates[0].Content.Parts {
		if part.Text != "" {
			if part.Thought {
				continue
			}
			texts = append(texts, part.Text)
		} else {
			if part.InlineData != nil {
				return "", errors.New("GenerateContentResponse.Text only supports text parts, but got InlineData")
			}
			if part.CodeExecutionResult != nil {
				return "", errors.New("GenerateContentResponse.Text only supports text parts, but got CodeExecutionResult")
			}
			if part.ExecutableCode != nil {
				return "", errors.New("GenerateContentResponse.Text only supports text parts, but got ExecutableCode")
			}
			if part.FileData != nil {
				return "", errors.New("GenerateContentResponse.Text only supports text parts, but got FileData")
			}
			if part.FunctionCall != nil {
				return "", errors.New("GenerateContentResponse.Text only supports text parts, but got FunctionCall")
			}
			if part.FunctionResponse != nil {
				return "", errors.New("GenerateContentResponse.Text only supports text parts, but got FunctionResponse")
			}
		}
	}

	if len(texts) == 0 {
		return "", nil
	}

	return strings.Join(texts, ""), nil
}

// FunctionCalls returns the list of function calls in the GenerateContentResponse.
func (r *GenerateContentResponse) FunctionCalls() []*FunctionCall {
	if len(r.Candidates) == 0 || r.Candidates[0].Content == nil || len(r.Candidates[0].Content.Parts) == 0 {
		return nil
	}

	if len(r.Candidates) > 1 {
		log.Printf("Warning: there are multiple candidates in the response, returning function calls from the first one.")
	}

	var functionCalls []*FunctionCall
	for _, part := range r.Candidates[0].Content.Parts {
		if part.FunctionCall != nil {
			functionCalls = append(functionCalls, part.FunctionCall)
		}
	}

	if len(functionCalls) == 0 {
		return nil
	}

	return functionCalls
}

// The configuration for generating images. You can find API default values and more
// details at
// VertexAI: https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/imagen-api.
// GeminiAPI: https://ai.google.dev/gemini-api/docs/imagen#imagen-model
type GenerateImagesConfig struct {
	// Cloud Storage URI used to store the generated images.
	OutputGCSURI string `json:"outputGcsUri,omitempty"`
	// Description of what to discourage in the generated images.
	NegativePrompt string `json:"negativePrompt,omitempty"`
	// Number of images to generate.
	NumberOfImages *int64 `json:"numberOfImages,omitempty"`
	// Controls how much the model adheres to the text prompt. Large
	// values increase output and prompt alignment, but may compromise image
	// quality.
	GuidanceScale *float64 `json:"guidanceScale,omitempty"`
	// Random seed for image generation. This is not available when
	// ``add_watermark`` is set to true.
	Seed *int64 `json:"seed,omitempty"`
	// Filter level for safety filtering.
	SafetyFilterLevel SafetyFilterLevel `json:"safetyFilterLevel,omitempty"`
	// Allows generation of people by the model.
	PersonGeneration PersonGeneration `json:"personGeneration,omitempty"`
	// Whether to report the safety scores of each image in the response.
	IncludeSafetyAttributes bool `json:"includeSafetyAttributes,omitempty"`
	// Whether to include the Responsible AI filter reason if the image
	// is filtered out of the response.
	IncludeRAIReason bool `json:"includeRaiReason,omitempty"`
	// Language of the text in the prompt.
	Language ImagePromptLanguage `json:"language,omitempty"`
	// MIME type of the generated image.
	OutputMIMEType string `json:"outputMimeType,omitempty"`
	// Compression quality of the generated image (for ``image/jpeg``
	// only).
	OutputCompressionQuality *int64 `json:"outputCompressionQuality,omitempty"`
	// Whether to add a watermark to the generated images.
	AddWatermark bool `json:"addWatermark,omitempty"`
	// Aspect ratio of the generated images.
	AspectRatio string `json:"aspectRatio,omitempty"`
	// Whether to use the prompt rewriting logic.
	EnhancePrompt bool `json:"enhancePrompt,omitempty"`
}

// The parameters for generating images.
type GenerateImagesParameters struct {
	// ID of the model to use. For a list of models, see `Google models
	// <https://cloud.google.com/vertex-ai/generative-ai/docs/learn/models>`_.
	Model string `json:"model,omitempty"`
	// Text prompt that typically describes the images to output.
	Prompt string `json:"prompt,omitempty"`
	// Configuration for generating images.
	Config *GenerateImagesConfig `json:"config,omitempty"`
}

// An image.
type Image struct {
	// The Cloud Storage URI of the image. ``Image`` can contain a value
	// for this field or the ``image_bytes`` field but not both.
	GCSURI string `json:"gcsUri,omitempty"`
	// The image bytes data. ``Image`` can contain a value for this field
	// or the ``gcs_uri`` field but not both.
	ImageBytes []byte `json:"imageBytes,omitempty"`
	// The MIME type of the image.
	MIMEType string `json:"mimeType,omitempty"`
}

// An output image.
type GeneratedImage struct {
	// The output image data.
	Image *Image `json:"image,omitempty"`
	// Responsible AI filter reason if the image is filtered out of the
	// response.
	RAIFilteredReason string `json:"raiFilteredReason,omitempty"`
	// The rewritten prompt used for the image generation if the prompt
	// enhancer is enabled.
	EnhancedPrompt string `json:"enhancedPrompt,omitempty"`
}

// The output images response.
type GenerateImagesResponse struct {
	// List of generated images.
	GeneratedImages []*GeneratedImage `json:"generatedImages,omitempty"`
}

// API config for UpscaleImage with fields not exposed to users.
// These fields require default values sent to the API which are not intended
// to be modifiable or exposed to users in the SDK method.
type upscaleImageAPIConfig struct {
	// Whether to include a reason for filtered-out images in the
	// response.
	IncludeRAIReason bool `json:"includeRaiReason,omitempty"`
	// The image format that the output should be saved as.
	OutputMIMEType string `json:"outputMimeType,omitempty"`
	// The level of compression if the ``output_mime_type`` is
	// ``image/jpeg``.
	OutputCompressionQuality *int64 `json:"outputCompressionQuality,omitempty"`

	NumberOfImages int64 `json:"numberOfImages,omitempty"`

	Mode string `json:"mode,omitempty"`
}

// API parameters for UpscaleImage.
type UpscaleImageAPIParameters struct {
	// The model to use.
	Model string `json:"model,omitempty"`
	// The input image to upscale.
	Image *Image `json:"image,omitempty"`
	// The factor to upscale the image (x2 or x4).
	UpscaleFactor string `json:"upscaleFactor,omitempty"`
	// Configuration for upscaling.
	Config *upscaleImageAPIConfig `json:"config,omitempty"`
}

type UpscaleImageResponse struct {
	// Generated images.
	GeneratedImages []*GeneratedImage `json:"generatedImages,omitempty"`
}

// Optional parameters for models.get method.
type GetModelConfig struct {
}

type GetModelParameters struct {
	Model string `json:"model,omitempty"`
	// Optional parameters for the request.
	Config *GetModelConfig `json:"config,omitempty"`
}

// An endpoint where you deploy models.
type Endpoint struct {
	// Resource name of the endpoint.
	Name string `json:"name,omitempty"`
	// ID of the model that's deployed to the endpoint.
	DeployedModelID string `json:"deployedModelId,omitempty"`
}

// A tuned machine learning model.
type TunedModelInfo struct {
	// ID of the base model that you want to tune.
	BaseModel string `json:"baseModel,omitempty"`
	// Date and time when the base model was created.
	CreateTime *time.Time `json:"createTime,omitempty"`
	// Date and time when the base model was last updated.
	UpdateTime *time.Time `json:"updateTime,omitempty"`
}

// A trained machine learning model.
type Model struct {
	// Resource name of the model.
	Name string `json:"name,omitempty"`
	// Display name of the model.
	DisplayName string `json:"displayName,omitempty"`
	// Description of the model.
	Description string `json:"description,omitempty"`
	// Version ID of the model. A new version is committed when a new
	// model version is uploaded or trained under an existing model ID. The
	// version ID is an auto-incrementing decimal number in string
	// representation.
	Version string `json:"version,omitempty"`
	// List of deployed models created from this base model. Note that a
	// model could have been deployed to endpoints in different locations.
	Endpoints []*Endpoint `json:"endpoints,omitempty"`
	// Labels with user-defined metadata to organize your models.
	Labels map[string]string `json:"labels,omitempty"`
	// Information about the tuned model from the base model.
	TunedModelInfo *TunedModelInfo `json:"tunedModelInfo,omitempty"`
	// The maximum number of input tokens that the model can handle.
	InputTokenLimit int64 `json:"inputTokenLimit,omitempty"`
	// The maximum number of output tokens that the model can generate.
	OutputTokenLimit int64 `json:"outputTokenLimit,omitempty"`
	// List of actions that are supported by the model.
	SupportedActions []string `json:"supportedActions,omitempty"`
}

type UpdateModelConfig struct {
	DisplayName string `json:"displayName,omitempty"`

	Description string `json:"description,omitempty"`
}

type UpdateModelParameters struct {
	Model string `json:"model,omitempty"`

	Config *UpdateModelConfig `json:"config,omitempty"`
}

type DeleteModelConfig struct {
}

type DeleteModelParameters struct {
	Model string `json:"model,omitempty"`
	// Optional parameters for the request.
	Config *DeleteModelConfig `json:"config,omitempty"`
}

type DeleteModelResponse struct {
}

// Generation config. You can find API default values and more details at https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/inference#generationconfig
// and https://cloud.google.com/vertex-ai/generative-ai/docs/multimodal/content-generation-parameters.
type GenerationConfig struct {
	// Optional. If enabled, audio timestamp will be included in the request to the model.
	AudioTimestamp bool `json:"audioTimestamp,omitempty"`
	// Optional. Number of candidates to generate.
	CandidateCount *int64 `json:"candidateCount,omitempty"`
	// Optional. Frequency penalties.
	FrequencyPenalty *float64 `json:"frequencyPenalty,omitempty"`
	// Optional. Logit probabilities.
	Logprobs *int64 `json:"logprobs,omitempty"`
	// Optional. The maximum number of output tokens to generate per message.
	MaxOutputTokens *int64 `json:"maxOutputTokens,omitempty"`
	// Optional. Positive penalties.
	PresencePenalty *float64 `json:"presencePenalty,omitempty"`
	// Optional. If true, export the logprobs results in response.
	ResponseLogprobs bool `json:"responseLogprobs,omitempty"`
	// Optional. Output response mimetype of the generated candidate text. Supported mimetype:
	// - `text/plain`: (default) Text output. - `application/json`: JSON response in the
	// candidates. The model needs to be prompted to output the appropriate response type,
	// otherwise the behavior is undefined. This is a preview feature.
	ResponseMIMEType string `json:"responseMimeType,omitempty"`
	// Optional. The `Schema` object allows the definition of input and output data types.
	// These types can be objects, but also primitives and arrays. Represents a select subset
	// of an [OpenAPI 3.0 schema object](https://spec.openapis.org/oas/v3.0.3#schema). If
	// set, a compatible response_mime_type must also be set. Compatible mimetypes: `application/json`:
	// Schema for JSON response.
	ResponseSchema *Schema `json:"responseSchema,omitempty"`
	// Optional. Routing configuration.
	RoutingConfig *GenerationConfigRoutingConfig `json:"routingConfig,omitempty"`
	// Optional. Seed.
	Seed *int64 `json:"seed,omitempty"`
	// Optional. Stop sequences.
	StopSequences []string `json:"stopSequences,omitempty"`
	// Optional. Controls the randomness of predictions.
	Temperature *float64 `json:"temperature,omitempty"`
	// Optional. If specified, top-k sampling will be used.
	TopK *float64 `json:"topK,omitempty"`
	// Optional. If specified, nucleus sampling will be used.
	TopP *float64 `json:"topP,omitempty"`
}

// Config for the count_tokens method.
type CountTokensConfig struct {
	// Instructions for the model to steer it toward better performance.
	SystemInstruction *Content `json:"systemInstruction,omitempty"`
	// Code that enables the system to interact with external systems to
	// perform an action outside of the knowledge and scope of the model.
	Tools []*Tool `json:"tools,omitempty"`
	// Configuration that the model uses to generate the response. Not
	// supported by the Gemini Developer API.
	GenerationConfig *GenerationConfig `json:"generationConfig,omitempty"`
}

// Parameters for counting tokens.
type CountTokensParameters struct {
	// ID of the model to use. For a list of models, see `Google models
	// <https://cloud.google.com/vertex-ai/generative-ai/docs/learn/models>`_.
	Model string `json:"model,omitempty"`
	// Input content.
	Contents []*Content `json:"contents,omitempty"`
	// Configuration for counting tokens.
	Config *CountTokensConfig `json:"config,omitempty"`
}

// Response for counting tokens.
type CountTokensResponse struct {
	// Total number of tokens.
	TotalTokens int64 `json:"totalTokens,omitempty"`
	// Number of tokens in the cached part of the prompt (the cached content). If nil, then
	// no CachedContentTokenCount is returned by the API.
	CachedContentTokenCount *int64 `json:"cachedContentTokenCount,omitempty"`
}

// Optional parameters for computing tokens.
type ComputeTokensConfig struct {
}

// Parameters for computing tokens.
type ComputeTokensParameters struct {
	// ID of the model to use. For a list of models, see `Google models
	// <https://cloud.google.com/vertex-ai/generative-ai/docs/learn/models>`_.
	Model string `json:"model,omitempty"`
	// Input content.
	Contents []*Content `json:"contents,omitempty"`
	// Optional parameters for the request.
	Config *ComputeTokensConfig `json:"config,omitempty"`
}

// Tokens info with a list of tokens and the corresponding list of token ids.
type TokensInfo struct {
	// Optional. Optional fields for the role from the corresponding Content.
	Role string `json:"role,omitempty"`
	// A list of token IDs from the input.
	TokenIDs []int64 `json:"tokenIds,omitempty"`
	// A list of tokens from the input.
	Tokens [][]byte `json:"tokens,omitempty"`
}

func (ti *TokensInfo) UnmarshalJSON(data []byte) error {
	aux := struct {
		Role     string   `json:"role,omitempty"`
		TokenIDs []string `json:"tokenIds,omitempty"`
		Tokens   [][]byte `json:"tokens,omitempty"`
	}{}
	if err := json.Unmarshal(data, &aux); err != nil {
		return err
	}

	// Convert the string value to int64
	tokenIDs := []int64{}
	for _, tokenID := range aux.TokenIDs {
		tokenIDInt, err := strconv.ParseInt(tokenID, 10, 64)
		if err != nil {
			return err
		}
		tokenIDs = append(tokenIDs, tokenIDInt)
	}
	ti.TokenIDs = tokenIDs
	ti.Role = aux.Role
	ti.Tokens = aux.Tokens
	return nil
}

// Response for computing tokens.
type ComputeTokensResponse struct {
	// Lists of tokens info from the input. A ComputeTokensRequest could have multiple instances
	// with a prompt in each instance. We also need to return lists of tokens info for the
	// request with multiple instances.
	TokensInfo []*TokensInfo `json:"tokensInfo,omitempty"`
}

// Optional configuration for cached content creation.
type CreateCachedContentConfig struct {
	// The TTL for this resource. The expiration time is computed: now + TTL.
	TTL string `json:"ttl,omitempty"`
	// Timestamp of when this resource is considered expired.
	ExpireTime *time.Time `json:"expireTime,omitempty"`
	// The user-generated meaningful display name of the cached content.
	DisplayName string `json:"displayName,omitempty"`
	// The content to cache.
	Contents []*Content `json:"contents,omitempty"`
	// Developer set system instruction.
	SystemInstruction *Content `json:"systemInstruction,omitempty"`
	// A list of `Tools` the model may use to generate the next response.
	Tools []*Tool `json:"tools,omitempty"`
	// Configuration for the tools to use. This config is shared for all tools.
	ToolConfig *ToolConfig `json:"toolConfig,omitempty"`
}

// Parameters for caches.create method.
type CreateCachedContentParameters struct {
	// ID of the model to use. Example: gemini-1.5-flash
	Model string `json:"model,omitempty"`
	// Configuration that contains optional parameters.
	Config *CreateCachedContentConfig `json:"config,omitempty"`
}

// Metadata on the usage of the cached content.
type CachedContentUsageMetadata struct {
	// Duration of audio in seconds. If nil, then no AudioDurationSeconds is returned by
	// the API.
	AudioDurationSeconds *int64 `json:"audioDurationSeconds,omitempty"`
	// Number of images. If nil, then no ImageCount is returned by the API.
	ImageCount *int64 `json:"imageCount,omitempty"`
	// Number of text characters. If nil, then no TextCount is returned by the API.
	TextCount *int64 `json:"textCount,omitempty"`
	// Total number of tokens that the cached content consumes.
	TotalTokenCount int64 `json:"totalTokenCount,omitempty"`
	// Duration of video in seconds. If nil, then no VideoDurationSeconds is returned by
	// the API.
	VideoDurationSeconds *int64 `json:"videoDurationSeconds,omitempty"`
}

// A resource used in LLM queries for users to explicitly specify what to cache.
type CachedContent struct {
	// The server-generated resource name of the cached content.
	Name string `json:"name,omitempty"`
	// The user-generated meaningful display name of the cached content.
	DisplayName string `json:"displayName,omitempty"`
	// The name of the publisher model to use for cached content.
	Model string `json:"model,omitempty"`
	// Creation time of the cache entry.
	CreateTime *time.Time `json:"createTime,omitempty"`
	// When the cache entry was last updated in UTC time.
	UpdateTime *time.Time `json:"updateTime,omitempty"`
	// Expiration time of the cached content.
	ExpireTime *time.Time `json:"expireTime,omitempty"`
	// Metadata on the usage of the cached content.
	UsageMetadata *CachedContentUsageMetadata `json:"usageMetadata,omitempty"`
}

// Optional parameters for caches.get method.
type GetCachedContentConfig struct {
}

// Parameters for caches.get method.
type GetCachedContentParameters struct {
	// The server-generated resource name of the cached content.
	Name string `json:"name,omitempty"`
	// Optional parameters for the request.
	Config *GetCachedContentConfig `json:"config,omitempty"`
}

// Optional parameters for caches.delete method.
type DeleteCachedContentConfig struct {
}

// Parameters for caches.delete method.
type DeleteCachedContentParameters struct {
	// The server-generated resource name of the cached content.
	Name string `json:"name,omitempty"`
	// Optional parameters for the request.
	Config *DeleteCachedContentConfig `json:"config,omitempty"`
}

// Empty response for caches.delete method.
type DeleteCachedContentResponse struct {
}

// Optional parameters for caches.update method.
type UpdateCachedContentConfig struct {
	// The TTL for this resource. The expiration time is computed: now + TTL.
	TTL string `json:"ttl,omitempty"`
	// Timestamp of when this resource is considered expired.
	ExpireTime *time.Time `json:"expireTime,omitempty"`
}

type UpdateCachedContentParameters struct {
	// The server-generated resource name of the cached content.
	Name string `json:"name,omitempty"`
	// Configuration that contains optional parameters.
	Config *UpdateCachedContentConfig `json:"config,omitempty"`
}

// Config for caches.list method.
type ListCachedContentsConfig struct {
	// PageSize specifies the maximum number of cached contents to return per API call.
	// This setting does not affect the total number of cached contents returned by the
	// All() function during iteration; it only controls how many items are retrieved in
	// each individual request to the server. If zero, the server will use a default value.
	// Setting a positive value can be useful for managing the size and frequency of API
	// calls.
	PageSize int64 `json:"pageSize,omitempty"`
	// PageToken represents a token used for pagination in API responses. It's an opaque
	// string that should be passed to subsequent requests to retrieve the next page of
	// results. An empty PageToken typically indicates that there are no further pages available.
	PageToken string `json:"pageToken,omitempty"`
}

// Parameters for caches.list method.
type ListCachedContentsParameters struct {
	// Configuration that contains optional parameters.
	Config *ListCachedContentsConfig `json:"config,omitempty"`
}

type ListCachedContentsResponse struct {
	NextPageToken string `json:"nextPageToken,omitempty"`
	// List of cached contents.
	CachedContents []*CachedContent `json:"cachedContents,omitempty"`
}

type testTableItem struct {
	// The name of the test. This is used to derive the replay id.
	Name string `json:"name,omitempty"`
	// The parameters to the test. Use pydantic models.
	Parameters map[string]any `json:"parameters,omitempty"`
	// Expects an exception for MLDev matching the string.
	ExceptionIfMLDev string `json:"exceptionIfMldev,omitempty"`
	// Expects an exception for Vertex matching the string.
	ExceptionIfVertex string `json:"exceptionIfVertex,omitempty"`
	// Use if you don't want to use the default replay ID which is derived from the test
	// name.
	OverrideReplayID string `json:"overrideReplayId,omitempty"`
	// True if the parameters contain an unsupported union type. This test will be skipped
	// for languages that do not support the union type.
	HasUnion bool `json:"hasUnion,omitempty"`
	// When set to a reason string, this test will be skipped in the API mode. Use this
	// flag for tests that can not be reproduced with the real API. E.g. a test that deletes
	// a resource.
	SkipInAPIMode string `json:"skipInApiMode,omitempty"`
}

type testTableFile struct {
	Comment string `json:"comment,omitempty"`

	TestMethod string `json:"testMethod,omitempty"`

	ParameterNames []string `json:"parameterNames,omitempty"`

	TestTable []*testTableItem `json:"testTable,omitempty"`
}

// Represents a single request in a replay.
type replayRequest struct {
	Method string `json:"method,omitempty"`

	URL string `json:"url,omitempty"`

	Headers map[string]string `json:"headers,omitempty"`

	BodySegments []map[string]any `json:"bodySegments,omitempty"`
}

// Represents a single response in a replay.
type replayResponse struct {
	StatusCode int64 `json:"statusCode,omitempty"`

	Headers map[string]string `json:"headers,omitempty"`

	BodySegments []map[string]any `json:"bodySegments,omitempty"`

	SDKResponseSegments []map[string]any `json:"sdkResponseSegments,omitempty"`
}

// Represents a single interaction, request and response in a replay.
type replayInteraction struct {
	Request *replayRequest `json:"request,omitempty"`

	Response *replayResponse `json:"response,omitempty"`
}

// Represents a recorded session.
type replayFile struct {
	ReplayID string `json:"replayId,omitempty"`

	Interactions []*replayInteraction `json:"interactions,omitempty"`
}

// Used to override the default configuration.
type UploadFileConfig struct {
	// The name of the file in the destination (e.g., 'files/sample-image'. If not provided
	// one will be generated.
	Name string `json:"name,omitempty"`
	// mime_type: The MIME type of the file. If not provided, it will be inferred from the
	// file extension.
	MIMEType string `json:"mimeType,omitempty"`
	// Optional display name of the file.
	DisplayName string `json:"displayName,omitempty"`
}

// Configuration for upscaling an image.
// For more information on this configuration, refer to
// the `Imagen API reference documentation
// <https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/imagen-api>`_.
type UpscaleImageConfig struct {
	// Whether to include a reason for filtered-out images in the
	// response.
	IncludeRAIReason bool `json:"includeRaiReason,omitempty"`
	// The image format that the output should be saved as.
	OutputMIMEType string `json:"outputMimeType,omitempty"`
	// The level of compression if the output_mime_type is image/jpeg. If nil, then API
	// will determine the default value.
	OutputCompressionQuality *int64 `json:"outputCompressionQuality,omitempty"`
}

// User-facing config UpscaleImageParameters.
type UpscaleImageParameters struct {
	// The model to use.
	Model string `json:"model,omitempty"`
	// The input image to upscale.
	Image *Image `json:"image,omitempty"`
	// The factor to upscale the image (x2 or x4).
	UpscaleFactor string `json:"upscaleFactor,omitempty"`
	// Configuration for upscaling.
	Config *UpscaleImageConfig `json:"config,omitempty"`
}

// A raw reference image.
// A raw reference image represents the base image to edit, provided by the user.
// It can optionally be provided in addition to a mask reference image or
// a style reference image.
type RawReferenceImage struct {
	// The reference image for the editing operation.
	ReferenceImage *Image `json:"referenceImage,omitempty"`
	// The ID of the reference image.
	ReferenceID int64 `json:"referenceId,omitempty"`
	// The type of the reference image.
	ReferenceType string `json:"referenceType,omitempty"`
}

// Configuration for a Mask reference image.
type MaskReferenceConfig struct {
	// Prompts the model to generate a mask instead of you needing to
	// provide one (unless MASK_MODE_USER_PROVIDED is used).
	MaskMode MaskReferenceMode `json:"maskMode,omitempty"`
	// A list of up to 5 class IDs to use for semantic segmentation.
	// Automatically creates an image mask based on specific objects.
	SegmentationClasses []int64 `json:"segmentationClasses,omitempty"`
	// Dilation percentage of the mask provided. Float between 0 and 1. If nil, then API
	// will determine the default value.
	MaskDilation *float64 `json:"maskDilation,omitempty"`
}

// A mask image is an image whose non-zero values indicate where to edit the base image.
// If the user provides a mask image, the mask must be in the same dimensions as the
// raw image. MaskReferenceImage encapsulates either a mask image provided by the user
// and configs for the user provided mask, or only config parameters for the model to
// generate a mask.
type MaskReferenceImage struct {
	// The reference image for the editing operation.
	ReferenceImage *Image `json:"referenceImage,omitempty"`
	// The ID of the reference image.
	ReferenceID int64 `json:"referenceId,omitempty"`
	// The type of the reference image.
	ReferenceType string `json:"referenceType,omitempty"`
	// Configuration for the mask reference image.
	Config *MaskReferenceConfig `json:"config,omitempty"`
}

// Configuration for a Control reference image.
type ControlReferenceConfig struct {
	// The type of control reference image to use.
	ControlType ControlReferenceType `json:"controlType,omitempty"`
	// Defaults to False. When set to True, the control image will be
	// computed by the model based on the control type. When set to False,
	// the control image must be provided by the user.
	EnableControlImageComputation bool `json:"enableControlImageComputation,omitempty"`
}

// A control image is an image that represents a sketch image of areas for the model
// to fill in based on the prompt.
// The image of the control reference image is either a control image provided by the
// user, or a regular image which the backend will use to generate a control image of.
// In the case of the latter, the enable_control_image_computation field in the config
// should be set to true.
type ControlReferenceImage struct {
	// The reference image for the editing operation.
	ReferenceImage *Image `json:"referenceImage,omitempty"`
	// The ID of the reference image.
	ReferenceID int64 `json:"referenceId,omitempty"`
	// The type of the reference image.
	ReferenceType string `json:"referenceType,omitempty"`
	// Configuration for the control reference image.
	Config *ControlReferenceConfig `json:"config,omitempty"`
}

// Configuration for a Style reference image.
type StyleReferenceConfig struct {
	// A text description of the style to use for the generated image.
	StyleDescription string `json:"styleDescription,omitempty"`
}

// A style reference image.
// This encapsulates a style reference image provided by the user, and
// additionally optional config parameters for the style reference image.
// A raw reference image can also be provided as a destination for the style to
// be applied to.
type StyleReferenceImage struct {
	// The reference image for the editing operation.
	ReferenceImage *Image `json:"referenceImage,omitempty"`
	// The ID of the reference image.
	ReferenceID int64 `json:"referenceId,omitempty"`
	// The type of the reference image.
	ReferenceType string `json:"referenceType,omitempty"`
	// Configuration for the style reference image.
	Config *StyleReferenceConfig `json:"config,omitempty"`
}

// Configuration for a Subject reference image.
type SubjectReferenceConfig struct {
	// The subject type of a subject reference image.
	SubjectType SubjectReferenceType `json:"subjectType,omitempty"`
	// Subject description for the image.
	SubjectDescription string `json:"subjectDescription,omitempty"`
}

// A subject reference image.
// This encapsulates a subject reference image provided by the user, and
// additionally optional config parameters for the subject reference image.
// A raw reference image can also be provided as a destination for the subject to
// be applied to.
type SubjectReferenceImage struct {
	// The reference image for the editing operation.
	ReferenceImage *Image `json:"referenceImage,omitempty"`
	// The ID of the reference image.
	ReferenceID int64 `json:"referenceId,omitempty"`
	// The type of the reference image.
	ReferenceType string `json:"referenceType,omitempty"`
	// Configuration for the subject reference image.
	Config *SubjectReferenceConfig `json:"config,omitempty"`
}

// Sent in response to a `LiveGenerateContentSetup` message from the client.
type LiveServerSetupComplete struct {
}

// Incremental server update generated by the model in response to client messages.
// Content is generated as quickly as possible, and not in real time. Clients
// may choose to buffer and play it out in real time.
type LiveServerContent struct {
	// The content that the model has generated as part of the current conversation with
	// the user.
	ModelTurn *Content `json:"modelTurn,omitempty"`
	// If true, indicates that the model is done generating. Generation will only start
	// in response to additional client messages. Can be set alongside `content`, indicating
	// that the `content` is the last in the turn.
	TurnComplete bool `json:"turnComplete,omitempty"`
	// If true, indicates that a client message has interrupted current model generation.
	// If the client is playing out the content in realtime, this is a good signal to stop
	// and empty the current queue.
	Interrupted bool `json:"interrupted,omitempty"`
}

// Request for the client to execute the `function_calls` and return the responses with
// the matching `id`s.
type LiveServerToolCall struct {
	// The function call to be executed.
	FunctionCalls []*FunctionCall `json:"functionCalls,omitempty"`
}

// Notification for the client that a previously issued `ToolCallMessage` with the specified
// `id`s should have been not executed and should be cancelled.
// If there were side-effects to those tool calls, clients may attempt to undo
// the tool calls. This message occurs only in cases where the clients interrupt
// server turns.
type LiveServerToolCallCancellation struct {
	// The IDs of the tool calls to be cancelled.
	IDs []string `json:"ids,omitempty"`
}

// Response message for API call.
type LiveServerMessage struct {
	// Sent in response to a `LiveClientSetup` message from the client.
	SetupComplete *LiveServerSetupComplete `json:"setupComplete,omitempty"`
	// Content generated by the model in response to client messages.
	ServerContent *LiveServerContent `json:"serverContent,omitempty"`
	// Request for the client to execute the `function_calls` and return the responses with
	// the matching `id`s.
	ToolCall *LiveServerToolCall `json:"toolCall,omitempty"`
	// Notification for the client that a previously issued `ToolCallMessage` with the specified
	// `id`s should have been not executed and should be cancelled.
	ToolCallCancellation *LiveServerToolCallCancellation `json:"toolCallCancellation,omitempty"`
}

// Message contains configuration that will apply for the duration of the streaming
// session.
type LiveClientSetup struct {
	// The fully qualified name of the publisher model or tuned model endpoint to
	// use.
	Model string `json:"model,omitempty"`
	// The generation configuration for the session.
	// The following fields are supported:
	//   - `response_logprobs`
	//   - `response_mime_type`
	//   - `logprobs`
	//   - `response_schema`
	//   - `stop_sequence`
	//   - `routing_config`
	//   - `audio_timestamp`
	GenerationConfig *GenerationConfig `json:"generationConfig,omitempty"`
	// The user provided system instructions for the model.
	// Note: only text should be used in parts and content in each part will be
	// in a separate paragraph.
	SystemInstruction *Content `json:"systemInstruction,omitempty"`
	// A list of `Tools` the model may use to generate the next response.
	// A `Tool` is a piece of code that enables the system to interact with
	// external systems to perform an action, or set of actions, outside of
	// knowledge and scope of the model.
	Tools []*Tool `json:"tools,omitempty"`
}

// Incremental update of the current conversation delivered from the client.
// All the content here will unconditionally be appended to the conversation
// history and used as part of the prompt to the model to generate content.
// A message here will interrupt any current model generation.
type LiveClientContent struct {
	// The content appended to the current conversation with the model.
	// For single-turn queries, this is a single instance. For multi-turn
	// queries, this is a repeated field that contains conversation history and
	// latest request.
	Turns []*Content `json:"turns,omitempty"`
	// If true, indicates that the server content generation should start with
	// the currently accumulated prompt. Otherwise, the server will await
	// additional messages before starting generation.
	TurnComplete bool `json:"turnComplete,omitempty"`
}

// User input that is sent in real time.
// This is different from `ClientContentUpdate` in a few ways:
//   - Can be sent continuously without interruption to model generation.
//   - If there is a need to mix data interleaved across the
//     `ClientContentUpdate` and the `RealtimeUpdate`, server attempts to
//     optimize for best response, but there are no guarantees.
//   - End of turn is not explicitly specified, but is rather derived from user
//     activity (for example, end of speech).
//   - Even before the end of turn, the data is processed incrementally
//     to optimize for a fast start of the response from the model.
//   - Is always assumed to be the user's input (cannot be used to populate
//     conversation history).
type LiveClientRealtimeInput struct {
	// Inlined bytes data for media input.
	MediaChunks []*Blob `json:"mediaChunks,omitempty"`
}

// Client generated response to a `ToolCall` received from the server.
// Individual `FunctionResponse` objects are matched to the respective
// `FunctionCall` objects by the `id` field.
// Note that in the unary and server-streaming GenerateContent APIs function
// calling happens by exchanging the `Content` parts, while in the bidi
// GenerateContent APIs function calling happens over this dedicated set of
// messages.
type LiveClientToolResponse struct {
	// The response to the function calls.
	FunctionResponses []*FunctionResponse `json:"functionResponses,omitempty"`
}

// Messages sent by the client in the API call.
type LiveClientMessage struct {
	// Message to be sent by the system when connecting to the API. SDK users should not
	// send this message.
	Setup *LiveClientSetup `json:"setup,omitempty"`
	// Incremental update of the current conversation delivered from the client.
	ClientContent *LiveClientContent `json:"clientContent,omitempty"`
	// User input that is sent in real time.
	RealtimeInput *LiveClientRealtimeInput `json:"realtimeInput,omitempty"`
	// Response to a `ToolCallMessage` received from the server.
	ToolResponse *LiveClientToolResponse `json:"toolResponse,omitempty"`
}

// Session config for the API connection.
type LiveConnectConfig struct {
	// The generation configuration for the session.
	GenerationConfig *GenerationConfig `json:"generationConfig,omitempty"`
	// The requested modalities of the response. Represents the set of
	// modalities that the model can return. Defaults to AUDIO if not specified.
	ResponseModalities []Modality `json:"responseModalities,omitempty"`
	// The speech generation configuration.
	SpeechConfig *SpeechConfig `json:"speechConfig,omitempty"`
	// The user provided system instructions for the model.
	// Note: only text should be used in parts and content in each part will be
	// in a separate paragraph.
	SystemInstruction *Content `json:"systemInstruction,omitempty"`
	// A list of `Tools` the model may use to generate the next response.
	// A `Tool` is a piece of code that enables the system to interact with
	// external systems to perform an action, or set of actions, outside of
	// knowledge and scope of the model.
	Tools []*Tool `json:"tools,omitempty"`
}