gpt3

type APIError struct {
	RateLimitHeaders RateLimitHeaders

	StatusCode int    `json:"status_code"`
	Message    string `json:"message"`
	Type       string `json:"type"`
}

type APIErrorResponse struct {
	Error APIError `json:"error"`
}

type ChatCompletionFunctionParameters struct {
	Type        string                                       `json:"type"`
	Description string                                       `json:"description,omitempty"`
	Properties  map[string]FunctionParameterPropertyMetadata `json:"properties"`
	Required    []string                                     `json:"required"`
}

type ChatCompletionFunctions struct {
	Name        string                           `json:"name"`
	Description string                           `json:"description,omitempty"`
	Parameters  ChatCompletionFunctionParameters `json:"parameters"`
}

type ChatCompletionRequest struct {
	// Model is the name of the model to use. If not specified, will default to gpt-3.5-turbo.
	Model string `json:"model"`

	// Messages is a list of messages to use as the context for the chat completion.
	Messages []ChatCompletionRequestMessage `json:"messages"`

	// Functions is a list of functions the model may generate JSON inputs for.
	Functions []ChatCompletionFunctions `json:"functions,omitempty"`

	// What sampling temperature to use, between 0 and 2. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic
	Temperature *float32 `json:"temperature,omitempty"`

	// An alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the tokens with top_p probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered.
	TopP float32 `json:"top_p,omitempty"`

	// Number of responses to generate
	N int `json:"n,omitempty"`

	// Whether or not to stream responses back as they are generated
	Stream bool `json:"stream,omitempty"`

	// Up to 4 sequences where the API will stop generating further tokens.
	Stop []string `json:"stop,omitempty"`

	// MaxTokens is the maximum number of tokens to return.
	MaxTokens int `json:"max_tokens,omitempty"`

	// (-2, 2) Penalize tokens that haven't appeared yet in the history.
	PresencePenalty float32 `json:"presence_penalty,omitempty"`

	// (-2, 2) Penalize tokens that appear too frequently in the history.
	FrequencyPenalty float32 `json:"frequency_penalty,omitempty"`

	// Modify the probability of specific tokens appearing in the completion.
	LogitBias map[string]float32 `json:"logit_bias,omitempty"`

	// Can be used to identify an end-user
	User string `json:"user,omitempty"`
}

type ChatCompletionRequestMessage struct {
	// Role is the role is the role of the the message. Can be "system", "user", or "assistant"
	Role string `json:"role"`

	// Content is the content of the message
	Content string `json:"content"`

	// FunctionCall is the name and arguments of a function that should be called, as generated by the model.
	FunctionCall *Function `json:"function_call,omitempty"`

	// Name is the the name of the author of this message. `name` is required if role is `function`, and it should be the name of the function whose response is in the `content`.
	Name string `json:"name,omitempty"`
}

type ChatCompletionResponse struct {
	RateLimitHeaders RateLimitHeaders

	ID      string                         `json:"id"`
	Object  string                         `json:"object"`
	Created int                            `json:"created"`
	Model   string                         `json:"model"`
	Choices []ChatCompletionResponseChoice `json:"choices"`
	Usage   ChatCompletionsResponseUsage   `json:"usage"`
}

type ChatCompletionResponseChoice struct {
	Index        int                           `json:"index"`
	FinishReason string                        `json:"finish_reason"`
	Message      ChatCompletionResponseMessage `json:"message"`
}

type ChatCompletionResponseMessage struct {
	Role         string    `json:"role"`
	Content      string    `json:"content"`
	FunctionCall *Function `json:"function_call,omitempty"`
}

type ChatCompletionStreamResponse struct {
	ID      string                               `json:"id"`
	Object  string                               `json:"object"`
	Created int                                  `json:"created"`
	Model   string                               `json:"model"`
	Choices []ChatCompletionStreamResponseChoice `json:"choices"`
	Usage   ChatCompletionsResponseUsage         `json:"usage"`
}

type ChatCompletionStreamResponseChoice struct {
	Index        int                           `json:"index"`
	FinishReason string                        `json:"finish_reason"`
	Delta        ChatCompletionResponseMessage `json:"delta"`
}

type ChatCompletionsResponseUsage struct {
	PromptTokens     int `json:"prompt_tokens"`
	CompletionTokens int `json:"completion_tokens"`
	TotalTokens      int `json:"total_tokens"`
}

type Client interface {
	// Engines lists the currently available engines, and provides basic information about each
	// option such as the owner and availability.
	Engines(ctx context.Context) (*EnginesResponse, error)

	// Engine retrieves an engine instance, providing basic information about the engine such
	// as the owner and availability.
	Engine(ctx context.Context, engine string) (*EngineObject, error)

	// ChatCompletion creates a completion with the Chat completion endpoint which
	// is what powers the ChatGPT experience.
	ChatCompletion(ctx context.Context, request ChatCompletionRequest) (*ChatCompletionResponse, error)

	// ChatCompletion creates a completion with the Chat completion endpoint which
	// is what powers the ChatGPT experience.
	ChatCompletionStream(ctx context.Context, request ChatCompletionRequest, onData func(*ChatCompletionStreamResponse) error) error

	// Completion creates a completion with the default engine. This is the main endpoint of the API
	// which auto-completes based on the given prompt.
	Completion(ctx context.Context, request CompletionRequest) (*CompletionResponse, error)

	// CompletionStream creates a completion with the default engine and streams the results through
	// multiple calls to onData.
	CompletionStream(ctx context.Context, request CompletionRequest, onData func(*CompletionResponse)) error

	// CompletionWithEngine is the same as Completion except allows overriding the default engine on the client
	CompletionWithEngine(ctx context.Context, engine string, request CompletionRequest) (*CompletionResponse, error)

	// CompletionStreamWithEngine is the same as CompletionStream except allows overriding the default engine on the client
	CompletionStreamWithEngine(ctx context.Context, engine string, request CompletionRequest, onData func(*CompletionResponse)) error

	// Given a prompt and an instruction, the model will return an edited version of the prompt.
	Edits(ctx context.Context, request EditsRequest) (*EditsResponse, error)

	// Search performs a semantic search over a list of documents with the default engine.
	Search(ctx context.Context, request SearchRequest) (*SearchResponse, error)

	// SearchWithEngine performs a semantic search over a list of documents with the specified engine.
	SearchWithEngine(ctx context.Context, engine string, request SearchRequest) (*SearchResponse, error)

	// Returns an embedding using the provided request.
	Embeddings(ctx context.Context, request EmbeddingsRequest) (*EmbeddingsResponse, error)

	// Moderation performs a moderation check on the given text against an OpenAI classifier to determine whether the
	// provided content complies with OpenAI's usage policies.
	Moderation(ctx context.Context, request ModerationRequest) (*ModerationResponse, error)
}

type ClientOption func(*client) error

type CompletionRequest struct {
	// A list of string prompts to use.
	// TODO there are other prompt types here for using token integers that we could add support for.
	Prompt []string `json:"prompt"`
	// How many tokens to complete up to. Max of 512
	MaxTokens *int `json:"max_tokens,omitempty"`
	// Sampling temperature to use
	Temperature *float32 `json:"temperature,omitempty"`
	// Alternative to temperature for nucleus sampling
	TopP *float32 `json:"top_p,omitempty"`
	// How many choice to create for each prompt
	N *int `json:"n"`
	// Include the probabilities of most likely tokens
	LogProbs *int `json:"logprobs"`
	// Echo back the prompt in addition to the completion
	Echo bool `json:"echo"`
	// Up to 4 sequences where the API will stop generating tokens. Response will not contain the stop sequence.
	Stop []string `json:"stop,omitempty"`
	// PresencePenalty number between 0 and 1 that penalizes tokens that have already appeared in the text so far.
	PresencePenalty float32 `json:"presence_penalty"`
	// FrequencyPenalty number between 0 and 1 that penalizes tokens on existing frequency in the text so far.
	FrequencyPenalty float32 `json:"frequency_penalty"`

	// Whether to stream back results or not. Don't set this value in the request yourself
	// as it will be overriden depending on if you use CompletionStream or Completion methods.
	Stream bool `json:"stream,omitempty"`
}

type CompletionResponse struct {
	RateLimitHeaders RateLimitHeaders

	ID      string                     `json:"id"`
	Object  string                     `json:"object"`
	Created int                        `json:"created"`
	Model   string                     `json:"model"`
	Choices []CompletionResponseChoice `json:"choices"`
	Usage   CompletionResponseUsage    `json:"usage"`
}

type CompletionResponseChoice struct {
	Text         string        `json:"text"`
	Index        int           `json:"index"`
	LogProbs     LogprobResult `json:"logprobs"`
	FinishReason string        `json:"finish_reason"`
}

type CompletionResponseUsage struct {
	PromptTokens     int `json:"prompt_tokens"`
	CompletionTokens int `json:"completion_tokens"`
	TotalTokens      int `json:"total_tokens"`
}

type EditsRequest struct {
	// ID of the model to use. You can use the List models API to see all of your available models, or see our Model overview for descriptions of them.
	Model string `json:"model"`
	// The input text to use as a starting point for the edit.
	Input string `json:"input"`
	// The instruction that tells the model how to edit the prompt.
	Instruction string `json:"instruction"`
	// Sampling temperature to use
	Temperature *float32 `json:"temperature,omitempty"`
	// Alternative to temperature for nucleus sampling
	TopP *float32 `json:"top_p,omitempty"`
	// How many edits to generate for the input and instruction. Defaults to 1
	N *int `json:"n"`
}

type EditsResponse struct {
	Object  string                `json:"object"`
	Created int                   `json:"created"`
	Choices []EditsResponseChoice `json:"choices"`
	Usage   EditsResponseUsage    `json:"usage"`
}

type EditsResponseChoice struct {
	Text  string `json:"text"`
	Index int    `json:"index"`
}

type EditsResponseUsage struct {
	PromptTokens     int `json:"prompt_tokens"`
	CompletionTokens int `json:"completion_tokens"`
	TotalTokens      int `json:"total_tokens"`
}

type EmbeddingEngine string

type EmbeddingsRequest struct {
	// Input text to get embeddings for, encoded as a string or array of tokens. To get embeddings
	// for multiple inputs in a single request, pass an array of strings or array of token arrays.
	// Each input must not exceed 2048 tokens in length.
	Input []string `json:"input"`
	// ID of the model to use
	Model string `json:"model"`
	// The request user is an optional parameter meant to be used to trace abusive requests
	// back to the originating user. OpenAI states:
	// "The [user] IDs should be a string that uniquely identifies each user. We recommend hashing
	// their username or email address, in order to avoid sending us any identifying information.
	// If you offer a preview of your product to non-logged in users, you can send a session ID
	// instead."
	User string `json:"user,omitempty"`
}

type EmbeddingsResponse struct {
	Object string             `json:"object"`
	Data   []EmbeddingsResult `json:"data"`
	Usage  EmbeddingsUsage    `json:"usage"`
}

type EmbeddingsResult struct {
	// The type of object returned (e.g., "list", "object")
	Object string `json:"object"`
	// The embedding data for the input
	Embedding []float64 `json:"embedding"`
	Index     int       `json:"index"`
}

type EmbeddingsUsage struct {
	// The number of tokens used by the prompt
	PromptTokens int `json:"prompt_tokens"`
	// The total tokens used
	TotalTokens int `json:"total_tokens"`
}

type EngineObject struct {
	ID     string `json:"id"`
	Object string `json:"object"`
	Owner  string `json:"owner"`
	Ready  bool   `json:"ready"`
}

type EnginesResponse struct {
	Data   []EngineObject `json:"data"`
	Object string         `json:"object"`
}

type Function struct {
	Name      string `json:"name"`
	Arguments string `json:"arguments"`
}

type FunctionParameterPropertyMetadata struct {
	Type        string   `json:"type"`
	Description string   `json:"description,omitempty"`
	Enum        []string `json:"enum,omitempty"`
}

type LogprobResult struct {
	Tokens        []string             `json:"tokens"`
	TokenLogprobs []float32            `json:"token_logprobs"`
	TopLogprobs   []map[string]float32 `json:"top_logprobs"`
	TextOffset    []int                `json:"text_offset"`
}

type ModerationCategoryResult struct {
	Hate            bool `json:"hate"`
	HateThreatening bool `json:"hate/threatening"`
	SelfHarm        bool `json:"self-harm"`
	Sexual          bool `json:"sexual"`
	SexualMinors    bool `json:"sexual/minors"`
	Violence        bool `json:"violence"`
	ViolenceGraphic bool `json:"violence/graphic"`
}

type ModerationCategoryScores struct {
	Hate            float32 `json:"hate"`
	HateThreatening float32 `json:"hate/threatening"`
	SelfHarm        float32 `json:"self-harm"`
	Sexual          float32 `json:"sexual"`
	SexualMinors    float32 `json:"sexual/minors"`
	Violence        float32 `json:"violence"`
	ViolenceGraphic float32 `json:"violence/graphic"`
}

type ModerationRequest struct {
	// Input is the input text that should be classified. Required.
	Input string `json:"input"`
	// Model is the content moderation model to use. If not specified, will default to OpenAI API defaults, which is
	// currently "text-moderation-latest".
	Model string `json:"model,omitempty"`
}

type ModerationResponse struct {
	ID      string             `json:"id"`
	Model   string             `json:"model"`
	Results []ModerationResult `json:"results"`
}

type ModerationResult struct {
	Flagged        bool                     `json:"flagged"`
	Categories     ModerationCategoryResult `json:"categories"`
	CategoryScores ModerationCategoryScores `json:"category_scores"`
}

type RateLimitHeaders struct {
	// x-ratelimit-limit-requests: The maximum number of requests that are permitted before exhausting the rate limit.
	LimitRequests int

	// x-ratelimit-limit-tokens: The maximum number of tokens that are permitted before exhausting the rate limit.
	LimitTokens int

	// x-ratelimit-remaining-requests: The remaining number of requests that are permitted before exhausting the rate limit.
	RemainingRequests int

	// x-ratelimit-remaining-tokens: The remaining number of tokens that are permitted before exhausting the rate limit.
	RemainingTokens int

	// x-ratelimit-reset-requests: The time until the rate limit (based on requests) resets to its initial state.
	ResetRequests time.Duration

	// x-ratelimit-reset-tokens: The time until the rate limit (based on tokens) resets to its initial state.
	ResetTokens time.Duration
}

type SearchData struct {
	Document int     `json:"document"`
	Object   string  `json:"object"`
	Score    float64 `json:"score"`
}

type SearchRequest struct {
	Documents []string `json:"documents"`
	Query     string   `json:"query"`
}

type SearchResponse struct {
	Data   []SearchData `json:"data"`
	Object string       `json:"object"`
}