Class OpenaiApiClient

A class representing an OpenAI API client.

Hierarchy

AbstractApiClient
- OpenaiApiClient
  - OpenaiApiClientExtended

Constructors

constructor

new OpenaiApiClient(options?): OpenaiApiClient
Create a new OpenaiApiClient instance.
Parameters
- options: IOpenaiApiClientOptions = {}
  
  The constructor options to use.
Returns OpenaiApiClient
Overrides AbstractApiClient.constructor
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:40

Properties

`Readonly` apiDefaults

apiDefaults: IOpenaiApiClientApiDefaultsOptions = ...

Defaults for API requests. Can be overriden in individual method calls.

`Optional` `Readonly` cache

cache?: ApiReponseCache<any>

API response cache

`Readonly` cacheDefaults

cacheDefaults: IResponseCacheOptions = ...

Default options for caching for api requests. Can be overriden in individual method calls.

`Readonly` client

client: OpenAIApi

API client instance

`Protected` concurrencyJustReduced

concurrencyJustReduced: boolean = false

`Readonly` events

events: EventEmitter = ...

Event emitter for cache events

`Readonly` queue

queue: PromiseQueue<IQueue<(() => Promise<unknown>), IQueueAddOptions>, IQueueAddOptions>

Global queue for sending requests to the openai api.

`Readonly` retryDefaults

retryDefaults: IAsyncRetryOptions = ...

Default options for async retry for api requests. Can be overriden in individual method calls.

`Protected` `Readonly` sendRequest

sendRequest: (<T>(options) => Promise<T>)

Type declaration

- <T>(options): Promise<T>
- Generic function for sending requests to the openai api. This is used for all the API endpoints. It handles retrying, cache, hashing, and emitting events. This method is bound to the instance on initialization because it gets wrapped with a concurrency controller in the constructor.
  Type Parameters
  - T
  Parameters
  - options: {
        apiRequest: (() => Promise<T>);
        args: any[];
        cache?: IResponseCacheOptions;
        retry?: IAsyncRetryOptions;
    }
    - apiRequest: (() => Promise<T>)
      
      (): Promise<T>
      
      Returns Promise<T>
    - args: any[]
    - Optional cache?: IResponseCacheOptions
    - Optional retry?: IAsyncRetryOptions
  Returns Promise<T>

`Static` `Readonly` concurrencyDefaults

concurrencyDefaults: IPromiseQueueOptions<IQueue<(() => Promise<unknown>), IQueueAddOptions>, IQueueAddOptions> = ...

Options for concurrency control. These affect all API requests.

Accessors

eventNames

get eventNames(): string[]
All emitted event names. Please note that the cache also emits events.

Returns string[]
Inherited from AbstractApiClient.eventNames
- Defined in packages/api-util/src/lib/AbstractApiClient.ts:26

Methods

`Protected` _chat

_chat(request, retry, cache): Promise<string>
Send chat request to the openai API. This is used by all the preset methods, the public methods: chat3_8, chat3_16, and chat4_8.
Parameters
- request: CreateChatCompletionRequest
  
  The request object to send to the openai api.
- retry: IAsyncRetryOptions
  
  The retry options.
- cache: IResponseCacheOptions
  
  The cache options.
Returns Promise<string>
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:133

`Protected` _transcribe

_transcribe(request, retry, cache): Promise<string>
Send transcribe (speech to text) request to the openai API.
Parameters
- request: IOpenaiTranscribeRequest
  
  The request object to send to the openai api.
- retry: IAsyncRetryOptions
  
  The retry options.
- cache: IResponseCacheOptions
  
  The cache options.
Returns Promise<string>
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:163

`Protected` assertReponseDataComplete

assertReponseDataComplete(data): void
Assert that the response data is complete by verifying that all returned choices have finish_reason: stop.
Parameters
- data: CreateChatCompletionResponse | CreateCompletionResponse
Returns void
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:259

countTokens

countTokens(string): number
Count the number of tokens in a string.
Parameters
- string: string
  
  The string to count tokens in.
Returns number
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:281

`Protected` deleteDefaultOrUndefinedOptions

deleteDefaultOrUndefinedOptions<T>(options, defaults?): T
Delete all options that are undefined or equal to the default value. The response cache uses hashed options to determine if the request has already been made. Removing default values and undefined values normalizes the options object so it hashes the same.
Type Parameters
- T extends Record<string, any>
Parameters
- options: T
  
  The options to delete from.
- defaults: Record<string, any> = {}
  
  The default values to compare against.
Returns T
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:235

`Protected` emit

emit<T>(event, arg): T
Emit an event but adds 'this' as an extra trailing argument.
Type Parameters
- T
Parameters
- event: string
  
  The event name.
- arg: T
  
  The argument to emit.
Returns T
Inherited from AbstractApiClient.emit
- Defined in packages/api-util/src/lib/AbstractApiClient.ts:182

gpt3-16k

gpt3-16k(options): Promise<string>
Send a chat completion request to the openai api with a max_tokens cap of 16384.
Parameters
- options: IOpenaiChatRequestOptions
  
  The options to use.
Returns Promise<string>
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:67

gpt3-8k

gpt3-8k(options): Promise<string>
Send a chat completion request to the openai api with a max_tokens cap of 4096.
Parameters
- options: IOpenaiChatRequestOptions
  
  The options to use.
Returns Promise<string>
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:58

gpt4-8k

gpt4-8k(options): Promise<string>
Send a gpt4 chat completion request to the openai api with a max_tokens cap of 8k.
Parameters
- options: IOpenaiChatRequestOptions
  
  The options to use.
Returns Promise<string>
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:76

`Protected` handleApiError

handleApiError(error): void
Parses API error codes. When the error is a rate limit error, lowers the concurrency.
Parameters
- error: any
  
  The error to parse.
Returns void
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:196

`Protected` handleCacheOptions

handleCacheOptions(cacheOptions?): IResponseCacheOptions
Handle cache options.
Parameters
- Optional cacheOptions: IResponseCacheOptions
  
  The cache options to handle.
Returns IResponseCacheOptions
Inherited from AbstractApiClient.handleCacheOptions
- Defined in packages/api-util/src/lib/AbstractApiClient.ts:173

`Protected` handleChatOptions

handleChatOptions(options): [CreateChatCompletionRequest, IAsyncRetryOptions, IResponseCacheOptions]
Handle chat options.
Parameters
- options: IOpenaiChatRequestOptions
  
  The options to handle.
Returns [CreateChatCompletionRequest, IAsyncRetryOptions, IResponseCacheOptions]
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:93

`Protected` handleOptions

handleOptions(options): IAbstractApiClientOptions
Handle the options passed to the constructor.
Parameters
- options: IAbstractApiClientOptions
  
  The options to handle.
Returns IAbstractApiClientOptions
Inherited from AbstractApiClient.handleOptions
- Defined in packages/api-util/src/lib/AbstractApiClient.ts:149

`Protected` handleRetryOptions

handleRetryOptions(retryOptions?): IAsyncRetryOptions
Handle retry options.
Parameters
- Optional retryOptions: IAsyncRetryOptions
  
  The retry options to handle.
Returns IAsyncRetryOptions
Inherited from AbstractApiClient.handleRetryOptions
- Defined in packages/api-util/src/lib/AbstractApiClient.ts:165

`Protected` handleTranscribeOptions

handleTranscribeOptions(options): [IOpenaiTranscribeRequest, IAsyncRetryOptions, IResponseCacheOptions]
Handle transcribe options.
Parameters
- options: IOpenaiTranscribeOptions
  
  The options to handle.
Returns [IOpenaiTranscribeRequest, IAsyncRetryOptions, IResponseCacheOptions]
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:117

listModels

listModels(): Promise<Model[]>
Returns a list of the currently available models, and provides basic information about each.

Returns Promise<Model[]>
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:49

`Protected` lowerConcurrency

lowerConcurrency(lowerBy?, raiseAgainBy?, delay?): void
Lower the concurrency to prevent rate limiting. Automatically raises the concurrency again after a delay.
Parameters
- lowerBy: number = 7
  
  The amount to lower the concurrency by.
- raiseAgainBy: number = 6
  
  The amount to raise the concurrency by after a delay.
- delay: number = ...
  
  The delay to wait before raising the concurrency again. This is randomized by +/- 5 seconds to prevent multiple requests from affecting the concurrency at the same time.
Returns void
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:213

stringTokens

stringTokens(string): number[]
Encode a string into tokens.
Parameters
- string: string
  
  The string to encode.
Returns number[]
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:273

transcribe

transcribe(options): Promise<string>
Send a transcribe completion request to the openai api.
Parameters
- options: IOpenaiTranscribeOptions
  
  The options to use.
Returns Promise<string>
- Defined in packages/openai-api-client/src/lib/OpenaiApiClient.ts:85

Class OpenaiApiClient

Hierarchy

Index

Constructors

Properties

Accessors

Methods

Constructors

constructor

Parameters

options: IOpenaiApiClientOptions = {}

Returns OpenaiApiClient

Properties

Readonly apiDefaults

Optional Readonly cache

Readonly cacheDefaults

Readonly client

Protected concurrencyJustReduced

Readonly events

Readonly queue

Readonly retryDefaults

Protected Readonly sendRequest

Type declaration

Type Parameters

T

Parameters

options: { apiRequest: (() => Promise<T>); args: any[]; cache?: IResponseCacheOptions; retry?: IAsyncRetryOptions; }

apiRequest: (() => Promise<T>)

Returns Promise<T>

args: any[]

Optional cache?: IResponseCacheOptions

Optional retry?: IAsyncRetryOptions

Returns Promise<T>

Static Readonly concurrencyDefaults

Accessors

eventNames

Returns string[]

Methods

Protected _chat

Parameters

request: CreateChatCompletionRequest

retry: IAsyncRetryOptions

cache: IResponseCacheOptions

Returns Promise<string>

Protected _transcribe

Parameters

request: IOpenaiTranscribeRequest

retry: IAsyncRetryOptions

cache: IResponseCacheOptions

Returns Promise<string>

Protected assertReponseDataComplete

Parameters

data: CreateChatCompletionResponse | CreateCompletionResponse

Returns void

countTokens

Parameters

string: string

Returns number

Protected deleteDefaultOrUndefinedOptions

Type Parameters

T extends Record<string, any>

Parameters

options: T

defaults: Record<string, any> = {}

Returns T

Protected emit

Type Parameters

T

Parameters

event: string

arg: T

Returns T

gpt3-16k

Parameters

options: IOpenaiChatRequestOptions

Returns Promise<string>

gpt3-8k

Parameters

options: IOpenaiChatRequestOptions

Returns Promise<string>

`Readonly` apiDefaults

`Optional` `Readonly` cache

`Readonly` cacheDefaults

`Readonly` client

`Protected` concurrencyJustReduced

`Readonly` events

`Readonly` queue

`Readonly` retryDefaults

`Protected` `Readonly` sendRequest

options: {
apiRequest: (() => Promise<T>);
args: any[];
cache?: IResponseCacheOptions;
retry?: IAsyncRetryOptions;
}

`Optional` cache?: IResponseCacheOptions

`Optional` retry?: IAsyncRetryOptions

`Static` `Readonly` concurrencyDefaults

`Protected` _chat

`Protected` _transcribe

`Protected` assertReponseDataComplete

`Protected` deleteDefaultOrUndefinedOptions

`Protected` emit

`Protected` handleApiError

`Protected` handleCacheOptions

`Optional` cacheOptions: IResponseCacheOptions

`Protected` handleChatOptions

`Protected` handleOptions

`Protected` handleRetryOptions

`Optional` retryOptions: IAsyncRetryOptions

`Protected` handleTranscribeOptions

`Protected` lowerConcurrency