langchain 0.0.140 → 0.0.141

package/dist/llms/openai.js

@@ -1,14 +1,13 @@
-import { Configuration, OpenAIApi, } from "openai";
+import { OpenAI as OpenAIClient } from "openai";
 import { calculateMaxTokens } from "../base_language/count_tokens.js";
 import { GenerationChunk } from "../schema/index.js";
-import fetchAdapter from "../util/axios-fetch-adapter.js";
 import { getEndpoint } from "../util/azure.js";
 import { chunkArray } from "../util/chunk.js";
-import { getEnvironmentVariable, isNode } from "../util/env.js";
+import { getEnvironmentVariable } from "../util/env.js";
 import { promptLayerTrackRequest } from "../util/prompt-layer.js";
-import { readableStreamToAsyncIterable } from "../util/stream.js";
 import { BaseLLM } from "./base.js";
 import { OpenAIChat } from "./openai-chat.js";
+import { wrapOpenAIClientError } from "../util/openai.js";
 /**
  * Wrapper around OpenAI large language models.
  *
@@ -268,6 +267,12 @@ export class OpenAI extends BaseLLM {
         this.clientConfig = {
             apiKey: this.openAIApiKey,
             organization: this.organization,
+            baseURL: configuration?.basePath ?? fields?.configuration?.basePath,
+            dangerouslyAllowBrowser: true,
+            defaultHeaders: configuration?.baseOptions?.headers ??
+                fields?.configuration?.baseOptions?.headers,
+            defaultQuery: configuration?.baseOptions?.params ??
+                fields?.configuration?.baseOptions?.params,
             ...configuration,
             ...fields?.configuration,
         };
@@ -292,6 +297,7 @@ export class OpenAI extends BaseLLM {
             ...this.modelKwargs,
         };
     }
+    /** @ignore */
     _identifyingParams() {
         return {
             model_name: this.modelName,
@@ -338,94 +344,62 @@ export class OpenAI extends BaseLLM {
         }
         for (let i = 0; i < subPrompts.length; i += 1) {
             const data = params.stream
-                ? await new Promise((resolve, reject) => {
+                ? await (async () => {
                     const choices = [];
                     let response;
-                    let rejected = false;
-                    let resolved = false;
-                    this.completionWithRetry({
+                    const stream = await this.completionWithRetry({
                         ...params,
+                        stream: true,
                         prompt: subPrompts[i],
-                    }, {
-                        signal: options.signal,
-                        ...options.options,
-                        adapter: fetchAdapter,
-                        responseType: "stream",
-                        onmessage: (event) => {
-                            if (event.data?.trim?.() === "[DONE]") {
-                                if (resolved || rejected) {
-                                    return;
-                                }
-                                resolved = true;
-                                resolve({
-                                    ...response,
-                                    choices,
-                                });
+                    }, options);
+                    for await (const message of stream) {
+                        // on the first message set the response properties
+                        if (!response) {
+                            response = {
+                                id: message.id,
+                                object: message.object,
+                                created: message.created,
+                                model: message.model,
+                            };
+                        }
+                        // on all messages, update choice
+                        for (const part of message.choices) {
+                            if (!choices[part.index]) {
+                                choices[part.index] = part;
                             }
                             else {
-                                const data = JSON.parse(event.data);
-                                if (data?.error) {
-                                    if (rejected) {
-                                        return;
-                                    }
-                                    rejected = true;
-                                    reject(data.error);
-                                    return;
-                                }
-                                const message = data;
-                                // on the first message set the response properties
-                                if (!response) {
-                                    response = {
-                                        id: message.id,
-                                        object: message.object,
-                                        created: message.created,
-                                        model: message.model,
-                                    };
-                                }
-                                // on all messages, update choice
-                                for (const part of message.choices) {
-                                    if (part != null && part.index != null) {
-                                        if (!choices[part.index])
-                                            choices[part.index] = {};
-                                        const choice = choices[part.index];
-                                        choice.text = (choice.text ?? "") + (part.text ?? "");
-                                        choice.finish_reason = part.finish_reason;
-                                        choice.logprobs = part.logprobs;
-                                        // eslint-disable-next-line no-void
-                                        void runManager?.handleLLMNewToken(part.text ?? "", {
-                                            prompt: Math.floor(part.index / this.n),
-                                            completion: part.index % this.n,
-                                        });
-                                    }
-                                }
-                                // when all messages are finished, resolve
-                                if (!resolved &&
-                                    !rejected &&
-                                    choices.every((c) => c.finish_reason != null)) {
-                                    resolved = true;
-                                    resolve({
-                                        ...response,
-                                        choices,
-                                    });
-                                }
+                                const choice = choices[part.index];
+                                choice.text += part.text;
+                                choice.finish_reason = part.finish_reason;
+                                choice.logprobs = part.logprobs;
                             }
-                        },
-                    }).catch((error) => {
-                        if (!rejected) {
-                            rejected = true;
-                            reject(error);
+                            void runManager?.handleLLMNewToken(part.text, {
+                                prompt: Math.floor(part.index / this.n),
+                                completion: part.index % this.n,
+                            });
                         }
-                    });
-                })
+                    }
+                    if (options.signal?.aborted) {
+                        throw new Error("AbortError");
+                    }
+                    return { ...response, choices };
+                })()
                 : await this.completionWithRetry({
                     ...params,
+                    stream: false,
                     prompt: subPrompts[i],
                 }, {
                     signal: options.signal,
                     ...options.options,
                 });
             choices.push(...data.choices);
-            const { completion_tokens: completionTokens, prompt_tokens: promptTokens, total_tokens: totalTokens, } = data.usage ?? {};
+            const { completion_tokens: completionTokens, prompt_tokens: promptTokens, total_tokens: totalTokens, } = data.usage
+                ? data.usage
+                : {
+                    completion_tokens: undefined,
+                    prompt_tokens: undefined,
+                    total_tokens: undefined,
+                };
             if (completionTokens) {
                 tokenUsage.completionTokens =
                     (tokenUsage.completionTokens ?? 0) + completionTokens;
@@ -449,18 +423,16 @@ export class OpenAI extends BaseLLM {
             llmOutput: { tokenUsage },
         };
     }
-    // TODO(jacoblee): Refactor with _generate(..., {stream: true}) implementation
-    // when we integrate OpenAI's new SDK.
+    // TODO(jacoblee): Refactor with _generate(..., {stream: true}) implementation?
     async *_streamResponseChunks(input, options, runManager) {
         const params = {
             ...this.invocationParams(options),
             prompt: input,
             stream: true,
         };
-        const streamIterable = this.startStream(params, options);
-        for await (const streamedResponse of streamIterable) {
-            const data = JSON.parse(streamedResponse);
-            const choice = data.choices?.[0];
+        const stream = await this.completionWithRetry(params, options);
+        for await (const data of stream) {
+            const choice = data.choices[0];
             if (!choice) {
                 continue;
             }
@@ -468,103 +440,71 @@ export class OpenAI extends BaseLLM {
                 text: choice.text,
                 generationInfo: {
                     finishReason: choice.finish_reason,
-                    logprobs: choice.logprobs,
                 },
             });
             yield chunk;
             // eslint-disable-next-line no-void
             void runManager?.handleLLMNewToken(chunk.text ?? "");
         }
+        if (options.signal?.aborted) {
+            throw new Error("AbortError");
+        }
     }
-    startStream(request, options) {
-        let done = false;
-        const stream = new TransformStream();
-        const writer = stream.writable.getWriter();
-        const iterable = readableStreamToAsyncIterable(stream.readable);
-        // eslint-disable-next-line @typescript-eslint/no-explicit-any
-        let err;
-        this.completionWithRetry(request, {
-            ...options,
-            adapter: fetchAdapter,
-            responseType: "stream",
-            onmessage: (event) => {
-                if (done)
-                    return;
-                if (event.data?.trim?.() === "[DONE]") {
-                    done = true;
-                    // eslint-disable-next-line no-void
-                    void writer.close();
-                }
-                else {
-                    const data = JSON.parse(event.data);
-                    if (data.error) {
-                        done = true;
-                        throw data.error;
-                    }
-                    // eslint-disable-next-line no-void
-                    void writer.write(event.data);
-                }
-            },
-        }).catch((error) => {
-            if (!done) {
-                err = error;
-                done = true;
-                // eslint-disable-next-line no-void
-                void writer.close();
+    async completionWithRetry(request, options) {
+        const requestOptions = this._getClientOptions(options);
+        return this.caller.call(async () => {
+            try {
+                const res = await this.client.completions.create(request, requestOptions);
+                return res;
+            }
+            catch (e) {
+                const error = wrapOpenAIClientError(e);
+                throw error;
             }
         });
-        return {
-            async next() {
-                const chunk = await iterable.next();
-                if (err) {
-                    throw err;
-                }
-                return chunk;
-            },
-            [Symbol.asyncIterator]() {
-                return this;
-            },
-        };
     }
-    /** @ignore */
-    async completionWithRetry(request, options) {
+    /**
+     * Calls the OpenAI API with retry logic in case of failures.
+     * @param request The request to send to the OpenAI API.
+     * @param options Optional configuration for the API call.
+     * @returns The response from the OpenAI API.
+     */
+    _getClientOptions(options) {
         if (!this.client) {
             const openAIEndpointConfig = {
                 azureOpenAIApiDeploymentName: this.azureOpenAIApiDeploymentName,
                 azureOpenAIApiInstanceName: this.azureOpenAIApiInstanceName,
                 azureOpenAIApiKey: this.azureOpenAIApiKey,
                 azureOpenAIBasePath: this.azureOpenAIBasePath,
-                basePath: this.clientConfig.basePath,
+                baseURL: this.clientConfig.baseURL,
             };
             const endpoint = getEndpoint(openAIEndpointConfig);
-            const clientConfig = new Configuration({
+            const params = {
                 ...this.clientConfig,
-                basePath: endpoint,
-                baseOptions: {
-                    timeout: this.timeout,
-                    ...this.clientConfig.baseOptions,
-                },
-            });
-            this.client = new OpenAIApi(clientConfig);
+                baseURL: endpoint,
+                timeout: this.timeout,
+                maxRetries: 0,
+            };
+            if (!params.baseURL) {
+                delete params.baseURL;
+            }
+            this.client = new OpenAIClient(params);
         }
-        const axiosOptions = {
-            adapter: isNode() ? undefined : fetchAdapter,
-            ...this.clientConfig.baseOptions,
+        const requestOptions = {
+            ...this.clientConfig,
             ...options,
         };
         if (this.azureOpenAIApiKey) {
-            axiosOptions.headers = {
+            requestOptions.headers = {
                 "api-key": this.azureOpenAIApiKey,
-                ...axiosOptions.headers,
+                ...requestOptions.headers,
             };
-            axiosOptions.params = {
+            requestOptions.query = {
                 "api-version": this.azureOpenAIApiVersion,
-                ...axiosOptions.params,
+                ...requestOptions.query,
             };
         }
-        return this.caller
-            .call(this.client.createCompletion.bind(this.client), request, axiosOptions)
-            .then((res) => res.data);
+        return requestOptions;
     }
     _llmType() {
         return "openai";
@@ -615,19 +555,6 @@ export class PromptLayerOpenAI extends OpenAI {
             throw new Error("Missing PromptLayer API key");
         }
     }
-    /**
-     * Calls the OpenAI API with retry logic in case of failures.
-     * @param request The request to send to the OpenAI API.
-     * @param options Optional configuration for the API call.
-     * @returns The response from the OpenAI API.
-     */
-    async completionWithRetry(request, options) {
-        if (request.stream) {
-            return super.completionWithRetry(request, options);
-        }
-        const response = await super.completionWithRetry(request);
-        return response;
-    }
     async _generate(prompts, options, runManager) {
         const requestStartTime = Date.now();
         const generations = await super._generate(prompts, options, runManager);
@@ -637,7 +564,9 @@ export class PromptLayerOpenAI extends OpenAI {
                 text: generations.generations[i][0].text,
                 llm_output: generations.llmOutput,
             };
-            const promptLayerRespBody = await promptLayerTrackRequest(this.caller, "langchain.PromptLayerOpenAI", [prompts[i]], this._identifyingParams(), this.plTags, parsedResp, requestStartTime, requestEndTime, this.promptLayerApiKey);
+            const promptLayerRespBody = await promptLayerTrackRequest(this.caller, "langchain.PromptLayerOpenAI", 
+            // eslint-disable-next-line @typescript-eslint/no-explicit-any
+            { ...this._identifyingParams(), prompt: prompts[i] }, this.plTags, parsedResp, requestStartTime, requestEndTime, this.promptLayerApiKey);
             let promptLayerRequestId;
             if (this.returnPromptLayerId === true) {
                 if (promptLayerRespBody && promptLayerRespBody.success === true) {

package/dist/load/import_map.cjs

@@ -24,8 +24,8 @@ var __importStar = (this && this.__importStar) || function (mod) {
     return result;
 };
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.retrievers__parent_document = exports.retrievers__multi_vector = exports.retrievers__document_compressors = exports.retrievers__contextual_compression = exports.retrievers__databerry = exports.retrievers__remote = exports.output_parsers = exports.callbacks = exports.schema__storage = exports.schema__runnable = exports.schema__retriever = exports.schema__query_constructor = exports.schema__output_parser = exports.schema = exports.chat_models__minimax = exports.chat_models__ollama = exports.chat_models__baiduwenxin = exports.chat_models__anthropic = exports.chat_models__openai = exports.chat_models__base = exports.document_transformers__openai_functions = exports.document_loaders__web__sort_xyz_blockchain = exports.document_loaders__web__serpapi = exports.document_loaders__base = exports.document = exports.memory = exports.text_splitter = exports.vectorstores__xata = exports.vectorstores__vectara = exports.vectorstores__prisma = exports.vectorstores__memory = exports.vectorstores__base = exports.prompts = exports.llms__ollama = exports.llms__aleph_alpha = exports.llms__ai21 = exports.llms__openai = exports.llms__base = exports.embeddings__minimax = exports.embeddings__openai = exports.embeddings__fake = exports.embeddings__cache_backed = exports.embeddings__base = exports.chains__openai_functions = exports.chains = exports.tools = exports.base_language = exports.agents__toolkits = exports.agents = exports.load__serializable = void 0;
-exports.evaluation = exports.experimental__plan_and_execute = exports.experimental__generative_agents = exports.experimental__babyagi = exports.experimental__autogpt = exports.util__math = exports.storage__in_memory = exports.stores__message__in_memory = exports.stores__file__in_memory = exports.stores__doc__in_memory = exports.cache = exports.retrievers__vespa = exports.retrievers__score_threshold = exports.retrievers__hyde = exports.retrievers__document_compressors__chain_extract = exports.retrievers__time_weighted = void 0;
+exports.retrievers__multi_vector = exports.retrievers__document_compressors = exports.retrievers__contextual_compression = exports.retrievers__databerry = exports.retrievers__remote = exports.output_parsers = exports.callbacks = exports.schema__storage = exports.schema__runnable = exports.schema__retriever = exports.schema__query_constructor = exports.schema__output_parser = exports.schema__document = exports.schema = exports.chat_models__minimax = exports.chat_models__ollama = exports.chat_models__baiduwenxin = exports.chat_models__anthropic = exports.chat_models__openai = exports.chat_models__base = exports.document_transformers__openai_functions = exports.document_loaders__web__sort_xyz_blockchain = exports.document_loaders__web__serpapi = exports.document_loaders__base = exports.document = exports.memory = exports.text_splitter = exports.vectorstores__xata = exports.vectorstores__vectara = exports.vectorstores__prisma = exports.vectorstores__memory = exports.vectorstores__base = exports.prompts = exports.llms__ollama = exports.llms__aleph_alpha = exports.llms__ai21 = exports.llms__openai = exports.llms__base = exports.embeddings__minimax = exports.embeddings__openai = exports.embeddings__fake = exports.embeddings__cache_backed = exports.embeddings__base = exports.chains__openai_functions = exports.chains = exports.tools = exports.base_language = exports.agents__toolkits = exports.agents = exports.load__serializable = void 0;
+exports.evaluation = exports.experimental__plan_and_execute = exports.experimental__generative_agents = exports.experimental__babyagi = exports.experimental__autogpt = exports.util__math = exports.storage__in_memory = exports.stores__message__in_memory = exports.stores__file__in_memory = exports.stores__doc__in_memory = exports.cache = exports.retrievers__vespa = exports.retrievers__score_threshold = exports.retrievers__hyde = exports.retrievers__document_compressors__chain_extract = exports.retrievers__time_weighted = exports.retrievers__parent_document = void 0;
 exports.load__serializable = __importStar(require("../load/serializable.cjs"));
 exports.agents = __importStar(require("../agents/index.cjs"));
 exports.agents__toolkits = __importStar(require("../agents/toolkits/index.cjs"));
@@ -63,6 +63,7 @@ exports.chat_models__baiduwenxin = __importStar(require("../chat_models/baiduwen
 exports.chat_models__ollama = __importStar(require("../chat_models/ollama.cjs"));
 exports.chat_models__minimax = __importStar(require("../chat_models/minimax.cjs"));
 exports.schema = __importStar(require("../schema/index.cjs"));
+exports.schema__document = __importStar(require("../schema/document.cjs"));
 exports.schema__output_parser = __importStar(require("../schema/output_parser.cjs"));
 exports.schema__query_constructor = __importStar(require("../schema/query_constructor.cjs"));
 exports.schema__retriever = __importStar(require("../schema/retriever.cjs"));

package/dist/load/import_map.d.ts

@@ -35,6 +35,7 @@ export * as chat_models__baiduwenxin from "../chat_models/baiduwenxin.js";
 export * as chat_models__ollama from "../chat_models/ollama.js";
 export * as chat_models__minimax from "../chat_models/minimax.js";
 export * as schema from "../schema/index.js";
+export * as schema__document from "../schema/document.js";
 export * as schema__output_parser from "../schema/output_parser.js";
 export * as schema__query_constructor from "../schema/query_constructor.js";
 export * as schema__retriever from "../schema/retriever.js";

package/dist/load/import_map.js

@@ -36,6 +36,7 @@ export * as chat_models__baiduwenxin from "../chat_models/baiduwenxin.js";
 export * as chat_models__ollama from "../chat_models/ollama.js";
 export * as chat_models__minimax from "../chat_models/minimax.js";
 export * as schema from "../schema/index.js";
+export * as schema__document from "../schema/document.js";
 export * as schema__output_parser from "../schema/output_parser.js";
 export * as schema__query_constructor from "../schema/query_constructor.js";
 export * as schema__retriever from "../schema/retriever.js";

package/dist/schema/index.d.ts

@@ -1,4 +1,4 @@
-import { ChatCompletionRequestMessageFunctionCall } from "openai";
+import type { OpenAI as OpenAIClient } from "openai";
 import { Document } from "../document.js";
 import { Serializable } from "../load/serializable.js";
 export declare const RUN_KEY = "__run";
@@ -64,7 +64,7 @@ export interface BaseMessageFields {
     content: string;
     name?: string;
     additional_kwargs?: {
-        function_call?: ChatCompletionRequestMessageFunctionCall;
+        function_call?: OpenAIClient.Chat.ChatCompletionMessage.FunctionCall;
         [key: string]: unknown;
     };
 }

package/dist/tools/convert_to_openai.d.ts

@@ -1,4 +1,4 @@
-import { ChatCompletionFunctions } from "openai";
+import type { OpenAI as OpenAIClient } from "openai";
 import { StructuredTool } from "./base.js";
 /**
  * Formats a `StructuredTool` instance into a format that is compatible
@@ -6,4 +6,4 @@ import { StructuredTool } from "./base.js";
  * function to convert the schema of the `StructuredTool` into a JSON
  * schema, which is then used as the parameters for the OpenAI function.
  */
-export declare function formatToOpenAIFunction(tool: StructuredTool): ChatCompletionFunctions;
+export declare function formatToOpenAIFunction(tool: StructuredTool): OpenAIClient.Chat.ChatCompletionCreateParams.Function;

package/dist/types/openai-types.d.ts

@@ -1,5 +1,4 @@
-import { AxiosRequestConfig } from "axios";
-import { ChatCompletionRequestMessage } from "openai";
+import type { OpenAI as OpenAIClient } from "openai";
 import { BaseLanguageModelCallOptions } from "../base_language/index.js";
 export declare interface OpenAIBaseInput {
     /** Sampling temperature to use */
@@ -42,11 +41,23 @@ export declare interface OpenAIBaseInput {
      */
     openAIApiKey?: string;
 }
+export type OpenAICoreRequestOptions<Req extends object = Record<string, unknown>> = {
+    path?: string;
+    query?: Req | undefined;
+    body?: Req | undefined;
+    headers?: Record<string, string | null | undefined> | undefined;
+    maxRetries?: number;
+    stream?: boolean | undefined;
+    timeout?: number;
+    httpAgent?: any;
+    signal?: AbortSignal | undefined | null;
+    idempotencyKey?: string;
+};
 export interface OpenAICallOptions extends BaseLanguageModelCallOptions {
     /**
      * Additional options to pass to the underlying axios request.
      */
-    options?: AxiosRequestConfig;
+    options?: OpenAICoreRequestOptions;
 }
 /**
  * Input to OpenAI class.
@@ -57,9 +68,21 @@ export declare interface OpenAIInput extends OpenAIBaseInput {
     /** Batch size to use when passing multiple documents to generate */
     batchSize: number;
 }
+/**
+ * @deprecated Use "baseURL", "defaultHeaders", and "defaultParams" instead.
+ */
+export interface LegacyOpenAIInput {
+    /** @deprecated Use baseURL instead */
+    basePath?: string;
+    /** @deprecated Use defaultHeaders and defaultQuery instead */
+    baseOptions?: {
+        headers?: Record<string, string>;
+        params?: Record<string, string>;
+    };
+}
 export interface OpenAIChatInput extends OpenAIBaseInput {
     /** ChatGPT messages to pass as a prefix to the prompt */
-    prefixMessages?: ChatCompletionRequestMessage[];
+    prefixMessages?: OpenAIClient.Chat.CreateChatCompletionRequestMessage[];
 }
 export declare interface AzureOpenAIInput {
     /**

package/dist/util/async_caller.cjs

@@ -9,6 +9,7 @@ const p_queue_1 = __importDefault(require("p-queue"));
 const STATUS_NO_RETRY = [
     400,
     401,
+    402,
     403,
     404,
     405,
@@ -69,24 +70,26 @@ class AsyncCaller {
             onFailedAttempt(error) {
                 if (error.message.startsWith("Cancel") ||
                     error.message.startsWith("TimeoutError") ||
-                    error.message.startsWith("AbortError")) {
+                    error.name === "TimeoutError" ||
+                    error.message.startsWith("AbortError") ||
+                    error.name === "AbortError") {
                     throw error;
                 }
                 // eslint-disable-next-line @typescript-eslint/no-explicit-any
                 if (error?.code === "ECONNABORTED") {
                     throw error;
                 }
+                const status = 
                 // eslint-disable-next-line @typescript-eslint/no-explicit-any
-                const status = error?.response?.status;
+                error?.response?.status ?? error?.status;
                 if (status && STATUS_NO_RETRY.includes(+status)) {
                     throw error;
                 }
                 // eslint-disable-next-line @typescript-eslint/no-explicit-any
-                const data = error?.response?.data;
-                if (data?.error?.code === "insufficient_quota") {
-                    const error = new Error(data?.error?.message);
-                    error.name = "InsufficientQuotaError";
-                    throw error;
+                if (error?.error?.code === "insufficient_quota") {
+                    const err = new Error(error?.message);
+                    err.name = "InsufficientQuotaError";
+                    throw err;
                 }
             },
             retries: this.maxRetries,

package/dist/util/async_caller.js

@@ -3,6 +3,7 @@ import PQueueMod from "p-queue";
 const STATUS_NO_RETRY = [
     400,
     401,
+    402,
     403,
     404,
     405,
@@ -63,24 +64,26 @@ export class AsyncCaller {
             onFailedAttempt(error) {
                 if (error.message.startsWith("Cancel") ||
                     error.message.startsWith("TimeoutError") ||
-                    error.message.startsWith("AbortError")) {
+                    error.name === "TimeoutError" ||
+                    error.message.startsWith("AbortError") ||
+                    error.name === "AbortError") {
                     throw error;
                 }
                 // eslint-disable-next-line @typescript-eslint/no-explicit-any
                 if (error?.code === "ECONNABORTED") {
                     throw error;
                 }
+                const status = 
                 // eslint-disable-next-line @typescript-eslint/no-explicit-any
-                const status = error?.response?.status;
+                error?.response?.status ?? error?.status;
                 if (status && STATUS_NO_RETRY.includes(+status)) {
                     throw error;
                 }
                 // eslint-disable-next-line @typescript-eslint/no-explicit-any
-                const data = error?.response?.data;
-                if (data?.error?.code === "insufficient_quota") {
-                    const error = new Error(data?.error?.message);
-                    error.name = "InsufficientQuotaError";
-                    throw error;
+                if (error?.error?.code === "insufficient_quota") {
+                    const err = new Error(error?.message);
+                    err.name = "InsufficientQuotaError";
+                    throw err;
                 }
             },
             retries: this.maxRetries,

package/dist/util/azure.cjs

@@ -11,12 +11,12 @@ exports.getEndpoint = void 0;
  * @property {string} config.azureOpenAIApiInstanceName - The instance name of Azure OpenAI.
  * @property {string} config.azureOpenAIApiKey - The API Key for Azure OpenAI.
  * @property {string} config.azureOpenAIBasePath - The base path for Azure OpenAI.
- * @property {string} config.basePath - Some other custom base path URL.
+ * @property {string} config.baseURL - Some other custom base path URL.
  *
  * The function operates as follows:
  * - If both `azureOpenAIBasePath` and `azureOpenAIApiDeploymentName` (plus `azureOpenAIApiKey`) are provided, it returns an URL combining these two parameters (`${azureOpenAIBasePath}/${azureOpenAIApiDeploymentName}`).
  * - If `azureOpenAIApiKey` is provided, it checks for `azureOpenAIApiInstanceName` and `azureOpenAIApiDeploymentName` and throws an error if any of these is missing. If both are provided, it generates an URL incorporating these parameters.
- * - If none of the above conditions are met, return any custom `basePath`.
+ * - If none of the above conditions are met, return any custom `baseURL`.
  * - The function returns the generated URL as a string, or undefined if no custom paths are specified.
  *
  * @throws Will throw an error if the necessary parameters for generating the URL are missing.
@@ -24,7 +24,7 @@ exports.getEndpoint = void 0;
  * @returns {string | undefined} The generated (Azure) OpenAI endpoint URL.
  */
 function getEndpoint(config) {
-    const { azureOpenAIApiDeploymentName, azureOpenAIApiInstanceName, azureOpenAIApiKey, azureOpenAIBasePath, basePath, } = config;
+    const { azureOpenAIApiDeploymentName, azureOpenAIApiInstanceName, azureOpenAIApiKey, azureOpenAIBasePath, baseURL, } = config;
     if (azureOpenAIApiKey &&
         azureOpenAIBasePath &&
         azureOpenAIApiDeploymentName) {
@@ -39,6 +39,6 @@ function getEndpoint(config) {
         }
         return `https://${azureOpenAIApiInstanceName}.openai.azure.com/openai/deployments/${azureOpenAIApiDeploymentName}`;
     }
-    return basePath;
+    return baseURL;
 }
 exports.getEndpoint = getEndpoint;

package/dist/util/azure.d.ts

@@ -3,7 +3,7 @@ export interface OpenAIEndpointConfig {
     azureOpenAIApiInstanceName?: string;
     azureOpenAIApiKey?: string;
     azureOpenAIBasePath?: string;
-    basePath?: string;
+    baseURL?: string;
 }
 /**
  * This function generates an endpoint URL for (Azure) OpenAI
@@ -15,12 +15,12 @@ export interface OpenAIEndpointConfig {
  * @property {string} config.azureOpenAIApiInstanceName - The instance name of Azure OpenAI.
  * @property {string} config.azureOpenAIApiKey - The API Key for Azure OpenAI.
  * @property {string} config.azureOpenAIBasePath - The base path for Azure OpenAI.
- * @property {string} config.basePath - Some other custom base path URL.
+ * @property {string} config.baseURL - Some other custom base path URL.
  *
  * The function operates as follows:
  * - If both `azureOpenAIBasePath` and `azureOpenAIApiDeploymentName` (plus `azureOpenAIApiKey`) are provided, it returns an URL combining these two parameters (`${azureOpenAIBasePath}/${azureOpenAIApiDeploymentName}`).
  * - If `azureOpenAIApiKey` is provided, it checks for `azureOpenAIApiInstanceName` and `azureOpenAIApiDeploymentName` and throws an error if any of these is missing. If both are provided, it generates an URL incorporating these parameters.
- * - If none of the above conditions are met, return any custom `basePath`.
+ * - If none of the above conditions are met, return any custom `baseURL`.
  * - The function returns the generated URL as a string, or undefined if no custom paths are specified.
  *
  * @throws Will throw an error if the necessary parameters for generating the URL are missing.