@falai/agent 0.1.0-alpha2

package/src/providers/OpenAIProvider.ts

@@ -0,0 +1,602 @@
+/**
+ * OpenAI provider implementation with retry and backup models
+ */
+
+import OpenAI from "openai";
+import type { ChatCompletionCreateParamsNonStreaming } from "openai/resources/chat/completions";
+
+import type {
+  AiProvider,
+  GenerateMessageInput,
+  GenerateMessageOutput,
+  GenerateMessageStreamChunk,
+  AgentStructuredResponse,
+  StructuredSchema,
+} from "../types";
+import { withTimeoutAndRetry, logger } from "../utils";
+import { FunctionParameters } from "openai/resources/shared.mjs";
+
+const DEFAULT_RETRY_CONFIG = {
+  timeout: 60000,
+  retries: 3,
+};
+
+/**
+ * Configuration options for OpenAI provider
+ * Uses types from openai package
+ */
+export interface OpenAIProviderOptions {
+  /** OpenAI API key */
+  apiKey: string;
+  /** Organization ID (optional) */
+  organization?: string;
+  /** Model to use (required) - e.g., "gpt-5", "gpt-5-mini" */
+  model: string;
+  /** Backup models to try if primary fails (default: []) */
+  backupModels?: string[];
+  /** Default parameters - uses ChatCompletionCreateParamsNonStreaming from openai package */
+  config?: Partial<
+    Omit<ChatCompletionCreateParamsNonStreaming, "model" | "messages">
+  >;
+  /** Retry configuration */
+  retryConfig?: {
+    timeout?: number;
+    retries?: number;
+  };
+}
+
+/**
+ * Type guard for errors with status/code properties
+ */
+interface ErrorWithStatus {
+  status?: number;
+  code?: string;
+  message?: string;
+  type?: string;
+}
+
+/**
+ * Type guard to check if error is ErrorWithStatus
+ */
+function isErrorWithStatus(error: unknown): error is ErrorWithStatus {
+  return (
+    typeof error === "object" &&
+    error !== null &&
+    ("status" in error || "code" in error || "message" in error)
+  );
+}
+
+/**
+ * Safely extract error message
+ */
+function getErrorMessage(error: unknown): string {
+  if (error instanceof Error) {
+    return error.message;
+  }
+  if (isErrorWithStatus(error) && error.message) {
+    return error.message;
+  }
+  return String(error);
+}
+
+/**
+ * Determines if an error should trigger backup model usage
+ */
+const shouldUseBackupModel = (error: unknown): boolean => {
+  if (!isErrorWithStatus(error)) {
+    return false;
+  }
+
+  // Server errors
+  if (error.status === 500 || error.status === 503) {
+    return true;
+  }
+
+  // Rate limiting
+  if (error.status === 429) {
+    return true;
+  }
+
+  // Model overloaded or unavailable
+  if (error.code === "model_not_found" || error.code === "model_overloaded") {
+    return true;
+  }
+
+  const message = getErrorMessage(error);
+  if (
+    message.includes("overloaded") ||
+    message.includes("unavailable") ||
+    message.includes("internal error") ||
+    message.includes("Internal error")
+  ) {
+    return true;
+  }
+
+  return false;
+};
+
+/**
+ * OpenAI provider implementation with backup models and retry logic
+ */
+export class OpenAIProvider implements AiProvider {
+  public readonly name = "openai";
+  private client: OpenAI;
+  private primaryModel: string;
+  private backupModels: string[];
+  private config?: Partial<
+    Omit<ChatCompletionCreateParamsNonStreaming, "model" | "messages">
+  >;
+  private retryConfig: { timeout: number; retries: number };
+
+  constructor(options: OpenAIProviderOptions) {
+    const {
+      apiKey,
+      organization,
+      model,
+      backupModels = [],
+      config,
+      retryConfig,
+    } = options;
+
+    if (!apiKey) {
+      throw new Error("OpenAI API key is required");
+    }
+
+    if (!model) {
+      throw new Error("Model is required. Example: 'gpt-5' or 'gpt-5-mini'");
+    }
+
+    // Dynamic import to avoid bundling issues
+
+    this.client = new OpenAI({
+      apiKey,
+      organization,
+    });
+    this.primaryModel = model;
+    this.backupModels = backupModels;
+    this.config = config;
+    this.retryConfig = {
+      timeout: retryConfig?.timeout || DEFAULT_RETRY_CONFIG.timeout,
+      retries: retryConfig?.retries || DEFAULT_RETRY_CONFIG.retries,
+    };
+  }
+
+  /**
+   * Adapt common schema format to OpenAI's format.
+   * OpenAI uses standard JSON Schema, so this is mostly a passthrough.
+   *
+   * @private
+   */
+  private adaptSchemaForOpenAI(
+    schema: StructuredSchema
+  ): Record<string, unknown> {
+    // OpenAI's responses.parse API uses standard JSON Schema
+    // Our StructuredSchema is already JSON Schema compatible
+    return schema as Record<string, unknown>;
+  }
+
+  async generateMessage<
+    TContext = unknown,
+    TStructured = AgentStructuredResponse
+  >(
+    input: GenerateMessageInput<TContext>
+  ): Promise<GenerateMessageOutput<TStructured>> {
+    return this.generateWithBackup<TContext, TStructured>(input);
+  }
+
+  async *generateMessageStream<
+    TContext = unknown,
+    TStructured = AgentStructuredResponse
+  >(
+    input: GenerateMessageInput<TContext>
+  ): AsyncGenerator<GenerateMessageStreamChunk<TStructured>> {
+    yield* this.generateStreamWithBackup<TContext, TStructured>(input);
+  }
+
+  private async generateWithBackup<
+    TContext = unknown,
+    TStructured = AgentStructuredResponse
+  >(
+    input: GenerateMessageInput<TContext>
+  ): Promise<GenerateMessageOutput<TStructured>> {
+    // Try primary model first
+    try {
+      return await this.generateWithModel<TContext, TStructured>(
+        this.primaryModel,
+        input
+      );
+    } catch (primaryError: unknown) {
+      const primaryErrMsg = getErrorMessage(primaryError);
+      logger.warn(
+        `[OPENAI] Primary model ${this.primaryModel} failed: ${primaryErrMsg}`
+      );
+
+      if (!shouldUseBackupModel(primaryError)) {
+        throw primaryError;
+      }
+
+      logger.debug(`[OPENAI] Trying backup models`);
+
+      let lastBackupError: unknown = primaryError;
+
+      for (let i = 0; i < this.backupModels.length; i++) {
+        const backupModel = this.backupModels[i];
+        logger.debug(
+          `[OPENAI] Trying backup model ${i + 1}/${
+            this.backupModels.length
+          }: ${backupModel}`
+        );
+
+        try {
+          const result = await this.generateWithModel(backupModel, input);
+          logger.debug(`[OPENAI] Backup model ${backupModel} succeeded`);
+          return result as GenerateMessageOutput<TStructured>;
+        } catch (backupError: unknown) {
+          const backupErrMsg = getErrorMessage(backupError);
+          logger.warn(
+            `[OPENAI] Backup model ${backupModel} failed: ${backupErrMsg}`
+          );
+          lastBackupError = backupError;
+
+          if (
+            !shouldUseBackupModel(backupError) &&
+            i < this.backupModels.length - 1
+          ) {
+            logger.debug(
+              `[OPENAI] Backup model error doesn't qualify for further attempts`
+            );
+            break;
+          }
+        }
+      }
+
+      const lastBackupErrMsg = getErrorMessage(lastBackupError);
+      logger.error(
+        `[OPENAI] All models failed. Primary: ${primaryErrMsg}, Last backup: ${lastBackupErrMsg}`
+      );
+      throw lastBackupError;
+    }
+  }
+
+  private async generateWithModel<
+    TContext = unknown,
+    TStructured = AgentStructuredResponse
+  >(
+    model: string,
+    input: GenerateMessageInput<TContext>
+  ): Promise<GenerateMessageOutput<TStructured>> {
+    const operation = async (): Promise<GenerateMessageOutput> => {
+      const params: ChatCompletionCreateParamsNonStreaming = {
+        model,
+        messages: [
+          {
+            role: "user",
+            content: input.prompt,
+          },
+        ],
+        ...this.config,
+      };
+
+      // Override with input parameters if provided
+      if (input.parameters?.maxOutputTokens !== undefined) {
+        params.max_tokens = input.parameters.maxOutputTokens;
+      }
+
+      // Add tools if provided
+      if (input.tools && input.tools.length > 0) {
+        params.tools = input.tools.map((tool) => ({
+          type: "function" as const,
+          function: {
+            name: tool.name || tool.id,
+            description: tool.description,
+            parameters: tool.parameters as FunctionParameters, // JSON schema
+          },
+        }));
+        params.tool_choice = "auto";
+      }
+
+      // Use structured output API if JSON schema is provided
+      if (input.parameters?.jsonSchema) {
+        const response = await this.client.responses.parse({
+          model,
+          instructions: input.prompt,
+          input: "",
+          reasoning: {
+            effort: input.parameters?.reasoning?.effort || "low",
+          },
+          text: {
+            format: {
+              type: "json_schema",
+              name: input.parameters?.schemaName || "structured_output",
+              // Adapt common schema format to OpenAI's format
+              schema: this.adaptSchemaForOpenAI(input.parameters.jsonSchema),
+            },
+          },
+        });
+
+        if (!response.output_parsed) {
+          throw new Error("No parsed output returned from OpenAI");
+        }
+
+        const structured = response.output_parsed as AgentStructuredResponse;
+        const message = structured.message;
+
+        return {
+          message,
+          metadata: {
+            model: response.model,
+            tokensUsed: response.usage?.total_tokens,
+            promptTokens: response.usage?.input_tokens,
+            completionTokens: response.usage?.output_tokens,
+          },
+          structured,
+        };
+      }
+
+      // Fall back to regular chat completions API if no schema provided
+      const response = await this.client.chat.completions.create(params);
+
+      const message = response.choices[0]?.message?.content;
+      if (!message) {
+        throw new Error("No response from OpenAI");
+      }
+
+      let toolCalls: Array<{
+        toolName: string;
+        arguments: Record<string, unknown>;
+      }> = [];
+      if (response.choices?.[0]?.message?.tool_calls) {
+        toolCalls = response.choices[0].message.tool_calls
+          .filter((toolCall) => toolCall.type === "function")
+          .map((toolCall) => {
+            let toolCallArguments: Record<string, unknown> = {};
+            try {
+              toolCallArguments = JSON.parse(
+                toolCall.function.arguments
+              ) as Record<string, unknown>;
+            } catch (error) {
+              logger.warn(
+                `[OPENAI] Failed to parse tool call arguments: ${getErrorMessage(
+                  error
+                )}`
+              );
+              toolCallArguments = {};
+            }
+            return {
+              toolName: toolCall.function.name,
+              arguments: toolCallArguments,
+            };
+          });
+      }
+      // Extract tool calls from response
+
+      return {
+        message,
+        metadata: {
+          model: response.model,
+          finishReason: response.choices[0]?.finish_reason,
+          tokensUsed: response.usage?.total_tokens,
+          promptTokens: response.usage?.prompt_tokens,
+          completionTokens: response.usage?.completion_tokens,
+        },
+        structured:
+          toolCalls.length > 0
+            ? ({ message, toolCalls } as AgentStructuredResponse)
+            : undefined,
+      };
+    };
+
+    return withTimeoutAndRetry(
+      operation,
+      this.retryConfig.timeout,
+      this.retryConfig.retries,
+      `OpenAI ${model}`
+    ) as Promise<GenerateMessageOutput<TStructured>>;
+  }
+
+  private async *generateStreamWithBackup<
+    TContext = unknown,
+    TStructured = AgentStructuredResponse
+  >(
+    input: GenerateMessageInput<TContext>
+  ): AsyncGenerator<GenerateMessageStreamChunk<TStructured>> {
+    // Try primary model first
+    try {
+      yield* this.generateStreamWithModel<TContext, TStructured>(
+        this.primaryModel,
+        input
+      );
+    } catch (primaryError: unknown) {
+      const primaryErrMsg = getErrorMessage(primaryError);
+      logger.warn(
+        `[OPENAI] Primary model ${this.primaryModel} failed: ${primaryErrMsg}`
+      );
+
+      if (!shouldUseBackupModel(primaryError)) {
+        throw primaryError;
+      }
+
+      logger.debug(`[OPENAI] Trying backup models for streaming`);
+
+      let lastBackupError: unknown = primaryError;
+
+      for (let i = 0; i < this.backupModels.length; i++) {
+        const backupModel = this.backupModels[i];
+        logger.debug(
+          `[OPENAI] Trying backup model ${i + 1}/${
+            this.backupModels.length
+          }: ${backupModel}`
+        );
+
+        try {
+          yield* this.generateStreamWithModel<TContext, TStructured>(
+            backupModel,
+            input
+          );
+          logger.debug(`[OPENAI] Backup model ${backupModel} succeeded`);
+          return;
+        } catch (backupError: unknown) {
+          const backupErrMsg = getErrorMessage(backupError);
+          logger.warn(
+            `[OPENAI] Backup model ${backupModel} failed: ${backupErrMsg}`
+          );
+          lastBackupError = backupError;
+
+          if (
+            !shouldUseBackupModel(backupError) &&
+            i < this.backupModels.length - 1
+          ) {
+            logger.debug(
+              `[OPENAI] Backup model error doesn't qualify for further attempts`
+            );
+            break;
+          }
+        }
+      }
+
+      const lastBackupErrMsg = getErrorMessage(lastBackupError);
+      logger.error(
+        `[OPENAI] All models failed. Primary: ${primaryErrMsg}, Last backup: ${lastBackupErrMsg}`
+      );
+      throw lastBackupError;
+    }
+  }
+
+  private async *generateStreamWithModel<
+    TContext = unknown,
+    TStructured = AgentStructuredResponse
+  >(
+    model: string,
+    input: GenerateMessageInput<TContext>
+  ): AsyncGenerator<GenerateMessageStreamChunk<TStructured>> {
+    const params = {
+      ...this.config,
+      model,
+      messages: [
+        {
+          role: "user" as const,
+          content: input.prompt,
+        },
+      ],
+      stream: true as const,
+    };
+
+    // Override with input parameters if provided
+    if (input.parameters?.maxOutputTokens !== undefined) {
+      params.max_tokens = input.parameters.maxOutputTokens;
+    }
+
+    // Add tools if provided
+    if (input.tools && input.tools.length > 0) {
+      params.tools = input.tools.map((tool) => ({
+        type: "function" as const,
+        function: {
+          name: tool.name || tool.id,
+          description: tool.description,
+          parameters: tool.parameters as FunctionParameters, // JSON schema
+        },
+      }));
+      params.tool_choice = "auto";
+    }
+
+    // Streaming path does not support responses.parse; if schema present,
+    // request JSON object and parse at the end.
+    if (input.parameters?.jsonSchema) {
+      params.response_format = { type: "json_object" };
+    }
+
+    const stream = await this.client.chat.completions.create(params);
+
+    let accumulated = "";
+    let currentModel = model;
+    let finishReason: string | undefined;
+    let promptTokens: number | undefined;
+    let completionTokens: number | undefined;
+    let totalTokens: number | undefined;
+    const toolCalls: Array<{
+      toolName: string;
+      arguments: Record<string, unknown>;
+    }> = [];
+
+    for await (const chunk of stream) {
+      currentModel = chunk.model;
+      const delta = chunk.choices[0]?.delta?.content || "";
+
+      // Extract tool calls from delta
+      if (chunk.choices[0]?.delta?.tool_calls) {
+        for (const toolCall of chunk.choices[0].delta.tool_calls) {
+          if (toolCall.function) {
+            let toolCallArguments: Record<string, unknown> = {};
+            try {
+              toolCallArguments = toolCall.function.arguments
+                ? (JSON.parse(toolCall.function.arguments) as Record<
+                    string,
+                    unknown
+                  >)
+                : {};
+            } catch (error) {
+              logger.warn(
+                `[OPENAI] Failed to parse tool call arguments in stream: ${getErrorMessage(
+                  error
+                )}`
+              );
+              toolCallArguments = {};
+            }
+            toolCalls.push({
+              toolName: toolCall.function.name || "",
+              arguments: toolCallArguments,
+            });
+          }
+        }
+      }
+
+      if (delta) {
+        accumulated += delta;
+        yield {
+          delta,
+          accumulated,
+          done: false,
+        };
+      }
+
+      if (chunk.choices[0]?.finish_reason) {
+        finishReason = chunk.choices[0].finish_reason;
+      }
+
+      // OpenAI includes usage in the final chunk for some models
+      if (chunk.usage) {
+        promptTokens = chunk.usage.prompt_tokens;
+        completionTokens = chunk.usage.completion_tokens;
+        totalTokens = chunk.usage.total_tokens;
+      }
+    }
+
+    // Parse JSON response if schema was provided
+    let structured: TStructured | undefined;
+    if (input.parameters?.jsonSchema && accumulated) {
+      try {
+        structured = JSON.parse(accumulated) as TStructured;
+      } catch (error) {
+        logger.warn(
+          "[OPENAI] Failed to parse JSON response in stream:",
+          error
+        );
+      }
+    }
+
+    // Yield final chunk
+    yield {
+      delta: "",
+      accumulated,
+      done: true,
+      metadata: {
+        model: currentModel,
+        finishReason,
+        tokensUsed: totalTokens,
+        promptTokens,
+        completionTokens,
+      },
+      structured: structured ? { ...structured, toolCalls } : undefined,
+    };
+  }
+}