Implementation:Ucbepic Docetl Generate Pipeline Config

Knowledge Sources	DocETL
Domains	Configuration, Portability
Last Updated	2026-02-08 01:40 GMT

Overview

Concrete API route and utility function for generating YAML pipeline configurations from the DocETL playground state.

Description

The POST /api/getPipelineConfig Next.js API route calls generatePipelineConfig() to convert UI state (operations, datasets, model settings) into a YAML configuration string. The function handles operation schema conversion, sample size injection, system prompt configuration, optimizer settings, and intermediate directory setup.

Usage

Called automatically when the user clicks Export/Download in the playground. Can also be called programmatically via the API to generate YAML from UI state.

Code Reference

Source Location

• Repository: docetl
• File: website/src/app/api/getPipelineConfig/route.ts (L4-62), website/src/app/api/utils.ts (L107+)

Signature

// API Route
export async function POST(request: Request): Promise<NextResponse>
// Returns: { pipelineConfig: string }

// Config generator
export function generatePipelineConfig(
  namespace: string,
  default_model: string,
  data: { path: string },
  operations: Operation[],
  operation_id: string,
  name: string,
  homeDir: string,
  sample_size: number | null,
  optimize?: boolean,
  clear_intermediate?: boolean,
  system_prompt?: { datasetDescription: string | null; persona: string | null } | null,
  apiKeys?: APIKey[],
  docetl_encryption_key?: string,
  enable_observability?: boolean,
  optimizerModel?: string,
  extraPipelineSettings?: Record<string, unknown> | null,
): { yamlString: string }

Import

import { generatePipelineConfig } from "@/app/api/utils";

I/O Contract

Inputs

Name	Type	Required	Description
default_model	string	Yes	LLM model for operations
data	{path: string}	Yes	Dataset file path
operations	Operation[]	Yes	UI operation objects
name	string	Yes	Pipeline name
namespace	string	Yes	File isolation namespace
sample_size	number or null	No	Optional dataset sampling

Outputs

Name	Type	Description
pipelineConfig	string	Generated YAML configuration

Usage Examples

// Frontend API call
const response = await fetch("/api/getPipelineConfig", {
  method: "POST",
  body: JSON.stringify({
    default_model: "gpt-4o",
    data: { path: "/path/to/data.json" },
    operations: [...operationsFromUI],
    name: "my_pipeline",
    namespace: "project1",
  }),
});
const { pipelineConfig } = await response.json();
// pipelineConfig is a YAML string ready for docetl run

Related Pages

Implements Principle

• Principle:Ucbepic_Docetl_Pipeline_Export

Requires Environment

• Environment:Ucbepic_Docetl_Frontend_Node_Environment