gRPC Client

Teckel Editor uses ConnectRPC for gRPC-Web communication with the Teckel Engine backend.

Proto definition

The gRPC service is defined in the proto file at:

../teckel-api/proto/teckel.proto

This defines the teckel.v1.TeckelService with all available RPC methods.

Code generation

Generated TypeScript files are produced by Buf:

npx buf generate

This reads the proto definition and generates:

• src/lib/api/gen/teckel_pb.ts — Protobuf message types
• src/lib/api/gen/teckel_connect.ts — ConnectRPC service client

Client wrapper

The generated client is wrapped in src/lib/api/teckel-client.ts for a cleaner interface. The createTeckelClient(baseUrl) factory function returns an object with all available methods:

const client = createTeckelClient('http://localhost:50051');

Available methods

health()

Health check. Returns {status, version}.

const health = await client.health();
// { status: "serving", version: "1.0.0" }

validate(yaml, variables?)

Synchronous pipeline validation. Returns {valid, error?}.

const result = await client.validate(yamlString, { INPUT_PATH: 's3://bucket' });
// { valid: true } or { valid: false, error: "..." }

explain(yaml, variables?)

Generate an execution plan. Returns {plan}.

const result = await client.explain(yamlString, variables);
// { plan: "Step 1: Read parquet from s3://..." }

submitJob(yaml, variables?)

Submit a pipeline for asynchronous execution. Returns {job_id, status}.

const result = await client.submitJob(yamlString, variables);
// { job_id: "abc123", status: "queued" }

getJob(jobId)

Poll the status of a submitted job. Returns the full JobResponse.

const job = await client.getJob('abc123');
// { id, status, error?, created_at, started_at?, completed_at?, duration_ms? }

waitJob(jobId, timeout?, signal?)

Long-poll for job completion. Blocks until the job finishes or the timeout expires.

const job = await client.waitJob('abc123', 30);

cancelJob(jobId)

Cancel a running job.

await client.cancelJob('abc123');

listJobs()

List all submitted jobs.

const { jobs } = await client.listJobs();

inspectSource(format, path, options?)

Inspect a data source to discover its schema (field names, types, row count).

const schema = await client.inspectSource('parquet', 's3://bucket/data/');
// { fields: [{ name: "id", data_type: "Int64", nullable: false }, ...], row_count: 1000000 }

Transport

The client uses createGrpcWebTransport from @connectrpc/connect-web:

import { createPromiseClient } from '@connectrpc/connect';
import { createGrpcWebTransport } from '@connectrpc/connect-web';
import { TeckelService } from './gen/teckel_connect';

const transport = createGrpcWebTransport({ baseUrl });
const client = createPromiseClient(TeckelService, transport);

This enables communication with the gRPC server from the browser using the gRPC-Web protocol.

Type exports

The client module exports several TypeScript types:

Type	Description
TeckelClient	Return type of createTeckelClient
JobStatus	'queued' | 'running' | 'completed' | 'failed' | 'cancelled'
ValidateResponse	{ valid: boolean; error?: string }
ExplainResponse	{ plan: string }
SubmitJobResponse	{ job_id: string; status: string }
JobResponse	Full job state with timestamps
HealthResponse	{ status: string; version: string }
FieldInfo	{ name: string; data_type: string; nullable: boolean }
InspectSourceResponse	{ fields: FieldInfo[]; row_count: number }