OriginTS

Why OriginTS?

Traditional data extraction pipelines silently coerce types, swallow errors, and make it impossible to trace how a value was derived. When something goes wrong, you’re left guessing.

OriginTS treats data extraction like compilation: build an immutable plan, execute it with full lineage tracking, and get structured failures instead of exceptions.

Two-Phase Architecture

Separate planning from execution. Build immutable plans with no I/O, then run them against actual data with full provenance.

Learn more →

Extraction System

A unified ExtractSpec type works across all formats — JSON, XLSX, CSV, YAML, HTML, Markdown, TOML. One API to learn, any format to extract.

Learn more →

Full Provenance

Every transformation step is recorded. Trace exactly how any output was derived — even when execution fails.

Learn more →

Explicit Failures

Seven structured failure kinds replace silent coercions and thrown exceptions. Fail fast, fail clearly.

Learn more →

Quick Example

import { Planner, load, run } from '@origints/core'

const plan = new Planner()
  .in(load({ name: 'Alice', age: 30 }))
  .emit((out, $) => out
    .add('name', $.get('name').string())
    .add('age', $.get('age').number())
  )
  .compile()

const result = await run(plan)

if (result.ok) {
  console.log(result.value)
  // { name: 'Alice', age: 30 }
}

Format Packages

OriginTS supports extraction from multiple data formats, each as a separate package:

CSV

RFC 4180 parsing, header detection, column-by-name access, predicate-based filtering.

@origints/csv

XLSX

Workbook navigation, cell predicates, eachSlice iteration, header-relative column lookup.

@origints/xlsx

YAML

Single and multi-document parsing, anchor/alias preservation, full source tracking.

@origints/yaml

HTML

CSS selector queries, attribute extraction, Markdown conversion.

@origints/html

Plus Markdown, TOML, and Mammoth (DOCX).