@knorby/inkeedecoder-client

A universal TypeScript client for scraping data from inkeedecoder.com — products, ingredients, search (simple + advanced), brands, and ingredient functions.

• Universal runtime. No Node.js-only APIs. HTTP runs through an injectable fetch (native in Node 20+, React Native, and browsers). HTML parsing uses a small cheerio-style adapter over htmlparser2
  • css-select — the same parse stack cheerio's slim build wraps, minus cheerio's heavy transitive tree (parse5, undici) — so the library bundles cleanly under React Native / Metro.
• Parse/transport split. Every parse* function is a pure (html) => data export, so you can pair it with any prefetch/cache layer. The createInkeedecoderClient() convenience wrapper fetches + parses for you.
• Options everywhere. Every object-returning method takes granular flags so lean output is the default and verbose sections (functions, skim table, CosIng data, long-form descriptions, …) are opt-in.
• Next-link pagination. The authoritative "Next page >>" URL is read from the HTML rather than reconstructed, so the library is robust across inkeedecoder.com's differing param schemes (ppage, uoffset, offset).

Not affiliated with inkeedecoder.com. This is an unofficial scraper. The site's robots.txt permits crawling /products, /ingredients, /search, /brands, and /ingredient-functions (it disallows only /auth/ and /products/recommend/). Be polite: keep request volume low, use the client's requestIntervalMs for auto-pagination, don't republish the site's prose, and prefer the lean option flags when you don't need verbose data.

Install

npm install @knorby/inkeedecoder-client

React Native

The library is Metro-safe out of the box (no undici/parse5 in the bundle). fetch is injected, so in environments without a global fetch pass one:

import { createInkeedecoderClient } from "@knorby/inkeedecoder-client";

const client = createInkeedecoderClient({
  // fetch: myFetch,            // optional; defaults to globalThis.fetch
  requestIntervalMs: 1000,      // politeness delay for auto-pagination
  maxPages: 25,                 // safety cap per allPages operation
});

Quick start

const client = createInkeedecoderClient();

// Lean product (defaults): name, brand, description, ingredient set, original
// image, hashtags.
const product = await client.getProduct("the-ordinary-retinol-1-in-squalane");
console.log(product.name, product.brand?.name, product.ingredients.length);

// Verbose product: opt into each section independently.
const full = await client.getProduct("the-ordinary-retinol-1-in-squalane", {
  images: "all",
  functions: true,
  skimTable: true,
  dates: true,
  tooltips: true,
  longDescriptions: true,
});

// Ingredient page.
const squalane = await client.getIngredient("squalane", {
  cosing: true,
  details: true,
  proof: true,
});

// Search (both products + ingredients tabs; each paginates independently).
const results = await client.search("the ordinary");
console.log(results.products.items.length, results.products.hasMore);

// ...or fetch every page of both tabs:
const all = await client.search("the ordinary", { allPages: true });

// ...or fetch a single tab — exactly one request at any page (the both-tabs
// form costs two requests from page 2 on):
const productsOnly = await client.search("the ordinary", {
  tab: "products",
  page: 3,
});

// Advanced product search (include/exclude use the site's display names —
// see the warning under "Methods").
const hits = await client.searchProducts({
  query: "the ordinary",
  include: ["Squalane"],
  exclude: ["Simple Alcohols"],
  includeMode: "all", // 'all' = ALL OF (default); 'any' = ANY OF
});

API

createInkeedecoderClient(options)

| Option | Default | Description | | ------------------ | -------------------------------- | ------------------------------------------------------ | | baseUrl | https://inkeedecoder.com | Base URL. | | fetch | globalThis.fetch | Custom fetch (cached/proxied/RN). | | headers | { User-Agent: … } | Extra request headers. | | requestIntervalMs | 1000 | Delay between requests during auto-pagination. | | maxPages | 25 | Safety cap on pages fetched per allPages operation. |

Methods

Flag semantics: identity fields (slug/name/path/url) are always returned; every other section is included only when its flag is on — a flag turned off omits the key entirely (no empty placeholders).

• search(query, opts?) → both products + ingredients tabs. Page 1 is one combined request; { page: 2 }+ fetches each tab separately (activetab + ppage); { allPages } follows each tab's next links from page 1. { tab: "products" | "ingredients" } selects a single tab: exactly one request at any page — half the cost of the both-tabs form at page ≥ 2 — and allPages walks only that tab. A tabbed call returns { query, tab, results } instead of { query, products, ingredients }.
• searchProducts(filters, opts?) → advanced search (include/exclude); { page } (uses ppage) and { allPages }.
• getProduct(id, opts?) → id is a slug, path, or URL. Flags (defaults in bold): brand true, description true, ingredients true, images "original" ("none" | "original" | "all"), hashtags true, fullName false, functions false, skimTable false, dates false, tooltips false, longDescriptions false.
• getIngredient(id, opts?) → ourTake is always present (null when the page has no rating). Flags: functions true, safety (irritancy + comedogenicity) true, alsoCalled true, image true, cosing false, details false, proof false.
• getIngredientProducts(id, opts?) → paginated "other products" list (uoffset); { includeKnownAmount: true } adds the known-amount carousel.
• getBrand(id, opts?) → brand product list (offset); { page } / { allPages }.
• getIngredientFunction(id, opts?) → ingredient-function ingredient list (offset); { page } / { allPages }.
• fetchHtml(target) → raw HTML for any path/URL.

Every list item is a { name, slug, path, url } ref — the slug is the stable key for the entity on inkeedecoder.com.

⚠️ Easy trap: searchProducts filters use the site's display names, not INCI names. include/exclude must match the ingredient names inkeedecoder.com itself uses in its advanced-search UI — e.g. "Squalane" or group entries like "Simple Alcohols", which have no INCI equivalent. A formal INCI name can silently match nothing. The name on any ingredient ref returned by this client is a reliable source of valid values (as is the ingredient's page title on the site).

Limitations

• No UPC/barcode lookup. inkeedecoder.com does not publish UPC/EAN/barcode identifiers anywhere (verified across product pages and search), so there is nothing to scrape. Match products by slug or name instead.

Pure parsers (no fetch)

For custom transport layers (RN prefetch, edge caches, raw HTML you already have), import the parse* functions directly:

import { parseProductPage, parseIngredientPage, parseSearchPage } from "@knorby/inkeedecoder-client";

const product = parseProductPage(htmlString, "https://inkeedecoder.com");

Also exported: parseIngredientProductsPage, parseSearchProductsPage, parseBrandPage, parseIngredientFunctionPage, parseHtml, findNextLink, makePaginated, paginateAll, and the text/URL utilities.

Testing

| Command | What it does | | --------------------- | -------------------------------------------------------------------- | | npm test | Offline fixture-based unit + client tests (no network). | | npm run test:live | Hits the live site (INKEEDECODER_LIVE=1) — a markup-change tripwire. | | npm run test:coverage | Coverage report. |

Fixtures live in tests/fixtures/ and are (re)captured with node scripts/capture-fixtures.mjs. The CI workflow runs only the offline suite.

Development

| Command | What it does | | -------------------- | --------------------------------------------------------------------- | | npm run build | Build (tsup — dual ESM/CJS + .d.ts/.d.cts). | | npm run lint | Lint with Biome. | | npm run check | Lint + format in one pass (writes changes). | | npm run typecheck | Type-check src/ + tests/ with tsc (no emit). | | npx changeset | Create a changeset (required for any change that affects published output). |

See AGENTS.md for AI-agent steering and CONTRIBUTING.md for the full development workflow.

License

Apache-2.0 — Copyright 2026 knorby.