📚 Datavist SDK – Quick‑Start Guide

You will need to create a Datavist account. It only takes a few seconds. Page extractions cost 1 cent per page. Deposit as little as $1 to get started and pay-as-you-go.

Datavist is an AI software-as-a-service that can reliably extract and monitor data from complex pages (modals, tables, etc.).

---

1️⃣ Getting Started

npm i @datavist/sdk
# or if you are using Yarn
yarn add @datavist/sdk

// src/index.js
import { DatavistClient } from "@datavist/sdk";

const client = new DatavistClient({
  apiKey: "YOUR_API_KEY",                 // <-- replace with yours
  baseUrl: "https://api.datavist.xyz/v1", // optional
  retry: true,                            // optional
});

---

2️⃣ Usage

There are two ways to use Datavist:

• Create/save a project that stores your settings and data in the cloud.
• Extract data in real-time--you are responsible for saving your own data and settings.

3️⃣ Create Project

3.1 Create Project With Prompt

import { DatavistClient } from "@datavist/sdk";

(async () => {
  const client = new DatavistClient({ apiKey: "YOUR_API_KEY_" });

  const proj = await client.createProjectWithPrompt({
    title: "Prompt Demo",
    urls: ["https://example.com/"],
    prompt: "describe the page.",
    frequency: "once",
    email: "you@example.com",                 // optional notifications
  });

  console.log("✅ Project created –", proj);

  proj.start() // you have to explicitly start project to begin the crawler
})()

3.2 Create Project With Schema

const schemaProj = await client.createProjectWithSchema({
  title: "Schema Demo",
  urls: ["https://example.com/products"],
  doc_type: "product",
  extraction_scope: "auto",
  properties: ['title', 'sku'],
  frequency: "daily",
});

console.log(schemaProj.id);

schemaProj.start()

Returned type: Project (flavour "schema"). Only the fields listed in UpdatableSchemaFields can be edited later (title,doc_type, properties, extraction_scope, frequency, email, webhook, max_pagination_pages, max_details_pages`).

4️⃣ Extract Data

Sometimes you just need a one‑off extraction and don’t want to store a project.

4.1 Extract Data With Prompt

const rawCsv = await client.extractDataWithPrompt({
  urls: ["https://example.com/"],
  prompt: "list all links on the page.",
});
console.log("One‑off CSV:", rawCsv);

4.2 Extract Data With Schema

const rawJson = await client.extractDataWithSchema({
  urls: ["https://example.com/products"],
  doc_type: "product",
  extraction_scope: "auto",
  properties: ['name', 'price'],
});
console.log("One‑off JSON‑L:", rawJson);

5️⃣ Usage

5.1 Get Project Dataset (CSV / JSON / JSONL)

import { writeFile } from "fs/promises";

// ---- CSV ---------------------------------------------------------
const csv = await proj.getDataset("csv");
await writeFile("./output.csv", csv);
console.log("✅ CSV written to ./output.csv");

// ---- JSON (array) ------------------------------------------------
const jsonArray = await proj.getDataset("json");
await writeFile("./output.json", JSON.stringify(jsonArray, null, 2));
console.log("✅ JSON array written to ./output.json");

// ---- Pagination example (only first 100 rows) -------------------
const first100 = await proj.getDataset("json", { limit: 100 });
console.log(`Fetched ${first100.length} rows`);

// ----‑JSONL ------------------------------------------------------
const jsonl = await proj.getDataset("jsonl");
await writeFile("./output.jsonl", jsonl);
console.log("✅ JSON‑L written to ./output.jsonl");

All three formats are supported by the same method (getDataset).
You can also pass offset / limit to page through large result sets.

5.2 Get Status

const status = await proj.getStatus();
console.log("🔄 Status:", status);

Possible values: "pending", "waiting", "running", "finished", "error", "disabled"

5.3 Get Row Count

const rowCount = await proj.getRowCount();
console.log("📊 Row count:", rowCount);

5.4 Modify/Update Project

proj.title = "Prompt Demo – Updated";
proj.frequency = "daily";          // allowed for prompt flavour
await proj.save();                 // only the dirty fields are PATCHed

console.log("✏️ Project updated – new title:", proj.title);

5.5 Delete Project

await proj.delete();
console.log("🗑️ Project deleted");

5.6 Get Project

const fetched = await client.getProject(promptProj.id);
console.log("Fetched project title:", fetched.title);

client.getProject returns the raw JSON (ProjectRaw).
If you need the high‑level wrapper, just do:

const proj = new Project(client, fetched, "prompt", ["https://example.com/"]);

5.7 Notifications

const proj = await client.createProjectWithPrompt({
  title: "Prompt Demo",
  urls: ["https://example.com/"],
  prompt: "describe the page.",
  frequency: "once",
  notifications: {email: "you@example.com", "webhook": "https://example.com/success"},                 // optional notifications
});

Alternatively...

const proj = await client.createProjectWithPrompt({
  title: "Prompt Demo",
  urls: ["https://example.com/"],
  prompt: "describe the page.",
  frequency: "once",
  email: "you@example.com",
  webhook: "https://example.com/success",
});

6️⃣ Project Object – Example

Project {
  dirty: Set(0) {},
  client: DatavistClient {
    apiKey: 'jl40924fb5cc4e3fb58a1b263b1325hk2',
    baseUrl: 'https://api.datavist.xyz/v1',
    retry: false
  },
  _data: {
    notifications: { email: 'you@example.com', webhook: '' },
    title: 'Prompt Demo',
    extraction_scope: 'auto',
    prompt: 'describe the page.',
    frequency: 'once',
    last_activity: null,
   : null,
    row_count: 1,
    status: 'finished',
    error_msg: '',
    max_details_pages: 0,
    max_pagination_pages: 0,
    total_project_revenue: 0.01,
    use_details_page: false,
    _id: '6929b7f98d95d700230fcc92',
    doc_type: 'custom',
    user: '01428235fc5ba20002342jkwjk2',
    createdAt: '2025-11-28T14:55:53.242Z',
    updatedAt: '2025-11-28T14:55:56.250Z',
    __v: 0,
    urls: 'https://example.com/',
    properties: { custom: [Array] }
  },
  _urls: 'https://example.com' ],
  whitelist: Set(6) { 'title', 'prompt', 'frequency', 'email', 'webhook', 'urls' }
}