Build reliable automations to extract web data

function transformString(input) {
  return input.replace(/ - Amendment #\d+/, '');
}

function transformEffectiveDate(input) {
  return input.replace('Effective Date: ', '');
}

function transformSupplierNumber(input) {
  return input.replace('Supplier Number: ', '');
}

const browser = await chromium.launch({headless: false});
// Create pages, interact with UI elements, assert values

const page = await browser.newPage();
await page.goto("https://demo-site-eta.vercel.app/lists/card");
await page.waitForTimeout(10000);

const listContainerLocator = page.locator(
  "#radix-\:r3\:-content-card > div");
const listItems = await listContainerLocator.locator("div").all();

const contractsList = [];

for (let i = 0; i < listItems.length; i++) {
  const itemLocator = listItems[i];
  const title = await itemLocator.locator("h2").textContent();
  const suppliers = await itemLocator.locator(
    "div:nth-child(1) > div:nth-child(2) > p:nth-child(1)"
  ).textContent();

  const effectiveDate = await itemLocator.locator(
    "div:nth-child(1) > div:nth-child(2) > p:nth-child(2)"
  ).textContent();

  const supplierNumber = await itemLocator.locator(
    "div:nth-child(1) > div:nth-child(2) > p:nth-child(3)"
  ).textContent();

  const status = await itemLocator.locator(
    "div:nth-child(1) > div:nth-child(2) > div"
  ).textContent();

  const link = await itemLocator.locator("a").getAttribute("href");

  contractsList.push({
    title: transformString(title),
    suppliers,
    effectiveDate: transformEffectiveDate(effectiveDate),
    supplierNumber: transformSupplierNumber(supplierNumber),
    status,
    link
  });

}

return contractsList;

await page.goto("https://demo-site-eta.vercel.app/lists/card");

const contractsList = await page.extractArrayOptimized({
  itemEntityName: "contract_info",
  label: "contract_info",
  itemEntitySchema: {
    type: "object",
    properties: {
      title: {
        type: "string", primary: true,
        description:
          "contract title, title should not include Amendment number",
      },
      supplier: {
        type: "string",
        description: "supplier of the contract",
      },
      state: {
        type: "string",
        description: "state of the contract",
      },
      effectiveDate: {
        type: "string",
        description: "effective date",
      },
    },
    required: ["title", "supplier", "state"],
  },
});

return contractsList;