Actions and user interactions

Click, fill, type, press, check, select, upload and drag - and which options you reach for when a real user would not click the centre of a button.

The everyday actions

import { test, expect } from "@playwright/test";

test("a user can update their profile", async ({ page }) => {
  await page.goto("/settings");

  const name = page.getByLabel("Display name");
  await name.fill("Ada Lovelace");          // clears, then types

  await page.getByLabel("Bio").pressSequentially("Mathematician", { delay: 20 });

  await page.getByLabel("Email me about updates").check();
  await page.getByLabel("Weekly").check();

  await page.getByLabel("Language").selectOption("en-GB");
  await page.getByLabel("Language").selectOption({ label: "English (UK)" });

  await page.getByRole("button", { name: "Save" }).click();
  await expect(page.getByRole("status")).toHaveText("Profile saved");
});

Action	Does	Use instead when
`fill`	Clears and sets the value in one go	Always, for text inputs
`pressSequentially`	Types character by character	A handler needs per-keystroke events
`press`	Sends one key or chord	Enter, Escape, Tab, Control+A
`check` / `uncheck`	Sets checkbox state idempotently	Always, for checkboxes
`selectOption`	Chooses by value, label or index	For native selects only
`setInputFiles`	Attaches files	File inputs and drag-drop zones

💡

type() is the old name for pressSequentially() and is deprecated. Prefer fill() for ordinary input - it is faster and, unlike typing, it cannot be racing the page's own formatting logic.

Clicks that are not a plain click

const row = page.getByRole("row", { name: "Invoice 1042" });

await row.click();                                   // left, centre, once
await row.dblclick();
await row.click({ button: "right" });                // context menu
await row.click({ modifiers: ["Shift"] });           // range select
await row.click({ position: { x: 10, y: 10 } });     // a specific spot
await row.hover();
await row.click({ trial: true });                    // check actionability, do nothing

// select a range the way a user does
await page.getByRole("row", { name: "Invoice 1042" }).click();
await page.getByRole("row", { name: "Invoice 1047" }).click({ modifiers: ["Shift"] });

// hover to reveal a menu, then click the item
await page.getByRole("button", { name: "Actions" }).hover();
await page.getByRole("menuitem", { name: "Duplicate" }).click();

// drag and drop
await page.getByTestId("card-1").dragTo(page.getByTestId("column-done"));

click({ trial: true }) performs the actionability checks without clicking - the fastest way to find out why a click is failing.
force: true skips the checks and clicks anyway. It makes a test pass that a user could not perform, so treat it as a temporary diagnostic, not a fix.
dragTo does the move-hold-move-release sequence for you; drop in a slow, animated target can still need a manual hover() first.
Playwright scrolls the element into view automatically before acting, so an explicit scroll is rarely needed.

Files, keyboard and scrolling

// a single file, or several
await page.getByLabel("Contract").setInputFiles("fixtures/contract.pdf");
await page.getByLabel("Photos").setInputFiles([
  "fixtures/one.jpg",
  "fixtures/two.jpg",
]);

// an in-memory file, no fixture on disk
await page.getByLabel("Import").setInputFiles({
  name: "data.csv",
  mimeType: "text/csv",
  buffer: Buffer.from("id,name\n1,ada\n"),
});

// clear the selection
await page.getByLabel("Photos").setInputFiles([]);

// keyboard focus and chords
await page.getByRole("textbox").focus();
await page.keyboard.press("Control+A");
await page.keyboard.press("Backspace");
await page.keyboard.type("replaced");

// global shortcut
await page.keyboard.press("Control+Shift+P");

// programmatic scrolling, when you need a specific element in view
await page.getByText("Terms and conditions").scrollIntoViewIfNeeded();

Need	Method
Set a text input	`locator.fill()`
Send keys to the focused element	`page.keyboard.press()`
Attach a file	`locator.setInputFiles()`
Bring an element into view	`locator.scrollIntoViewIfNeeded()`
Move the mouse somewhere	`page.mouse.move(x, y)`
Select text range	`locator.selectText()`

Everything on the page object - keyboard, mouse, touchscreen - acts on the page as a whole and bypasses locator checks. That makes them powerful and easy to misuse: prefer a locator action whenever one exists, and reach for the page-level API only when the interaction genuinely is not element-based.

FAQ

Why does my click time out even though the element is visible?

Visibility is only one of Playwright's actionability checks. A click also requires the element to be stable (not animating), to receive events (nothing overlapping it) and to be enabled. Use click({ trial: true }) or the trace to see which check failed.

When is force: true acceptable?

Almost never in a committed test. It exists for situations where you deliberately test an element a user cannot reach, such as a hidden control exercised through keyboard navigation. If you find yourself adding it to get past a real overlay, fix the overlay interaction instead.

Auto-waiting, timeouts and flakiness Locators and assertions

Last refreshed 2026-09-18.

Actions and user interactions

The everyday actions

Clicks that are not a plain click

Files, keyboard and scrolling

FAQ

Related