XPath

XPath (XML Path Language) describes a path through a document tree to the nodes you want. In test automation it is one of the ways to write a locator, and its main advantage over CSS is that it can match on text content and walk in any direction: up to a parent, across to a sibling, or down to a descendant. Browsers implement XPath 1.0, so functions from later versions, such as matches() or lower-case(), are not available.

Example

//button[normalize-space()='Sign in']
//input[@name='email']
//label[text()='Password']/following-sibling::input
//tr[td[normalize-space()='Order 1042']]//button[@aria-label='Cancel']
//a[contains(@href, '/invoices/')]

The fourth line shows why people reach for XPath: find the table row that contains a given cell, then the button inside that row.

WebElement cancel = driver.findElement(
    By.xpath("//tr[td[normalize-space()='Order 1042']]//button[@aria-label='Cancel']"));

Absolute versus relative

An absolute XPath starts at the root (/html/body/div[2]/form/button) and breaks as soon as any wrapper changes. A relative XPath starts with // and anchors on something meaningful. Copying the “full XPath” from DevTools gives you the absolute form, which is almost never what you want in a test.

Common mistakes

• Index-based paths such as (//div)[7]. They encode layout, not meaning.
• text() on elements with nested markup. //button[text()='Save'] fails if the label sits in a child <span>. normalize-space() (or .) uses the full string value instead.
• Matching a class with @class='btn'. That only matches when btn is the entire attribute. Use contains(concat(' ', normalize-space(@class), ' '), ' btn '), or switch to CSS .btn.
• Defaulting to XPath out of habit. In Playwright, getByRole and getByText cover most text-based cases and read better.