XPath Master Cheat Sheet:
Functions, Axes, Operators, Examples

1. Anatomy of an XPath

//tag[@attribute='value'][position()]/axis::tag
│  │   │                  │          │
│  │   │                  │          │── walk to related nodes (parent::, following-sibling::)
│  │   │                  │── pick one from the matched set ([1], [last()])
│  │   │── predicate: the condition in [ ] (@id='x', text()='y', contains(...))
│  │── the element tag, or * for any
│── // = search at ANY depth   / = direct child only

• Absolute paths start at the root: /html/body/div[2]/form/input. One DOM change breaks them. Never ship one.
• Relative paths start with // and anchor on something stable: //input[@name='email']. This is the default.

2. Quick reference: the basics

3. Text matching

4. All the functions you will actually use

5. All 13 axes, with the family tree

Axes are XPath's superpower over CSS: from any node you can walk in every direction. Read them as axis::node, "from here, go that way".

6. Operators

7. Indexing, and the gotcha that breaks everyone once

// XPath counts from 1, not 0
//table//tr[1]                 // first row
//table//tr[last()]            // last row
//table//tr[last()-1]          // second to last
//table//tr[position()<=3]     // first three

// THE GOTCHA: [n] binds to the nearest step, not the whole expression
//div/a[2]      // the 2nd link INSIDE EACH div (can match many!)
(//div/a)[2]    // the 2nd link of ALL matched links (exactly one)

8. Dynamic element patterns

9. Table strategies, end to end

Same demo table as the CSS sheet, so you can compare the two languages line by line:

// The row containing Anita (cell text anchors the row)
//tr[td[normalize-space()='Anita']]

// Anita's STATUS (3rd cell of that row)
//tr[td[normalize-space()='Anita']]/td[3]

// Click Delete on Anita's row
//tr[td[normalize-space()='Anita']]//button

// Same, but climbing UP from the anchor instead (ancestor axis)
//td[normalize-space()='Anita']/ancestor::tr//button

// Column by HEADER NAME, not position: Status cell of Anita's row
//tr[td[.='Anita']]/td[count(//th[normalize-space()='Status']/preceding-sibling::th)+1]

// Second Delivered row (parentheses first, then index)
(//tr[td[normalize-space()='Delivered']])[2]

// All data rows, skipping the header row
//tr[not(th)]

10. Using XPath in Selenium and Playwright

# Selenium (Python)
driver.find_element(By.XPATH, "//tr[td[normalize-space()='Anita']]//button").click()

// Selenium (Java)
driver.findElement(By.xpath("//label[text()='Email']/following-sibling::input")).sendKeys("[email protected]");

// Playwright (TypeScript) - xpath= prefix, or // is auto-detected
await page.locator("xpath=//tr[td[normalize-space()='Anita']]/td[3]").innerText();
await page.locator("//h2[text()='Orders']/following::table[1]").screenshot();

// DevTools console - test any XPath live
$x("//tr[td[normalize-space()='Anita']]")

11. XPath vs CSS: when to use which

12. FAQ

What is the difference between / and // ?

/ steps to a direct child (from the root when leading); // searches any depth. //input finds every input on the page; /html/body/input only a direct child of body.

Does XPath start at 0 or 1?

1. //tr[1] is the first row. And remember the binding gotcha: (//a)[2] vs //div/a[2] are different questions.

How do I select by text?

//tag[normalize-space()='Exact'] for exact, //tag[contains(normalize-space(.),'part')] for partial. Avoid bare text()= on real apps: whitespace and nested spans break it.

Which axes should I learn first?

ancestor:: and following-sibling::. They solve the two everyday problems: climb from a unique cell to its row, and hop from a label to its input.

Drill these on real markup in the tables practice page and the Playwright practice library, or let SnapLocator generate and heal locators for you.