XPATH expressions - the good, the bad and the ugly

A good XPATH expression, as per Chat GPT, should be

1. Specific

2. Simple

3. Reusable

4. Efficient

5. Readable

Awesome!

Now, what does this mean?

“You need to find this by yourself, Alex, this is as far as I can take you!”

---

Joking aside,

For an XPATH expression to have these attributes, it should use

• specific tags (no *)

• specific attributes that are created for test automation purposes (no class, type, etc)

• unique and complete attribute values

---

It should not use

• indexes

• XPATH axes (sibling, child, parent, etc)

• XPATH functions (contains, starts-with, etc)

• logical operators

• text values

---

It should be short with

• 1 search, maximum 2 acceptable

• one attribute ideally

---

It should work in similar contexts.

For example:

1. XPATH expression for the 1st result in a results page should work for 1st, 2nd and 5th results pages

2. XPATH expression for search textbox should work if the site is used in English but also in French

---

Some examples follow.

The good

//button[@testid = ’username’]

//div[@testid = 'header']//input[@testid='email'] 

//div[@testid = 'header']/div[@testid='menu'] 

The bad

//button[contains(@testid,'username')]  

//input[starts-with(@testid,'username')] 

//span[text()='UserName'] 

//button[@testid = ’signup-button’]/a

//span[@class = 'form']

//input[@value="email" and contains(@class,’web’)] 

(//input[contains(@testid,'result')])[2] 

The ugly

//form[contains(@class,'form')]//following::div//input[@id='userpassword'] 

//form[contains(@class,'form')]//following::div//button/../input[@id='userpassword'] 

(//div[contains(@class,'custom__border')]//following-sibling::div)[1]

//button[@data-testid='signup-button']//preceding::input[@type='password'] 

//div[contains(@class,'custom__border')]//preceding-sibling::div

//div[@aria-labelledby='sign_up_with_google_label']//child::span

//button[@data-testid='signup-button']//parent::div//preceding-sibling::div//input[@type='password']

//div[contains(@class,'overflow-hidden')]//descendant::span

//input[@type='password']//ancestor::div//input[@id='email']

/html//div/div/div/div/div/div/div/img 

---

What should you do if the HTML of the site does not allow creating good XPATH expressions?

First, ask the dev team to improve the HTML code by adding id attributes with static values to all important page elements.

If this is not possible, annotate each test with @Flaky so it is clear how the test will be.

Thanks for reading.

PS:

If you are interested in a full Selenium project built from A - Z, please find it here.