Take screenshot of full page with Selenium Python

Question

After trying out various approaches... I have stumbled upon this page to take full-page screenshot with chromedriver, selenium and python.

The original code is here. (and I copy the code in this posting below)

It uses PIL and it works great! However, there is one issue... which is it captures fixed headers and repeats for the whole page and also misses some parts of the page during page change. sample url to take a screenshot:

http://www.w3schools.com/js/default.asp

How to avoid the repeated headers with this code... Or is there any better option which uses python only... ( i don't know java and do not want to use java).

Please see the screenshot of the current result and sample code below.

test.py

"""
This script uses a simplified version of the one here:
https://snipt.net/restrada/python-selenium-workaround-for-full-page-screenshot-using-chromedriver-2x/

It contains the *crucial* correction added in the comments by Jason Coutu.
"""

import sys

from selenium import webdriver
import unittest

import util

class Test(unittest.TestCase):
    """ Demonstration: Get Chrome to generate fullscreen screenshot """

    def setUp(self):
        self.driver = webdriver.Chrome()

    def tearDown(self):
        self.driver.quit()

    def test_fullpage_screenshot(self):
        ''' Generate document-height screenshot '''
        #url = "http://effbot.org/imagingbook/introduction.htm"
        url = "http://www.w3schools.com/js/default.asp"
        self.driver.get(url)
        util.fullpage_screenshot(self.driver, "test.png")


if __name__ == "__main__":
    unittest.main(argv=[sys.argv[0]])

util.py

import os
import time

from PIL import Image

def fullpage_screenshot(driver, file):

        print("Starting chrome full page screenshot workaround ...")

        total_width = driver.execute_script("return document.body.offsetWidth")
        total_height = driver.execute_script("return document.body.parentNode.scrollHeight")
        viewport_width = driver.execute_script("return document.body.clientWidth")
        viewport_height = driver.execute_script("return window.innerHeight")
        print("Total: ({0}, {1}), Viewport: ({2},{3})".format(total_width, total_height,viewport_width,viewport_height))
        rectangles = []

        i = 0
        while i < total_height:
            ii = 0
            top_height = i + viewport_height

            if top_height > total_height:
                top_height = total_height

            while ii < total_width:
                top_width = ii + viewport_width

                if top_width > total_width:
                    top_width = total_width

                print("Appending rectangle ({0},{1},{2},{3})".format(ii, i, top_width, top_height))
                rectangles.append((ii, i, top_width,top_height))

                ii = ii + viewport_width

            i = i + viewport_height

        stitched_image = Image.new('RGB', (total_width, total_height))
        previous = None
        part = 0

        for rectangle in rectangles:
            if not previous is None:
                driver.execute_script("window.scrollTo({0}, {1})".format(rectangle[0], rectangle[1]))
                print("Scrolled To ({0},{1})".format(rectangle[0], rectangle[1]))
                time.sleep(0.2)

            file_name = "part_{0}.png".format(part)
            print("Capturing {0} ...".format(file_name))

            driver.get_screenshot_as_file(file_name)
            screenshot = Image.open(file_name)

            if rectangle[1] + viewport_height > total_height:
                offset = (rectangle[0], total_height - viewport_height)
            else:
                offset = (rectangle[0], rectangle[1])

            print("Adding to stitched image with offset ({0}, {1})".format(offset[0],offset[1]))
            stitched_image.paste(screenshot, offset)

            del screenshot
            os.remove(file_name)
            part = part + 1
            previous = rectangle

        stitched_image.save(file)
        print("Finishing chrome full page screenshot workaround...")
        return True

Answer 1

I have modified the answer given by @ihightower, instead of saving the screenshot in that function, return the total height and total width of the webpage and then set the window size to total height and total width.

from PIL import Image
from io import BytesIO

from selenium import webdriver
from selenium.webdriver.chrome.options import Options

def open_url(url):
    options = Options()

    options.headless = True

    driver = webdriver.Chrome(chrome_options=options)

    driver.maximize_window()
    driver.get(url)
    save_screenshot(driver, 'screen.png')

def save_screenshot(driver, file_name):
    height, width = scroll_down(driver)
    driver.set_window_size(width, height)
    img_binary = driver.get_screenshot_as_png()
    img = Image.open(BytesIO(img_binary))
    img.save(file_name)
    # print(file_name)
    print(" screenshot saved ")


def scroll_down(driver):
    total_width = driver.execute_script("return document.body.offsetWidth")
    total_height = driver.execute_script("return document.body.parentNode.scrollHeight")
    viewport_width = driver.execute_script("return document.body.clientWidth")
    viewport_height = driver.execute_script("return window.innerHeight")

    rectangles = []

    i = 0
    while i < total_height:
        ii = 0
        top_height = i + viewport_height

        if top_height > total_height:
            top_height = total_height

        while ii < total_width:
            top_width = ii + viewport_width

            if top_width > total_width:
                top_width = total_width

            rectangles.append((ii, i, top_width, top_height))

            ii = ii + viewport_width

        i = i + viewport_height

    previous = None
    part = 0

    for rectangle in rectangles:
        if not previous is None:
            driver.execute_script("window.scrollTo({0}, {1})".format(rectangle[0], rectangle[1]))
            time.sleep(0.5)
        # time.sleep(0.2)

        if rectangle[1] + viewport_height > total_height:
            offset = (rectangle[0], total_height - viewport_height)
        else:
            offset = (rectangle[0], rectangle[1])

        previous = rectangle

    return (total_height, total_width)

open_url("https://www.medium.com")

Answer 2

This answer improves upon prior answers by am05mhz and Javed Karim.

It assumes headless mode, and that a window-size option was not initially set. Before calling this function, ensure the page has loaded fully or sufficiently.

It attempts to set the width and height both to what is necessary. The screenshot of the entire page can sometimes include a needless vertical scrollbar. One way to generally avoid the scrollbar is by taking a screenshot of the body element instead. After saving a screenshot, it reverts the size to what it was originally, failing which the size for the next screenshot may not set correctly.

Ultimately this technique may still not work perfectly well for some examples.

def save_screenshot(driver: webdriver.Chrome, path: str = '/tmp/screenshot.png') -> None:
    # Ref: https://stackoverflow.com/a/52572919/
    original_size = driver.get_window_size()
    required_width = driver.execute_script('return document.body.parentNode.scrollWidth')
    required_height = driver.execute_script('return document.body.parentNode.scrollHeight')
    driver.set_window_size(required_width, required_height)
    # driver.save_screenshot(path)  # has scrollbar
    driver.find_element_by_tag_name('body').screenshot(path)  # avoids scrollbar
    driver.set_window_size(original_size['width'], original_size['height'])

If using Python older than 3.6, remove the type annotations from the function definition.

Answer 3

After knowing the approach of @Moshisho.

My full standalone working script is... (added sleep 0.2 after each scroll and position)

import sys
from selenium import webdriver
import util
import os
import time
from PIL import Image

def fullpage_screenshot(driver, file):

        print("Starting chrome full page screenshot workaround ...")

        total_width = driver.execute_script("return document.body.offsetWidth")
        total_height = driver.execute_script("return document.body.parentNode.scrollHeight")
        viewport_width = driver.execute_script("return document.body.clientWidth")
        viewport_height = driver.execute_script("return window.innerHeight")
        print("Total: ({0}, {1}), Viewport: ({2},{3})".format(total_width, total_height,viewport_width,viewport_height))
        rectangles = []

        i = 0
        while i < total_height:
            ii = 0
            top_height = i + viewport_height

            if top_height > total_height:
                top_height = total_height

            while ii < total_width:
                top_width = ii + viewport_width

                if top_width > total_width:
                    top_width = total_width

                print("Appending rectangle ({0},{1},{2},{3})".format(ii, i, top_width, top_height))
                rectangles.append((ii, i, top_width,top_height))

                ii = ii + viewport_width

            i = i + viewport_height

        stitched_image = Image.new('RGB', (total_width, total_height))
        previous = None
        part = 0

        for rectangle in rectangles:
            if not previous is None:
                driver.execute_script("window.scrollTo({0}, {1})".format(rectangle[0], rectangle[1]))
                time.sleep(0.2)
                driver.execute_script("document.getElementById('topnav').setAttribute('style', 'position: absolute; top: 0px;');")
                time.sleep(0.2)
                print("Scrolled To ({0},{1})".format(rectangle[0], rectangle[1]))
                time.sleep(0.2)

            file_name = "part_{0}.png".format(part)
            print("Capturing {0} ...".format(file_name))

            driver.get_screenshot_as_file(file_name)
            screenshot = Image.open(file_name)

            if rectangle[1] + viewport_height > total_height:
                offset = (rectangle[0], total_height - viewport_height)
            else:
                offset = (rectangle[0], rectangle[1])

            print("Adding to stitched image with offset ({0}, {1})".format(offset[0],offset[1]))
            stitched_image.paste(screenshot, offset)

            del screenshot
            os.remove(file_name)
            part = part + 1
            previous = rectangle

        stitched_image.save(file)
        print("Finishing chrome full page screenshot workaround...")
        return True


driver = webdriver.Chrome()

''' Generate document-height screenshot '''
url = "http://effbot.org/imagingbook/introduction.htm"
url = "http://www.w3schools.com/js/default.asp"
driver.get(url)
fullpage_screenshot(driver, "test1236.png")

Answer 4

Not sure if people are still having this issue. I've done a small hack that works pretty well and that plays nicely with dynamic zones. Hope it helps

# 1. get dimensions
browser = webdriver.Chrome(chrome_options=options)
browser.set_window_size(default_width, default_height)
browser.get(url)
time.sleep(sometime)
total_height = browser.execute_script("return document.body.parentNode.scrollHeight")
browser.quit()

# 2. get screenshot
browser = webdriver.Chrome(chrome_options=options)
browser.set_window_size(default_width, total_height)
browser.get(url)  
browser.save_screenshot(screenshot_path)

Answer 5

element=driver.find_element_by_tag_name('body')
element_png = element.screenshot_as_png
with open("test2.png", "wb") as file:
    file.write(element_png)

There was an error in the code suggested earlier in line 2. Here is the corrected one. Being a noob here, not able to edit my own post as yet.

Sometimes the baove doesn't get best results. So can use another method to get height of all elements and sum them to set the capture height as below:

element=driver.find_elements_by_xpath("/html/child::*/child::*")
    eheight=set()
    for e in element:
        eheight.add(round(e.size["height"]))
    print (eheight)
    total_height = sum(eheight)
    driver.execute_script("document.getElementsByTagName('html')[0].setAttribute('style', 'height:"+str(total_height)+"px')")
    element=driver.find_element_by_tag_name('body')
    element_png = element.screenshot_as_png
    with open(fname, "wb") as file:
        file.write(element_png)

BTW, it works on FF.

Answer 6

You can achieve this by changing the CSS of the header before the screenshot:

topnav = driver.find_element_by_id("topnav")
driver.execute_script("arguments[0].setAttribute('style', 'position: absolute; top: 0px;')", topnav) 

EDIT: Put this line after your window scroll:

driver.execute_script("document.getElementById('topnav').setAttribute('style', 'position: absolute; top: 0px;');")

So in your util.py it will be:

driver.execute_script("window.scrollTo({0}, {1})".format(rectangle[0], rectangle[1]))
driver.execute_script("document.getElementById('topnav').setAttribute('style', 'position: absolute; top: 0px;');")

If the site is using the header tag, you can do it with find_element_by_tag_name("header")