DEV Community

Cover image for Scrape Multiple Google Answer Box Layouts with Python
Dmitriy Zub ☀️
Dmitriy Zub ☀️

Posted on • Updated on • Originally published at serpapi.com

Scrape Multiple Google Answer Box Layouts with Python

#python #tutorial #webdev #webscraping

This blog post covers how to scrape: weather, stock, converter, calculator, phone number, address, population, translation, dictionary, matches, flights, formulas, and multiple organic layout answer box results using Python.

Note: Not every possible layout is covered here. Some of them may just not seen by me.

Contents:

Intro

You'll see usage of if/elif/else, try/except, list comprehension, usage of zip() function. Simple stuff, but if you're a bit new to this, be sure to see what some of the things are doing since there won't be any proper explanation. An alternative API solution will be shown.

Prerequisites

$ pip install requests
$ pip install lxml 
$ pip install beautifulsoup4
$ pip install google-search-results
Enter fullscreen mode Exit fullscreen mode

Make sure you have a basic knowledge of libraries mentioned above (except google-search-results), since this blog post is not exactly a tutorial for beginners, so be sure you have a basic familiarity with them. I'll try my best to show in code that it's not that difficult.

Also, make sure you have a basic understanding of CSS selectors because of select()/select_one() beautifulsoup methods which accepts CSS selectors. CSS selectors reference.

Imports

import requests, lxml
from bs4 import BeautifulSoup
from serpapi import GoogleSearch # API soulution
Enter fullscreen mode Exit fullscreen mode

Scrape Google Phone Number Answer Box

There're several HTML layouts for phone answer box I stumble upon. All of this layouts will be scraped in the code below.

image

image

image

image

import requests, lxml
from bs4 import BeautifulSoup

headers = {
  "User-agent":
  "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582"
}

test_queries = [
  'skyscanner customer service phone number',
  'ryanair phone number uk',
  'wizz air phone number',
  'apple online customer service phone number',
  'amazon phone number',
  'walmart online phone number',
  'imb bank online phone number',
  'target online phone number',
  'yelp online phone number'
]

def get_phone_answer_box():

  for query in test_queries:
    params = {
      "q": query,
      "hl": "en",
      "gl": "us"
    }

    html = requests.get('https://www.google.com/search', headers=headers, params=params)
    soup = BeautifulSoup(html.text, 'lxml')

    # several CSS selectors to cover multiple layouts
    phone_number = soup.select_one('.d9FyLd b, .IZ6rdc, .EfDVh, .mw31Ze, .hgKElc b').text
    print(phone_number)

get_phone_answer_box()

-----------
'''
1-800-455-2720
+44 1279358438
0330 977 0444
(800) MY–APPLE (800–692–7753)
1 (206) 922-0880
1 (800) 925-6278
011 61 2 4227 9111
1 (800) 440-0680
(877) 767-9357
'''
Enter fullscreen mode Exit fullscreen mode

Scrape Google Calculator Answer Box

image

import requests, lxml
from bs4 import BeautifulSoup

headers = {
  "User-agent":
  "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582"
}

params = {
  "q": "32*3/3+12*332-1995",
}

def get_calculator_answerbox():
  html = requests.get('https://www.google.com/search', headers=headers, params=params)
  soup = BeautifulSoup(html.text, 'lxml')

  math_expression = soup.select_one('.XH1CIc').text.strip().replace(' =', '')
  calc_answer = soup.select_one('#cwos').text.strip()

  print(f"Expression: {math_expression}\nAnswer: {calc_answer}")


get_calculator_answerbox()

'''
Expression: ((32 * 3) / 3) + (12 * 332) - 1995
Answer: 2021
'''
Enter fullscreen mode Exit fullscreen mode

Scrape Google Weather Answer Box

image

from bs4 import BeautifulSoup
import requests, lxml

headers = {
  "User-Agent":
  "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582"
}

params = {
  "q": "london weather", # query
  "gl": "uk"             # country to search from (United Kingdom)
}

def get_weather_answerbox():
    response = requests.get('https://www.google.com/search', headers=headers, params=params)
    soup = BeautifulSoup(response.text, 'lxml')

    location = soup.select_one('#wob_loc').text
    weather_condition = soup.select_one('#wob_dc').text
    tempature = soup.select_one('#wob_tm').text
    precipitation = soup.select_one('#wob_pp').text
    humidity = soup.select_one('#wob_hm').text
    wind = soup.select_one('#wob_ws').text
    current_time = soup.select_one('#wob_dts').text

    print(f'Location: {location}\n'
          f'Weather condition: {weather_condition}\n'
          f'Temperature: {tempature}°C\n'
          f'Precipitation: {precipitation}\n'
          f'Humidity: {humidity}\n'
          f'Wind speed: {wind}\n'
          f'Current time: {current_time}\n')

    print('Forcast wind:')
    for wind_speed_direction in soup.select('.wob_noe .wob_hw'):
      try:
        wind_speed = wind_speed_direction.select_one('.wob_t').text

        '''
        extracts elemenets, splits the string by a SPACE and grabs 2nd and 4th index,
        and then joins via SPACE.

        Example:
        7 mph From northwest Sunday 9:00 AM ---> From northeast
        '''
        wind_direction = ' '.join(wind_speed_direction.select_one('.wob_t')['aria-label'].split(' ')[2:4])

        print(f"Wind Speed: {wind_speed}\nWind Direction: {wind_direction}\n")
      except:
        pass # or return None instead

    print('Forcast temperature:')
    for forecast in soup.select('.wob_df'):
      day = forecast.select_one('.Z1VzSb')['aria-label']
      weather = forecast.select_one('img.uW5pk')['alt']

      # check if selector exists in "current" HTML layout
      if forecast.select_one('.vk_gy .wob_t:nth-child(1)') is None:
        # different HTML layout selector
        high_temp = forecast.select_one('.gNCp2e .wob_t').text
      else:
        # different HTML layout selector
        high_temp = forecast.select_one('.vk_gy .wob_t:nth-child(1)').text

      low_temp = forecast.select_one('.QrNVmd .wob_t:nth-child(1)').text

      print(f'Day: {day}\nWeather: {weather}\nHigh: {high_temp}, Low: {low_temp}\n')


    print("Forecast Precipitation:")
    for recipitation_forecast in soup.select('.wob_hw'):
      try:
        recipitation = recipitation_forecast.select_one('.XwOqJe')['aria-label'].split(' ')[0]
      except: recipitation = None
      print(recipitation)


get_weather_answerbox()

---------
'''
Location: London, UK
Weather condition: Partly cloudy
Temperature: 20°C
Precipitation: 0%
Humidity: 55%
Wind speed: 7 mph
Current time: Monday 15:00

Forcast wind:
Wind Speed: 8 mph
Wind Direction: From east
...
Forcast temperature:
Day: Monday
Weather: Partly cloudy
High: 21, Low: 15
...
Forecast Precipitation:
0%
1%
1%
2%
2%
4%
12%
'''
Enter fullscreen mode Exit fullscreen mode

Scrape Google Stock Answer Box

There are currently two layouts depending on the country from which the search was initiated ('gl'='COUNTRY' query parameter):

image

image

import requests, lxml
from bs4 import BeautifulSoup

headers = {
  'User-agent':
  'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582'
}

stock_queries = [
  'netflix stock',
  'amazon stock',
  'alibaba stock',
  'cloudflare stock',
  'apple stock',
  'walmart stock'
]


def get_stock_answerbox():

  for query in stock_queries:
    params = {
      'q': query,
      'gl': 'us'
    }

    html = requests.get('https://www.google.com/search', headers=headers, params=params)
    soup = BeautifulSoup(html.text, 'lxml')

    title = soup.select_one('.oPhL2e').text.replace(u'\xa0', u'').split('>')[1]
    link = soup.select_one('.tiS4rf')['href']
    date_time = soup.select_one('[jsname=ihIZgd]').text.replace(' ·', '')
    market_status = soup.select_one('.TgMHGc span:nth-child(1)').text.strip().replace(':', '')
    currency = soup.select_one('.knFDje').text.strip()

    # two selectors which will handle two layouts
    current_price = soup.select_one('.wT3VGc, .XcVN5d').text

    price_change = soup.select_one('.WlRRw > span:nth-child(1)').text
    price_change_percent = soup.select_one('.jBBUv span:nth-child(1)').text.replace('(', '').replace(')', '')
    price_change_date = soup.select_one('.jdUcZd span').text.strip().capitalize()

    price_movement = 'Down' if '−' in price_change else 'Up'

    # different exchange & stock ticker layout handling (US website has a different stock layout)
    if soup.select_one('.HfMth') is None:
      exchange = soup.select_one('.EFkvDd').text
      stock_ticker = soup.select_one('.WuDkNe').text
    else:
      stock_exchange_ticker = soup.select_one('.HfMth').text
      exchange = stock_exchange_ticker.split(': ')[0]
      stock_ticker = stock_exchange_ticker.split(': ')[1]

    print(f'\nTitle: {title}\n'
        f'Link: {link}\n'
        f'Stock status: {market_status}\n'
        f'Current time: {date_time}\n'
        f'Exchange: {exchange}\n'
        f'Stock ticker: {stock_ticker}\n'
        f'Currency: {currency}\n'
        f'Current price: {current_price}\n'
        f'Price change: {price_change}\n'
        f'Percent price change: {price_change_percent}\n'
        f'Price movement: {price_movement}\n'
        f'Price change date: {price_change_date}\n')

    for stock_table_key, stock_table_value in zip(soup.select('.JgXcPd'), soup.select('.iyjjgb')):
        stock_key = stock_table_key.text
        stock_value = stock_table_value.text
        print(f"{stock_key}: {stock_value}")


get_stock_answerbox()

-------------
'''
Title: Netflix Inc
Link: https://www.google.com/finance/quote/NFLX:NASDAQ?sa=X&ved=2ahUKEwj1npu7oP7yAhWEcc0KHT1uDiEQ3ecFegQIGxAS
Stock status: Closed
Current time: Sep 14, 5:42 AM EDT 
Exchange: NASDAQ
Stock ticker: NFLX
Currency: USD
Current price: 589.29
Price change: −9.43 
Percent price change: 1.58%
Price movement: Down
Price change date: Today

Open: 598.57
High: 598.57
Low: 582.78
Mkt cap: 260.82B
P/E ratio: 61.10
Div yield: -
Prev close: 598.72
52-wk high: 615.60
52-wk low: 458.60

Title: Amazon.com, Inc.
Link: https://www.google.com/finance/quote/AMZN:NASDAQ?sa=X&ved=2ahUKEwjeoOW7oP7yAhVRK80KHZoSCSEQ3ecFegQIKRAS
Stock status: Closed
Current time: Sep 13, 6:54 PM EDT 
Exchange: NASDAQ
Stock ticker: AMZN
Currency: USD
Current price: 3,457.17
Price change: −11.98 
Percent price change: 0.35%
Price movement: Down
Price change date: Today

Open: 3,482.80
High: 3,497.96
Low: 3,438.00
Mkt cap: 1.75T
P/E ratio: 60.25
Div yield: -
52-wk high: 3,773.08
52-wk low: 2,871.00
...
'''
Enter fullscreen mode Exit fullscreen mode

Scrape Google Population Answer Box

image

import requests, lxml
from bs4 import BeautifulSoup

headers = {
  "User-agent":
  "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582"
}

params = {
  "q": "population of london",
}

def get_population_answerbox():
  html = requests.get('https://www.google.com/search', headers=headers, params=params)
  soup = BeautifulSoup(html.text, 'lxml')

  place = soup.select_one('.GzssTd span').text
  population_year = soup.select_one('.KBXm4e').text.split(' ')
  population = population_year[0]
  year = population_year[1].replace('(', '').replace(')', '')

  explore_more_link = soup.select_one('.tiS4rf')['href']
  sources = [source.text for source in soup.select('.kno-ftr span a')]

  print(f'{place}\nCurrent population: {population}\nCaptured in {year}\n{sources}\n{explore_more_link}\n')

  for other_city, other_population in zip(soup.select('.AleqXe'), soup.select('.kpd-lv')):
    other_place_city = other_city.text.strip()
    other_place_population = other_population.text

    print(f'{other_place_city}: {other_place_population}')


get_population_answerbox()

-----------
'''
London
Current population: 8.982 million
Captured in 2019
['Eurostat', 'United States Census Bureau', 'Feedback']
https://datacommons.org/place?utm_medium=explore&dcid=nuts/UKI&mprop=count&popt=Person&hl=en

London: 8.982 million‌
New York: 8.419 million‌
Scotland: 5.454 million‌
'''
Enter fullscreen mode Exit fullscreen mode

Scrape Google Converter Answer Box

image

import requests, lxml
from bs4 import BeautifulSoup

headers = {
  "User-agent":
  "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582"
}

params = {
  "q": "100 usd in gbp",
    "gl": "us"
}

def get_converter_answerbox():
  html = requests.get('https://www.google.com/search', headers=headers, params=params)
  soup = BeautifulSoup(html.text, 'lxml')

  conversion = soup.select_one('.SwHCTb').text
  conversion_currency = soup.select_one('.MWvIVe').text
  print(f"{conversion} {conversion_currency}")

get_converter_answerbox()

# 72.16 Pound sterling
Enter fullscreen mode Exit fullscreen mode

Scrape Google Dictionary Answer Box

image

import requests, lxml
from bs4 import BeautifulSoup

headers = {
  'User-agent':
  'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582'
}

params = {
  'q': 'swagger definition',
  'gl': 'us'
}

def get_dictionary_answerbox():
    html = requests.get('https://www.google.com/search', headers=headers, params=params)
    soup = BeautifulSoup(html.text, 'lxml')

    data = []

    for result in soup.select('.VpH2eb.vmod'):
        syllables = result.select_one('.DgZBFd span').text
        audio_link = f"https:{result.select_one('.brWULd audio source')['src']}"
        phonetic = result.select_one('.S23sjd .LTKOO span').text
        word_types = [word_type.text for word_type in result.select('.vdBwhd .YrbPuc')]

        definitions = [definition.text for definition in result.select('.LTKOO div[data-dobid=dfn]')]
        sentence_examples = [example.text for example in result.select('.ubHt5c')]
        similar_words = [similar_word.text for similar_word in result.select('.p9F8Cd span')]

        data.append({
          'syllables': syllables,
          'audio_link': audio_link,
          'phonetic': phonetic,
          'examples': {
              'word_type': [word_types],
              'definitions': [definitions],
              'sentence_examples': [sentence_examples],
              'similar_words': [similar_words]
          }
        })

    print(json.dumps(data, indent = 2, ensure_ascii = False))

get_dictionary_answerbox()

---------
'''
[
  {
    "syllables": "swag·ger",
    "audio_link": "https://ssl.gstatic.com/dictionary/static/sounds/20200429/swagger--_us_1.mp3",
    "phonetic": "ˈswaɡər",
    "examples": {
      "word_type": [
        [
          "verb",
          "noun",
          "adjective"
        ]
      ],
      "definitions": [
        [
          "walk or behave in a very confident and typically arrogant or aggressive way.",
          "a very confident and typically arrogant or aggressive gait or manner.",
          "denoting a coat or jacket cut with a loose flare from the shoulders."
        ]
      ],
      "sentence_examples": [
        [
          "\"he swaggered along the corridor\"",
          "\"they strolled around the camp with an exaggerated swagger\""
        ]
      ],
      "similar_words": [
        [
          "strut",
          "parade",
          "stride",
          "roll",
          "prance",
          "sashay",
          "swash",
          "boast",
          "brag",
          "bray",
          "bluster",
          "crow",
          "gloat",
          "posture",
          "pose",
          "blow one's own trumpet",
          "show off",
          "swank",
          "play to the gallery",
          "rodomontade",
          "strut",
          "parading",
          "roll",
          "prancing",
          "confidence",
          "arrogance",
          "self-assurance",
          "show",
          "ostentation",
          "boasting",
          "bragging",
          "bluster",
          "swashbuckling",
          "vainglory",
          "puffery",
          "swank",
          "braggadocio",
          "rodomontade",
          "gasconade"
        ]
      ]
    }
  }
]

'''
Enter fullscreen mode Exit fullscreen mode

Scrape Google First Organic Answer Box

image

image

import requests, lxml
from bs4 import BeautifulSoup

headers = {
  'User-agent':
  'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582'
}

params = {
    'q': 'luke skywalker lightsaber color',
    'gl': 'us'
}

def get_organic_result_answerbox():
  html = requests.get('https://www.google.com/search', headers=headers, params=params)
  soup = BeautifulSoup(html.text, 'lxml')

  answer = soup.select_one('.XcVN5d').text
  title = soup.select_one('.DKV0Md').text
  link = soup.select_one('.yuRUbf a')['href']
  snippet = soup.select_one('.hgKElc').text
  print(f"{answer}\n{title}\n{link}\n{snippet}")

get_organic_result_answerbox()

----------
'''
Green
Luke Skywalker's Lightsaber (Green Blade) | StarWars.com
https://www.starwars.com/databank/luke-skywalkers-lightsaber
After losing his father's lightsaber on Cloud City, Luke Skywalker constructed a replacement with a green plasma blade, its hilt similar to that of Obi-Wan Kenobi's weapon.
'''
Enter fullscreen mode Exit fullscreen mode

Scrape Google Second Organic Answer Box

image

image

import requests, lxml
from bs4 import BeautifulSoup

headers = {
  'User-agent':
  'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582'
}

params = {
    'q': 'Who is using IBM',
    'gl': 'us'
}

def get_second_organic_result_answerbox():
  html = requests.get('https://www.google.com/search', headers=headers, params=params)
  soup = BeautifulSoup(html.text, 'lxml')

  title = soup.select_one('.xpdopen .DKV0Md').text
  link = soup.select_one('.xpdopen .yuRUbf a')['href']
  displayed_link = soup.select_one('.xpdopen .iUh30').text

  # snippet layout handling + bulletpoints
  # example: shorturl.at/kIJQ9
  if soup.select_one('.xpdopen .co8aDb b') and soup.select_one('.TrT0Xe') is not None:
    snippet = soup.select_one('.xpdopen .co8aDb b').text
    bullet_points = '\n'.join([bullet_point.text for bullet_point in soup.select('.TrT0Xe')])
  else: 
    snippet = soup.select_one('.xpdopen .iKJnec').text
    bullet_points = None

  print(f'{title}\n{link}\n{displayed_link}\n{snippet}\n\nBullet points:\n{bullet_points}')


  if soup.select_one('#rso td:nth-child(1)') is None:
    pass
  else:
    print('\nTable:')
    for table_key, table_value in zip(
      soup.select('#rso td:nth-child(1)'), 
      soup.select('#rso td+ td')):

      key = table_key.text
      value = table_value.text
      print(f'{key}: {value}')


get_second_organic_result_answerbox()

----------
'''
Companies using IBM Watson and its marketshare - Enlyft
https://enlyft.com/tech/products/ibm-watson
https://enlyft.com › All Products › Machine Learning
The companies using IBM Watson are most often found in United States and in the Computer Software industry. IBM Watson is most often used by companies with 10-50 employees and 1M-10M dollars in revenue....Who uses IBM Watson?

Website: redcross.org
Country: United States
Revenue: >1000M
Company Size: >10000

# second layout
Types of Content Writing - DemandJump
https://www.demandjump.com/blog/types-of-content-writing
https://www.demandjump.com › blog › types-of-conten...
What are the types of content writing?

Bullet points:
Blogging. Creating blog posts is a staple of content writing. ... 
Copywriting. ... 
Technical Writing/Long Form. ... 
Social Media Posts. ... 
Emails.
SEO Content Writer. An SEO writer is one of the most important content writer types out there. ... 
Blog Content Writer. ... 
Long-Form Content Writer. ... 
Social Media Content Writer. ... 
Copywriter. ... 
Email Marketing Writer. ... 
Technical Writers. ... 
Press Release Writer.
'''
Enter fullscreen mode Exit fullscreen mode

Scrape Google Third Organic Answer Box

image

import requests, lxml
from bs4 import BeautifulSoup

headers = {
  'User-agent':
  'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582'
}

params = {
    'q': 'How much is a large pizza at Papa Johns',
    'gl': 'us'
}

def get_third_organic_result_answerbox():
  html = requests.get('https://www.google.com/search', headers=headers, params=params)
  soup = BeautifulSoup(html.text, 'lxml')

  title = soup.select_one('.ifM9O .LC20lb').text
  link = soup.select_one('.ifM9O .yuRUbf a')['href']
  displayed_link = soup.select_one('.ifM9O .iUh30').text
  snippet = soup.select_one('.ifM9O .iKJnec').text

  print(title, link, displayed_link, snippet, sep='\n')

  for table_key, table_value, table_value_price in zip(
    soup.select('.ztXv9~ tr+ tr td:nth-child(1)'), 
    soup.select('td:nth-child(2)'),
    soup.select('td~ td+ td')):

    key = table_key.text
    value = table_value.text
    price = table_value_price.text

    print(f"{key}: {value}, {price}")


get_third_organic_result_answerbox()

--------------
'''
Papa John's - Fast Food Menu Prices
https://www.fastfoodmenuprices.com/papa-johns-prices/
https://www.fastfoodmenuprices.com › papa-johns-prices
Papa John's Menu Prices
Cheese (Original): Medium, $13.00
Cheese (Original): Large, $15.00
Cheese (Original): Extra Large, $17.00
'''
Enter fullscreen mode Exit fullscreen mode

Scrape Google Translation Answer Box

image

import requests, lxml
from bs4 import BeautifulSoup

headers = {
  'User-agent':
  'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582'
}

params = {
    'q': 'hello in french',
    'gl': 'us'
}

def get_translation_answerbox():
  html = requests.get('https://www.google.com/search', headers=headers, params=params)
  soup = BeautifulSoup(html.text, 'lxml')

  detected_language = soup.select_one('.source-language').text
  text = soup.select_one('#tw-source-text-ta').text
  pronounce = soup.select_one('#tw-source-rmn .Y2IQFc').text

  target_language = soup.select_one('.target-language').text
  target_text = soup.select_one('#tw-target-text .Y2IQFc').text

  print(detected_language, text, pronounce, target_language, target_text, sep='\n')

  for result in soup.select('.DFDLnc'):
      title = result.next_sibling.select_one('.hrcAhc').text
      words = result.next_sibling.select_one('.MaH2Hf').text

      print(title, words, sep='\n')

get_translation_answerbox()

------------
'''
English - detected
hello
həˈlō
French
Bonjour

Bonjour!
Hello!, Hi!, Good morning!, Good afternoon!, How do you do?, Hallo!
Salut!
Hi!, Hello!, Salute!, All the best!, Hallo!, Hullo!
Tiens!
Hallo!, Hello!, Hullo!, Why!
Allô!
Hello!, Hullo!, Hallo!
'''
Enter fullscreen mode Exit fullscreen mode

Scrape Google Formula Answer Boxes

There're currently two layouts of formulas that I found, this and the one below. The first one is straightforward, the second one is a bit tricky.

image

import requests, lxml, numpy, time
from bs4 import BeautifulSoup
from selenium import webdriver

headers = {
  'User-agent':
  'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582'
}

params = {
    'q': 'formula of distance',
    'gl': 'us'
}

# example query: shorturl.at/euPX1
def get_formula_answerbox():
  html = requests.get('https://www.google.com/search', headers=headers, params=params)
  soup = BeautifulSoup(html.text, 'lxml')

  formula_img = soup.select_one('.IGEbUc')['src'].strip()
  formula = soup.select_one('.IGEbUc')['alt']
  print(formula, formula_img, sep='\n')

  temp_list = []
  clear_list = []

  for symbol, definition in zip(
    soup.select('img.ZFAGbf'),
    soup.select('.GQLwEe')):

    '''
    appending to temporary list:
    1. formula symbol name
    2. formula symbol (plain text)
    3. formula symbol image (svg)
    '''
    temp_list.append([definition.text, symbol['alt'], symbol['src']])

  # https://stackoverflow.com/a/45323085/15164646
  # multidimensional list -> flat list
  for item in list(numpy.array(temp_list).flat):
    if item not in clear_list:
      clear_list.append(item)

  for result in clear_list:
    print(result)


get_formula_answerbox()

--------------
'''
d = \sqrt{(x_2 - x_1)^2 + (y_2-y_1)^2}
https://www.gstatic.com/education/formulas2/397133473/en/distance_formula.svg
distance
d
https://www.gstatic.com/education/formulas2/397133473/en/distance_formula_d.svg
coordinates of the first point
(x_1, y_1)
https://www.gstatic.com/education/formulas2/397133473/en/distance_formula_x1y1.svg
coordinates of the second point
(x_2, y_2)
https://www.gstatic.com/education/formulas2/397133473/en/distance_formula_x2y2.svg
'''
Enter fullscreen mode Exit fullscreen mode

The second layout is different because parsing such formulas (screenshot below) in such layout don't have <img> option to scrape, instead, it's a plain text that doesn't make any sense.

To make it kinda work, we can use selenium to take a screenshot via .screenshot('filename.png') and mathpix to make a post request and process that screenshot image to a Latex code.

Note: Using selenium slows down everything strong enough.

image

# example URL: shorturl.at/bfvAU
def get_formula_solve_answerbox():
  html = requests.get('https://www.google.com/search', headers=headers, params=params)
  soup = BeautifulSoup(html.text, 'lxml')

  if soup.select_one('#HE4HM') is not None:
    title = soup.select_one('.vk_h').text
    solving = soup.select_one('.goog-menu-button-caption').text
    print(title, solving, sep='\n')

    for i in soup.select('.lr-fy-ol.Gi02q'):
        print(i.text)

  else: print('no formula found')


  # ----------------------------
  # Another way using Selenium

  driver = webdriver.Chrome(executable_path = 'path/to/chromedriver.exe')

    CURRENT_DATE = time.strftime("%d/%m/%y")

    queries = [
        'volume of a cylinder',
        'volume of a sphere',
        'volume of a cone',
        'volume of a cube',
        'volume of a pyramid',
        'sphere surface area',
        'cube surface area'
    ]

    for query in queries:

        current_time = time.strftime("%H:%M:%S")
        print(current_time)

        driver.get(f"https://www.google.com/search?q={query}&gl=us")

        plain_text_formula = driver.find_element_by_css_selector('.lr-fy-ol.Gi02q').text.replace('\n', '')
        print(plain_text_formula)

        driver.find_element_by_xpath('//*[@id="HE4HM"]/div/div[3]/div[2]').screenshot(f'formula_{CURRENT_DATE}_{current_time}.png')
        time.sleep(1.3)

    driver.quit()


get_formula_solve_answerbox()


----------
'''
# Bs4 output
Vπr2h
Va3
V43πr3

# Selenium output
V=πr2h
V=43πr3
V=πr2h3
V=a3
V=lwh3
A=4πr2
'''
Enter fullscreen mode Exit fullscreen mode

Screenshots from Selenium:

image

And few examples of actual screenshots:
image
image
image
image

To convert images using mathpix, we need to process image and convert screenshots to Latex code.

  1. Make authorization.
  2. Make a post request and send already taken screenshots. 
import requests
import json

r = requests.post("https://api.mathpix.com/v3/text",
    files={"file": open("YOUR_IMAGE_FILE.png","rb")},
    data={
      "options_json": json.dumps({
        "math_inline_delimiters": ["$", "$"],
        "rm_spaces": True
      })
    },
    headers={
        "app_id": "APP_ID",
        "app_key": "APP_KEY"
    }
)
print(json.dumps(r.json(), indent=4, sort_keys=True))
Enter fullscreen mode Exit fullscreen mode

Get response:

{
  "auto_rotate_confidence": 0.0046418966250847404,
  "auto_rotate_degrees": 0,
  "confidence": 0.9849104881286621,
  "confidence_rate": 0.9849104881286621,
  "is_handwritten": true,
  "is_printed": false,
  "latex_styled": "f(x)=\\left\\{\\begin{array}{ll}\nx^{2} & \\text { if } x<0 \\\\\n2 x & \\text { if } x \\geq 0\n\\end{array}\\right.",
  "request_id": "bd02af63ef187492c085331c60151d98",
  "text": "$f(x)=\\left\\{\\begin{array}{ll}x^{2} & \\text { if } x<0 \\\\ 2 x & \\text { if } x \\geq 0\\end{array}\\right.$"
}
...
Enter fullscreen mode Exit fullscreen mode

Have a look at other available request parameters, and format description. Their API pricing.

Note: possibly using mathpix will be enough for your task, but if it's not, use Wolfram Alpha instead, this is their subject area of expertise.

Scrape Google Flights Answer Boxes

image

import requests, lxml
from bs4 import BeautifulSoup

headers = {
  'User-agent':
  'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582'
}

params = {
    'q': 'flight from london to new york',
    'gl': 'us'
}

def get_flights_answerbox():
  html = requests.get('https://www.google.com/search', headers=headers, params=params)
  soup = BeautifulSoup(html.text, 'lxml')

  title = soup.select_one('.N00E5e .mfMhoc').text.strip()
  print(f"{title}\n")

  for flight in soup.select('.ikUyY'):
    link = flight['href']
    airline = flight.select_one('.ps0VMc').text.strip()
    flight_duration = flight.select_one('.sRcB8').text.strip()
    flight_option = flight.select_one('.u85UCd').text.strip()
    price = flight.select_one('.xqqLDd').text.strip()

    print(f'{airline}\n{flight_duration}\n{flight_option}\n{price}\n{link}\n')


get_flights_answerbox()

-----------
'''
Flights from London, United Kingdom (all airports) to New York, NY (all airports)

Tap Air Portugal
12h 15m+
Connecting
from $327
https://www.google.com/flights?gl=us&source=flun&uitype=cuAA&hl=en&curr=USD&tfs=CAEQAhotEgoyMDIxLTExLTAxMgJUUGoMCAISCC9tLzA0anBscg0IAhIJL20vMDJfMjg2Gi0SCjIwMjEtMTEtMDgyAlRQag0IAhIJL20vMDJfMjg2cgwIAhIIL20vMDRqcGx6aENqUklXbkpCZDA1WmRFeENRbTlCUmpnMVRIZENSeTB0TFMwdExTMHRMVzkxYldjeE9FRkJRVUZCUjBaQ2NtaEZSMDU1VDBGQkVnTmpWRkFhQ3dpOC93RVFBaG9EVlZORU9BTnd2UDhC&sa=X&ved=2ahUKEwjXrLmSx4DzAhWQGs0KHZZVAQYQ1RUoAHoECBcQGg

JetBlue
8h 48m
Nonstop
from $402
https://www.google.com/flights?gl=us&source=flun&uitype=cuAA&hl=en&curr=USD&tfs=CAEQAhotEgoyMDIxLTExLTAxMgJCNmoMCAISCC9tLzA0anBscg0IAhIJL20vMDJfMjg2Gi0SCjIwMjEtMTEtMDgyAkI2ag0IAhIJL20vMDJfMjg2cgwIAhIIL20vMDRqcGx6aENqUklXbkpCZDA1WmRFeENRbTlCUmpnMVRIZENSeTB0TFMwdExTMHRMVzkxYldjeE9FRkJRVUZCUjBaQ2NtaEZSMDU1VDBGQkVnTnVRallhQ3dpSXVnSVFBaG9EVlZORU9BTndpTG9D&sa=X&ved=2ahUKEwjXrLmSx4DzAhWQGs0KHZZVAQYQ1RUoAXoECBcQGw
...
'''
Enter fullscreen mode Exit fullscreen mode

Scrape Google Matches Answer Boxes

image

import requests, lxml
from bs4 import BeautifulSoup

headers = {
  'User-agent':
  'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582'
}

params = {
    'q': 'nfl matches',
    'gl': 'us'
}


''' 
Properly working while each team has scores, for example: 
Chiefs: 29
Browns: 29

But when only date and time is shown it won't scrape it, for example:
Today
20:10

It would only scrape team which HAVE scores but the scores attached to teams won't be correct. 
'''
def get_sport_matches_answerbox():
  html = requests.get('https://www.google.com/search', headers=headers, params=params)
  soup = BeautifulSoup(html.text, 'lxml')

  title = soup.select_one('.ofy7ae').text

  if soup.select_one('.mKwiob'):
    league = soup.select_one('.mKwiob').text
  else: league = 'Not mentioned'

  print(title, league, sep='\n')
  print()

  # zip() will get data in parallel but it will find all needed elements at once
  for first_team, second_team, first_score, second_score, status, match_date, video_highlight in zip(
    soup.select('.L5Kkcd+ .L5Kkcd span'),
    soup.select('.L5Kkcd:nth-child(5) span'),
    soup.select('.L5Kkcd:nth-child(5) .imspo_mt__t-sc .imspo_mt__tt-w'),
    soup.select('.imspo_mt__lt-t .imspo_mt__tt-w'),
    soup.select('.imspo_mt__match-status'),
    soup.select('.imspo_mt__ms-w div div :nth-child(1)'),
    soup.select('.BbrjBe')):

    match_status = status.text
    match_game_date = match_date.text
    match_video_highlight = video_highlight.select_one('a')['href']

    print(match_status, match_game_date, match_video_highlight, sep='\n')
    print(f"{first_team.text}: {first_score.text}")
    print(f"{second_team.text}: {second_score.text}")
    print()


get_sport_matches_answerbox()

----------
'''
NFL
Week 1 of 18

Final
Sun, Sep 12
https://www.youtube.com/watch?v=4g2R-kn_-0c&feature=onebox
Chiefs: 29
Browns: 29

Final
Sun, Sep 12
https://www.youtube.com/watch?v=dc_e6GdLzls&feature=onebox
Patriots: 17
Dolphins: 16
...
'''
Enter fullscreen mode Exit fullscreen mode

Scrape Google Address Answer Box

image

import requests, lxml
from bs4 import BeautifulSoup

headers = {
  "User-agent":
  "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582"
}

params = {
  "q": "spotlight 29 casino address",
}

def get_address_answerbox():
  html = requests.get('https://www.google.com/search', headers=headers, params=params)
  soup = BeautifulSoup(html.text, 'lxml')

  address = soup.select_one('.sXLaOe').text
  print(address)

get_address_answerbox()

# 46-200 Harrison Pl, Coachella, CA 92236
Enter fullscreen mode Exit fullscreen mode

Scrape Google Age Answer Box

Image description

import requests, lxml, json
from bs4 import BeautifulSoup

headers = {
  "User-agent":
  "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582"
}

params = {
  "q": "selena gomez age",
}

def get_age_answerbox():
  html = requests.get('https://www.google.com/search', headers=headers, params=params)
  soup = BeautifulSoup(html.text, 'lxml')

  data = []

  age = soup.select_one('.XcVN5d').text
  birth_date = soup.select_one('.kZ91ed').text

  people_also_search_for = [{
    "title": search.select_one('.rOVRL').text,
    "link": f"https://www.google.com{search.select_one('a')['href']}",
    "age": search.select_one('.C9iYEe').text
  } for search in soup.select('.PZPZlf.lLRSIb')]

  data.append({
    "age": age,
    "birth_date": birth_date,
    "people_also_search_for": people_also_search_for
  })

  print(json.dumps(data, indent=2, ensure_ascii=False))

---------
'''
[
  {
    "age": "29 years",
    "birth_date": "July 22, 1992",
    "people_also_search_for": [
      {
        "title": "Justin Bieber",
        "link": "https://www.google.com/search?q=justin+bieber+age&stick=H4sIAAAAAAAAAONgFuLUz9U3SC82KytS4gIxzcqNivNMtfgCUouK8_OCM1NSyxMrixcximYnW-kXpOYX5KQCKZCcVWJ66iJWwazS4pLMPIWkzNSk1CIFoBgAbN1xHFUAAAA&sa=X&ved=2ahUKEwjdp4aN9N3zAhXHGs0KHUzUAEEQxA16BAgmEAQ",
        "age": "27 years"
      },
      {
        "title": "Ariana Grande",
        "link": "https://www.google.com/search?q=ariana+grande+age&stick=H4sIAAAAAAAAAONgFuLUz9U3SC82KytS4gIxLdOzUypNtPgCUouK8_OCM1NSyxMrixcximYnW-kXpOYX5KQCKZCcVWJ66iJWwcSizMS8RIX0osS8lFQFoBgASMcE2FUAAAA&sa=X&ved=2ahUKEwjdp4aN9N3zAhXHGs0KHUzUAEEQxA16BAgmEAY",
        "age": "28 years"
      },
      {
        "title": "Taylor Swift",
        "link": "https://www.google.com/search?q=taylor+swift+age&stick=H4sIAAAAAAAAAONgFuLUz9U3SC82KytSAjNTckzNzLX4AlKLivPzgjNTUssTK4sXMYpmJ1vpF6TmF-SkAimQnFVieuoiVoGSxMqc_CKF4vLMtBIFoBAA8NTXJlMAAAA&sa=X&ved=2ahUKEwjdp4aN9N3zAhXHGs0KHUzUAEEQxA16BAgmEAg",
        "age": "31 years"
      }
    ]
  }
]
'''
Enter fullscreen mode Exit fullscreen mode

Scrape Google Answer Box Using SerpApi

SerpApi is a paid API with a free plan.

The difference is that everything (except the second formula layout) that was done above and more is already present by default and there's no need to maintain the parser over time if something in the HTML layout will be changed thus break the parser.

Instead, the only thing that really needs to be done is to iterate over structured JSON and get the data you want fast.

Example code for all mentioned layouts except the second formula layout:

from serpapi import GoogleSearch
import os, json

def serpapi_answerbox_example():
  params = {
    "api_key": os.getenv("API_KEY"),         # API key environment
    "engine": "google",                      # search engine
    "q": "flight from london to new york",   # query
    "location": "United States",             # from where search to originate
    "google_domain": "google.com",           # domain name
    "gl": "us",                              # location to search from
    "hl": "en"                               # language
  }

  search = GoogleSearch(params)
  results = search.get_dict()

  print(json.dumps(results['answer_box'], indent = 2, ensure_ascii = False))

-----------
'''
{
  "type": "google_flights",
  "title": "Flights from London, United Kingdom (all airports) to New York, NY (all airports)",
  "flights": [
    {
      "link": "https://www.google.com/flights?gl=us&hl=en&source=flun&uitype=cuAA&curr=USD&tfs=CAEQAhotEgoyMDIxLTExLTAxMgJUUGoMCAISCC9tLzA0anBscg0IAhIJL20vMDJfMjg2Gi0SCjIwMjEtMTEtMDgyAlRQag0IAhIJL20vMDJfMjg2cgwIAhIIL20vMDRqcGx6aENqUklZVzVMZWtNM1YzVXhhVFJCUmpObGJHZENSeTB0TFMwdExTMHRMUzF2ZFhSa04wRkJRVUZCUjBaRE0wOWpURzFVYjBGQkVnTmpWRkFhQ3dpOC93RVFBaG9EVlZORU9BTnd2UDhC&sa=X&ved=2ahUKEwir8vL554LzAhUVnWoFHbhSAnAQ1RUoAHoECBYQGg",
      "flight_info": [
        "Tap Air Portugal",
        "12h 15m+",
        "Connecting",
        "from $327"
      ]
    },
    {
      "link": "https://www.google.com/flights?gl=us&hl=en&source=flun&uitype=cuAA&curr=USD&tfs=CAEQAhotEgoyMDIxLTExLTAxMgJCNmoMCAISCC9tLzA0anBscg0IAhIJL20vMDJfMjg2Gi0SCjIwMjEtMTEtMDgyAkI2ag0IAhIJL20vMDJfMjg2cgwIAhIIL20vMDRqcGx6aENqUklZVzVMZWtNM1YzVXhhVFJCUmpObGJHZENSeTB0TFMwdExTMHRMUzF2ZFhSa04wRkJRVUZCUjBaRE0wOWpURzFVYjBGQkVnTnVRallhQ3dpSXVnSVFBaG9EVlZORU9BTndpTG9D&sa=X&ved=2ahUKEwir8vL554LzAhUVnWoFHbhSAnAQ1RUoAXoECBYQGw",
      "flight_info": [
        "JetBlue",
        "8h 48m",
        "Nonstop",
        "from $402"
      ]
    }
  ]
}
'''
Enter fullscreen mode Exit fullscreen mode

Code in the online IDE • Google Direct Answer Box API • Mathpix • Wolfram Alpha

Outro

If you have any questions or something isn't working correctly or you have any suggestions, feel free to drop a comment in the comment section or via Twitter at @serp_api. If you found a bug using SerpApi, please, add this bug on SerpApi Forum.

Yours,
Dimitry, and the rest of SerpApi Team.

Top comments (0)

Read next

hyperkai profile image

Warning Filter in Python

Super Kai (Kazuya Ito) -

dee_codes profile image

Setting up Redis in a Next.js Application

Hadiza Mohammed -

yassinesallami profile image

How to create an image authentication system with python streamlit and canva!

Yassine Sallami -

paharihacker profile image

A Complete Guide to React useEffect vs useLayoutEffect for Beginners

Rajesh Dhiman -