scrape_age_and_bio.py

import csv, random, time
from inspect import getmodule
from tinderbotz.session import Session
from tinderbotz.helpers.constants_helper import *

# CITY_INFO = ("New York", 40.76187, -73.98576)
# CITY_INFO = ("Seoul", 37.5665, 126.9780)
CITY_INFO = ("Tokyo", 35.69126, 139.70103)
MAX_COUNT = 2000

if __name__ == "__main__":

    # creates instance of session
    session = Session()
    # session = Session(headless=True)

    # set a custom location
    session.set_custom_location(latitude=CITY_INFO[1], longitude=CITY_INFO[2])

    # replace this with your own email and password!
    email = "<email address>"
    password = "<password>"

    # login using your google account with a verified email! Alternatively, you can use Facebook login
    session.login_using_facebook(email, password)

    # adjust allowed distance for geomatches
    # Note: PARAMETER IS IN KILOMETERS!
    session.set_distance_range(km=15)

    # set range of prefered age
    session.set_age_range(18, 28)

    # set interested in gender(s) -> options are: WOMEN, MEN, EVERYONE
    session.set_sexuality(Sexuality.WOMEN)

    # Create a csv file
    file_name = "/tmp/{city}_{count}.csv"
    f = open(file_name.format(city = CITY_INFO[0], count = MAX_COUNT), 'w')
    writer = csv.writer(f)

    # Add header
    writer.writerow(['age', 'bio'])

    # start scraping as much geomatches as possible
    count = 0
    refresh_count = 0
    while count < MAX_COUNT:
        geomatch = session.get_geomatch()

        # check if crucial data is not empty (This will rarely be the case tho, but we want a 'clean' dataset
        if geomatch.get_age() is not None \
                and geomatch.get_bio() is not None:
            count += 1

            print(str(count) + ": " + str(geomatch.get_age()) + ", " + geomatch.get_bio())
            # Add line to the csv file.
            writer.writerow([geomatch.get_age(), geomatch.get_bio()])

            # account is scraped, now dislike and go next (since dislikes are infinite)
            # NOTE: if no amount is passed, it will dislike once -> same as => dislike(amount=1)
            # NOTE: if you have TinderPlus, -Gold or -Platinum you could also use session.like()
            session.dislike()

        else:
            # refresh webpage once in five times and go for another geomatch
            refresh_count += 1
            session.dislike()
            if refresh_count > 5:
                refresh_count = 0
                session.browser.refresh()

        # make a random sleep between dislikes between 0 and 3 seconds to mimic looks human-like, not spammy behaviour
        sleepy_time = random.random() * 3
        time.sleep(sleepy_time)

    f.close()