User:Rita Graca/gradproject/prototyping/twittertrends

From XPUB & Lens-Based wiki

Twitter trends

API

Cancel culture happens in Twitter through design features such as hashtags and trending topics.
To investigate better this movement, I understood I had to inform myself about which topics/people/things were being cancelled, how was the engagement with this topic, what were the language and strategies used.


1. Using the Twitter API I could get the current trends in the US.

Steps:

  • Create a Twitter developer account
  • Get keys and tokens from Twitter
  • Install Ruby
  • Install Twurl
  • Install JQ to read JSON
  • Use the command line
  twurl "/1.1/trends/place.json?id=23424977" | jq


2. I was only interested in the trends related to cancel culture, so I used Python to develop the script a bit more.

Steps:

  • Use Python library Tweepy
  • Get trends
  • Look for trends with words related with cancel culture


3. It was useful to save the trends. Instead of saving them in a .txt file, it made more sense to post them back in a Twitter account.

Steps:

  • Create a status with the search results (a status is a tweet in the library)


4. To make it look for trends regularly I created a cron job on my computer.

  46 * * * * /usr/local/bin/python3 /Users/0972516/desktop/ritaiscancelled/trends.py


Outcome:
The account @CancelledWho looks for trends related to my topic and posts them. This way I can be always monitoring an important topic of my research.

#!/usr/bin/python

import tweepy
import key # this is a pyhton file with my API passwords
import time

# using the passwords to OAuth process, authentication
auth = tweepy.OAuthHandler(key.consumer_key, key.consumer_secret)
auth.set_access_token(key.access_token, key.access_token_secret)
api = tweepy.API(auth)


trends1 = api.trends_place(23424977)  # american woeid id

trends = set([trend['name'] for trend in trends1[0]['trends']]) # just getting the name, not timestamp, author, etc.

trendsLower = [item.lower() for item in trends] # makes everything lowercase, important for then to match with cancelwords.txt
trendsLine = '\n'.join(trendsLower) # makes it more readable, puts the names with line breaks

#print(trendsLine)

cancelwords = ["cancelled", "canceled", "cancel", "isoverparty", "booed", "boycott"]
#print(cancelwords)


for line in trendsLine.splitlines():
            #print(line)
    for word in cancelwords:
        if  word in line:
            try:
                status = "Who are we fighting today? " + line
                print(status)
                api.update_status(status) # Creates a tweet, a status is a tweet
                time.sleep(5)
            except tweepy.TweepError as e: # the error is occuring when the last status is the same
                print("ups, you already tweeted this")
                break
    #    time.sleep(3600) # so the script will wait 1h to run again if catches error


Outcome


Get trends from historical archive

I wanted to also have access to old trending topics. e.g. trending topics from last year.

Steps:

  • Have sandbox subscription from Twitter (attention: this only allows 50 requests)
  • Install searchtweets Python library (wrapper for the Twitter premium search APIs)


I could see it was possible before, but with the current limitations of Twitter API I could only search for tweets, not trends.
So, mission failed.


Script for tweets in general:

from searchtweets import ResultStream, gen_rule_payload, load_credentials, collect_results

import requests

premium_search_args = load_credentials("~/twitter_keys.yaml",
                                       yaml_key="search_tweets_premium",
                                       env_overwrite=False)


rule = gen_rule_payload("isoverparty", from_date="2019-09-07", to_date="2019-09-09", results_per_call=10) # testing with a sandbox account

print(rule)

from searchtweets import collect_results

tweets = collect_results(rule,
                         max_results=10,
                         result_stream_args=premium_search_args)

# print(tweets.all_text)

[print(tweet.all_text, end='\n\n') for tweet in tweets[0:10]];

Scrape from existing website

(less accurate, abandoned)

When the API Archive failed, I started looking for existing projects scraping trends. There's a website that has been saving daily trends from Twitter.
Using Selenium I could go through the website and scrape the trends related with cancel culture.
I stopped the prototype because I couldn't rely on the website accuracy.

Script to go through different pages with Selenium:

from selenium import webdriver
from selenium.webdriver.common.keys import Keys
import os
import time
import datetime
from pprint import pprint
import requests
import multiprocessing
import base64

m = 1
d = 1

driver = webdriver.Firefox(executable_path=os.path.dirname(os.path.realpath(__file__)) + '/geckodriver')

for m in range(0, 12):  # for every month until 12

    for d in range(0, 31): # for every day until 31
        url = ("https://us.trend-calendar.com/trend/2019-{0:02d}-{1:02d}.html".format(m+1, d+1))
        print(url)
        # Implicit wait tells Selenium how long it should wait before it throws an exception
        driver.implicitly_wait(5)
        # driver.get(url)
        time.sleep(3)

        #driver.find_element_by_xpath("/html/body/div[1]/div/div/main/article/div/section/div[1]/a").click(); # click the 'More' button
        #print ('opening day......')
        #time.sleep(3)
        #driver.close()
        #print("DONE! Closing Window")

    else:
        print("Month finished")


else:
    print("Year finished")


Use existing database

As I couldn't search in the API archive I looked for existing archives.
https://archive.org/details/twitterstream