@moeshitbot annotated source code

send_tweet.py

#       ___ _       ______  ____ 
#      /   | |     / / __ \/ __ \
#     / /| | | /| / / / / / / / /
#    / ___ | |/ |/ / /_/ / /_/ / 
#   /_/  |_|__/|__/\____/\____/  
#
#       @moeshitbot, by timgor |  timgor.moe
#       annotated source code
#       
#   this is my shitty cute anime pics twitter bot.
#   it makes requests to my timgor.moe api and yeets them up on to twitter.
#   a lot of the heavy lifting is done by tweepy (www.tweepy.org), 
#   a twitter api wrapper for python. it makes doing this all an absolute breeze
#   and is the gold standard for making twitter bots.
#   
#   also huge thanks to saucenao.com for letting me procedurally get the source (sauce)
#   for all the images. i would never make this kind of bot without giving the original
#   artists credit for their hard work, and saucenao provides me with that ability.
#   for those unaquainted, saucenao is a reverse image search for artwork that
#   is a goddamn blessing for figuring out who drew that amzing lewd you just downloaded.
#   

print('sending a tweet')

import json
import tweepy
import requests
from io import BytesIO
import urllib
from random import choice

# load the saucenao api creds from a local json file
with open('saucenao_credentials.json') as f:
    saucenao_credentials = json.load(f)
    
# request to saucenao api
# tries with up to 100 different images just in case it can't find the sauce for one
sauce = ''
try_num = 0
num_tries = 100
while not sauce:  # try until we hit sauce
    url = requests.get('https://timgor.moe/api/random')  # get image url from timgor.moe api
    
    # make the request to the saucenao api
    # probably should've encoded the query parameters more readably
    # basically means:
    #   json format
    #   one result
    #   >80% similarity
    #   all databases
    sn_url = 'http://saucenao.com/search.php?output_type=2&numres=1&minsim=80!&db=999&api_key='+saucenao_credentials['api_key']+'&url='+urllib.parse.quote(url.content)
    sn_response = requests.get(sn_url)
    #  it's json parsing time
    sn_json = json.loads(sn_response.content)
    result = sn_json['results'][0]
    # set sauce if we got anything with high enough similarity
    if float(result['header']['similarity']) > 80:
        sauce = result['data']['ext_urls'][0]  
    
    try_num += 1
    # exit if retries exceeded
    if num_tries < try_num:
        print('retries exceeded')
        exit(1)

# download the image and proceed if successfull
data = requests.get(url.content)
if data.status_code == 200:
    
    # this code is designed to be as portable as possible, 
    # and runnable in places where you don't neccesarilly have
    # write access. to have a file object usable with tweepy, 
    # i create a BytesIO object which acts a file, but is kept only in RAM
    
    image = BytesIO(data.content)
    
    # load twitter api and user creds from a local json file
    with open('twitter_credentials.json') as f:
        twitter_credentials = json.load(f)
        
    # the witty messages are also kept externally
    # it's nice to have these kinds of things in a seperate file so it's harder 
    # (not impossible) to break the whole program adding one
    # this also maximizes code prettyness and reusability
    with open('messages.json') as f:
        messages = json.load(f)
    
    # messages are all encoded as string.format() templates so i can easily insert the sauce url, like so:
    # "programmatically selected from my anime pics folder. \nsauce: {0} (courtesy of saucenao.com)"
    message = choice(messages).format(sauce)  # choose random message
    
    # tweepy wants a filename with an extension for guessing MIME content output_type
    # i need to extract it from the headers of the response i got earlier
    # this clusterfuck of a listcomp does that, i don't remember how
    filename = [a for a in data.headers['Content-Disposition'].split(';') if 'filename=' in a][0].replace('filename=','').replace('"','')
    
    # tweepy handles all the authorization for you
    # manually dealing with oauth2 is like having razor blades that 
    # have been heated to 1000 C shoved up your dick hole. it's not fun i assure you
    # i am very grateful for tweepy
    auth = tweepy.OAuthHandler(twitter_credentials['consumer_key'], twitter_credentials['consumer_secret'])
    auth.set_access_token(twitter_credentials['access_token'], twitter_credentials['access_token_secret'])
    api = tweepy.API(auth)
    
    
    try:
        # sends the tweet with the BytesIO object and filename from earlier
        api.update_with_media(filename,status=message,file=image)  
    except:
        # that's right kids, except: pass in actual production code
        # please never do this
        # do actual error handling
        # don't be like me
        pass

print('tweet sent')  # we did it, reddit!