Botgiornissimo (v0.3)
Posted
I’ve been experimenting with a telegram bot, I figured this might be as good a place as any to lay down the code for it and some thoughts about it.
I started with an R version, sadly the R package that one uses to connect to the Telegram API does not have as many functionalities as the python version from which it’s inspired - there is no function to send polls, and I wanted the bot to be able to send polls.
The idea of the bot is to have cheesy “Good morning” images sent out, well, every morning. I was part of a group where they were usually - unironically - shared, and they’re so bad they’re good, in a way. The true mark of the kitsch.
The R version of the bot had the images sent as a response to the command “/buongiornissimo”. I still had the images stored locally on my computer, the bot took all the generic ones plus the ones linked to the particular day of the week - some of them read “Good Wednesday” or something like that - and then chose one at random. The code looks like this:
# General settings
library(telegram.bot)
library(magrittr)
library(lubridate)
bg_token <- Sys.getenv("BOTGIORNISSIMO_TOKEN")
botgiornissimo <- Bot(token = bg_token)
setwd("~/Documents/R/botgiornissimo")
# buongiornissimo command
buongiornissimo_df <- data.frame(Chat_id = 0,
Count = 0,
Time = Sys.time())
buongiornissimo <- function(bot, update) {
chat_id <- update$message$chat_id
if(!(chat_id %in% buongiornissimo_df$Chat_id)) {
buongiornissimo_df <<- buongiornissimo_df %>%
mutate(Time = as_datetime(Time)) %>%
add_row(
Chat_id = chat_id,
Count = 0,
Time = as_datetime(Sys.time())
)
}
buongiornissimo_count <- buongiornissimo_df %>% filter(Chat_id == chat_id) %$% Count
buongiornissimo_time <- buongiornissimo_df %>% filter(Chat_id == chat_id) %$% Time
if (day(as_datetime(buongiornissimo_time)) < day(Sys.time())) {
buongiornissimo_count <- 0
}
if (buongiornissimo_count == 0) {
buongiornissimo_df <<- buongiornissimo_df %>%
mutate(Count = ifelse(Chat_id == chat_id,
1,
Count),
Time = ifelse(Chat_id == chat_id,
Sys.time(),
Time))
image <- switch(weekdays(Sys.time()),
Monday = list.files("./buongiornissimo") %>%
str_subset(pattern = "(^LU|^X|^IMG)"),
Tuesday = list.files("./buongiornissimo") %>%
str_subset(pattern = "(^MA|^X|^IMG)"),
Wednesday = list.files("./buongiornissimo") %>%
str_subset(pattern = "(^ME|^X|^IMG)"),
Thursday = list.files("./buongiornissimo") %>%
str_subset(pattern = "(^GI|^X|^IMG)"),
Friday = list.files("./buongiornissimo") %>%
str_subset(pattern = "(^VE|^X|^IMG)"),
Saturday = list.files("./buongiornissimo") %>%
str_subset(pattern = "(^SA|^X|^IMG)"),
Sunday = list.files("./buongiornissimo") %>%
str_subset(pattern = "(^DO|^X|^IMG)"),
) %>%
sample(1)
bot$sendPhoto(chat_id = chat_id,
photo = str_c("~/Documents/R/botgiornissimo/buongiornissimo/", image),
sprintf("Buongiornissimo %s!", update$message$from$first_name))
} else {
bot$sendMessage(chat_id = update$message$chat_id,
text = sprintf("Torna domani per un nuovo buongiornissimo!"))
}
}
buongiornissimo_handler <- CommandHandler("buongiornissimo", buongiornissimo)
updaterissimo <- Updater(token = bg_token) %>%
add(buongiornissimo_handler)
# Execute to start the bot
updaterissimo$start_polling()What happens it that the bot builds a dataset which has an entry for every open chat, and fills it with a 0 or 1 depending on whether that day it’s already been asked to deliver a buongiornissimo. Depending on the status of the variable and the day one either gets an image or a message declining the request. The bot also addresses the user by name.
The first problem here is that this is just marginally better than having to send an image every morning - you have to send a message whose consequence is that an image gets sent. My problem was that I kept forgetting sending the image, and I just as well kept forgetting sending the message.
The second, arguably bigger, problem is that the bot stops running whenever my laptop goes in standby. This can be solved by putting it somewhere on a server and have it run there - I googled around and this seemed easier to do with Python or Ruby than R, so I tried my hand with Python.
I’m not really sure how everything is called in Python, and I don’t have much experience progrmaming in it, so the code is quite suboptimal. Anyway, here it is:
# General settings
import telegram
import datetime
import github
import random
import time
g = github.Github("username", "password")
repo = g.get_user().get_repo("repository")
images = list(map(lambda x : x.path, repo.get_contents("repository_with_pictures")))
urls = list(map(lambda x: "beginning of the URL" + x, images))
names = list(map(lambda x : x[16:100], [x for x in images if not(x.endswith("Store"))]))
image_dict = dict(zip(names, urls))
bot = telegram.Bot(token="bot token")
from telegram.ext import Updater
updater = Updater(token="bot token,
use_context=True)
dispatcher = updater.dispatcher
# Start command
def start(update, context):
chat_id = update.effective_chat.id
context.bot.send_message(chat_id,
text = "Botgiornissimo è stato attivato. (v0.3)")
while 1:
current_time=datetime.datetime.now()
hour = current_time.hour
weekday = current_time.weekday()
if (weekday == 0) :
dict_subset = {key: value for key, value in image_dict.items() if (key.startswith("LU") | key.startswith("X") | key.startswith("IMG"))}
elif (weekday == 1) :
dict_subset = {key: value for key, value in image_dict.items() if (key.startswith("MA") | key.startswith("X") | key.startswith("IMG"))}
elif (weekday == 2) :
dict_subset = {key: value for key, value in image_dict.items() if (key.startswith("ME") | key.startswith("X") | key.startswith("IMG"))}
elif (weekday == 3) :
dict_subset = {key: value for key, value in image_dict.items() if (key.startswith("GI") | key.startswith("X") | key.startswith("IMG"))}
elif (weekday == 4) :
dict_subset = {key: value for key, value in image_dict.items() if (key.startswith("VE") | key.startswith("X") | key.startswith("IMG"))}
elif (weekday== 5) :
dict_subset = {key: value for key, value in image_dict.items() if (key.startswith("SA") | key.startswith("X") | key.startswith("IMG"))}
elif (weekday == 6) :
dict_subset = {key: value for key, value in image_dict.items() if (key.startswith("DO") | key.startswith("X") | key.startswith("IMG"))}
image = random.choice(list(dict_subset.values()))
if((hour == 7)):
bot.send_photo(chat_id=chat_id,
photo=image,
caption = "Buongiornissimo!")
if (weekday == 3):
bot.send_poll(chat_id=chat_id,
question="Smashissimo questa settimana?",
options = ["Venerdì alle 21", "Sabato alle 21", "No", "Altro (specificare)"],
allows_multiple_answers = True)
time.sleep(43000)
from telegram.ext import CommandHandler
start_handler = CommandHandler('start', start)
dispatcher.add_handler(start_handler)
# execute to activate bot
updater.start_polling()This time I had the images hosted on github, so that the bot can access them if it’s hosted online somewhere. Python has dictionaries, so I built one having the names of the pictures as keys and their URL as values. From the keys one understands if an image is generic or relates to a particular day of the week, and filter the URLs accordingly. Just as before, a picture is then selected as random.
The user does not need to write /buongiornissimo, they just have to start the bot. The bot keeps checking for the time, and at 7 AM sends a buongiornissimo picture and goes to sleep for something less than 12 hours - just to be safe -, then wakes up, waits until 7, sends the pictures and goes back to sleep. On Wednesdays it also sends a poll with options for weekend planning.
The first problem is thus solved, but it generates a lot of other problems in turn. The bot is perennemente stuck in the while loop or sleeping, so that it can apparently only serve one chat at a time, and for the same reason cannot accept any other command. This isn’t really functional and needs to be changed.
As far as the second problem is concerned, I just found a software that keeps the laptop from going in standby. This is probably going to cause the death of my already elderly machine, but I hope it can survive until Black Friday and that I can get a Raspberry Pi by then… let’s see.