rstats-spam.R

library(rtweet)
library(tidyverse)

auth_setup_default()
json <- search_tweets("#rstats", n = 5000, include_rts = FALSE, parse = FALSE)[1:5]

tweets <- tibble(json = json) %>% 
  unnest_wider(json) %>% 
  select(statuses) %>% 
  unnest_longer(statuses) %>% 
  unpack(statuses) %>% 
  unpack(entities)


bad_tags <- c("5G", "Serverless", "IoT", "IIoT", "AI", "100DaysOfCode", "DEVCommunity", "FemTech")

simple <- tweets %>% 
  mutate(id = id_str, user_id = user$id_str, screen_name = user$screen_name, full_text, hashtags = map(hashtags, "text"), .keep = "none") %>% 
  mutate(n_bad = map_int(hashtags, \(tags) sum(tags %in% bad_tags)))
  
existing <- rtweet:::TWIT_get(NULL, "/1.1/mutes/users/ids", params = list(stringify_ids = TRUE))

bad <- simple %>% 
  filter(n_bad > 2) %>% 
  filter(!user_id %in% existing$ids) %>% 
  count(screen_name, user_id, sort = TRUE)
bad
  
walk(bad$user_id, function(id) {
  message("Muting ", id)
  post_mute(id)
})