Delete all tweets older than 30 days unless they have certain hashtags

delete_tweets.R

library(here)
library(rtweet)
library(tidyverse)
library(assertthat)



#---- prep ----

my_dir <- here()

# read in vector of 'keeper' hashtags we will use to filter
data_tags <- read_lines("hashtags_to_keep.txt")

date_cutoff <- Sys.Date() - 30

tweets <- get_timeline(
 "jeremy_data",
 n = 10000
)

# ids of tweets to keep
data_related_ids <- tweets %>% 
 select(status_id, hashtags) %>% 
 filter(!is.na(hashtags)) %>%                    # must have a hashtag
 unnest(hashtags) %>% 
 mutate(hashtags = str_to_lower(hashtags)) %>% 
 filter(hashtags %in% data_tags) %>%             # must be a data-related hashtag 
 unique() %>% 
 pull(status_id)

ids_to_delete <- tweets %>%
 select(status_id, created_at) %>% 
 filter(as.Date(created_at) < as.Date(date_cutoff)) %>%  # are older than 30 days
 filter(!status_id %in% data_related_ids) %>%            # are not data-related
 pull(status_id)

total <- length(ids_to_delete)

# confirm that tweets to delete really are older than 30 days
assert_that(
 tweets %>% filter(status_id == sample(ids_to_delete, 1)) %>% pull(created_at) %>% 
  as.Date() < date_cutoff
)



#---- work ----

for (i in seq_along(ids_to_delete)) {
 
 skip_to_next <- FALSE
 
 Sys.sleep(.5)
 
 message("... ", i, " of ", total, " ... trying to delete ",  ids_to_delete[i])
 
 tryCatch(
  expr = post_tweet("", destroy_id = ids_to_delete[i]), # non-intuitive post request to twitter api
  error = function(e) {
   message("... Destroy seems to have failed for ", i, ":\n")
   write(x = paste0(ids_to_delete[i], "\n", e),
         file = paste0(my_dir, "/log.txt"),
         append = TRUE)
   
   skip_to_next <<- TRUE
  }
 ) # end tryCatch
 
 if(skip_to_next) next
 
} # end loop



#---- rest ----

closeAllConnections()

cat("Enjoy!")