CODEWITHSUNDEEP
r
jqb101
jqb101

Reputation: 1

How do you handle errors with twitter API?

I have script I'm using to pull and classify tweets. It uses a CSV where I have a series of usernames and other info. However some of the usernames are inactive and result in an empty data frame. This crashes my script. Any advice on error handling? Basically if the "temp" <-get_all_tweets gets 0 tweets I need to "next". What I have below doesn't work.

  for (i in 1:length(usernames))
      {
      #get tweets and subset to text, ID, creation  (may drop ID)
      rm(temp)
      temp<- get_all_tweets(users = usernames[i], start_tweets = start_date[i], end_tweets = end_date[i], bearer_token = get_bearer(),file="test",data_path="Tweets",is_retweet=FALSE, is_reply=FALSE,n = 15000) %>% subset(select=c("text","id","created_at"))
      #error control
      if (!exists("temp")){next}

rest of script works

Upvotes: 0

Views: 139

Answers (1)

r2evans
r2evans

Reputation: 160607

This is a good example of error-handling using try or tryCatch. (There are other methods for working with signals, but they get a bit ... circuitous :-)

Try this:

for (i in seq_along(usernames)) {
  temp <- tryCatch(
    get_all_tweets(
      users = usernames[i], start_tweets = start_date[i],
      end_tweets = end_date[i], bearer_token = get_bearer(),
      file = "test", data_path = "Tweets", is_retweet = FALSE,
      is_reply = FALSE, n = 15000),
    error = function(e) e)
  if (inherits(temp, "error")) {
    warning("error retrieving data for ", sQuote(usernames[i], FALSE), ": ",
            conditionMessage(temp), call. = FALSE)
    # perhaps something more here?
  } else {
    temp <- subset(temp, select = c("text", "id", "created_at"))
    # rest of your code here
  }
}

This code has a simple if statement branching on success of the call to get_all_tweets. In this case, I choose to do nothing more if an error occurs. There are times when you may want to do something else (e.g., call again with different args, or use default-data), then one might do something like this instead:

for (i in seq_along(usernames)) {
  temp <- tryCatch(
    get_all_tweets(
      users = usernames[i], start_tweets = start_date[i],
      end_tweets = end_date[i], bearer_token = get_bearer(),
      file = "test", data_path = "Tweets", is_retweet = FALSE,
      is_reply = FALSE, n = 15000),
    error = function(e) e)
  if (inherits(temp, "error")) {
    warning("error retrieving data for ", sQuote(usernames[i], FALSE), ": ",
            conditionMessage(temp), call. = FALSE)
    # assign to temp from some other method such that
    # the rest of your code works
  } else {
    temp <- subset(temp, select = c("text", "id", "created_at"))
  }
  # rest of your code here
}

Upvotes: 1

Related Questions