Delete method

[wibeasley]

A recent post on an old community thread made we realize REDCapR doesn't implement the delete function. I'm kinda surprised no one has requested it.

[joundso]

Very raw suggestion (funnily I just needed it a few days ago):

install.packages("httr")
 
remove_ids <- c(9, 10, 11)
 
url <- Sys.getenv("REDCAP_API_URL")
formData <-
  list(
    token = Sys.getenv("REDCAP_API_KEY"),
    action = "delete",
    content = "record"
    # ,'records[0]' = '2HA9FF0A',
    # 'records[1]' = '2J9YPFTE',
    # 'records[2]' = '2J49KFRP'
  )
i <- 0
for (id in remove_ids) {
  formData[[paste0("records[", i, "]")]] <- id
  i <- i + 1
}
 
response <- httr::POST(url, body = formData, encode = "form")
if (httr::http_error(response)) {
  stop("Error in HTTP request to REDCap.")
} else {
  print(httr::content(x = response, as = "text"))
}

[wibeasley]

@joundso, thanks for thinking & working on this, I just noticed #373 in GitHub Actions that you're working on this too. I should have refreshed this report before I pushed what I was working on last night.

It looks like we're about in the same place too: the single-arm deletion is working, the parameter for the multiple-arm project needs to be added. Do you want to work on that? If so, I'll step away and let you do your thing. I'll use your code, and feel free to take anything from my dev branch that you find helpful.

Two things to note:

1. you may prefer the way I add records to the post_body without declaring a loop.
2. I'm thinking that a error should be thrown (with stop()) if it's unsuccessful. But I'm open to ideas. I think overall REDCapR needs to do a better job throwing & communicating errors. I thought I 'd try errors for this function, and see how if works well with other people.

[joundso]

Hi @wibeasley - wow - great work! Sorry for interrupting/overlapping.
I really like the way you wrote the functions!

1. Absolutely right. My loop is bad performing:

## Cleanup the backend in RStudio:
cat("\014") # Clears the console (imitates CTR + L)
rm(list = ls()) # Clears the Global Environment/variables/data
invisible(gc()) # Garbage collector/Clear unused RAM

records_to_delete <- sample(x = 1:1e10,
                            size = 1e3,
                            replace = FALSE)

loop_assignment <- function(vec) {
  i <- 0
  res <- list()
  for (id in vec) {
    res[[paste0("records[", i, "]")]] <- id
    i <- i + 1
  }
  return(res)
}

name_assignment <- function(vec) {
  return(stats::setNames(vec,
                         sprintf("records[%i]", seq_along(vec) - 1)))
}


microbenchmark::microbenchmark(loop_assignment(records_to_delete),
                               name_assignment(records_to_delete))
#> Unit: microseconds
#>                                expr     min       lq     mean   median      uq
#>  loop_assignment(records_to_delete) 10574.7 10855.15 11702.27 11157.45 11929.3
#>  name_assignment(records_to_delete)   247.3   269.55   318.45   292.50   313.7
#>      max neval
#>  18396.4   100
#>   2613.0   100

^{Created on 2021-11-24 by the reprex package (v2.0.1)}

2. I would like to see more errors being thrown. So I can only endorse the suggestion to include stop().

Unfortunately I won't be able to contribute much meaningful code in the coming days - but I will continue to follow your developments with interest! 👀

[wibeasley]

Cool. Thanks for the feedback, I'll take it from here. I saw things in your code that I like, and I'll try to do it so you show up in the commits.

And I wouldn't worry too much about the loop/vector performance. Even though you say my way is 1sec faster, I'm guessing this delay will be dwarfed by the network latency & delete operation on MySQL.