Record names that do not comply to the REDCapR pattern

[AlexBalhuizen]

Hi,
I tried to use REDCapR_write and REDCapR_write_oneshot to batch upload pdfs.
Each file has the record id as its filename. When I use the REDcap playground code or redcapAPI it works, but here it is only one file per time.
The record names are uniquely numbered as: ABC123-X01B-M01 as they are specific for sampling sites, occasions, and vial numbers.

When I run the code it uses ABC123, and if I do "" as "ABC123-X01B-M01", it returns an error.

REDCapR_write
_#Define some constants
uri <- "https://redcapxxx/api/"
token <- "TOKEN" # The simple project -pid
field <- "portion_files_fj"
event <- "sorting_arm_1'" # only for longitudinal events
#Upload a single image file.
record <- "ABC123-X01B-M01"
file_path <- system.file("C://FJpdfs//ABC123-X01B-M01.pdf", package="REDCapR")

It gives me the following error:

Error in REDCapR::redcap_upload_file(file_name = file_path, record = record, : Assertion on 'file_name' failed: Must comply to pattern '^.{1,}$'.

Is it possible to run batch uploading with these record names?
I

[wibeasley]

Thanks for explaining it, @AlexBalhuizen. I'm not sure how REDCapR::redcap_write() is related to this, so for now I'll concentrate on the error message that the file names aren't complying.

I'm stumped because that is a very permissive regex. It's requiring only that the string has at least one character.

REDCapR/R/redcap-upload-file-oneshot.R

Line 102 in 44d4cc2

checkmate::assert_character(file_name , any.missing=FALSE, len=1, pattern="^.{1,}$")

In isolation, that file name passes. Can you double check that is the error message associated with the ABC123... file?

checkmate::assert_character("C://FJpdfs//ABC123-X01B-M01.pdf", any.missing=FALSE, len=1, pattern="^.{1,}$")

If that doesn't solve it, please make sure REDCapR is the latest version and run the documentation example for REDCapR::redcap_upload_file_oneshot().

If that passes, please provide more information about the error with your dataset, such as traceback().

[AlexBalhuizen]

I tried a couple of versions of the record name. It seems there are two problems one is that I use a dash - in the record name. Due to this, I have to use single quotes or double quotes to have R use the full record name, but it seems then that redcapR doesn't work with either single quotes or double quotes to define the record name.

For example:
If I run a quoted "P" as record
record <- "P"
Error in REDCapR::redcap_upload_file(file_name = file_path, record = record, : Assertion on 'file_name' failed: Must comply to pattern '^.{1,}$'.

While redcapr processes everything without quotes or dashes, e.g. with underscore instead of dash. Here, error not found as I did run it without a token.

record <- ABC0123
Error: object 'ABC0123' not found
#Upload a single image file.
record <- ABC0123-X01B-M01
Error: object 'ABC0123' not found
record <- ABC0123_X01B_M01
Error: object 'ABC0123_X01B_M01' not found
record <- "ABC0123_X01B_M01"

If I run checkmate it returns empty.

[wibeasley]

My guess is that we shouldn't worry about record now. The file_path argument is throwing the error.

Can you peek at the value your passing to file_path first? If it's like the code you've included in the first post, there's no file there, so system.file() returns a zero-length character. That's why it's failing the regex (that requires only that the file path have at least one character).

file_path <- system.file("C://FJpdfs//ABC123-X01B-M01.pdf", package="REDCapR")

message("File path : `", file_path, "`")
#> File path : ``

checkmate::assert_file_exists(file_path)
#> Error in eval(ei, envir) : 
#>   Assertion on 'file_path' failed: File does not exist: ''.

That first line (that I copied from your post) is asking for the "C:/...pdf" file to be found in the REDCapR directory. I don't think that's what you mean to do though.

[AlexBalhuizen]

Thanks, yes it works to remove the system.files( and package="REDCapR"). and only keeping the path.
I will then need to figure out how to generate the records from the file names.

[wibeasley]

I'm going to close this, since it no longer seems like a REDCap API issue.

Here's a nudge towards "how to generate the records from the file names"

I like the fs package. This will strip off the directory and the extension from the full path.

file_path <- "C://FJpdfs//ABC123-X01B-M01.pdf"

fs::path_ext_remove(fs::path_file(file_path))
#> "ABC123-X01B-M01"