Count Backwards in String until pattern R - r

Question

I'm trying to extract UPCs from item descriptions. There is a varying number of /'s in the front of the description, but the UPC is always right before the last /, so I was using a count of characters, however, there is a variable number of characters at the end based on pack size. In the replication, you can see on the first row what this is supposed to look like at the end, but the second row has dropped the first digit of the UPC and picked up the /. Looking for a way to do this inline with DPLYR. My original code is under the replication.
test <- structure(list(Month = structure(c(17987, 17987), class = "Date"),store_id = c("7005", "7005"), UPC = c("000004150860081","00001200050404/"), `Item Description` = c("ACQUA PANNA SPRING WATER/EACH/000004150860081/1","AQUAFINA 24PK/24PK/000001200050404/24"), `Cals Item Description` = c(NA_character_,NA_character_), `Sub-Category` = c(NA_character_, NA_character_), Category = c(NA_character_, NA_character_), Department = c(NA_character_,NA_character_), `Sales Dollars` = c(17.43, 131.78), Units = c(7,528), Cost = c(8.4, 112.2), `Gross Margin` = c(9.03, 19.58), `Gross Margin %` = c(0.5181, 0.1486)), row.names = c(NA,-2L), class = c("tbl_df", "tbl", "data.frame"))
foo <- list.files(pattern = "*.csv", full.names = T) %>%
map_df(~read_csv(.)) %>%
mutate(date = lubridate::mdy(str_sub(textbox43, start = -10))) %>%
mutate(store_id = str_sub(textbox6, start = 1, end = 4)) %>%
mutate(item_desc = textbox57) %>%
filter(!is.na(item_desc), item_desc != "") %>%
mutate(dollars = textbox58,
units = textbox59,
cost = textbox61,
gm = textbox66,
gm_pct = textbox67) %>%
mutate(UPC = str_sub(item_desc, start = -17, end = -3))

Answer 1

Is this what you want?
sub("^.*/([^/]+)/[^/]*$",
"\\1",
test$`Item Description`)
Returns:
[1] "000004150860081" "000001200050404"
Edit: You were asking for dplyr style:
test %>%
mutate(item_id = sub("^.*/([^/]+)/[^/]*$",
"\\1",
test$`Item Description`))