指定使用 map_dfr 时哪些列相同

Question

I have two folders each with hundreds of CSVs and I want to merge them all in one data frame. I have used the following:

tbl <-
  list.files(path = c("./reports_0", "./reports_1"),
             pattern = "*.csv", 
             full.names = T) %>%
  map_dfr(~read_csv(., col_types = cols(.default = "c")))

Now I realized that some of those CSVs have their column name as Firmware Version and some as Firmware version (upper and lowercase).

I would like to specify that those are the same and can be combined in one called Firmware Version.

the

by =

does not work and I could not find a solution.

Hope there is someone that can help, thanks!

EDIT
My workaround is:

tbl <- tbl %>% 
  unite(`Firmware Version`, `Firmware version`, na.rm = T) %>% 
  mutate(`Firmware Version` = replace(`Firmware Version`, `Firmware Version`=="", NA_character_))

However, I still wonder whether there is a nicer, more straightforward way.

Answer 1

you could use janitor::make_clean_names() to convert columnnames to the same format (for example camelCase), and then rowbind.

for example:

library(data.table)
library(janitor)
ftr <- list.files(path = c("./reports_0", "./reports_1"), 
   pattern = ".*\\.csv$", 
   names = TRUE)

DT <- rbindlist(
  lapply(ftr, function(x) {
    tempDT <- fread(x)
    setnames(tempDT, names(tempDT), janitor::make_clean_names(names(tempDT)))
    return(tempDT)
  }), use.names = TRUE, fill = TRUE)

proof of concept

convert names to snake_case

> janitor::make_clean_names("Firmware Version")
[1] "firmware_version"
> janitor::make_clean_names("Firmware version")
[1] "firmware_version"