Bdtjtk

I am trying to dynamically clean up some column names for a large number of tables and I get the above error.

I have a gut feeling that I should be using quo but I have no idea on how to do that.

Any ideas?

The apply_alias applies a set of business rules to clean the names.

apply_alias <-  function(l){

  which(l=="Geography")

  l[which(l=="Geography")]  <- "GEO"



  toupper(l)

}

The cleanup_column_names_tbl applies the alias_function to a list of table

cleanup_column_names_tbl <- function(PID){

  for(p in PID){

    names(get(paste0("tbl_",p))) <- apply_alias(names(get(paste0("tbl_",p))))

  }

}



cleanup_column_names_tbl("14100287")

When I try to run it i get the following error message:

> cleanup_column_names_tbl("14100287")

Error in names(get(paste0("tbl_", p))) <- apply_alias(names(get(paste0("tbl_",  : 

  target of assignment expands to non-language object

Sample data:

> dput(tbl_14100287[1,])

structure(list(V1 = 0L, REF_DATE = "1976-01", GEO = "Canada", 

    DGUID = "2016A000011124", `Labour force characteristics` = "Population", 

    Sex = "Both sexes", `Age group` = "15 years and over", Statistics = "Estimate", 

    `Data type` = "Seasonally adjusted", UOM = "Persons", UOM_ID = 249L, 

    SCALAR_FACTOR = "thousands", SCALAR_ID = 3L, VECTOR = "v2062809", 

    COORDINATE = "1.1.1.1.1.1", VALUE = 16852.4, STATUS = "", 

    SYMBOL = NA, TERMINATED = NA, DECIMALS = 1L), class = c("data.table", 

"data.frame"), row.names = c(NA, -1L), .internal.selfref = <pointer: 0x000002123cf21ef0>)

You cannot assign a value to get since there is no function get<-. The right way of doing it would be something like the following.

apply_alias <-  function(l){

  l[which(l == "Geography")]  <- "GEO"

  toupper(l)

}



cleanup_column_names_tbl <- function(PID, envir = .GlobalEnv){

  pid_full <- paste0("tbl_", PID)

  res <- lapply(pid_full, function(p){

    nms <- apply_alias(names(get(p)))

    DF <- get(p)

    names(DF) <- nms

    DF

  })

  names(res) <- pid_full

  list2env(res, envir = envir)

  invisible(NULL)

}





cleanup_column_names_tbl("14100287")



names(tbl_14100287)

# [1] "V1"                           "REF_DATE"                    

# [3] "GEO"                          "DGUID"                       

# [5] "LABOUR FORCE CHARACTERISTICS" "SEX"                         

# [7] "AGE GROUP"                    "STATISTICS"                  

# [9] "DATA TYPE"                    "UOM"                         

#[11] "UOM_ID"                       "SCALAR_FACTOR"               

#[13] "SCALAR_ID"                    "VECTOR"                      

#[15] "COORDINATE"                   "VALUE"                       

#[17] "STATUS"                       "SYMBOL"                      

#[19] "TERMINATED"                   "DECIMALS"  

My solution:

Creating and expression and evaluating it. It is short but somehow does not feel like it is the proper way of doing things since I am breaking away from the functional paradigm of R.

cleanup_column_names_tbl <- function(PID){

  for(p in PID){

    expr1 <- paste0("names(", paste0("tbl_",p), ") <- apply_alias(names(", paste0("tbl_",p),"))")

    eval(rlang::parse_expr(expr1))

    }

}

Edit:

A slightly different way that:

• avoids using strings


• a little more flexible since it allows the use of strings and symbol

library(rlang)

test_df <- data.frame(a=1:10,b=1:10)

test_df2 <- data.frame(a=1:10,b=1:10)





fix_names <-  function(df){



  x <- ensym(df)

  expr1 <- expr(names(!!x) <- toupper(names(!!x)))

  eval(expr1, envir = parent.env(environment()))

  # expr1

}



fix_names(test_df)

fix_names("test_df2")



names(test_df)

#> [1] "A" "B"

names(test_df2)

#> [1] "A" "B"