pass column name to dplyr filter function from within a function

Question

I'm writing a function to add some extra rows to a data frame i am building.

I've read many questions and answer's from previous OPs. But all answers, tips and tricks i found there do not work for me.

For this question i have the following test data frame:

tst <- data.frame("col 1" = c("a","a", "c"), "keyword test" = c("What", "Why", "how"), check.names = F)
> tst
  col 1 keyword test
1     a         What
2     a          Why
3     c          how

As you can see i have spaces in the data frame which i cannot remove, since the next tool is expecting spaces in the column names (DON'T ask me why!).

Now i want for example to filter all rows starting with "how" and replace "how" with "no_idea". This happens inside a temp DF. So that later on i can add the row "c" "no_idea" to the original data frame.

The function i wrote for this looks like this:

add_keyword <- function(df, filterColumn, filterValue,replacement){
  library(plyr)
  library(dplyr)
  temp_df <- dplyr::filter_(df, filterColumn == filterValue)
  temp_df$`Target keyword` <- gsub(as.character(filterValue), as.character(replacement), temp_df$`Target keyword`)
  df_out <- rbind(df, temp_df)
  return(df_out)
}

tst2 <- add_keyword(tst, "keyword test", "how", "no_idea")

Of course if I run the lines inside the function outside the function it works perfect.

The result i would like to have inside tst2

> tst2
  col 1 keyword test
1     a         What
2     a         Why
3     c         how
4     c         no_idea 

Answer 1

We can do this with interp from lazyeval:

library(dplyr)
library(lazyeval)

add_keyword <- function(df, filterColumn, filterValue,replacement){
    temp_df <- df %>%
        filter_(interp(~ var == fval, var = as.name(filterColumn), fval = filterValue))

    temp_df[[filterColumn]] <- gsub(filterValue, replacement = replacement, temp_df[[filterColumn]])
    rbind(df, temp_df)
}
add_keyword(tst, "keyword test", "how", "no_idea")
#   col 1 keyword test
# 1     a         What
# 2     a          Why
# 3     c          how
# 4     c      no_idea

If we do not want to create additional rows, we can also try:

add_keyword <- function(df, filterColumn, filterValue, replacement){
    df <- df %>%
        mutate_(
            .dots = setNames(
                list(
                    interp(~ ifelse(startsWith(as.character(var), fval), rval, as.character(var)),
                           var = as.name(filterColumn), fval = filterValue, rval = replacement)),
                filterColumn
            )
        )
    df
}
add_keyword(tst, "keyword test", "how", "no_idea")


#   col 1 keyword test
# 1     a         What
# 2     a          Why
# 3     c      no_idea