Proper way to convert string to expression for dplyr in R?

Question

I'm trying to create dynamic SQL pulls in R. I'm using dplyr to create the pulls. However, I haven't found a way to properly convert a string to expression with dplyr.

For example, the following code will work.

  data("CO2")
  
  CO2_mod <- CO2 %>%
    mutate(new_c = trimws(paste0(Plant, Type)))

However, when I try any of these below none of them will work. Does anyone know how to convert this string to something I can use in a dplyr pull.

  command <- 'mutate(new_c = trimws(paste0(Plant, Type)))'
  expr <- parse_expr(command)
  expr <- str2lang(command)
  expr <- str2expression(command)

  CO2_mod <- CO2 %>%
    expr #Tried this too for all combinations as well eval(expr)

Firstly

Answer 1

Define your own piping operator:

"%>e%" <- function(lhs, rhs) {
    do.call("%>%", list(lhs, str2lang(rhs)))
}

command <- 'mutate(new_c = trimws(paste0(Plant, Type)))'
CO2 %>e% command %>% head

giving

  Plant   Type  Treatment conc uptake     new_c
1   Qn1 Quebec nonchilled   95   16.0 Qn1Quebec
2   Qn1 Quebec nonchilled  175   30.4 Qn1Quebec
3   Qn1 Quebec nonchilled  250   34.8 Qn1Quebec
4   Qn1 Quebec nonchilled  350   37.2 Qn1Quebec
5   Qn1 Quebec nonchilled  500   35.3 Qn1Quebec
6   Qn1 Quebec nonchilled  675   39.2 Qn1Quebec

Visa

Answer 2

this seems to be what you are looking for:

library(dplyr)

df <- data.frame(
  Plant = c("Rose", "Tulip", "Daisy"),
  Type = c("Flower", "Flower", "Flower")
)

command <- 'df <- mutate(df, new_c = trimws(paste0(Plant, Type)))'

expr <- parse(text = command)

eval(expr)

print(df)

  Plant   Type       new_c
1  Rose Flower  RoseFlower
2 Tulip Flower TulipFlower
3 Daisy Flower DaisyFlower

The steps are to include the data frame in the string, then parse the string as an expression using the parse() function and then evaluate the expression using the eval() function.

When the data frame is printed to the console, you will find that it the mutate operation has been performed on it. I hope this helps!

Embody

Answer 3

If the object returned by str2lang or by str2expression are of class "call" or "expression" you need to evaluate it.

library(tidyverse)

data(CO2)

command <- 'mutate(CO2, new_c = trimws(paste0(Plant, Type)))'
expr <- str2lang(command)
class(expr)
#> [1] "call"

CO2_mod1 <- eval(expr)
head(CO2_mod1)
#>   Plant   Type  Treatment conc uptake     new_c
#> 1   Qn1 Quebec nonchilled   95   16.0 Qn1Quebec
#> 2   Qn1 Quebec nonchilled  175   30.4 Qn1Quebec
#> 3   Qn1 Quebec nonchilled  250   34.8 Qn1Quebec
#> 4   Qn1 Quebec nonchilled  350   37.2 Qn1Quebec
#> 5   Qn1 Quebec nonchilled  500   35.3 Qn1Quebec
#> 6   Qn1 Quebec nonchilled  675   39.2 Qn1Quebec

rlang::eval_tidy(expr) |> head()
#>   Plant   Type  Treatment conc uptake     new_c
#> 1   Qn1 Quebec nonchilled   95   16.0 Qn1Quebec
#> 2   Qn1 Quebec nonchilled  175   30.4 Qn1Quebec
#> 3   Qn1 Quebec nonchilled  250   34.8 Qn1Quebec
#> 4   Qn1 Quebec nonchilled  350   37.2 Qn1Quebec
#> 5   Qn1 Quebec nonchilled  500   35.3 Qn1Quebec
#> 6   Qn1 Quebec nonchilled  675   39.2 Qn1Quebec

^{Created on 2024-08-29 with reprex v2.1.0}

And an object of class "expression".

library(tidyverse)
command <- 'mutate(CO2, new_c = trimws(paste0(Plant, Type)))'

expr <- str2expression(command)
class(expr)
#> [1] "expression"

CO2_mod2 <- eval(expr)
head(CO2_mod2)
#>   Plant   Type  Treatment conc uptake     new_c
#> 1   Qn1 Quebec nonchilled   95   16.0 Qn1Quebec
#> 2   Qn1 Quebec nonchilled  175   30.4 Qn1Quebec
#> 3   Qn1 Quebec nonchilled  250   34.8 Qn1Quebec
#> 4   Qn1 Quebec nonchilled  350   37.2 Qn1Quebec
#> 5   Qn1 Quebec nonchilled  500   35.3 Qn1Quebec
#> 6   Qn1 Quebec nonchilled  675   39.2 Qn1Quebec

^{Created on 2024-08-29 with reprex v2.1.0}

Coppice