Curly curly passing a column name to mutate or regex_left_join returns error, could not find assignment operator `:=`

Question

I am getting an error in console:

Error :=({ : could not find function ":="

I am using a fuzzyjoin (by David Robinson) and tidyverse packages only. The function is accepted with no syntax errors. On execution the error is thrown at me. What could be the reason?

 df_recode_brand_name <- 
    tibble(
      regex_name  = c("wacoal|tempt", "calvin", "hanky", "victoria"),
      return_name = c("wacoal",       "calvin", "hanky", "victoria"))
  
  df_recode_product_category <- 
    tibble(
      regex_name  = c("lingerie", "pant", "bra", "undies"),
      return_name = c("lingerie", "pant", "bra", "undies"))
  
  
fn_regex_recode <- function(df, df_recode_dic, col_name) {
    df %>% regex_left_join(
      df_recode_dic, 
      by = c({{col_name}} := "regex_name"),
      ignore_case = T ) %>% 
    mutate({{col_name}} := if_else(is.na(return_name), str_to_lower({{col_name}}), return_name)) %>% 
    select(-regex_name, -return_name)  
  }

df_raw2 <-  
    df_raw %>%  
    fn_regex_recode(df_recode_brand_name,brand_name) %>% 
    fn_regex_recode(df_recode_product_category, product_category)

Here is the example data

df_raw <-
tibble::tribble(
                ~brand_name,                                  ~product_category,           ~description,
             "Calvin Klein",              "Women - Lingerie & Shapewear - Bras", "Practice bold, fearl",
                   "Wacoal",                                             "Bras", "Beautiful all over c",
        "Victoria's Secret",                           "Add-2-Cups Push-Up Bra", "The ultimate lift-lo",
                   "Wacoal",     "Women - Lingerie & Shapewear - Sexy Lingerie", "With luscious lace a",
        "Victoria's Secret",                 "Incredible by Victoria Sport Bra", "Tackle high-intensit",
             "Calvin Klein", "Women - Lingerie & Shapewear - Designer Lingerie", "Moderate coverage th",
        "Victoria's Secret",                        "Wicked Unlined Uplift Bra", "A little lift goes a",
        "Victoria's Secret",                     "Crochet Lace Cheekster Panty", "The prettiest croche",
        "Victoria's Secret",                           "Curved-hem Thong Panty",  "Seriously sleek and",
        "Victoria's Secret",                           "Add-2-Cups Push-Up Bra", "The ultimate lift-lo",
        "Victoria's Secret",                             "Perfect Coverage Bra", "Our fullest coverage",
               "US TOPSHOP",                                         "Lingerie", "Revamp your lingerie",
             "Calvin Klein",                                        "Sleepwear", "a modern cotton loun",
                    "AERIE",           "Everyday Loves Undies 7 for $27.50 USD", "Introducing Everyday",
      "b.tempt'd by Wacoal", "Women - Lingerie & Shapewear - Designer Lingerie",  "Sheer and sexy, the",
                   "Wacoal",              "Women - Lingerie & Shapewear - Bras", "Discover the glove-l",
   "Victoria's Secret Pink",                      "Wear Everywhere Push-Up Bra", "An everyday fave wit",
                   "WACOAL",                                          "PANTIES", "A sheer lace panty t",
        "Victoria's Secret",                          "Add-1?-Cups Push-Up Bra", "This push-up gives y",
                   "Wacoal",                                             "Bras", "Sport bra offers gre"
  )

I believe the substitution works right but then you're still left with a `:=` in your `c() ` call, which causes the issue. Maybe if you use a tidyverse function such as `vctrs::vec_c() ` it ll work. Else try Ronak's answer, where you should be able to use {{}} instead of substitute. — moodymudskipper, Oct 29 '20 at 11:19

Ronak Shah · Accepted Answer · 2020-10-29T12:12:32.983

1

Try using setNames in by to pass a named vector.

library(dplyr)
library(fuzzyjoin)
library(rlang)

  fn_regex_recode <- function(df, df_recode_dic, col_name) {
    val <- deparse(substitute(col_name))
    df %>% 
      regex_left_join(
        df_recode_dic, 
        by = setNames('regex_name', val),ignore_case = TRUE) %>%
      mutate({{col_name}} := if_else(is.na(return_name), str_to_lower({{col_name}}), return_name)) %>% 
      select(-regex_name, -return_name) 
  }

edited Oct 29 '20 at 12:12

answered Oct 29 '20 at 10:45

Ronak Shah

377,200
20
156
213

It throws a syntax error Error: unexpected '}' in: " select(-regex_name, -return_name) }" – Jacek Kotowski Oct 29 '20 at 11:47
Can you provide example of `df_raw` ? – Ronak Shah Oct 29 '20 at 11:48
Thanks for help. I have attached the data sample to question. – Jacek Kotowski Oct 29 '20 at 11:56
BTW I see that the problem is only with regex_left_join but not with dplyr::mutate – Jacek Kotowski Oct 29 '20 at 12:02
It works sweet!!! Thanks, lots of health. I need to understand deparse(substitute) and then set names... for cases like that, not mainstream tidyverse functions. – Jacek Kotowski Oct 29 '20 at 12:18
1

`deparse(substitute()` is not robust, it might return a vector of length > 1. `as_label(substitute(arg))` or `as_string(ensym(arg))` do not have this problem. – Lionel Henry Oct 29 '20 at 13:52

Curly curly passing a column name to mutate or regex_left_join returns error, could not find assignment operator `:=`

1 Answers1