My test data:

df = structure(list(field_name = c("fruit1", "fruit2"), field_label = c("Yellow fruits", 
"Green fruits"), value = c("1, Pineapple | 2, Mango | 3, Peach", 
"1, Lemon | 2, Apple")), class = c("tbl_df", "tbl", "data.frame"
), row.names = c(NA, -2L))

I need to separate the value column based on the pipe delimiter and create multiple lines (pivot_longer) but considering that the each value has a number that i would like to append to the field_name

My desired output:

df_out=structure(list(field_name = c("fruit1_1", "fruit1_2", "fruit1_3", 
"fruit2_1", "fruit2_2"), field_label = c("Yellow fruits", "Yellow fruits", 
"Yellow fruits", "Green fruits", "Green fruits"), value = c("Pineapple", 
"Mango", "Peach", "Lemon", "Apple")), class = c("tbl_df", "tbl", 
"data.frame"), row.names = c(NA, -5L))

CodePudding user response：

You can first separate rows using tidyr::separate_rows only on the pipe and then use extract to create two columns - number and fruit, then mutate to add the number to the field_name

library(tidyr)
library(dplyr)
df %>% 
    separate_rows("value", sep = " \\| ") %>% 
    extract(value, into = c("first", "value"), regex = "(\\d ), (.*)") %>% 
    mutate(field_name = paste(field_name, first, sep = "_")) %>% 
    select(-first)

# A tibble: 5 x 3
# field_name field_label   value    
# <chr>      <chr>         <chr>    
#  fruit1_1   Yellow fruits Pineapple
#  fruit1_2   Yellow fruits Mango    
#  fruit1_3   Yellow fruits Peach    
#  fruit2_1   Green fruits  Lemon    
#  fruit2_2   Green fruits  Apple

CodePudding user response：

Here is an alternative approach:

1. Using separate_rows from tidyr package we could separate value
2. Use lead to bring value on the same line
3. with filter we keep all even row numbers
4. paste for field_name
5. tweak with select

library(dplyr)
library(tidyr)

df %>% 
  separate_rows(value) %>% 
  mutate(value1 = lead(value)) %>% 
  filter(row_number() %% 2 == 1) %>% 
  mutate(field_name = paste(field_name, value, sep = "_")) %>% 
  select(-value, value = value1)
  

 field_name field_label   value    
  <chr>      <chr>         <chr>    
1 fruit1_1   Yellow fruits Pineapple
2 fruit1_2   Yellow fruits Mango    
3 fruit1_3   Yellow fruits Peach    
4 fruit2_1   Green fruits  Lemon    
5 fruit2_2   Green fruits  Apple  

CodePudding user response：

Using by.

do.call(rbind, c(by(df, df$field_name, \(x) {
  s <- do.call(rbind, strsplit(el(strsplit(x$value, " \\| ")), ", "))
  r <- Reduce(\(x, y) paste0(x, "_", y), data.frame(x$field_name, s[, 1]))
  setNames(cbind.data.frame(r, x$field_label, s[, 2]), names(x))
}), make.row.names=FALSE))
#   field_name   field_label     value
# 1   fruit1_1 Yellow fruits Pineapple
# 2   fruit1_2 Yellow fruits     Mango
# 3   fruit1_3 Yellow fruits     Peach
# 4   fruit2_1  Green fruits     Lemon
# 5   fruit2_2  Green fruits     Apple

CodePudding user response：

Here is a tidyverse way. Use str_extract_all to get what are not numbers, unlist the result and finally, put the output values in the required formats.

df = structure(list(field_name = c("fruit1", "fruit2"), 
                    field_label = c("Yellow fruits", 
                                    "Green fruits"), 
                    value = c("1, Pineapple | 2, Mango | 3, Peach", 
                              "1, Lemon | 2, Apple")), 
               class = c("tbl_df", "tbl", "data.frame"), 
               row.names = c(NA, -2L))

suppressPackageStartupMessages(library(dplyr))
library(tidyr)

df %>%
  mutate(value = stringr::str_extract_all(value, pattern = "\\D ")) %>%
  unnest(cols = value) %>%
  group_by(field_label) %>%
  mutate(field_name = paste(sub("\\d$", "", field_name), row_number(), sep = "_"),
         value = gsub(", | \\| ", "", value))
#> # A tibble: 5 x 3
#> # Groups:   field_label [2]
#>   field_name field_label   value    
#>   <chr>      <chr>         <chr>    
#> 1 fruit_1    Yellow fruits Pineapple
#> 2 fruit_2    Yellow fruits Mango    
#> 3 fruit_3    Yellow fruits Peach    
#> 4 fruit_1    Green fruits  Lemon    
#> 5 fruit_2    Green fruits  Apple

^{Created on 2022-03-28 by the reprex package (v2.0.1)}