How to replace values under severals conditions using purrr? - r

Question

The post has been edit at Aug 17, 2020 to make the example looks more like my actual data.
The days always come first either with 1 or 2 digits. The months always come second either in full or in part and in French. The years always come third either with 2 or 4 digits.
I'm learning to code with tidyverse packages. I'm trying to replace every elements in a variable by an other string if they match specific conditions. The problem is that I can only do it one condition at the time. I would like to know how to achieve it at severals condition a the time.
Here's a reproductible exemple :
library(tidyverse)
library(magrittr)
tib <- tibble(
ID = 1:6,
Date = c("1-JAN-20", "15-JUILL-20", "30 DEC 2020",
"1-JAN-20", "15-JUILL-20", "30 DEC 2020"),
Comm = c("Should be 2020-01-01", "Should be 2020-06-15", "Should be 2020-12-30",
"Should be 2020-01-01", "Should be 2020-06-15", "Should be 2020-12-30"))
head(tib)
# A tibble: 6 x 3
ID Date Comm
<int> <chr> <chr>
1 1 1-JAN-20 Should be 2020-01-01
2 2 15-JUILL-20 Should be 2020-06-15
3 3 30 DEC 2020 Should be 2020-12-30
4 4 1-JAN-20 Should be 2020-01-01
5 5 15-JUILL-20 Should be 2020-06-15
6 6 30 DEC 2020 Should be 2020-12-30
# Returns the unique values of the character variables execept the "Comm" one. So, it
# returns only one in that case, but my original data have severals ones.
tib %>% select(where(is.character), -Comm) %>% map(~ unique(.x))
$Date
[1] "1-JAN-20" "15-JUILL-20" "30 DEC 2020"
Here we are! The following code works, but I wonder if there's a better way to atcheive it instead of copy/pass the same code line every time and changing it.
tib <- tib %>% mutate(Date = case_when(Date == "1-JAN-20" ~ "2020-01-01",
Date == "15-JUILL-20" ~ "2020-06-15",
Date == "30 DEC 2020" ~ "2020-12-01"))
head(tib)
# A tibble: 6 x 3
ID Date Comm
<int> <chr> <chr>
1 1 2020-01-01 Should be 2020-01-01
2 2 2020-06-15 Should be 2020-06-15
3 3 2020-12-01 Should be 2020-12-30
4 4 2020-01-01 Should be 2020-01-01
5 5 2020-06-15 Should be 2020-06-15
6 6 2020-12-01 Should be 2020-12-30
Since I will have to do this manipulation on other variables, how could I build a function that would accomplish this?
Also, I would like to know if you know some good documentations/tutorials to learn Purrr package?
Thank you and have a good day!

Answer 1

While handling dates/times you should use standard date time functions for manipulation. Don't replace dates one by one using str_replace. Imagine you have 1000's of dates with different years, it is not practically possible to list each one of them. In this case, you can use lubridate::dmy to convert them to date object, for more complicated cases there is lubridate::parse_date_time which can convert variables in different format to dates.
tib %>% dplyr::mutate(new_date = lubridate::dmy(Date))
# ID Date Comm new_date
# <int> <chr> <chr> <date>
#1 1 01-JAN-20 Should be 2020-01-01 2020-01-01
#2 2 15-JUN-20 Should be 2020-06-15 2020-06-15
#3 3 30 DEC 2020 Should be 2020-12-30 2020-12-30
#4 4 01-JAN-20 Should be 2020-01-01 2020-01-01
#5 5 15-JUN-20 Should be 2020-06-15 2020-06-15
#6 6 30 DEC 2020 Should be 2020-12-30 2020-12-30
If you want dates in specific format, you can use the format function on new_date.

Answer 2

Maybe you could try dplyr::case_when:
library(magrittr)
library(purrr)
# A tibble that looks like my data.
tib <- tibble(
ID = 1:6,
Date = c("01-JAN-20", "15-JUN-20", "30 DEC 2020",
"01-JAN-20", "15-JUN-20", "30 DEC 2020"),
Comm = c("Should be 2020-01-01", "Should be 2020-06-15", "Should be 2020-12-30",
"Should be 2020-01-01", "Should be 2020-06-15", "Should be 2020-12-30"))
head(tib)
tib %>% select(where(is.character), -Comm) %>% map(~ unique(.x))
tib <- tib %>% mutate(Date = dplyr::case_when(Date == "01-JAN-20" ~ "2020-01-01",
Date == "15-JUN-20" ~ "2020-06-15",
Date == "30 DEC 2020" ~ "2020-12-01"))
> tib
# A tibble: 6 x 3
ID Date Comm
<int> <chr> <chr>
1 1 2020-01-01 Should be 2020-01-01
2 2 2020-06-15 Should be 2020-06-15
3 3 2020-12-01 Should be 2020-12-30
4 4 2020-01-01 Should be 2020-01-01
5 5 2020-06-15 Should be 2020-06-15
6 6 2020-12-01 Should be 2020-12-30
The best thing to try to do here is to transform your Date column into Date class using the "anytime" package. Although you would have to manually fix your Date column so all years have 4 digits. If years are always in the last place of the date, that can be an easy thing to do.