Removing dates ( in any format) form a text column

Question

Hope everyone is well. In my dataset there is column including free texts. My goal is to remove all dates in any format form the text. this is a snapshot of the data

df <- data.frame(
  text=c('tommorow is 2022 11 03',"I married on 2020-01-01",
         'why not going there on 2023/01/14','2023 08 01 will be great'))
df %>% select(text)

                               text
1            tommorow is 2022 11 03
2           I married on 2020-01-01
3 why not going there on 2023/01/14
4          2023 08 01 will be great

The outcome should look like

               text
1            tommorow is 
2            I married on 
3            why not going there on 
4            will be great

Thank you!

Answer 1

Best approach would perhaps be to have a sensitive regex pattern:

df <- data.frame(
  text=c('tommorow is 2022 11 03',"I married on 2020-01-01",
         'why not going there on 2023/01/14','2023 08 01 will be great'))

library(tidyverse)

df |>
  mutate(left_text = str_trim(str_remove(text, "\\d{1,4}\\D\\d{1,2}\\D\\d{1,4}")))

#>                                text              left_text
#> 1            tommorow is 2022 11 03            tommorow is
#> 2           I married on 2020-01-01           I married on
#> 3 why not going there on 2023/01/14 why not going there on
#> 4          2023 08 01 will be great          will be great

This will match dates by:

• \\d{1,4} = starting with either month (1-2 numeric characters), day (1-2 characters) or year (2-4 characters); followed by
• \\D = anything that's not a number, ie the separator; followed by
• \\d{1,2} = day or month (1-2 chars); followed by
• \\D again; ending with
• \\d{1,4} = day or year (1-2 or 2-4 chars)

The challenge is balancing sensitivity with specificity. This should not take out numbers which are clearly not dates, but might miss out:

• dates with no year
• dates with no separators
• dates with double spaces between parts

But hopefully should catch every sensible date in your text column!

Further date detection examples:

library(tidyverse)

df <- data.frame(
  text = c(
    'tommorow is 2022 11 03',
    "I married on 2020-01-01",
    'why not going there on 2023/01/14',
    '2023 08 01 will be great',
    'A trickier example: January 05,2020',
    'or try Oct 2010',
    'dec 21/22 is another date'
  )
)


df |>
  mutate(left_text = str_remove(text, "\\d{1,4}\\D\\d{1,2}\\D\\d{1,4}") |> 
           str_remove(regex(paste0("(", paste(month.name, collapse = "|"),
                                   ")(\\D+\\d{1,2})?\\D+\\d{1,4}"),
                            ignore_case = TRUE)) |> 
           str_remove(regex(paste0("(", paste(month.abb, collapse = "|"),
                                   ")(\\D+\\d{1,2})?\\D+\\d{1,4}"),
                            ignore_case = TRUE)) |> 
           str_trim())

#>                                  text              left_text
#> 1              tommorow is 2022 11 03            tommorow is
#> 2             I married on 2020-01-01           I married on
#> 3   why not going there on 2023/01/14 why not going there on
#> 4            2023 08 01 will be great          will be great
#> 5 A trickier example: January 05,2020    A trickier example:
#> 6                     or try Oct 2010                 or try
#> 7           dec 21/22 is another date        is another date

Edit 2 - doing replace (with temporary placeholders)

df |>
  mutate(left_text = str_replace(text, "\\d{1,4}\\D\\d{1,2}\\D\\d{1,4}", "REP_DATE") |> 
           str_replace(regex(paste0("(", paste(month.name, collapse = "|"),
                                   ")(\\D+\\d{1,2})?\\D+\\d{1,4}"),
                            ignore_case = TRUE), "REP_DATE") |> 
           str_replace(regex(paste0("(", paste(month.abb, collapse = "|"),
                                   ")(\\D+\\d{1,2})?\\D+\\d{1,4}"),
                            ignore_case = TRUE), "REP_DATE") |> 
           str_replace("REP_DATE", "25th October 2022"))