Convert Specific values in a column to UNIX timestamp-CodePudding

I have a column in my dataframe which is a mix of some dates and string values. I want to specifically choose the dates and convert into a UNIX timestamp and leave the string values as such. How can this be accomplished ?

Sample data

|column1|
---------
|2020-12-21 00:00:00|
|test1|
|test2|
|test3|
|2021-12-21 00:00:00|

Expected Result

|Column1|
---------------
|1608508800|
|test1|
|test2|
|test3|
|1608508800|

CodePudding user response：

x = read.table(text = 'column1
2020-12-21 00:00:00
test1
test2
test3
2021-12-21 00:00:00', sep = ";", header = T)

uts = as.numeric(as.POSIXct(x$column1, format = "%Y-%m-%d %H:%M:%S", tz = "UTC"))
uts_i = which(!is.na(uts))
x$column1[uts_i] = uts[uts_i]
x
#      column1
# 1 1608508800
# 2      test1
# 3      test2
# 4      test3
# 5 1640044800

Or with dplyr:

x %>%
  mutate(
    uts = as.numeric(as.POSIXct(x$column1, format = "%Y-%m-%d %H:%M:%S", tz = "UTC")),
    column1 = coalesce(as.character(uts), column1)
  ) %>%
  select(-uts)
#      column1
# 1 1608508800
# 2      test1
# 3      test2
# 4      test3
# 5 1640044800