parse_date_time function

Parse date-times from Elasticsearch records

Given a data.table with date-time strings, this function converts those dates-times to type POSIXct with the appropriate time zone. Assumption is that dates are of the form "2016-07-25T22:15:19Z" where T is just a separator and the last letter is a military timezone.

This is a side-effect-free function: it returns a new data.table and the input data.table is unmodified.

parse_date_time(input_df, date_cols, assume_tz = "UTC")

Arguments

• input_df: a data.table with one or more date-time columns you want to convert
• date_cols: Character vector of column names to convert. Columns should have string dates of the form "2016-07-25T22:15:19Z".
• assume_tz: Timezone to convert to if parsing fails. Default is UTC

Examples

# Sample es_search(), chomp_hits(), or chomp_aggs() output:
someDT <- data.table::data.table(id = 1:5
                                 , company = c("Apple", "Apple", "Banana", "Banana", "Cucumber")
                                 , timestamp = c("2015-03-14T09:26:53B", "2015-03-14T09:26:54B"
                                                 , "2031-06-28T08:53:07Z", "2031-06-28T08:53:08Z"
                                                 , "2000-01-01"))

# Note that the date field is character right now
str(someDT)

# Let's fix that!
someDT <- parse_date_time(input_df = someDT
                          , date_cols = "timestamp"
                          , assume_tz = "UTC")
str(someDT)

References

https://www.timeanddate.com/time/zones/military

https://en.wikipedia.org/wiki/List_of_tz_database_time_zones