R subset custom function with optional wildcard param - r

Question

I want to write an R function with some optional params. It should subset some data by two core params and then I want to have the option to pass additional constraints. Eg.
filter_func <- function(start_datetime, end_datetime, user=*, type=*){
as.data.frame(subset(df, format(df$datetime,"%Y-%m-%d %H:%M:%S") > start_datetime &
format(df$datetime,"%Y-%m-%d %H:%M:%S") < end_datetime) &
df$user == user &
df$type == type)
So... if i pass a param it constrains it to that column on either user or type, but if i don't it uses the wildcard and gets everything in the column?
I've seen examples here that use %in% or grepl() but those seem more aimed at where you have part of a string and then want the rest... like new_york gets both new_york_city and new_york_state... i don't want to get any values that don't exactly match the param!
edit: now with examples
So... ideally go from something like this...
start | end | user | type |
-----------------|------------------|------|------|
2017-01-01 11:00 | 2017-01-01 20:00 | usr1 | typ1 |
2017-01-01 12:00 | 2017-01-01 19:00 | usr2 | typ2 |
2017-01-01 02:00 | 2017-01-01 03:00 | usr2 | typ1 |
2017-03-01 01:00 | 2017-03-01 09:00 | usr1 | typ2 |
2017-04-01 05:00 | 2017-04-01 07:00 | usr3 | typ4 |
2017-05-01 01:00 | 2017-05-01 08:00 | usr2 | typ5 |
use my function filter_func(2017-01-01 00:00, 2017-01-01 23:59)
gets me:
start | end | user | type |
-----------------|------------------|------|------|
2017-01-01 11:00 | 2017-01-01 20:00 | usr1 | typ1 |
2017-01-01 12:00 | 2017-01-01 19:00 | usr2 | typ2 |
2017-01-01 02:00 | 2017-01-01 03:00 | usr2 | typ1 |
but if i add a param filter_func(2017-01-01 00:00, 2017-01-01 23:59, usr2)
start | end | user | type |
-----------------|------------------|------|------|
2017-01-01 12:00 | 2017-01-01 19:00 | usr2 | typ2 |
2017-01-01 02:00 | 2017-01-01 03:00 | usr2 | typ1 |
or even filter_func(2017-01-01 00:00, 2017-01-01 23:59, usr2, typ2)
start | end | user | type |
-----------------|------------------|------|------|
2017-01-01 12:00 | 2017-01-01 19:00 | usr2 | typ2 |

Answer 1

Firstly,
[ is safer for programmatic use than subset.
You don't need format, which turns datetime objects into string; you need as.POSIXct or the like, which parse strings into datetimes. You could do this in the function, but you should do this before the function, as you'll always want your datetimes parsed, and there's no point in doing it repeatedly.
You can update a version of the data.frame internal to the function in several steps, which allows you to use control flow like if. You'll still need to check if the variable exists. Two options:
Use missing, which is built for checking whether function parameters exist.
Supply a default value of NULL and use is.null.
You'll need to pass in quoted strings or parsed datetimes (< operators will try to coerce objects that don't match to the same class).
I added a parameter to pass in the data.frame first, which gives the function broader use, but is not necessary.
Altogether, then,
df <- data.frame(start = c("2017-01-01 11:00", "2017-01-01 12:00", "2017-01-01 02:00",
"2017-03-01 01:00", "2017-04-01 05:00", "2017-05-01 01:00"),
end = c("2017-01-01 20:00", "2017-01-01 19:00", "2017-01-01 03:00",
"2017-03-01 09:00", "2017-04-01 07:00", "2017-05-01 08:00"),
user = c("usr1", "usr2", "usr2", "usr1", "usr3", "usr2"),
type = c( "typ1", "typ2", "typ1", "typ2", "typ4", "typ5"))
# parse in two steps if you like, e.g. df$start <- as.POSIXct(df$start)
df[1:2] <- lapply(df[1:2], as.POSIXct)
filter_func <- function(x, start_time, end_time, usr, typ = NULL){
x <- x[x$start > start_time & x$end < end_time, ]
if (!missing(usr)) {
x <- x[x$user %in% usr, ]
}
if (!is.null(typ)) {
x <- x[x$type %in% typ, ]
}
x
}
and test it:
str(df)
#> 'data.frame': 6 obs. of 4 variables:
#> $ start: POSIXct, format: "2017-01-01 11:00:00" "2017-01-01 12:00:00" ...
#> $ end : POSIXct, format: "2017-01-01 20:00:00" "2017-01-01 19:00:00" ...
#> $ user : Factor w/ 3 levels "usr1","usr2",..: 1 2 2 1 3 2
#> $ type : Factor w/ 4 levels "typ1","typ2",..: 1 2 1 2 3 4
filter_func(df, as.POSIXct('2017-01-01 00:00'), as.POSIXct('2017-01-01 23:59'))
#> start end user type
#> 1 2017-01-01 11:00:00 2017-01-01 20:00:00 usr1 typ1
#> 2 2017-01-01 12:00:00 2017-01-01 19:00:00 usr2 typ2
#> 3 2017-01-01 02:00:00 2017-01-01 03:00:00 usr2 typ1
filter_func(df, '2017-01-01 00:00', '2017-01-01 23:59')
#> start end user type
#> 1 2017-01-01 11:00:00 2017-01-01 20:00:00 usr1 typ1
#> 2 2017-01-01 12:00:00 2017-01-01 19:00:00 usr2 typ2
#> 3 2017-01-01 02:00:00 2017-01-01 03:00:00 usr2 typ1
filter_func(df, '2017-01-01 00:00', '2017-01-01 23:59', 'usr2')
#> start end user type
#> 2 2017-01-01 12:00:00 2017-01-01 19:00:00 usr2 typ2
#> 3 2017-01-01 02:00:00 2017-01-01 03:00:00 usr2 typ1
filter_func(df, '2017-01-01 00:00', '2017-01-01 23:59', 'usr2', 'typ2')
#> start end user type
#> 2 2017-01-01 12:00:00 2017-01-01 19:00:00 usr2 typ2

Answer 2

You need to use grepl() for pattern matching.
filter_func <- function(start_datetime, end_datetime, user_='*', type_='*'){
subset(df, as.POSIXlt(df$start) > as.POSIXlt(start_datetime) &
as.POSIXlt(df$end) < as.POSIXlt(end_datetime) &
grepl(user_, df$user) &
grepl(type_, df$type))
}
filter_func(start='2017-01-01 00:00', end='2017-01-01 23:59')
# start end user type
#1 2017-01-01 11:00 2017-01-01 20:00 usr1 typ1
#2 2017-01-01 12:00 2017-01-01 19:00 usr2 typ2
#3 2017-01-01 02:00 2017-01-01 03:00 usr2 typ1
filter_func(start='2017-01-01 00:00', end='2017-01-01 23:59', user='usr2')
# start end user type
#2 2017-01-01 12:00 2017-01-01 19:00 usr2 typ2
#3 2017-01-01 02:00 2017-01-01 03:00 usr2 typ1
filter_func(start='2017-01-01 00:00', end='2017-01-01 23:59', user='usr2', type='typ2')
# start end user type
#2 2017-01-01 12:00 2017-01-01 19:00 usr2 typ2