ARROW-15800 [R] bindings for lubridate::as_date() and lubridate::as_datetime() - abandoned

r/NEWS.md

@@ -23,6 +23,7 @@
 * `lubridate`:
  * component extraction functions: `tz()` (timezone), `semester()` (semester), `dst()` (daylight savings time indicator), `date()` (extract date), `epiyear()` (epiyear), improvements to `month()`, which now works with integer inputs.
  * `make_date()` & `make_datetime()` + `ISOdatetime()` & `ISOdate()` to create date-times from numeric representations. 
+ * `as_date()` and `as_datetime()`
 * date-time functionality:
  * `difftime` and `as.difftime()` 
  * `as.Date()` to convert to date

r/R/dplyr-funcs-datetime.R

@@ -191,6 +191,70 @@ register_bindings_datetime <- function() {
 }
 
 register_bindings_duration <- function() {
+ register_binding("difftime", function(time1,
+ time2,
+ tz,
+ units = "secs") {
+ if (units != "secs") {
+ abort("`difftime()` with units other than `secs` not supported in Arrow")
+ }
+
+ if (!missing(tz)) {
+ warn("`tz` argument is not supported in Arrow, so it will be ignored")
+ }
+
+ # cast to timestamp if time1 and time2 are not dates or timestamp expressions
+ # (the subtraction of which would output a `duration`)
+ if (!call_binding("is.instant", time1)) {
+ time1 <- build_expr("cast", time1, options = cast_options(to_type = timestamp()))
+ }
+
+ if (!call_binding("is.instant", time2)) {
+ time2 <- build_expr("cast", time2, options = cast_options(to_type = timestamp()))
+ }
+
+ # we need to go build the subtract expression instead of `time1 - time2` to
+ # prevent complaints when we try to subtract an R object from an Expression
+ subtract_output <- build_expr("-", time1, time2)
+ build_expr("cast", subtract_output, options = cast_options(to_type = duration("s")))
+ })
+ register_binding("as.difftime", function(x,
+ format = "%X",
+ units = "secs") {
+ # windows doesn't seem to like "%X"
+ if (format == "%X" & tolower(Sys.info()[["sysname"]]) == "windows") {
+ format <- "%H:%M:%S"
+ }
+
+ if (units != "secs") {
+ abort("`as.difftime()` with units other than 'secs' not supported in Arrow")
+ }
+
+ if (call_binding("is.character", x)) {
+ x <- build_expr("strptime", x, options = list(format = format, unit = 0L))
+ # complex casting only due to cast type restrictions: time64 -> int64 -> duration(us)
+ # and then we cast to duration ("s") at the end
+ x <- x$cast(time64("us"))$cast(int64())$cast(duration("us"))
+ }
+
+ # numeric -> duration not supported in Arrow yet so we use int64() as an
+ # intermediate step
+ # TODO revisit if https://issues.apache.org/jira/browse/ARROW-15862 results
+ # in numeric -> duration support
+
+ if (call_binding("is.numeric", x)) {
+ # coerce x to be int64(). it should work for integer-like doubles and fail
+ # for pure doubles
+ # if we abort for all doubles, we risk erroring in cases in which
+ # coercion to int64() would work
+ x <- build_expr("cast", x, options = cast_options(to_type = int64()))
+ }
+
+ build_expr("cast", x, options = cast_options(to_type = duration(unit = "s")))
+ })
+}
+
+register_bindings_datetime_helpers <- function() {
  register_binding("make_datetime", function(year = 1970L,
  month = 1L,
  day = 1L,
@@ -239,66 +303,108 @@ register_bindings_duration <- function() {
  tz = "UTC") {
  call_binding("make_datetime", year, month, day, hour, min, sec, tz)
  })
- register_binding("difftime", function(time1,
- time2,
- tz,
- units = "secs") {
- if (units != "secs") {
- abort("`difftime()` with units other than `secs` not supported in Arrow")
- }
+ register_binding("as.Date", function(x,
+ format = NULL,
+ tryFormats = "%Y-%m-%d",
+ origin = "1970-01-01",
+ tz = "UTC") {
 
- if (!missing(tz)) {
- warn("`tz` argument is not supported in Arrow, so it will be ignored")
+ # the origin argument will be better supported once we implement temporal
+ # arithmetic (https://issues.apache.org/jira/browse/ARROW-14947)
+ # TODO revisit once the above has been sorted
+ if (call_binding("is.numeric", x) & origin != "1970-01-01") {
+ abort("`as.Date()` with an `origin` different than '1970-01-01' is not supported in Arrow")
  }
 
- # cast to timestamp if time1 and time2 are not dates or timestamp expressions
- # (the subtraction of which would output a `duration`)
- if (!call_binding("is.instant", time1)) {
- time1 <- build_expr("cast", time1, options = cast_options(to_type = timestamp(timezone = "UTC")))
+ # this could be improved with tryFormats once strptime returns NA and we
+ # can use coalesce - https://issues.apache.org/jira/browse/ARROW-15659
+ # TODO revisit once https://issues.apache.org/jira/browse/ARROW-15659 is done
+ if (is.null(format) && length(tryFormats) > 1) {
+ abort("`as.Date()` with multiple `tryFormats` is not supported in Arrow")
  }
 
- if (!call_binding("is.instant", time2)) {
- time2 <- build_expr("cast", time2, options = cast_options(to_type = timestamp(timezone = "UTC")))
- }
+ if (call_binding("is.Date", x)) {
+ return(x)
 
- # we need to go build the subtract expression instead of `time1 - time2` to
- # prevent complaints when we try to subtract an R object from an Expression
- subtract_output <- build_expr("-", time1, time2)
- build_expr("cast", subtract_output, options = cast_options(to_type = duration("s")))
+ # cast from POSIXct
+ } else if (call_binding("is.POSIXct", x)) {
+ # base::as.Date() first converts to the desired timezone and then extracts
+ # the date, which is why we need to go through timestamp() first
+ x <- build_expr("cast", x, options = cast_options(to_type = timestamp(timezone = tz)))
+
+ # cast from character
+ } else if (call_binding("is.character", x)) {
+ format <- format %||% tryFormats[[1]]
+ # unit = 0L is the identifier for seconds in valid_time32_units
+ x <- build_expr("strptime", x, options = list(format = format, unit = 0L))
+
+ # cast from numeric
+ } else if (call_binding("is.numeric", x) & !call_binding("is.integer", x)) {
+ # Arrow does not support direct casting from double to date32(), but for
+ # integer-like values we can go via int32()
+ # https://issues.apache.org/jira/browse/ARROW-15798
+ # TODO revisit if arrow decides to support double -> date casting
+ x <- build_expr("cast", x, options = cast_options(to_type = int32()))
+ }
+ build_expr("cast", x, options = cast_options(to_type = date32()))
  })
- register_binding("as.difftime", function(x,
- format = "%X",
- units = "secs") {
- # windows doesn't seem to like "%X"
- if (format == "%X" & tolower(Sys.info()[["sysname"]]) == "windows") {
- format <- "%H:%M:%S"
+ register_binding("as_date", function(x,
+ format = NULL,
+ origin = "1970-01-01",
+ tz = "UTC") {
+ # the origin argument will be better supported once we implement temporal
+ # arithmetic (https://issues.apache.org/jira/browse/ARROW-14947)
+ # TODO revisit once the above has been sorted
+ if (call_binding("is.numeric", x) & origin != "1970-01-01") {
+ abort("`as.Date()` with an `origin` different than '1970-01-01' is not supported in Arrow")
  }
 
- if (units != "secs") {
- abort("`as.difftime()` with units other than 'secs' not supported in Arrow")
+ # assume format is ISO if unspecified (to align with lubridate::as_date)
+ if (is.null(format)) {
+ format <- "%Y-%m-%d"
  }
 
- if (call_binding("is.character", x)) {
- x <- build_expr("strptime", x, options = list(format = format, unit = 0L))
- # complex casting only due to cast type restrictions: time64 -> int64 -> duration(us)
- # and then we cast to duration ("s") at the end
- x <- x$cast(time64("us"))$cast(int64())$cast(duration("us"))
- }
+ if (call_binding("is.Date", x)) {
+ return(x)
 
- # numeric -> duration not supported in Arrow yet so we use int64() as an
- # intermediate step
- # TODO revisit if https://issues.apache.org/jira/browse/ARROW-15862 results
- # in numeric -> duration support
+ # cast from POSIXct
+ } else if (call_binding("is.POSIXct", x)) {
+ # this is where as_date() differs from as.Date()
+ if (!missing(tz)) {
+ x <- build_expr("cast", x, options = cast_options(to_type = timestamp(timezone = tz)))
+ }
+ # POSIXct is of type double -> we need this to prevent going down the
+ # "double" branch
+ x <- x
+
+ # cast from character
+ } else if (call_binding("is.character", x)) {
+ # unit = 0L is the identifier for seconds in valid_time32_units
+ x <- build_expr("strptime", x, options = list(format = format, unit = 0L))
 
+ # cast from numeric
+ } else if (call_binding("is.numeric", x) & !call_binding("is.integer", x)) {
+ # Arrow does not support direct casting from double to date32(), but for
+ # integer-like values we can go via int32()
+ # https://issues.apache.org/jira/browse/ARROW-15798
+ # TODO revisit if arrow decides to support double -> date casting
+ x <- build_expr("cast", x, options = cast_options(to_type = int32()))
+ }
+ build_expr("cast", x, options = cast_options(to_type = date32()))
+ })
+ register_binding("as_datetime", function(x,
+ origin = "1970-01-01",
+ tz = "UTC") {
  if (call_binding("is.numeric", x)) {
- # coerce x to be int64(). it should work for integer-like doubles and fail
- # for pure doubles
- # if we abort for all doubles, we risk erroring in cases in which
- # coercion to int64() would work
+ delta <- call_binding("difftime", origin, "1970-01-01")
+ delta <- build_expr("cast", delta, options = cast_options(to_type = int64()))
  x <- build_expr("cast", x, options = cast_options(to_type = int64()))
+ output <- build_expr("+", x, delta)
+ output <- build_expr("cast", output, options = cast_options(to_type = timestamp()))
+ } else {
+ output <- build_expr("cast", x, options = cast_options(to_type = timestamp()))
  }
-
- build_expr("cast", x, options = cast_options(to_type = duration(unit = "s")))
+ build_expr("assume_timezone", output, options = list(timezone = tz))
  })
 }
 

r/R/dplyr-funcs-type.R

@@ -77,51 +77,6 @@ register_bindings_type_cast <- function() {
  register_binding("as.numeric", function(x) {
  build_expr("cast", x, options = cast_options(to_type = float64()))
  })
- register_binding("as.Date", function(x,
- format = NULL,
- tryFormats = "%Y-%m-%d",
- origin = "1970-01-01",
- tz = "UTC") {
-
- # the origin argument will be better supported once we implement temporal
- # arithmetic (https://issues.apache.org/jira/browse/ARROW-14947)
- # TODO revisit once the above has been sorted
- if (call_binding("is.numeric", x) & origin != "1970-01-01") {
- abort("`as.Date()` with an `origin` different than '1970-01-01' is not supported in Arrow")
- }
-
- # this could be improved with tryFormats once strptime returns NA and we
- # can use coalesce - https://issues.apache.org/jira/browse/ARROW-15659
- # TODO revisit once https://issues.apache.org/jira/browse/ARROW-15659 is done
- if (is.null(format) && length(tryFormats) > 1) {
- abort("`as.Date()` with multiple `tryFormats` is not supported in Arrow")
- }
-
- if (call_binding("is.Date", x)) {
- return(x)
-
- # cast from POSIXct
- } else if (call_binding("is.POSIXct", x)) {
- # base::as.Date() first converts to the desired timezone and then extracts
- # the date, which is why we need to go through timestamp() first
- x <- build_expr("cast", x, options = cast_options(to_type = timestamp(timezone = tz)))
-
- # cast from character
- } else if (call_binding("is.character", x)) {
- format <- format %||% tryFormats[[1]]
- # unit = 0L is the identifier for seconds in valid_time32_units
- x <- build_expr("strptime", x, options = list(format = format, unit = 0L))
-
- # cast from numeric
- } else if (call_binding("is.numeric", x) & !call_binding("is.integer", x)) {
- # Arrow does not support direct casting from double to date32()
- # https://issues.apache.org/jira/browse/ARROW-15798
- # TODO revisit if arrow decides to support double -> date casting
- abort("`as.Date()` with double/float is not supported in Arrow")
- }
- build_expr("cast", x, options = cast_options(to_type = date32()))
- })
-
  register_binding("is", function(object, class2) {
  if (is.string(class2)) {
  switch(class2,

r/R/dplyr-funcs.R

@@ -107,6 +107,7 @@ create_binding_cache <- function() {
  register_bindings_conditional()
  register_bindings_datetime()
  register_bindings_duration()
+ register_bindings_datetime_helpers()
  register_bindings_math()
  register_bindings_string()
  register_bindings_type()