Infer cols from named order in setorderv() (#6932) (#7861)

* updated changes

* ..

* ..

* ..

* update tests

* refine comment

---------

Co-authored-by: Benjamin Schwendinger <benjaminschwe@gmail.com>

Author: venom1204

NEWS.md

@@ -42,6 +42,8 @@
 
 10. `subset()` method for data.tables supports `drop = TRUE` for consistency to data.frame, [#7859](https://github.com/Rdatatable/data.table/issues/7859). Thanks @MichaelChirico for the report and fix.
 
+11. `setorderv()` now accepts a named vector for the `order` argument. When provided, the names are used to identify the columns, allowing the `cols` argument to be omitted, [#6932](https://github.com/Rdatatable/data.table/issues/6932). Thanks to @MichaelChirico for the suggestion and @venom1204 for the implementation.
+
 ### BUG FIXES
 
 1. `fread()` with `skip=0` and `(header=TRUE|FALSE)` no longer skips the first row when it has fewer fields than subsequent rows, [#7463](https://github.com/Rdatatable/data.table/issues/7463). Thanks @emayerhofer for the report and @ben-schwen for the fix.

R/setkey.R

@@ -256,6 +256,10 @@ setorder = function(x, ..., na.last=FALSE)
 
 setorderv = function(x, cols = colnames(x), order=1L, na.last=FALSE)
 {
+  if (missing(cols) && !is.null(names(order))) {
+    cols = names(order)
+    if (anyDuplicated(cols)) stopf("order argument has named duplicates: %s", brackify(duplicated_values(cols)))
+  }
   if (is.null(cols)) return(x)
   if (!is.data.frame(x)) stopf("x must be a data.frame or data.table")
   na.last = as.logical(na.last)

inst/tests/tests.Rraw

@@ -21939,3 +21939,16 @@ test(2384.3, print(DT, na.print=".", topn=2, col.names="none", row.names=FALSE,
   output=c("   .\n   e\n ---\n   w\n   ."))
 DT = data.table(a=c("x\ny","z"), b=1:2)
 test(2384.4, print(DT, col.names="none", row.names=FALSE, class=FALSE), output=c(" x\\ny 1\n    z 2"))
+
+# setorderv() could take order as named column order mapping, #6932
+DT = data.table(a=c(2,1,1,2), b=c("d","c","b","a"), c=c(1,2,1,2))
+m = c(c=-1L, a=1L)
+test(2385.01, setorderv(copy(DT), order=m), setorderv(copy(DT), cols=c("c", "a"), order=m))
+DT=data.table(x=c(2,1,2), y=c(2,1,1))
+test(2385.02, setorderv(DT, order=c(x=1L,y=-1L)), data.table(x=c(1,2,2), y=c(1,2,1)))
+DT = data.table(x=1:3)
+test(2385.03, setorderv(DT, order=c(y=1L)), error="some columns are not in the data.table")
+test(2385.04, setorderv(DT, order=c(x=1L, x=-1L)), error="order argument has named duplicates")
+test(2385.05, setorderv(DT, order=c(x=2L)), error="Must be +1 or -1")
+DT = data.table(a=c(2,1,2), b=3:1)
+test(2385.06, setorderv(copy(DT), order=c(a=1L)), setorderv(copy(DT), cols="a", order=1L))

man/setorder.Rd

@@ -42,11 +42,13 @@ is missing (ex: \code{setorder(x)}), \code{x} is rearranged based on all
 columns in ascending order by default. To sort by a column in descending order
 prefix the symbol \code{"-"} which means "descending" (\emph{not} "negative", in this context), i.e., \code{setorder(x, a, -b, c)}. The \code{-b} works
 when \code{b} is of type \code{character} as well. }
-\item{cols}{ A character vector of column names of \code{x} by which to order. By default, sorts over all columns; \code{cols = NULL} will return \code{x} untouched. Do not add \code{"-"} here. Use \code{order} argument instead. }
+\item{cols}{ A character vector of column names of \code{x} by which to order. By default, sorts over all columns; \code{cols = NULL} will return \code{x} untouched. Do not add \code{"-"} here. Use \code{order} argument instead. This argument can be omitted if \code{order} is a named vector. }
 \item{order}{ An integer vector with only possible values of \code{1} and
 \code{-1}, corresponding to ascending and descending order. The length of
 \code{order} must be either \code{1} or equal to that of \code{cols}. If
-\code{length(order) == 1}, it is recycled to \code{length(cols)}. }
+\code{length(order) == 1}, it is recycled to \code{length(cols)}. \code{order} 
+can also be a named vector, in which case the names are used as \code{cols} 
+and the \code{cols} argument can be omitted. }
 \item{na.last}{ \code{logical}. If \code{TRUE}, missing values in the data are placed last; if \code{FALSE}, they are placed first; if \code{NA} they are removed.
 \code{na.last=NA} is valid only for \code{x[order(., na.last)]} and related \code{sort_by(x, .)} (\R \ifelse{html}{\out{≥}}{\eqn{\ge}} 4.4.0) and its
 default is \code{TRUE}. \code{setorder} and \code{setorderv} only accept
@@ -65,7 +67,9 @@ Note that \code{-b} also works with columns of type \code{character} unlike
 \code{\link[base]{order}}, which requires \code{-xtfrm(y)} instead (which is slow).
 
 \code{setorderv} in turn accepts a character vector of column names and an
-integer vector of column order separately.
+integer vector of column order separately. Additionally, \code{setorderv} can 
+accept a named integer vector for \code{order}, in which case \code{cols} is 
+inferred from the names and can be omitted.
 
 Note that \code{\link{setkey}} still requires and will always sort only in
 ascending order, and is different from \code{setorder} in that it additionally
@@ -133,6 +137,10 @@ setorder(DT, A, -B)
 
 # same as above, but using setorderv
 setorderv(DT, c("A", "B"), c(1, -1))
+
+# infer cols from named order mapping
+mapping = c(A = 1L, B = -1L)
+setorderv(DT, order = mapping)
 }
 \keyword{ data }