Allow 1-D arrays in j=list() form while grouping (#6054)

NEWS.md

@@ -30,6 +30,8 @@
 
 7. `fread`'s `fill` argument now also accepts an `integer` in addition to boolean values. `fread` always guesses the number of columns based on reading a sample of rows in the file. When `fill=TRUE`, `fread` stops reading and ignores subsequent rows when this estimate winds up too low, e.g. when the sampled rows happen to exclude some rows that are even wider, [#2727](https://github.com/Rdatatable/data.table/issues/2727) [#2691](https://github.com/Rdatatable/data.table/issues/2691) [#4130](https://github.com/Rdatatable/data.table/issues/4130) [#3436](https://github.com/Rdatatable/data.table/issues/3436). Providing an `integer` as argument for `fill` allows for a manual estimate of the number of columns instead, [#1812](https://github.com/Rdatatable/data.table/issues/1812) [#5378](https://github.com/Rdatatable/data.table/issues/5378). Thanks to @jangorecki, @christellacaze, @Yiguan, @alexdthomas, @ibombonato, @Befrancesco, @TobiasGold for reporting/requesting, and Benjamin Schwendinger for the PR.
 
+8. Computations in `j` can return a matrix or array _if it is one-dimensional_, e.g. a row or column vector, when `j` is a list of columns during grouping, [#783](https://github.com/Rdatatable/data.table/issues/783). Previously a matrix could be provided `DT[, expr, by]` form, but not `DT[, list(expr), by]` form; this resolves that inconsistency. It is still an error to return a "true" array, e.g. a `2x3` matrix.
+
 ## BUG FIXES
 
 1. `unique()` returns a copy the case when `nrows(x) <= 1` instead of a mutable alias, [#5932](https://github.com/Rdatatable/data.table/pull/5932). This is consistent with existing `unique()` behavior when the input has no duplicates but more than one row. Thanks to @brookslogan for the report and @dshemetov for the fix.

inst/tests/tests.Rraw

@@ -18476,3 +18476,16 @@ DT = data.table(strrep(ja_ko, 1:3L), strrep(ja_n, 2:4L), strrep(accented_a, 3))
 test(2253.17, options=list(datatable.prettyprint.char = 4L), gsub(clean_regex, "", capture.output(print(DT))[-1L]), c("こ     んん ááá", "ここ   んんん ááá", "こここ んんんん ááá"))
 test(2253.18, options=list(datatable.prettyprint.char = 3L), gsub(clean_regex, "", capture.output(print(DT))[-1L]), c("こ      んん ááá", "ここ    んんん ááá", "こここ んんん... ááá"))
 test(2253.19, options=list(datatable.prettyprint.char = 1L), gsub(clean_regex, "", capture.output(print(DT))[-1L]), c("こ ん... á...", "こ... ん... á...", "こ... ん... á..."))
+
+# allow 1-D matrix in j for consistency, #783
+DT=data.table(a = rep(1:2, 3), b = 1:6)
+test(2254.1, DT[, .(cbind(b, b)), by=a], error="Entry 1 for group 1.*2 dimensions > 1")
+test(2254.2, DT[, .(replicate(.GRP, b)), by=a], error="Entry 1 for group 2.*2 dimensions > 1")
+test(2254.3, DT[, .(b, cbind(b, b)), by=a], error="Entry 2 for group 1.*2 dimensions > 1")
+test(2254.4, DT[, .(b, replicate(.GRP, b)), by=a], error="Entry 2 for group 2.*2 dimensions > 1")
+test(2254.5, DT[, .(array(dim=2:4)), by=a], error="3 dimensions > 1")
+test(2254.6, DT[, .(array(dim=rep(1:2, c(10L, 2L)))), by=a], error="2 dimensions > 1")
+# but 1-D matrix is fine
+test(2254.7, DT[, .(b = cbind(b)), by=a], DT[order(a)])
+test(2254.8, DT[, .(b = rbind(b)), by=a], DT[order(a)])
+test(2254.9, DT[, .(b = array(b, dim=rep(c(1L, .N), c(10L, 1L)))), by=a], DT[order(a)])

src/dogroups.c

@@ -275,8 +275,17 @@ SEXP dogroups(SEXP dt, SEXP dtcols, SEXP groups, SEXP grpcols, SEXP jiscols, SEX
       }
       for (int j=0; j<LENGTH(jval); ++j) {
         thiscol = VECTOR_ELT(jval,j);
-        if (!isNull(thiscol) && (!isVector(thiscol) || isFrame(thiscol) || isArray(thiscol) ))
-          error(_("All items in j=list(...) should be atomic vectors or lists. If you are trying something like j=list(.SD,newcol=mean(colA)) then use := by group instead (much quicker), or cbind or merge afterwards."));
+        if (isNull(thiscol)) continue;
+        if (!isVector(thiscol) || isFrame(thiscol))
+          error(_("Entry %d for group %d in j=list(...) should be atomic vector or list. If you are trying something like j=list(.SD,newcol=mean(colA)) then use := by group instead (much quicker), or cbind or merge afterwards."), j+1, i+1);
+        if (isArray(thiscol)) {
+          SEXP dims = PROTECT(getAttrib(thiscol, R_DimSymbol));
+          int nDimensions=0;
+          for (int d=0; d<LENGTH(dims); ++d) if (INTEGER(dims)[d] > 1) ++nDimensions;
+          UNPROTECT(1);
+          if (nDimensions > 1)
+            error(_("Entry %d for group %d in j=list(...) is an array with %d dimensions > 1, which is disallowed. \"Break\" the array yourself with c() or as.vector() if that is intentional."), j+1, i+1, nDimensions);
+        }
       }
     }
     if (!isNull(lhs)) {