From 9fa392dfffee6c783108581c7e762802ec1f58f4 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Aman=20=7C=20=E0=A4=85=E0=A4=AE=E0=A4=A8?=
 <121532863+beingamanforever@users.noreply.github.com>
Date: Tue, 20 Jan 2026 06:37:08 +0530
Subject: [PATCH] tabulate: fix categorical and string edge cases

---
 inst/tabulate.m | 68 ++++++++++++++++++++++++++++---------------------
 1 file changed, 39 insertions(+), 29 deletions(-)

diff --git a/inst/tabulate.m b/inst/tabulate.m
index 1e378a6c..3ea3f3e4 100644
--- a/inst/tabulate.m
+++ b/inst/tabulate.m
@@ -84,7 +84,11 @@
     endif
 
     total = sum (counts);
-    percents = 100 * counts ./ total;
+    if (total == 0)
+      percents = zeros (size (counts));
+    else
+      percents = 100 * counts ./ total;
+    endif
 
     ## Output format: Cell array
     out = cell (length (vals), 3);
@@ -96,40 +100,21 @@
     ## Handle string arrays
     x(ismissing (x)) = [];
 
-    ## Convert to cellstr and use grp2idx which is robust
-    [idx, vals] = grp2idx (cellstr (x));
-
-    if (isempty (idx))
-      counts = [];
-      percents = [];
+    if (isempty (x))
+      out = cell (0, 3);
     else
-      counts = accumarray (idx, 1);
-      total = sum (counts);
-      percents = 100 * counts ./ total;
-    endif
+      ## Convert to cellstr and use grp2idx which is robust
+      [idx, vals] = grp2idx (cellstr (x));
 
-    ## Output format: Cell array
-    vals_cell = vals;
-    out = cell (length (vals_cell), 3);
-    out(:,1) = vals_cell;
-    out(:,2) = num2cell (counts);
-    out(:,3) = num2cell (percents);
-
-    if (isempty (idx))
-      counts = [];
-      percents = [];
-    else
       counts = accumarray (idx, 1);
       total = sum (counts);
       percents = 100 * counts ./ total;
-    endif
 
-    ## Output format: Cell array
-    vals_cell = vals;
-    out = cell (length (vals_cell), 3);
-    out(:,1) = vals_cell;
-    out(:,2) = num2cell (counts);
-    out(:,3) = num2cell (percents);
+      out = cell (length (vals), 3);
+      out(:,1) = vals;
+      out(:,2) = num2cell (counts);
+      out(:,3) = num2cell (percents);
+    endif
 
   elseif (islogical (x))
     ## Handle logical arrays
@@ -407,3 +392,28 @@
 %!error<tabulate: X must be either a numeric vector> tabulate ({1, 2, 3, 4})
 %!error<tabulate: X must be either a numeric vector> ...
 %! tabulate ({"a", "b"; "a", "c"})
+
+## Test categorical with all undefined values (should return zero counts/percents)
+%!test
+%! x = categorical ({'a','b','c'});
+%! x(:) = categorical (missing);
+%! tbl = tabulate (x);
+%! assert (iscell (tbl));
+%! assert ([tbl{:,2}]', [0; 0; 0]);
+%! assert ([tbl{:,3}]', [0; 0; 0]);
+
+## Test categorical with defined categories but no data
+%!test
+%! x = categorical ({}, {'low','med','high'});
+%! tbl = tabulate (x);
+%! assert (iscell (tbl));
+%! assert ([tbl{:,2}]', [0; 0; 0]);
+%! assert ([tbl{:,3}]', [0; 0; 0]);
+
+## Test string array with all missing values (should return empty table)
+%!test
+%! x = string ({'a','b'});
+%! x(:) = missing;
+%! tbl = tabulate (x);
+%! assert (iscell (tbl));
+%! assert (isempty (tbl));