Fix non-equi join error with equi + 2 inequality conditions (#7642)

* Fix non-equi join error with equi + 2 inequality conditions (#7641)

When rightcols contains duplicates (same x column in multiple non-equi
conditions), chmatchdup remaps them into the expanded result namespace.
This overwrote rightcols, causing downstream code (e.g. .shallow(x, rightcols)
in .join_result_key) to reference columns beyond ncol(x).

Introduce ansrightcols for the remapped indices and keep rightcols as
original column indices into x. Only icolsAns needs the remapped values;
all other uses (xcols, names_x[rightcols], .join_result_key) need the
original x-relative indices.

Closes #7641

* Add contributor credit in DESCRIPTION and NEWS

Per reviewer request: thank reporter in NEWS.md and add
Tarun Thammisetty as contributor to DESCRIPTION.

Author: tarun-t

DESCRIPTION

@@ -107,5 +107,6 @@ Authors@R: c(
   person("Vinit", "Thakur",        role="ctb"),
   person("Mukul", "Kumar",         role="ctb"),
   person("Ildikó", "Czeller",      role="ctb"),
-  person("Manmita", "Das",         role="ctb")
+  person("Manmita", "Das",         role="ctb"),
+  person("Tarun", "Thammisetty",   role="ctb")
   )

NEWS.md

@@ -38,6 +38,8 @@
 
 4. `rowwiseDT()` now provides a helpful error message when a complex object that is not a list (e.g., a function) is provided as a cell value, instructing the user to wrap it in `list()`, [#7219](https://github.com/Rdatatable/data.table/issues/7219). Thanks @kylebutts for the report and @venom1204 for the fix.
 
+5. Non-equi joins combining an equality condition with two inequality conditions on the same column (e.g., `on = .(id == id, val >= lo, val <= hi)`) no longer error, [#7641](https://github.com/Rdatatable/data.table/issues/7641). The internal `chmatchdup` remapping of duplicate `rightcols` was overwriting the original column indices, causing downstream code to reference non-existent columns. Thanks @tarun-t for the report and fix, and @aitap for the diagnosis.
+
 ### Notes
 
 1. {data.table} now depends on R 3.5.0 (2018).

R/data.table.R

@@ -1075,12 +1075,15 @@ replace_dot_alias = function(e) {
       if (length(tt)) jisvars[tt] = paste0("i.",jisvars[tt])
       if (length(duprightcols <- rightcols[duplicated(rightcols)])) {
         nx = c(names_x, names_x[duprightcols])
-        rightcols = chmatchdup(names_x[rightcols], nx)
+        ansrightcols = chmatchdup(names_x[rightcols], nx) # indices into result namespace nx, #7641
         nx = make.unique(nx)
-      } else nx = names_x
+      } else {
+        nx = names_x
+        ansrightcols = rightcols
+      }
       ansvars = make.unique(c(nx, jisvars))
       icols = c(leftcols, seq_along(i)[-leftcols])
-      icolsAns = c(rightcols, seq.int(length(nx)+1L, length.out=ncol(i)-length(unique(leftcols))))
+      icolsAns = c(ansrightcols, seq.int(length(nx)+1L, length.out=ncol(i)-length(unique(leftcols))))
       xcols = xcolsAns = seq_along(x)[-rightcols]
     }
     ansvals = chmatch(ansvars, nx)

inst/tests/tests.Rraw

@@ -6954,6 +6954,11 @@ DT1 <- data.table(a=1L, key='a')
 DT2 <- data.table(a=2.0, key='a')
 test(1483.83, DT1[DT2, roll='nearest'], data.table(a=2L, key='a'))
 
+# non-equi join with equi + 2 inequality conditions should not error, #7641
+dt_x = data.table(id = c("A","A","B","B"), val = c(1,5,2,4), key = c("id","val"))
+dt_i = data.table(id = c("A","B"), lo = c(2,1), hi = c(6,3))
+test(1483.91, nrow(dt_x[dt_i, on = .(id == id, val >= lo, val <= hi)]), 2L)
+
 # NULL items should be removed when making data.table from list, #842
 # Original fix for #842 added a branch in as.data.table.list() using point()
 # Then PR#3471 moved logic from data.table() into as.data.table.list() and now removes NULL items up front, so longer need for the branch