SERVER-56819 Make indexOfCP's classic engine impl always return -1 when end is less than start

(cherry picked from commit 934d6de41bb3adbecee23e3837852e29c7ff4af6)
(cherry picked from commit 2141bb0d88dc9938fe568a7c4f4584d3c191319a)
(cherry picked from commit 55c75b382453b4fe9ae83b1e73437a20a028d9dd)
(cherry picked from commit 823909a49c1cd29218c2a696516060511b52580d)

2 files changed, 21 insertions, 1 deletions

diff --git a/jstests/aggregation/expressions/indexof_codepoints.js b/jstests/aggregation/expressions/indexof_codepoints.js
index 506b1a13cfa..bd6a261c121 100644
--- a/jstests/aggregation/expressions/indexof_codepoints.js
+++ b/jstests/aggregation/expressions/indexof_codepoints.js
@@ -30,7 +30,21 @@ load("jstests/aggregation/extras/utils.js");  // For assertErrorCode and testExp
         }
     }
 
-    var coll = db.indexofcp;
+    const coll = db.indexofcp;
+    coll.drop();
+    assert.commandWorked(coll.insert({item: 'foobar foobar', emptyStr: ''}));
+
+    // Test the edge case of searching for an empty string inside an empty string, where the start
+    // index is past the end index. These cases are designed to reproduce SERVER-56819.
+    assert.eq(-1,
+              coll.aggregate({$project: {byteLocation: {$indexOfCP: ['', '$emptyStr', 3]}}})
+                  .toArray()[0]
+                  .byteLocation);
+    assert.eq(-1,
+              coll.aggregate({$project: {byteLocation: {$indexOfCP: ['', '$emptyStr', 3, 1]}}})
+                  .toArray()[0]
+                  .byteLocation);
+
     coll.drop();
 
     // Insert a dummy document so something flows through the pipeline.
diff --git a/src/mongo/db/pipeline/expression.cpp b/src/mongo/db/pipeline/expression.cpp
index 33659fe89e2..c378c7a94c6 100644
--- a/src/mongo/db/pipeline/expression.cpp
+++ b/src/mongo/db/pipeline/expression.cpp
@@ -3045,6 +3045,12 @@ Value ExpressionIndexOfCP::evaluate(const Document& root, Variables* variables)
             std::min(codePointLength, static_cast<size_t>(endIndexArg.coerceToInt()));
     }
 
+    // If the start index is past the end, then always return -1 since 'token' does not exist within
+    // these invalid bounds.
+    if (endCodePointIndex < startCodePointIndex) {
+        return Value(-1);
+    }
+
     if (startByteIndex == 0 && input.empty() && token.empty()) {
         // If we are finding the index of "" in the string "", the below loop will not loop, so we
         // need a special case for this.