SERVER-120668 Disallow $trim with too large mask string (#51722)

GitOrigin-RevId: b63bdeaac30bf821c12f059235459bb809f126f5

etc/backports_required_for_multiversion_tests.yml

@@ -659,6 +659,8 @@ last-continuous:
     ticket: SERVER-119413
   - test_file: jstests/core/index/index_on_incorrect_collection.js
     ticket: SERVER-116327
+  - test_file: jstests/aggregation/expressions/trim.js
+    ticket: SERVER-120668
   suites: null
 last-lts:
   all:
@@ -1370,4 +1372,6 @@ last-lts:
     ticket: SERVER-119413
   - test_file: jstests/core/index/index_on_incorrect_collection.js
     ticket: SERVER-116327
+  - test_file: jstests/aggregation/expressions/trim.js
+    ticket: SERVER-120668
   suites: null

jstests/aggregation/expressions/trim.js

@@ -88,3 +88,12 @@ assertErrorCode(coll, [{$project: {x: {$trim: {input: {$add: [4, 2]}}}}}], 50699
 assertErrorCode(coll, [{$project: {x: {$trim: {input: "$_id"}}}}], 50699);
 assertErrorCode(coll, [{$project: {x: {$trim: {input: " x ", chars: "$_id"}}}}], 50700);
 }());
+
+assert(coll.drop());
+assert.commandWorked(coll.insert([{_id: 0, str: " x "}]));
+
+// Test that characters limit is triggered correctly (for all of $trim, $ltrim, $rtrim).
+for (const op of ["$trim", "$ltrim", "$rtrim"]) {
+    assertErrorCode(
+        coll, [{$project: {x: {[op]: {input: "$str", chars: " ".repeat(4097)}}}}], 12066800);
+}

src/mongo/db/pipeline/expression.cpp

@@ -6169,8 +6169,14 @@ Value ExpressionTrim::evaluate(const Document& root, Variables* variables) const
                           << typeName(unvalidatedUserChars.getType()) << ") instead.",
             unvalidatedUserChars.getType() == BSONType::String);
 
-    return Value(
-        doTrim(input, extractCodePointsFromChars(unvalidatedUserChars.getStringData(), _name)));
+    auto unvalidatedUserCharsStringData = unvalidatedUserChars.getStringData();
+    uassert(12066800,
+            str::stream() << _name << " requires 'chars' to be not greater than "
+                          << _kMaximumAllowedTrimStringBytes << " bytes, got "
+                          << unvalidatedUserCharsStringData.size() << " bytes instead.",
+            unvalidatedUserCharsStringData.size() <= _kMaximumAllowedTrimStringBytes);
+
+    return Value(doTrim(input, extractCodePointsFromChars(unvalidatedUserCharsStringData, _name)));
 }
 
 bool ExpressionTrim::codePointMatchesAtIndex(const StringData& input,

src/mongo/db/pipeline/expression.h

@@ -3689,6 +3689,8 @@ private:
     static constexpr size_t _kInput = 0;
     static constexpr size_t _kCharacters = 1;  // Optional, null if not specified.
 
+    static constexpr size_t _kMaximumAllowedTrimStringBytes = 4096;
+
     TrimType _trimType;
     std::string _name;  // "$trim", "$ltrim", or "$rtrim".
 };