SERVER-40343 Better handling of errors from PCRE in $regex expressions

2 files changed, 11 insertions, 6 deletions

diff --git a/src/mongo/db/pipeline/expression.cpp b/src/mongo/db/pipeline/expression.cpp
index cebc922084f..f842c88c1c9 100644
--- a/src/mongo/db/pipeline/expression.cpp
+++ b/src/mongo/db/pipeline/expression.cpp
@@ -5681,10 +5681,15 @@ public:
                                    0,  // No need to overwrite the options set during pcre_compile.
                                    &_capturesBuffer.front(),
                                    _capturesBuffer.size());
-        // The 'execResult' will be '(_numCaptures + 1)' if there is a match, negative if there is
-        // no match, and zero if _capturesBuffer's capacity is not sufficient to hold all the
-        // results; the latter scenario should never actually occur.
-        invariant(execResult < 0 || execResult == _numCaptures + 1);
+        // The 'execResult' will be (_numCaptures + 1) if there is a match, -1 if there is no
+        // match, negative if there is an error during execution, and zero if _capturesBuffer's
+        // capacity is not sufficient to hold all the results. The latter scenario should never
+        // occur.
+        uassert(
+            51156,
+            str::stream() << "Error occurred while executing the regular expression. Result code:"
+                          << execResult,
+            execResult == -1 || execResult == (_numCaptures + 1));
         return execResult;
     }
 
diff --git a/src/mongo/db/pipeline/expression_test.cpp b/src/mongo/db/pipeline/expression_test.cpp
index d0ee16fc2fa..5c2426217ab 100644
--- a/src/mongo/db/pipeline/expression_test.cpp
+++ b/src/mongo/db/pipeline/expression_test.cpp
@@ -6019,8 +6019,8 @@ TEST(ExpressionRegexFindAllTest, InvalidUTF8InInput) {
     intrusive_ptr<ExpressionContextForTest> expCtx(new ExpressionContextForTest());
     ExpressionRegexFindAll regexF(expCtx);
     regexF.addOperand(ExpressionConstant::create(expCtx, input));
-    // Verify no match if there is an invalid UTF-8 character in input.
-    ASSERT_VALUE_EQ(regexF.evaluate(Document()), Value(std::vector<Value>()));
+    // Verify that PCRE will error during execution if input is not a valid UTF-8.
+    ASSERT_THROWS_CODE(regexF.evaluate(Document()), DBException, 51156);
 }
 
 TEST(ExpressionRegexFindAllTest, InvalidUTF8InRegex) {