fix(NODE-5839): support for multibyte code-points in stringifyWithMax…

…Len (#3979)
mongodb · Jan 26, 2024 · aed1cf0 · aed1cf0
1 parent 8f7bb59
commit aed1cf0
Show file tree

Hide file tree

Showing 3 changed files with 55 additions and 15 deletions.
diff --git a/src/mongo_logger.ts b/src/mongo_logger.ts
@@ -420,10 +420,29 @@ export function stringifyWithMaxLen(
 ): string {
   let strToTruncate = '';
 
-  try {
-    strToTruncate = typeof value !== 'function' ? EJSON.stringify(value, options) : value.name;
-  } catch (e) {
-    strToTruncate = `Extended JSON serialization failed with: ${e.message}`;
+  if (typeof value === 'string') {
+    strToTruncate = value;
+  } else if (typeof value === 'function') {
+    strToTruncate = value.name;
+  } else {
+    try {
+      strToTruncate = EJSON.stringify(value, options);
+    } catch (e) {
+      strToTruncate = `Extended JSON serialization failed with: ${e.message}`;
+    }
+  }
+
+  // handle truncation that occurs in the middle of multi-byte codepoints
+  if (
+    maxDocumentLength !== 0 &&
+    strToTruncate.length > maxDocumentLength &&
+    strToTruncate.charCodeAt(maxDocumentLength - 1) !==
+      strToTruncate.codePointAt(maxDocumentLength - 1)
+  ) {
+    maxDocumentLength--;
+    if (maxDocumentLength === 0) {
+      return '';
+    }
   }
 
   return maxDocumentLength !== 0 && strToTruncate.length > maxDocumentLength

diff --git a/test/integration/command-logging-and-monitoring/command_logging_and_monitoring.prose.test.ts b/test/integration/command-logging-and-monitoring/command_logging_and_monitoring.prose.test.ts
@@ -156,7 +156,7 @@ describe('Command Logging and Monitoring Prose Tests', function () {
     });
   });
 
-  context.skip('Truncation with multi-byte codepoints', function () {
+  context('Truncation with multi-byte codepoints', function () {
     /*
     A specific test case is not provided here due to the allowed variations in truncation logic
      as well as varying extended JSON whitespace usage.
@@ -216,12 +216,12 @@ describe('Command Logging and Monitoring Prose Tests', function () {
       );
 
       // multi-byte codepoint in middle of truncated string
-      expect(insertManyCommandStarted?.command.charCodeAt(maxDocLength - 1)).to.equal(
-        firstByteChar
-      );
-      expect(insertManyCommandStarted?.command.charCodeAt(maxDocLength - 1)).to.equal(
+      expect(insertManyCommandStarted?.command.charCodeAt(maxDocLength - 2)).to.equal(
         secondByteChar
       );
+      expect(insertManyCommandStarted?.command.charCodeAt(maxDocLength - 3)).to.equal(
+        firstByteChar
+      );
 
       const insertManyCommandSucceeded = writable.buffer[1];
       expect(insertManyCommandSucceeded?.message).to.equal('Command succeeded');
@@ -230,5 +230,5 @@ describe('Command Logging and Monitoring Prose Tests', function () {
         maxDocLength + ELLIPSES_LENGTH
       );
     });
-  }).skipReason = 'todo(NODE-5839)';
+  });
 });
diff --git a/test/unit/mongo_logger.test.ts b/test/unit/mongo_logger.test.ts
@@ -1268,14 +1268,36 @@ describe('class MongoLogger', async function () {
 
     context('when maxDocumentLength is non-zero', function () {
       context('when document has length greater than maxDocumentLength', function () {
-        it('truncates ejson string to length of maxDocumentLength + 3', function () {
-          expect(stringifyWithMaxLen(largeDoc, DEFAULT_MAX_DOCUMENT_LENGTH)).to.have.lengthOf(
-            DEFAULT_MAX_DOCUMENT_LENGTH + 3
-          );
+        context('when truncation does not occur mid-multibyte codepoint', function () {
+          it('truncates ejson string to length of maxDocumentLength + 3', function () {
+            expect(stringifyWithMaxLen(largeDoc, DEFAULT_MAX_DOCUMENT_LENGTH)).to.have.lengthOf(
+              DEFAULT_MAX_DOCUMENT_LENGTH + 3
+            );
+          });
         });
+
         it('ends with "..."', function () {
           expect(stringifyWithMaxLen(largeDoc, DEFAULT_MAX_DOCUMENT_LENGTH)).to.match(/^.*\.\.\.$/);
         });
+
+        context('when truncation occurs mid-multibyte codepoint', function () {
+          const multiByteCodePoint = '\ud83d\ude0d'; // heart eyes emoji
+          context('when maxDocumentLength = 1 but greater than 0', function () {
+            it('should return an empty string', function () {
+              expect(stringifyWithMaxLen(multiByteCodePoint, 1, { relaxed: true })).to.equal('');
+            });
+          });
+
+          context('when maxDocumentLength > 1', function () {
+            it('should round down maxDocLength to previous codepoint', function () {
+              const randomStringMinusACodePoint = `random ${multiByteCodePoint}random random${multiByteCodePoint}`;
+              const randomString = `${randomStringMinusACodePoint}${multiByteCodePoint}`;
+              expect(
+                stringifyWithMaxLen(randomString, randomString.length - 1, { relaxed: true })
+              ).to.equal(`${randomStringMinusACodePoint}...`);
+            });
+          });
+        });
       });
 
       context('when document has length less than or equal to maxDocumentLength', function () {
@@ -1289,7 +1311,6 @@ describe('class MongoLogger', async function () {
             /^.*\.\.\./
           );
         });
-
         it('produces valid relaxed EJSON', function () {
           expect(() => {
             EJSON.parse(stringifyWithMaxLen(smallDoc, DEFAULT_MAX_DOCUMENT_LENGTH));