util: escape lone surrogate code points using .inspect()

Unpaired surrogate code points have no representation in UTF8. Therefore, such code points are just "random" output that is unreadable. Instead, escape the code points similar to C0 and C1 control characters. Refs: https://unicodebook.readthedocs.io/unicode_encodings.html#utf-16-surrogate-pairs Signed-off-by: Ruben Bridgewater <ruben@bridgewater.de> PR-URL: #41001 Reviewed-By: James M Snell <jasnell@gmail.com>
nodejs · Dec 13, 2021 · a0326f0 · a0326f0
1 parent 668284b
commit a0326f0
Show file tree

Hide file tree

Showing 2 changed files with 62 additions and 8 deletions.
diff --git a/lib/internal/util/inspect.js b/lib/internal/util/inspect.js
@@ -179,10 +179,10 @@ const kArrayType = 1;
 const kArrayExtrasType = 2;
 
 /* eslint-disable no-control-regex */
-const strEscapeSequencesRegExp = /[\x00-\x1f\x27\x5c\x7f-\x9f]/;
-const strEscapeSequencesReplacer = /[\x00-\x1f\x27\x5c\x7f-\x9f]/g;
-const strEscapeSequencesRegExpSingle = /[\x00-\x1f\x5c\x7f-\x9f]/;
-const strEscapeSequencesReplacerSingle = /[\x00-\x1f\x5c\x7f-\x9f]/g;
+const strEscapeSequencesRegExp = /[\x00-\x1f\x27\x5c\x7f-\x9f]|[\ud800-\udbff](?![\udc00-\udfff])|(?<![\ud800-\udbff])[\udc00-\udfff]/;
+const strEscapeSequencesReplacer = /[\x00-\x1f\x27\x5c\x7f-\x9f]|[\ud800-\udbff](?![\udc00-\udfff])|(?<![\ud800-\udbff])[\udc00-\udfff]/g;
+const strEscapeSequencesRegExpSingle = /[\x00-\x1f\x5c\x7f-\x9f]|[\ud800-\udbff](?![\udc00-\udfff])|(?<![\ud800-\udbff])[\udc00-\udfff]/;
+const strEscapeSequencesReplacerSingle = /[\x00-\x1f\x5c\x7f-\x9f]|[\ud800-\udbff](?![\udc00-\udfff])|(?<![\ud800-\udbff])[\udc00-\udfff]/g;
 /* eslint-enable no-control-regex */
 
 const keyStrRegExp = /^[a-zA-Z_][a-zA-Z_0-9]*$/;
@@ -463,7 +463,10 @@ function addQuotes(str, quotes) {
   return `'${str}'`;
 }
 
-const escapeFn = (str) => meta[StringPrototypeCharCodeAt(str)];
+function escapeFn(str) {
+  const charCode = StringPrototypeCharCodeAt(str);
+  return meta.length > charCode ? meta[charCode] : `\\u${charCode.toString(16)}`;
+}
 
 // Escape control characters, single quotes and the backslash.
 // This is similar to JSON stringify escaping.
@@ -501,8 +504,7 @@ function strEscape(str) {
 
   let result = '';
   let last = 0;
-  const lastIndex = str.length;
-  for (let i = 0; i < lastIndex; i++) {
+  for (let i = 0; i < str.length; i++) {
     const point = StringPrototypeCharCodeAt(str, i);
     if (point === singleQuote ||
         point === 92 ||
@@ -514,10 +516,20 @@ function strEscape(str) {
         result += `${StringPrototypeSlice(str, last, i)}${meta[point]}`;
       }
       last = i + 1;
+    } else if (point >= 0xd800 && point <= 0xdfff) {
+      if (point <= 0xdbff && i + 1 < str.length) {
+        const point = StringPrototypeCharCodeAt(str, i + 1);
+        if (point >= 0xdc00 && point <= 0xdfff) {
+          i++;
+          continue;
+        }
+      }
+      result += `${StringPrototypeSlice(str, last, i)}${`\\u${point.toString(16)}`}`;
+      last = i + 1;
     }
   }
 
-  if (last !== lastIndex) {
+  if (last !== str.length) {
     result += StringPrototypeSlice(str, last);
   }
   return addQuotes(result, singleQuote);

diff --git a/test/parallel/test-util-inspect.js b/test/parallel/test-util-inspect.js
@@ -837,6 +837,48 @@ assert.strictEqual(util.inspect(Object.create(Date.prototype)), 'Date {}');
   );
 }
 
+// Escape unpaired surrogate pairs.
+{
+  const edgeChar = String.fromCharCode(0xd799);
+
+  for (let charCode = 0xD800; charCode < 0xDFFF; charCode++) {
+    const surrogate = String.fromCharCode(charCode);
+
+    assert.strictEqual(
+      util.inspect(surrogate),
+      `'\\u${charCode.toString(16)}'`
+    );
+    assert.strictEqual(
+      util.inspect(`${'a'.repeat(200)}${surrogate}`),
+      `'${'a'.repeat(200)}\\u${charCode.toString(16)}'`
+    );
+    assert.strictEqual(
+      util.inspect(`${surrogate}${'a'.repeat(200)}`),
+      `'\\u${charCode.toString(16)}${'a'.repeat(200)}'`
+    );
+    if (charCode < 0xdc00) {
+      const highSurrogate = surrogate;
+      const lowSurrogate = String.fromCharCode(charCode + 1024);
+      assert(
+        !util.inspect(
+          `${edgeChar}${highSurrogate}${lowSurrogate}${edgeChar}`
+        ).includes('\\u')
+      );
+      assert.strictEqual(
+        (util.inspect(
+          `${highSurrogate}${highSurrogate}${lowSurrogate}`
+        ).match(/\\u/g) ?? []).length,
+        1
+      );
+    } else {
+      assert.strictEqual(
+        util.inspect(`${edgeChar}${surrogate}${edgeChar}`),
+        `'${edgeChar}\\u${charCode.toString(16)}${edgeChar}'`
+      );
+    }
+  }
+}
+
 // Test util.inspect.styles and util.inspect.colors.
 {
   function testColorStyle(style, input, implicit) {