bson 6.2.0 → 6.3.0

package/lib/bson.mjs

@@ -95,8 +95,8 @@ class BSONError extends Error {
     get name() {
         return 'BSONError';
     }
-    constructor(message) {
-        super(message);
+    constructor(message, options) {
+        super(message, options);
     }
     static isBSONError(value) {
         return (value != null &&
@@ -125,6 +125,79 @@ class BSONRuntimeError extends BSONError {
     }
 }
 
+const FIRST_BIT = 0x80;
+const FIRST_TWO_BITS = 0xc0;
+const FIRST_THREE_BITS = 0xe0;
+const FIRST_FOUR_BITS = 0xf0;
+const FIRST_FIVE_BITS = 0xf8;
+const TWO_BIT_CHAR = 0xc0;
+const THREE_BIT_CHAR = 0xe0;
+const FOUR_BIT_CHAR = 0xf0;
+const CONTINUING_CHAR = 0x80;
+function validateUtf8(bytes, start, end) {
+    let continuation = 0;
+    for (let i = start; i < end; i += 1) {
+        const byte = bytes[i];
+        if (continuation) {
+            if ((byte & FIRST_TWO_BITS) !== CONTINUING_CHAR) {
+                return false;
+            }
+            continuation -= 1;
+        }
+        else if (byte & FIRST_BIT) {
+            if ((byte & FIRST_THREE_BITS) === TWO_BIT_CHAR) {
+                continuation = 1;
+            }
+            else if ((byte & FIRST_FOUR_BITS) === THREE_BIT_CHAR) {
+                continuation = 2;
+            }
+            else if ((byte & FIRST_FIVE_BITS) === FOUR_BIT_CHAR) {
+                continuation = 3;
+            }
+            else {
+                return false;
+            }
+        }
+    }
+    return !continuation;
+}
+
+function tryLatin(uint8array, start, end) {
+    if (uint8array.length === 0) {
+        return '';
+    }
+    const stringByteLength = end - start;
+    if (stringByteLength === 0) {
+        return '';
+    }
+    if (stringByteLength > 20) {
+        return null;
+    }
+    if (stringByteLength === 1 && uint8array[start] < 128) {
+        return String.fromCharCode(uint8array[start]);
+    }
+    if (stringByteLength === 2 && uint8array[start] < 128 && uint8array[start + 1] < 128) {
+        return String.fromCharCode(uint8array[start]) + String.fromCharCode(uint8array[start + 1]);
+    }
+    if (stringByteLength === 3 &&
+        uint8array[start] < 128 &&
+        uint8array[start + 1] < 128 &&
+        uint8array[start + 2] < 128) {
+        return (String.fromCharCode(uint8array[start]) +
+            String.fromCharCode(uint8array[start + 1]) +
+            String.fromCharCode(uint8array[start + 2]));
+    }
+    const latinBytes = [];
+    for (let i = start; i < end; i++) {
+        const byte = uint8array[i];
+        if (byte > 127) {
+            return null;
+        }
+        latinBytes.push(byte);
+    }
+    return String.fromCharCode(...latinBytes);
+}
+
 function nodejsMathRandomBytes(byteLength) {
     return nodeJsByteUtils.fromNumberArray(Array.from({ length: byteLength }, () => Math.floor(Math.random() * 256)));
 }
@@ -183,8 +256,23 @@ const nodeJsByteUtils = {
     fromUTF8(text) {
         return Buffer.from(text, 'utf8');
     },
-    toUTF8(buffer, start, end) {
-        return nodeJsByteUtils.toLocalBufferType(buffer).toString('utf8', start, end);
+    toUTF8(buffer, start, end, fatal) {
+        const basicLatin = end - start <= 20 ? tryLatin(buffer, start, end) : null;
+        if (basicLatin != null) {
+            return basicLatin;
+        }
+        const string = nodeJsByteUtils.toLocalBufferType(buffer).toString('utf8', start, end);
+        if (fatal) {
+            for (let i = 0; i < string.length; i++) {
+                if (string.charCodeAt(i) === 0xfffd) {
+                    if (!validateUtf8(buffer, start, end)) {
+                        throw new BSONError('Invalid UTF-8 string in BSON document');
+                    }
+                    break;
+                }
+            }
+        }
+        return string;
     },
     utf8ByteLength(input) {
         return Buffer.byteLength(input, 'utf8');
@@ -294,8 +382,20 @@ const webByteUtils = {
     fromUTF8(text) {
         return new TextEncoder().encode(text);
     },
-    toUTF8(uint8array, start, end) {
-        return new TextDecoder('utf8', { fatal: false }).decode(uint8array.slice(start, end));
+    toUTF8(uint8array, start, end, fatal) {
+        const basicLatin = end - start <= 20 ? tryLatin(uint8array, start, end) : null;
+        if (basicLatin != null) {
+            return basicLatin;
+        }
+        if (fatal) {
+            try {
+                return new TextDecoder('utf8', { fatal }).decode(uint8array.slice(start, end));
+            }
+            catch (cause) {
+                throw new BSONError('Invalid UTF-8 string in BSON document', { cause });
+            }
+        }
+        return new TextDecoder('utf8', { fatal }).decode(uint8array.slice(start, end));
     },
     utf8ByteLength(input) {
         return webByteUtils.fromUTF8(input).byteLength;
@@ -416,8 +516,8 @@ class Binary extends BSONValue {
         if (encoding === 'base64')
             return ByteUtils.toBase64(this.buffer);
         if (encoding === 'utf8' || encoding === 'utf-8')
-            return ByteUtils.toUTF8(this.buffer, 0, this.buffer.byteLength);
-        return ByteUtils.toUTF8(this.buffer, 0, this.buffer.byteLength);
+            return ByteUtils.toUTF8(this.buffer, 0, this.buffer.byteLength, false);
+        return ByteUtils.toUTF8(this.buffer, 0, this.buffer.byteLength, false);
     }
     toExtendedJSON(options) {
         options = options || {};
@@ -2505,43 +2605,6 @@ class Timestamp extends LongWithoutOverridesClass {
 }
 Timestamp.MAX_VALUE = Long.MAX_UNSIGNED_VALUE;
 
-const FIRST_BIT = 0x80;
-const FIRST_TWO_BITS = 0xc0;
-const FIRST_THREE_BITS = 0xe0;
-const FIRST_FOUR_BITS = 0xf0;
-const FIRST_FIVE_BITS = 0xf8;
-const TWO_BIT_CHAR = 0xc0;
-const THREE_BIT_CHAR = 0xe0;
-const FOUR_BIT_CHAR = 0xf0;
-const CONTINUING_CHAR = 0x80;
-function validateUtf8(bytes, start, end) {
-    let continuation = 0;
-    for (let i = start; i < end; i += 1) {
-        const byte = bytes[i];
-        if (continuation) {
-            if ((byte & FIRST_TWO_BITS) !== CONTINUING_CHAR) {
-                return false;
-            }
-            continuation -= 1;
-        }
-        else if (byte & FIRST_BIT) {
-            if ((byte & FIRST_THREE_BITS) === TWO_BIT_CHAR) {
-                continuation = 1;
-            }
-            else if ((byte & FIRST_FOUR_BITS) === THREE_BIT_CHAR) {
-                continuation = 2;
-            }
-            else if ((byte & FIRST_FIVE_BITS) === FOUR_BIT_CHAR) {
-                continuation = 3;
-            }
-            else {
-                return false;
-            }
-        }
-    }
-    return !continuation;
-}
-
 const JS_INT_MAX_LONG = Long.fromNumber(JS_INT_MAX);
 const JS_INT_MIN_LONG = Long.fromNumber(JS_INT_MIN);
 function internalDeserialize(buffer, options, isArray) {
@@ -2633,7 +2696,7 @@ function deserializeObject(buffer, index, options, isArray = false) {
         }
         if (i >= buffer.byteLength)
             throw new BSONError('Bad BSON Document: illegal CString');
-        const name = isArray ? arrayIndex++ : ByteUtils.toUTF8(buffer, index, i);
+        const name = isArray ? arrayIndex++ : ByteUtils.toUTF8(buffer, index, i, false);
         let shouldValidateKey = true;
         if (globalUTFValidation || utf8KeysSet.has(name)) {
             shouldValidateKey = validationSetting;
@@ -2656,7 +2719,7 @@ function deserializeObject(buffer, index, options, isArray = false) {
                 buffer[index + stringSize - 1] !== 0) {
                 throw new BSONError('bad string length in bson');
             }
-            value = getValidatedString(buffer, index, index + stringSize - 1, shouldValidateKey);
+            value = ByteUtils.toUTF8(buffer, index, index + stringSize - 1, shouldValidateKey);
             index = index + stringSize;
         }
         else if (elementType === BSON_DATA_OID) {
@@ -2848,7 +2911,7 @@ function deserializeObject(buffer, index, options, isArray = false) {
             }
             if (i >= buffer.length)
                 throw new BSONError('Bad BSON Document: illegal CString');
-            const source = ByteUtils.toUTF8(buffer, index, i);
+            const source = ByteUtils.toUTF8(buffer, index, i, false);
             index = i + 1;
             i = index;
             while (buffer[i] !== 0x00 && i < buffer.length) {
@@ -2856,7 +2919,7 @@ function deserializeObject(buffer, index, options, isArray = false) {
             }
             if (i >= buffer.length)
                 throw new BSONError('Bad BSON Document: illegal CString');
-            const regExpOptions = ByteUtils.toUTF8(buffer, index, i);
+            const regExpOptions = ByteUtils.toUTF8(buffer, index, i, false);
             index = i + 1;
             const optionsArray = new Array(regExpOptions.length);
             for (i = 0; i < regExpOptions.length; i++) {
@@ -2881,7 +2944,7 @@ function deserializeObject(buffer, index, options, isArray = false) {
             }
             if (i >= buffer.length)
                 throw new BSONError('Bad BSON Document: illegal CString');
-            const source = ByteUtils.toUTF8(buffer, index, i);
+            const source = ByteUtils.toUTF8(buffer, index, i, false);
             index = i + 1;
             i = index;
             while (buffer[i] !== 0x00 && i < buffer.length) {
@@ -2889,7 +2952,7 @@ function deserializeObject(buffer, index, options, isArray = false) {
             }
             if (i >= buffer.length)
                 throw new BSONError('Bad BSON Document: illegal CString');
-            const regExpOptions = ByteUtils.toUTF8(buffer, index, i);
+            const regExpOptions = ByteUtils.toUTF8(buffer, index, i, false);
             index = i + 1;
             value = new BSONRegExp(source, regExpOptions);
         }
@@ -2903,7 +2966,7 @@ function deserializeObject(buffer, index, options, isArray = false) {
                 buffer[index + stringSize - 1] !== 0) {
                 throw new BSONError('bad string length in bson');
             }
-            const symbol = getValidatedString(buffer, index, index + stringSize - 1, shouldValidateKey);
+            const symbol = ByteUtils.toUTF8(buffer, index, index + stringSize - 1, shouldValidateKey);
             value = promoteValues ? symbol : new BSONSymbol(symbol);
             index = index + stringSize;
         }
@@ -2934,7 +2997,7 @@ function deserializeObject(buffer, index, options, isArray = false) {
                 buffer[index + stringSize - 1] !== 0) {
                 throw new BSONError('bad string length in bson');
             }
-            const functionString = getValidatedString(buffer, index, index + stringSize - 1, shouldValidateKey);
+            const functionString = ByteUtils.toUTF8(buffer, index, index + stringSize - 1, shouldValidateKey);
             value = new Code(functionString);
             index = index + stringSize;
         }
@@ -2955,7 +3018,7 @@ function deserializeObject(buffer, index, options, isArray = false) {
                 buffer[index + stringSize - 1] !== 0) {
                 throw new BSONError('bad string length in bson');
             }
-            const functionString = getValidatedString(buffer, index, index + stringSize - 1, shouldValidateKey);
+            const functionString = ByteUtils.toUTF8(buffer, index, index + stringSize - 1, shouldValidateKey);
             index = index + stringSize;
             const _index = index;
             const objectSize = buffer[index] |
@@ -2986,7 +3049,7 @@ function deserializeObject(buffer, index, options, isArray = false) {
                     throw new BSONError('Invalid UTF-8 string in BSON document');
                 }
             }
-            const namespace = ByteUtils.toUTF8(buffer, index, index + stringSize - 1);
+            const namespace = ByteUtils.toUTF8(buffer, index, index + stringSize - 1, false);
             index = index + stringSize;
             const oidBuffer = ByteUtils.allocate(12);
             oidBuffer.set(buffer.subarray(index, index + 12), 0);
@@ -3025,20 +3088,6 @@ function deserializeObject(buffer, index, options, isArray = false) {
     }
     return object;
 }
-function getValidatedString(buffer, start, end, shouldValidateUtf8) {
-    const value = ByteUtils.toUTF8(buffer, start, end);
-    if (shouldValidateUtf8) {
-        for (let i = 0; i < value.length; i++) {
-            if (value.charCodeAt(i) === 0xfffd) {
-                if (!validateUtf8(buffer, start, end)) {
-                    throw new BSONError('Invalid UTF-8 string in BSON document');
-                }
-                break;
-            }
-        }
-    }
-    return value;
-}
 
 const regexp = /\x00/;
 const ignoreKeys = new Set(['$db', '$ref', '$id', '$clusterTime']);