feat(NODE-5957): add parse to elements API

Author: nbbeeken

src/bson.ts

@@ -54,6 +54,7 @@ export { BSONValue } from './bson_value';
 export { BSONError, BSONVersionError, BSONRuntimeError } from './error';
 export { BSONType } from './constants';
 export { EJSON } from './extended_json';
+export { onDemand } from './parser/on_demand/index';
 
 /** @public */
 export interface Document {

src/error.ts

@@ -81,3 +81,25 @@ export class BSONRuntimeError extends BSONError {
     super(message);
   }
 }
+
+/**
+ * @public
+ * @category Error
+ *
+ * @experimental
+ *
+ * An error generated when BSON bytes are invalid.
+ * Reports the offset the parser was able to reach before encountering the error.
+ */
+export class BSONOffsetError extends BSONError {
+  public get name(): 'BSONOffsetError' {
+    return 'BSONOffsetError';
+  }
+
+  public offset: number;
+
+  constructor(message: string, offset: number) {
+    super(`${message}. offset: ${offset}`);
+    this.offset = offset;
+  }
+}

src/parser/on_demand/index.ts

@@ -0,0 +1,28 @@
+import { type BSONError, BSONOffsetError } from '../../error';
+import { type BSONElement, parseToElements } from './parse_to_elements';
+/**
+ * @experimental
+ * @public
+ *
+ * A new set of BSON APIs that are currently experimental and not intended for production use.
+ */
+export type OnDemand = {
+  BSONOffsetError: {
+    new (message: string, offset: number): BSONOffsetError;
+    isBSONError(value: unknown): value is BSONError;
+  };
+  parseToElements: (this: void, bytes: Uint8Array, startOffset?: number) => Iterable<BSONElement>;
+};
+
+/**
+ * @experimental
+ * @public
+ */
+const onDemand: OnDemand = Object.create(null);
+
+onDemand.parseToElements = parseToElements;
+onDemand.BSONOffsetError = BSONOffsetError;
+
+Object.freeze(onDemand);
+
+export { onDemand };

src/parser/on_demand/parse_to_elements.ts

@@ -0,0 +1,131 @@
+import { BSONOffsetError } from '../../error';
+import { NumberUtils } from '../../utils/number_utils';
+
+/**
+ * @public
+ * @experimental
+ */
+export type BSONElement = [
+  type: number,
+  nameOffset: number,
+  nameLength: number,
+  offset: number,
+  length: number
+];
+
+/**
+ * Searches for null terminator.
+ * **Does not** bounds check since this should **ONLY** be used within parseToElements which has asserted that `bytes` ends with a `0x00`.
+ * So this will at most iterate to the document's terminator and error if that is the offset reached.
+ */
+function findNull(bytes: Uint8Array, offset: number): number {
+  let nullTerminatorOffset = offset;
+
+  for (; bytes[nullTerminatorOffset] !== 0x00; nullTerminatorOffset++);
+
+  if (nullTerminatorOffset === bytes.length - 1) {
+    throw new BSONOffsetError('Null terminator not found', offset);
+  }
+
+  return nullTerminatorOffset;
+}
+
+/**
+ * @public
+ * @experimental
+ */
+export function parseToElements(bytes: Uint8Array, startOffset = 0): Iterable<BSONElement> {
+  if (bytes.length < 5) {
+    throw new BSONOffsetError(
+      `Input must be at least 5 bytes, got ${bytes.length} bytes`,
+      startOffset
+    );
+  }
+
+  const documentSize = NumberUtils.getSize(bytes, startOffset);
+
+  if (documentSize > bytes.length - startOffset) {
+    throw new BSONOffsetError(
+      `Parsed documentSize (${documentSize} bytes) does not match input length (${bytes.length} bytes)`,
+      startOffset
+    );
+  }
+
+  if (bytes[startOffset + documentSize - 1] !== 0x00) {
+    throw new BSONOffsetError('BSON documents must end in 0x00', startOffset + documentSize);
+  }
+
+  const elements: BSONElement[] = [];
+  let offset = startOffset + 4;
+
+  while (offset <= documentSize + startOffset) {
+    const type = bytes[offset];
+    offset += 1;
+
+    if (type === 0) {
+      if (offset - startOffset !== documentSize) {
+        throw new BSONOffsetError(`Invalid 0x00 type byte`, offset);
+      }
+      break;
+    }
+
+    const nameOffset = offset;
+    const nameLength = findNull(bytes, offset) - nameOffset;
+    offset += nameLength + 1;
+
+    let length: number;
+
+    if (type === 1 || type === 18 || type === 9 || type === 17) {
+      // double, long, date, timestamp
+      length = 8;
+    } else if (type === 16) {
+      // int
+      length = 4;
+    } else if (type === 7) {
+      // objectId
+      length = 12;
+    } else if (type === 19) {
+      // decimal128
+      length = 16;
+    } else if (type === 8) {
+      // boolean
+      length = 1;
+    } else if (type === 10 || type === 6 || type === 127 || type === 255) {
+      // null, undefined, maxKey, minKey
+      length = 0;
+    }
+    // Needs a size calculation
+    else if (type === 11) {
+      // regex
+      length = findNull(bytes, findNull(bytes, offset) + 1) + 1 - offset;
+    } else if (type === 3 || type === 4 || type === 15) {
+      // object, array, code_w_scope
+      length = NumberUtils.getSize(bytes, offset);
+    } else if (type === 2 || type === 5 || type === 12 || type === 13 || type === 14) {
+      // string, binary, dbpointer, code, symbol
+      length = NumberUtils.getSize(bytes, offset) + 4;
+      if (type === 5) {
+        // binary subtype
+        length += 1;
+      }
+      if (type === 12) {
+        // dbPointer's objectId
+        length += 12;
+      }
+    } else {
+      throw new BSONOffsetError(
+        `Invalid 0x${type.toString(16).padStart(2, '0')} type byte`,
+        offset
+      );
+    }
+
+    if (length > documentSize) {
+      throw new BSONOffsetError('value reports length larger than document', offset);
+    }
+
+    elements.push([type, nameOffset, nameLength, offset, length]);
+    offset += length;
+  }
+
+  return elements;
+}

src/utils/number_utils.ts

@@ -1,3 +1,5 @@
+import { BSONOffsetError } from '../error';
+
 const FLOAT = new Float64Array(1);
 const FLOAT_BYTES = new Uint8Array(FLOAT.buffer, 0, 8);
 
@@ -7,6 +9,13 @@ const FLOAT_BYTES = new Uint8Array(FLOAT.buffer, 0, 8);
  * @internal
  */
 export const NumberUtils = {
+  getSize(source: Uint8Array, offset: number): number {
+    if (source[offset + 3] > 127) {
+      throw new BSONOffsetError('BSON size cannot be negative', offset);
+    }
+    return NumberUtils.getInt32LE(source, offset);
+  },
+
   /** Reads a little-endian 32-bit integer from source */
   getInt32LE(source: Uint8Array, offset: number): number {
     return (

test/node/error.test.ts

@@ -1,7 +1,13 @@
 import { expect } from 'chai';
 import { loadESModuleBSON } from '../load_bson';
 
-import { __isWeb__, BSONError, BSONVersionError, BSONRuntimeError } from '../register-bson';
+import {
+  __isWeb__,
+  BSONError,
+  BSONVersionError,
+  BSONRuntimeError,
+  onDemand
+} from '../register-bson';
 
 const instanceOfChecksWork = !__isWeb__;
 
@@ -102,4 +108,24 @@ describe('BSONError', function () {
       expect(new BSONRuntimeError('Woops!')).to.have.property('name', 'BSONRuntimeError');
     });
   });
+
+  describe('class BSONOffsetError', () => {
+    it('is a BSONError instance', function () {
+      expect(BSONError.isBSONError(new onDemand.BSONOffsetError('Oopsie', 3))).to.be.true;
+    });
+
+    it('has a name property equal to "BSONOffsetError"', function () {
+      expect(new onDemand.BSONOffsetError('Woops!', 3)).to.have.property('name', 'BSONOffsetError');
+    });
+
+    it('sets the offset property', function () {
+      expect(new onDemand.BSONOffsetError('Woops!', 3)).to.have.property('offset', 3);
+    });
+
+    it('includes the offset in the message', function () {
+      expect(new onDemand.BSONOffsetError('Woops!', 3))
+        .to.have.property('message')
+        .that.matches(/offset: 3/i);
+    });
+  });
 });

test/node/exports.test.ts

@@ -18,6 +18,7 @@ const EXPECTED_EXPORTS = [
   'DBRef',
   'Binary',
   'ObjectId',
+  'onDemand',
   'UUID',
   'Long',
   'Timestamp',