opusdev/vector-similarity-api

1import { Binary } from '../binary';2import type { Document } from '../bson';3import { BSONError, BSONVersionError } from '../error';4import * as constants from '../constants';5import { ByteUtils } from '../utils/byte_utils';6import { isAnyArrayBuffer, isDate, isRegExp } from './utils';7 8export function internalCalculateObjectSize(9  object: Document,10  serializeFunctions?: boolean,11  ignoreUndefined?: boolean12): number {13  let totalLength = 4 + 1;14 15  if (Array.isArray(object)) {16    for (let i = 0; i < object.length; i++) {17      totalLength += calculateElement(18        i.toString(),19        object[i],20        serializeFunctions,21        true,22        ignoreUndefined23      );24    }25  } else {26    // If we have toBSON defined, override the current object27 28    if (typeof object?.toBSON === 'function') {29      object = object.toBSON();30    }31 32    // Calculate size33    for (const key of Object.keys(object)) {34      totalLength += calculateElement(key, object[key], serializeFunctions, false, ignoreUndefined);35    }36  }37 38  return totalLength;39}40 41/** @internal */42function calculateElement(43  name: string,44  // eslint-disable-next-line @typescript-eslint/no-explicit-any45  value: any,46  serializeFunctions = false,47  isArray = false,48  ignoreUndefined = false49) {50  // If we have toBSON defined, override the current object51  if (typeof value?.toBSON === 'function') {52    value = value.toBSON();53  }54 55  switch (typeof value) {56    case 'string':57      return 1 + ByteUtils.utf8ByteLength(name) + 1 + 4 + ByteUtils.utf8ByteLength(value) + 1;58    case 'number':59      if (60        Math.floor(value) === value &&61        value >= constants.JS_INT_MIN &&62        value <= constants.JS_INT_MAX63      ) {64        if (value >= constants.BSON_INT32_MIN && value <= constants.BSON_INT32_MAX) {65          // 32 bit66          return (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) + (4 + 1);67        } else {68          return (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) + (8 + 1);69        }70      } else {71        // 64 bit72        return (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) + (8 + 1);73      }74    case 'undefined':75      if (isArray || !ignoreUndefined)76        return (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) + 1;77      return 0;78    case 'boolean':79      return (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) + (1 + 1);80    case 'object':81      if (82        value != null &&83        typeof value._bsontype === 'string' &&84        value[constants.BSON_VERSION_SYMBOL] !== constants.BSON_MAJOR_VERSION85      ) {86        throw new BSONVersionError();87      } else if (value == null || value._bsontype === 'MinKey' || value._bsontype === 'MaxKey') {88        return (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) + 1;89      } else if (value._bsontype === 'ObjectId') {90        return (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) + (12 + 1);91      } else if (value instanceof Date || isDate(value)) {92        return (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) + (8 + 1);93      } else if (94        ArrayBuffer.isView(value) ||95        value instanceof ArrayBuffer ||96        isAnyArrayBuffer(value)97      ) {98        return (99          (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) + (1 + 4 + 1) + value.byteLength100        );101      } else if (102        value._bsontype === 'Long' ||103        value._bsontype === 'Double' ||104        value._bsontype === 'Timestamp'105      ) {106        return (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) + (8 + 1);107      } else if (value._bsontype === 'Decimal128') {108        return (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) + (16 + 1);109      } else if (value._bsontype === 'Code') {110        // Calculate size depending on the availability of a scope111        if (value.scope != null && Object.keys(value.scope).length > 0) {112          return (113            (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) +114            1 +115            4 +116            4 +117            ByteUtils.utf8ByteLength(value.code.toString()) +118            1 +119            internalCalculateObjectSize(value.scope, serializeFunctions, ignoreUndefined)120          );121        } else {122          return (123            (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) +124            1 +125            4 +126            ByteUtils.utf8ByteLength(value.code.toString()) +127            1128          );129        }130      } else if (value._bsontype === 'Binary') {131        const binary: Binary = value;132        // Check what kind of subtype we have133        if (binary.sub_type === Binary.SUBTYPE_BYTE_ARRAY) {134          return (135            (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) +136            (binary.position + 1 + 4 + 1 + 4)137          );138        } else {139          return (140            (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) + (binary.position + 1 + 4 + 1)141          );142        }143      } else if (value._bsontype === 'Symbol') {144        return (145          (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) +146          ByteUtils.utf8ByteLength(value.value) +147          4 +148          1 +149          1150        );151      } else if (value._bsontype === 'DBRef') {152        // Set up correct object for serialization153        const ordered_values = Object.assign(154          {155            $ref: value.collection,156            $id: value.oid157          },158          value.fields159        );160 161        // Add db reference if it exists162        if (value.db != null) {163          ordered_values['$db'] = value.db;164        }165 166        return (167          (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) +168          1 +169          internalCalculateObjectSize(ordered_values, serializeFunctions, ignoreUndefined)170        );171      } else if (value instanceof RegExp || isRegExp(value)) {172        return (173          (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) +174          1 +175          ByteUtils.utf8ByteLength(value.source) +176          1 +177          (value.global ? 1 : 0) +178          (value.ignoreCase ? 1 : 0) +179          (value.multiline ? 1 : 0) +180          1181        );182      } else if (value._bsontype === 'BSONRegExp') {183        return (184          (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) +185          1 +186          ByteUtils.utf8ByteLength(value.pattern) +187          1 +188          ByteUtils.utf8ByteLength(value.options) +189          1190        );191      } else {192        return (193          (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) +194          internalCalculateObjectSize(value, serializeFunctions, ignoreUndefined) +195          1196        );197      }198    case 'function':199      if (serializeFunctions) {200        return (201          (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) +202          1 +203          4 +204          ByteUtils.utf8ByteLength(value.toString()) +205          1206        );207      }208      return 0;209    case 'bigint':210      return (name != null ? ByteUtils.utf8ByteLength(name) + 1 : 0) + (8 + 1);211    case 'symbol':212      return 0;213    default:214      throw new BSONError(`Unrecognized JS type: ${typeof value}`);215  }216 217  return 0;218}219