@bsv/wallet-toolbox 1.5.20 → 1.6.1

package/src/services/chaintracker/chaintracks/util/BulkFileDataManager.ts

@@ -0,0 +1,938 @@
+import { ChaintracksFetchApi } from '../Api/ChaintracksFetchApi'
+import { BlockHeader, Chain, WERR_INTERNAL, WERR_INVALID_OPERATION, WERR_INVALID_PARAMETER } from '../../../../sdk'
+import { Hash } from '@bsv/sdk'
+import { asArray, asString, asUint8Array } from '../../../../utility/utilityHelpers.noBuffer'
+import { BulkHeaderFileInfo, BulkHeaderFilesInfo } from './BulkHeaderFile'
+import { isKnownValidBulkHeaderFile, validBulkHeaderFiles } from './validBulkHeaderFilesByFileHash'
+import { HeightRange } from './HeightRange'
+import {
+  addWork,
+  convertBitsToWork,
+  deserializeBlockHeader,
+  serializeBaseBlockHeaders,
+  subWork,
+  validateBufferOfHeaders,
+  validateGenesisHeader
+} from './blockHeaderUtilities'
+import { ChaintracksStorageBulkFileApi } from '../Api/ChaintracksStorageApi'
+import { ChaintracksFetch } from './ChaintracksFetch'
+import { logger } from '../../../../../test/utils/TestUtilsWalletStorage'
+import { ChaintracksFsApi } from '../Api/ChaintracksFsApi'
+import { SingleWriterMultiReaderLock } from './SingleWriterMultiReaderLock'
+
+export interface BulkFileDataManagerOptions {
+  chain: Chain
+  maxPerFile: number
+  maxRetained?: number
+  fetch?: ChaintracksFetchApi
+  fromKnownSourceUrl?: string
+}
+
+/**
+ * Manages bulk file data (typically 8MB chunks of 100,000 headers each).
+ *
+ * If not cached in memory,
+ * optionally fetches data by `sourceUrl` from CDN on demand,
+ * optionally finds data by `fileId` in a database on demand,
+ * and retains a limited number of files in memory,
+ * subject to the optional `maxRetained` limit.
+ */
+export class BulkFileDataManager {
+  static createDefaultOptions(chain: Chain): BulkFileDataManagerOptions {
+    return {
+      chain,
+      maxPerFile: 100000,
+      maxRetained: 2,
+      fetch: new ChaintracksFetch(),
+      fromKnownSourceUrl: 'https://cdn.projectbabbage.com/blockheaders'
+    }
+  }
+
+  private bfds: BulkFileData[] = []
+  private fileHashToIndex: Record<string, number> = {}
+  private lock: SingleWriterMultiReaderLock = new SingleWriterMultiReaderLock()
+  private storage?: ChaintracksStorageBulkFileApi
+
+  readonly chain: Chain
+  readonly maxPerFile: number
+  readonly fetch?: ChaintracksFetchApi
+  readonly maxRetained?: number
+  readonly fromKnownSourceUrl?: string
+
+  constructor(options: BulkFileDataManagerOptions | Chain) {
+    if (typeof options === 'object') options = options as BulkFileDataManagerOptions
+    else options = BulkFileDataManager.createDefaultOptions(options as Chain)
+    this.chain = options.chain
+    this.maxPerFile = options.maxPerFile
+    this.maxRetained = options.maxRetained
+    this.fromKnownSourceUrl = options.fromKnownSourceUrl
+    this.fetch = options.fetch
+
+    this.deleteBulkFilesNoLock()
+  }
+
+  async createReader(range?: HeightRange, maxBufferSize?: number): Promise<BulkFileDataReader> {
+    range = range || (await this.getHeightRange())
+    maxBufferSize = maxBufferSize || 1000000 * 80 // 100,000 headers, 8MB
+    return new BulkFileDataReader(this, range, maxBufferSize)
+  }
+
+  async updateFromUrl(cdnUrl: string): Promise<void> {
+    if (!this.fetch) throw new WERR_INVALID_OPERATION('fetch is not defined in the BulkFileDataManager.')
+
+    const toUrl = (file: string) => this.fetch!.pathJoin(cdnUrl, file)
+    const url = toUrl(`${this.chain}NetBlockHeaders.json`)
+
+    const availableBulkFiles = (await this.fetch.fetchJson(url)) as BulkHeaderFilesInfo
+    if (!availableBulkFiles)
+      throw new WERR_INVALID_PARAMETER(`cdnUrl`, `a valid BulkHeaderFilesInfo JSON resource available from ${url}`)
+
+    const selectedFiles = selectBulkHeaderFiles(
+      availableBulkFiles.files,
+      this.chain,
+      this.maxPerFile || availableBulkFiles.headersPerFile
+    )
+    for (const bf of selectedFiles) {
+      if (!bf.fileHash) {
+        throw new WERR_INVALID_PARAMETER(`fileHash`, `valid for all files in json downloaded from ${url}`)
+      }
+      if (!bf.chain || bf.chain !== this.chain) {
+        throw new WERR_INVALID_PARAMETER(`chain`, `"${this.chain}" for all files in json downloaded from ${url}`)
+      }
+      if (!bf.sourceUrl || bf.sourceUrl !== cdnUrl) bf.sourceUrl = cdnUrl
+    }
+
+    const rangeBefore = await this.getHeightRange()
+    const r = await this.merge(selectedFiles)
+    const rangeAfter = await this.getHeightRange()
+
+    let log = 'BulkDataFileManager.updateFromUrl\n'
+    log += `  url: ${url}\n`
+    log += `  bulk range before: ${rangeBefore}\n`
+    log += `  bulk range after:  ${rangeAfter}\n`
+    logger(log)
+  }
+
+  async setStorage(storage: ChaintracksStorageBulkFileApi): Promise<void> {
+    return this.lock.withWriteLock(async () => this.setStorageNoLock(storage))
+  }
+
+  private async setStorageNoLock(storage: ChaintracksStorageBulkFileApi): Promise<void> {
+    this.storage = storage
+    // Sync bfds with storage. Two scenarios supported:
+    const sfs = await this.storage.getBulkFiles()
+    if (sfs.length === 0) {
+      // 1. Storage has no files: Update storage to reflect bfds.
+      for (const bfd of this.bfds) {
+        await this.ensureData(bfd)
+        bfd.fileId = await this.storage.insertBulkFile(bfdToInfo(bfd, true))
+      }
+    } else {
+      // 2. bfds are a prefix of storage, including last bfd having same firstHeight but possibly fewer headers: Merge storage to bfds.
+      const r = await this.mergeNoLock(sfs)
+    }
+  }
+
+  async deleteBulkFiles(): Promise<void> {
+    return this.lock.withWriteLock(async () => this.deleteBulkFilesNoLock())
+  }
+
+  private deleteBulkFilesNoLock(): void {
+    this.bfds = []
+    this.fileHashToIndex = {}
+
+    if (this.fromKnownSourceUrl) {
+      const files = selectBulkHeaderFiles(
+        validBulkHeaderFiles.filter(f => f.sourceUrl === this.fromKnownSourceUrl),
+        this.chain,
+        this.maxPerFile
+      )
+      for (const file of files) {
+        this.add({ ...file, fileHash: file.fileHash!, mru: Date.now() })
+      }
+    }
+  }
+
+  async merge(files: BulkHeaderFileInfo[]): Promise<BulkFileDataManagerMergeResult> {
+    return this.lock.withWriteLock(async () => this.mergeNoLock(files))
+  }
+
+  private async mergeNoLock(files: BulkHeaderFileInfo[]): Promise<BulkFileDataManagerMergeResult> {
+    const r: BulkFileDataManagerMergeResult = { inserted: [], updated: [], unchanged: [], dropped: [] }
+    for (const file of files) {
+      const hbf = this.getBfdForHeight(file.firstHeight)
+      if (hbf && file.fileId) hbf.fileId = file.fileId // Always update fileId if provided
+      const lbf = this.getLastBfd()
+      if (
+        hbf &&
+        hbf.fileHash === file.fileHash &&
+        hbf.count === file.count &&
+        hbf.lastHash === file.lastHash &&
+        hbf.lastChainWork === file.lastChainWork
+      ) {
+        // We already have an identical matching file...
+        r.unchanged.push(bfdToInfo(hbf))
+        continue
+      }
+      const vbf: BulkFileData = await this.validateFileInfo(file)
+      if (hbf) {
+        // We have a matching file by firstHeight but count and fileHash differ
+        await this.update(vbf, hbf, r)
+      } else if (isBdfIncremental(vbf) && lbf && isBdfIncremental(lbf)) {
+        await this.mergeIncremental(lbf, vbf, r)
+      } else {
+        const added = this.add(vbf)
+        r.inserted.push(added)
+        if (this.storage) {
+          vbf.fileId = await this.storage.insertBulkFile(added)
+        }
+      }
+    }
+    logger(`BulkFileDataManager.merge:\n${this.toLogString(r)}\n`)
+    return r
+  }
+
+  private async mergeIncremental(lbf: BulkFileData, vbf: BulkFileData, r: BulkFileDataManagerMergeResult) {
+    lbf.count += vbf.count
+    lbf.lastHash = vbf.lastHash
+    lbf.lastChainWork = vbf.lastChainWork
+    await this.ensureData(lbf)
+    const newData = new Uint8Array(lbf.data!.length + vbf.data!.length)
+    newData.set(lbf.data!)
+    newData.set(vbf.data!, lbf.data!.length)
+    lbf.data = newData
+    delete this.fileHashToIndex[lbf.fileHash!]
+    lbf.fileHash = asString(Hash.sha256(asArray(newData)), 'base64')
+    this.fileHashToIndex[lbf.fileHash] = this.bfds.length - 1
+    lbf.mru = Date.now()
+    const lbfInfo = bfdToInfo(lbf, true)
+    r.updated.push(lbfInfo)
+    if (this.storage && lbf.fileId) {
+      await this.storage.updateBulkFile(lbf.fileId, lbfInfo)
+    }
+  }
+
+  toLogString(what?: BulkFileDataManagerMergeResult | BulkFileData[] | BulkHeaderFileInfo[]): string {
+    let log = ''
+    if (!what) {
+      log += this.toLogString(this.bfds)
+    } else if (what['updated']) {
+      what = what as BulkFileDataManagerMergeResult
+      for (const { category, bfds } of [
+        { category: 'unchanged', bfds: what.unchanged },
+        { category: 'dropped', bfds: what.dropped },
+        { category: 'updated', bfds: what.updated },
+        { category: 'inserted', bfds: what.inserted }
+      ]) {
+        if (bfds.length > 0) {
+          log += `  ${category}:\n`
+          log += this.toLogString(bfds)
+        }
+      }
+    } else if (Array.isArray(what)) {
+      what = what as BulkHeaderFileInfo[]
+      let i = -1
+      for (const bfd of what) {
+        i++
+        log += `  ${i}: ${bfd.fileName} fileId=${bfd.fileId} ${bfd.firstHeight}-${bfd.firstHeight + bfd.count - 1}\n`
+      }
+    }
+
+    return log
+  }
+
+  async mergeIncrementalBlockHeaders(newBulkHeaders: BlockHeader[], incrementalChainWork?: string): Promise<void> {
+    if (newBulkHeaders.length === 0) return
+    return this.lock.withWriteLock(async () => {
+      const lbf = this.getLastFileNoLock()
+      const nextHeight = lbf ? lbf.firstHeight + lbf.count : 0
+      if (nextHeight > 0 && newBulkHeaders.length > 0 && newBulkHeaders[0].height < nextHeight) {
+        // Don't modify the incoming array...
+        newBulkHeaders = [...newBulkHeaders]
+        // If we have more headers than we need, drop the incoming headers.
+        while (newBulkHeaders.length > 0 && newBulkHeaders[0].height < nextHeight) {
+          const h = newBulkHeaders.shift()
+          if (h && incrementalChainWork) {
+            incrementalChainWork = subWork(incrementalChainWork, convertBitsToWork(h.bits))
+          }
+        }
+      }
+      if (newBulkHeaders.length === 0) return
+      if (!lbf || nextHeight !== newBulkHeaders[0].height)
+        throw new WERR_INVALID_PARAMETER('newBulkHeaders', 'an extension of existing bulk headers')
+      if (!lbf.lastHash) throw new WERR_INTERNAL(`lastHash is not defined for the last bulk file ${lbf.fileName}`)
+
+      const fbh = newBulkHeaders[0]
+      const lbh = newBulkHeaders.slice(-1)[0]
+      let lastChainWork = lbf.lastChainWork
+      if (incrementalChainWork) {
+        lastChainWork = addWork(incrementalChainWork, lastChainWork)
+      } else {
+        // If lastChainWork is not provided, calculate it from the last file with basic validation.
+        let lastHeight = lbf.firstHeight + lbf.count - 1
+        let lastHash = lbf.lastHash
+        for (const h of newBulkHeaders) {
+          if (h.height !== lastHeight + 1 || h.previousHash !== lastHash) {
+            throw new WERR_INVALID_PARAMETER(
+              'headers',
+              `an extension of existing bulk headers, header with height ${h.height} is non-sequential`
+            )
+          }
+          lastChainWork = addWork(lastChainWork, convertBitsToWork(h.bits))
+          lastHeight = h.height
+          lastHash = h.hash
+        }
+      }
+      const data = serializeBaseBlockHeaders(newBulkHeaders)
+
+      const fileHash = asString(Hash.sha256(asArray(data)), 'base64')
+      const bf: BulkHeaderFileInfo = {
+        fileId: undefined,
+        chain: this.chain,
+        sourceUrl: undefined,
+        fileName: 'incremental',
+        firstHeight: fbh.height,
+        count: newBulkHeaders.length,
+        prevChainWork: lbf.lastChainWork,
+        lastChainWork,
+        prevHash: lbf.lastHash,
+        lastHash: lbh.hash,
+        fileHash,
+        data
+      }
+      await this.mergeNoLock([bf])
+    })
+  }
+
+  async getBulkFiles(keepData?: boolean): Promise<BulkHeaderFileInfo[]> {
+    return this.lock.withReadLock(async () => {
+      return this.bfds.map(bfd => bfdToInfo(bfd, keepData))
+    })
+  }
+
+  async getHeightRange(): Promise<HeightRange> {
+    return this.lock.withReadLock(async () => {
+      if (this.bfds.length === 0) return HeightRange.empty
+      const first = this.bfds[0]
+      const last = this.bfds[this.bfds.length - 1]
+      return new HeightRange(first.firstHeight, last.firstHeight + last.count - 1)
+    })
+  }
+
+  async getDataFromFile(file: BulkHeaderFileInfo, offset?: number, length?: number): Promise<Uint8Array | undefined> {
+    const bfd = await this.getBfdForHeight(file.firstHeight)
+    if (!bfd || bfd.count < file.count)
+      throw new WERR_INVALID_PARAMETER(
+        'file',
+        `a match for ${file.firstHeight}, ${file.count} in the BulkFileDataManager.`
+      )
+    return this.lock.withReadLock(async () => this.getDataFromFileNoLock(bfd, offset, length))
+  }
+
+  private async getDataFromFileNoLock(
+    bfd: BulkFileData,
+    offset?: number,
+    length?: number
+  ): Promise<Uint8Array | undefined> {
+    const fileLength = bfd.count * 80
+    offset = offset || 0
+    if (offset > fileLength - 1) return undefined
+    length = length || bfd.count * 80 - offset
+    length = Math.min(length, fileLength - offset)
+    let data: Uint8Array | undefined
+    if (bfd.data) {
+      data = bfd.data.slice(offset, offset + length)
+    } else if (bfd.fileId && this.storage) {
+      data = await this.storage.getBulkFileData(bfd.fileId, offset, length)
+    }
+    if (!data) {
+      await this.ensureData(bfd)
+      if (bfd.data) data = bfd.data.slice(offset, offset + length)
+    }
+    if (!data) return undefined
+    return data
+  }
+
+  async findHeaderForHeightOrUndefined(height: number): Promise<BlockHeader | undefined> {
+    return this.lock.withReadLock(async () => {
+      if (!Number.isInteger(height) || height < 0)
+        throw new WERR_INVALID_PARAMETER('height', `a non-negative integer (${height}).`)
+      const file = this.bfds.find(f => f.firstHeight <= height && f.firstHeight + f.count > height)
+      if (!file) return undefined
+      const offset = (height - file.firstHeight) * 80
+      const data = await this.getDataFromFileNoLock(file, offset, 80)
+      if (!data) return undefined
+      const header = deserializeBlockHeader(data, 0, height)
+      return header
+    })
+  }
+
+  async getFileForHeight(height: number): Promise<BulkHeaderFileInfo | undefined> {
+    return this.lock.withReadLock(async () => {
+      const bfd = this.getBfdForHeight(height)
+      if (!bfd) return undefined
+      return bfdToInfo(bfd)
+    })
+  }
+
+  private getBfdForHeight(height: number): BulkFileData | undefined {
+    if (!Number.isInteger(height) || height < 0)
+      throw new WERR_INVALID_PARAMETER('height', `a non-negative integer (${height}).`)
+    const file = this.bfds.find(f => f.firstHeight <= height && f.firstHeight + f.count > height)
+    return file
+  }
+
+  private getLastBfd(fromEnd = 1): BulkFileData | undefined {
+    if (this.bfds.length < fromEnd) return undefined
+    const bfd = this.bfds[this.bfds.length - fromEnd]
+    return bfd
+  }
+
+  async getLastFile(fromEnd = 1): Promise<BulkHeaderFileInfo | undefined> {
+    return this.lock.withReadLock(async () => this.getLastFileNoLock(fromEnd))
+  }
+
+  private getLastFileNoLock(fromEnd = 1): BulkHeaderFileInfo | undefined {
+    const bfd = this.getLastBfd(fromEnd)
+    if (!bfd) return undefined
+    return bfdToInfo(bfd)
+  }
+
+  private async getDataByFileHash(fileHash: string): Promise<Uint8Array | undefined> {
+    const index = this.fileHashToIndex[fileHash]
+    if (index === undefined)
+      throw new WERR_INVALID_PARAMETER('fileHash', `known to the BulkFileDataManager. ${fileHash} is unknown.`)
+    const bfd = this.bfds[index]
+    const data = await this.ensureData(bfd)
+    return data
+  }
+
+  private async getDataByFileId(fileId: number): Promise<Uint8Array | undefined> {
+    const bfd = this.bfds.find(f => f.fileId === fileId)
+    if (bfd === undefined)
+      throw new WERR_INVALID_PARAMETER('fileId', `known to the BulkFileDataManager. ${fileId} is unknown.`)
+    const data = await this.ensureData(bfd)
+    return data
+  }
+
+  private async validateFileInfo(file: BulkHeaderFileInfo): Promise<BulkFileData> {
+    if (file.chain !== this.chain) throw new WERR_INVALID_PARAMETER('chain', `${this.chain}`)
+    if (file.count <= 0)
+      throw new WERR_INVALID_PARAMETER('bf.count', `expected count to be greater than 0, but got ${file.count}`)
+    if (file.count > this.maxPerFile && file.fileName !== 'incremental')
+      throw new WERR_INVALID_PARAMETER('count', `less than or equal to maxPerFile ${this.maxPerFile}`)
+    if (!file.fileHash) throw new WERR_INVALID_PARAMETER('fileHash', `defined`)
+    if (!file.sourceUrl && !file.fileId && !file.data)
+      throw new WERR_INVALID_PARAMETER('data', `defined when sourceUrl and fileId are undefined`)
+
+    let bfd: BulkFileData = {
+      ...file,
+      fileHash: file.fileHash,
+      mru: Date.now()
+    }
+
+    if (!bfd.validated) {
+      await this.ensureData(bfd)
+
+      if (!bfd.data || bfd.data.length !== bfd.count * 80)
+        throw new WERR_INVALID_PARAMETER(
+          'file.data',
+          `bulk file ${bfd.fileName} data length ${bfd.data?.length} does not match expected count ${bfd.count}`
+        )
+
+      bfd.fileHash = asString(Hash.sha256(asArray(bfd.data)), 'base64')
+      if (file.fileHash && file.fileHash !== bfd.fileHash)
+        throw new WERR_INVALID_PARAMETER('file.fileHash', `expected ${file.fileHash} but got ${bfd.fileHash}`)
+
+      if (!isKnownValidBulkHeaderFile(bfd)) {
+        const pbf = bfd.firstHeight > 0 ? this.getBfdForHeight(bfd.firstHeight - 1) : undefined
+        const prevHash = pbf ? pbf.lastHash! : '00'.repeat(32)
+        const prevChainWork = pbf ? pbf.lastChainWork : '00'.repeat(32)
+
+        const { lastHeaderHash, lastChainWork } = validateBufferOfHeaders(
+          bfd.data,
+          prevHash,
+          0,
+          undefined,
+          prevChainWork
+        )
+
+        if (bfd.lastHash && bfd.lastHash !== lastHeaderHash)
+          throw new WERR_INVALID_PARAMETER('file.lastHash', `expected ${bfd.lastHash} but got ${lastHeaderHash}`)
+        if (bfd.lastChainWork && bfd.lastChainWork !== lastChainWork)
+          throw new WERR_INVALID_PARAMETER(
+            'file.lastChainWork',
+            `expected ${bfd.lastChainWork} but got ${lastChainWork}`
+          )
+
+        bfd.lastHash = lastHeaderHash
+        bfd.lastChainWork = lastChainWork!
+
+        if (bfd.firstHeight === 0) {
+          validateGenesisHeader(bfd.data, bfd.chain!)
+        }
+      }
+      bfd.validated = true
+    }
+
+    return bfd
+  }
+
+  async ReValidate(): Promise<void> {
+    return this.lock.withReadLock(async () => this.ReValidateNoLock())
+  }
+
+  private async ReValidateNoLock(): Promise<void> {
+    for (const file of this.bfds) {
+      await this.ensureData(file)
+      file.validated = false // Reset validation to re-validate on next access
+      const bfd = await this.validateFileInfo(file)
+      if (!bfd.validated) throw new WERR_INTERNAL(`BulkFileDataManager.ReValidate failed for file ${bfd.fileName}`)
+      file.validated = true
+    }
+  }
+
+  private validateBfdForAdd(bfd: BulkFileData): void {
+    if (this.bfds.length === 0 && bfd.firstHeight !== 0)
+      throw new WERR_INVALID_PARAMETER('firstHeight', `0 for the first file`)
+    if (this.bfds.length > 0) {
+      const last = this.bfds[this.bfds.length - 1]
+      if (bfd.firstHeight !== last.firstHeight + last.count)
+        throw new WERR_INVALID_PARAMETER('firstHeight', `the last file's firstHeight + count`)
+      if (bfd.prevHash !== last.lastHash || bfd.prevChainWork !== last.lastChainWork)
+        throw new WERR_INVALID_PARAMETER('prevHash/prevChainWork', `the last file's lastHash/lastChainWork`)
+    }
+  }
+
+  private add(bfd: BulkFileData): BulkHeaderFileInfo {
+    this.validateBfdForAdd(bfd)
+    const index = this.bfds.length
+    this.bfds.push(bfd)
+    this.fileHashToIndex[bfd.fileHash] = index
+    this.ensureMaxRetained()
+    return bfdToInfo(bfd, true)
+  }
+
+  private replaceBfdAtIndex(index: number, update: BulkFileData): void {
+    const oldBfd = this.bfds[index]
+    delete this.fileHashToIndex[oldBfd.fileHash]
+    this.bfds[index] = update
+    this.fileHashToIndex[update.fileHash] = index
+  }
+
+  /**
+   * Updating an existing file occurs in two specific contexts:
+   *
+   * 1. CDN Update: CDN files of a specific `maxPerFile` series typically ends in a partial file
+   * which may periodically add more headers until the next file is started.
+   * If the CDN update is the second to last file (followed by an incremental file),
+   * then the incremental file is updated or deleted and also returned as the result (with a count of zero if deleted).
+   *
+   * 2. Incremental Update: The last bulk file is almost always an "incremental" file
+   * which is not limited by "maxPerFile" and holds all non-CDN bulk headers.
+   * If is updated with new bulk headers which come either from non CDN ingestors or from live header migration to bulk.
+   *
+   * Updating preserves the following properties:
+   *
+   * - Any existing headers following this update are preserved and must form an unbroken chain.
+   * - There can be at most one incremental file and it must be the last file.
+   * - The update start conditions (height, prevHash, prevChainWork) must match an existing file which may be either CDN or internal.
+   * - The update fileId must match, it may be undefind.
+   * - The fileName does not need to match.
+   * - The incremental file must always have fileName "incremental" and sourceUrl must be undefined.
+   * - The update count must be greater than 0.
+   * - The update count must be greater than current count for CDN to CDN update.
+   *
+   * @param update new validated BulkFileData to update.
+   * @param hbf corresponding existing BulkFileData to update.
+   */
+  private async update(update: BulkFileData, hbf: BulkFileData, r: BulkFileDataManagerMergeResult): Promise<void> {
+    if (
+      !hbf ||
+      hbf.firstHeight !== update.firstHeight ||
+      hbf.prevChainWork !== update.prevChainWork ||
+      hbf.prevHash !== update.prevHash
+    )
+      throw new WERR_INVALID_PARAMETER('file', `an existing file by height, prevChainWork and prevHash`)
+    if (isBdfCdn(update) === isBdfCdn(hbf) && update.count <= hbf.count)
+      throw new WERR_INVALID_PARAMETER('file.count', `greater than the current count ${hbf.count}`)
+
+    const lbf = this.getLastBfd()!
+    let index = this.bfds.length - 1
+    let truncate: BulkFileData | undefined = undefined
+    let replaced: BulkFileData | undefined = undefined
+    let drop: BulkFileData | undefined = undefined
+
+    if (hbf.firstHeight === lbf.firstHeight) {
+      // If the update is for the last file, there are three cases:
+
+      if (isBdfIncremental(update)) {
+        // 1. Incremental file may only be extended with more incremental headers.
+        if (!isBdfIncremental(lbf))
+          throw new WERR_INVALID_PARAMETER('file', `an incremental file to update an existing incremental file`)
+      } else {
+        // The update is a CDN bulk file.
+        if (isBdfCdn(lbf)) {
+          // 2. An updated CDN file replaces a partial CDN file.
+          if (update.count <= lbf.count)
+            throw new WERR_INVALID_PARAMETER(
+              'update.count',
+              `CDN update must have more headers. ${update.count} <= ${lbf.count}`
+            )
+        } else {
+          // 3. A new CDN file replaces some or all of current incremental file.
+          // Retain extra incremental headers if any.
+          if (update.count < lbf.count) {
+            // The new CDN partially replaces the last incremental file, prepare to shift work and re-add it.
+            await this.ensureData(lbf)
+            truncate = lbf
+          }
+        }
+      }
+    } else {
+      // If the update is NOT for the last file, then it MUST be for the second to last file which MUST be a CDN file:
+      // - it must be a CDN file update with more headers than the current CDN file.
+      // - the last file must be an incremental file which is updated or deleted. The updated (or deleted) last file is returned.
+      const lbf2 = this.getLastBfd(2)
+      if (!lbf2 || hbf.firstHeight !== lbf2.firstHeight)
+        throw new WERR_INVALID_PARAMETER('file', `an update to last or second to last file`)
+      if (!isBdfCdn(update) || !isBdfCdn(lbf2) || update.count <= lbf2.count)
+        throw new WERR_INVALID_PARAMETER('file', `a CDN file update with more headers than the current CDN file`)
+      if (!isBdfIncremental(lbf))
+        throw new WERR_INVALID_PARAMETER('file', `a CDN file update followed by an incremental file`)
+      if (!update.fileId) update.fileId = lbf2.fileId // Update fileId if not provided
+      if (update.count >= lbf2.count + lbf.count) {
+        // The current last file is fully replaced by the CDN update.
+        drop = lbf
+      } else {
+        // If the update doesn't fully replace the last incremental file, make sure data is available to be truncated.
+        await this.ensureData(lbf)
+        truncate = lbf
+        // The existing second to last file is fully replaced by the update.
+        replaced = lbf2
+      }
+
+      index = index - 1 // The update replaces the second to last file.
+    }
+
+    // In all cases the bulk file at the current fileId if any is updated.
+    this.replaceBfdAtIndex(index, update)
+    if (truncate) {
+      // If there is a bulk file to be truncated, it becomes the new (reduced) last file.
+      await this.shiftWork(update, truncate, replaced)
+    }
+    if (drop) {
+      this.dropLastBulkFile(drop)
+    }
+
+    const updateInfo = bfdToInfo(update, true)
+    const truncateInfo = truncate ? bfdToInfo(truncate, true) : undefined
+
+    if (this.storage) {
+      // Keep storage in sync.
+      if (update.fileId) {
+        await this.storage.updateBulkFile(update.fileId, updateInfo)
+      }
+      if (truncate && truncateInfo) {
+        if (replaced) {
+          await this.storage.updateBulkFile(truncate.fileId!, truncateInfo)
+        } else {
+          truncateInfo.fileId = undefined // Make sure truncate is a new file.
+          truncate.fileId = await this.storage.insertBulkFile(truncateInfo)
+        }
+      }
+      if (drop && drop.fileId) {
+        await this.storage.deleteBulkFile(drop.fileId)
+      }
+    }
+
+    if (r) {
+      // Update results for logging...
+      r.updated.push(updateInfo)
+      if (truncateInfo) {
+        if (replaced) {
+          r.updated.push(truncateInfo)
+        } else {
+          r.inserted.push(truncateInfo)
+        }
+      }
+      if (drop) {
+        r.dropped.push(bfdToInfo(drop))
+      }
+    }
+
+    this.ensureMaxRetained()
+  }
+
+  private dropLastBulkFile(lbf: BulkFileData): void {
+    delete this.fileHashToIndex[lbf.fileHash]
+    const index = this.bfds.indexOf(lbf)
+    if (index !== this.bfds.length - 1)
+      throw new WERR_INTERNAL(`dropLastBulkFile requires lbf is the current last file.`)
+    this.bfds.pop()
+  }
+
+  /**
+   * Remove work (and headers) from `truncate` that now exists in `update`.
+   * There are two scenarios:
+   * 1. `replaced` is undefined: update is a CDN file that splits an incremental file that must be truncated.
+   * 2. `replaced` is valid: update is a CDN update that replaced an existing CDN file and splits an incremental file that must be truncated.
+   * @param update the new CDN update file.
+   * @param truncate the incremental file to be truncated (losing work which now exists in `update`).
+   * @param replaced the existing CDN file that was replaced by `update` (if any).
+   */
+  private async shiftWork(update: BulkFileData, truncate: BulkFileData, replaced?: BulkFileData): Promise<void> {
+    const updateIndex = this.fileHashToIndex[update.fileHash]
+    // replaced will be valid if the update replaced it and it must become the new last file.
+    // truncateIndex will be updateIndex + 1 if the existing last file is being truncated and update is second to last.
+    const truncateIndex = this.fileHashToIndex[truncate.fileHash]
+    if (truncateIndex !== undefined && truncateIndex !== updateIndex + 1)
+      throw new WERR_INTERNAL(`shiftWork requires update to have replaced truncate or truncate to follow update`)
+    if (truncateIndex !== undefined && !replaced)
+      throw new WERR_INTERNAL(`shiftWork requires valid replaced when update hasn't replaced truncate`)
+
+    truncate.prevHash = update.lastHash!
+    truncate.prevChainWork = update.lastChainWork
+    // truncate.lastChainWork, truncate.lastHash remain unchanged
+    let count = update.count
+    if (replaced) {
+      count -= replaced.count
+    } else {
+      // The truncated file is itself being replaced by the update and must be inserted as a new file.
+      truncate.fileId = undefined
+      this.bfds.push(truncate) // Add the truncated file as a new entry.
+    }
+    truncate.count -= count
+    truncate.firstHeight += count
+
+    truncate.data = truncate.data?.slice(count * 80)
+    delete this.fileHashToIndex[truncate.fileHash]
+    truncate.fileHash = asString(Hash.sha256(asArray(truncate.data!)), 'base64')
+    this.fileHashToIndex[truncate.fileHash] = updateIndex + 1
+  }
+
+  /**
+   *
+   * @param bfd
+   * @returns
+   */
+  private async ensureData(bfd: BulkFileData): Promise<Uint8Array> {
+    if (bfd.data) return bfd.data
+
+    if (this.storage && bfd.fileId) {
+      bfd.data = await this.storage.getBulkFileData(bfd.fileId)
+      if (!bfd.data) throw new WERR_INVALID_PARAMETER('fileId', `valid, data not found for fileId ${bfd.fileId}`)
+    }
+
+    if (!bfd.data && this.fetch && bfd.sourceUrl) {
+      // TODO - restore this change
+      const url = this.fetch.pathJoin(bfd.sourceUrl, bfd.fileName)
+      //const url = this.fetch.pathJoin('http://localhost:8842/blockheaders', bfd.fileName)
+
+      try {
+        bfd.data = await this.fetch.download(url)
+      } catch (err) {
+        bfd.data = await this.fetch.download(url)
+      }
+      if (!bfd.data) throw new WERR_INVALID_PARAMETER('sourceUrl', `data not found for sourceUrl ${url}`)
+    }
+
+    if (!bfd.data) throw new WERR_INVALID_PARAMETER('data', `defined. Unable to retrieve data for ${bfd.fileName}`)
+
+    bfd.mru = Date.now()
+
+    // Validate retrieved data.
+    const fileHash = asString(Hash.sha256(asArray(bfd.data)), 'base64')
+    if (fileHash !== bfd.fileHash)
+      throw new WERR_INVALID_PARAMETER('fileHash', `a match for retrieved data for ${bfd.fileName}`)
+
+    this.ensureMaxRetained()
+    return bfd.data
+  }
+
+  private ensureMaxRetained(): void {
+    if (this.maxRetained === undefined) return
+    let withData = this.bfds.filter(bfd => bfd.data && (bfd.fileId || bfd.sourceUrl))
+    let countToRelease = withData.length - this.maxRetained
+    if (countToRelease <= 0) return
+    const sorted = withData.sort((a, b) => a.mru - b.mru)
+    while (countToRelease-- > 0 && sorted.length > 0) {
+      const oldest = sorted.shift()!
+      // Release the least recently used data
+      oldest.data = undefined // Release the data
+    }
+  }
+
+  async exportHeadersToFs(
+    toFs: ChaintracksFsApi,
+    toHeadersPerFile: number,
+    toFolder: string,
+    sourceUrl?: string,
+    maxHeight?: number
+  ): Promise<void> {
+    const chain = this.chain
+    const toFileName = (i: number) => `${chain}Net_${i}.headers`
+    const toPath = (i: number) => toFs.pathJoin(toFolder, toFileName(i))
+    const toJsonPath = () => toFs.pathJoin(toFolder, `${chain}NetBlockHeaders.json`)
+
+    const toBulkFiles: BulkHeaderFilesInfo = {
+      rootFolder: sourceUrl || toFolder,
+      jsonFilename: `${chain}NetBlockHeaders.json`,
+      headersPerFile: toHeadersPerFile,
+      files: []
+    }
+
+    let range = await this.getHeightRange()
+    if (maxHeight) range = range.intersect(new HeightRange(0, maxHeight))
+    const reader = await this.createReader(range, toHeadersPerFile * 80)
+
+    let firstHeight = 0
+    let lastHeaderHash = '00'.repeat(32)
+    let lastChainWork = '00'.repeat(32)
+
+    let i = -1
+    for (;;) {
+      i++
+      const data = await reader.read()
+      if (!data || data.length === 0) {
+        break
+      }
+
+      const last = validateBufferOfHeaders(data, lastHeaderHash, 0, undefined, lastChainWork)
+
+      await toFs.writeFile(toPath(i), data)
+
+      const fileHash = asString(Hash.sha256(asArray(data)), 'base64')
+      const file: BulkHeaderFileInfo = {
+        chain,
+        count: data.length / 80,
+        fileHash,
+        fileName: toFileName(i),
+        firstHeight,
+        lastChainWork: last.lastChainWork!,
+        lastHash: last.lastHeaderHash,
+        prevChainWork: lastChainWork,
+        prevHash: lastHeaderHash,
+        sourceUrl
+      }
+      toBulkFiles.files.push(file)
+      firstHeight += file.count
+      lastHeaderHash = file.lastHash!
+      lastChainWork = file.lastChainWork!
+    }
+
+    await toFs.writeFile(toJsonPath(), asUint8Array(JSON.stringify(toBulkFiles), 'utf8'))
+  }
+}
+
+interface BulkFileData extends BulkHeaderFileInfo {
+  mru: number
+  fileHash: string
+}
+
+export function selectBulkHeaderFiles(
+  files: BulkHeaderFileInfo[],
+  chain: Chain,
+  maxPerFile: number
+): BulkHeaderFileInfo[] {
+  const r: BulkHeaderFileInfo[] = []
+  let height = 0
+  for (;;) {
+    const choices = files.filter(f => f.firstHeight === height && f.count <= maxPerFile && f.chain === chain)
+    // Pick the file with the maximum count
+    const choice = choices.reduce((a, b) => (a.count > b.count ? a : b), choices[0])
+    if (!choice) break // no more files to select
+    r.push(choice)
+    height += choice.count
+  }
+  return r
+}
+
+function isBdfIncremental(bfd: BulkFileData | BulkHeaderFileInfo): boolean {
+  return bfd.fileName === 'incremental' && !bfd.sourceUrl
+}
+
+function isBdfCdn(bfd: BulkFileData | BulkHeaderFileInfo): boolean {
+  return !isBdfIncremental(bfd)
+}
+
+function bfdToInfo(bfd: BulkFileData, keepData?: boolean): BulkHeaderFileInfo {
+  return {
+    chain: bfd.chain,
+    fileHash: bfd.fileHash,
+    fileName: bfd.fileName,
+    sourceUrl: bfd.sourceUrl,
+    fileId: bfd.fileId,
+    count: bfd.count,
+    prevChainWork: bfd.prevChainWork,
+    lastChainWork: bfd.lastChainWork,
+    firstHeight: bfd.firstHeight,
+    prevHash: bfd.prevHash,
+    lastHash: bfd.lastHash,
+    validated: bfd.validated || false,
+    data: keepData ? bfd.data : undefined
+  }
+}
+
+export interface BulkFileDataManagerMergeResult {
+  unchanged: BulkHeaderFileInfo[]
+  inserted: BulkHeaderFileInfo[]
+  updated: BulkHeaderFileInfo[]
+  dropped: BulkHeaderFileInfo[]
+}
+
+export class BulkFileDataReader {
+  readonly manager: BulkFileDataManager
+  readonly range: HeightRange
+  readonly maxBufferSize: number
+  nextHeight: number
+
+  constructor(manager: BulkFileDataManager, range: HeightRange, maxBufferSize: number) {
+    this.manager = manager
+    this.range = range
+    this.maxBufferSize = maxBufferSize
+    this.nextHeight = range.minHeight
+  }
+
+  /**
+   * Returns the Buffer of block headers from the given `file` for the given `range`.
+   * If `range` is undefined, the file's full height range is read.
+   * The returned Buffer will only contain headers in `file` and in `range`
+   * @param file
+   * @param range
+   */
+  private async readBufferFromFile(file: BulkHeaderFileInfo, range?: HeightRange): Promise<Uint8Array | undefined> {
+    // Constrain the range to the file's contents...
+    let fileRange = new HeightRange(file.firstHeight, file.firstHeight + file.count - 1)
+    if (range) fileRange = fileRange.intersect(range)
+    if (fileRange.isEmpty) return undefined
+    const offset = (fileRange.minHeight - file.firstHeight) * 80
+    const length = fileRange.length * 80
+    return await this.manager.getDataFromFile(file, offset, length)
+  }
+
+  /**
+   * @returns an array containing the next `maxBufferSize` bytes of headers from the files.
+   */
+  async read(): Promise<Uint8Array | undefined> {
+    if (this.nextHeight === undefined || !this.range || this.range.isEmpty || this.nextHeight > this.range.maxHeight)
+      return undefined
+    let lastHeight = this.nextHeight + this.maxBufferSize / 80 - 1
+    lastHeight = Math.min(lastHeight, this.range.maxHeight)
+    let file = await this.manager.getFileForHeight(this.nextHeight)
+    if (!file) throw new WERR_INTERNAL(`logic error`)
+    const readRange = new HeightRange(this.nextHeight, lastHeight)
+    let buffers = new Uint8Array(readRange.length * 80)
+    let offset = 0
+    while (file) {
+      const buffer = await this.readBufferFromFile(file, readRange)
+      if (!buffer) break
+      buffers.set(buffer, offset)
+      offset += buffer.length
+      file = await this.manager.getFileForHeight(file.firstHeight + file.count)
+    }
+    if (!buffers.length || offset !== readRange.length * 80) return undefined
+    this.nextHeight = lastHeight + 1
+    return buffers
+  }
+}