XLSX worker protocol

1import type { CompactSheet } from "@/lib/xlsx-workbook";
2 
3export interface XlsxParseLimits {
4  maxRowMajorIndex: number;
5  maxNonEmptyCells: number;
6  maxTextChars: number;
7}
8 
9export const XLSX_PARSE_LIMITS: XlsxParseLimits = {
10  maxRowMajorIndex: 0xffffffff,
11  maxNonEmptyCells: 1_000_000,
12  maxTextChars: 50_000_000,
13};
14 
15export type XlsxWorkerErrorCode =
16  | "parse_failed"
17  | "range_too_large"
18  | "too_many_cells"
19  | "text_too_large";
20 
21export type XlsxWorkerRequest = {
22  type: "parse_workbook";
23  buffer: ArrayBuffer;
24};
25 
26export type XlsxWorkerResponse =
27  | { type: "workbook"; sheets: CompactSheet[] }
28  | { type: "error"; message: string; code: XlsxWorkerErrorCode };
29 
30export class XlsxWorkerError extends Error {
31  constructor(
32    readonly code: XlsxWorkerErrorCode,
33    message: string,
34  ) {
35    super(message);
36    this.name = "XlsxWorkerError";
37  }
38}
39 
40// ZIP local-file-header variants ("PK\x03\x04", empty, and spanned archives)
41// cover .xlsx/.xlsm/.xlsb/.ods. The OLE2 / Compound File Binary header covers
42// legacy binary .xls.
43const SPREADSHEET_CONTAINER_PREFIXES: ReadonlyArray<ReadonlyArray<number>> = [
44  [0x50, 0x4b, 0x03, 0x04],
45  [0x50, 0x4b, 0x05, 0x06],
46  [0x50, 0x4b, 0x07, 0x08],
47  [0xd0, 0xcf, 0x11, 0xe0, 0xa1, 0xb1, 0x1a, 0xe1],
48];
49 
50function hasBytePrefix(
51  bytes: Uint8Array,
52  prefix: ReadonlyArray<number>,
53): boolean {
54  if (bytes.length < prefix.length) return false;
55  for (let i = 0; i < prefix.length; i++) {
56    if (bytes[i] !== prefix[i]) return false;
57  }
58  return true;
59}
60 
61/**
62 * True when the buffer begins with a recognized spreadsheet container
63 * signature: a ZIP archive (.xlsx/.xlsm/.xlsb/.ods) or an OLE2 compound file
64 * (legacy .xls). This guards SheetJS's lenient format sniffer, which would
65 * otherwise coerce arbitrary bytes — a mislabeled PDF, image, or text file —
66 * into a degenerate single-cell sheet instead of surfacing a parse error.
67 */
68export function isSpreadsheetContainer(buffer: ArrayBuffer): boolean {
69  const bytes = new Uint8Array(buffer);
70  return SPREADSHEET_CONTAINER_PREFIXES.some((prefix) =>
71    hasBytePrefix(bytes, prefix),
72  );
73}