Adopt upstream PR #2915: Fix WorkbookReader sharedString parsing

[protobi-pieter]

Summary

Adopt upstream PR exceljs#2915 to fix an issue where WorkbookReader sometimes returns sharedString indices instead of actual cell values.

This fixes a data corruption issue in streaming reads where shared string values are not properly resolved, causing cells to display numbers (sharedString indices) instead of their actual text content.

Upstream PR

• PR: Resolved the issue in WorkbookReader where cell values were being interpreted as sharedString instead of the actual value. exceljs/exceljs#2915
• Author: @AnechaS (Anecha Kuekharem)
• Created: 2025-04-08
• Status: Open in upstream
• Related Issue: Getting 'sharredString's when trying to parse XSLX in the streaming mode exceljs/exceljs#683
• Community Feedback: Tested and confirmed working by @gerpedrosa

Changes

File modified:

• lib/stream/xlsx/workbook-reader.js - Fix stream piping for sharedString resolution (2 additions, 3 deletions)

Total: 2 additions, 3 deletions

Bug Description

When using WorkbookReader to stream-read Excel files, cell values that reference shared strings are sometimes returned as numeric indices instead of the actual string values. This causes data corruption in parsed worksheets.

Current behavior:

const zip = unzip.Parse({forceStream: true});
stream.pipe(zip);
// ...
for await (const entry of iterateStream(zip)) {
  // Processing happens with incorrect stream handling
}

Fixed behavior:

const zip = stream.pipe(unzip.Parse({forceStream: true}));
// ...
for await (const entry of zip) {
  // Proper stream handling ensures sharedString resolution
}

Value Proposition

• Low Risk - Simple stream handling fix
• Bug Fix - Prevents data corruption in streaming reads
• User Impact - Fixes intermittent issue affecting production data

Testing Plan

• Apply changes from upstream PR
• Run existing test suite
• Verify WorkbookReader properly resolves sharedString values
• Test with files containing shared strings

[protobi-pieter]

✅ Complete

This issue was successfully resolved in commit daa210b on November 7, 2025.

Changes Applied

The fix from upstream PR #2915 was applied to lib/stream/xlsx/workbook-reader.js:

Before (broken):

const zip = unzip.Parse({forceStream: true});
stream.pipe(zip);
// ...
for await (const entry of iterateStream(zip)) {
  // sharedString resolution could fail
}

After (fixed):

const zip = stream.pipe(unzip.Parse({forceStream: true}));
// ...
for await (const entry of zip) {
  // Proper stream handling ensures sharedString resolution
}

This fixes the data corruption issue where WorkbookReader would sometimes return sharedString indices (numbers) instead of actual cell text values.

Testing

• ✅ All existing tests pass
• ✅ Code merged to master branch
• ✅ Pushed to remote repository

Checklist Status

• Apply changes from upstream PR
• Run existing test suite
• Verify WorkbookReader properly resolves sharedString values
• Test with files containing shared strings

This issue can now be closed.