Issue Review: cloudflare/workers-sdk#9578

Summary

D1 export fails with cell.replace is not a function error when database contains values with mismatched data types.

Findings

• Created: 2025-06-12
• Updated: 2025-07-07
• Version: wrangler 4.16.1 / 4.19.2 / 4.20.0 -> 4.60.0 (current)
• Component: D1 (wrangler)
• Labels: bug, internal, d1
• Comments: 0

Key Evidence

• PR #9866 (merged 2025-07-18) and backport PR #10035 (merged 2025-07-22) fix this exact issue: "fix: escape column names and handle mismatched data types in D1 SQL dump"
• The fix addresses the root cause: the code was checking the column's declared type (colType) instead of the actual runtime type (cellType) before formatting cells
• When a non-string value was encountered in a TEXT column, the code would call cell.replace() on a non-string, causing the error
• The fix changes the logic to check typeof cell === "string" instead of colType === "TEXT"
• Issue was reported on 2025-06-12, fix was merged on 2025-07-18 (36 days after the issue was created)
• The PR explicitly fixes related issues #8377, #6357, #5920, and mentions #8226

Recommendation

Status: NEEDS VERIFICATION

Reasoning: The cell.replace is not a function error is fixed by PR #9866, which handles SQLite's dynamic typing properly. The fix was merged over 6 months ago and is available in wrangler versions after 4.20.0 (which the reporter tested). However, since the issue was not explicitly closed by the PR and the fix was for local/miniflare D1 export (not remote), there may be a different code path for remote exports that still has this bug.

Action: Ask the reporter to verify with the latest wrangler version (4.60.0). If the issue persists for remote D1 exports specifically, this may indicate a separate bug in the remote export API path.

Suggested Comment

Hi @axww! A fix for the cell.replace is not a function error was merged in PR #9866 and backported in PR #10035. This fix addresses SQLite's dynamic typing where column values may have different runtime types than their declared column types.

Could you please test with the latest wrangler version (4.60.0) and let us know if the issue is resolved?

npm install wrangler@latest
npx wrangler d1 export <database-name> --remote --output=export.sql

If the issue persists, it would be helpful to know:

1. What data types are in your database tables (particularly any columns that might have mixed types)?
2. Does the export work for a smaller database or specific tables?

Thank you!