fix(dump): put _updated_at last in whole-table downloads

[luccasmmg]

Problem

QA report: after appending records via the API and downloading the resource as CSV, the first two columns were _id and _updated_at. _updated_at should be the last column.

Cause

BigQuery stores the system columns first (_id, _updated_at, then user fields). GET /dump/{resource_id} built both its EXPORT DATA SELECT list and the CSV header from that physical order. datastore_search already moves _updated_at after the user columns; dump did not.

Fix

• New _dump_column_order(schema) in bigquery/export.py moves _updated_at last.
• dump() uses it for both the SELECT list and the CSV header, so the two stay aligned. This applies to csv, gzip, ndjson and parquet.
• /dump/query is unchanged: it keeps the user's SQL projection order.

Notes

• Cached exports are keyed on table.modified, so an unchanged table can still serve a previously cached file with the old order until the table is written to again or the cache entry ages out.
• Not tested against real BigQuery; covered by unit tests only.

Tests

• Added test_dump_column_order_moves_updated_at_last and test_dump_column_order_without_updated_at_is_unchanged.
• pytest tests/test_datastore_dump.py tests/engines: 124 passed.

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

• Configuration used: defaults
• Review profile: CHILL
• Plan: Advanced
• Run ID: e22eafe5-daba-41cb-b66b-112f4a09009d

---

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

---

_{Comment @coderabbitai help to get the list of available commands.}