Synthetic editorial fixture, not a real business database or benchmark of ChatGPT.
Dialect SQLite. All datetime TEXT values use YYYY-MM-DD HH:MM:SS and UTC. Money is integer kopecks; no tax/conversion.
Scope: completed orders created_at >=2026-10-01 00:00:00 and <2026-11-01 00:00:00.
All item rows belong to an order; qty and price non-null. Sum qty*price_cents per order.
Net payment = sum of ALL posted amounts attached to selected order, including negative refunds; failed ignored. It is NOT cashflow during October because payments have no date field.
Missing posted payment means zero. Unknown customer_id remains NULL and must not exclude an order.
Q1: one ordered row per qualifying order: id, goods_cents, paid_cents.
Q2: one summary row: order_count, goods_cents, paid_cents.
Q3: selected orders with no posted payment ROW, not net payment zero; return id, customer_id.
Only SELECT for reports, separate setup fixture. No production DB, external credentials or personal data.
