bindbox-game/.planning/ROADMAP.md
win b99bcbd06f docs(01-core-pnl-functions): create phase 1 plans
4 plans across 3 waves:
- 01-01 (wave 1): package scaffold — types.go, service.go, service_test.go
- 01-02 (wave 2): QueryUserProfitLoss — query_user.go + user integration tests
- 01-03 (wave 2, parallel): QueryActivityProfitLoss — query_activity.go + activity tests
- 01-04 (wave 3): phase verification — static checks + full test suite gate

Covers all 20 Phase 1 requirements: PNL-01..08, DIM-01..04, RET-01/03,
AST-01, QUA-01..05.
2026-03-21 17:27:58 +08:00

4.3 KiB
Raw Blame History

Roadmap: Bindbox Game 盈亏统计函数

Overview

Two reusable service-layer functions — QueryUserProfitLoss and QueryActivityProfitLoss — are built in a new internal/service/finance/ package. Phase 1 delivers correct, working functions covering the full P&L computation path. Phase 2 populates the per-asset-type breakdown slice, including Fragment synthesis cost. The result is a clean, testable service that callers can invoke without touching HTTP handler logic.

Phases

Phase Numbering:

  • Integer phases (1, 2, 3): Planned milestone work
  • Decimal phases (2.1, 2.2): Urgent insertions (marked with INSERTED)

Decimal phases appear between their surrounding integers in numeric order.

  • Phase 1: Core P&L Functions - Scaffold the finance package and deliver working QueryUserProfitLoss / QueryActivityProfitLoss with correct revenue, cost, and profit
  • Phase 2: Per-Asset-Type Breakdown - Populate the ProfitLossBreakdown slice for all 5 asset types, including Fragment synthesis cost from its own table

Phase Details

Phase 1: Core P&L Functions

Goal: Callers can invoke QueryUserProfitLoss and QueryActivityProfitLoss and receive a correct ProfitLossResult with total revenue, cost, profit, and profit rate — with all edge cases handled Depends on: Nothing (first phase) Requirements: PNL-01, PNL-02, PNL-03, PNL-04, PNL-05, PNL-06, PNL-07, PNL-08, DIM-01, DIM-02, DIM-03, DIM-04, RET-01, RET-03, AST-01, QUA-01, QUA-02, QUA-03, QUA-04, QUA-05 Success Criteria (what must be TRUE):

  1. Calling QueryUserProfitLoss with a list of user IDs returns a ProfitLossResult where Revenue equals actual_amount + discount_amount for non-refunded orders only, and game-pass orders contribute draw_count × activity_price instead of cash revenue
  2. Calling QueryActivityProfitLoss with an activity ID returns a ProfitLossResult where Revenue is attributed directly to the order's activity (1:1 per D-01 — no proration subquery)
  3. Both functions return an error (not silent zero) when any db.Scan() call fails
  4. Passing nil for StartTime/EndTime applies no time filter; passing nil/0 for AssetType returns aggregated totals across all asset types
  5. The package contains no call to GetDbW() — all queries route through the injected DbR handle; voided inventory (remark LIKE '%void%' or status=2) and refunded orders (status=3,4) are excluded from cost and revenue respectively Plans: 4 plans

Plans:

  • 01-01-PLAN.md — Package scaffold: types.go (AssetType enum + param/result structs) + service.go (interface + read-only constructor) + service_test.go (SQLite test infrastructure)
  • 01-02-PLAN.md — QueryUserProfitLoss: query_user.go with 4 fan-out scans (revenue, inventory cost, points cost, coupon cost) + integration tests
  • 01-03-PLAN.md — QueryActivityProfitLoss: query_activity.go with 4 fan-out scans attributed to activity dimension + integration tests
  • 01-04-PLAN.md — Phase 1 verification: full test suite + static checks (no GetDbW, fan-out count, finance functions reused, int64 monetary types)

Phase 2: Per-Asset-Type Breakdown

Goal: The ProfitLossBreakdown slice in every ProfitLossResult contains one entry per relevant asset type (Points, Coupon, ItemCard, Product, Fragment), with correct per-type cost including Fragment synthesis cost sourced from fragment_synthesis_logs Depends on: Phase 1 Requirements: AST-02, AST-03, RET-02 Success Criteria (what must be TRUE):

  1. A ProfitLossResult for an activity that awarded Points, Coupons, and Items contains exactly one ProfitLossBreakdown entry per asset type that had non-zero cost, with the correct cost value for each
  2. Fragment cost in the breakdown is sourced from fragment_synthesis_logs using the verified join path (not from user_inventory), and matches manual spot-checks against the database
  3. Summing all breakdown entries' Cost fields equals the total Cost field in the parent ProfitLossResult (no gaps or double-counting between breakdown and totals) Plans: TBD

Progress

Execution Order: Phases execute in numeric order: 1 → 2

Phase Plans Complete Status Completed
1. Core P&L Functions 0/4 Planning complete -
2. Per-Asset-Type Breakdown 0/TBD Not started -