andreagrandi / andreagrandi/draftomen
Prove multi-format drafts end to end
- 主要语言
- Python
- 星标
- 0
- 派生
- 0
- 平均合并
- 1 小时 23 分钟
- 30 天内合并 PR
- 181
描述
## Problem
Slice-level parser, state, scoring, and adapter tests do not prove that a real Arena log reaches the complete Draft Omen workflow. Raw player logs also cannot be committed to make CI perform that check.
## Proposed change
Add deterministic, hand-authored synthetic end-to-end fixtures for all four formats, then run the actual CLI/TUI and desktop workflows locally against the private raw corpus. Document a repeatable private verification command/process without publishing the logs or copied excerpts.
## Acceptance Criteria
- [ ] Synthetic end-to-end fixtures exercise Quick, Premier, Traditional, and Pick-Two from ingestion through presentation.
- [ ] Existing Quick Draft regression behavior remains intact.
- [ ] Private real-log checks parse both Premier samples, both Pick-Two samples, and the verified Traditional sample.
- [ ] Premier/Traditional retain one selected card per logical pick; Pick-Two retains two.
- [ ] Completion, missing-pick, duplicate, unknown-event, and truncation cases are covered at the appropriate boundary.
- [ ] Actual TUI/watch and desktop user-facing workflows are exercised and observed.
- [ ] `git status --untracked-files=all` proves no raw log or private derived artifact enters the changeset.
- [ ] The changelog records completed multi-format support.
## Scope notes
No new production behavior except integration fixes strictly required by the end-to-end workflow.
贡献指南
这个仓库没有索引到贡献指南
评估
这个 Issue 还没有评估数据。