From 1458d94b0ebb283bd78bd1cd002892ebfcc9d143 Mon Sep 17 00:00:00 2001 From: DKChan <524912269@qq.com> Date: Fri, 16 Oct 2026 15:41:01 +0800 Subject: [PATCH] =?UTF-8?q?feat(=E8=BF=AD=E4=BB=A329=20T4):=20L3=20?= =?UTF-8?q?=E5=B1=8F=E5=A3=B3=E5=88=87=E6=8D=A2=E6=8E=A2=E9=92=88=20screen?= =?UTF-8?q?=5Fflow=5Fprobe=EF=BC=8C=E5=85=A5=E8=BD=BB=E9=97=B8=E9=BB=98?= =?UTF-8?q?=E8=AE=A4=E6=AE=B5?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit 新建 game/smoke/screen_flow_probe.gd/.tscn:走 菜单→营地→菜单→冒险→(关卡屏⇄战斗屏)→放弃→终局→菜单 六步主流程与两条回退路径,每步断言屏级互斥 + 流程状态位。与 ui_click_probe 分工:那件走「点某个按钮 有没有反应」的行为走查(含战斗推进与结算数值面),本件只走屏壳与流程状态,不推进战斗不碰结算数值 ——全库 change_scene 零命中、屏切换全靠 visible 互斥,这正是最容易被装配顺序改动悄悄破坏的面。 判据两处订正(首跑实测抓出,都写进注释防复发): ① 屏标记不能裸看 visible——战斗容器 battle_root 在局外也 visible,会把 boot 态误判成「菜单与冒险 同时在场」;冒险屏改取流程状态位 run_scope.is_adventure。 ② 互斥不能用「可见屏计数 ≤1」——终局覆盖层合法叠在冒险内容之上(终局态 is_adventure 尚未收束, 是设计如此);改为冲突对判定,合法叠加仅「终局+冒险」这一对。 变异验证:boot 后强置 camp_overlay.visible=true → 精确命中「菜单与营地同时在场」两条断言见红。 run_all.gd:负面套件(negative=true)全文输出折叠、只回显结论行——它输出里的 Parse Error 是预期产物, 回显出去会让编排层的日志扫描把集成/负面用例的正常输出误判成真实脚本错误(首跑即撞)。 入轻闸默认段(ui_contrast 之后);轻闸 --quick 墙钟 363s → 405s(新探针约 40s),轻闸上限内。 --- game/smoke/screen_flow_probe.gd | 186 ++++++++++++++++++++++++++++++ game/smoke/screen_flow_probe.tscn | 6 + game/tests/run_all.gd | 10 +- tools/probe_gate.sh | 6 + 4 files changed, 206 insertions(+), 2 deletions(-) create mode 100644 game/smoke/screen_flow_probe.gd create mode 100644 game/smoke/screen_flow_probe.tscn diff --git a/game/smoke/screen_flow_probe.gd b/game/smoke/screen_flow_probe.gd new file mode 100644 index 00000000..b633a9c0 --- /dev/null +++ b/game/smoke/screen_flow_probe.gd @@ -0,0 +1,186 @@ +# screen_flow_probe.gd — L3 屏壳切换探针(迭代 29 T4) +# 判的是「跑起来之后,屏与场景的切换对不对」:菜单 ⇄ 营地 ⇄ 冒险 ⇄ 终局 的主流程与回退路径, +# 每一步断言 **屏级互斥**(同一时刻最多一个屏可见)+ 流程状态位落到正确值。 +# 与 ui_click_probe 的分工:那件走的是「点某个按钮有没有反应」的行为走查(含战斗推进与结算数值面); +# 本件只走屏壳与流程状态,不推进战斗、不碰结算数值——屏级互斥是最容易被装配顺序改动悄悄破坏 +# 又最难被单测覆盖的面(全库 change_scene 零命中,屏切换全靠 visible 互斥)。 +# 运行:Godot_console.exe --headless --path . res://smoke/screen_flow_probe.tscn +extends Node + +const ProbeUiUtil := preload("res://smoke/probe_ui_util.gd") + +var main: Node +var _problems: Array[String] = [] + + +func _ready() -> void: + await get_tree().process_frame + main = load("res://presentation/main.tscn").instantiate() + add_child(main) + await get_tree().process_frame + await get_tree().process_frame + + # 隔离存档(全 worktree 共享同一份 user://save.json——探针必须换注,禁碰真实档) + ProbeUiUtil.isolate_save(main, "smoke_screen_flow", false, false, false, false) + await get_tree().process_frame + + await _step_boot_menu() + await _step_menu_to_camp() + await _step_camp_back_to_menu() + await _step_menu_to_adventure() + await _step_level_to_battle_and_back() + await _step_abandon_to_end() + await _step_end_back_to_menu() + + if _problems.is_empty(): + print("[screen_flow_probe] PASS — 六步主流程+回退路径全绿(屏级互斥:任意时刻 ≤1 屏可见)") + get_tree().quit(0) + return + print("[screen_flow_probe] FAIL — %d 处问题:" % _problems.size()) + for p in _problems: + print("[screen_flow_probe] - " + p) + get_tree().quit(1) + + +# ---- 屏级互斥判据 ---- +# 屏标记取法:菜单/营地/终局看各自主内容层的 visible;冒险取流程状态位 run_scope.is_adventure +#(战斗容器 battle_root 在局外也 visible——裸看 visible 会把 boot 态误判成「菜单与冒险同时在场」)。 +# **合法叠加只有一对**:终局覆盖层盖在冒险内容之上(终局态 is_adventure 尚未收束,是设计如此)。 +# 其余任意两屏同时为真=互斥失守。 +func _screen_flags() -> Dictionary: + return { + "菜单": main._start_overlay != null and main._start_overlay.visible, + "营地": main.camp_overlay != null and main.camp_overlay.visible, + "冒险": main.run_scope != null and main.run_scope.is_adventure, + "终局": main._end_overlay != null and main._end_overlay.visible, + } + + +func _flags_text(f: Dictionary) -> String: + var on: Array[String] = [] + for k in ["菜单", "营地", "冒险", "终局"]: + if bool(f[k]): + on.append(k) + return "/".join(on) if not on.is_empty() else "无" + + +# ---- 流程状态位(与可见性是两回事,逐步单独判)---- +func _expect_adventure_flag(step: String, expect: bool) -> void: + var got: bool = main.run_scope != null and main.run_scope.is_adventure + if got != expect: + _problems.append("%s:流程状态位 is_adventure 应为 %s,实测 %s" % [step, str(expect), str(got)]) + + +func _expect_exclusive(step: String, expect: String) -> void: + var f: Dictionary = _screen_flags() + var pairs := [["菜单", "营地"], ["菜单", "冒险"], ["营地", "冒险"], ["终局", "营地"], ["终局", "菜单"]] + for pair in pairs: + var a: String = pair[0] + var b: String = pair[1] + if bool(f[a]) and bool(f[b]): + _problems.append("%s:屏级互斥失守(%s 与 %s 同时在场)" % [step, a, b]) + if not bool(f[expect]): + _problems.append("%s:期望屏「%s」不在场(实测 %s)" % [step, expect, _flags_text(f)]) + return + for k in ["菜单", "营地", "冒险", "终局"]: + if k == expect: + continue + if k == "冒险" and expect == "终局": + continue # 合法叠加:终局覆盖在冒险内容之上 + if bool(f[k]): + _problems.append("%s:屏「%s」不该在场(实测 %s)" % [step, k, _flags_text(f)]) + + +func _frames(n: int = 2) -> void: + for _i in n: + await get_tree().process_frame + + +# ---- ① boot:菜单独占 ---- +func _step_boot_menu() -> void: + _expect_exclusive("①boot", "菜单") + _expect_adventure_flag("①boot", false) + if main.debug_overlay != null and main.debug_overlay.visible: + _problems.append("①boot:调试浮窗 boot 自动开(菜单优先失守)") + + +# ---- ② 菜单 → 营地(真实按钮信号路径)---- +func _step_menu_to_camp() -> void: + var campbtn: Button = ProbeUiUtil.find_button(main._start_overlay, "营地", "prefix") + if campbtn == null: + _problems.append("②菜单→营地:营地按钮缺失") + return + campbtn.pressed.emit() + await _frames() + _expect_exclusive("②菜单→营地", "营地") + + +# ---- ③ 营地 → 菜单(回退路径:营地关闭发自覆盖层自身信号)---- +func _step_camp_back_to_menu() -> void: + if main.camp_overlay == null: + _problems.append("③营地→菜单:营地覆盖层句柄缺失") + return + main.camp_overlay.closed.emit() + await _frames() + _expect_exclusive("③营地→菜单(回退)", "菜单") + + +# ---- ④ 菜单 → 冒险(开新冒险 → 入口停点落关卡地图屏)---- +func _step_menu_to_adventure() -> void: + var lvbtn: Button = ProbeUiUtil.find_button(main._start_overlay, "开始冒险", "contains") + if lvbtn == null: + lvbtn = ProbeUiUtil.find_button(main._start_overlay, "冒险", "contains") + if lvbtn == null: + _problems.append("④菜单→冒险:开始冒险按钮缺失") + return + lvbtn.pressed.emit() + await _frames(3) + _expect_exclusive("④菜单→冒险", "冒险") + _expect_adventure_flag("④菜单→冒险", true) + if main.run_scope != null and main.run_scope.adventure == null: + _problems.append("④菜单→冒险:冒险实例未建立(入口停点未达)") + if main.level_screen != null and not main.level_screen.visible: + _problems.append("④菜单→冒险:入口停点未落在关卡地图屏(level_screen 不可见)") + + +# ---- ⑤ 冒险内屏切换:关卡地图屏 ⇄ 战斗屏(屏内两态,屏级仍独占)---- +func _step_level_to_battle_and_back() -> void: + if not main.level_screen.visible: + _problems.append("⑤冒险内切换:前置不在关卡地图屏") + return + main._exit_level_screen() + await _frames() + if main.level_screen.visible: + _problems.append("⑤冒险内切换:退出关卡屏后 level_screen 仍可见") + if main._battle_root != null and not main._battle_root.visible: + _problems.append("⑤冒险内切换:战斗屏未接手(battle_root 不可见)") + _expect_exclusive("⑤冒险内切换(关卡→战斗)", "冒险") + main._enter_level_screen() + await _frames() + if not main.level_screen.visible: + _problems.append("⑤冒险内切换:回到关卡地图屏失败") + _expect_exclusive("⑤冒险内切换(战斗→关卡)", "冒险") + + +# ---- ⑥ 放弃 → 终局(放弃路径=回退路径的非正常出口)---- +func _step_abandon_to_end() -> void: + if main.run_scope == null or main.run_scope.adventure == null: + _problems.append("⑥放弃→终局:前置无冒险实例") + return + main._on_abandon_requested() + await _frames(4) + _expect_exclusive("⑥放弃→终局", "终局") + if main._end_overlay != null and not main._end_overlay.visible: + _problems.append("⑥放弃→终局:终局覆盖层未开") + + +# ---- ⑦ 终局 → 菜单(终局出口返回开始界面)---- +func _step_end_back_to_menu() -> void: + if main._end_overlay == null or not main._end_overlay.visible: + _problems.append("⑦终局→菜单:前置不在终局") + return + main._end_overlay.continue_pressed.emit() + await _frames(3) + _expect_exclusive("⑦终局→菜单", "菜单") + if main.run_scope != null and main.run_scope.is_adventure: + _problems.append("⑦终局→菜单:冒险态未收束(run_scope 仍标 is_adventure)") diff --git a/game/smoke/screen_flow_probe.tscn b/game/smoke/screen_flow_probe.tscn new file mode 100644 index 00000000..5d847fb3 --- /dev/null +++ b/game/smoke/screen_flow_probe.tscn @@ -0,0 +1,6 @@ +[gd_scene load_steps=2 format=3] + +[ext_resource type="Script" path="res://smoke/screen_flow_probe.gd" id="1"] + +[node name="ScreenFlowProbe" type="Node"] +script = ExtResource("1") diff --git a/game/tests/run_all.gd b/game/tests/run_all.gd index 8dd341f8..4c48139e 100644 --- a/game/tests/run_all.gd +++ b/game/tests/run_all.gd @@ -93,8 +93,14 @@ func _run_suite(suite: Dictionary, idx: int) -> int: var out := PackedStringArray() var code: int = OS.execute("cmd.exe", PackedStringArray(["/c", cmd]), out, true, false) var text := FileAccess.get_file_as_string(log) if FileAccess.file_exists(log) else "" - print(text) - if not bool(suite.get("negative", false)) and _has_script_error(text): + var negative := bool(suite.get("negative", false)) + # 负面套件(坏表/违规输入是它的测试目标)全文折叠,只回显结论行—— + # 它输出里的 Parse Error 是预期产物,回显出去会让编排层的日志扫描误判成真实脚本错误。 + if negative: + print("[run_all] %s:负面套件——全文输出已折叠(输出中的错误为其测试目标,非缺陷)" % path) + else: + print(text) + if not negative and _has_script_error(text): push_error("[run_all] FAIL %s — 输出含脚本错误标记:%s" % [path, _script_error_line(text)]) return 1 if code != expect: diff --git a/tools/probe_gate.sh b/tools/probe_gate.sh index 484904e7..7b69c3b4 100644 --- a/tools/probe_gate.sh +++ b/tools/probe_gate.sh @@ -110,6 +110,12 @@ st=$?; kill_orphans grep -E "PASS|FAIL" "$TMP/ctp.log" | head -12 verdict "ui_contrast_probe" "$TMP/ctp.log" "$st" || _rc=1 +echo "======== screen_flow_probe(L3 屏壳切换·迭代 29 T4) ========" +timeout 120 "$G" --headless --path "$GAME_WIN" res://smoke/screen_flow_probe.tscn >"$TMP/sfp.log" 2>&1 +st=$?; kill_orphans +grep -E "screen_flow_probe" "$TMP/sfp.log" | tail -3 +verdict "screen_flow_probe" "$TMP/sfp.log" "$st" "screen_flow_probe" || _rc=1 + if [ "$WITH_CLICK" -eq 1 ]; then echo "======== ui_click_probe 行为走查 ========" timeout 240 "$G" --headless --path "$GAME_WIN" res://smoke/ui_click_probe.tscn >"$TMP/clk.log" 2>&1