SoundGame/tools/check_all_assets.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""全项目音频资产终检（R5）。

检查 11 个游戏 web/<game>/audio/（批次 3 起含 hunt；hunt 的 bot 喊话落
audio/sfx/npc_*，不在 voice 平铺豁免范围，23-hunt.md §15 裁决 7）：
  1. 每个 .ogg 都有同名 .m4a（反之亦然）
  2. voice/male 与 voice/female 的 id 集合一致（storyline/rpg 的平铺 NPC/角色句除外）
  3. 键盘变体目录 voice/<gender>/key/（P1 路线 a）：
     male/female 两性别 key 集合一致；key 集合 ⊆ 主（性别目录）集合——变体只
     覆盖含操作词的句子，其余 id 由引擎回退主资产
  4. 无 0 字节文件；ogg/m4a 均可被 ffprobe 解码且时长 > 0
输出 Markdown 报告到 stdout（重定向进 docs/audio/final-asset-check.md）。
"""
import os
import subprocess
import sys

ROOT = "/home/xsl/SoundGame/web"
GAMES = ["storyline", "escape", "simon", "rhythm", "space", "rpg", "radio", "dungeon",
         "abyss", "detective", "hunt", "cipher", "teahouse", "keeper", "bridge"]
# 平铺 voice 文件（角色/NPC，单音色，不做性别集合一致性要求）
FLAT_OK = {"storyline", "rpg", "radio", "abyss", "detective",
           "cipher", "teahouse", "keeper", "bridge"}


def ids(d, ext):
    if not os.path.isdir(d):
        return set()
    return {f[: -len(ext)] for f in os.listdir(d) if f.endswith(ext)}


def main():
    problems = []
    print("# 全项目音频资产终检（R5 建立，批次 1 起含 radio/dungeon，批次 2 起含 abyss/detective，批次 3 起含 hunt，2026-09-13；批次 3/4 起含 cipher/teahouse/keeper/bridge，2026-09-14）\n")
    print("| 游戏 | ogg | m4a | 0字节 | 解码失败 | voice male/female id 一致 | key male/female 一致且 ⊆ 主集 |")
    print("|---|---|---|---|---|---|---|")
    for g in GAMES:
        base = os.path.join(ROOT, g, "audio")
        oggs, m4as, zero, bad = [], [], [], []
        for dirpath, _dirs, files in os.walk(base):
            for f in files:
                p = os.path.join(dirpath, f)
                if os.path.getsize(p) == 0:
                    zero.append(p)
                if f.endswith(".ogg"):
                    oggs.append(p)
                elif f.endswith(".m4a"):
                    m4as.append(p)
        # 双格式配对
        o_set = {p[:-4] for p in oggs}
        m_set = {p[:-4] for p in m4as}
        for p in sorted(o_set - m_set):
            problems.append(f"{g}: 缺 m4a —— {p}.ogg")
        for p in sorted(m_set - o_set):
            problems.append(f"{g}: 缺 ogg —— {p}.m4a")
        # 解码校验（抽查全量 ffprobe）
        for p in oggs + m4as:
            r = subprocess.run(["ffprobe", "-v", "error", "-show_entries", "format=duration",
                                "-of", "csv=p=0", p], capture_output=True, text=True)
            try:
                if float(r.stdout.strip()) <= 0:
                    bad.append(p)
            except ValueError:
                bad.append(p)
        for p in zero:
            problems.append(f"{g}: 0 字节 —— {p}")
        for p in bad:
            problems.append(f"{g}: 解码失败/时长为 0 —— {p}")
        # voice 性别集合一致性
        vm = ids(os.path.join(base, "voice", "male"), ".ogg")
        vf = ids(os.path.join(base, "voice", "female"), ".ogg")
        if vm == vf and vm:
            pair = f"✅ 各 {len(vm)}"
        else:
            pair = f"❌ male={len(vm)} female={len(vf)}"
            for i in sorted(vm - vf):
                problems.append(f"{g}: voice 缺 female/{i}")
            for i in sorted(vf - vm):
                problems.append(f"{g}: voice 缺 male/{i}")
        # 键盘变体目录（P1 路线 a）：两性别 key 集合一致 + key ⊆ 主集合
        km = ids(os.path.join(base, "voice", "male", "key"), ".ogg")
        kf = ids(os.path.join(base, "voice", "female", "key"), ".ogg")
        if km == kf and km <= vm:
            kpair = f"✅ 各 {len(km)} ⊆ 主集" if km else "—（无变体目录）"
        else:
            kpair = f"❌ male={len(km)} female={len(kf)}"
            for i in sorted(km - kf):
                problems.append(f"{g}: key 变体缺 female/key/{i}")
            for i in sorted(kf - km):
                problems.append(f"{g}: key 变体缺 male/key/{i}")
            for i in sorted((km | kf) - (vm | vf)):
                problems.append(f"{g}: key 变体 {i} 不在主集合（key 必须 ⊆ 主集）")
        flat = ids(os.path.join(base, "voice"), ".ogg")
        flat_note = f"（平铺 {len(flat)} 个角色/NPC 句，豁免）" if g in FLAT_OK and flat else \
                    (f"（⚠️ 非豁免游戏存在平铺 {len(flat)} 个）" if flat and g not in FLAT_OK else "")
        print(f"| {g} | {len(oggs)} | {len(m4as)} | {len(zero)} | {len(bad)} | {pair}{flat_note} | {kpair} |")
    print()
    if problems:
        print(f"## 问题清单（{len(problems)} 条）\n")
        for p in problems:
            print(f"- {p}")
        return 1
    print("## 结论\n\n全部通过：双格式配对完整、无 0 字节、全部可解码、voice 性别目录 id 集合一致（storyline/rpg/radio/abyss/detective 平铺角色句按约定豁免）、键盘变体目录两性别一致且 ⊆ 主集合。")
    return 0


if __name__ == "__main__":
    sys.exit(main())