卷压缩模式: metadata --vol-compress (16MB切卷+逐卷7z+逐卷加密, context=vol:N 派生nonce, 服务端零改动) + 下载端 vol_mode 逐卷解密+解压+拼接

This commit is contained in:
lou
2026-08-10 20:37:04 +08:00
parent 66cb9cd22e
commit d5e5732308
2 changed files with 151 additions and 0 deletions
+96
View File
@@ -126,6 +126,96 @@ def sha256_of(path: str | Path) -> str:
return h.hexdigest()
def _build_vol_init(src_path: Path, args: argparse.Namespace) -> int:
"""卷压缩模式: 流式切卷 -> 逐卷 7z 压缩 -> 逐卷加密
每卷独立 context (7z-encrypt:v1:vol:N), nonce 由 context 派生,
下载端按卷号重建 context 解密, 服务端零改动。
"""
import subprocess
import tempfile
from crypto import CONTEXT
vol_size = args.vol_compress * 1024 * 1024
engine = CryptoEngine()
try:
engine.get_key(args.key_id)
except KeyNotFoundError:
engine.generate_key(args.key_id)
print(f"[密钥库] 已生成新密钥 -> {engine.keyring_path}")
# 卷工作目录: 磁盘优先 (/root), Android /tmp 是 tmpfs 但卷小无碍, 统一临时目录
workdir = Path(tempfile.gettempdir())
init_path = src_path.with_name(src_path.name + ".init.json")
manifest_path = src_path.with_name(src_path.name + "_manifest.json")
src_st = src_path.stat().st_size
print(f"=== 卷压缩模式: {src_path.name} ({src_st / 1048576:.1f} MB), 每卷 {args.vol_compress} MB ===")
chunks: list[dict[str, Any]] = []
total_sha = hashlib.sha256()
idx = 0
pbar = tqdm(total=src_st, desc="[压缩+加密]", unit="B", unit_scale=True, leave=False)
with open(src_path, "rb") as src:
while True:
raw = src.read(vol_size)
if not raw:
break
idx += 1
raw_path = workdir / f"szvol_{src_path.name}_{idx:04d}.bin"
z7_path = raw_path.with_suffix(".7z")
chunk_path = src_path.with_name(f"{src_path.name}.chunk{idx:04d}")
try:
raw_path.write_bytes(raw)
r = subprocess.run(
["7z", "a", "-y", "-bd", "-mx=9", str(z7_path), str(raw_path)],
capture_output=True, text=True,
)
if r.returncode != 0:
raise MetadataError(f"{idx} 7z 压缩失败: {r.stderr[-200:]}")
raw_path.unlink()
with open(z7_path, "rb") as s, open(chunk_path, "wb") as d:
engine.encrypt_stream(
s, d, args.key_id, context=f"{CONTEXT.decode()}:vol:{idx}".encode())
z7_path.unlink()
with open(chunk_path, "rb") as f:
while True:
b = f.read(1 << 20)
if not b:
break
total_sha.update(b)
chunks.append({
"index": idx,
"size": chunk_path.stat().st_size,
"sha256": sha256_of(chunk_path),
"filename": str(chunk_path),
})
finally:
raw_path.unlink(missing_ok=True)
z7_path.unlink(missing_ok=True)
pbar.update(len(raw))
pbar.close()
with open(manifest_path, "w", encoding="utf-8") as f:
json.dump(chunks, f, indent=2, ensure_ascii=False)
enc_params: dict[str, Any] = {
"alg": "cobblestone-aes256gcm", "key_id": args.key_id,
"context": CONTEXT.decode(), "compressed": True, "vol_mode": True,
}
file_name = args.file_name or src_path.name
if not args.no_encrypt_name:
file_name = engine.encrypt_name(file_name)
init_json = build_init_json(
file_name=file_name,
file_size=src_st,
total_sha256=total_sha.hexdigest(),
enc=enc_params,
chunks=chunks,
)
with open(init_path, "w", encoding="utf-8") as f:
json.dump(init_json, f, indent=2, ensure_ascii=False)
print(f"[完成] {idx} 卷 -> {init_path.name}, 清单 -> {manifest_path.name}")
return 0
def main() -> int:
ap = argparse.ArgumentParser(description="元数据生成: 加密 -> 分卷 -> 组装 init json")
ap.add_argument("file", nargs="?", default="index.mp4", help="要处理的文件 (默认 index.mp4)")
@@ -135,6 +225,8 @@ def main() -> int:
ap.add_argument("--file-name", help="覆盖 init json 的 file_name (压缩场景用原名)")
ap.add_argument("--compressed", action="store_true",
help="标记 7z 压缩 (enc_params 加 compressed, 下载端自动解压)")
ap.add_argument("--vol-compress", type=int, default=0,
help="卷压缩模式 MB: 流式切卷(每卷MB) -> 逐卷7z压缩 -> 逐卷加密 (规避大文件OOM)")
ap.add_argument("--no-encrypt-name", action="store_true",
help="不加密文件名 (默认加密成 enc: 前缀)")
args = ap.parse_args()
@@ -144,6 +236,10 @@ def main() -> int:
print(f"[错误] 文件不存在: {src_path}")
return 1
# 卷压缩模式: 流式切卷 -> 逐卷 7z 压缩 -> 逐卷加密 (每卷内存 16MB 级, 规避大文件 OOM)
if args.vol_compress > 0:
return _build_vol_init(src_path, args)
chunk_size = args.chunk_size * 1024 * 1024
enc_path = src_path.with_name(src_path.name + ".enc")
manifest_path = src_path.with_name(src_path.name + "_manifest.json")
+55
View File
@@ -212,6 +212,10 @@ class TransferClient:
elif dest.is_dir():
dest = dest / name
assert dest is not None
# 卷压缩模式: 每卷独立解密+7z解压+拼接明文 (卷1直接处理, 其余 aria2/逐卷拉取)
if enc_params.get("vol_mode"):
self._download_vol_mode(file_id, dest, chunk_count, enc_params)
return dest, enc_params
outer = tqdm(total=chunk_count, desc="[下载]", unit="",
position=0, leave=False)
with open(dest, "wb") as f:
@@ -276,6 +280,45 @@ class TransferClient:
outer.update(1)
return done
def _download_vol_mode(self, file_id: str, dest: Path, chunk_count: int,
enc_params: dict[str, Any]) -> None:
"""卷压缩模式下载: 逐卷拉密文 -> 逐卷解密(context=vol:N) -> 7z x -> 拼接明文
每卷 16MB 级内存, 规避大文件 OOM; 卷 context 与上传端对称派生。
"""
import subprocess
import tempfile
from crypto import CryptoEngine
base_ctx = enc_params.get("context") or "7z-encrypt:v1"
key_id = enc_params.get("key_id") or "default_key"
engine = CryptoEngine(_load_keyring_path())
with tempfile.TemporaryDirectory(prefix="sz-vol-") as td:
td = Path(td)
open(dest, "wb").close()
pbar = tqdm(total=chunk_count, desc="[下载+解密]", unit="", leave=False)
for idx in range(1, chunk_count + 1):
enc_path = td / f"v{idx}.enc"
with self._open_chunk(file_id, idx) as resp:
with open(enc_path, "wb") as f:
shutil.copyfileobj(resp, f, 1 << 20)
z7_path = td / f"v{idx}.7z"
with open(enc_path, "rb") as s, open(z7_path, "wb") as d:
engine.decrypt_stream(s, d, key_id, f"{base_ctx}:vol:{idx}".encode())
out_dir = td / f"out{idx}"
r = subprocess.run(
["7z", "x", "-y", "-bd", f"-o{out_dir}", str(z7_path)],
capture_output=True, text=True,
)
if r.returncode != 0:
raise TransferError(f"{idx} 7z 解压失败: {r.stderr[-200:]}")
with open(dest, "ab") as f:
for p in sorted(out_dir.iterdir()):
if p.is_file():
with open(p, "rb") as s:
shutil.copyfileobj(s, f, 1 << 20)
pbar.update(1)
pbar.close()
def _aria2_pull(self, file_id: str, chunk_range: range, workdir: Path) -> list[Path]:
"""aria2 并发拉取卷文件 (多任务打满带宽), 返回按序的卷文件列表
@@ -662,6 +705,18 @@ def main() -> int:
final = dest_arg
else:
final = Path(name)
if enc_params.get("vol_mode"):
# 卷压缩模式: download() 内部已逐卷解密+解压+拼接明文到 enc_path
if dest_arg and dest_arg.is_dir():
final = dest_arg / name
elif dest_arg:
final = dest_arg
else:
final = Path(name)
final.parent.mkdir(parents=True, exist_ok=True)
shutil.move(enc_path, final)
print(f"[下载] 完成 -> {final} ({final.stat().st_size / 1048576:.1f} MB, 卷解密+解压还原)")
return 0
from crypto import CryptoEngine
engine = CryptoEngine(_load_keyring_path())
if enc_params.get("compressed"):