option to save a copy of corrupted uploads

copyparty/__main__.py

@@ -750,6 +750,9 @@ def run_argparse(argv: list[str], formatter: Any, retry: bool) -> argparse.Names
     ap2.add_argument("--stackmon", metavar="P,S", type=u, help="write stacktrace to Path every S second, for example --stackmon=./st/%%Y-%%m/%%d/%%H%%M.xz,60")
     ap2.add_argument("--log-thrs", metavar="SEC", type=float, help="list active threads every SEC")
     ap2.add_argument("--log-fk", metavar="REGEX", type=u, default="", help="log filekey params for files where path matches REGEX; '.' (a single dot) = all files")
+    ap2.add_argument("--bak-flips", action="store_true", help="[up2k] if a client uploads a bitflipped/corrupted chunk, store a copy according to --bf-nc and --bf-dir")
+    ap2.add_argument("--bf-nc", metavar="NUM", type=int, default=200, help="bak-flips: stop if there's more than NUM files at --kf-dir already; default: 6.3 GiB max (200*32M)")
+    ap2.add_argument("--bf-dir", metavar="PATH", type=u, default="bf", help="bak-flips: store corrupted chunks at PATH; default: folder named 'bf' wherever copyparty was started")
     # fmt: on
 
     ap2 = ap.add_argument_group("help sections")

copyparty/httpcli.py

@@ -82,6 +82,7 @@ from .util import (
 
 try:
     from typing import Any, Generator, Match, Optional, Pattern, Type, Union
+    import typing
 except:
     pass
 
@@ -920,6 +921,38 @@ class HttpCli(object):
         self.reply(t.encode("utf-8"))
         return True
 
+    def bakflip(self, f: typing.BinaryIO, ofs: int, sz: int, sha: str) -> None:
+        if not self.args.bak_flips or self.args.nw:
+            return
+
+        sdir = self.args.bf_dir
+        fp = os.path.join(sdir, sha)
+        if bos.path.exists(fp):
+            return self.log("no bakflip; have it", 6)
+
+        if not bos.path.isdir(sdir):
+            bos.makedirs(sdir)
+
+        if len(bos.listdir(sdir)) >= self.args.bf_nc:
+            return self.log("no bakflip; too many", 3)
+
+        nrem = sz
+        f.seek(ofs)
+        with open(fp, "wb") as fo:
+            while nrem:
+                buf = f.read(min(nrem, 512 * 1024))
+                if not buf:
+                    break
+
+                nrem -= len(buf)
+                fo.write(buf)
+
+        if nrem:
+            self.log("bakflip truncated; {} remains".format(nrem), 1)
+            atomic_move(fp, fp + ".trunc")
+        else:
+            self.log("bakflip ok", 2)
+
     def rand_name(self, fdir: str, fn: str, rnd: int) -> str:
         ok = False
         try:
@@ -1177,6 +1210,11 @@ class HttpCli(object):
                 post_sz, _, sha_b64 = hashcopy(reader, f, self.args.s_wr_slp)
 
                 if sha_b64 != chash:
+                    try:
+                        self.bakflip(f, cstart[0], post_sz, sha_b64)
+                    except:
+                        self.log("bakflip failed: " + min_ex())
+
                     t = "your chunk got corrupted somehow (received {} bytes); expected vs received hash:\n{}\n{}"
                     raise Pebkac(400, t.format(post_sz, chash, sha_b64))
 

copyparty/svchub.py

@@ -24,7 +24,7 @@ try:
 except:
     pass
 
-from .__init__ import ANYWIN, MACOS, PY2, VT100, WINDOWS, EnvParams, unicode
+from .__init__ import ANYWIN, MACOS, VT100, EnvParams, unicode
 from .authsrv import AuthSrv
 from .mtag import HAVE_FFMPEG, HAVE_FFPROBE
 from .tcpsrv import TcpSrv

copyparty/util.py

@@ -1786,7 +1786,7 @@ def yieldfile(fn: str) -> Generator[bytes, None, None]:
 
 
 def hashcopy(
-    fin: Union[typing.BinaryIO, Generator[bytes, None, None]],
+    fin: Generator[bytes, None, None],
     fout: Union[typing.BinaryIO, typing.IO[Any]],
     slp: int = 0,
     max_sz: int = 0,