copyparty 1.12.1__py3-none-any.whl → 1.13.0__py3-none-any.whl

copyparty/__main__.py

@@ -850,8 +850,9 @@ def add_qr(ap, tty):
 
 def add_fs(ap):
     ap2 = ap.add_argument_group("filesystem options")
-    rm_re_def = "5/0.1" if ANYWIN else "0/0"
+    rm_re_def = "15/0.1" if ANYWIN else "0/0"
     ap2.add_argument("--rm-retry", metavar="T/R", type=u, default=rm_re_def, help="if a file cannot be deleted because it is busy, continue trying for \033[33mT\033[0m seconds, retry every \033[33mR\033[0m seconds; disable with 0/0 (volflag=rm_retry)")
+    ap2.add_argument("--mv-retry", metavar="T/R", type=u, default=rm_re_def, help="if a file cannot be renamed because it is busy, continue trying for \033[33mT\033[0m seconds, retry every \033[33mR\033[0m seconds; disable with 0/0 (volflag=mv_retry)")
     ap2.add_argument("--iobuf", metavar="BYTES", type=int, default=256*1024, help="file I/O buffer-size; if your volumes are on a network drive, try increasing to \033[32m524288\033[0m or even \033[32m4194304\033[0m (and let me know if that improves your performance)")
 
 
@@ -943,6 +944,8 @@ def add_auth(ap):
     ap2.add_argument("--idp-h-grp", metavar="HN", type=u, default="", help="assume the request-header \033[33mHN\033[0m contains the groupname of the requesting user; can be referenced in config files for group-based access control")
     ap2.add_argument("--idp-h-key", metavar="HN", type=u, default="", help="optional but recommended safeguard; your reverse-proxy will insert a secret header named \033[33mHN\033[0m into all requests, and the other IdP headers will be ignored if this header is not present")
     ap2.add_argument("--idp-gsep", metavar="RE", type=u, default="|:;+,", help="if there are multiple groups in \033[33m--idp-h-grp\033[0m, they are separated by one of the characters in \033[33mRE\033[0m")
+    ap2.add_argument("--no-bauth", action="store_true", help="disable basic-authentication support; do not accept passwords from the 'Authenticate' header at all. NOTE: This breaks support for the android app")
+    ap2.add_argument("--bauth-last", action="store_true", help="keeps basic-authentication enabled, but only as a last-resort; if a cookie is also provided then the cookie wins")
 
 
 def add_zeroconf(ap):
@@ -1082,6 +1085,8 @@ def add_optouts(ap):
     ap2.add_argument("--no-zip", action="store_true", help="disable download as zip/tar")
     ap2.add_argument("--no-tarcmp", action="store_true", help="disable download as compressed tar (?tar=gz, ?tar=bz2, ?tar=xz, ?tar=gz:9, ...)")
     ap2.add_argument("--no-lifetime", action="store_true", help="do not allow clients (or server config) to schedule an upload to be deleted after a given time")
+    ap2.add_argument("--no-pipe", action="store_true", help="disable race-the-beam (lockstep download of files which are currently being uploaded) (volflag=nopipe)")
+    ap2.add_argument("--no-db-ip", action="store_true", help="do not write uploader IPs into the database")
 
 
 def add_safety(ap):
@@ -1207,7 +1212,7 @@ def add_db_general(ap, hcores):
     ap2.add_argument("--no-hash", metavar="PTN", type=u, help="regex: disable hashing of matching absolute-filesystem-paths during e2ds folder scans (volflag=nohash)")
     ap2.add_argument("--no-idx", metavar="PTN", type=u, default=noidx, help="regex: disable indexing of matching absolute-filesystem-paths during e2ds folder scans (volflag=noidx)")
     ap2.add_argument("--no-dhash", action="store_true", help="disable rescan acceleration; do full database integrity check -- makes the db ~5%% smaller and bootup/rescans 3~10x slower")
-    ap2.add_argument("--re-dhash", action="store_true", help="rebuild the cache if it gets out of sync (for example crash on startup during metadata scanning)")
+    ap2.add_argument("--re-dhash", action="store_true", help="force a cache rebuild on startup; enable this once if it gets out of sync (should never be necessary)")
     ap2.add_argument("--no-forget", action="store_true", help="never forget indexed files, even when deleted from disk -- makes it impossible to ever upload the same file twice -- only useful for offloading uploads to a cloud service or something (volflag=noforget)")
     ap2.add_argument("--dbd", metavar="PROFILE", default="wal", help="database durability profile; sets the tradeoff between robustness and speed, see \033[33m--help-dbd\033[0m (volflag=dbd)")
     ap2.add_argument("--xlink", action="store_true", help="on upload: check all volumes for dupes, not just the target volume (volflag=xlink)")

copyparty/__version__.py

@@ -1,8 +1,8 @@
 # coding: utf-8
 
-VERSION = (1, 12, 1)
-CODENAME = "locksmith"
-BUILD_DT = (2024, 4, 9)
+VERSION = (1, 13, 0)
+CODENAME = "race the beam"
+BUILD_DT = (2024, 4, 20)
 
 S_VERSION = ".".join(map(str, VERSION))
 S_BUILD_DT = "{0:04d}-{1:02d}-{2:02d}".format(*BUILD_DT)

copyparty/authsrv.py

@@ -1757,13 +1757,14 @@ class AuthSrv(object):
                 if k in vol.flags:
                     vol.flags[k] = float(vol.flags[k])
 
-            try:
-                zs1, zs2 = vol.flags["rm_retry"].split("/")
-                vol.flags["rm_re_t"] = float(zs1)
-                vol.flags["rm_re_r"] = float(zs2)
-            except:
-                t = 'volume "/%s" has invalid rm_retry [%s]'
-                raise Exception(t % (vol.vpath, vol.flags.get("rm_retry")))
+            for k in ("mv_re", "rm_re"):
+                try:
+                    zs1, zs2 = vol.flags[k + "try"].split("/")
+                    vol.flags[k + "_t"] = float(zs1)
+                    vol.flags[k + "_r"] = float(zs2)
+                except:
+                    t = 'volume "/%s" has invalid %stry [%s]'
+                    raise Exception(t % (vol.vpath, k, vol.flags.get(k + "try")))
 
             for k1, k2 in IMPLICATIONS:
                 if k1 in vol.flags:

copyparty/cert.py

@@ -6,10 +6,17 @@ import os
 import shutil
 import time
 
-from .util import Netdev, runcmd
+from .__init__ import ANYWIN
+from .util import Netdev, runcmd, wrename, wunlink
 
 HAVE_CFSSL = True
 
+if ANYWIN:
+    VF = {"mv_re_t": 5, "rm_re_t": 5, "mv_re_r": 0.1, "rm_re_r": 0.1}
+else:
+    VF = {"mv_re_t": 0, "rm_re_t": 0}
+
+
 def ensure_cert(log , args)  :
     """
     the default cert (and the entire TLS support) is only here to enable the
@@ -101,8 +108,12 @@ def _gen_ca(log , args):
         raise Exception("failed to translate ca-cert: {}, {}".format(rc, se), 3)
 
     bname = os.path.join(args.crt_dir, "ca")
-    os.rename(bname + "-key.pem", bname + ".key")
-    os.unlink(bname + ".csr")
+    try:
+        wunlink(log, bname + ".key", VF)
+    except:
+        pass
+    wrename(log, bname + "-key.pem", bname + ".key", VF)
+    wunlink(log, bname + ".csr", VF)
 
     log("cert", "new ca OK", 2)
 
@@ -181,11 +192,11 @@ def _gen_srv(log , args, netdevs  ):
 
     bname = os.path.join(args.crt_dir, "srv")
     try:
-        os.unlink(bname + ".key")
+        wunlink(log, bname + ".key", VF)
     except:
         pass
-    os.rename(bname + "-key.pem", bname + ".key")
-    os.unlink(bname + ".csr")
+    wrename(log, bname + "-key.pem", bname + ".key", VF)
+    wunlink(log, bname + ".csr", VF)
 
     with open(os.path.join(args.crt_dir, "ca.pem"), "rb") as f:
         ca = f.read()

copyparty/cfg.py

@@ -16,6 +16,7 @@ def vf_bmap()   :
         "no_dedup": "copydupes",
         "no_dupe": "nodupe",
         "no_forget": "noforget",
+        "no_pipe": "nopipe",
         "no_robots": "norobots",
         "no_thumb": "dthumb",
         "no_vthumb": "dvthumb",
@@ -63,6 +64,7 @@ def vf_vmap()   :
         "lg_sbf",
         "md_sbf",
         "nrand",
+        "mv_retry",
         "rm_retry",
         "sort",
         "unlist",
@@ -214,6 +216,7 @@ flagcats = {
         "dots": "allow all users with read-access to\nenable the option to show dotfiles in listings",
         "fk=8": 'generates per-file accesskeys,\nwhich are then required at the "g" permission;\nkeys are invalidated if filesize or inode changes',
         "fka=8": 'generates slightly weaker per-file accesskeys,\nwhich are then required at the "g" permission;\nnot affected by filesize or inode numbers',
+        "mv_retry": "ms-windows: timeout for renaming busy files",
         "rm_retry": "ms-windows: timeout for deleting busy files",
         "davauth": "ask webdav clients to login for all folders",
         "davrt": "show lastmod time of symlink destination, not the link itself\n(note: this option is always enabled for recursive listings)",

copyparty/httpcli.py

@@ -36,6 +36,7 @@ from .bos import bos
 from .star import StreamTar
 from .sutil import StreamArc, gfilter
 from .szip import StreamZip
+from .up2k import up2k_chunksize
 from .util import unquote  # type: ignore
 from .util import (
     APPLESAN_RE,
@@ -89,6 +90,7 @@ from .util import (
     vjoin,
     vol_san,
     vsplit,
+    wrename,
     wunlink,
     yieldfile,
 )
@@ -122,6 +124,7 @@ class HttpCli(object):
         self.ico = conn.ico  # mypy404
         self.thumbcli = conn.thumbcli  # mypy404
         self.u2fh = conn.u2fh  # mypy404
+        self.pipes = conn.pipes  # mypy404
         self.log_func = conn.log_func  # mypy404
         self.log_src = conn.log_src  # mypy404
         self.gen_fk = self._gen_fk if self.args.log_fk else gen_filekey
@@ -439,7 +442,11 @@ class HttpCli(object):
 
         zso = self.headers.get("authorization")
         bauth = ""
-        if zso:
+        if (
+            zso
+            and not self.args.no_bauth
+            and (not cookie_pw or not self.args.bauth_last)
+        ):
             try:
                 zb = zso.split(" ")[1].encode("ascii")
                 zs = base64.b64decode(zb).decode("utf-8")
@@ -1796,7 +1803,7 @@ class HttpCli(object):
                     f, fn = zfw["orz"]
 
                 path2 = os.path.join(fdir, fn2)
-                atomic_move(path, path2)
+                atomic_move(self.log, path, path2, vfs.flags)
                 fn = fn2
                 path = path2
 
@@ -1877,7 +1884,9 @@ class HttpCli(object):
         self.reply(t.encode("utf-8"), 201, headers=h)
         return True
 
-    def bakflip(self, f , ofs , sz , sha )  :
+    def bakflip(
+        self, f , ofs , sz , sha , flags  
+    )  :
         if not self.args.bak_flips or self.args.nw:
             return
 
@@ -1905,7 +1914,7 @@ class HttpCli(object):
 
         if nrem:
             self.log("bakflip truncated; {} remains".format(nrem), 1)
-            atomic_move(fp, fp + ".trunc")
+            atomic_move(self.log, fp, fp + ".trunc", flags)
         else:
             self.log("bakflip ok", 2)
 
@@ -2171,7 +2180,7 @@ class HttpCli(object):
 
                 if sha_b64 != chash:
                     try:
-                        self.bakflip(f, cstart[0], post_sz, sha_b64)
+                        self.bakflip(f, cstart[0], post_sz, sha_b64, vfs.flags)
                     except:
                         self.log("bakflip failed: " + min_ex())
 
@@ -2523,7 +2532,7 @@ class HttpCli(object):
                             raise
 
                     if not nullwrite:
-                        atomic_move(tabspath, abspath)
+                        atomic_move(self.log, tabspath, abspath, vfs.flags)
 
                     tabspath = ""
 
@@ -2763,7 +2772,7 @@ class HttpCli(object):
                 hidedir(dp)
             except:
                 pass
-            bos.rename(fp, os.path.join(mdir, ".hist", mfile2))
+            wrename(self.log, fp, os.path.join(mdir, ".hist", mfile2), vfs.flags)
 
         assert self.parser.gen
         p_field, _, p_data = next(self.parser.gen)
@@ -2918,17 +2927,42 @@ class HttpCli(object):
 
         return txt
 
-    def tx_file(self, req_path )  :
+    def tx_file(self, req_path , ptop  = None)  :
         status = 200
         logmsg = "{:4} {} ".format("", self.req)
         logtail = ""
 
+        if ptop is not None:
+            try:
+                dp, fn = os.path.split(req_path)
+                tnam = fn + ".PARTIAL"
+                if self.args.dotpart:
+                    tnam = "." + tnam
+                ap_data = os.path.join(dp, tnam)
+                st_data = bos.stat(ap_data)
+                if not st_data.st_size:
+                    raise Exception("partial is empty")
+                x = self.conn.hsrv.broker.ask("up2k.find_job_by_ap", ptop, req_path)
+                job = json.loads(x.get())
+                if not job:
+                    raise Exception("not found in registry")
+                self.pipes.set(req_path, job)
+            except Exception as ex:
+                self.log("will not pipe [%s]; %s" % (ap_data, ex), 6)
+                ptop = None
+
         #
         # if request is for foo.js, check if we have foo.js.gz
 
         file_ts = 0.0
         editions    = {}
         for ext in ("", ".gz"):
+            if ptop is not None:
+                sz = job["size"]
+                file_ts = job["lmod"]
+                editions["plain"] = (ap_data, sz)
+                break
+
             try:
                 fs_path = req_path + ext
                 st = bos.stat(fs_path)
@@ -3085,6 +3119,11 @@ class HttpCli(object):
             self.send_headers(length=upper - lower, status=status, mime=mime)
             return True
 
+        if ptop is not None:
+            return self.tx_pipe(
+                ptop, req_path, ap_data, job, lower, upper, status, mime, logmsg
+            )
+
         ret = True
         with open_func(*open_args) as f:
             self.send_headers(length=upper - lower, status=status, mime=mime)
@@ -3104,6 +3143,143 @@ class HttpCli(object):
 
         return ret
 
+    def tx_pipe(
+        self,
+        ptop ,
+        req_path ,
+        ap_data ,
+        job  ,
+        lower ,
+        upper ,
+        status ,
+        mime ,
+        logmsg ,
+    )  :
+        M = 1048576
+        self.send_headers(length=upper - lower, status=status, mime=mime)
+        wr_slp = self.args.s_wr_slp
+        wr_sz = self.args.s_wr_sz
+        file_size = job["size"]
+        chunk_size = up2k_chunksize(file_size)
+        num_need = -1
+        data_end = 0
+        remains = upper - lower
+        broken = False
+        spins = 0
+        tier = 0
+        tiers = ["uncapped", "reduced speed", "one byte per sec"]
+
+        while lower < upper and not broken:
+            with self.u2mutex:
+                job = self.pipes.get(req_path)
+                if not job:
+                    x = self.conn.hsrv.broker.ask("up2k.find_job_by_ap", ptop, req_path)
+                    job = json.loads(x.get())
+                    if job:
+                        self.pipes.set(req_path, job)
+
+            if not job:
+                t = "pipe: OK, upload has finished; yeeting remainder"
+                self.log(t, 2)
+                data_end = file_size
+                break
+
+            if num_need != len(job["need"]):
+                num_need = len(job["need"])
+                data_end = 0
+                for cid in job["hash"]:
+                    if cid in job["need"]:
+                        break
+                    data_end += chunk_size
+                t = "pipe: can stream %.2f MiB; requested range is %.2f to %.2f"
+                self.log(t % (data_end / M, lower / M, upper / M), 6)
+                with self.u2mutex:
+                    if data_end > self.u2fh.aps.get(ap_data, data_end):
+                        try:
+                            fhs = self.u2fh.cache[ap_data].all_fhs
+                            for fh in fhs:
+                                fh.flush()
+                            self.u2fh.aps[ap_data] = data_end
+                            self.log("pipe: flushed %d up2k-FDs" % (len(fhs),))
+                        except Exception as ex:
+                            self.log("pipe: u2fh flush failed: %r" % (ex,))
+
+            if lower >= data_end:
+                if data_end:
+                    t = "pipe: uploader is too slow; aborting download at %.2f MiB"
+                    self.log(t % (data_end / M))
+                    raise Pebkac(416, "uploader is too slow")
+
+                raise Pebkac(416, "no data available yet; please retry in a bit")
+
+            slack = data_end - lower
+            if slack >= 8 * M:
+                ntier = 0
+                winsz = M
+                bufsz = wr_sz
+                slp = wr_slp
+            else:
+                winsz = max(40, int(M * (slack / (12 * M))))
+                base_rate = M if not wr_slp else wr_sz / wr_slp
+                if winsz > base_rate:
+                    ntier = 0
+                    bufsz = wr_sz
+                    slp = wr_slp
+                elif winsz > 300:
+                    ntier = 1
+                    bufsz = winsz // 5
+                    slp = 0.2
+                else:
+                    ntier = 2
+                    bufsz = winsz = slp = 1
+
+            if tier != ntier:
+                tier = ntier
+                self.log("moved to tier %d (%s)" % (tier, tiers[tier]))
+
+            try:
+                with open(ap_data, "rb", self.args.iobuf) as f:
+                    f.seek(lower)
+                    page = f.read(min(winsz, data_end - lower, upper - lower))
+                if not page:
+                    raise Exception("got 0 bytes (EOF?)")
+            except Exception as ex:
+                self.log("pipe: read failed at %.2f MiB: %s" % (lower / M, ex), 3)
+                with self.u2mutex:
+                    self.pipes.c.pop(req_path, None)
+                spins += 1
+                if spins > 3:
+                    raise Pebkac(500, "file became unreadable")
+                time.sleep(2)
+                continue
+
+            spins = 0
+            pofs = 0
+            while pofs < len(page):
+                if slp:
+                    time.sleep(slp)
+
+                try:
+                    buf = page[pofs : pofs + bufsz]
+                    self.s.sendall(buf)
+                    zi = len(buf)
+                    remains -= zi
+                    lower += zi
+                    pofs += zi
+                except:
+                    broken = True
+                    break
+
+        if lower < upper and not broken:
+            with open(req_path, "rb") as f:
+                remains = sendfile_py(self.log, lower, upper, f, self.s, wr_sz, wr_slp)
+
+        spd = self._spd((upper - lower) - remains)
+        if self.do_log:
+            self.log("{},  {}".format(logmsg, spd))
+
+        return not broken
+
     def tx_zip(
         self,
         fmt ,
@@ -3741,7 +3917,7 @@ class HttpCli(object):
         if not allvols:
             ret = [{"kinshi": 1}]
 
-        jtxt = '{"u":%s,"c":%s}' % (uret, json.dumps(ret, indent=0))
+        jtxt = '{"u":%s,"c":%s}' % (uret, json.dumps(ret, separators=(",\n", ": ")))
         zi = len(uret.split('\n"pd":')) - 1
         self.log("%s #%d+%d %.2fsec" % (lm, zi, len(ret), time.time() - t0))
         self.reply(jtxt.encode("utf-8", "replace"), mime="application/json")
@@ -4020,7 +4196,9 @@ class HttpCli(object):
             ):
                 return self.tx_md(vn, abspath)
 
-            return self.tx_file(abspath)
+            return self.tx_file(
+                abspath, None if st.st_size or "nopipe" in vn.flags else vn.realpath
+            )
 
         elif is_dir and not self.can_read:
             if self._use_dirkey(abspath):

copyparty/httpconn.py

@@ -52,6 +52,7 @@ class HttpConn(object):
         self.E  = self.args.E
         self.asrv  = hsrv.asrv  # mypy404
         self.u2fh  = hsrv.u2fh  # mypy404
+        self.pipes  = hsrv.pipes  # mypy404
         self.ipa_nm  = hsrv.ipa_nm
         self.xff_nm  = hsrv.xff_nm
         self.xff_lan  = hsrv.xff_lan  # type: ignore

copyparty/httpsrv.py

@@ -61,6 +61,7 @@ from .u2idx import U2idx
 from .util import (
     E_SCK,
     FHC,
+    CachedDict,
     Daemon,
     Garda,
     Magician,
@@ -126,6 +127,7 @@ class HttpSrv(object):
         self.t_periodic  = None
 
         self.u2fh = FHC()
+        self.pipes = CachedDict(0.2)
         self.metrics = Metrics(self)
         self.nreq = 0
         self.nsus = 0

copyparty/svchub.py

@@ -418,6 +418,12 @@ class SvcHub(object):
             t = "WARNING: found config files in [%s]: %s\n  config files are not expected here, and will NOT be loaded (unless your setup is intentionally hella funky)"
             self.log("root", t % (E.cfg, ", ".join(hits)), 3)
 
+        if self.args.no_bauth:
+            t = "WARNING: --no-bauth disables support for the Android app; you may want to use --bauth-last instead"
+            self.log("root", t, 3)
+            if self.args.bauth_last:
+                self.log("root", "WARNING: ignoring --bauth-last due to --no-bauth", 3)
+
     def _process_config(self)  :
         al = self.args
 
@@ -538,6 +544,13 @@ class SvcHub(object):
         except:
             raise Exception("invalid --rm-retry [%s]" % (self.args.rm_retry,))
 
+        try:
+            zf1, zf2 = self.args.mv_retry.split("/")
+            self.args.mv_re_t = float(zf1)
+            self.args.mv_re_r = float(zf2)
+        except:
+            raise Exception("invalid --mv-retry [%s]" % (self.args.mv_retry,))
+
         return True
 
     def _ipa2re(self, txt)  :

copyparty/th_srv.py

@@ -28,6 +28,7 @@ from .util import (
     runcmd,
     statdir,
     vsplit,
+    wrename,
     wunlink,
 )
 
@@ -343,7 +344,7 @@ class ThumbSrv(object):
                             pass
 
             try:
-                bos.rename(ttpath, tpath)
+                wrename(self.log, ttpath, tpath, vn.flags)
             except:
                 pass