v0.7.7

.vscode/launch.json

@@ -12,12 +12,21 @@
                 //"-nw",
                 "-ed",
                 "-emp",
+                "-e2d",
+                "-e2s",
                 "-a",
                 "ed:wark",
                 "-v",
-                "srv::r:aed"
+                "srv::r:aed:cnodupe"
             ]
         },
+        {
+            "name": "No debug",
+            "preLaunchTask": "no_dbg",
+            "type": "python",
+            //"request": "attach", "port": 42069
+            // fork: nc -l 42069 </dev/null
+        },
         {
             "name": "Run active unit test",
             "type": "python",

.vscode/settings.json

@@ -50,11 +50,9 @@
     "files.associations": {
         "*.makefile": "makefile"
     },
-    "editor.codeActionsOnSaveTimeout": 9001,
-    "editor.formatOnSaveTimeout": 9001,
-    //
-    //  things you may wanna edit:
-    //
-    "python.pythonPath": "/usr/bin/python3",
-    //"python.linting.enabled": true,
+    "python.formatting.blackArgs": [
+        "-t",
+        "py27"
+    ],
+    "python.linting.enabled": true,
 }

.vscode/tasks.json

@@ -0,0 +1,15 @@
+{
+    "version": "2.0.0",
+    "tasks": [
+        {
+            "label": "pre",
+            "command": "true;rm -rf inc/* inc/.hist/;mkdir -p inc;",
+            "type": "shell"
+        },
+        {
+            "label": "no_dbg",
+            "command": "${config:python.pythonPath} -m copyparty -ed -emp -e2d -e2s -a ed:wark -v srv::r:aed:cnodupe ;exit 1",
+            "type": "shell"
+        }
+    ]
+}

README.md

@@ -8,7 +8,7 @@
 
 turn your phone or raspi into a portable file server with resumable uploads/downloads using IE6 or any other browser
 
-* server runs on anything with `py2.7` or `py3.2+`
+* server runs on anything with `py2.7` or `py3.3+`
 * *resumable* uploads need `firefox 12+` / `chrome 6+` / `safari 6+` / `IE 10+`
 * code standard: `black`
 
@@ -68,17 +68,16 @@ summary: it works! you can use it! (but technically not even close to beta)
 # dependencies
 
 * `jinja2`
-  * pulls in `markupsafe` as of v2.7; use jinja 2.6 on py3.2
 
-optional, enables thumbnails:
+optional, will eventually enable thumbnails:
 * `Pillow` (requires py2.7 or py3.5+)
 
 
 # sfx
 
 currently there are two self-contained binaries:
-* `copyparty-sfx.sh` for unix (linux and osx) -- smaller, more robust
-* `copyparty-sfx.py` for windows (unix too) -- crossplatform, beta
+* [copyparty-sfx.py](https://github.com/9001/copyparty/releases/latest/download/copyparty-sfx.py) -- pure python, works everywhere
+* [copyparty-sfx.sh](https://github.com/9001/copyparty/releases/latest/download/copyparty-sfx.sh) -- smaller, but only for linux and macos
 
 launch either of them (**use sfx.py on systemd**) and it'll unpack and run copyparty, assuming you have python installed of course
 
@@ -142,10 +141,9 @@ roughly sorted by priority
   * terminate client on bad data
 * drop onto folders
 * `os.copy_file_range` for up2k cloning
+* up2k partials ui
 * support pillow-simd
 * cache sha512 chunks on client
-* ~~symlink existing files on upload~~
-  * ok at runtime, up2k db still not persisted
 * comment field
 * ~~look into android thumbnail cache file format~~ bad idea
 * figure out the deal with pixel3a not being connectable as hotspot

bin/copyparty-fuse-streaming.py

@@ -1067,7 +1067,7 @@ def main():
         dbg = null_log
 
     if WINDOWS:
-        os.system("")
+        os.system("rem")
 
         for ch in '<>:"\\|?*':
             # microsoft maps illegal characters to f0xx

bin/copyparty-fuse.py

@@ -980,7 +980,7 @@ def main():
         dbg = null_log
 
     if WINDOWS:
-        os.system("")
+        os.system("rem")
 
         for ch in '<>:"\\|?*':
             # microsoft maps illegal characters to f0xx

contrib/README.md

@@ -10,7 +10,12 @@
 * modify `10.13.1.1` as necessary if you wish to support browsers without javascript
 
 ### [`explorer-nothumbs-nofoldertypes.reg`](explorer-nothumbs-nofoldertypes.reg)
-disables thumbnails and folder-type detection in windows explorer, makes it way faster (especially for slow/networked locations (such as copyparty-fuse))
+* disables thumbnails and folder-type detection in windows explorer
+* makes it way faster (especially for slow/networked locations (such as copyparty-fuse))
+
+### [`cfssl.sh`](cfssl.sh)
+* creates CA and server certificates using cfssl
+* give a 3rd argument to install it to your copyparty config
 
 # OS integration
 init-scripts to start copyparty as a service

contrib/cfssl.sh

@@ -0,0 +1,72 @@
+#!/bin/bash
+set -e
+
+# ca-name and server-name
+ca_name="$1"
+srv_name="$2"
+
+[ -z "$srv_name" ] && {
+	echo "need arg 1: ca name"
+	echo "need arg 2: server name"
+	exit 1
+}
+
+
+gen_ca() {
+	(tee /dev/stderr <<EOF
+{"CN": "$ca_name ca",
+"CA": {"expiry":"87600h", "pathlen":0},
+"key": {"algo":"rsa", "size":4096},
+"names": [{"O":"$ca_name ca"}]}
+EOF
+	)|
+	cfssl gencert -initca - |
+	cfssljson -bare ca
+	
+	mv ca-key.pem ca.key
+	rm ca.csr
+}
+
+
+gen_srv() {
+	(tee /dev/stderr <<EOF
+{"key": {"algo":"rsa", "size":4096},
+"names": [{"O":"$ca_name - $srv_name"}]}
+EOF
+	)|
+	cfssl gencert -ca ca.pem -ca-key ca.key \
+		-profile=www -hostname="$srv_name.$ca_name" - |
+	cfssljson -bare "$srv_name"
+
+	mv "$srv_name-key.pem" "$srv_name.key"
+	rm "$srv_name.csr"
+}
+
+
+# create ca if not exist
+[ -e ca.key ] ||
+	gen_ca
+
+# always create server cert
+gen_srv
+
+
+# dump cert info
+show() {
+	openssl x509 -text -noout -in $1 |
+	awk '!o; {o=0} /[0-9a-f:]{16}/{o=1}'
+}
+show ca.pem
+show "$srv_name.pem"
+
+
+# write cert into copyparty config
+[ -z "$3" ] || {
+	mkdir -p ~/.config/copyparty
+	cat "$srv_name".{key,pem} ca.pem >~/.config/copyparty/cert.pem 
+}
+
+
+# rm *.key *.pem
+# cfssl print-defaults config
+# cfssl print-defaults csr

copyparty/__main__.py

@@ -8,7 +8,9 @@ __copyright__ = 2019
 __license__ = "MIT"
 __url__ = "https://github.com/9001/copyparty/"
 
+import re
 import os
+import sys
 import time
 import shutil
 import filecmp
@@ -19,7 +21,13 @@ from textwrap import dedent
 from .__init__ import E, WINDOWS, VT100
 from .__version__ import S_VERSION, S_BUILD_DT, CODENAME
 from .svchub import SvcHub
-from .util import py_desc
+from .util import py_desc, align_tab
+
+HAVE_SSL = True
+try:
+    import ssl
+except:
+    HAVE_SSL = False
 
 
 class RiceFormatter(argparse.HelpFormatter):
@@ -85,10 +93,77 @@ def ensure_cert():
     # printf 'NO\n.\n.\n.\n.\ncopyparty-insecure\n.\n' | faketime '2000-01-01 00:00:00' openssl req -x509 -sha256 -newkey rsa:2048 -keyout insecure.pem -out insecure.pem -days $((($(printf %d 0x7fffffff)-$(date +%s --date=2000-01-01T00:00:00Z))/(60*60*24))) -nodes && ls -al insecure.pem && openssl x509 -in insecure.pem -text -noout
 
 
+def configure_ssl_ver(al):
+    def terse_sslver(txt):
+        txt = txt.lower()
+        for c in ["_", "v", "."]:
+            txt = txt.replace(c, "")
+
+        return txt.replace("tls10", "tls1")
+
+    # oh man i love openssl
+    # check this out
+    # hold my beer
+    ptn = re.compile(r"^OP_NO_(TLS|SSL)v")
+    sslver = terse_sslver(al.ssl_ver).split(",")
+    flags = [k for k in ssl.__dict__ if ptn.match(k)]
+    # SSLv2 SSLv3 TLSv1 TLSv1_1 TLSv1_2 TLSv1_3
+    if "help" in sslver:
+        avail = [terse_sslver(x[6:]) for x in flags]
+        avail = " ".join(sorted(avail) + ["all"])
+        print("\navailable ssl/tls versions:\n  " + avail)
+        sys.exit(0)
+
+    al.ssl_flags_en = 0
+    al.ssl_flags_de = 0
+    for flag in sorted(flags):
+        ver = terse_sslver(flag[6:])
+        num = getattr(ssl, flag)
+        if ver in sslver:
+            al.ssl_flags_en |= num
+        else:
+            al.ssl_flags_de |= num
+
+    if sslver == ["all"]:
+        x = al.ssl_flags_en
+        al.ssl_flags_en = al.ssl_flags_de
+        al.ssl_flags_de = x
+
+    for k in ["ssl_flags_en", "ssl_flags_de"]:
+        num = getattr(al, k)
+        print("{}: {:8x} ({})".format(k, num, num))
+
+    # think i need that beer now
+
+
+def configure_ssl_ciphers(al):
+    ctx = ssl.create_default_context(ssl.Purpose.CLIENT_AUTH)
+    if al.ssl_ver:
+        ctx.options &= ~al.ssl_flags_en
+        ctx.options |= al.ssl_flags_de
+
+    is_help = al.ciphers == "help"
+
+    if al.ciphers and not is_help:
+        try:
+            ctx.set_ciphers(al.ciphers)
+        except:
+            print("\n\033[1;31mfailed to set ciphers\033[0m\n")
+
+    if not hasattr(ctx, "get_ciphers"):
+        print("cannot read cipher list: openssl or python too old")
+    else:
+        ciphers = [x["description"] for x in ctx.get_ciphers()]
+        print("\n  ".join(["\nenabled ciphers:"] + align_tab(ciphers) + [""]))
+
+    if is_help:
+        sys.exit(0)
+
+
 def main():
     time.strptime("19970815", "%Y%m%d")  # python#7980
     if WINDOWS:
-        os.system("")  # enables colors
+        os.system("rem")  # enables colors
 
     desc = py_desc().replace("[", "\033[1;30m[")
 
@@ -96,7 +171,8 @@ def main():
     print(f.format(S_VERSION, CODENAME, S_BUILD_DT, desc))
 
     ensure_locale()
-    ensure_cert()
+    if HAVE_SSL:
+        ensure_cert()
 
     ap = argparse.ArgumentParser(
         formatter_class=RiceFormatter,
@@ -105,29 +181,45 @@ def main():
         epilog=dedent(
             """
             -a takes username:password,
-            -v takes src:dst:permset:permset:... where "permset" is
-               accesslevel followed by username (no separator)
+            -v takes src:dst:permset:permset:cflag:cflag:...
+               where "permset" is accesslevel followed by username (no separator)
+               and "cflag" is config flags to set on this volume
             
+            list of cflags:
+              cnodupe rejects existing files (instead of symlinking them)
+
             example:\033[35m
-              -a ed:hunter2 -v .::r:aed -v ../inc:dump:w:aed  \033[36m
+              -a ed:hunter2 -v .::r:aed -v ../inc:dump:w:aed:cnodupe  \033[36m
               mount current directory at "/" with
                * r (read-only) for everyone
                * a (read+write) for ed
               mount ../inc at "/dump" with
                * w (write-only) for everyone
-               * a (read+write) for ed  \033[0m
+               * a (read+write) for ed
+               * reject duplicate files  \033[0m
             
             if no accounts or volumes are configured,
             current folder will be read/write for everyone
 
             consider the config file for more flexible account/volume management,
             including dynamic reload at runtime (and being more readable w)
+
+            values for --urlform:
+              "stash" dumps the data to file and returns length + checksum
+              "save,get" dumps to file and returns the page like a GET
+              "print,get" prints the data in the log and returns GET
+              (leave out the ",get" to return an error instead)
+
+            --ciphers help = available ssl/tls ciphers,
+            --ssl-ver help = available ssl/tls versions,
+              default is what python considers safe, usually >= TLS1
             """
         ),
     )
+    # fmt: off
     ap.add_argument("-c", metavar="PATH", type=str, action="append", help="add config file")
-    ap.add_argument("-i", metavar="IP", type=str, default="0.0.0.0", help="ip to bind")
-    ap.add_argument("-p", metavar="PORT", type=int, default=3923, help="port to bind")
+    ap.add_argument("-i", metavar="IP", type=str, default="0.0.0.0", help="ip to bind (comma-sep.)")
+    ap.add_argument("-p", metavar="PORT", type=str, default="3923", help="ports to bind (comma/range)")
     ap.add_argument("-nc", metavar="NUM", type=int, default=64, help="max num clients")
     ap.add_argument("-j", metavar="CORES", type=int, default=1, help="max num cpu cores")
     ap.add_argument("-a", metavar="ACCT", type=str, action="append", help="add account")
@@ -135,12 +227,43 @@ def main():
     ap.add_argument("-q", action="store_true", help="quiet")
     ap.add_argument("-ed", action="store_true", help="enable ?dots")
     ap.add_argument("-emp", action="store_true", help="enable markdown plugins")
+    ap.add_argument("-e2d", action="store_true", help="enable up2k database")
+    ap.add_argument("-e2s", action="store_true", help="enable up2k db-scanner")
     ap.add_argument("-mcr", metavar="SEC", type=int, default=60, help="md-editor mod-chk rate")
     ap.add_argument("-nw", action="store_true", help="disable writes (benchmark)")
     ap.add_argument("-nih", action="store_true", help="no info hostname")
     ap.add_argument("-nid", action="store_true", help="no info disk-usage")
     ap.add_argument("--no-sendfile", action="store_true", help="disable sendfile")
+    ap.add_argument("--urlform", type=str, default="print,get", help="how to handle url-forms")
+
+    ap2 = ap.add_argument_group('SSL/TLS options')
+    ap2.add_argument("--http-only", action="store_true", help="disable ssl/tls")
+    ap2.add_argument("--https-only", action="store_true", help="disable plaintext")
+    ap2.add_argument("--ssl-ver", type=str, help="ssl/tls versions to allow")
+    ap2.add_argument("--ciphers", metavar="LIST", help="set allowed ciphers")
+    ap2.add_argument("--ssl-dbg", action="store_true", help="dump some tls info")
+    ap2.add_argument("--ssl-log", metavar="PATH", help="log master secrets")
     al = ap.parse_args()
+    # fmt: on
+
+    al.i = al.i.split(",")
+    try:
+        if "-" in al.p:
+            lo, hi = [int(x) for x in al.p.split("-")]
+            al.p = list(range(lo, hi + 1))
+        else:
+            al.p = [int(x) for x in al.p.split(",")]
+    except:
+        raise Exception("invalid value for -p")
+
+    if HAVE_SSL:
+        if al.ssl_ver:
+            configure_ssl_ver(al)
+
+        if al.ciphers:
+            configure_ssl_ciphers(al)
+    else:
+        print("\033[33m  ssl module does not exist; cannot enable https\033[0m\n")
 
     SvcHub(al).run()
 

copyparty/__version__.py

@@ -1,8 +1,8 @@
 # coding: utf-8
 
-VERSION = (0, 6, 3)
-CODENAME = "CHRISTMAAAAAS"
-BUILD_DT = (2021, 1, 7)
+VERSION = (0, 7, 7)
+CODENAME = "keeping track"
+BUILD_DT = (2021, 2, 14)
 
 S_VERSION = ".".join(map(str, VERSION))
 S_BUILD_DT = "{0:04d}-{1:02d}-{2:02d}".format(*BUILD_DT)

copyparty/authsrv.py

@@ -12,11 +12,12 @@ from .util import undot, Pebkac, fsdec, fsenc
 class VFS(object):
     """single level in the virtual fs"""
 
-    def __init__(self, realpath, vpath, uread=[], uwrite=[]):
+    def __init__(self, realpath, vpath, uread=[], uwrite=[], flags={}):
         self.realpath = realpath  # absolute path on host filesystem
         self.vpath = vpath  # absolute path in the virtual filesystem
         self.uread = uread  # users who can read this
         self.uwrite = uwrite  # users who can write this
+        self.flags = flags  # config switches
         self.nodes = {}  # child nodes
 
     def add(self, src, dst):
@@ -36,6 +37,7 @@ class VFS(object):
                 "{}/{}".format(self.vpath, name).lstrip("/"),
                 self.uread,
                 self.uwrite,
+                self.flags,
             )
             self.nodes[name] = vn
             return vn.add(src, dst)
@@ -128,11 +130,10 @@ class VFS(object):
 class AuthSrv(object):
     """verifies users against given paths"""
 
-    def __init__(self, args, log_func):
-        self.log_func = log_func
+    def __init__(self, args, log_func, warn_anonwrite=True):
         self.args = args
-
-        self.warn_anonwrite = True
+        self.log_func = log_func
+        self.warn_anonwrite = warn_anonwrite
 
         if WINDOWS:
             self.re_vol = re.compile(r"^([a-zA-Z]:[\\/][^:]*|[^:]*):([^:]*):(.*)$")
@@ -161,7 +162,7 @@ class AuthSrv(object):
 
         yield prev, True
 
-    def _parse_config_file(self, fd, user, mread, mwrite, mount):
+    def _parse_config_file(self, fd, user, mread, mwrite, mflags, mount):
         vol_src = None
         vol_dst = None
         for ln in [x.decode("utf-8").strip() for x in fd]:
@@ -191,6 +192,7 @@ class AuthSrv(object):
                 mount[vol_dst] = vol_src
                 mread[vol_dst] = []
                 mwrite[vol_dst] = []
+                mflags[vol_dst] = {}
                 continue
 
             lvl, uname = ln.split(" ")
@@ -198,6 +200,9 @@ class AuthSrv(object):
                 mread[vol_dst].append(uname)
             if lvl in "wa":
                 mwrite[vol_dst].append(uname)
+            if lvl == "c":
+                # config option, currently switches only
+                mflags[vol_dst][uname] = True
 
     def reload(self):
         """
@@ -210,6 +215,7 @@ class AuthSrv(object):
         user = {}  # username:password
         mread = {}  # mountpoint:[username]
         mwrite = {}  # mountpoint:[username]
+        mflags = {}  # mountpoint:[flag]
         mount = {}  # dst:src (mountpoint:realpath)
 
         if self.args.a:
@@ -232,9 +238,13 @@ class AuthSrv(object):
                 mount[dst] = src
                 mread[dst] = []
                 mwrite[dst] = []
+                mflags[dst] = {}
 
                 perms = perms.split(":")
                 for (lvl, uname) in [[x[0], x[1:]] for x in perms]:
+                    if lvl == "c":
+                        # config option, currently switches only
+                        mflags[dst][uname] = True
                     if uname == "":
                         uname = "*"
                     if lvl in "ra":
@@ -245,14 +255,15 @@ class AuthSrv(object):
         if self.args.c:
             for cfg_fn in self.args.c:
                 with open(cfg_fn, "rb") as f:
-                    self._parse_config_file(f, user, mread, mwrite, mount)
+                    self._parse_config_file(f, user, mread, mwrite, mflags, mount)
 
+        self.all_writable = []
         if not mount:
             # -h says our defaults are CWD at root and read/write for everyone
             vfs = VFS(os.path.abspath("."), "", ["*"], ["*"])
         elif "" not in mount:
             # there's volumes but no root; make root inaccessible
-            vfs = VFS(os.path.abspath("."), "", [], [])
+            vfs = VFS(os.path.abspath("."), "")
 
         maxdepth = 0
         for dst in sorted(mount.keys(), key=lambda x: (x.count("/"), len(x))):
@@ -262,12 +273,18 @@ class AuthSrv(object):
 
             if dst == "":
                 # rootfs was mapped; fully replaces the default CWD vfs
-                vfs = VFS(mount[dst], dst, mread[dst], mwrite[dst])
+                vfs = VFS(mount[dst], dst, mread[dst], mwrite[dst], mflags[dst])
                 continue
 
             v = vfs.add(mount[dst], dst)
             v.uread = mread[dst]
             v.uwrite = mwrite[dst]
+            v.flags = mflags[dst]
+            if v.uwrite:
+                self.all_writable.append(v)
+
+        if vfs.uwrite and vfs not in self.all_writable:
+            self.all_writable.append(vfs)
 
         missing_users = {}
         for d in [mread, mwrite]:

copyparty/broker_mpw.py

@@ -73,7 +73,7 @@ class MpWorker(object):
                 if PY2:
                     sck = pickle.loads(sck)  # nosec
 
-                self.log("%s %s" % addr, "-" * 4 + "C-qpop")
+                self.log("%s %s" % addr, "\033[1;30m|%sC-qpop\033[0m" % ("-" * 4,))
                 self.httpsrv.accept(sck, addr)
 
                 with self.mutex:

copyparty/broker_thr.py

@@ -28,7 +28,7 @@ class BrokerThr(object):
     def put(self, want_retval, dest, *args):
         if dest == "httpconn":
             sck, addr = args
-            self.log("%s %s" % addr, "-" * 4 + "C-qpop")
+            self.log("%s %s" % addr, "\033[1;30m|%sC-qpop\033[0m" % ("-" * 4,))
             self.httpsrv.accept(sck, addr)
 
         else:

copyparty/httpcli.py

@@ -28,11 +28,13 @@ class HttpCli(object):
         self.conn = conn
         self.s = conn.s
         self.sr = conn.sr
+        self.ip = conn.addr[0]
         self.addr = conn.addr
         self.args = conn.args
         self.auth = conn.auth
         self.log_func = conn.log_func
         self.log_src = conn.log_src
+        self.tls = hasattr(self.s, "cipher")
 
         self.bufsz = 1024 * 32
         self.absolute_urls = False
@@ -42,7 +44,7 @@ class HttpCli(object):
         self.log_func(self.log_src, msg)
 
     def _check_nonfatal(self, ex):
-        return ex.code in [404]
+        return ex.code < 400 or ex.code == 404
 
     def _assert_safe_rem(self, rem):
         # sanity check to prevent any disasters
@@ -85,7 +87,8 @@ class HttpCli(object):
 
         v = self.headers.get("x-forwarded-for", None)
         if v is not None and self.conn.addr[0] in ["127.0.0.1", "::1"]:
-            self.log_src = self.conn.set_rproxy(v.split(",")[0])
+            self.ip = v.split(",")[0]
+            self.log_src = self.conn.set_rproxy(self.ip)
 
         self.uname = "*"
         if "cookie" in self.headers:
@@ -292,26 +295,51 @@ class HttpCli(object):
         if "application/octet-stream" in ctype:
             return self.handle_post_binary()
 
-        raise Pebkac(405, "don't know how to handle {} POST".format(ctype))
+        if "application/x-www-form-urlencoded" in ctype:
+            opt = self.args.urlform
+            if "stash" in opt:
+                return self.handle_stash()
 
-    def handle_stash(self):
+            if "save" in opt:
+                post_sz, _, _, path = self.dump_to_file()
+                self.log("urlform: {} bytes, {}".format(post_sz, path))
+            elif "print" in opt:
+                reader, _ = self.get_body_reader()
+                for buf in reader:
+                    buf = buf.decode("utf-8", "replace")
+                    self.log("urlform:\n  {}\n".format(buf))
+
+            if "get" in opt:
+                return self.handle_get()
+
+            raise Pebkac(405, "POST({}) is disabled".format(ctype))
+
+        raise Pebkac(405, "don't know how to handle POST({})".format(ctype))
+
+    def get_body_reader(self):
         remains = int(self.headers.get("content-length", None))
         if remains is None:
-            reader = read_socket_unbounded(self.sr)
             self.keepalive = False
+            return read_socket_unbounded(self.sr), remains
         else:
-            reader = read_socket(self.sr, remains)
+            return read_socket(self.sr, remains), remains
 
+    def dump_to_file(self):
+        reader, remains = self.get_body_reader()
         vfs, rem = self.conn.auth.vfs.get(self.vpath, self.uname, False, True)
         fdir = os.path.join(vfs.realpath, rem)
 
-        addr = self.conn.addr[0].replace(":", ".")
+        addr = self.ip.replace(":", ".")
         fn = "put-{:.6f}-{}.bin".format(time.time(), addr)
         path = os.path.join(fdir, fn)
 
         with open(path, "wb", 512 * 1024) as f:
             post_sz, _, sha_b64 = hashcopy(self.conn, reader, f)
 
+        return post_sz, sha_b64, remains, path
+
+    def handle_stash(self):
+        post_sz, sha_b64, remains, path = self.dump_to_file()
         spd = self._spd(post_sz)
         self.log("{} wrote {}/{} bytes to {}".format(spd, post_sz, remains, path))
         self.reply("{}\n{}\n".format(post_sz, sha_b64).encode("utf-8"))
@@ -384,9 +412,11 @@ class HttpCli(object):
 
         vfs, rem = self.conn.auth.vfs.get(self.vpath, self.uname, False, True)
 
-        body["vdir"] = self.vpath
-        body["rdir"] = os.path.join(vfs.realpath, rem)
-        body["addr"] = self.addr[0]
+        body["vtop"] = vfs.vpath
+        body["ptop"] = vfs.realpath
+        body["prel"] = rem
+        body["addr"] = self.ip
+        body["flag"] = vfs.flags
 
         x = self.conn.hsrv.broker.put(True, "up2k.handle_json", body)
         response = x.get()
@@ -408,7 +438,10 @@ class HttpCli(object):
         except KeyError:
             raise Pebkac(400, "need hash and wark headers for binary POST")
 
-        x = self.conn.hsrv.broker.put(True, "up2k.handle_chunk", wark, chash)
+        vfs, _ = self.conn.auth.vfs.get(self.vpath, self.uname, False, True)
+        ptop = vfs.realpath
+
+        x = self.conn.hsrv.broker.put(True, "up2k.handle_chunk", ptop, wark, chash)
         response = x.get()
         chunksize, cstart, path, lastmod = response
 
@@ -453,8 +486,13 @@ class HttpCli(object):
 
                 self.log("clone {} done".format(cstart[0]))
 
-        x = self.conn.hsrv.broker.put(True, "up2k.confirm_chunk", wark, chash)
-        num_left = x.get()
+        x = self.conn.hsrv.broker.put(True, "up2k.confirm_chunk", ptop, wark, chash)
+        x = x.get()
+        try:
+            num_left, path = x
+        except:
+            self.loud_reply(x, status=500)
+            return False
 
         if not WINDOWS and num_left == 0:
             times = (int(time.time()), int(lastmod))
@@ -510,10 +548,9 @@ class HttpCli(object):
                 raise Pebkac(500, "mkdir failed, check the logs")
 
         vpath = "{}/{}".format(self.vpath, sanitized).lstrip("/")
+        esc_paths = [quotep(vpath), html_escape(vpath)]
         html = self.conn.tpl_msg.render(
-            h2='<a href="/{}">go to /{}</a>'.format(
-                quotep(vpath), html_escape(vpath)
-            ),
+            h2='<a href="/{}">go to /{}</a>'.format(*esc_paths),
             pre="aight",
             click=True,
         )
@@ -575,7 +612,7 @@ class HttpCli(object):
                     if not os.path.isdir(fsenc(fdir)):
                         raise Pebkac(404, "that folder does not exist")
 
-                    suffix = ".{:.6f}-{}".format(time.time(), self.addr[0])
+                    suffix = ".{:.6f}-{}".format(time.time(), self.ip)
                     open_args = {"fdir": fdir, "suffix": suffix}
                 else:
                     open_args = {}
@@ -637,7 +674,7 @@ class HttpCli(object):
                         "\n".join(
                             unicode(x)
                             for x in [
-                                ":".join(unicode(x) for x in self.addr),
+                                ":".join(unicode(x) for x in [self.ip, self.addr[1]]),
                                 msg.rstrip(),
                             ]
                         )
@@ -790,6 +827,8 @@ class HttpCli(object):
                 editions[ext or "plain"] = [fs_path, st.st_size]
             except:
                 pass
+            if not self.vpath.startswith(".cpr/"):
+                break
 
         if not editions:
             raise Pebkac(404)
@@ -894,8 +933,11 @@ class HttpCli(object):
             open_func = open
             # 512 kB is optimal for huge files, use 64k
             open_args = [fsenc(fs_path), "rb", 64 * 1024]
-            if hasattr(os, 'sendfile'):
-                use_sendfile = not self.args.no_sendfile
+            use_sendfile = (
+                not self.tls  #
+                and not self.args.no_sendfile
+                and hasattr(os, "sendfile")
+            )
 
         #
         # send reply
@@ -922,7 +964,7 @@ class HttpCli(object):
                 remains = sendfile_kern(lower, upper, f, self.s)
             else:
                 remains = sendfile_py(lower, upper, f, self.s)
-        
+
         if remains > 0:
             logmsg += " \033[31m" + str(upper - remains) + "\033[0m"
 
@@ -1020,6 +1062,10 @@ class HttpCli(object):
             if abspath.endswith(".md") and "raw" not in self.uparam:
                 return self.tx_md(abspath)
 
+            bad = "{0}.hist{0}up2k.".format(os.sep)
+            if abspath.endswith(bad + "db") or abspath.endswith(bad + "snap"):
+                raise Pebkac(403)
+
             return self.tx_file(abspath)
 
         fsroot, vfs_ls, vfs_virt = vn.ls(rem, self.uname)
@@ -1045,6 +1091,10 @@ class HttpCli(object):
         if not self.args.ed or "dots" not in self.uparam:
             vfs_ls = exclude_dotfiles(vfs_ls)
 
+        hidden = []
+        if fsroot.endswith(str(os.sep) + ".hist"):
+            hidden = ["up2k.db", "up2k.snap"]
+
         dirs = []
         files = []
         for fn in vfs_ls:
@@ -1056,6 +1106,8 @@ class HttpCli(object):
 
             if fn in vfs_virt:
                 fspath = vfs_virt[fn].realpath
+            elif fn in hidden:
+                continue
             else:
                 fspath = fsroot + "/" + fn
 

copyparty/httpconn.py

@@ -3,10 +3,15 @@ from __future__ import print_function, unicode_literals
 
 import os
 import sys
-import ssl
 import time
 import socket
 
+HAVE_SSL = True
+try:
+    import ssl
+except:
+    HAVE_SSL = False
+
 try:
     import jinja2
 except ImportError:
@@ -65,6 +70,7 @@ class HttpConn(object):
             color = 34
             self.rproxy = ip
 
+        self.ip = ip
         self.log_src = "{} \033[{}m{}".format(ip, color, self.addr[1]).ljust(26)
         return self.log_src
 
@@ -74,9 +80,8 @@ class HttpConn(object):
     def log(self, msg):
         self.log_func(self.log_src, msg)
 
-    def run(self):
+    def _detect_https(self):
         method = None
-        self.sr = None
         if self.cert_path:
             try:
                 method = self.s.recv(4, socket.MSG_PEEK)
@@ -101,16 +106,58 @@ class HttpConn(object):
                 self.s.send(b"HTTP/1.1 400 Bad Request\r\n\r\n" + err.encode("utf-8"))
                 return
 
-        if method not in [None, b"GET ", b"HEAD", b"POST", b"PUT ", b"OPTI"]:
+        return method not in [None, b"GET ", b"HEAD", b"POST", b"PUT ", b"OPTI"]
+
+    def run(self):
+        self.sr = None
+        if self.args.https_only:
+            is_https = True
+        elif self.args.http_only or not HAVE_SSL:
+            is_https = False
+        else:
+            is_https = self._detect_https()
+
+        if is_https:
             if self.sr:
                 self.log("\033[1;31mTODO: cannot do https in jython\033[0m")
                 return
 
             self.log_src = self.log_src.replace("[36m", "[35m")
             try:
-                self.s = ssl.wrap_socket(
-                    self.s, server_side=True, certfile=self.cert_path
-                )
+                ctx = ssl.create_default_context(ssl.Purpose.CLIENT_AUTH)
+                ctx.load_cert_chain(self.cert_path)
+                if self.args.ssl_ver:
+                    ctx.options &= ~self.args.ssl_flags_en
+                    ctx.options |= self.args.ssl_flags_de
+                    # print(repr(ctx.options))
+
+                if self.args.ssl_log:
+                    try:
+                        ctx.keylog_filename = self.args.ssl_log
+                    except:
+                        self.log("keylog failed; openssl or python too old")
+
+                if self.args.ciphers:
+                    ctx.set_ciphers(self.args.ciphers)
+
+                self.s = ctx.wrap_socket(self.s, server_side=True)
+                msg = [
+                    "\033[1;3{:d}m{}".format(c, s)
+                    for c, s in zip([0, 5, 0], self.s.cipher())
+                ]
+                self.log(" ".join(msg) + "\033[0m")
+
+                if self.args.ssl_dbg and hasattr(self.s, "shared_ciphers"):
+                    overlap = [y[::-1] for y in self.s.shared_ciphers()]
+                    lines = [str(x) for x in (["TLS cipher overlap:"] + overlap)]
+                    self.log("\n".join(lines))
+                    for k, v in [
+                        ["compression", self.s.compression()],
+                        ["ALPN proto", self.s.selected_alpn_protocol()],
+                        ["NPN proto", self.s.selected_npn_protocol()],
+                    ]:
+                        self.log("TLS {}: {}".format(k, v or "nah"))
+
             except Exception as ex:
                 em = str(ex)
 

copyparty/httpsrv.py

@@ -38,7 +38,7 @@ class HttpSrv(object):
 
     def accept(self, sck, addr):
         """takes an incoming tcp connection and creates a thread to handle it"""
-        self.log("%s %s" % addr, "-" * 5 + "C-cthr")
+        self.log("%s %s" % addr, "\033[1;30m|%sC-cthr\033[0m" % ("-" * 5,))
         thr = threading.Thread(target=self.thr_client, args=(sck, addr))
         thr.daemon = True
         thr.start()
@@ -66,11 +66,11 @@ class HttpSrv(object):
                 thr.start()
 
         try:
-            self.log("%s %s" % addr, "-" * 6 + "C-crun")
+            self.log("%s %s" % addr, "\033[1;30m|%sC-crun\033[0m" % ("-" * 6,))
             cli.run()
 
         finally:
-            self.log("%s %s" % addr, "-" * 7 + "C-done")
+            self.log("%s %s" % addr, "\033[1;30m|%sC-cdone\033[0m" % ("-" * 7,))
             try:
                 sck.shutdown(socket.SHUT_RDWR)
                 sck.close()
@@ -78,7 +78,7 @@ class HttpSrv(object):
                 if not MACOS:
                     self.log(
                         "%s %s" % addr,
-                        "shut_rdwr err:\n  {}\n  {}".format(repr(sck), ex),
+                        "\033[1;30mshut({}): {}\033[0m".format(sck.fileno(), ex),
                     )
                 if ex.errno not in [10038, 10054, 107, 57, 9]:
                     # 10038 No longer considered a socket

copyparty/svchub.py

@@ -9,6 +9,7 @@ from datetime import datetime, timedelta
 import calendar
 
 from .__init__ import PY2, WINDOWS, MACOS, VT100
+from .authsrv import AuthSrv
 from .tcpsrv import TcpSrv
 from .up2k import Up2k
 from .util import mp
@@ -38,6 +39,10 @@ class SvcHub(object):
         self.tcpsrv = TcpSrv(self)
         self.up2k = Up2k(self)
 
+        if self.args.e2d and self.args.e2s:
+            auth = AuthSrv(self.args, self.log, False)
+            self.up2k.build_indexes(auth.all_writable)
+
         # decide which worker impl to use
         if self.check_mp_enable():
             from .broker_mp import BrokerMp as Broker

copyparty/tcpsrv.py

@@ -4,6 +4,7 @@ from __future__ import print_function, unicode_literals
 import re
 import time
 import socket
+import select
 
 from .util import chkcmd, Counter
 
@@ -23,56 +24,73 @@ class TcpSrv(object):
 
         ip = "127.0.0.1"
         eps = {ip: "local only"}
-        if self.args.i != ip:
-            eps = self.detect_interfaces(self.args.i) or {self.args.i: "external"}
+        nonlocals = [x for x in self.args.i if x != ip]
+        if nonlocals:
+            eps = self.detect_interfaces(self.args.i)
+            if not eps:
+                for x in nonlocals:
+                    eps[x] = "external"
 
         for ip, desc in sorted(eps.items(), key=lambda x: x[1]):
-            self.log(
-                "tcpsrv",
-                "available @ http://{}:{}/  (\033[33m{}\033[0m)".format(
-                    ip, self.args.p, desc
-                ),
-            )
+            for port in sorted(self.args.p):
+                self.log(
+                    "tcpsrv",
+                    "available @ http://{}:{}/  (\033[33m{}\033[0m)".format(
+                        ip, port, desc
+                    ),
+                )
 
-        self.srv = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
-        self.srv.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
-        self.srv.setsockopt(socket.IPPROTO_TCP, socket.TCP_NODELAY, 1)
+        self.srv = []
+        for ip in self.args.i:
+            for port in self.args.p:
+                self.srv.append(self._listen(ip, port))
+
+    def _listen(self, ip, port):
+        srv = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+        srv.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
+        srv.setsockopt(socket.IPPROTO_TCP, socket.TCP_NODELAY, 1)
         try:
-            self.srv.bind((self.args.i, self.args.p))
+            srv.bind((ip, port))
+            return srv
         except (OSError, socket.error) as ex:
-            if ex.errno == 98:
-                raise Exception(
-                    "\033[1;31mport {} is busy on interface {}\033[0m".format(
-                        self.args.p, self.args.i
-                    )
-                )
-
-            if ex.errno == 99:
-                raise Exception(
-                    "\033[1;31minterface {} does not exist\033[0m".format(self.args.i)
-                )
+            if ex.errno in [98, 48]:
+                e = "\033[1;31mport {} is busy on interface {}\033[0m".format(port, ip)
+            elif ex.errno in [99, 49]:
+                e = "\033[1;31minterface {} does not exist\033[0m".format(ip)
+            else:
+                raise
+            raise Exception(e)
 
     def run(self):
-        self.srv.listen(self.args.nc)
-
-        self.log("tcpsrv", "listening @ {0}:{1}".format(self.args.i, self.args.p))
+        for srv in self.srv:
+            srv.listen(self.args.nc)
+            ip, port = srv.getsockname()
+            self.log("tcpsrv", "listening @ {0}:{1}".format(ip, port))
 
         while True:
-            self.log("tcpsrv", "-" * 1 + "C-ncli")
+            self.log("tcpsrv", "\033[1;30m|%sC-ncli\033[0m" % ("-" * 1,))
             if self.num_clients.v >= self.args.nc:
                 time.sleep(0.1)
                 continue
 
-            self.log("tcpsrv", "-" * 2 + "C-acc1")
-            sck, addr = self.srv.accept()
-            self.log("%s %s" % addr, "-" * 3 + "C-acc2")
-            self.num_clients.add()
-            self.hub.broker.put(False, "httpconn", sck, addr)
+            self.log("tcpsrv", "\033[1;30m|%sC-acc1\033[0m" % ("-" * 2,))
+            ready, _, _ = select.select(self.srv, [], [])
+            for srv in ready:
+                sck, addr = srv.accept()
+                sip, sport = srv.getsockname()
+                self.log(
+                    "%s %s" % addr,
+                    "\033[1;30m|{}C-acc2 \033[0;36m{} \033[3{}m{}".format(
+                        "-" * 3, sip, sport % 8, sport
+                    ),
+                )
+                self.num_clients.add()
+                self.hub.broker.put(False, "httpconn", sck, addr)
 
     def shutdown(self):
         self.log("tcpsrv", "ok bye")
 
-    def detect_interfaces(self, listen_ip):
+    def detect_interfaces(self, listen_ips):
         eps = {}
 
         # get all ips and their interfaces
@@ -86,8 +104,9 @@ class TcpSrv(object):
             for ln in ip_addr.split("\n"):
                 try:
                     ip, dev = r.match(ln.rstrip()).groups()
-                    if listen_ip in ["0.0.0.0", ip]:
-                        eps[ip] = dev
+                    for lip in listen_ips:
+                        if lip in ["0.0.0.0", ip]:
+                            eps[ip] = dev
                 except:
                     pass
 
@@ -114,11 +133,12 @@ class TcpSrv(object):
 
         s.close()
 
-        if default_route and listen_ip in ["0.0.0.0", default_route]:
-            desc = "\033[32mexternal"
-            try:
-                eps[default_route] += ", " + desc
-            except:
-                eps[default_route] = desc
+        for lip in listen_ips:
+            if default_route and lip in ["0.0.0.0", default_route]:
+                desc = "\033[32mexternal"
+                try:
+                    eps[default_route] += ", " + desc
+                except:
+                    eps[default_route] = desc
 
         return eps

copyparty/up2k.py

@@ -2,10 +2,14 @@
 from __future__ import print_function, unicode_literals
 
 
-import os
 import re
+import os
+import sys
 import time
 import math
+import json
+import gzip
+import stat
 import shutil
 import base64
 import hashlib
@@ -13,7 +17,15 @@ import threading
 from copy import deepcopy
 
 from .__init__ import WINDOWS
-from .util import Pebkac, Queue, fsenc, sanitize_fn, ren_open
+from .util import Pebkac, Queue, fsdec, fsenc, sanitize_fn, ren_open, atomic_move
+
+HAVE_SQLITE3 = False
+try:
+    import sqlite3
+
+    HAVE_SQLITE3 = True
+except:
+    pass
 
 
 class Up2k(object):
@@ -22,20 +34,21 @@ class Up2k(object):
       * documentation
       * registry persistence
         * ~/.config flatfiles for active jobs
-        * wark->path database for finished uploads
     """
 
     def __init__(self, broker):
         self.broker = broker
         self.args = broker.args
         self.log = broker.log
+        self.persist = self.args.e2d
 
         # config
         self.salt = "hunter2"  # TODO: config
 
         # state
-        self.registry = {}
         self.mutex = threading.Lock()
+        self.registry = {}
+        self.db = {}
 
         if WINDOWS:
             # usually fails to set lastmod too quickly
@@ -44,53 +57,313 @@ class Up2k(object):
             thr.daemon = True
             thr.start()
 
+        if self.persist:
+            thr = threading.Thread(target=self._snapshot)
+            thr.daemon = True
+            thr.start()
+
         # static
         self.r_hash = re.compile("^[0-9a-zA-Z_-]{43}$")
 
+        if self.persist and not HAVE_SQLITE3:
+            m = "could not initialize sqlite3, will use in-memory registry only"
+            self.log("up2k", m)
+
+    def _vis_job_progress(self, job):
+        perc = 100 - (len(job["need"]) * 100.0 / len(job["hash"]))
+        path = os.path.join(job["ptop"], job["prel"], job["name"])
+        return "{:5.1f}% {}".format(perc, path)
+
+    def _vis_reg_progress(self, reg):
+        ret = []
+        for _, job in reg.items():
+            ret.append(self._vis_job_progress(job))
+
+        return ret
+
+    def register_vpath(self, ptop):
+        with self.mutex:
+            if ptop in self.registry:
+                return None
+
+            reg = {}
+            path = os.path.join(ptop, ".hist", "up2k.snap")
+            if self.persist and os.path.exists(path):
+                with gzip.GzipFile(path, "rb") as f:
+                    j = f.read().decode("utf-8")
+
+                reg = json.loads(j)
+                for _, job in reg.items():
+                    job["poke"] = time.time()
+
+                m = "loaded snap {} |{}|".format(path, len(reg.keys()))
+                m = [m] + self._vis_reg_progress(reg)
+                self.log("up2k", "\n".join(m))
+
+            self.registry[ptop] = reg
+            if not self.persist or not HAVE_SQLITE3:
+                return None
+
+            try:
+                os.mkdir(os.path.join(ptop, ".hist"))
+            except:
+                pass
+
+            db_path = os.path.join(ptop, ".hist", "up2k.db")
+            if ptop in self.db:
+                # self.db[ptop].close()
+                return None
+
+            try:
+                db = self._open_db(db_path)
+                self.db[ptop] = db
+                return db
+            except Exception as ex:
+                m = "failed to open [{}]: {}".format(ptop, repr(ex))
+                self.log("up2k", m)
+
+            return None
+
+    def build_indexes(self, writeables):
+        tops = [d.realpath for d in writeables]
+        for top in tops:
+            db = self.register_vpath(top)
+            if db:
+                # can be symlink so don't `and d.startswith(top)``
+                excl = set([d for d in tops if d != top])
+                dbw = [db, 0, time.time()]
+                self._build_dir(dbw, top, excl, top)
+                self._drop_lost(db, top)
+                if dbw[1]:
+                    self.log("up2k", "commit {} new files".format(dbw[1]))
+
+                db.commit()
+
+    def _build_dir(self, dbw, top, excl, cdir):
+        try:
+            inodes = [fsdec(x) for x in os.listdir(fsenc(cdir))]
+        except Exception as ex:
+            self.log("up2k", "listdir: {} @ [{}]".format(repr(ex), cdir))
+            return
+
+        histdir = os.path.join(top, ".hist")
+        for inode in inodes:
+            abspath = os.path.join(cdir, inode)
+            try:
+                inf = os.stat(fsenc(abspath))
+            except Exception as ex:
+                self.log("up2k", "stat: {} @ [{}]".format(repr(ex), abspath))
+                continue
+
+            if stat.S_ISDIR(inf.st_mode):
+                if abspath in excl or abspath == histdir:
+                    continue
+                # self.log("up2k", " dir: {}".format(abspath))
+                self._build_dir(dbw, top, excl, abspath)
+            else:
+                # self.log("up2k", "file: {}".format(abspath))
+                rp = abspath[len(top) :].replace("\\", "/").strip("/")
+                c = dbw[0].execute("select * from up where rp = ?", (rp,))
+                in_db = list(c.fetchall())
+                if in_db:
+                    _, dts, dsz, _ = in_db[0]
+                    if len(in_db) > 1:
+                        m = "WARN: multiple entries: [{}] => [{}] ({})"
+                        self.log("up2k", m.format(top, rp, len(in_db)))
+                        dts = -1
+
+                    if dts == inf.st_mtime and dsz == inf.st_size:
+                        continue
+
+                    m = "reindex [{}] => [{}] ({}/{}) ({}/{})".format(
+                        top, rp, dts, inf.st_mtime, dsz, inf.st_size
+                    )
+                    self.log("up2k", m)
+                    self.db_rm(dbw[0], rp)
+                    dbw[1] += 1
+                    in_db = None
+
+                self.log("up2k", "file: {}".format(abspath))
+                try:
+                    hashes = self._hashlist_from_file(abspath)
+                except Exception as ex:
+                    self.log("up2k", "hash: {} @ [{}]".format(repr(ex), abspath))
+                    continue
+
+                wark = self._wark_from_hashlist(inf.st_size, hashes)
+                self.db_add(dbw[0], wark, rp, inf.st_mtime, inf.st_size)
+                dbw[1] += 1
+                td = time.time() - dbw[2]
+                if dbw[1] > 1024 or td > 60:
+                    self.log("up2k", "commit {} new files".format(dbw[1]))
+                    dbw[0].commit()
+                    dbw[1] = 0
+                    dbw[2] = time.time()
+
+    def _drop_lost(self, db, top):
+        rm = []
+        c = db.execute("select * from up")
+        for dwark, dts, dsz, drp in c:
+            abspath = os.path.join(top, drp)
+            try:
+                if not os.path.exists(fsenc(abspath)):
+                    rm.append(drp)
+            except Exception as ex:
+                self.log("up2k", "stat-rm: {} @ [{}]".format(repr(ex), abspath))
+
+        if not rm:
+            return
+
+        self.log("up2k", "forgetting {} deleted files".format(len(rm)))
+        for rp in rm:
+            self.db_rm(db, rp)
+
+    def _open_db(self, db_path):
+        conn = sqlite3.connect(db_path, check_same_thread=False)
+        try:
+            c = conn.execute(r"select * from kv where k = 'sver'")
+            rows = c.fetchall()
+            if rows:
+                ver = rows[0][1]
+            else:
+                self.log("up2k", "WARN: no sver in kv, DB corrupt?")
+                ver = "unknown"
+
+            if ver == "1":
+                try:
+                    nfiles = next(conn.execute("select count(w) from up"))[0]
+                    self.log("up2k", "found DB at {} |{}|".format(db_path, nfiles))
+                    return conn
+                except Exception as ex:
+                    m = "WARN: could not list files, DB corrupt?\n  " + repr(ex)
+                    self.log("up2k", m)
+
+            m = "REPLACING unsupported DB (v.{}) at {}".format(ver, db_path)
+            self.log("up2k", m)
+            conn.close()
+            os.unlink(db_path)
+            conn = sqlite3.connect(db_path, check_same_thread=False)
+        except:
+            pass
+
+        # sqlite is variable-width only, no point in using char/nchar/varchar
+        for cmd in [
+            r"create table kv (k text, v text)",
+            r"create table up (w text, mt int, sz int, rp text)",
+            r"insert into kv values ('sver', '1')",
+            r"create index up_w on up(w)",
+        ]:
+            conn.execute(cmd)
+
+        conn.commit()
+        self.log("up2k", "created DB at {}".format(db_path))
+        return conn
+
     def handle_json(self, cj):
+        self.register_vpath(cj["ptop"])
         cj["name"] = sanitize_fn(cj["name"])
+        cj["poke"] = time.time()
         wark = self._get_wark(cj)
         now = time.time()
+        job = None
         with self.mutex:
-            # TODO use registry persistence here to symlink any matching wark
-            if wark in self.registry:
-                job = self.registry[wark]
-                if job["rdir"] != cj["rdir"] or job["name"] != cj["name"]:
-                    src = os.path.join(job["rdir"], job["name"])
-                    dst = os.path.join(cj["rdir"], cj["name"])
+            db = self.db.get(cj["ptop"], None)
+            reg = self.registry[cj["ptop"]]
+            if db:
+                cur = db.execute(r"select * from up where w = ?", (wark,))
+                for _, dtime, dsize, dp_rel in cur:
+                    dp_abs = os.path.join(cj["ptop"], dp_rel).replace("\\", "/")
+                    # relying on path.exists to return false on broken symlinks
+                    if os.path.exists(fsenc(dp_abs)):
+                        try:
+                            prel, name = dp_rel.rsplit("/", 1)
+                        except:
+                            prel = ""
+                            name = dp_rel
+
+                        job = {
+                            "name": name,
+                            "prel": prel,
+                            "vtop": cj["vtop"],
+                            "ptop": cj["ptop"],
+                            "flag": cj["flag"],
+                            "size": dsize,
+                            "lmod": dtime,
+                            "hash": [],
+                            "need": [],
+                        }
+                        break
+
+                if job and wark in reg:
+                    del reg[wark]
+
+            if job or wark in reg:
+                job = job or reg[wark]
+                if job["prel"] == cj["prel"] and job["name"] == cj["name"]:
+                    # ensure the files haven't been deleted manually
+                    names = [job[x] for x in ["name", "tnam"] if x in job]
+                    for fn in names:
+                        path = os.path.join(job["ptop"], job["prel"], fn)
+                        try:
+                            if os.path.getsize(path) > 0:
+                                # upload completed or both present
+                                break
+                        except:
+                            # missing; restart
+                            job = None
+                            break
+                else:
+                    # file contents match, but not the path
+                    src = os.path.join(job["ptop"], job["prel"], job["name"])
+                    dst = os.path.join(cj["ptop"], cj["prel"], cj["name"])
+                    vsrc = os.path.join(job["vtop"], job["prel"], job["name"])
+                    vsrc = vsrc.replace("\\", "/")  # just for prints anyways
                     if job["need"]:
                         self.log("up2k", "unfinished:\n  {0}\n  {1}".format(src, dst))
-                        err = "partial upload exists at a different location; please resume uploading here instead:\n{0}{1} ".format(
-                            job["vdir"], job["name"]
-                        )
+                        err = "partial upload exists at a different location; please resume uploading here instead:\n"
+                        err += vsrc + " "
+                        raise Pebkac(400, err)
+                    elif "nodupe" in job["flag"]:
+                        self.log("up2k", "dupe-reject:\n  {0}\n  {1}".format(src, dst))
+                        err = "upload rejected, file already exists:\n " + vsrc + " "
                         raise Pebkac(400, err)
                     else:
                         # symlink to the client-provided name,
                         # returning the previous upload info
                         job = deepcopy(job)
-                        job["rdir"] = cj["rdir"]
-                        job["name"] = self._untaken(cj["rdir"], cj["name"], now, cj["addr"])
-                        dst = os.path.join(job["rdir"], job["name"])
+                        for k in ["ptop", "vtop", "prel"]:
+                            job[k] = cj[k]
+
+                        pdir = os.path.join(cj["ptop"], cj["prel"])
+                        job["name"] = self._untaken(pdir, cj["name"], now, cj["addr"])
+                        dst = os.path.join(job["ptop"], job["prel"], job["name"])
                         os.unlink(fsenc(dst))  # TODO ed pls
                         self._symlink(src, dst)
-            else:
+
+            if not job:
                 job = {
                     "wark": wark,
                     "t0": now,
-                    "addr": cj["addr"],
-                    "vdir": cj["vdir"],
-                    "rdir": cj["rdir"],
-                    # client-provided, sanitized by _get_wark:
-                    "name": cj["name"],
-                    "size": cj["size"],
-                    "lmod": cj["lmod"],
                     "hash": deepcopy(cj["hash"]),
+                    "need": [],
                 }
+                # client-provided, sanitized by _get_wark: name, size, lmod
+                for k in [
+                    "addr",
+                    "vtop",
+                    "ptop",
+                    "prel",
+                    "flag",
+                    "name",
+                    "size",
+                    "lmod",
+                    "poke",
+                ]:
+                    job[k] = cj[k]
 
                 # one chunk may occur multiple times in a file;
                 # filter to unique values for the list of missing chunks
                 # (preserve order to reduce disk thrashing)
-                job["need"] = []
                 lut = {}
                 for k in cj["hash"]:
                     if k not in lut:
@@ -142,37 +415,62 @@ class Up2k(object):
             self.log("up2k", "cannot symlink; creating copy: " + repr(ex))
             shutil.copy2(fsenc(src), fsenc(dst))
 
-    def handle_chunk(self, wark, chash):
+    def handle_chunk(self, ptop, wark, chash):
         with self.mutex:
-            job = self.registry.get(wark)
+            job = self.registry[ptop].get(wark, None)
             if not job:
-                raise Pebkac(404, "unknown wark")
+                raise Pebkac(400, "unknown wark")
 
             if chash not in job["need"]:
-                raise Pebkac(200, "already got that but thanks??")
+                raise Pebkac(400, "already got that but thanks??")
 
             nchunk = [n for n, v in enumerate(job["hash"]) if v == chash]
             if not nchunk:
-                raise Pebkac(404, "unknown chunk")
+                raise Pebkac(400, "unknown chunk")
+
+        job["poke"] = time.time()
 
         chunksize = self._get_chunksize(job["size"])
         ofs = [chunksize * x for x in nchunk]
 
-        path = os.path.join(job["rdir"], job["name"])
+        path = os.path.join(job["ptop"], job["prel"], job["tnam"])
 
         return [chunksize, ofs, path, job["lmod"]]
 
-    def confirm_chunk(self, wark, chash):
+    def confirm_chunk(self, ptop, wark, chash):
         with self.mutex:
-            job = self.registry[wark]
-            job["need"].remove(chash)
+            try:
+                job = self.registry[ptop][wark]
+                pdir = os.path.join(job["ptop"], job["prel"])
+                src = os.path.join(pdir, job["tnam"])
+                dst = os.path.join(pdir, job["name"])
+            except Exception as ex:
+                return "confirm_chunk, wark, " + repr(ex)
+
+            try:
+                job["need"].remove(chash)
+            except Exception as ex:
+                return "confirm_chunk, chash, " + repr(ex)
+
             ret = len(job["need"])
+            if ret > 0:
+                return ret, src
 
-            if WINDOWS and ret == 0:
-                path = os.path.join(job["rdir"], job["name"])
-                self.lastmod_q.put([path, (int(time.time()), int(job["lmod"]))])
+            atomic_move(src, dst)
 
-            return ret
+            if WINDOWS:
+                self.lastmod_q.put([dst, (int(time.time()), int(job["lmod"]))])
+
+            db = self.db.get(job["ptop"], None)
+            if db:
+                rp = os.path.join(job["prel"], job["name"]).replace("\\", "/")
+                self.db_rm(db, rp)
+                self.db_add(db, job["wark"], rp, job["lmod"], job["size"])
+                db.commit()
+                del self.registry[ptop][wark]
+                # in-memory registry is reserved for unfinished uploads
+
+            return ret, dst
 
     def _get_chunksize(self, filesize):
         chunksize = 1024 * 1024
@@ -186,6 +484,13 @@ class Up2k(object):
                 chunksize += stepsize
                 stepsize *= mul
 
+    def db_rm(self, db, rp):
+        db.execute("delete from up where rp = ?", (rp,))
+
+    def db_add(self, db, wark, rp, ts, sz):
+        v = (wark, ts, sz, rp)
+        db.execute("insert into up values (?,?,?,?)", v)
+
     def _get_wark(self, cj):
         if len(cj["name"]) > 1024 or len(cj["hash"]) > 512 * 1024:  # 16TiB
             raise Pebkac(400, "name or numchunks not according to spec")
@@ -202,9 +507,13 @@ class Up2k(object):
         except:
             cj["lmod"] = int(time.time())
 
-        # server-reproducible file identifier, independent of name or location
-        ident = [self.salt, str(cj["size"])]
-        ident.extend(cj["hash"])
+        wark = self._wark_from_hashlist(cj["size"], cj["hash"])
+        return wark
+
+    def _wark_from_hashlist(self, filesize, hashes):
+        """ server-reproducible file identifier, independent of name or location """
+        ident = [self.salt, str(filesize)]
+        ident.extend(hashes)
         ident = "\n".join(ident)
 
         hasher = hashlib.sha512()
@@ -214,11 +523,48 @@ class Up2k(object):
         wark = base64.urlsafe_b64encode(digest)
         return wark.decode("utf-8").rstrip("=")
 
+    def _hashlist_from_file(self, path):
+        fsz = os.path.getsize(path)
+        csz = self._get_chunksize(fsz)
+        ret = []
+        last_print = time.time()
+        with open(path, "rb", 512 * 1024) as f:
+            while fsz > 0:
+                now = time.time()
+                td = now - last_print
+                if td >= 0.1:
+                    last_print = now
+                    msg = " {} MB   \r".format(int(fsz / 1024 / 1024))
+                    print(msg, end="", file=sys.stderr)
+
+                hashobj = hashlib.sha512()
+                rem = min(csz, fsz)
+                fsz -= rem
+                while rem > 0:
+                    buf = f.read(min(rem, 64 * 1024))
+                    if not buf:
+                        raise Exception("EOF at " + str(f.tell()))
+
+                    hashobj.update(buf)
+                    rem -= len(buf)
+
+                digest = hashobj.digest()[:32]
+                digest = base64.urlsafe_b64encode(digest)
+                ret.append(digest.decode("utf-8").rstrip("="))
+
+        return ret
+
     def _new_upload(self, job):
-        self.registry[job["wark"]] = job
+        self.registry[job["ptop"]][job["wark"]] = job
+        pdir = os.path.join(job["ptop"], job["prel"])
+        job["name"] = self._untaken(pdir, job["name"], job["t0"], job["addr"])
+        # if len(job["name"].split(".")) > 8:
+        #    raise Exception("aaa")
+
+        tnam = job["name"] + ".PARTIAL"
         suffix = ".{:.6f}-{}".format(job["t0"], job["addr"])
-        with ren_open(job["name"], "wb", fdir=job["rdir"], suffix=suffix) as f:
-            f, job["name"] = f["orz"]
+        with ren_open(tnam, "wb", fdir=pdir, suffix=suffix) as f:
+            f, job["tnam"] = f["orz"]
             f.seek(job["size"] - 1)
             f.write(b"e")
 
@@ -231,7 +577,63 @@ class Up2k(object):
             # self.log("lmod", "got {}".format(len(ready)))
             time.sleep(5)
             for path, times in ready:
+                self.log("lmod", "setting times {} on {}".format(times, path))
                 try:
                     os.utime(fsenc(path), times)
                 except:
                     self.log("lmod", "failed to utime ({}, {})".format(path, times))
+
+    def _snapshot(self):
+        persist_interval = 30  # persist unfinished uploads index every 30 sec
+        discard_interval = 21600  # drop unfinished uploads after 6 hours inactivity
+        prev = {}
+        while True:
+            time.sleep(persist_interval)
+            with self.mutex:
+                for k, reg in self.registry.items():
+                    self._snap_reg(prev, k, reg, discard_interval)
+
+    def _snap_reg(self, prev, k, reg, discard_interval):
+        now = time.time()
+        rm = [x for x in reg.values() if now - x["poke"] > discard_interval]
+        if rm:
+            m = "dropping {} abandoned uploads in {}".format(len(rm), k)
+            vis = [self._vis_job_progress(x) for x in rm]
+            self.log("up2k", "\n".join([m] + vis))
+            for job in rm:
+                del reg[job["wark"]]
+                try:
+                    # remove the filename reservation
+                    path = os.path.join(job["ptop"], job["prel"], job["name"])
+                    if os.path.getsize(path) == 0:
+                        os.unlink(path)
+
+                    if len(job["hash"]) == len(job["need"]):
+                        # PARTIAL is empty, delete that too
+                        path = os.path.join(job["ptop"], job["prel"], job["tnam"])
+                        os.unlink(path)
+                except:
+                    pass
+
+        path = os.path.join(k, ".hist", "up2k.snap")
+        if not reg:
+            if k not in prev or prev[k] is not None:
+                prev[k] = None
+                if os.path.exists(path):
+                    os.unlink(path)
+            return
+
+        newest = max(x["poke"] for _, x in reg.items()) if reg else 0
+        etag = [len(reg), newest]
+        if etag == prev.get(k, None):
+            return
+
+        path2 = "{}.{}".format(path, os.getpid())
+        j = json.dumps(reg, indent=2, sort_keys=True).encode("utf-8")
+        with gzip.GzipFile(path2, "wb") as f:
+            f.write(j)
+
+        atomic_move(path2, path)
+
+        self.log("up2k", "snap: {} |{}|".format(path, len(reg.keys())))
+        prev[k] = etag

copyparty/util.py

@@ -108,7 +108,7 @@ def ren_open(fname, *args, **kwargs):
         with open(fname, *args, **kwargs) as f:
             yield {"orz": [f, fname]}
             return
-    
+
     orig_name = fname
     bname = fname
     ext = ""
@@ -549,6 +549,16 @@ else:
     fsdec = w8dec
 
 
+def atomic_move(src, dst):
+    if not PY2:
+        os.replace(src, dst)
+    else:
+        if os.path.exists(dst):
+            os.unlink(dst)
+
+        os.rename(src, dst)
+
+
 def read_socket(sr, total_size):
     remains = total_size
     while remains > 0:
@@ -622,10 +632,10 @@ def sendfile_kern(lower, upper, f, s):
         except Exception as ex:
             # print("sendfile: " + repr(ex))
             n = 0
-        
+
         if n <= 0:
             return upper - ofs
-        
+
         ofs += n
         # print("sendfile: ok, sent {} now, {} total, {} remains".format(n, ofs - lower, upper - ofs))
 
@@ -708,7 +718,26 @@ def py_desc():
     )
 
 
+def align_tab(lines):
+    rows = []
+    ncols = 0
+    for ln in lines:
+        row = [x for x in ln.split(" ") if x]
+        ncols = max(ncols, len(row))
+        rows.append(row)
+
+    lens = [0] * ncols
+    for row in rows:
+        for n, col in enumerate(row):
+            lens[n] = max(lens[n], len(col))
+
+    return ["".join(x.ljust(y + 2) for x, y in zip(row, lens)) for row in rows]
+
+
 class Pebkac(Exception):
     def __init__(self, code, msg=None):
         super(Pebkac, self).__init__(msg or HTTPCODE[code])
         self.code = code
+
+    def __repr__(self):
+        return "Pebkac({}, {})".format(self.code, repr(self.args))

copyparty/web/browser.js

@@ -8,7 +8,14 @@ function dbg(msg) {
 
 function ev(e) {
 	e = e || window.event;
-	e.preventDefault ? e.preventDefault() : (e.returnValue = false);
+
+	if (e.preventDefault)
+		e.preventDefault()
+
+	if (e.stopPropagation)
+		e.stopPropagation();
+
+	e.returnValue = false;
 	return e;
 }
 
@@ -447,7 +454,8 @@ function play(tid, call_depth) {
 	mp.au.tid = tid;
 	mp.au.src = url;
 	mp.au.volume = mp.expvol();
-	setclass('trk' + tid, 'play act');
+	var oid = 'trk' + tid;
+	setclass(oid, 'play act');
 
 	try {
 		if (hack_attempt_play)
@@ -456,7 +464,11 @@ function play(tid, call_depth) {
 		if (mp.au.paused)
 			autoplay_blocked();
 
-		location.hash = 'trk' + tid;
+		var o = ebi(oid);
+		o.setAttribute('id', 'thx_js');
+		location.hash = oid;
+		o.setAttribute('id', oid);
+
 		pbar.drawbuf();
 		return true;
 	}

copyparty/web/up2k.js

@@ -89,6 +89,104 @@ catch (ex) {
 }
 
 
+function up2k_flagbus() {
+    var flag = {
+        "id": Math.floor(Math.random() * 1024 * 1024 * 1023 * 2),
+        "ch": new BroadcastChannel("up2k_flagbus"),
+        "ours": false,
+        "owner": null,
+        "wants": null,
+        "act": false,
+        "last_tx": ["x", null]
+    };
+    var dbg = function (who, msg) {
+        console.log('flagbus(' + flag.id + '): [' + who + '] ' + msg);
+    };
+    flag.ch.onmessage = function (ev) {
+        var who = ev.data[0],
+            what = ev.data[1];
+
+        if (who == flag.id) {
+            dbg(who, 'hi me (??)');
+            return;
+        }
+        flag.act = new Date().getTime();
+        if (what == "want") {
+            // lowest id wins, don't care if that's us
+            if (who < flag.id) {
+                dbg(who, 'wants (ack)');
+                flag.wants = [who, flag.act];
+            }
+            else {
+                dbg(who, 'wants (ign)');
+            }
+        }
+        else if (what == "have") {
+            dbg(who, 'have');
+            flag.owner = [who, flag.act];
+        }
+        else if (what == "give") {
+            if (flag.owner && flag.owner[0] == who) {
+                flag.owner = null;
+                dbg(who, 'give (ok)');
+            }
+            else {
+                dbg(who, 'give, INVALID, ' + flag.owner);
+            }
+        }
+        else if (what == "hi") {
+            dbg(who, 'hi');
+            flag.ch.postMessage([flag.id, "hey"]);
+        }
+        else {
+            dbg('?', ev.data);
+        }
+    };
+    var tx = function (now, msg) {
+        var td = now - flag.last_tx[1];
+        if (td > 500 || flag.last_tx[0] != msg) {
+            dbg('*', 'tx ' + msg);
+            flag.ch.postMessage([flag.id, msg]);
+            flag.last_tx = [msg, now];
+        }
+    };
+    var do_take = function (now) {
+        //dbg('*', 'do_take');
+        tx(now, "have");
+        flag.owner = [flag.id, now];
+        flag.ours = true;
+    };
+    var do_want = function (now) {
+        //dbg('*', 'do_want');
+        tx(now, "want");
+    };
+    flag.take = function (now) {
+        if (flag.ours) {
+            do_take(now);
+            return;
+        }
+        if (flag.owner && now - flag.owner[1] > 5000) {
+            flag.owner = null;
+        }
+        if (flag.wants && now - flag.wants[1] > 5000) {
+            flag.wants = null;
+        }
+        if (!flag.owner && !flag.wants) {
+            do_take(now);
+            return;
+        }
+        do_want(now);
+    };
+    flag.give = function () {
+        dbg('#', 'put give');
+        flag.ch.postMessage([flag.id, "give"]);
+        flag.owner = null;
+        flag.ours = false;
+    };
+    flag.ch.postMessage([flag.id, 'hi']);
+    return flag;
+}
+
 function up2k_init(have_crypto) {
     //have_crypto = false;
     var need_filereader_cache = undefined;
@@ -201,6 +299,8 @@ function up2k_init(have_crypto) {
 
     var parallel_uploads = cfg_get('nthread');
     var multitask = bcfg_get('multitask', true);
+    var ask_up = bcfg_get('ask_up', true);
+    var flag_en = bcfg_get('flag_en', false);
 
     var col_hashing = '#00bbff';
     var col_hashed = '#004466';
@@ -218,6 +318,10 @@ function up2k_init(have_crypto) {
             "hash": [],
             "handshake": [],
             "upload": []
+        },
+        "bytes": {
+            "hashed": 0,
+            "uploaded": 0
         }
     };
 
@@ -228,6 +332,9 @@ function up2k_init(have_crypto) {
     if (!bobslice || !window.FileReader || !window.FileList)
         return un2k("this is the basic uploader; up2k needs at least<br />chrome 21 // firefox 13 // edge 12 // opera 12 // safari 5.1");
 
+    var flag = false;
+    apply_flag_cfg();
+
     function nav() {
         ebi('file' + fdom_ctr).click();
     }
@@ -262,6 +369,7 @@ function up2k_init(have_crypto) {
 
         more_one_file();
         var bad_files = [];
+        var good_files = [];
         for (var a = 0; a < files.length; a++) {
             var fobj = files[a];
             if (is_itemlist) {
@@ -275,9 +383,32 @@ function up2k_init(have_crypto) {
                     throw 1;
             }
             catch (ex) {
-                bad_files.push([a, fobj.name]);
+                bad_files.push(fobj.name);
                 continue;
             }
+            good_files.push(fobj);
+        }
+
+        if (bad_files.length > 0) {
+            var msg = 'These {0} files (of {1} total) were skipped because they are empty:\n'.format(bad_files.length, files.length);
+            for (var a = 0; a < bad_files.length; a++)
+                msg += '-- ' + bad_files[a] + '\n';
+
+            if (files.length - bad_files.length <= 1 && /(android)/i.test(navigator.userAgent))
+                msg += '\nFirefox-Android has a bug which prevents selecting multiple files. Try selecting one file at a time. For more info, see firefox bug 1456557';
+
+            alert(msg);
+        }
+
+        var msg = ['upload these ' + good_files.length + ' files?'];
+        for (var a = 0; a < good_files.length; a++)
+            msg.push(good_files[a].name);
+
+        if (ask_up && !confirm(msg.join('\n')))
+            return;
+
+        for (var a = 0; a < good_files.length; a++) {
+            var fobj = good_files[a];
             var now = new Date().getTime();
             var lmod = fobj.lastModified || now;
             var entry = {
@@ -307,17 +438,6 @@ function up2k_init(have_crypto) {
             st.files.push(entry);
             st.todo.hash.push(entry);
         }
-
-        if (bad_files.length > 0) {
-            var msg = 'These {0} files (of {1} total) were skipped because they are empty:\n'.format(bad_files.length, files.length);
-            for (var a = 0; a < bad_files.length; a++)
-                msg += '-- ' + bad_files[a][1] + '\n';
-
-            if (files.length - bad_files.length <= 1 && /(android)/i.test(navigator.userAgent))
-                msg += '\nFirefox-Android has a bug which prevents selecting multiple files. Try selecting one file at a time. For more info, see firefox bug 1456557';
-
-            alert(msg);
-        }
     }
     ebi('u2btn').addEventListener('drop', gotfile, false);
 
@@ -336,16 +456,20 @@ function up2k_init(have_crypto) {
     //
 
     function handshakes_permitted() {
-        return multitask || (
-            st.todo.upload.length == 0 &&
-            st.busy.upload.length == 0);
+        var lim = multitask ? 1 : 0;
+        return lim >=
+            st.todo.upload.length +
+            st.busy.upload.length;
     }
 
     function hashing_permitted() {
-        return multitask || (
-            handshakes_permitted() &&
-            st.todo.handshake.length == 0 &&
-            st.busy.handshake.length == 0);
+        if (multitask) {
+            var ahead = st.bytes.hashed - st.bytes.uploaded;
+            return ahead < 1024 * 1024 * 128;
+        }
+        return handshakes_permitted() && 0 ==
+            st.todo.handshake.length +
+            st.busy.handshake.length;
     }
 
     var tasker = (function () {
@@ -357,8 +481,54 @@ function up2k_init(have_crypto) {
 
             mutex = true;
             while (true) {
+                if (false) {
+                    ebi('srv_info').innerHTML =
+                        new Date().getTime() + ", " +
+                        st.todo.hash.length + ", " +
+                        st.todo.handshake.length + ", " +
+                        st.todo.upload.length + ", " +
+                        st.busy.hash.length + ", " +
+                        st.busy.handshake.length + ", " +
+                        st.busy.upload.length;
+                }
+
+                if (flag) {
+                    var need_flag = 0 !=
+                        st.todo.hash.length +
+                        st.todo.handshake.length +
+                        st.todo.upload.length +
+                        st.busy.hash.length +
+                        st.busy.handshake.length +
+                        st.busy.upload.length;
+
+                    if (need_flag) {
+                        var now = new Date().getTime();
+                        flag.take(now);
+                        if (!flag.ours) {
+                            setTimeout(taskerd, 100);
+                            mutex = false;
+                            return;
+                        }
+                    }
+                    else if (flag.ours) {
+                        flag.give();
+                    }
+                }
+
                 var mou_ikkai = false;
 
+                if (st.todo.handshake.length > 0 &&
+                    st.busy.handshake.length == 0 && (
+                        st.todo.handshake[0].t3 || (
+                            handshakes_permitted() &&
+                            st.busy.upload.length < parallel_uploads
+                        )
+                    )
+                ) {
+                    exec_handshake();
+                    mou_ikkai = true;
+                }
+
                 if (handshakes_permitted() &&
                     st.todo.handshake.length > 0 &&
                     st.busy.handshake.length == 0 &&
@@ -497,6 +667,8 @@ function up2k_init(have_crypto) {
 
         var t = st.todo.hash.shift();
         st.busy.hash.push(t);
+        st.bytes.hashed += t.size;
+        t.bytes_uploaded = 0;
         t.t1 = new Date().getTime();
 
         var nchunk = 0;
@@ -660,18 +832,25 @@ function up2k_init(have_crypto) {
                 st.busy.handshake.splice(st.busy.handshake.indexOf(t), 1);
 
                 if (done) {
+                    st.bytes.uploaded += t.size - t.bytes_uploaded;
                     var spd1 = (t.size / ((t.t2 - t.t1) / 1000.)) / (1024 * 1024.);
                     var spd2 = (t.size / ((t.t3 - t.t2) / 1000.)) / (1024 * 1024.);
                     ebi('f{0}p'.format(t.n)).innerHTML = 'hash {0}, up {1} MB/s'.format(
                         spd1.toFixed(2), spd2.toFixed(2));
                 }
+                else t.t3 = undefined;
+
                 tasker();
             }
             else {
                 var err = "";
                 var rsp = (xhr.responseText + '');
-                if (rsp.indexOf('partial upload exists') !== -1) {
-                    err = rsp.slice(5);
+                if (rsp.indexOf('partial upload exists') !== -1 ||
+                    rsp.indexOf('file already exists') !== -1) {
+                    err = rsp;
+                    var ofs = err.lastIndexOf(' : ');
+                    if (ofs > 0)
+                        err = err.slice(0, ofs);
                 }
                 if (err != "") {
                     ebi('f{0}t'.format(t.n)).innerHTML = "ERROR";
@@ -733,12 +912,14 @@ function up2k_init(have_crypto) {
             xhr.onload = function (xev) {
                 if (xhr.status == 200) {
                     prog(t.n, npart, col_uploaded);
+                    st.bytes.uploaded += cdr - car;
+                    t.bytes_uploaded += cdr - car;
                     st.busy.upload.splice(st.busy.upload.indexOf(upt), 1);
                     t.postlist.splice(t.postlist.indexOf(npart), 1);
                     if (t.postlist.length == 0) {
                         t.t3 = new Date().getTime();
                         ebi('f{0}t'.format(t.n)).innerHTML = 'verifying';
-                        st.todo.handshake.push(t);
+                        st.todo.handshake.unshift(t);
                     }
                     tasker();
                 }
@@ -821,6 +1002,33 @@ function up2k_init(have_crypto) {
         bcfg_set('multitask', multitask);
     }
 
+    function tgl_ask_up() {
+        ask_up = !ask_up;
+        bcfg_set('ask_up', ask_up);
+    }
+
+    function tgl_flag_en() {
+        flag_en = !flag_en;
+        bcfg_set('flag_en', flag_en);
+        apply_flag_cfg();
+    }
+
+    function apply_flag_cfg() {
+        if (flag_en && !flag) {
+            try {
+                flag = up2k_flagbus();
+            }
+            catch (ex) {
+                console.log("flag error: " + ex.toString());
+                tgl_flag_en();
+            }
+        }
+        else if (!flag_en && flag) {
+            flag.ch.close();
+            flag = false;
+        }
+    }
+
     function nop(ev) {
         ev.preventDefault();
         this.click();
@@ -837,6 +1045,8 @@ function up2k_init(have_crypto) {
 
     ebi('nthread').addEventListener('input', bumpthread, false);
     ebi('multitask').addEventListener('click', tgl_multitask, false);
+    ebi('ask_up').addEventListener('click', tgl_ask_up, false);
+    ebi('flag_en').addEventListener('click', tgl_flag_en, false);
 
     var nodes = ebi('u2conf').getElementsByTagName('a');
     for (var a = nodes.length - 1; a >= 0; a--)

copyparty/web/upload.css

@@ -194,6 +194,12 @@
 #u2conf input+a {
 	background: #d80;
 }
+#u2conf input[type="checkbox"]+label {
+	color: #f5a;
+}
+#u2conf input[type="checkbox"]:checked+label {
+	color: #fc5;
+}
 #u2foot {
 	color: #fff;
 	font-style: italic;

copyparty/web/upload.html

@@ -3,7 +3,8 @@
         href="#" data-dest="up2k">up2k</a><i></i><a
         href="#" data-dest="bup">bup</a><i></i><a
         href="#" data-dest="mkdir">mkdir</a><i></i><a
-        href="#" data-dest="new_md">new.md</a></div>
+        href="#" data-dest="new_md">new.md</a><i></i><a
+        href="#" data-dest="msg">msg</a></div>
 
     <div id="op_bup" class="opview opbox act">
         <div id="u2err"></div>
@@ -30,6 +31,13 @@
         </form>
     </div>
 
+    <div id="op_msg" class="opview opbox">
+        <form method="post" enctype="application/x-www-form-urlencoded" accept-charset="utf-8" action="/{{ vdir }}">
+            <input type="text" name="msg" size="30">
+            <input type="submit" value="send">
+        </form>
+    </div>
+
     <div id="op_up2k" class="opview">
         <form id="u2form" method="post" enctype="multipart/form-data" onsubmit="return false;"></form>
 
@@ -43,9 +51,17 @@
                         <input class="txtbox" id="nthread" value="2" />
                         <a href="#" id="nthread_add">+</a>
                     </td>
-                    <td rowspan="2">
+                    <td rowspan="2" style="padding-left:1.5em">
                         <input type="checkbox" id="multitask" />
-                        <label for="multitask">hash while<br />uploading</label>
+                        <label for="multitask">hash<br />while<br />upping</label>
+                    </td>
+                    <td rowspan="2">
+                        <input type="checkbox" id="ask_up" />
+                        <label for="ask_up">ask<br />before<br />start</label>
+                    </td>
+                    <td rowspan="2">
+                        <input type="checkbox" id="flag_en" />
+                        <label for="flag_en">only<br />one tab<br />at once</label>
                     </td>
                 </tr>
             </table>

docs/notes.sh

@@ -3,6 +3,14 @@ echo not a script
 exit 1
 
 
+##
+## delete all partial uploads
+##  (supports linux/macos, probably windows+msys2)
+
+gzip -d < .hist/up2k.snap | jq -r '.[].tnam' | while IFS= read -r f; do rm -f -- "$f"; done
+gzip -d < .hist/up2k.snap | jq -r '.[].name' | while IFS= read -r f; do wc -c -- "$f" | grep -qiE '^[^0-9a-z]*0' && rm -f -- "$f"; done
+
+
 ##
 ## create a test payload
 

scripts/make-pypi-release.sh

@@ -3,12 +3,15 @@ set -e
 echo
 
 # osx support
-command -v gtar  >/dev/null &&
-command -v gfind >/dev/null && {
-	tar()  { gtar  "$@"; }
+# port install gnutar findutils gsed coreutils
+gtar=$(command -v gtar || command -v gnutar) || true
+[ ! -z "$gtar" ] && command -v gfind >/dev/null && {
+	tar()  { $gtar "$@"; }
 	sed()  { gsed  "$@"; }
 	find() { gfind "$@"; }
 	sort() { gsort "$@"; }
+	command -v grealpath >/dev/null &&
+		realpath() { grealpath "$@"; }
 }
 
 which md5sum 2>/dev/null >/dev/null &&

scripts/make-sfx.sh

@@ -18,13 +18,16 @@ echo
 #   (the fancy markdown editor)
 
 
-command -v gtar  >/dev/null &&
-command -v gfind >/dev/null && {
-	tar()  { gtar  "$@"; }
+# port install gnutar findutils gsed coreutils
+gtar=$(command -v gtar || command -v gnutar) || true
+[ ! -z "$gtar" ] && command -v gfind >/dev/null && {
+	tar()  { $gtar "$@"; }
 	sed()  { gsed  "$@"; }
 	find() { gfind "$@"; }
 	sort() { gsort "$@"; }
 	unexpand() { gunexpand "$@"; }
+	command -v grealpath >/dev/null &&
+		realpath() { grealpath "$@"; }
 }
 
 [ -e copyparty/__main__.py ] || cd ..
@@ -59,28 +62,32 @@ cd sfx
 	)/pe-copyparty"
 
 	echo "repack of files in $old"
-	cp -pR "$old/"*{jinja2,copyparty} .
-	mv {x.,}jinja2 2>/dev/null || true
+	cp -pR "$old/"*{dep-j2,copyparty} .
 }
 
 [ $repack ] || {
 	echo collecting jinja2
-	f="../build/Jinja2-2.6.tar.gz"
+	f="../build/Jinja2-2.11.3.tar.gz"
 	[ -e "$f" ] ||
-		(url=https://files.pythonhosted.org/packages/25/c8/212b1c2fd6df9eaf536384b6c6619c4e70a3afd2dffdd00e5296ffbae940/Jinja2-2.6.tar.gz;
+		(url=https://files.pythonhosted.org/packages/4f/e7/65300e6b32e69768ded990494809106f87da1d436418d5f1367ed3966fd7/Jinja2-2.11.3.tar.gz;
 		wget -O$f "$url" || curl -L "$url" >$f)
 
 	tar -zxf $f
-	mv Jinja2-*/jinja2 .
-	rm -rf Jinja2-* jinja2/testsuite jinja2/_markupsafe/tests.py jinja2/_stringdefs.py
+	mv Jinja2-*/src/jinja2 .
+	rm -rf Jinja2-*
 	
-	f=jinja2/lexer.py
-	sed -r '/.*föö.*/    raise SyntaxError/' <$f >t
-	tmv $f
-	
-	f=jinja2/_markupsafe/_constants.py
-	awk '!/: [0-9]+,?$/ || /(amp|gt|lt|quot|apos|nbsp).:/' <$f >t
-	tmv $f
+	echo collecting markupsafe
+	f="../build/MarkupSafe-1.1.1.tar.gz"
+	[ -e "$f" ] ||
+		(url=https://files.pythonhosted.org/packages/b9/2e/64db92e53b86efccfaea71321f597fa2e1b2bd3853d8ce658568f7a13094/MarkupSafe-1.1.1.tar.gz;
+		wget -O$f "$url" || curl -L "$url" >$f)
+
+	tar -zxf $f
+	mv MarkupSafe-*/src/markupsafe .
+	rm -rf MarkupSafe-* markupsafe/_speedups.c
+
+	mkdir dep-j2/
+	mv {markupsafe,jinja2} dep-j2/
 
 	# msys2 tar is bad, make the best of it
 	echo collecting source
@@ -162,6 +169,15 @@ done
 	sed -r '/edit2">edit \(fancy/d' <$f >t && tmv "$f"
 }
 
+find | grep -E '\.py$' |
+  grep -vE '__version__' |
+  tr '\n' '\0' |
+  xargs -0 python ../scripts/uncomment.py
+
+f=dep-j2/jinja2/constants.py
+awk '/^LOREM_IPSUM_WORDS/{o=1;print "LOREM_IPSUM_WORDS = u\"a\"";next} !o; /"""/{o=0}' <$f >t
+tmv "$f"
+
 # up2k goes from 28k to 22k laff
 echo entabbening
 find | grep -E '\.(js|css|html|py)$' | while IFS= read -r f; do
@@ -174,7 +190,7 @@ args=(--owner=1000 --group=1000)
 [ "$OSTYPE" = msys ] &&
 	args=()
 
-tar -cf tar "${args[@]}" --numeric-owner copyparty jinja2
+tar -cf tar "${args[@]}" --numeric-owner copyparty dep-j2
 
 echo compressing tar
 # detect best level; bzip2 -7 is usually better than -9

scripts/make-tgz-release.sh

@@ -2,12 +2,16 @@
 set -e
 echo
 
-command -v gtar  >/dev/null &&
-command -v gfind >/dev/null && {
-	tar()  { gtar  "$@"; }
+# osx support
+# port install gnutar findutils gsed coreutils
+gtar=$(command -v gtar || command -v gnutar) || true
+[ ! -z "$gtar" ] && command -v gfind >/dev/null && {
+	tar()  { $gtar "$@"; }
 	sed()  { gsed  "$@"; }
 	find() { gfind "$@"; }
 	sort() { gsort "$@"; }
+	command -v grealpath >/dev/null &&
+		realpath() { grealpath "$@"; }
 }
 
 which md5sum 2>/dev/null >/dev/null &&

scripts/sfx.py

@@ -2,7 +2,7 @@
 # coding: utf-8
 from __future__ import print_function, unicode_literals
 
-import re, os, sys, time, shutil, signal, tarfile, hashlib, platform, tempfile
+import os, sys, time, shutil, signal, tarfile, hashlib, platform, tempfile
 import subprocess as sp
 
 """
@@ -202,93 +202,6 @@ def u8(gen):
             yield s
 
 
-def get_py_win(ret):
-    tops = []
-    p = str(os.getenv("LocalAppdata"))
-    if p:
-        tops.append(os.path.join(p, "Programs", "Python"))
-
-    progfiles = {}
-    for p in ["ProgramFiles", "ProgramFiles(x86)"]:
-        p = str(os.getenv(p))
-        if p:
-            progfiles[p] = 1
-            # 32bit apps get x86 for both
-            if p.endswith(" (x86)"):
-                progfiles[p[:-6]] = 1
-
-    tops += list(progfiles.keys())
-
-    for sysroot in [me, sys.executable]:
-        sysroot = sysroot[:3].upper()
-        if sysroot[1] == ":" and sysroot not in tops:
-            tops.append(sysroot)
-
-    # $WIRESHARK_SLOGAN
-    for top in tops:
-        try:
-            for name1 in u8(sorted(os.listdir(top), reverse=True)):
-                if name1.lower().startswith("python"):
-                    path1 = os.path.join(top, name1)
-                    try:
-                        for name2 in u8(os.listdir(path1)):
-                            if name2.lower() == "python.exe":
-                                path2 = os.path.join(path1, name2)
-                                ret[path2.lower()] = path2
-                    except:
-                        pass
-        except:
-            pass
-
-
-def get_py_nix(ret):
-    ptn = re.compile(r"^(python|pypy)[0-9\.-]*$")
-    for bindir in os.getenv("PATH").split(":"):
-        if not bindir:
-            next
-
-        try:
-            for fn in u8(os.listdir(bindir)):
-                if ptn.match(fn):
-                    fn = os.path.join(bindir, fn)
-                    ret[fn.lower()] = fn
-        except:
-            pass
-
-
-def read_py(binp):
-    cmd = [
-        binp,
-        "-c",
-        "import sys; sys.stdout.write(' '.join(str(x) for x in sys.version_info)); import jinja2",
-    ]
-    p = sp.Popen(cmd, stdout=sp.PIPE, stderr=sp.PIPE)
-    ver, _ = p.communicate()
-    ver = ver.decode("utf-8").split(" ")[:3]
-    ver = [int(x) if x.isdigit() else 0 for x in ver]
-    return ver, p.returncode == 0
-
-
-def get_pys():
-    ver, chk = read_py(sys.executable)
-    if chk or PY2:
-        return [[chk, ver, sys.executable]]
-
-    hits = {sys.executable.lower(): sys.executable}
-    if platform.system() == "Windows":
-        get_py_win(hits)
-    else:
-        get_py_nix(hits)
-
-    ret = []
-    for binp in hits.values():
-        ver, chk = read_py(binp)
-        ret.append([chk, ver, binp])
-        msg("\t".join(str(x) for x in ret[-1]))
-
-    return ret
-
-
 def yieldfile(fn):
     with open(fn, "rb") as f:
         for block in iter(lambda: f.read(64 * 1024), b""):
@@ -440,12 +353,11 @@ def confirm():
         pass
 
 
-def run(tmp, py):
+def run(tmp, j2ver):
     global cpp
 
-    msg("OK")
-    msg("will use:", py)
-    msg("bound to:", tmp)
+    msg("jinja2:", j2ver or "bundled")
+    msg("sfxdir:", tmp)
 
     # "systemd-tmpfiles-clean.timer"?? HOW do you even come up with this shit
     try:
@@ -457,24 +369,20 @@ def run(tmp, py):
     except:
         pass
 
-    fp_py = os.path.join(tmp, "py")
-    try:
-        with open(fp_py, "wb") as f:
-            f.write(py.encode("utf-8") + b"\n")
-    except:
-        pass
+    ld = [tmp, os.path.join(tmp, "dep-j2")]
+    if j2ver:
+        del ld[-1]
 
-    # avoid loading ./copyparty.py
-    cmd = [
-        py,
-        "-c",
-        'import sys, runpy; sys.path.insert(0, r"'
-        + tmp
-        + '"); runpy.run_module("copyparty", run_name="__main__")',
-    ] + list(sys.argv[1:])
+    cmd = (
+        "import sys, runpy; "
+        + "".join(['sys.path.insert(0, r"' + x + '"); ' for x in ld])
+        + 'runpy.run_module("copyparty", run_name="__main__")'
+    )
+    cmd = [sys.executable, "-c", cmd] + list(sys.argv[1:])
 
+    cmd = [str(x) for x in cmd]
     msg("\n", cmd, "\n")
-    cpp = sp.Popen(str(x) for x in cmd)
+    cpp = sp.Popen(cmd)
     try:
         cpp.wait()
     except:
@@ -494,7 +402,6 @@ def bye(sig, frame):
 def main():
     sysver = str(sys.version).replace("\n", "\n" + " " * 18)
     pktime = time.strftime("%Y-%m-%d, %H:%M:%S", time.gmtime(STAMP))
-    os.system("")
     msg()
     msg("   this is: copyparty", VER)
     msg(" packed at:", pktime, "UTC,", STAMP)
@@ -526,33 +433,13 @@ def main():
     signal.signal(signal.SIGTERM, bye)
 
     tmp = unpack()
-    fp_py = os.path.join(tmp, "py")
-    if os.path.exists(fp_py):
-        with open(fp_py, "rb") as f:
-            py = f.read().decode("utf-8").rstrip()
 
-        return run(tmp, py)
+    try:
+        from jinja2 import __version__ as j2ver
+    except:
+        j2ver = None
 
-    pys = get_pys()
-    pys.sort(reverse=True)
-    j2, ver, py = pys[0]
-    if j2:
-        try:
-            os.rename(os.path.join(tmp, "jinja2"), os.path.join(tmp, "x.jinja2"))
-        except:
-            pass
-
-        return run(tmp, py)
-
-    msg("\n  could not find jinja2; will use py2 + the bundled version\n")
-    for _, ver, py in pys:
-        if ver > [2, 7] and ver < [3, 0]:
-            return run(tmp, py)
-
-    m = "\033[1;31m\n\n\ncould not find a python with jinja2 installed; please do one of these:\n\n  pip install --user jinja2\n\n  install python2\n\n\033[0m"
-    msg(m)
-    confirm()
-    sys.exit(1)
+    return run(tmp, j2ver)
 
 
 if __name__ == "__main__":

scripts/uncomment.py

@@ -0,0 +1,77 @@
+#!/usr/bin/env python
+# coding: utf-8
+from __future__ import print_function, unicode_literals
+
+import io
+import sys
+import tokenize
+
+
+def uncomment(fpath):
+    """ modified https://stackoverflow.com/a/62074206 """
+
+    with open(fpath, "rb") as f:
+        orig = f.read().decode("utf-8")
+
+    out = ""
+    for ln in orig.split("\n"):
+        if not ln.startswith("#"):
+            break
+
+        out += ln + "\n"
+
+    io_obj = io.StringIO(orig)
+    prev_toktype = tokenize.INDENT
+    last_lineno = -1
+    last_col = 0
+    for tok in tokenize.generate_tokens(io_obj.readline):
+        # print(repr(tok))
+        token_type = tok[0]
+        token_string = tok[1]
+        start_line, start_col = tok[2]
+        end_line, end_col = tok[3]
+
+        if start_line > last_lineno:
+            last_col = 0
+
+        if start_col > last_col:
+            out += " " * (start_col - last_col)
+
+        is_legalese = (
+            "copyright" in token_string.lower() or "license" in token_string.lower()
+        )
+
+        if token_type == tokenize.STRING:
+            if (
+                prev_toktype != tokenize.INDENT
+                and prev_toktype != tokenize.NEWLINE
+                and start_col > 0
+                or is_legalese
+            ):
+                out += token_string
+            else:
+                out += '"a"'
+        elif token_type != tokenize.COMMENT or is_legalese:
+            out += token_string
+
+        prev_toktype = token_type
+        last_lineno = end_line
+        last_col = end_col
+
+    # out = "\n".join(x for x in out.splitlines() if x.strip())
+
+    with open(fpath, "wb") as f:
+        f.write(out.encode("utf-8"))
+
+
+def main():
+    print("uncommenting", end="")
+    for f in sys.argv[1:]:
+        print(".", end="")
+        uncomment(f)
+
+    print("k")
+
+
+if __name__ == "__main__":
+    main()

setup.py

@@ -2,10 +2,8 @@
 # coding: utf-8
 from __future__ import print_function
 
-import io
 import os
 import sys
-from glob import glob
 from shutil import rmtree
 
 setuptools_available = True
@@ -49,7 +47,7 @@ with open(here + "/README.md", "rb") as f:
 about = {}
 if not VERSION:
     with open(os.path.join(here, NAME, "__version__.py"), "rb") as f:
-        exec(f.read().decode("utf-8").split("\n\n", 1)[1], about)
+        exec (f.read().decode("utf-8").split("\n\n", 1)[1], about)
 else:
     about["__version__"] = VERSION
 
@@ -110,13 +108,13 @@ args = {
         "Programming Language :: Python :: 2",
         "Programming Language :: Python :: 2.7",
         "Programming Language :: Python :: 3",
-        "Programming Language :: Python :: 3.2",
         "Programming Language :: Python :: 3.3",
         "Programming Language :: Python :: 3.4",
         "Programming Language :: Python :: 3.5",
         "Programming Language :: Python :: 3.6",
         "Programming Language :: Python :: 3.7",
         "Programming Language :: Python :: 3.8",
+        "Programming Language :: Python :: 3.9",
         "Programming Language :: Python :: Implementation :: CPython",
         "Programming Language :: Python :: Implementation :: PyPy",
         "Environment :: Console",