v0.11.7

README.md

@@ -520,20 +520,25 @@ in the `scripts` folder:
 
 roughly sorted by priority
 
-* separate sqlite table per tag
-* audio fingerprinting
 * readme.md as epilogue
+* single sha512 across all up2k chunks? maybe
 * reduce up2k roundtrips
   * start from a chunk index and just go
   * terminate client on bad data
-* `os.copy_file_range` for up2k cloning
-* single sha512 across all up2k chunks? maybe
-* figure out the deal with pixel3a not being connectable as hotspot
-  * pixel3a having unpredictable 3sec latency in general :||||
 
 discarded ideas
 
+* separate sqlite table per tag
+  * performance fixed by skipping some indexes (`+mt.k`)
+* audio fingerprinting
+  * only makes sense if there can be a wasm client and that doesn't exist yet (except for olaf which is agpl hence counts as not existing)
+* `os.copy_file_range` for up2k cloning
+  * almost never hit this path anyways
 * up2k partials ui
+  * feels like there isn't much point
 * cache sha512 chunks on client
+  * too dangerous
 * comment field
+  * nah
 * look into android thumbnail cache file format
+  * absolutely not

copyparty/__main__.py

@@ -225,6 +225,19 @@ def run_argparse(argv, formatter):
             --ciphers help = available ssl/tls ciphers,
             --ssl-ver help = available ssl/tls versions,
               default is what python considers safe, usually >= TLS1
+            
+            values for --ls:
+              "USR" is a user to browse as; * is anonymous, ** is all users
+              "VOL" is a single volume to scan, default is * (all vols)
+              "FLAG" is flags;
+                "v" in addition to realpaths, print usernames and vpaths
+                "ln" only prints symlinks leaving the volume mountpoint
+                "p" exits 1 if any such symlinks are found
+                "r" resumes startup after the listing
+            examples:
+              --ls '**'          # list all files which are possible to read
+              --ls '**,*,ln'     # check for dangerous symlinks
+              --ls '**,*,ln,p,r' # check, then start normally if safe
             """
         ),
     )
@@ -288,6 +301,7 @@ def run_argparse(argv, formatter):
     ap2.add_argument("--ssl-log", metavar="PATH", help="log master secrets")
 
     ap2 = ap.add_argument_group('debug options')
+    ap2.add_argument("--ls", metavar="U[,V[,F]]", help="scan all volumes")
     ap2.add_argument("--log-conn", action="store_true", help="print tcp-server msgs")
     ap2.add_argument("--no-sendfile", action="store_true", help="disable sendfile")
     ap2.add_argument("--no-scandir", action="store_true", help="disable scandir")

copyparty/__version__.py

@@ -1,8 +1,8 @@
 # coding: utf-8
 
-VERSION = (0, 11, 4)
+VERSION = (0, 11, 7)
 CODENAME = "the grid"
-BUILD_DT = (2021, 6, 1)
+BUILD_DT = (2021, 6, 5)
 
 S_VERSION = ".".join(map(str, VERSION))
 S_BUILD_DT = "{0:04d}-{1:02d}-{2:02d}".format(*BUILD_DT)

copyparty/authsrv.py

@@ -228,21 +228,19 @@ class VFS(object):
             for f in [{"vp": v, "ap": a, "st": n[1]} for v, a, n in files]:
                 yield f
 
-    def user_tree(self, uname, readable=False, writable=False, admin=False):
+    def user_tree(self, uname, readable, writable, admin):
-        ret = []
+        is_readable = False
-        opt1 = readable and (uname in self.uread or "*" in self.uread)
+        if uname in self.uread or "*" in self.uread:
-        opt2 = writable and (uname in self.uwrite or "*" in self.uwrite)
+            readable.append(self.vpath)
-        if admin:
+            is_readable = True
-            if opt1 and opt2:
+
-                ret.append(self.vpath)
+        if uname in self.uwrite or "*" in self.uwrite:
-        else:
+            writable.append(self.vpath)
-            if opt1 or opt2:
+            if is_readable:
-                ret.append(self.vpath)
+                admin.append(self.vpath)
 
         for _, vn in sorted(self.nodes.items()):
-            ret.extend(vn.user_tree(uname, readable, writable, admin))
+            vn.user_tree(uname, readable, writable, admin)
-
-        return ret
 
 
 class AuthSrv(object):
@@ -559,3 +557,90 @@ class AuthSrv(object):
 
         # import pprint
         # pprint.pprint({"usr": user, "rd": mread, "wr": mwrite, "mnt": mount})
+
+    def dbg_ls(self):
+        users = self.args.ls
+        vols = "*"
+        flags = []
+
+        try:
+            users, vols = users.split(",", 1)
+        except:
+            pass
+
+        try:
+            vols, flags = vols.split(",", 1)
+            flags = flags.split(",")
+        except:
+            pass
+
+        if users == "**":
+            users = list(self.user.keys()) + ["*"]
+        else:
+            users = [users]
+
+        for u in users:
+            if u not in self.user and u != "*":
+                raise Exception("user not found: " + u)
+
+        if vols == "*":
+            vols = ["/" + x for x in self.vfs.all_vols.keys()]
+        else:
+            vols = [vols]
+
+        for v in vols:
+            if not v.startswith("/"):
+                raise Exception("volumes must start with /")
+
+            if v[1:] not in self.vfs.all_vols:
+                raise Exception("volume not found: " + v)
+
+        self.log({"users": users, "vols": vols, "flags": flags})
+        for k, v in self.vfs.all_vols.items():
+            self.log("/{}: read({}) write({})".format(k, v.uread, v.uwrite))
+
+        flag_v = "v" in flags
+        flag_ln = "ln" in flags
+        flag_p = "p" in flags
+        flag_r = "r" in flags
+
+        n_bads = 0
+        for v in vols:
+            v = v[1:]
+            vtop = "/{}/".format(v) if v else "/"
+            for u in users:
+                self.log("checking /{} as {}".format(v, u))
+                try:
+                    vn, _ = self.vfs.get(v, u, True, False)
+                except:
+                    continue
+
+                atop = vn.realpath
+                g = vn.walk("", "", u, True, not self.args.no_scandir, lstat=False)
+                for vpath, apath, files, _, _ in g:
+                    fnames = [n[0] for n in files]
+                    vpaths = [vpath + "/" + n for n in fnames] if vpath else fnames
+                    vpaths = [vtop + x for x in vpaths]
+                    apaths = [os.path.join(apath, n) for n in fnames]
+                    files = list(zip(vpaths, apaths))
+
+                    if flag_ln:
+                        files = [x for x in files if not x[1].startswith(atop + os.sep)]
+                        n_bads += len(files)
+
+                    if flag_v:
+                        msg = [
+                            '# user "{}", vpath "{}"\n{}'.format(u, vp, ap)
+                            for vp, ap in files
+                        ]
+                    else:
+                        msg = [x[1] for x in files]
+
+                    if msg:
+                        nuprint("\n".join(msg))
+
+                if n_bads and flag_p:
+                    raise Exception("found symlink leaving volume, and strict is set")
+
+        if not flag_r:
+            sys.exit(0)

copyparty/httpcli.py

@@ -153,10 +153,8 @@ class HttpCli(object):
 
         pwd = uparam.get("pw")
         self.uname = self.auth.iuser.get(pwd, "*")
-        if self.uname:
+        self.rvol, self.wvol, self.avol = [[], [], []]
-            self.rvol = self.auth.vfs.user_tree(self.uname, readable=True)
+        self.auth.vfs.user_tree(self.uname, self.rvol, self.wvol, self.avol)
-            self.wvol = self.auth.vfs.user_tree(self.uname, writable=True)
-            self.avol = self.auth.vfs.user_tree(self.uname, True, True, True)
 
         ua = self.headers.get("user-agent", "")
         self.is_rclone = ua.startswith("rclone/")

copyparty/svchub.py

@@ -39,6 +39,8 @@ class SvcHub(object):
 
         # jank goes here
         auth = AuthSrv(self.args, self.log, False)
+        if args.ls:
+            auth.dbg_ls()
 
         # initiate all services to manage
         self.tcpsrv = TcpSrv(self)

copyparty/u2idx.py

@@ -126,10 +126,11 @@ class U2idx(object):
                     joins += "inner join mt mt{} on {} = {} ".format(
                         mt_ctr, mt_keycmp, mt_keycmp2
                     )
+                    mt_keycmp = mt_keycmp2
                     if v == "tags":
                         v = "mt{0}.v".format(mt_ctr)
                     else:
-                        v = "mt{0}.k = '{1}' and mt{0}.v".format(mt_ctr, v)
+                        v = "+mt{0}.k = '{1}' and mt{0}.v".format(mt_ctr, v)
 
                 else:
                     raise Pebkac(400, "invalid key [" + v + "]")
@@ -195,7 +196,7 @@ class U2idx(object):
         thr.daemon = True
         thr.start()
 
-        if not uq:
+        if not uq or not uv:
             q = "select * from up"
             v = ()
         else:

copyparty/up2k.py

@@ -552,6 +552,7 @@ class Up2k(object):
                     last_write = time.time()
                     n_buf = 0
 
+            if mpool:
                 self._stop_mpool(mpool)
                 with self.mutex:
                     n_add += len(self._flush_mpool(c3))

copyparty/web/browser.js

@@ -803,8 +803,11 @@ var thegrid = (function () {
 			r.sz = v;
 			swrite('gridsz', r.sz);
 		}
+		try {
 			document.documentElement.style.setProperty('--grid-sz', r.sz + 'em');
 		}
+		catch (ex) { }
+	}
 	setsz();
 
 	function seltgl(e) {
@@ -820,10 +823,18 @@ var thegrid = (function () {
 		this.setAttribute('class', tr.getAttribute('class'));
 	}
 
+	function bgopen(e) {
+		ev(e);
+		var url = this.getAttribute('href');
+		window.open(url, '_blank');
+	}
+
 	r.loadsel = function () {
-		var ths = QSA('#ggrid>a');
+		var ths = QSA('#ggrid>a'),
+			have_sel = !!QS('#files tr.sel');
+
 		for (var a = 0, aa = ths.length; a < aa; a++) {
-			ths[a].onclick = r.sel ? seltgl : null;
+			ths[a].onclick = r.sel ? seltgl : have_sel ? bgopen : null;
 			ths[a].setAttribute('class', ebi(ths[a].getAttribute('ref')).parentNode.parentNode.getAttribute('class'));
 		}
 		var uns = QS('#ggrid a[ref="unsearch"]');
@@ -965,7 +976,7 @@ document.onkeydown = function (e) {
 	if (k == 'KeyT')
 		return ebi('thumbs').click();
 
-	if (window['thegrid'] && thegrid.en) {
+	if (thegrid.en) {
 		if (k == 'KeyS')
 			return ebi('gridsel').click();
 
@@ -1437,7 +1448,7 @@ var treectl = (function () {
 		if (hpush)
 			get_tree('.', xhr.top);
 
-		enspin('#files');
+		enspin(thegrid.en ? '#gfiles' : '#files');
 	}
 
 	function treegrow(e) {
@@ -1517,6 +1528,7 @@ var treectl = (function () {
 
 		apply_perms(res.perms);
 		despin('#files');
+		despin('#gfiles');
 
 		ebi('pro').innerHTML = res.logues ? res.logues[0] || "" : "";
 		ebi('epi').innerHTML = res.logues ? res.logues[1] || "" : "";
@@ -2051,7 +2063,6 @@ var msel = (function () {
 	}
 	function selui() {
 		clmod(ebi('wtoggle'), 'sel', getsel().length);
-		if (window['thegrid'])
 		thegrid.loadsel();
 	}
 	function seltgl(e) {

copyparty/web/up2k.js

@@ -17,6 +17,7 @@ function goto_up2k() {
 // chrome requires https to use crypto.subtle,
 // usually it's undefined but some chromes throw on invoke
 var up2k = null;
+var sha_js = window.WebAssembly ? 'hw' : 'ac';  // ff53,c57,sa11
 try {
     var cf = crypto.subtle || crypto.webkitSubtle;
     cf.digest('SHA-512', new Uint8Array(1)).then(
@@ -430,13 +431,15 @@ function up2k_init(subtle) {
     // upload ui hidden by default, clicking the header shows it
     function init_deps() {
         if (!subtle && !window.asmCrypto) {
-            showmodal('<h1>loading sha512.js</h1><h2>since ' + shame + '</h2><h4>thanks chrome</h4>');
+            var fn = 'sha512.' + sha_js + '.js';
-            import_js('/.cpr/deps/sha512.js', unmodal);
+            showmodal('<h1>loading ' + fn + '</h1><h2>since ' + shame + '</h2><h4>thanks chrome</h4>');
+            import_js('/.cpr/deps/' + fn, unmodal);
 
             if (is_https)
                 ebi('u2foot').innerHTML = shame + ' so <em>this</em> uploader will do like 500kB/s at best';
             else
-                ebi('u2foot').innerHTML = 'seems like ' + shame + ' so do that if you want more performance';
+                ebi('u2foot').innerHTML = 'seems like ' + shame + ' so do that if you want more performance <span style="color:#' +
+                    (sha_js == 'ac' ? 'c84">(expecting 20' : '8a5">(but dont worry too much, expect 100') + ' MiB/s)</span>';
         }
     }
 
@@ -886,6 +889,10 @@ function up2k_init(subtle) {
         return base64;
     }
 
+    function hex2u8(txt) {
+        return new Uint8Array(txt.match(/.{2}/g).map(function (b) { return parseInt(b, 16); }));
+    }
+
     function get_chunksize(filesize) {
         var chunksize = 1024 * 1024,
             stepsize = 512 * 1024;
@@ -987,10 +994,18 @@ function up2k_init(subtle) {
             if (subtle)
                 subtle.digest('SHA-512', buf).then(hash_done);
             else setTimeout(function () {
+                var u8buf = new Uint8Array(buf);
+                if (sha_js == 'hw') {
+                    hashwasm.sha512(u8buf).then(function (v) {
+                        hash_done(hex2u8(v))
+                    });
+                }
+                else {
                     var hasher = new asmCrypto.Sha512();
-                hasher.process(new Uint8Array(buf));
+                    hasher.process(u8buf);
                     hasher.finish();
                     hash_done(hasher.result);
+                }
             }, 1);
         };
 

copyparty/web/upload.css

@@ -238,6 +238,10 @@
 	color: #fff;
 	font-style: italic;
 }
+#u2foot span {
+	color: #999;
+	font-size: .9em;
+}
 #u2footfoot {
 	margin-bottom: -1em;
 }

scripts/deps-docker/Dockerfile

@@ -9,6 +9,12 @@ ENV     ver_asmcrypto=5b994303a9d3e27e0915f72a10b6c2c51535a4dc \
         ver_zopfli=1.0.3
 
 
+# TODO
+# sha512.hw.js https://github.com/Daninet/hash-wasm
+# sha512.kc.js https://github.com/chm-diederichs/sha3-wasm
+# awk '/HMAC state/{o=1}  /var HEAP/{o=0}  /function hmac_reset/{o=1}  /return \{/{o=0}  /var __extends =/{o=1}  /var Hash =/{o=0}  /hmac_|pbkdf2_/{next}  o{next}  {gsub(/IllegalStateError/,"Exception")}  {sub(/^ +/,"");sub(/^\/\/ .*/,"");sub(/;$/," ;")}  1' <sha512.ac.js.orig >sha512.ac.js; for fn in sha512.ac.js.orig sha512.ac.js; do wc -c <$fn; wc -c <$fn.gz ; for n in {1..9}; do printf '%8d %d bz\n' $(bzip2 -c$n <$fn | wc -c) $n; done; done
+
+
 # download;
 # the scp url is latin from https://fonts.googleapis.com/css2?family=Source+Code+Pro&display=swap
 RUN     mkdir -p /z/dist/no-pk \

scripts/install-githooks.sh

@@ -0,0 +1,12 @@
+#!/bin/bash
+set -ex
+
+[ -e setup.py ] || ..
+[ -e setup.py ] || {
+    echo u wot
+    exit 1
+}
+
+cd .git/hooks
+rm -f pre-commit
+ln -s ../../scripts/run-tests.sh pre-commit

scripts/profile.py

@@ -0,0 +1,34 @@
+#!/usr/bin/env python3
+
+import sys
+
+sys.path.insert(0, ".")
+cmd = sys.argv[1]
+
+if cmd == "cpp":
+    from copyparty.__main__ import main
+
+    argv = ["__main__", "-v", "srv::r", "-v", "../../yt:yt:r"]
+    main(argv=argv)
+
+elif cmd == "test":
+    from unittest import main
+
+    argv = ["__main__", "discover", "-s", "tests"]
+    main(module=None, argv=argv)
+
+else:
+    raise Exception()
+
+# import dis; print(dis.dis(main))
+
+
+# macos:
+#   option1) python3.9 -m pip install --user -U vmprof==0.4.9
+#   option2) python3.9 -m pip install --user -U https://github.com/vmprof/vmprof-python/archive/refs/heads/master.zip
+#
+# python -m vmprof -o prof --lines ./scripts/profile.py test
+
+# linux: ~/.local/bin/vmprofshow prof tree | grep -vF '[1m  0.'
+# macos: ~/Library/Python/3.9/bin/vmprofshow prof tree | grep -vF '[1m  0.'
+#   win: %appdata%\..\Roaming\Python\Python39\Scripts\vmprofshow.exe prof tree

scripts/run-tests.sh

@@ -0,0 +1,12 @@
+#!/bin/bash
+set -ex
+
+pids=()
+for py in python{2,3}; do
+    $py -m unittest discover -s tests >/dev/null &
+    pids+=($!)
+done
+
+for pid in ${pids[@]}; do
+    wait $pid
+done

tests/test_httpcli.py

@@ -8,13 +8,13 @@ import time
 import shutil
 import pprint
 import tarfile
+import tempfile
 import unittest
-
 from argparse import Namespace
-from copyparty.authsrv import AuthSrv
-from copyparty.httpcli import HttpCli
 
 from tests import util as tu
+from copyparty.authsrv import AuthSrv
+from copyparty.httpcli import HttpCli
 
 
 def hdr(query):
@@ -42,13 +42,15 @@ class Cfg(Namespace):
 
 
 class TestHttpCli(unittest.TestCase):
-    def test(self):
+    def setUp(self):
-        td = os.path.join(tu.get_ramdisk(), "vfs")
+        self.td = tu.get_ramdisk()
-        try:
-            shutil.rmtree(td)
-        except OSError:
-            pass
 
+    def tearDown(self):
+        os.chdir(tempfile.gettempdir())
+        shutil.rmtree(self.td)
+
+    def test(self):
+        td = os.path.join(self.td, "vfs")
         os.mkdir(td)
         os.chdir(td)
 

tests/test_vfs.py

@@ -7,13 +7,12 @@ import json
 import shutil
 import tempfile
 import unittest
-
 from textwrap import dedent
 from argparse import Namespace
-from copyparty.authsrv import AuthSrv
-from copyparty import util
 
 from tests import util as tu
+from copyparty.authsrv import AuthSrv
+from copyparty import util
 
 
 class Cfg(Namespace):
@@ -25,6 +24,13 @@ class Cfg(Namespace):
 
 
 class TestVFS(unittest.TestCase):
+    def setUp(self):
+        self.td = tu.get_ramdisk()
+
+    def tearDown(self):
+        os.chdir(tempfile.gettempdir())
+        shutil.rmtree(self.td)
+
     def dump(self, vfs):
         print(json.dumps(vfs, indent=4, sort_keys=True, default=lambda o: o.__dict__))
 
@@ -55,12 +61,7 @@ class TestVFS(unittest.TestCase):
         pass
 
     def test(self):
-        td = os.path.join(tu.get_ramdisk(), "vfs")
+        td = os.path.join(self.td, "vfs")
-        try:
-            shutil.rmtree(td)
-        except OSError:
-            pass
-
         os.mkdir(td)
         os.chdir(td)
 
@@ -227,7 +228,7 @@ class TestVFS(unittest.TestCase):
         self.assertEqual(list(v1), list(v2))
 
         # config file parser
-        cfg_path = os.path.join(tu.get_ramdisk(), "test.cfg")
+        cfg_path = os.path.join(self.td, "test.cfg")
         with open(cfg_path, "wb") as f:
             f.write(
                 dedent(
@@ -260,6 +261,4 @@ class TestVFS(unittest.TestCase):
         self.assertEqual(n.uwrite, ["asd"])
         self.assertEqual(len(n.nodes), 0)
 
-        os.chdir(tempfile.gettempdir())
-        shutil.rmtree(td)
         os.unlink(cfg_path)

tests/util.py

@@ -1,16 +1,36 @@
 import os
+import sys
 import time
+import shutil
 import jinja2
 import tempfile
+import platform
 import subprocess as sp
 
-from copyparty.util import Unrecv
 
+WINDOWS = platform.system() == "Windows"
+ANYWIN = WINDOWS or sys.platform in ["msys"]
+MACOS = platform.system() == "Darwin"
 
 J2_ENV = jinja2.Environment(loader=jinja2.BaseLoader)
 J2_FILES = J2_ENV.from_string("{{ files|join('\n') }}")
 
 
+def nah(*a, **ka):
+    return False
+
+
+if MACOS:
+    import posixpath
+
+    posixpath.islink = nah
+    os.path.islink = nah
+    # 25% faster; until any tests do symlink stuff
+
+
+from copyparty.util import Unrecv
+
+
 def runcmd(*argv):
     p = sp.Popen(argv, stdout=sp.PIPE, stderr=sp.PIPE)
     stdout, stderr = p.communicate()
@@ -28,18 +48,25 @@ def chkcmd(*argv):
 
 
 def get_ramdisk():
+    def subdir(top):
+        ret = os.path.join(top, "cptd-{}".format(os.getpid()))
+        shutil.rmtree(ret, True)
+        os.mkdir(ret)
+        return ret
+
     for vol in ["/dev/shm", "/Volumes/cptd"]:  # nosec (singleton test)
         if os.path.exists(vol):
-            return vol
+            return subdir(vol)
 
     if os.path.exists("/Volumes"):
-        devname, _ = chkcmd("hdiutil", "attach", "-nomount", "ram://32768")
+        # hdiutil eject /Volumes/cptd/
+        devname, _ = chkcmd("hdiutil", "attach", "-nomount", "ram://65536")
         devname = devname.strip()
         print("devname: [{}]".format(devname))
         for _ in range(10):
             try:
                 _, _ = chkcmd("diskutil", "eraseVolume", "HFS+", "cptd", devname)
-                return "/Volumes/cptd"
+                return subdir("/Volumes/cptd")
             except Exception as ex:
                 print(repr(ex))
                 time.sleep(0.25)
@@ -50,7 +77,7 @@ def get_ramdisk():
     try:
         os.mkdir(ret)
     finally:
-        return ret
+        return subdir(ret)
 
 
 class NullBroker(object):