404/403 can be handled with plugins

README.md

@@ -66,6 +66,7 @@ turn almost any device into a file server with resumable uploads/downloads using
     * [file parser plugins](#file-parser-plugins) - provide custom parsers to index additional tags
     * [event hooks](#event-hooks) - trigger a program on uploads, renames etc ([examples](./bin/hooks/))
         * [upload events](#upload-events) - the older, more powerful approach ([examples](./bin/mtag/))
+    * [handlers](#handlers) - redefine behavior with plugins ([examples](./bin/handlers/))
     * [hiding from google](#hiding-from-google) - tell search engines you dont wanna be indexed
     * [themes](#themes)
     * [complete examples](#complete-examples)
@@ -1127,6 +1128,13 @@ note that this is way more complicated than the new [event hooks](#event-hooks)
 note that it will occupy the parsing threads, so fork anything expensive (or set `kn` to have copyparty fork it for you) -- otoh if you want to intentionally queue/singlethread you can combine it with `--mtag-mt 1`
 
 
+## handlers
+
+redefine behavior with plugins ([examples](./bin/handlers/))
+
+replace 404 and 403 errors with something completely different (that's it for now)
+
+
 ## hiding from google
 
 tell search engines you dont wanna be indexed,  either using the good old [robots.txt](https://www.robotstxt.org/robotstxt.html) or through copyparty settings:

bin/handlers/README.md

@@ -0,0 +1,35 @@
+replace the standard 404 / 403 responses with plugins
+
+
+# usage
+
+load plugins either globally with `--on404 ~/dev/copyparty/bin/handlers/sorry.py` or for a specific volume with `:c,on404=~/handlers/sorry.py`
+
+
+# api
+
+each plugin must define a `main()` which takes 3 arguments;
+
+* `cli` is an instance of [copyparty/httpcli.py](https://github.com/9001/copyparty/blob/hovudstraum/copyparty/httpcli.py) (the monstrosity itself)
+* `vn` is the VFS which overlaps with the requested URL, and
+* `rem` is the URL remainder below the VFS mountpoint
+    * so `vn.vpath + rem` == `cli.vpath` == original request
+
+
+# examples
+
+## on404
+
+* [sorry.py](answer.py) replies with a custom message instead of the usual 404
+* [nooo.py](nooo.py) replies with an endless noooooooooooooo
+* [never404.py](never404.py) 100% guarantee that 404 will never be a thing again as it automatically creates dummy files whenever necessary
+* [caching-proxy.py](caching-proxy.py) transforms copyparty into a squid/varnish knockoff
+
+## on403
+
+* [ip-ok.py](ip-ok.py) disables security checks if client-ip is 1.2.3.4
+
+
+# notes
+
+* on403 only works for trivial stuff (basic http access) since I haven't been able to think of any good usecases for it (was just easy to add while doing on404)

bin/handlers/caching-proxy.py

@@ -0,0 +1,36 @@
+# assume each requested file exists on another webserver and
+# download + mirror them as they're requested
+# (basically pretend we're warnish)
+
+import os
+import requests
+
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from copyparty.httpcli import HttpCli
+
+
+def main(cli: "HttpCli", vn, rem):
+    url = "https://mirrors.edge.kernel.org/alpine/" + rem
+    abspath = os.path.join(vn.realpath, rem)
+
+    # sneaky trick to preserve a requests-session between downloads
+    # so it doesn't have to spend ages reopening https connections;
+    # luckily we can stash it inside the copyparty client session,
+    # name just has to be definitely unused so "hacapo_req_s" it is
+    req_s = getattr(cli.conn, "hacapo_req_s", None) or requests.Session()
+    setattr(cli.conn, "hacapo_req_s", req_s)
+
+    try:
+        os.makedirs(os.path.dirname(abspath), exist_ok=True)
+        with req_s.get(url, stream=True, timeout=69) as r:
+            r.raise_for_status()
+            with open(abspath, "wb", 64 * 1024) as f:
+                for buf in r.iter_content(chunk_size=64 * 1024):
+                    f.write(buf)
+    except:
+        os.unlink(abspath)
+        return "false"
+
+    return "retry"

bin/handlers/ip-ok.py

@@ -0,0 +1,6 @@
+# disable permission checks and allow access if client-ip is 1.2.3.4
+
+
+def main(cli, vn, rem):
+    if cli.ip == "1.2.3.4":
+        return "allow"

bin/handlers/never404.py

@@ -0,0 +1,11 @@
+# create a dummy file and let copyparty return it
+
+
+def main(cli, vn, rem):
+    print("hello", cli.ip)
+
+    abspath = vn.canonical(rem)
+    with open(abspath, "wb") as f:
+        f.write(b"404? not on MY watch!")
+
+    return "retry"

bin/handlers/nooo.py

@@ -0,0 +1,16 @@
+# reply with an endless "noooooooooooooooooooooooo"
+
+
+def say_no():
+    yield b"n"
+    while True:
+        yield b"o" * 4096
+
+
+def main(cli, vn, rem):
+    cli.send_headers(None, 404, "text/plain")
+
+    for chunk in say_no():
+        cli.s.sendall(chunk)
+
+    return "false"

bin/handlers/sorry.py

@@ -0,0 +1,7 @@
+# sends a custom response instead of the usual 404
+
+
+def main(cli, vn, rem):
+    msg = f"sorry {cli.ip} but {cli.vpath} doesn't exist"
+
+    return str(cli.reply(msg.encode("utf-8"), 404, "text/plain"))

copyparty/__main__.py

@@ -528,6 +528,51 @@ def get_sects():
             ).rstrip()
             + build_flags_desc(),
         ],
+        [
+            "handlers",
+            "use plugins to handle certain events",
+            dedent(
+                """
+            usually copyparty returns a 404 if a file does not exist, and
+            403 if a user tries to access a file they don't have access to
+
+            you can load a plugin which will be invoked right before this
+            happens, and the plugin can choose to override this behavior
+
+            load the plugin using --args or volflags; for example \033[36m
+             --on404 ~/partyhandlers/not404.py
+             -v .::r:c,on404=~/partyhandlers/not404.py
+            \033[0m
+
+            the file must define a `main(cli,vn,rem)` function:
+             cli: the HttpCli instance
+             vn:  the VFS which overlaps with the requested URL
+             rem: the remainder of the URL below the VFS mountpoint
+
+            `main` must return a string; one of the following:
+
+            > "true": the plugin has responded to the request,
+                and the TCP connection should be kept open
+
+            > "false": the plugin has responded to the request,
+                and the TCP connection should be terminated
+
+            > "retry": the plugin has done something to resolve the 404
+                situation, and copyparty should reattempt reading the file.
+                if it still fails, a regular 404 will be returned
+
+            > "allow": should ignore the insufficient permissions
+                and let the client continue anyways
+
+            > "": the plugin has not handled the request;
+                try the next plugin or return the usual 404 or 403
+
+            PS! the folder that contains the python file should ideally
+              not contain many other python files, and especially nothing
+              with filenames that overlap with modules used by copyparty
+            """
+            ),
+        ],
         [
             "hooks",
             "execute commands before/after various events",
@@ -858,6 +903,13 @@ def add_smb(ap):
     ap2.add_argument("--smbvvv", action="store_true", help="verbosest")
 
 
+def add_handlers(ap):
+    ap2 = ap.add_argument_group('handlers (see --help-handlers)')
+    ap2.add_argument("--on404", metavar="PY", type=u, action="append", help="handle 404s by executing PY file")
+    ap2.add_argument("--on403", metavar="PY", type=u, action="append", help="handle 403s by executing PY file")
+    ap2.add_argument("--hot-handlers", action="store_true", help="reload handlers on each request -- expensive but convenient when hacking on stuff")
+
+
 def add_hooks(ap):
     ap2 = ap.add_argument_group('event hooks (see --help-hooks)')
     ap2.add_argument("--xbu", metavar="CMD", type=u, action="append", help="execute CMD before a file upload starts")
@@ -1113,6 +1165,7 @@ def run_argparse(
     add_optouts(ap)
     add_shutdown(ap)
     add_yolo(ap)
+    add_handlers(ap)
     add_hooks(ap)
     add_ui(ap, retry)
     add_admin(ap)

copyparty/authsrv.py

@@ -1047,7 +1047,7 @@ class AuthSrv(object):
             flags[name] = True
             return
 
-        if name not in "mtp xbu xau xiu xbr xar xbd xad xm".split():
+        if name not in "mtp xbu xau xiu xbr xar xbd xad xm on404 on403".split():
             if value is True:
                 t = "└─add volflag [{}] = {}  ({})"
             else:
@@ -1446,7 +1446,7 @@ class AuthSrv(object):
 
             # append additive args from argv to volflags
             hooks = "xbu xau xiu xbr xar xbd xad xm".split()
-            for name in ["mtp"] + hooks:
+            for name in "mtp on404 on403".split() + hooks:
                 self._read_volflag(vol.flags, name, getattr(self.args, name), True)
 
             for hn in hooks:
@@ -1835,7 +1835,7 @@ class AuthSrv(object):
         ]
 
         csv = set("i p".split())
-        lst = set("c ihead mtm mtp xad xar xau xiu xbd xbr xbu xm".split())
+        lst = set("c ihead mtm mtp xad xar xau xiu xbd xbr xbu xm on404 on403".split())
         askip = set("a v c vc cgen theme".split())
 
         # keymap from argv to vflag

copyparty/cfg.py

@@ -125,6 +125,10 @@ flagcats = {
         "dathumb": "disables audio thumbnails (spectrograms)",
         "dithumb": "disables image thumbnails",
     },
+    "handlers\n(better explained in --help-handlers)": {
+        "on404=PY": "handle 404s by executing PY file",
+        "on403=PY": "handle 403s by executing PY file",
+    },
     "event hooks\n(better explained in --help-hooks)": {
         "xbu=CMD": "execute CMD before a file upload starts",
         "xau=CMD": "execute CMD after  a file upload finishes",

copyparty/httpcli.py

@@ -58,6 +58,7 @@ from .util import (
     html_escape,
     humansize,
     ipnorm,
+    loadpy,
     min_ex,
     quotep,
     rand_name,
@@ -767,11 +768,27 @@ class HttpCli(object):
             return True
 
         if not self.can_read and not self.can_write and not self.can_get:
-            if self.vpath:
+            t = "@{} has no access to [{}]"
-                self.log("inaccessible: [{}]".format(self.vpath))
+            self.log(t.format(self.uname, self.vpath))
-                return self.tx_404(True)
 
-            self.uparam["h"] = ""
+            if self.avn and "on403" in self.avn.flags:
+                vn, rem = self.asrv.vfs.get(self.vpath, self.uname, False, False)
+                ret = self.on40x(vn.flags["on403"], vn, rem)
+                if ret == "true":
+                    return True
+                elif ret == "false":
+                    return False
+                elif ret == "allow":
+                    self.log("plugin override; access permitted")
+                    self.can_read = self.can_write = self.can_move = True
+                    self.can_delete = self.can_get = self.can_upget = True
+                else:
+                    return self.tx_404(True)
+            else:
+                if self.vpath:
+                    return self.tx_404(True)
+
+                self.uparam["h"] = ""
 
         if "tree" in self.uparam:
             return self.tx_tree()
@@ -3056,6 +3073,20 @@ class HttpCli(object):
         self.reply(html.encode("utf-8"), status=rc)
         return True
 
+    def on40x(self, mods: list[str], vn: VFS, rem: str) -> str:
+        for mpath in mods:
+            try:
+                mod = loadpy(mpath, self.args.hot_handlers)
+            except Exception as ex:
+                self.log("import failed: {!r}".format(ex))
+                continue
+
+            ret = mod.main(self, vn, rem)
+            if ret:
+                return ret.lower()
+
+        return ""  # unhandled / fallthrough
+
     def scanvol(self) -> bool:
         if not self.can_read or not self.can_write:
             raise Pebkac(403, "not allowed for user " + self.uname)
@@ -3373,7 +3404,21 @@ class HttpCli(object):
         try:
             st = bos.stat(abspath)
         except:
-            return self.tx_404()
+            if "on404" not in vn.flags:
+                return self.tx_404()
+
+            ret = self.on40x(vn.flags["on404"], vn, rem)
+            if ret == "true":
+                return True
+            elif ret == "false":
+                return False
+            elif ret == "retry":
+                try:
+                    st = bos.stat(abspath)
+                except:
+                    return self.tx_404()
+            else:
+                return self.tx_404()
 
         if rem.startswith(".hist/up2k.") or (
             rem.endswith("/dir.txt") and rem.startswith(".hist/th/")

copyparty/util.py

@@ -2727,6 +2727,34 @@ def runhook(
     return True
 
 
+def loadpy(ap: str, hot: bool) -> Any:
+    """
+    a nice can of worms capable of causing all sorts of bugs
+    depending on what other inconveniently named files happen
+    to be in the same folder
+    """
+    if ap.startswith("~"):
+        ap = os.path.expanduser(ap)
+
+    mdir, mfile = os.path.split(absreal(ap))
+    mname = mfile.rsplit(".", 1)[0]
+    sys.path.insert(0, mdir)
+
+    if PY2:
+        mod = __import__(mname)
+        if hot:
+            reload(mod)
+    else:
+        import importlib
+
+        mod = importlib.import_module(mname)
+        if hot:
+            importlib.reload(mod)
+
+    sys.path.remove(mdir)
+    return mod
+
+
 def gzip_orig_sz(fn: str) -> int:
     with open(fsenc(fn), "rb") as f:
         f.seek(-4, 2)

tests/util.py

@@ -116,7 +116,7 @@ class Cfg(Namespace):
         ex = "ah_alg doctitle favico html_head lg_sbf log_fk md_sbf mth textfiles unlist R RS SR"
         ka.update(**{k: "" for k in ex.split()})
 
-        ex = "xad xar xau xbd xbr xbu xiu xm"
+        ex = "on403 on404 xad xar xau xbd xbr xbu xiu xm"
         ka.update(**{k: [] for k in ex.split()})
 
         super(Cfg, self).__init__(