google isn't taking the hint

specifically google, but also some others, have started ignoring
rel="nofollow" while also understanding just enough javascript to
try viewing binary files as text

copyparty/__main__.py

@@ -40,6 +40,7 @@ from .cfg import flagcats, onedash
 from .svchub import SvcHub
 from .util import (
     APPLESAN_TXT,
+    BAD_BOTS,
     DEF_EXP,
     DEF_MTE,
     DEF_MTH,
@@ -1244,6 +1245,7 @@ def add_optouts(ap):
     ap2.add_argument("--zipmaxt", metavar="TXT", type=u, default="", help="custom errormessage when download size exceeds max (volflag=zipmaxt)")
     ap2.add_argument("--zipmaxu", action="store_true", help="authenticated users bypass the zip size limit (volflag=zipmaxu)")
     ap2.add_argument("--zip-who", metavar="LVL", type=int, default=3, help="who can download as zip/tar? [\033[32m0\033[0m]=nobody, [\033[32m1\033[0m]=admins, [\033[32m2\033[0m]=authenticated-with-read-access, [\033[32m3\033[0m]=everyone-with-read-access (volflag=zip_who)\n\033[1;31mWARNING:\033[0m if a nested volume has a more restrictive value than a parent volume, then this will be \033[33mignored\033[0m if the download is initiated from the parent, more lenient volume")
+    ap2.add_argument("--ua-nozip", metavar="PTN", type=u, default=BAD_BOTS, help="regex of user-agents to reject from download-as-zip/tar; disable with [\033[32mno\033[0m] or blank")
     ap2.add_argument("--no-zip", action="store_true", help="disable download as zip/tar; same as \033[33m--zip-who=0\033[0m")
     ap2.add_argument("--no-tarcmp", action="store_true", help="disable download as compressed tar (?tar=gz, ?tar=bz2, ?tar=xz, ?tar=gz:9, ...)")
     ap2.add_argument("--no-lifetime", action="store_true", help="do not allow clients (or server config) to schedule an upload to be deleted after a given time")
@@ -1434,6 +1436,7 @@ def add_txt(ap):
     ap2.add_argument("--exp", action="store_true", help="enable textfile expansion -- replace {{self.ip}} and such; see \033[33m--help-exp\033[0m (volflag=exp)")
     ap2.add_argument("--exp-md", metavar="V,V,V", type=u, default=DEF_EXP, help="comma/space-separated list of placeholders to expand in markdown files; add/remove stuff on the default list with +hdr_foo or /vf.scan (volflag=exp_md)")
     ap2.add_argument("--exp-lg", metavar="V,V,V", type=u, default=DEF_EXP, help="comma/space-separated list of placeholders to expand in prologue/epilogue files (volflag=exp_lg)")
+    ap2.add_argument("--ua-nodoc", metavar="PTN", type=u, default=BAD_BOTS, help="regex of user-agents to reject from viewing documents through ?doc=[...]; disable with [\033[32mno\033[0m] or blank")
 
 
 def add_og(ap):

copyparty/httpcli.py

@@ -3807,6 +3807,9 @@ class HttpCli(object):
             return "download-as-zip/tar is admin-only on this server"
         elif lvl <= 2 and self.uname in ("", "*"):
             return "you must be authenticated to download-as-zip/tar on this server"
+        elif self.args.ua_nozip and self.args.ua_nozip.search(self.ua):
+            t = "this URL contains no valuable information for bots/crawlers"
+            raise Pebkac(403, t)
         return ""
 
     def tx_res(self, req_path: str) -> bool:
@@ -6291,6 +6294,10 @@ class HttpCli(object):
 
         doc = self.uparam.get("doc") if self.can_read else None
         if doc:
+            zp = self.args.ua_nodoc
+            if zp and zp.search(self.ua):
+                t = "this URL contains no valuable information for bots/crawlers"
+                raise Pebkac(403, t)
             j2a["docname"] = doc
             doctxt = None
             dfn = lnames.get(doc.lower())

copyparty/svchub.py

@@ -769,7 +769,8 @@ class SvcHub(object):
                 vs = os.path.expandvars(os.path.expanduser(vs))
                 setattr(al, k, vs)
 
-        for k in "dav_ua1 sus_urls nonsus_urls".split(" "):
+        zs = "dav_ua1 sus_urls nonsus_urls ua_nodoc ua_nozip"
+        for k in zs.split(" "):
             vs = getattr(al, k)
             if not vs or vs == "no":
                 setattr(al, k, None)

copyparty/util.py

@@ -245,6 +245,9 @@ SYMTIME = PY36 and os.utime in os.supports_follow_symlinks
 
 META_NOBOTS = '<meta name="robots" content="noindex, nofollow">\n'
 
+# smart enough to understand javascript while also ignoring rel="nofollow"
+BAD_BOTS = r"Barkrowler|bingbot|BLEXBot|Googlebot|GPTBot|PetalBot|SeekportBot|SemrushBot|YandexBot"
+
 FFMPEG_URL = "https://www.gyan.dev/ffmpeg/builds/ffmpeg-git-full.7z"
 
 URL_PRJ = "https://github.com/9001/copyparty"

tests/util.py

@@ -135,7 +135,7 @@ class Cfg(Namespace):
         ex = "dav_inf dedup dotpart dotsrch hook_v no_dhash no_fastboot no_fpool no_htp no_rescan no_sendfile no_ses no_snap no_up_list no_voldump re_dhash plain_ip"
         ka.update(**{k: True for k in ex.split()})
 
-        ex = "ah_cli ah_gen css_browser hist ipu js_browser js_other mime mimes no_forget no_hash no_idx nonsus_urls og_tpl og_ua"
+        ex = "ah_cli ah_gen css_browser hist ipu js_browser js_other mime mimes no_forget no_hash no_idx nonsus_urls og_tpl og_ua ua_nodoc ua_nozip"
         ka.update(**{k: None for k in ex.split()})
 
         ex = "hash_mt hsortn safe_dedup srch_time u2abort u2j u2sz"