tools/user-activity/arvados_user_activity/main.py

   1 #!/usr/bin/env python3
   2 # Copyright (C) The Arvados Authors. All rights reserved.
   3 #
   4 # SPDX-License-Identifier: AGPL-3.0
   5
   6 import argparse
   7 import sys
   8
   9 import arvados
  10 import arvados.util
  11 import datetime
  12 import ciso8601
  13
  14 def parse_arguments(arguments):
  15     arg_parser = argparse.ArgumentParser()
  16     arg_parser.add_argument('--start', help='Start date for the report in YYYY-MM-DD format (UTC)')
  17     arg_parser.add_argument('--end', help='End date for the report in YYYY-MM-DD format (UTC)')
  18     arg_parser.add_argument('--days', type=int, help='Number of days before now() to start the report')
  19     args = arg_parser.parse_args(arguments)
  20
  21     if args.days and (args.start or args.end):
  22         arg_parser.print_help()
  23         print("Error: either specify --days or both --start and --end")
  24         exit(1)
  25
  26     if not args.days and (not args.start or not args.end):
  27         arg_parser.print_help()
  28         print("\nError: either specify --days or both --start and --end")
  29         exit(1)
  30
  31     if (args.start and not args.end) or (args.end and not args.start):
  32         arg_parser.print_help()
  33         print("\nError: no start or end date found, either specify --days or both --start and --end")
  34         exit(1)
  35
  36     if args.days:
  37         to = datetime.datetime.utcnow()
  38         since = to - datetime.timedelta(days=args.days)
  39
  40     if args.start:
  41         try:
  42             since = datetime.datetime.strptime(args.start,"%Y-%m-%d")
  43         except:
  44             arg_parser.print_help()
  45             print("\nError: start date must be in YYYY-MM-DD format")
  46             exit(1)
  47
  48     if args.end:
  49         try:
  50             to = datetime.datetime.strptime(args.end,"%Y-%m-%d")
  51         except:
  52             arg_parser.print_help()
  53             print("\nError: end date must be in YYYY-MM-DD format")
  54             exit(1)
  55
  56     return args, since, to
  57
  58 def getowner(arv, uuid, owners):
  59     if uuid is None:
  60         return None
  61     if uuid[6:11] == "tpzed":
  62         return uuid
  63
  64     if uuid not in owners:
  65         try:
  66             gp = arv.groups().get(uuid=uuid).execute()
  67             owners[uuid] = gp["owner_uuid"]
  68         except:
  69             owners[uuid] = None
  70
  71     return getowner(arv, owners[uuid], owners)
  72
  73 def getuserinfo(arv, uuid):
  74     try:
  75         u = arv.users().get(uuid=uuid).execute()
  76     except:
  77         return "deleted user (%susers/%s)" % (arv.config()["Services"]["Workbench1"]["ExternalURL"],
  78                                                        uuid)
  79     prof = "\n".join("  %s: \"%s\"" % (k, v) for k, v in u["prefs"].get("profile", {}).items() if v)
  80     if prof:
  81         prof = "\n"+prof+"\n"
  82     return "%s %s <%s> (%susers/%s)%s" % (u["first_name"], u["last_name"], u["email"],
  83                                                        arv.config()["Services"]["Workbench1"]["ExternalURL"],
  84                                                        uuid, prof)
  85
  86 collectionNameCache = {}
  87 def getCollectionName(arv, uuid, pdh):
  88     lookupField = uuid
  89     filters = [["uuid","=",uuid]]
  90     cached = uuid in collectionNameCache
  91     # look up by uuid if it is available, fall back to look up by pdh
  92     if len(uuid) != 27:
  93         # Look up by pdh. Note that this can be misleading; the download could
  94         # have happened from a collection with the same pdh but different name.
  95         # We arbitrarily pick the oldest collection with the pdh to lookup the
  96         # name, if the uuid for the request is not known.
  97         lookupField = pdh
  98         filters = [["portable_data_hash","=",pdh]]
  99         cached = pdh in collectionNameCache
 100
 101     if not cached:
 102         u = arv.collections().list(filters=filters,order="created_at",limit=1).execute().get("items")
 103         if len(u) < 1:
 104             return "(deleted)"
 105         collectionNameCache[lookupField] = u[0]["name"]
 106     return collectionNameCache[lookupField]
 107
 108 def getname(u):
 109     return "\"%s\" (%s)" % (u["name"], u["uuid"])
 110
 111 def main(arguments=None):
 112     if arguments is None:
 113         arguments = sys.argv[1:]
 114
 115     args, since, to = parse_arguments(arguments)
 116
 117     arv = arvados.api()
 118
 119     print("User activity on %s between %s and %s\n" % (arv.config()["ClusterID"],
 120                                                        since.isoformat(sep=" ", timespec="minutes"),
 121                                                        to.isoformat(sep=" ", timespec="minutes")))
 122
 123     events = arvados.util.keyset_list_all(arv.logs().list, filters=[["created_at", ">=", since.isoformat()],["created_at", "<", to.isoformat()]])
 124
 125     users = {}
 126     owners = {}
 127
 128     for e in events:
 129         owner = getowner(arv, e["object_owner_uuid"], owners)
 130         users.setdefault(owner, [])
 131         event_at = ciso8601.parse_datetime(e["event_at"]).astimezone().isoformat(sep=" ", timespec="minutes")
 132         # loguuid = e["uuid"]
 133         loguuid = ""
 134
 135         if e["event_type"] == "create" and e["object_uuid"][6:11] == "tpzed":
 136             users.setdefault(e["object_uuid"], [])
 137             users[e["object_uuid"]].append("%s User account created" % event_at)
 138
 139         elif e["event_type"] == "update" and e["object_uuid"][6:11] == "tpzed":
 140             pass
 141
 142         elif e["event_type"] == "create" and e["object_uuid"][6:11] == "xvhdp":
 143             if e["properties"]["new_attributes"]["requesting_container_uuid"] is None:
 144                 users[owner].append("%s Ran container %s %s" % (event_at, getname(e["properties"]["new_attributes"]), loguuid))
 145
 146         elif e["event_type"] == "update" and e["object_uuid"][6:11] == "xvhdp":
 147             pass
 148
 149         elif e["event_type"] == "create" and e["object_uuid"][6:11] == "j7d0g":
 150             users[owner].append("%s Created project %s" %  (event_at, getname(e["properties"]["new_attributes"])))
 151
 152         elif e["event_type"] == "delete" and e["object_uuid"][6:11] == "j7d0g":
 153             users[owner].append("%s Deleted project %s" % (event_at, getname(e["properties"]["old_attributes"])))
 154
 155         elif e["event_type"] == "update" and e["object_uuid"][6:11] == "j7d0g":
 156             users[owner].append("%s Updated project %s" % (event_at, getname(e["properties"]["new_attributes"])))
 157
 158         elif e["event_type"] in ("create", "update") and e["object_uuid"][6:11] == "gj3su":
 159             since_last = None
 160             if len(users[owner]) > 0 and users[owner][-1].endswith("activity"):
 161                 sp = users[owner][-1].split(" ")
 162                 start = sp[0]+" "+sp[1]
 163                 since_last = ciso8601.parse_datetime(event_at) - ciso8601.parse_datetime(sp[3]+" "+sp[4])
 164                 span = ciso8601.parse_datetime(event_at) - ciso8601.parse_datetime(start)
 165
 166             if since_last is not None and since_last < datetime.timedelta(minutes=61):
 167                 users[owner][-1] = "%s to %s (%02d:%02d) Account activity" % (start, event_at, span.days*24 + int(span.seconds/3600), int((span.seconds % 3600)/60))
 168             else:
 169                 users[owner].append("%s to %s (0:00) Account activity" % (event_at, event_at))
 170
 171         elif e["event_type"] == "create" and e["object_uuid"][6:11] == "o0j2j":
 172             if e["properties"]["new_attributes"]["link_class"] == "tag":
 173                 users[owner].append("%s Tagged %s" % (event_at, e["properties"]["new_attributes"]["head_uuid"]))
 174             elif e["properties"]["new_attributes"]["link_class"] == "permission":
 175                 users[owner].append("%s Shared %s with %s" % (event_at, e["properties"]["new_attributes"]["tail_uuid"], e["properties"]["new_attributes"]["head_uuid"]))
 176             else:
 177                 users[owner].append("%s %s %s %s %s" % (event_at, e["event_type"], e["object_kind"], e["object_uuid"], loguuid))
 178
 179         elif e["event_type"] == "delete" and e["object_uuid"][6:11] == "o0j2j":
 180             if e["properties"]["old_attributes"]["link_class"] == "tag":
 181                 users[owner].append("%s Untagged %s" % (event_at, e["properties"]["old_attributes"]["head_uuid"]))
 182             elif e["properties"]["old_attributes"]["link_class"] == "permission":
 183                 users[owner].append("%s Unshared %s with %s" % (event_at, e["properties"]["old_attributes"]["tail_uuid"], e["properties"]["old_attributes"]["head_uuid"]))
 184             else:
 185                 users[owner].append("%s %s %s %s %s" % (event_at, e["event_type"], e["object_kind"], e["object_uuid"], loguuid))
 186
 187         elif e["event_type"] == "create" and e["object_uuid"][6:11] == "4zz18":
 188             if e["properties"]["new_attributes"]["properties"].get("type") in ("log", "output", "intermediate"):
 189                 pass
 190             else:
 191                 users[owner].append("%s Created collection %s %s" % (event_at, getname(e["properties"]["new_attributes"]), loguuid))
 192
 193         elif e["event_type"] == "update" and e["object_uuid"][6:11] == "4zz18":
 194             users[owner].append("%s Updated collection %s %s" % (event_at, getname(e["properties"]["new_attributes"]), loguuid))
 195
 196         elif e["event_type"] == "delete" and e["object_uuid"][6:11] == "4zz18":
 197             if e["properties"]["old_attributes"]["properties"].get("type") in ("log", "output", "intermediate"):
 198                 pass
 199             else:
 200                 users[owner].append("%s Deleted collection %s %s" % (event_at, getname(e["properties"]["old_attributes"]), loguuid))
 201
 202         elif e["event_type"] == "file_download":
 203                 users.setdefault(e["object_uuid"], [])
 204                 users[e["object_uuid"]].append("%s Downloaded file \"%s\" from \"%s\" (%s) (%s)" % (event_at,
 205                                                                                        e["properties"].get("collection_file_path") or e["properties"].get("reqPath"),
 206                                                                                        getCollectionName(arv, e["properties"].get("collection_uuid"), e["properties"].get("portable_data_hash")),
 207                                                                                        e["properties"].get("collection_uuid"),
 208                                                                                        e["properties"].get("portable_data_hash")))
 209
 210         elif e["event_type"] == "file_upload":
 211                 users.setdefault(e["object_uuid"], [])
 212                 users[e["object_uuid"]].append("%s Uploaded file \"%s\" to \"%s\" (%s)" % (event_at,
 213                                                                                     e["properties"].get("collection_file_path") or e["properties"].get("reqPath"),
 214                                                                                     getCollectionName(arv, e["properties"].get("collection_uuid"), e["properties"].get("portable_data_hash")),
 215                                                                                     e["properties"].get("collection_uuid")))
 216
 217         else:
 218             users[owner].append("%s %s %s %s %s" % (event_at, e["event_type"], e["object_kind"], e["object_uuid"], loguuid))
 219
 220     for k,v in users.items():
 221         if k is None or k.endswith("-tpzed-000000000000000"):
 222             continue
 223         print(getuserinfo(arv, k))
 224         for ev in v:
 225             print("  %s" % ev)
 226         print("")
 227
 228 if __name__ == "__main__":
 229     main()