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')
  17     arg_parser.add_argument('--end', help='End date for the report in YYYY-MM-DD format')
  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         p.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         p.print_help()
  28         print("Error: 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         p.print_help()
  33         print("Error: no start or end date found, either specify --days or both --start and --end")
  34         exit(1)
  35
  36     return args
  37
  38 def getowner(arv, uuid, owners):
  39     if uuid is None:
  40         return None
  41     if uuid[6:11] == "tpzed":
  42         return uuid
  43
  44     if uuid not in owners:
  45         try:
  46             gp = arv.groups().get(uuid=uuid).execute()
  47             owners[uuid] = gp["owner_uuid"]
  48         except:
  49             owners[uuid] = None
  50
  51     return getowner(arv, owners[uuid], owners)
  52
  53 def getuserinfo(arv, uuid):
  54     try:
  55         u = arv.users().get(uuid=uuid).execute()
  56     except:
  57         return "deleted user (%susers/%s)" % (arv.config()["Services"]["Workbench1"]["ExternalURL"],
  58                                                        uuid)
  59     prof = "\n".join("  %s: \"%s\"" % (k, v) for k, v in u["prefs"].get("profile", {}).items() if v)
  60     if prof:
  61         prof = "\n"+prof+"\n"
  62     return "%s %s <%s> (%susers/%s)%s" % (u["first_name"], u["last_name"], u["email"],
  63                                                        arv.config()["Services"]["Workbench1"]["ExternalURL"],
  64                                                        uuid, prof)
  65
  66 collectionNameCache = {}
  67 def getCollectionName(arv, pdh):
  68     if pdh not in collectionNameCache:
  69         u = arv.collections().list(filters=[["portable_data_hash","=",pdh]]).execute().get("items")
  70         if len(u) < 1:
  71             return "(deleted)"
  72         collectionNameCache[pdh] = u[0]["name"]
  73     return collectionNameCache[pdh]
  74
  75 def getname(u):
  76     return "\"%s\" (%s)" % (u["name"], u["uuid"])
  77
  78 def main(arguments=None):
  79     if arguments is None:
  80         arguments = sys.argv[1:]
  81
  82     args = parse_arguments(arguments)
  83
  84     arv = arvados.api()
  85
  86     if args.days:
  87         to = datetime.datetime.utcnow()
  88         since = to - datetime.timedelta(days=args.days)
  89
  90     if args.start:
  91         try:
  92             since = datetime.datetime.strptime(args.start,"%Y-%m-%d")
  93         except:
  94             p.print_help()
  95             print("Error: start date must be in YYYY-MM-DD format")
  96             exit(1)
  97
  98     if args.end:
  99         try:
 100             to = datetime.datetime.strptime(args.end,"%Y-%m-%d")
 101         except:
 102             p.print_help()
 103             print("Error: end date must be in YYYY-MM-DD format")
 104             exit(1)
 105
 106     print("User activity on %s between %s and %s\n" % (arv.config()["ClusterID"],
 107                                                        since.isoformat(sep=" ", timespec="minutes"),
 108                                                        to.isoformat(sep=" ", timespec="minutes")))
 109
 110     events = arvados.util.keyset_list_all(arv.logs().list, filters=[["created_at", ">=", since.isoformat()],["created_at", "<", to.isoformat()]])
 111
 112     users = {}
 113     owners = {}
 114
 115     for e in events:
 116         owner = getowner(arv, e["object_owner_uuid"], owners)
 117         users.setdefault(owner, [])
 118         event_at = ciso8601.parse_datetime(e["event_at"]).astimezone().isoformat(sep=" ", timespec="minutes")
 119         # loguuid = e["uuid"]
 120         loguuid = ""
 121
 122         if e["event_type"] == "create" and e["object_uuid"][6:11] == "tpzed":
 123             users.setdefault(e["object_uuid"], [])
 124             users[e["object_uuid"]].append("%s User account created" % event_at)
 125
 126         elif e["event_type"] == "update" and e["object_uuid"][6:11] == "tpzed":
 127             pass
 128
 129         elif e["event_type"] == "create" and e["object_uuid"][6:11] == "xvhdp":
 130             if e["properties"]["new_attributes"]["requesting_container_uuid"] is None:
 131                 users[owner].append("%s Ran container %s %s" % (event_at, getname(e["properties"]["new_attributes"]), loguuid))
 132
 133         elif e["event_type"] == "update" and e["object_uuid"][6:11] == "xvhdp":
 134             pass
 135
 136         elif e["event_type"] == "create" and e["object_uuid"][6:11] == "j7d0g":
 137             users[owner].append("%s Created project %s" %  (event_at, getname(e["properties"]["new_attributes"])))
 138
 139         elif e["event_type"] == "delete" and e["object_uuid"][6:11] == "j7d0g":
 140             users[owner].append("%s Deleted project %s" % (event_at, getname(e["properties"]["old_attributes"])))
 141
 142         elif e["event_type"] == "update" and e["object_uuid"][6:11] == "j7d0g":
 143             users[owner].append("%s Updated project %s" % (event_at, getname(e["properties"]["new_attributes"])))
 144
 145         elif e["event_type"] in ("create", "update") and e["object_uuid"][6:11] == "gj3su":
 146             since_last = None
 147             if len(users[owner]) > 0 and users[owner][-1].endswith("activity"):
 148                 sp = users[owner][-1].split(" ")
 149                 start = sp[0]+" "+sp[1]
 150                 since_last = ciso8601.parse_datetime(event_at) - ciso8601.parse_datetime(sp[3]+" "+sp[4])
 151                 span = ciso8601.parse_datetime(event_at) - ciso8601.parse_datetime(start)
 152
 153             if since_last is not None and since_last < datetime.timedelta(minutes=61):
 154                 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))
 155             else:
 156                 users[owner].append("%s to %s (0:00) Account activity" % (event_at, event_at))
 157
 158         elif e["event_type"] == "create" and e["object_uuid"][6:11] == "o0j2j":
 159             if e["properties"]["new_attributes"]["link_class"] == "tag":
 160                 users[owner].append("%s Tagged %s" % (event_at, e["properties"]["new_attributes"]["head_uuid"]))
 161             elif e["properties"]["new_attributes"]["link_class"] == "permission":
 162                 users[owner].append("%s Shared %s with %s" % (event_at, e["properties"]["new_attributes"]["tail_uuid"], e["properties"]["new_attributes"]["head_uuid"]))
 163             else:
 164                 users[owner].append("%s %s %s %s %s" % (event_at, e["event_type"], e["object_kind"], e["object_uuid"], loguuid))
 165
 166         elif e["event_type"] == "delete" and e["object_uuid"][6:11] == "o0j2j":
 167             if e["properties"]["old_attributes"]["link_class"] == "tag":
 168                 users[owner].append("%s Untagged %s" % (event_at, e["properties"]["old_attributes"]["head_uuid"]))
 169             elif e["properties"]["old_attributes"]["link_class"] == "permission":
 170                 users[owner].append("%s Unshared %s with %s" % (event_at, e["properties"]["old_attributes"]["tail_uuid"], e["properties"]["old_attributes"]["head_uuid"]))
 171             else:
 172                 users[owner].append("%s %s %s %s %s" % (event_at, e["event_type"], e["object_kind"], e["object_uuid"], loguuid))
 173
 174         elif e["event_type"] == "create" and e["object_uuid"][6:11] == "4zz18":
 175             if e["properties"]["new_attributes"]["properties"].get("type") in ("log", "output", "intermediate"):
 176                 pass
 177             else:
 178                 users[owner].append("%s Created collection %s %s" % (event_at, getname(e["properties"]["new_attributes"]), loguuid))
 179
 180         elif e["event_type"] == "update" and e["object_uuid"][6:11] == "4zz18":
 181             users[owner].append("%s Updated collection %s %s" % (event_at, getname(e["properties"]["new_attributes"]), loguuid))
 182
 183         elif e["event_type"] == "delete" and e["object_uuid"][6:11] == "4zz18":
 184             if e["properties"]["old_attributes"]["properties"].get("type") in ("log", "output", "intermediate"):
 185                 pass
 186             else:
 187                 users[owner].append("%s Deleted collection %s %s" % (event_at, getname(e["properties"]["old_attributes"]), loguuid))
 188
 189         elif e["event_type"] == "file_download":
 190                 users.setdefault(e["object_uuid"], [])
 191                 users[e["object_uuid"]].append("%s Downloaded file \"%s\" from \"%s\" (%s) (%s)" % (event_at,
 192                                                                                        e["properties"].get("collection_file_path") or e["properties"].get("reqPath"),
 193                                                                                        getCollectionName(arv, e["properties"].get("portable_data_hash")),
 194                                                                                        e["properties"].get("collection_uuid"),
 195                                                                                        e["properties"].get("portable_data_hash")))
 196
 197         elif e["event_type"] == "file_upload":
 198                 users.setdefault(e["object_uuid"], [])
 199                 users[e["object_uuid"]].append("%s Uploaded file \"%s\" to \"%s\" (%s)" % (event_at,
 200                                                                                     e["properties"].get("collection_file_path") or e["properties"].get("reqPath"),
 201                                                                                     getCollectionName(arv, e["properties"].get("portable_data_hash")),
 202                                                                                     e["properties"].get("collection_uuid")))
 203
 204         else:
 205             users[owner].append("%s %s %s %s %s" % (event_at, e["event_type"], e["object_kind"], e["object_uuid"], loguuid))
 206
 207     for k,v in users.items():
 208         if k is None or k.endswith("-tpzed-000000000000000"):
 209             continue
 210         print(getuserinfo(arv, k))
 211         for ev in v:
 212             print("  %s" % ev)
 213         print("")
 214
 215 if __name__ == "__main__":
 216     main()