bin/djbdns-logparse.py

   1 #!/usr/bin/python3
   2 """
   3 Convert tinydns and dnscache logs to human-readable form
   4 """
   5
   6 import re
   7 from struct import pack
   8 from time import strftime, gmtime
   9 from subprocess import Popen, PIPE
  10
  11
  12 ## Regular expressions for matching tinydns/dnscache log lines. We
  13 ## compile these once here rather than within the corresponding
  14 ## matching functions, because the latter get executed repeatedly.
  15
  16 # This first pattern is used to match the timestamp format that the
  17 # tai64nlocal program produces. It appears in both dnscache and
  18 # tinydns lines, after they've been piped through tai64nlocal, of
  19 # course.
  20 timestamp_pat = r'[\d-]+ [\d:\.]+'
  21
  22 # The regex to match dnscache log lines.
  23 dnscache_log_re = re.compile(fr'({timestamp_pat}) (\w+)(.*)')
  24
  25 # The "hex4" pattern matches a string of four hexadecimal digits. This
  26 # is used, for example, by tinydns to encode the query type
  27 # identifier.
  28 hex4_pat = r'[0-9a-f]{4}'
  29
  30 # The IP pattern matches a string of either 8 or 32 hexadecimal
  31 # characters, which correspond to IPv4 and IPv6 addresses,
  32 # respectively, in tinydns logs.
  33 ip_pat = r'[0-9a-f]{8,32}'
  34
  35 # The regex to match tinydns log lines.
  36 tinydns_log_re = re.compile(
  37     rf'({timestamp_pat}) ({ip_pat}):({hex4_pat}):({hex4_pat}) ([\+\-IC/]) ({hex4_pat}) (.*)'
  38 )
  39
  40 # A dictionary mapping query type identifiers, in decimal, to their
  41 # friendly names for tinydns. Reference:
  42 #
  43 #   https://en.wikipedia.org/wiki/List_of_DNS_record_types
  44 #
  45 # Note that mapping here is non-exhaustive, and that tinydns will
  46 # log responses for record types that it does not know about.
  47 query_type = {
  48       1: "a",
  49       2: "ns",
  50       5: "cname",
  51       6: "soa",
  52      12: "ptr",
  53      13: "hinfo",
  54      15: "mx",
  55      16: "txt",
  56      17: "rp",
  57      24: "sig",
  58      25: "key",
  59      28: "aaaa",
  60      33: "srv",
  61      35: "naptr",
  62      38: "a6",
  63      48: "dnskey",
  64      52: "tlsa",
  65      65: "https",
  66     252: "axfr",
  67     255: "any",
  68     257: "caa"
  69 }
  70
  71 # tinydns can drop a query for one of three reasons; this dictionary
  72 # maps the symbol that gets logged in each case to a human-readable
  73 # reason.
  74 query_drop_reason = {
  75     "-": "no authority",
  76     "I": "invalid query",
  77     "C": "invalid class",
  78 }
  79
  80
  81 def convert_ip(ip : str):
  82     """
  83     Convert a hex string representing an IP address to conventional
  84     human-readable form, ie. dotted-quad decimal for IPv4, and
  85     8 colon-separated hex shorts for IPv6.
  86
  87     Examples
  88     --------
  89
  90         >>> convert_ip("7f000001")
  91         '127.0.0.1'
  92         >>> convert_ip("00000000000000000000ffff7f000001")
  93         '0000:0000:0000:0000:0000:ffff:7f00:0001'
  94
  95     """
  96     if len(ip) == 8:
  97         # IPv4, eg. "7f000001" -> "7f 00 00 01" -> "127.0.0.1"
  98         return "%d.%d.%d.%d" % tuple(pack(">L", int(ip, 16)))
  99     elif len(ip) == 32:
 100         # IPv6 is actually simpler -- it's just a string-slicing operation.
 101         return ":".join([ip[(4*i) : (4*i+4)] for i in range(8)])
 102
 103
 104 def decode_client(words, i):
 105     chunks = words[i].split(":")
 106     if len(chunks) == 2:                # ip:port
 107         words[i] = "%s:%d" % (convert_ip(chunks[0]), int(chunks[1], 16))
 108     elif len(chunks) == 3:
 109         words[i] = "%s:%d (id %d)" % (convert_ip(chunks[0]),
 110                                       int(chunks[1], 16),
 111                                       int(chunks[2], 16))
 112
 113 def decode_ip(words, i):
 114     words[i] = convert_ip(words[i])
 115
 116 def decode_ttl(words, i):
 117     words[i] = "TTL=%s" % words[i]
 118
 119 def decode_serial(words, i):
 120     serial = int(words[i])
 121     words[i] = "#%d" % serial
 122
 123 def decode_type(words, i):
 124     qt = words[i]
 125     words[i] = query_type.get(int(qt), qt)
 126
 127 def handle_dnscache_log(line, match):
 128     (timestamp, event, data) = match.groups()
 129
 130     words = data.split()
 131     if event == "cached":
 132         if words[0] not in ("cname", "ns", "nxdomain"):
 133             decode_type(words, 0)
 134
 135     elif event == "drop":
 136         decode_serial(words, 0)
 137
 138     elif event == "lame":
 139         decode_ip(words, 0)
 140
 141     elif event == "nodata":
 142         decode_ip(words, 0)
 143         decode_ttl(words, 1)
 144         decode_type(words, 2)
 145
 146     elif event == "nxdomain":
 147         decode_ip(words, 0)
 148         decode_ttl(words, 1)
 149
 150     elif event == "query":
 151         decode_serial(words, 0)
 152         decode_client(words, 1)
 153         decode_type(words, 2)
 154
 155     elif event == "rr":
 156         decode_ip(words, 0)
 157         decode_ttl(words, 1)
 158         if words[2] not in ("cname", "mx", "ns", "ptr", "soa"):
 159             decode_type(words, 2)
 160             if words[2] == "a":         # decode answer to an A query
 161                 decode_ip(words, 4)
 162             if words[2] == "txt":       # text record
 163                 response = words[4]
 164                 if response.endswith("..."):
 165                     ellipsis = "..."
 166                     response = response[0:-3]
 167                 else:
 168                     ellipsis = ""
 169                 length = int(response[0:2], 16)
 170                 chars = []
 171                 for i in range(1, len(response)/2):
 172                     chars.append(chr(int(response[2*i : (2*i)+2], 16)))
 173                 words[4] = "%d:\"%s%s\"" % (length, "".join(chars), ellipsis)
 174
 175     elif event == "sent":
 176         decode_serial(words, 0)
 177
 178     elif event == "stats":
 179         words[0] = "count=%s" % words[0]
 180         words[1] = "motion=%s" % words[1]
 181         words[2] = "udp-active=%s" % words[2]
 182         words[3] = "tcp-active=%s" % words[3]
 183
 184     elif event == "tx":
 185         words[0] = "g=%s" % words[0]
 186         decode_type(words, 1)
 187         # words[2] = name
 188         # words[3] = control (domain for which these servers are believed
 189         #            to be authoritative)
 190         for i in range(4, len(words)):
 191             decode_ip(words, i)
 192
 193     elif event in ("tcpopen", "tcpclose"):
 194         decode_client(words, 0)
 195
 196     print(timestamp, event, " ".join(words))
 197
 198
 199 def handle_tinydns_log(line : str, match: re.Match):
 200     """
 201     Handle a line that matched the ``tinydns_log_re`` regex.
 202
 203     Parameters
 204     ----------
 205
 206     line : string
 207         The tinydns log line that matched ``tinydns_log_re``.
 208
 209     match : re.Match
 210         The match object that was returned when ``line`` was
 211         tested against ``tinydns_log_re``.
 212
 213     Examples
 214     --------
 215
 216         >>> line = "2022-09-14 21:04:40.206516500 7f000001:9d61:be69 - 0001 www.example.com"
 217         >>> match = tinydns_log_re.match(line)
 218         >>> handle_tinydns_log(line, match)
 219         2022-09-14 21:04:40.206516500 dropped query (no authority) from 127.0.0.1:40289 (id 48745): a www.example.com
 220
 221     """
 222     (timestamp, ip, port, id, code, type, name) = match.groups()
 223     ip = convert_ip(ip)
 224     port = int(port, 16)
 225     id = int(id, 16)
 226
 227     # Convert the "type" field to a human-readable record type name
 228     # using the query_type dictionary. If the right name isn't present
 229     # in the dictionary, we use the (decimal) type id instead.
 230     type = int(type, 16)                # "001c" -> 28
 231     type = query_type.get(type, type)   # 28 -> "aaaa"
 232
 233     print(timestamp, end=' ')
 234
 235     if code == "+":
 236         print ("sent response to %s:%s (id %s): %s %s"
 237                % (ip, port, id, type, name))
 238     elif code in ("-", "I", "C"):
 239         reason = query_drop_reason[code]
 240         print ("dropped query (%s) from %s:%s (id %s): %s %s"
 241                % (reason, ip, port, id, type, name))
 242     elif code == "/":
 243         print ("dropped query (couldn't parse) from %s:%s"
 244                % (ip, port))
 245     else:
 246         print ("%s from %s:%s (id %s): %s %s"
 247                % (code, ip, port, id, type, name))
 248
 249
 250 def parse_logfile(file):
 251     # Open pipe to tai64nlocal: we will write lines of our input (the
 252     # raw log file) to it, and read log lines with readable timestamps
 253     # from it.
 254     tai = Popen(["tai64nlocal"], stdin=PIPE, stdout=PIPE, text=True, bufsize=0)
 255
 256     for line in file:
 257         tai.stdin.write(line)
 258         line = tai.stdout.readline()
 259
 260         match = tinydns_log_re.match(line)
 261         if match:
 262             handle_tinydns_log(line, match)
 263             continue
 264
 265         match = dnscache_log_re.match(line)
 266         if match:
 267             handle_dnscache_log(line, match)
 268             continue
 269
 270         print(line)
 271
 272 def main():
 273     # Create an argument parser using the file's docsctring as its
 274     # description.
 275     from argparse import ArgumentParser, FileType
 276     parser = ArgumentParser(description = __doc__)
 277
 278     # Parse zero or more positional arguments into a list of
 279     # "logfiles". If none are given, read from stdin instead.
 280     from sys import stdin
 281     parser.add_argument("logfiles",
 282                         metavar="LOGFILE",
 283                         type=FileType("r"),
 284                         nargs="*",
 285                         default=[stdin],
 286                         help="djbdns logfile to process (default: stdin)")
 287
 288     args = parser.parse_args()
 289     for f in args.logfiles:
 290         parse_logfile(f)
 291
 292
 293
 294
 295 if __name__ == "__main__":
 296     main()