bin/djbdns-logparse

   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 # common components of line-matching regexes
  13 timestamp_pat = r'[\d-]+ [\d:\.]+'      # output of tai64nlocal
  14 hex4_pat = r'[0-9a-f]{4}'
  15 ip_pat = r'[0-9a-f]{8,32}'              # IPv4 or IPv6 addresses in hex
  16
  17 # discriminate between dnscache and tinydns log lines
  18 tinydns_log_re = re.compile(
  19     r'(%s) (%s):(%s):(%s) ([\+\-IC/]) (%s) (.*)'
  20     % (timestamp_pat, ip_pat, hex4_pat, hex4_pat, hex4_pat))
  21 dnscache_log_re = re.compile(r'(%s) (\w+)(.*)' % timestamp_pat)
  22
  23 query_type = {
  24       1: "a",
  25       2: "ns",
  26       5: "cname",
  27       6: "soa",
  28      12: "ptr",
  29      13: "hinfo",
  30      15: "mx",
  31      16: "txt",
  32      17: "rp",
  33      24: "sig",
  34      25: "key",
  35      28: "aaaa",
  36      38: "a6",
  37     252: "axfr",
  38     255: "any",
  39 }
  40
  41 # for tinydns only
  42 query_drop_reason = {
  43     "-": "no authority",
  44     "I": "invalid query",
  45     "C": "invalid class",
  46     }
  47
  48
  49 def convert_ip(ip):
  50     """Convert a hex string representing an IP address to conventional
  51     human-readable form, ie. dotted-quad decimal for IPv4, and
  52     8 colon-separated hex shorts for IPv6.
  53     """
  54     if len(ip) == 8:
  55         # IPv4, eg. "7f000001" -> "127.0.0.1"
  56         return "%d.%d.%d.%d" % tuple(pack(">L", int(ip, 16)))
  57     elif len(ip) == 32:
  58         # IPv6 is actually simpler -- it's just a string-slicing operation,
  59         # eg. "00000000000000000000ffff7f000001" ->
  60         # "0000:0000:0000:0000:0000:ffff:7f00:0001"
  61         return ":".join([ip[(4*i) : (4*i+4)] for i in range(8)])
  62
  63
  64 def _cvt_ip(match):
  65     return convert_ip(match.group(1))
  66
  67 def _cvt_port(match):
  68     return ":" + str(int(match.group(1), 16))
  69
  70 def decode_client(words, i):
  71     chunks = words[i].split(":")
  72     if len(chunks) == 2:                # ip:port
  73         words[i] = "%s:%d" % (convert_ip(chunks[0]), int(chunks[1], 16))
  74     elif len(chunks) == 3:
  75         words[i] = "%s:%d (id %d)" % (convert_ip(chunks[0]),
  76                                       int(chunks[1], 16),
  77                                       int(chunks[2], 16))
  78
  79 def decode_ip(words, i):
  80     words[i] = convert_ip(words[i])
  81
  82 def decode_ttl(words, i):
  83     words[i] = "TTL=%s" % words[i]
  84
  85 def decode_serial(words, i):
  86     serial = int(words[i])
  87     words[i] = "#%d" % serial
  88
  89 def decode_type(words, i):
  90     qt = words[i]
  91     words[i] = query_type.get(int(qt), qt)
  92
  93 def handle_dnscache_log(line, match):
  94     (timestamp, event, data) = match.groups()
  95
  96     words = data.split()
  97     if event == "cached":
  98         if words[0] not in ("cname", "ns", "nxdomain"):
  99             decode_type(words, 0)
 100
 101     elif event == "drop":
 102         decode_serial(words, 0)
 103
 104     elif event == "lame":
 105         decode_ip(words, 0)
 106
 107     elif event == "nodata":
 108         decode_ip(words, 0)
 109         decode_ttl(words, 1)
 110         decode_type(words, 2)
 111
 112     elif event == "nxdomain":
 113         decode_ip(words, 0)
 114         decode_ttl(words, 1)
 115
 116     elif event == "query":
 117         decode_serial(words, 0)
 118         decode_client(words, 1)
 119         decode_type(words, 2)
 120
 121     elif event == "rr":
 122         decode_ip(words, 0)
 123         decode_ttl(words, 1)
 124         if words[2] not in ("cname", "mx", "ns", "ptr", "soa"):
 125             decode_type(words, 2)
 126             if words[2] == "a":         # decode answer to an A query
 127                 decode_ip(words, 4)
 128             if words[2] == "txt":       # text record
 129                 response = words[4]
 130                 if response.endswith("..."):
 131                     ellipsis = "..."
 132                     response = response[0:-3]
 133                 else:
 134                     ellipsis = ""
 135                 length = int(response[0:2], 16)
 136                 chars = []
 137                 for i in range(1, len(response)/2):
 138                     chars.append(chr(int(response[2*i : (2*i)+2], 16)))
 139                 words[4] = "%d:\"%s%s\"" % (length, "".join(chars), ellipsis)
 140
 141     elif event == "sent":
 142         decode_serial(words, 0)
 143
 144     elif event == "stats":
 145         words[0] = "count=%s" % words[0]
 146         words[1] = "motion=%s" % words[1]
 147         words[2] = "udp-active=%s" % words[2]
 148         words[3] = "tcp-active=%s" % words[3]
 149
 150     elif event == "tx":
 151         words[0] = "g=%s" % words[0]
 152         decode_type(words, 1)
 153         # words[2] = name
 154         # words[3] = control (domain for which these servers are believed
 155         #            to be authoritative)
 156         for i in range(4, len(words)):
 157             decode_ip(words, i)
 158
 159     elif event in ("tcpopen", "tcpclose"):
 160         decode_client(words, 0)
 161
 162     print(timestamp, event, " ".join(words))
 163
 164
 165 def handle_tinydns_log(line, match):
 166     (timestamp, ip, port, id, code, type, name) = match.groups()
 167     ip = convert_ip(ip)
 168     port = int(port, 16)
 169     id = int(id, 16)
 170     type = int(type, 16)                # "001c" -> 28
 171     type = query_type.get(type, type)   # 28 -> "aaaa"
 172
 173     print(timestamp, end=' ')
 174
 175     if code == "+":
 176         print ("sent response to %s:%s (id %s): %s %s"
 177                % (ip, port, id, type, name))
 178     elif code in ("-", "I", "C"):
 179         reason = query_drop_reason[code]
 180         print ("dropped query (%s) from %s:%s (id %s): %s %s"
 181                % (reason, ip, port, id, type, name))
 182     elif code == "/":
 183         print ("dropped query (couldn't parse) from %s:%s"
 184                % (ip, port))
 185     else:
 186         print ("%s from %s:%s (id %s): %s %s"
 187                % (code, ip, port, id, type, name))
 188
 189
 190 def parse_logfile(file):
 191     # Open pipe to tai64nlocal: we will write lines of our input (the
 192     # raw log file) to it, and read log lines with readable timestamps
 193     # from it.
 194     tai = Popen(["tai64nlocal"], stdin=PIPE, stdout=PIPE, text=True, bufsize=0)
 195
 196     for line in file:
 197         tai.stdin.write(line)
 198         line = tai.stdout.readline()
 199
 200         match = tinydns_log_re.match(line)
 201         if match:
 202             handle_tinydns_log(line, match)
 203             continue
 204
 205         match = dnscache_log_re.match(line)
 206         if match:
 207             handle_dnscache_log(line, match)
 208             continue
 209
 210         print(line)
 211
 212 def main():
 213     # Create an argument parser using the file's docsctring as its
 214     # description.
 215     from argparse import ArgumentParser, FileType
 216     parser = ArgumentParser(description = __doc__)
 217
 218     # Parse zero or more positional arguments into a list of
 219     # "logfiles". If none are given, read from stdin instead.
 220     from sys import stdin
 221     parser.add_argument("logfiles",
 222                         metavar="LOGFILE",
 223                         type=FileType("r"),
 224                         nargs="*",
 225                         default=[stdin],
 226                         help="djbdns logfile to process (default: stdin)")
 227
 228     args = parser.parse_args()
 229     for f in args.logfiles:
 230         parse_logfile(f)
 231
 232
 233
 234
 235 if __name__ == "__main__":
 236     main()