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 # 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 : str):
  50     """
  51     Convert a hex string representing an IP address to conventional
  52     human-readable form, ie. dotted-quad decimal for IPv4, and
  53     8 colon-separated hex shorts for IPv6.
  54
  55     Examples
  56     --------
  57
  58         >>> convert_ip("7f000001")
  59         '127.0.0.1'
  60         >>> convert_ip("00000000000000000000ffff7f000001")
  61         '0000:0000:0000:0000:0000:ffff:7f00:0001'
  62
  63     """
  64     if len(ip) == 8:
  65         # IPv4, eg. "7f000001" -> "127.0.0.1"
  66         return "%d.%d.%d.%d" % tuple(pack(">L", int(ip, 16)))
  67     elif len(ip) == 32:
  68         # IPv6 is actually simpler -- it's just a string-slicing operation,
  69         # eg. "00000000000000000000ffff7f000001" ->
  70         # "0000:0000:0000:0000:0000:ffff:7f00:0001"
  71         return ":".join([ip[(4*i) : (4*i+4)] for i in range(8)])
  72
  73
  74 def _cvt_ip(match):
  75     return convert_ip(match.group(1))
  76
  77 def _cvt_port(match):
  78     return ":" + str(int(match.group(1), 16))
  79
  80 def decode_client(words, i):
  81     chunks = words[i].split(":")
  82     if len(chunks) == 2:                # ip:port
  83         words[i] = "%s:%d" % (convert_ip(chunks[0]), int(chunks[1], 16))
  84     elif len(chunks) == 3:
  85         words[i] = "%s:%d (id %d)" % (convert_ip(chunks[0]),
  86                                       int(chunks[1], 16),
  87                                       int(chunks[2], 16))
  88
  89 def decode_ip(words, i):
  90     words[i] = convert_ip(words[i])
  91
  92 def decode_ttl(words, i):
  93     words[i] = "TTL=%s" % words[i]
  94
  95 def decode_serial(words, i):
  96     serial = int(words[i])
  97     words[i] = "#%d" % serial
  98
  99 def decode_type(words, i):
 100     qt = words[i]
 101     words[i] = query_type.get(int(qt), qt)
 102
 103 def handle_dnscache_log(line, match):
 104     (timestamp, event, data) = match.groups()
 105
 106     words = data.split()
 107     if event == "cached":
 108         if words[0] not in ("cname", "ns", "nxdomain"):
 109             decode_type(words, 0)
 110
 111     elif event == "drop":
 112         decode_serial(words, 0)
 113
 114     elif event == "lame":
 115         decode_ip(words, 0)
 116
 117     elif event == "nodata":
 118         decode_ip(words, 0)
 119         decode_ttl(words, 1)
 120         decode_type(words, 2)
 121
 122     elif event == "nxdomain":
 123         decode_ip(words, 0)
 124         decode_ttl(words, 1)
 125
 126     elif event == "query":
 127         decode_serial(words, 0)
 128         decode_client(words, 1)
 129         decode_type(words, 2)
 130
 131     elif event == "rr":
 132         decode_ip(words, 0)
 133         decode_ttl(words, 1)
 134         if words[2] not in ("cname", "mx", "ns", "ptr", "soa"):
 135             decode_type(words, 2)
 136             if words[2] == "a":         # decode answer to an A query
 137                 decode_ip(words, 4)
 138             if words[2] == "txt":       # text record
 139                 response = words[4]
 140                 if response.endswith("..."):
 141                     ellipsis = "..."
 142                     response = response[0:-3]
 143                 else:
 144                     ellipsis = ""
 145                 length = int(response[0:2], 16)
 146                 chars = []
 147                 for i in range(1, len(response)/2):
 148                     chars.append(chr(int(response[2*i : (2*i)+2], 16)))
 149                 words[4] = "%d:\"%s%s\"" % (length, "".join(chars), ellipsis)
 150
 151     elif event == "sent":
 152         decode_serial(words, 0)
 153
 154     elif event == "stats":
 155         words[0] = "count=%s" % words[0]
 156         words[1] = "motion=%s" % words[1]
 157         words[2] = "udp-active=%s" % words[2]
 158         words[3] = "tcp-active=%s" % words[3]
 159
 160     elif event == "tx":
 161         words[0] = "g=%s" % words[0]
 162         decode_type(words, 1)
 163         # words[2] = name
 164         # words[3] = control (domain for which these servers are believed
 165         #            to be authoritative)
 166         for i in range(4, len(words)):
 167             decode_ip(words, i)
 168
 169     elif event in ("tcpopen", "tcpclose"):
 170         decode_client(words, 0)
 171
 172     print(timestamp, event, " ".join(words))
 173
 174
 175 def handle_tinydns_log(line, match):
 176     (timestamp, ip, port, id, code, type, name) = match.groups()
 177     ip = convert_ip(ip)
 178     port = int(port, 16)
 179     id = int(id, 16)
 180     type = int(type, 16)                # "001c" -> 28
 181     type = query_type.get(type, type)   # 28 -> "aaaa"
 182
 183     print(timestamp, end=' ')
 184
 185     if code == "+":
 186         print ("sent response to %s:%s (id %s): %s %s"
 187                % (ip, port, id, type, name))
 188     elif code in ("-", "I", "C"):
 189         reason = query_drop_reason[code]
 190         print ("dropped query (%s) from %s:%s (id %s): %s %s"
 191                % (reason, ip, port, id, type, name))
 192     elif code == "/":
 193         print ("dropped query (couldn't parse) from %s:%s"
 194                % (ip, port))
 195     else:
 196         print ("%s from %s:%s (id %s): %s %s"
 197                % (code, ip, port, id, type, name))
 198
 199
 200 def parse_logfile(file):
 201     # Open pipe to tai64nlocal: we will write lines of our input (the
 202     # raw log file) to it, and read log lines with readable timestamps
 203     # from it.
 204     tai = Popen(["tai64nlocal"], stdin=PIPE, stdout=PIPE, text=True, bufsize=0)
 205
 206     for line in file:
 207         tai.stdin.write(line)
 208         line = tai.stdout.readline()
 209
 210         match = tinydns_log_re.match(line)
 211         if match:
 212             handle_tinydns_log(line, match)
 213             continue
 214
 215         match = dnscache_log_re.match(line)
 216         if match:
 217             handle_dnscache_log(line, match)
 218             continue
 219
 220         print(line)
 221
 222 def main():
 223     # Create an argument parser using the file's docsctring as its
 224     # description.
 225     from argparse import ArgumentParser, FileType
 226     parser = ArgumentParser(description = __doc__)
 227
 228     # Parse zero or more positional arguments into a list of
 229     # "logfiles". If none are given, read from stdin instead.
 230     from sys import stdin
 231     parser.add_argument("logfiles",
 232                         metavar="LOGFILE",
 233                         type=FileType("r"),
 234                         nargs="*",
 235                         default=[stdin],
 236                         help="djbdns logfile to process (default: stdin)")
 237
 238     args = parser.parse_args()
 239     for f in args.logfiles:
 240         parse_logfile(f)
 241
 242
 243
 244
 245 if __name__ == "__main__":
 246     main()