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      33: "srv",
  37      35: "naptr",
  38      38: "a6",
  39      48: "dnskey",
  40      52: "tlsa",
  41      65: "https",
  42     252: "axfr",
  43     255: "any",
  44     257: "caa"
  45 }
  46
  47 # for tinydns only
  48 query_drop_reason = {
  49     "-": "no authority",
  50     "I": "invalid query",
  51     "C": "invalid class",
  52     }
  53
  54
  55 def convert_ip(ip : str):
  56     """
  57     Convert a hex string representing an IP address to conventional
  58     human-readable form, ie. dotted-quad decimal for IPv4, and
  59     8 colon-separated hex shorts for IPv6.
  60
  61     Examples
  62     --------
  63
  64         >>> convert_ip("7f000001")
  65         '127.0.0.1'
  66         >>> convert_ip("00000000000000000000ffff7f000001")
  67         '0000:0000:0000:0000:0000:ffff:7f00:0001'
  68
  69     """
  70     if len(ip) == 8:
  71         # IPv4, eg. "7f000001" -> "7f 00 00 01" -> "127.0.0.1"
  72         return "%d.%d.%d.%d" % tuple(pack(">L", int(ip, 16)))
  73     elif len(ip) == 32:
  74         # IPv6 is actually simpler -- it's just a string-slicing operation.
  75         return ":".join([ip[(4*i) : (4*i+4)] for i in range(8)])
  76
  77
  78 def _cvt_ip(match):
  79     return convert_ip(match.group(1))
  80
  81 def _cvt_port(match):
  82     return ":" + str(int(match.group(1), 16))
  83
  84 def decode_client(words, i):
  85     chunks = words[i].split(":")
  86     if len(chunks) == 2:                # ip:port
  87         words[i] = "%s:%d" % (convert_ip(chunks[0]), int(chunks[1], 16))
  88     elif len(chunks) == 3:
  89         words[i] = "%s:%d (id %d)" % (convert_ip(chunks[0]),
  90                                       int(chunks[1], 16),
  91                                       int(chunks[2], 16))
  92
  93 def decode_ip(words, i):
  94     words[i] = convert_ip(words[i])
  95
  96 def decode_ttl(words, i):
  97     words[i] = "TTL=%s" % words[i]
  98
  99 def decode_serial(words, i):
 100     serial = int(words[i])
 101     words[i] = "#%d" % serial
 102
 103 def decode_type(words, i):
 104     qt = words[i]
 105     words[i] = query_type.get(int(qt), qt)
 106
 107 def handle_dnscache_log(line, match):
 108     (timestamp, event, data) = match.groups()
 109
 110     words = data.split()
 111     if event == "cached":
 112         if words[0] not in ("cname", "ns", "nxdomain"):
 113             decode_type(words, 0)
 114
 115     elif event == "drop":
 116         decode_serial(words, 0)
 117
 118     elif event == "lame":
 119         decode_ip(words, 0)
 120
 121     elif event == "nodata":
 122         decode_ip(words, 0)
 123         decode_ttl(words, 1)
 124         decode_type(words, 2)
 125
 126     elif event == "nxdomain":
 127         decode_ip(words, 0)
 128         decode_ttl(words, 1)
 129
 130     elif event == "query":
 131         decode_serial(words, 0)
 132         decode_client(words, 1)
 133         decode_type(words, 2)
 134
 135     elif event == "rr":
 136         decode_ip(words, 0)
 137         decode_ttl(words, 1)
 138         if words[2] not in ("cname", "mx", "ns", "ptr", "soa"):
 139             decode_type(words, 2)
 140             if words[2] == "a":         # decode answer to an A query
 141                 decode_ip(words, 4)
 142             if words[2] == "txt":       # text record
 143                 response = words[4]
 144                 if response.endswith("..."):
 145                     ellipsis = "..."
 146                     response = response[0:-3]
 147                 else:
 148                     ellipsis = ""
 149                 length = int(response[0:2], 16)
 150                 chars = []
 151                 for i in range(1, len(response)/2):
 152                     chars.append(chr(int(response[2*i : (2*i)+2], 16)))
 153                 words[4] = "%d:\"%s%s\"" % (length, "".join(chars), ellipsis)
 154
 155     elif event == "sent":
 156         decode_serial(words, 0)
 157
 158     elif event == "stats":
 159         words[0] = "count=%s" % words[0]
 160         words[1] = "motion=%s" % words[1]
 161         words[2] = "udp-active=%s" % words[2]
 162         words[3] = "tcp-active=%s" % words[3]
 163
 164     elif event == "tx":
 165         words[0] = "g=%s" % words[0]
 166         decode_type(words, 1)
 167         # words[2] = name
 168         # words[3] = control (domain for which these servers are believed
 169         #            to be authoritative)
 170         for i in range(4, len(words)):
 171             decode_ip(words, i)
 172
 173     elif event in ("tcpopen", "tcpclose"):
 174         decode_client(words, 0)
 175
 176     print(timestamp, event, " ".join(words))
 177
 178
 179 def handle_tinydns_log(line, match):
 180     (timestamp, ip, port, id, code, type, name) = match.groups()
 181     ip = convert_ip(ip)
 182     port = int(port, 16)
 183     id = int(id, 16)
 184     type = int(type, 16)                # "001c" -> 28
 185     type = query_type.get(type, type)   # 28 -> "aaaa"
 186
 187     print(timestamp, end=' ')
 188
 189     if code == "+":
 190         print ("sent response to %s:%s (id %s): %s %s"
 191                % (ip, port, id, type, name))
 192     elif code in ("-", "I", "C"):
 193         reason = query_drop_reason[code]
 194         print ("dropped query (%s) from %s:%s (id %s): %s %s"
 195                % (reason, ip, port, id, type, name))
 196     elif code == "/":
 197         print ("dropped query (couldn't parse) from %s:%s"
 198                % (ip, port))
 199     else:
 200         print ("%s from %s:%s (id %s): %s %s"
 201                % (code, ip, port, id, type, name))
 202
 203
 204 def parse_logfile(file):
 205     # Open pipe to tai64nlocal: we will write lines of our input (the
 206     # raw log file) to it, and read log lines with readable timestamps
 207     # from it.
 208     tai = Popen(["tai64nlocal"], stdin=PIPE, stdout=PIPE, text=True, bufsize=0)
 209
 210     for line in file:
 211         tai.stdin.write(line)
 212         line = tai.stdout.readline()
 213
 214         match = tinydns_log_re.match(line)
 215         if match:
 216             handle_tinydns_log(line, match)
 217             continue
 218
 219         match = dnscache_log_re.match(line)
 220         if match:
 221             handle_dnscache_log(line, match)
 222             continue
 223
 224         print(line)
 225
 226 def main():
 227     # Create an argument parser using the file's docsctring as its
 228     # description.
 229     from argparse import ArgumentParser, FileType
 230     parser = ArgumentParser(description = __doc__)
 231
 232     # Parse zero or more positional arguments into a list of
 233     # "logfiles". If none are given, read from stdin instead.
 234     from sys import stdin
 235     parser.add_argument("logfiles",
 236                         metavar="LOGFILE",
 237                         type=FileType("r"),
 238                         nargs="*",
 239                         default=[stdin],
 240                         help="djbdns logfile to process (default: stdin)")
 241
 242     args = parser.parse_args()
 243     for f in args.logfiles:
 244         parse_logfile(f)
 245
 246
 247
 248
 249 if __name__ == "__main__":
 250     main()