Convert tinydns and dnscache logs to human-readable form
"""
-import re
+import re, typing
from struct import pack
from time import strftime, gmtime
from subprocess import Popen, PIPE
-# common components of line-matching regexes
-timestamp_pat = r'[\d-]+ [\d:\.]+' # output of tai64nlocal
+## Regular expressions for matching tinydns/dnscache log lines. We
+## compile these once here rather than within the corresponding
+## matching functions, because the latter get executed repeatedly.
+
+# This first pattern is used to match the timestamp format that the
+# tai64nlocal program produces. It appears in both dnscache and
+# tinydns lines, after they've been piped through tai64nlocal, of
+# course.
+timestamp_pat = r'[\d-]+ [\d:\.]+'
+
+# The regex to match dnscache log lines.
+dnscache_log_re = re.compile(fr'({timestamp_pat}) (\w+)(.*)')
+
+# The "hex4" pattern matches a string of four hexadecimal digits. This
+# is used, for example, by tinydns to encode the query type
+# identifier.
hex4_pat = r'[0-9a-f]{4}'
-ip_pat = r'[0-9a-f]{8,32}' # IPv4 or IPv6 addresses in hex
-# discriminate between dnscache and tinydns log lines
-tinydns_log_re = re.compile(
- r'(%s) (%s):(%s):(%s) ([\+\-IC/]) (%s) (.*)'
- % (timestamp_pat, ip_pat, hex4_pat, hex4_pat, hex4_pat))
-dnscache_log_re = re.compile(r'(%s) (\w+)(.*)' % timestamp_pat)
+# The IP pattern matches a string of either 8 or 32 hexadecimal
+# characters, which correspond to IPv4 and IPv6 addresses,
+# respectively, in tinydns logs.
+ip_pat = r'[0-9a-f]{8,32}'
+# The regex to match tinydns log lines.
+tinydns_log_re = re.compile(
+ rf'({timestamp_pat}) ({ip_pat}):({hex4_pat}):({hex4_pat}) ([\+\-IC/]) ({hex4_pat}) (.*)'
+)
+
+# A dictionary mapping query type identifiers, in decimal, to their
+# friendly names for tinydns. Reference:
+#
+# https://en.wikipedia.org/wiki/List_of_DNS_record_types
+#
+# Note that mapping here is non-exhaustive, and that tinydns will
+# log responses for record types that it does not know about.
query_type = {
1: "a",
2: "ns",
24: "sig",
25: "key",
28: "aaaa",
+ 33: "srv",
+ 35: "naptr",
38: "a6",
+ 48: "dnskey",
+ 52: "tlsa",
+ 65: "https",
252: "axfr",
255: "any",
+ 257: "caa"
}
-# for tinydns only
+# tinydns can drop a query for one of three reasons; this dictionary
+# maps the symbol that gets logged in each case to a human-readable
+# reason. We include the "+" case here, indicating that the query was
+# NOT dropped, to avoid a special case later on when we're formatting
+# the human-readable output.
query_drop_reason = {
+ "+": None,
"-": "no authority",
"I": "invalid query",
"C": "invalid class",
- }
+ "/": "couldn't parse"
+}
def convert_ip(ip : str):
"""
if len(ip) == 8:
- # IPv4, eg. "7f000001" -> "127.0.0.1"
+ # IPv4, eg. "7f000001" -> "7f 00 00 01" -> "127.0.0.1"
return "%d.%d.%d.%d" % tuple(pack(">L", int(ip, 16)))
elif len(ip) == 32:
- # IPv6 is actually simpler -- it's just a string-slicing operation,
- # eg. "00000000000000000000ffff7f000001" ->
- # "0000:0000:0000:0000:0000:ffff:7f00:0001"
+ # IPv6 is actually simpler -- it's just a string-slicing operation.
return ":".join([ip[(4*i) : (4*i+4)] for i in range(8)])
-def _cvt_ip(match):
- return convert_ip(match.group(1))
-
-def _cvt_port(match):
- return ":" + str(int(match.group(1), 16))
-
def decode_client(words, i):
chunks = words[i].split(":")
if len(chunks) == 2: # ip:port
print(timestamp, event, " ".join(words))
-def handle_tinydns_log(line, match):
+def handle_tinydns_log(line : str, match: re.Match):
+ """
+ Handle a line that matched the ``tinydns_log_re`` regex.
+
+ Parameters
+ ----------
+
+ line : string
+ The tinydns log line that matched ``tinydns_log_re``.
+
+ match : re.Match
+ The match object that was returned when ``line`` was
+ tested against ``tinydns_log_re``.
+
+ Examples
+ --------
+
+ >>> line = "2022-09-14 21:04:40.206516500 7f000001:9d61:be69 - 0001 www.example.com"
+ >>> match = tinydns_log_re.match(line)
+ >>> handle_tinydns_log(line, match)
+ 2022-09-14 21:04:40.206516500 dropped query (no authority) from 127.0.0.1:40289 (id 48745): a www.example.com
+
+ """
(timestamp, ip, port, id, code, type, name) = match.groups()
ip = convert_ip(ip)
port = int(port, 16)
id = int(id, 16)
+
+ # Convert the "type" field to a human-readable record type name
+ # using the query_type dictionary. If the right name isn't present
+ # in the dictionary, we use the (decimal) type id instead.
type = int(type, 16) # "001c" -> 28
type = query_type.get(type, type) # 28 -> "aaaa"
print(timestamp, end=' ')
+ reason = query_drop_reason[code]
if code == "+":
- print ("sent response to %s:%s (id %s): %s %s"
- % (ip, port, id, type, name))
- elif code in ("-", "I", "C"):
- reason = query_drop_reason[code]
- print ("dropped query (%s) from %s:%s (id %s): %s %s"
- % (reason, ip, port, id, type, name))
- elif code == "/":
- print ("dropped query (couldn't parse) from %s:%s"
- % (ip, port))
+ line_tpl = "sent response to {ip}:{port} (id {id}): {type} {name}"
else:
- print ("%s from %s:%s (id %s): %s %s"
- % (code, ip, port, id, type, name))
+ line_tpl = "dropped query ({reason}) from {ip}:{port}"
+ if code != "/":
+ # If the query can actually be parsed, the log line is a
+ # bit more informative than it would have been otherwise.
+ line_tpl += " (id {id}): {type} {name}"
+
+ print(line_tpl.format(reason=reason,
+ ip=ip,
+ port=port,
+ id=id,
+ type=type,
+ name=name))
+
+
+def parse_logfile(file : typing.TextIO):
+ r"""
+ Process a single log ``file``.
+
+ Parameters
+ ----------
+ file : typing.TextIO
+ An open log file, or stdin.
-def parse_logfile(file):
+ Examples
+ --------
+
+ >>> line = "@4000000063227a320c4f3114 7f000001:9d61:be69 - 0001 www.example.com\n"
+ >>> from tempfile import NamedTemporaryFile
+ >>> with NamedTemporaryFile(mode="w", delete=False) as f:
+ ... _ = f.write(line)
+ >>> f = open(f.name, 'r')
+ >>> parse_logfile(f)
+ 2022-09-14 21:04:40.206516500 dropped query (no authority) from 127.0.0.1:40289 (id 48745): a www.example.com
+ >>> f.close()
+ >>> from os import remove
+ >>> remove(f.name)
+
+ """
# Open pipe to tai64nlocal: we will write lines of our input (the
# raw log file) to it, and read log lines with readable timestamps
# from it.
default=[stdin],
help="djbdns logfile to process (default: stdin)")
+ # Warning: argparse automatically opens its file arguments here,
+ # and they only get closed when the program terminates. There's no
+ # real benefit to closing them one-at-a-time after calling
+ # parse_logfile(), because the "scarce" resource of open file
+ # descriptors gets consumed immediately, before any processing has
+ # happened. In other words, if you're going to run out of file
+ # descriptors, it's going to happen right now.
+ #
+ # So anyway, don't run this on several million logfiles.
args = parser.parse_args()
for f in args.logfiles:
parse_logfile(f)
-
-
if __name__ == "__main__":
main()