From 855a443bce100c00e4753780957ad02a78a8b8f4 Mon Sep 17 00:00:00 2001
From: Michael Orlitzky <michael@orlitzky.com>
Date: Thu, 15 Sep 2022 09:30:18 -0400
Subject: [PATCH] bin/djbdns-logparse.py: use f-strings for regex patterns; add
 comments.

---
 bin/djbdns-logparse.py | 46 +++++++++++++++++++++++++++++++++---------
 1 file changed, 36 insertions(+), 10 deletions(-)

diff --git a/bin/djbdns-logparse.py b/bin/djbdns-logparse.py
index ed85381..8380124 100755
--- a/bin/djbdns-logparse.py
+++ b/bin/djbdns-logparse.py
@@ -9,17 +9,41 @@ 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",
@@ -44,12 +68,14 @@ query_type = {
     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.
 query_drop_reason = {
     "-": "no authority",
     "I": "invalid query",
     "C": "invalid class",
-    }
+}
 
 
 def convert_ip(ip : str):
-- 
2.44.2