Read and apply IANA's caching instructions. Closes #3

ianardap.py

@@ -1,7 +1,7 @@
 #!/usr/bin/env python3
 
 """A simple module to get the RDAP server for a given domain name,
-from the IANA database specified in RFC 7484.
+from the IANA database specified in RFC 9224.
 
 """
 
@@ -15,13 +15,35 @@ import sys
 import time
 import fcntl
 import pickle
+import pathlib
 
 IANABASE = "https://data.iana.org/rdap/dns.json"
 CACHE = os.environ["HOME"] + "/.ianardapcache" 
-MAXAGE = 24 # Hours
+MAXAGE = 24 # Hours. Used only if the server no longer gives the information.
 IANATIMEOUT = 10 # Seconds
 MAXTESTS = 3 # Maximum attempts to get the database
 
+# Don't touch
+HTTP_DATE_FORMAT = "%a, %d %b %Y %H:%M:%S %Z"
+
+# RFC 9111, section 5.2
+def parse_cachecontrol(h):
+    result = {}
+    directives = h.split(",")
+    for directive in directives:
+        directive = directive.strip()
+        if "=" in directive:
+            (key, value) = directive.split("=")
+        else:
+            key = directive
+            value = None
+        result[key.lower()] = value
+    return result
+
+def parse_expires(h):
+    d = datetime.datetime.strptime(h, HTTP_DATE_FORMAT)
+    return d
+
 class IanaRDAPDatabase():
 
     def __init__(self, maxage=MAXAGE, cachefile=CACHE, pickleformat=False):
@@ -36,14 +58,16 @@ write in the file (see the documentation of the module)."""
         else:
             self.cachefile = cachefile + ".json"
         self.lockname = self.cachefile + ".lock"
+        self.expirationfile = self.cachefile + ".expires"
         loaded = False
         tests = 0
         errmsg = "No error"
         while not loaded and tests < MAXTESTS:
             self.lock()
             if os.path.exists(self.cachefile) and \
-               datetime.datetime.fromtimestamp(os.path.getmtime(self.cachefile)) > \
+               (pathlib.Path(self.expirationfile).exists() and \
-               (datetime.datetime.now() - datetime.timedelta(hours = maxage)):
+                datetime.datetime.fromtimestamp(os.path.getmtime(self.expirationfile)) > \
+                datetime.datetime.now()):
                 cache = open(self.cachefile, "rb")
                 content = cache.read()
                 cache.close()
@@ -75,6 +99,18 @@ write in the file (see the documentation of the module)."""
             else:
                 self.unlock()
                 response = requests.get(IANABASE, timeout=IANATIMEOUT)
+                expirationtime = None
+                if "cache-control" in response.headers:
+                    directives = parse_cachecontrol(response.headers["cache-control"])
+                    if "max-age" in directives:
+                        maxage = int(directives["max-age"])
+                        expirationtime = datetime.datetime.now() + datetime.timedelta(seconds=maxage)
+                if not expirationtime:
+                    if "expires" in response.headers:
+                        expirationtime = parse_expires(response.headers["expires"])
+                    else:
+                        expirationtime = datetime.datetime.now() + datetime.timedelta(hours=MAXAGE)
+                self.expirationtime = time.mktime(expirationtime.timetuple())
                 if response.status_code != 200:
                     time.sleep(2)
                     tests += 1
@@ -86,6 +122,9 @@ write in the file (see the documentation of the module)."""
                 try:
                     content = response.content
                     database = json.loads(content)
+                    with open(self.expirationfile, 'w'):
+                        os.utime(self.expirationfile,
+                                 times = (self.expirationtime, self.expirationtime))
                 except json.decoder.JSONDecodeError:
                     tests += 1
                     errmsg = "Invalid JSON retrieved from %s" % IANABASE