New code to retrieve the database, more robust

check_expire

@@ -54,8 +54,8 @@ def usage(msg=None):
 
 def details():
     if verbose:
-        print(" RDAP database \"%s\", version %s published on %s, RDAP server is %s" % \
+        print(" RDAP database \"%s\", version %s published on %s, retrieved on %s, RDAP server is %s" % \
-              (database.description, database.version, database.publication, server))
+              (database.description, database.version, database.publication, database.retrieved, server))
     else:
         print("")
 

ianardap.py

@@ -12,12 +12,14 @@ import datetime
 import json
 import os
 import sys
+import time
 import fcntl
 
 IANABASE = "https://data.iana.org/rdap/dns.json"
 CACHE = os.environ["HOME"] + "/.ianardapcache.json" 
 MAXAGE = 24 # Hours
 IANATIMEOUT = 10 # Seconds
+MAXTESTS = 3 # Maximum attempts to get the database
 
 class IanaRDAPDatabase():
 
@@ -26,22 +28,49 @@ class IanaRDAPDatabase():
         cache_valid = False
         self.cachefile = cachefile
         self.lockname = self.cachefile + ".lock"
-        self.lock()
+        loaded = False
-        if os.path.exists(cachefile) and \
+        tests = 0
-           datetime.datetime.fromtimestamp(os.path.getmtime(cachefile)) >= \
+        errmsg = "No error"
-           (datetime.datetime.utcnow() - datetime.timedelta(hours = maxage)):
+        while not loaded and tests < MAXTESTS:
-            cache = open(cachefile, "rb")
+            self.lock()
-            content = cache.read()
+            if os.path.exists(cachefile) and \
-            cache.close()
+               datetime.datetime.fromtimestamp(os.path.getmtime(cachefile)) > \
-            cache_valid = True
+               (datetime.datetime.now() - datetime.timedelta(hours = maxage)):
-            self.unlock()
+                cache = open(cachefile, "rb")
-        else:
+                content = cache.read()
-            self.unlock()
+                cache.close()
-            response = requests.get(IANABASE, timeout=IANATIMEOUT)
+                self.unlock()
-            if response.status_code != 200:
+                try:
-                raise Exception("Invalid HTTPS return code when trying to get %s: %s" % (IANABASE, response.status_code))
+                    database = json.loads(content)
-            content = response.content
+                    loaded = True
-        database = json.loads(content)
+                    self.retrieved = datetime.datetime.fromtimestamp(os.path.getmtime(cachefile))
+                    cache_valid = True
+                except json.decoder.JSONDecodeError:
+                    tests += 1
+                    errmsg = "Invalid JSON content in %s" % cachefile
+                    # Delete it without mercy
+                    os.remove(cachefile)
+                    continue
+            else:
+                self.unlock()
+                response = requests.get(IANABASE, timeout=IANATIMEOUT)
+                if response.status_code != 200:
+                    time.sleep(2)
+                    tests += 1
+                    errmsg = "Invalid HTTPS return code when trying to get %s: %s" % (IANABASE, response.status_code)
+                    continue
+                else:
+                    loaded = True
+                    self.retrieved = datetime.datetime.now()
+                try:
+                    content = response.content
+                    database = json.loads(content)
+                except json.decoder.JSONDecodeError:
+                    tests += 1
+                    errmsg = "Invalid JSON retrieved from %s" % IANABASE
+                    continue
+        if not loaded:
+            raise Exception("Cannot read IANA database: %s" % errmsg)
         self.description = database["description"]
         self.publication = database["publication"]
         self.version = database["version"]
@@ -78,8 +107,8 @@ class IanaRDAPDatabase():
 
 if __name__ == "__main__":
     rdap = IanaRDAPDatabase(maxage=1)
-    print("Database \"%s\", version %s published on %s, %i services" % \
+    print("Database \"%s\", version %s published on %s, retrieved on %s, %i services" % \
-          (rdap.description, rdap.version, rdap.publication, len(rdap.services)))
+          (rdap.description, rdap.version, rdap.publication, rdap.retrieved, len(rdap.services)))
     for domain in sys.argv[1:]:
         print("%s -> %s" % (domain, rdap.find(domain)))
 

test_ianardap.py

@@ -0,0 +1,43 @@
+#!/usr/bin/env python3
+
+"""Test by executing 'pytest' https://docs.pytest.org/
+
+Don't forget to set PYTHONPATH if you want to test the development
+tree and not the installed package.  """
+
+import ianardap
+
+import tempfile
+import os.path
+import datetime
+
+def test_basic():
+    database = ianardap.IanaRDAPDatabase()
+    assert database.description.startswith("RDAP bootstrap file") and database.version == "1.0" and \
+        len(database.services) > 1000
+    
+def test_alternative_cache():
+    tmpfile = tempfile.NamedTemporaryFile(suffix=".testianacache", delete=False)
+    database = ianardap.IanaRDAPDatabase(cachefile=tmpfile.name, maxage=0)
+    assert os.path.exists(tmpfile.name) and \
+        datetime.datetime.fromtimestamp(os.path.getmtime(tmpfile.name)) > \
+        (datetime.datetime.now() - datetime.timedelta(minutes=1))
+    os.remove(tmpfile.name)
+    os.remove(tmpfile.name + ".lock")
+
+def test_refresh():
+    # Force a resfresh
+    database = ianardap.IanaRDAPDatabase(maxage=0)
+    assert (database.retrieved > (datetime.datetime.now() - datetime.timedelta(minutes=1))) and \
+        (datetime.datetime.fromtimestamp(os.path.getmtime(database.cachefile)) > \
+        (datetime.datetime.now() - datetime.timedelta(minutes=1)))
+
+def test_find_exists():
+    database = ianardap.IanaRDAPDatabase()
+    server = database.find("www.foobar.ar")
+    assert server == "https://rdap.nic.ar/"
+
+def test_find_not_exists():
+    database = ianardap.IanaRDAPDatabase()
+    server = database.find("www.foobar.example")
+    assert server is None