Merge pull request #227 from windgmbh/add-new-tlds

Add support for new Top Level Domains

Author: MooCow

tests/ok-domains.txt

@@ -49,6 +49,7 @@
     google.eu
     google.fi
     google.fm
+    google.fo
     google.fr
     google.game
     google.global
@@ -61,6 +62,8 @@
     google.is
     google.it
     google.jp
+    google.kg
+    google.kn
     google.link
     google.lt
     google.me
@@ -88,6 +91,7 @@
     google.tel
     google.theatre
     google.tickets
+    google.tokyo
     google.trade
     google.tv
     google.us

whois/_3_adjust.py

@@ -159,7 +159,8 @@ def __init__(
     "%Y%m%d%H%M%S",  # 20071224102432 used in edu_ua
     "%Y-%m-%d %H:%M:%S (%Z%z)",  # .tw uses (UTC+8) but we need (UTC+0800) for %z match
     "%d %B %Y at %H:%M:%S.%f",  # 07 january 2020 at 23:38:30.772
-    "%Y-%m-%d %H:%M:%S.%f %Z",  # 2022-09-18 22:38:18.0 UTC (sn Senegal)
+    "%Y-%m-%d %H:%M:%S.%f %Z",  # 2022-09-18 22:38:18.0 UTC (sn Senegal),
+    "%a %b %d %H:%M:%S %Y",  # Thu Oct 21 05:54:20 2032 (kg Kyrgyzstan)
 ]
 
 CUSTOM_DATE_FORMATS = {
@@ -195,6 +196,9 @@ def str_to_date(text: str, tld: Optional[str] = None) -> Optional[datetime.datet
     # better here https://stackoverflow.com/questions/1258199/python-datetime-strptime-wildcard
     text = re.sub(r"(\d+)(st|nd|rd|th) ", r"\1 ", text)
 
+    # Remove consecutive whitespace
+    text = re.sub(r"\s\s+", r" ", text)
+
     # 07 january 2020 at 23:38:30.772
     # %d %B %Y at %H:%M %S.%f
     if tld and tld in CUSTOM_DATE_FORMATS:

whois/tld_regexpr.py

@@ -404,6 +404,11 @@
     "extend": "com",
 }
 
+fo = {
+    "extend": "com",
+    "registrant": None,
+}
+
 fr = {
     "extend": "com",
     "domain_name": r"domain:\s?(.+)",
@@ -586,6 +591,23 @@
     "extend": "com",
 }
 
+kg = {
+    "extend": None,
+    "domain_name": r"Domain\s(.+)\s\(",
+    "registrar": r"Billing\sContact:\n.*\n\s+Name:\s(.+)\n",
+    "registrant_country": None,
+    "expiration_date": r"Record expires on:\s+(.+)",
+    "creation_date": r"Record created:\s+(.+)",
+    "updated_date": r"Record last updated on:\s+(.+)",
+    "name_servers": None,
+    "status": None,
+}
+
+# Saint Kitts and Nevis
+kn = {
+    "extend": "com",
+}
+
 kr = {
     "extend": "com",
     "domain_name": r"Domain Name\s*:\s?(.+)",
@@ -964,6 +986,10 @@
     "emails": r"[\w.-]+@[\w.-]+\.[\w]{2,4}",
 }
 
+tokyo = {
+    "extend": "com",
+}
+
 top = {
     "extend": "com",
 }