update

idn_mappings.py

@@ -0,0 +1,119 @@
+#!/usr/bin/env python3
+"""
+Script to map Punycode TLDs to their Unicode representations
+"""
+
+import idna
+
+# Known IDN TLD mappings
+IDN_MAPPINGS = {
+    'xn--p1ai': '.рф',      # Russia
+    'xn--fiqs8s': '.中国',   # China 
+    'xn--fiqz9s': '.中國',   # China (traditional)
+    'xn--lgbbat1ad8j': '.الجزائر', # Algeria
+    'xn--yfro4i67o': '.קום', # Israel (KOM)
+    'xn--4gbrim': '.مصر',   # Egypt
+    'xn--55qx5d': '.موريتانيا', # Mauritania
+    'xn--80akhbyknj4f': '.հայ', # Armenia
+    'xn--80asehdb': '.бел', # Belarus
+    'xn--90a3ac': '.мкд',   # Macedonia
+    'xn--45brj9c': '.бг',   # Bulgaria
+    'xn--p1ai': '.рф',      # Russia (duplicate)
+    'xn--hlcj6aya': '.سوريا', # Syria
+    'xn--mgbcpq6gpa1a': '.السعودية', # Saudi Arabia
+    'xn--ogbpf8fl': '.سودان', # Sudan
+    'xn--kprw13d': '.გე',   # Georgia
+    'xn--kpry57d': '.გე',   # Georgia (alternative)
+    'xn--o1ac': '.ελ',      # Greece
+    'xn--80ao21a': '.қаз',  # Kazakhstan
+    'xn--fgbp6a': '.مغرب',  # Morocco
+    'xn--j1amh': '.укр',    # Ukraine
+    'xn--mix891f': '.ไทย',  # Thailand
+    'xn--mix082f': '.ไทย',  # Thailand (alternative)
+    'xn--mxtq1m': '.新加坡', # Singapore
+    'xn--node': '.नेट',    # India (NET)
+    'xn--j6w193g': '.香港', # Hong Kong
+    'xn--55qw42g': '.中国',  # China (alternative)
+    'xn--5tzm5g': '.台灣',  # Taiwan
+    'xn--6frz82g': '.ලංකා', # Sri Lanka
+    'xn--80adxhks': '.мкд', # Macedonia (alternative)
+    'xn--l1acc': '.мон',    # Mongolia
+    'xn--9t4b11yi5a': '.இலங்கை', # Sri Lanka (alternative)
+    'xn--rhqv96g': '.世博',  # Expo
+    'xn--0zwm56d': '.澳洲',  # Australia
+    'xn--czru2d': '.कोम',   # India (COM)
+    'xn--d1acj3b': '.дети',  # Kids
+    'xn--d1alf': '.москва',  # Moscow
+    'xn--h2brj9c': '.срб',   # Serbia
+    'xn--h2breg3eve': '.срб', # Serbia (alternative)
+    'xn--k1x57d': '.新加坡', # Singapore (alternative)
+    'xn--mgbbh1a71e': '.امارات', # UAE
+    'xn--mgbaam7a8h': '.الاردن', # Jordan
+    'xn--mgbayh7gpa': '.الاردن', # Jordan (alternative)
+    'xn--y9a3aq': '.հայ',   # Armenia (alternative)
+    'xn--mgbx4cd0ab': '.مليسيا', # Malaysia
+    'xn--54b7fta0cc': '.بھارت', # India
+    'xn--90ae5b': '.بازار',  # Iran (Bazaar)
+    'xn--l1nej': '.موقع',   # Iran (Site)
+    'xn--mgbgu82a': '.شبكة', # Iran (Network)
+    'xn--fiq64b': '.कॉम',   # India (COM alternative)
+    'xn--kcrx77d1x4a': '.சிங்கப்பூர்', # Singapore (Tamil)
+    'xn--i1b6b1a6a2e': '.संगठन', # India (Organization)
+    'xn--nqv7f': '.فلسطين', # Palestine
+    'xn--qqh11a': '.مصر',   # Egypt (alternative)
+    'xn--c1avg': '.бел',    # Belarus (alternative)
+    'xn--e1a4c': '.ею',     # European Union
+    'xn--8h0a': '.ايران',   # Iran
+    'xn--1qqw23a': '.游戏',  # China (Game)
+    'xn--3bst00m': '.公司',  # China (Company)
+    'xn--45br5cyl': '.бг',   # Bulgaria (alternative)
+    'xn--s9brj9c': '.срб',   # Serbia (alternative)
+    'xn--czrs0t': '.कोम',   # India (COM alternative)
+    'xn--czr694b': '.कॉम',   # India (COM alternative)
+    'xn--gecrj9c': '.克罗地亚', # Croatia
+    'xn--p1ai': '.рф',      # Russia (duplicate)
+    'xn--9krt00a': '.日本',  # Japan
+    'xn--xkc2dl3a5ee0h': '.ಭಾರತ', # India (Kannada)
+    'xn--fzys8d69uvgm': '.تونس', # Tunisia
+    'xn--fzc2c9e2c': '.السعودية', # Saudi Arabia (alternative)
+}
+
+def punycode_to_unicode(punycode):
+    """Convert Punycode to Unicode representation"""
+    try:
+        if punycode.startswith('xn--'):
+            return idna.decode(punycode)
+        return punycode
+    except:
+        return IDN_MAPPINGS.get(punycode, punycode)
+
+def get_all_idn_tlds():
+    """Get all IDN TLDs with their Unicode representations"""
+    import requests
+    
+    response = requests.get('https://data.iana.org/TLD/tlds-alpha-by-domain.txt')
+    lines = response.text.strip().split('\n')
+    tlds = []
+    
+    for line in lines:
+        line = line.strip()
+        if line and not line.startswith('#'):
+            tlds.append(line.lower())
+    
+    idn_tlds = []
+    for tld in tlds:
+        if tld.startswith('xn--'):
+            unicode_form = punycode_to_unicode(tld)
+            idn_tlds.append({
+                'punycode': tld,
+                'unicode': unicode_form,
+                'display': f"{tld} ({unicode_form})"
+            })
+    
+    return idn_tlds
+
+if __name__ == "__main__":
+    idn_tlds = get_all_idn_tlds()
+    print(f"Found {len(idn_tlds)} IDN TLDs:")
+    for tld in idn_tlds[:20]:
+        print(f"  {tld['display']}")