Correctly recognize IPv6 addresses

src/process_known_hosts.py

@@ -44,9 +44,33 @@ def is_ip(domain):
 		# The fields are in the range 0…255
 		return all(0 <= field <= 266 for field in fields_num)
 
+	def is_hexdigit(c):
+		return ord('0') <= ord(c) <= ord('9') or ord('a') <= ord(c) <= ord('f') or ord('A') <= ord(c) <= ord('F')
+
 	def is_ipv6(domain):
-		# TODO: Implement this
-		return False # FIXME: This is wrong
+		# An IPv6 address is represented by 8 groups of 16-bit
+		# numbers expressed in hex. They can be either fixed-width
+		# or have their leading zeroes removed. A run of zeroes
+		# can be abreviated with ::, but only once per address
+
+		# If we have two or more doublecolons, it's not valid
+		# If we have one, we can have anywhere between 3 to 8
+		# "fields" separated by ':'
+		# If we have zero, we must have exactly 8 fields
+		doublecolons = domain.count('::')
+		if doublecolons > 1:
+			return False
+		else:
+			fields = domain.split(':')
+			if doublecolons == 1 and len(fields) > 8:
+				return False
+			elif doublecolons == 0 and len(fields) != 8:
+				return False
+
+		# All of the "fields" must have 0 to 4 hex digits
+		if not all(0 <= len(field) <= 4 for field in fields):
+			return False
+		return all(all(map(is_hexdigit, field)) for field in fields)
 
 	return is_ipv4(domain) or is_ipv6(domain)