From 71b2e3b1420c3188db7b15743cf3dd382d425a9a Mon Sep 17 00:00:00 2001
From: Sven Slootweg <jamsoftgamedev@gmail.com>
Date: Thu, 21 Nov 2013 07:52:44 +0100
Subject: [PATCH] Data normalization

---
 pwhois               |  2 +-
 pythonwhois/parse.py | 38 +++++++++++++++++++++++++++++++++++++-
 2 files changed, 38 insertions(+), 2 deletions(-)

diff --git a/pwhois b/pwhois
index 1f3c077..768300d 100755
--- a/pwhois
+++ b/pwhois
@@ -18,7 +18,7 @@ else:
 if args.raw == True:
 	print "\n--\n".join(data)
 else:
-	parsed = pythonwhois.parse.parse_raw_whois(data)
+	parsed = pythonwhois.parse.parse_raw_whois(data, normalized=True)
 	data_map = OrderedDict({})
 	
 	# This defines the fields shown in the output
diff --git a/pythonwhois/parse.py b/pythonwhois/parse.py
index 6ac0bff..02dccc2 100644
--- a/pythonwhois/parse.py
+++ b/pythonwhois/parse.py
@@ -116,7 +116,7 @@ grammar = {
 	}
 }
 	
-def parse_raw_whois(raw_data):
+def parse_raw_whois(raw_data, normalized=[]):
 	data = {}
 	
 	raw_data = [segment.replace("\r", "") for segment in raw_data] # Carriage returns are the devil
@@ -178,6 +178,42 @@ def parse_raw_whois(raw_data):
 	
 	data["raw"] = raw_data
 	
+	if normalized != []:
+		data = normalize_data(data, normalized)
+	
+	return data
+
+def normalize_data(data, normalized):
+	for key in ("name_servers", "emails", "whois_server"):
+		if key in data and data[key] is not None and (normalized == True or key in normalized):
+			if isinstance(data[key], basestring):
+				data[key] = data[key].lower()
+			else:
+				data[key] = [item.lower() for item in data[key]]
+	
+	for key in ("registrar", "status"):
+		if key in data and data[key] is not None and (normalized == True or key in normalized):
+			if isinstance(data[key], basestring) and data[key].isupper():
+				data[key] = " ".join(word.capitalize() for word in data[key].split(" "))
+			else:
+				data[key] = [" ".join(word.capitalize() for word in item.split(" ")) for item in data[key] if item.isupper()] + [item for item in data[key] if not item.isupper()]
+	
+	for contact_type, contact in data['contacts'].iteritems():
+		if contact is not None:
+			for key in ("email",):
+				if key in contact and contact[key] is not None and (normalized == True or key in normalized):
+					if isinstance(contact[key], basestring):
+						contact[key] = contact[key].lower()
+					else:
+						contact[key] = [item.lower() for item in contact[key]]
+			
+			for key in ("name", "street", "city", "state"):
+				if key in contact and contact[key] is not None and (normalized == True or key in normalized):
+					if isinstance(contact[key], basestring) and contact[key].isupper():
+						contact[key] = " ".join(word.capitalize() for word in contact[key].split(" "))
+					else:
+						contact[key] = [" ".join(word.capitalize() for word in item.split(" ")) for item in contact[key] if item.isupper()] + [item for item in contact[key] if not item.isupper()]
+	
 	return data
 
 def parse_dates(dates):