--- MPIWGWeb/Attic/nameSplitter.py	2006/08/27 05:40:45	1.1.2.1
+++ MPIWGWeb/Attic/nameSplitter.py	2006/09/13 08:17:33	1.1.2.2
@@ -31,6 +31,7 @@ def getSupportedEncoding(encodings):
 splitter for lastnames in database
 
 """
+import re
 
 def quote(str):
     str=str.replace("'","\\\'")
@@ -40,22 +41,27 @@ class nameSplitter:
     default_encoding = "utf-8"
 
     def process(self, lsttmp):
+        print "XXX"
         result = []
         o = psycopg.connect('dbname=authorities user=dwinter password=3333',serialize=0) 
         c = o.cursor() 
-        replaceStr="<>;.:()"
+#        replaceStr="<>;.:(),"
         lst=" ".join(lsttmp)
-        for x in replaceStr:
-            lst=lst.replace(x," ")
-        
+#        for x in replaceStr:
+#            lst=lst.replace(x," ")
+        lst=re.sub("[<|>|;|.|:|\(|\|)|,]", " ", lst)
         for s in lst.split():
+
             if type(s) is not StringType: # not unicode
                 s = s.decode(self.default_encoding)
+
             if s not in result: # check for database entry
-                c.execute("select lastname from persons where lower(lastname) = '%s'"%quote(s))
+
+                #c.execute("select lastname from persons where lower(lastname) = '%s'"%quote(s.lower()))
+                c.execute("select lastname from persons where lastname = '%s'"%quote(s))
                 if c.fetchone():
                     print "found",s
-                    result.append(lastname)
+                    result.append(s)
         return result