Work on improving the API

bokonon/being.py

@@ -82,12 +82,14 @@ def groupMerge(universe, pred,extract, description=None,logging=None):
             print(txt)
             print("")
 
-def windowMerge(universe, pred, extract, windowSize, maxDistance, description=None):
+def windowMerge(universe, selector, extract, windowSize, maxDistance,
+                pred = lambda x,y: True, description=None,logging=None):
+
     if description != None:
             print(description)        
             start = countTypes(universe)
 
-    nodes = filter(lambda t: pred(t[1]),universe.nodes(data=True))
+    nodes = filter(lambda t: selector(t[1]),universe.nodes(data=True))
     d = {}
     for k,v in nodes:
         for s in extract(v):
@@ -100,11 +102,15 @@ def windowMerge(universe, pred, extract, windowSize, maxDistance, description=No
             b = items[i+j]
             dQ = distance(a[0],b[0]) <= maxDistance
             bQ = findBeing(universe,a[1]) != findBeing(universe,b[1])
-            lQ = len(a[0]) > 5 and len(b[0]) > 5 
-            if  dQ and bQ and lQ:
-                # print(a[0])
-                # print(b[0])
-                # print("\n")
+            lQ = len(a[0]) > 5 and len(b[0]) > 5
+
+            an = universe.node[a[1]]
+            bn = universe.node[b[1]]            
+            if  dQ and bQ and lQ and pred(an,bn):
+                if logging != None:
+                    print(logging(an))
+                    print(logging(bn))                    
+                    print("")
                 mergeTheirBeings(universe,a[1],b[1])
 
     cullHermits(universe)

bokonon/graph.py

@@ -13,32 +13,43 @@ def main():
     print("Loading universe...")
     universe = loadData()
 
+    #Solid matching
     groupMerge(universe,
                matchTypeAndHasFields("client",["name"]),               
                lambda v: [v["name"]],
                description="Merged clients based on exact name match")
 
+    #Surprisingly solid
     groupMerge(universe,
                matchTypeAndHasFields("client",["name"]),               
-               lambda v: extractNames(v["name"]),
-               description="Merged clients based on extracted and cleaned name match")
+               lambda v: [re.sub(" ","",v["name"])],
+               description="Merged clients based on exact match without spaces")
 
+    #Solid
     groupMerge(universe,
                matchTypeAndHasFields("client",["name"]),               
-               lambda v: [re.sub(" ","",v["name"])],
-               description="Merged clients based on exact match without spaces",
-               logging=represent)
-
+               lambda v: [re.sub("'","",v["name"])],
+               description="Merged clients based on exact match without 's")
+
+    #Most likely solid
+    groupMerge(universe,
+               matchTypeAndHasFields("client",["name"]),               
+               lambda v: extractNames(v["name"]),
+               description="Merged clients based on extracted and cleaned name match"
+               )
+
+    #Not great
     windowMerge(universe,
                matchTypeAndHasFields("client",["name"]),               
                lambda v: extractNames(v["name"]),
                 5,
                 1,
-               description="Merged clients based on windowed extracted name matchs")
+                pred=lambda v,w: v["state"] == w["state"] and v["city"] == w["city"] and v["address"] == w["address"],
+                description="Merged clients based on windowed extracted name matchs",
+                logging=represent)
 
     project(universe,"clientnames.txt", lambda v: v["type"] == "client", represent)
 
 
 if __name__ == "__main__":
-    main()
-
+    main()