arcpy.da.updatecursor to delete duplicate rows where values switch columns

Question

I'm trying to dissolve polygons using polygon neighbors in arcgis 10.1, but don't want to use duplicate rows. However, the field values will switch when they're duplicated.

If I run this:

arcpy.env.workspace = "C:\Users\Ant\Documents\ArcGIS\ITN.gdb"
fc = r"BLPUs_PolygonNeighbors"
fields = ["src_OBJECTID","nbr_OBJECTID"]
with arcpy.da.SearchCursor(fc,fields) as cursor:
 for row in cursor:
 print "{0}, {1}".format(row[0],row[1])

I get:

1, 2
1, 4
2, 1
2, 3
2, 4
3, 2
3, 4
4, 1
4, 2
4, 3

Then when I run my script to dissolve based on these values, I will end up with duplicate polygons such as 1 & 2 as well as 2 & 1. Can someone please help me with writing the UpdateCursor to go through and delete these duplicates? I don't know how to look through when the fields have then switched.

Also, what if I then have three polygons to merge? If I had three columns such as OID1, OID2 and OID3, is there an SQL expression to capture whether the three values in these have been repeated in a different order but in the same row previously? Thanks

Answer 1

This should do it with a single pass.

arcpy.env.workspace = r"C:\Users\Ant\Documents\ArcGIS\ITN.gdb"
fc = r"BLPUs_PolygonNeighbors"
fields = ["src_OBJECTID","nbr_OBJECTID"]
row_pairs = set()
with arcpy.da.UpdateCursor(fc,fields) as cursor:
 for row in cursor:
 row_pair = tuple(sorted(row))
 if row_pair in row_pairs:
 cursor.deleteRow()
 else:
 row_pairs.add(row_pair)

• Paul: By creating an empty set, will row_pair only be added if it doesn't already exist in row_pairs? I'm trying to understand why to use that over creating an empty list.
• Jason Scheirer: If the row_pair is not yet in the set, it gets added to the set. But if it's already in the set, the row is a known duplicate and it's deleted. A set is useful because it will only ever have one copy of each value so it won't get excessively big and uses a hash to look up values so searching it is faster than searching a list.
• user2581350: thanks, that worked - had to use UpdateCursor, not SearchCursor though. Edited the answer. Cheers
• Paul: +1 Hmm, I need to look more into sets then, thanks!
• user2581350: i guess it was for the deleteRow :)

Answer 2

You can use sorted() (which is a built in Python function) in a generator expression to sort the sub-lists separately and then use sorted() on the entire list so that it's in ascending order. Using your example:

cursordata = [[1, 2], [1, 4], [2, 1], [2, 3], [2, 4], [3, 2], [3, 4], [4, 1], [4, 2], [4, 3]]
print sorted((sorted(x) for x in cursordata))
[[1, 2], [1, 2], [1, 4], [1, 4], [2, 3], [2, 3], [2, 4], [2, 4], [3, 4], [3, 4]]

This can easily be extended to sublists of length n.