Take the 2-minute tour ×
Geographic Information Systems Stack Exchange is a question and answer site for cartographers, geographers and GIS professionals. It's 100% free, no registration required.

UPDATED BELOW

I have 9 shapefiles, each shapefile has 700,000+ rows and I have roughly 12 attribute fields in each shapefile.

There are three attribute fields, common to every shapefile, that I am interested in - Type, Code, Area. In the majority of records, there is one entry for each field (| B | X2 | 3.45 |) but in some cases you have more than one entry in a field (up to 8) separated by comma;

| A, B, B | X2, ZZ9, GGR | 1.2, 3.1, 0.23 |

It is nearly always the case that if there is 3 records in one field, the other two fields concerned will also have 3 records. If this does not apply, then the data is incomplete and can be ignored. It is also the case that the 1st entry in one field is associated with the 1st entry in another field.

I have found how to ArcGIS 10 Python split string and ArcMap 10.1 Python split inconsistent string but that results in deconcatenating the string by the comma delimiter and creating new attribute fields for every comma separated entry.

The thing i need is for the deconcatenated data to create new records NOT new fields. In the instance above;

  1. | A, B, B | X2, ZZ9, GGR | 1.2, 3.1, 0.23 |

would become

  1. | A | X2 | 1.2 |
  2. | B | ZZ9 | 3.1 |
  3. | B | GGR | 0.23 |

the top one is the original record with the superfluous data removed into two new records below it. I need the new records to retain all the other information from the other 9 fields that dont need to be touched and also the geometry as they refer to the same polygon space.

thanks very much, i'm stuck!

(Answers for ArcGIS 10.1, FME, R, Python all gladly accepted)

EDIT 9th APR:

i used the following code to split concatenated entries (by a comma) in fields 'TYPE' and 'AREA' into new rows (there are between 1 and 8 concatenated entries in TYPE and AREA):

import arcpy
searchC =  arcpy.da.SearchCursor("FeatureClass", ("TYPE","AREA", "SHAPE@"))

insertC = arcpy.da.InsertCursor("newFeatureClass", ("TYPE","AREA", "SHAPE@")) 

for row in searchC:
   A = [x.split(",") for x in row[:-1] ]

   for i in range(min(len(A[0]),len(A[1]))):
       newRow = [A[0][i] , A[1][i], row[2]]
       insertC.insertRow(newRow)
del insertC

this produces the correct amount of new rows based upon the content of the original concatenated rows.

but when doing this, i only get the 2 fields TYPE and AREA next to a new FID. each entry also has a uniqueID that has to come across as well, so i tried (amateurishly) this;

import arcpy
searchC =  arcpy.da.SearchCursor("FeatureClass", ("UNIQUEID", "TYPE", "AREA", "SHAPE@"))

insertC = arcpy.da.InsertCursor("newFeatureClass", ("UNIQUEID", "TYPE","AREA", "SHAPE@")) 

for row in searchC:
    A = [x.split(",") for x in row[:-1] ]

    for i in range(min(len(A[0]),len(A[1]),len(A[2]))):
        newRow = [A[0][i] , A[1][i], A[2][i], row[3]]
        insertC.insertRow(newRow)
del insertC

However, this now leaves the original concatenated row plus one new row that takes only the first entry from TYpe and AREA. UNIQUEID does populate both rows.

Any help? i want to split out the concatenated entries but make sure the UNIQUEID that applies to all is carried across, thanks very much

share|improve this question
    
Am I getting this right; in your example you would want three identical polygon features in your shapefile, one with each of the three groups of attributes? Instead of the single polygon with all attributes combined? –  Martin Jan 16 at 12:55
    
i know it sounds odd, but yes. the values concatenated cause lots of problems for mapping (As one eg, instead of 150 unique entries for the 'Code' field, you get up to 10,000 unique entries due to the different combinations). –  Sam Jan 16 at 13:01
    
It doesn't sound odd; it is in fact best practice to model this as a many-to-one relationship. The real question is whether the geometry should remain on the "one" side or be duplicated. Implementing this is easy, but determining if it's the correct path isn't. –  Vince Jan 16 at 13:14
    
@ Vince, could you elaborate on the 'easy' part? :) –  Sam Jan 16 at 13:53

1 Answer 1

you can use something like below :

import arcpy
searchC =  arcpy.da.SearchCursor(featureclass, ("type","code","area" , "SHAPE@")) #plus other attribute if you need

insertC = arcpy.da.InsertCursor(newfeatureclass, ("type","code","area" , "SHAPE@")) 

for row in searchC:
    A = [x.split(",") for x in row[:-1] ]

    for i in range(min(len(A[0]),len(A[1]),len(A[2]))):
        newRow = [A[0][i] , A[1][i], A[2][i], row[3]]
        insertC.insertRow(newRow)
del insertC

you can use a similar code to create a table instead of a feature class, as mentioned by Vince, so that you do not duplicate the geometries but instead create a one to many relation.

share|improve this answer
    
thanks, i'm not sure i understand Search and Insert Cursor very well, so i'm struggling a bit, but something might come of it. –  Sam Jan 16 at 15:37
2  
searchCursor() iterates through all features in a set, one by one, relatively quickly. It won't make changes to the feature set. insertCursor() will make changes. In the @radouxju example , new features ("newRow" instances) are added to the end of the pre-existing feature set w/insertCursor(). One could search w/insertCursor(), but it's slower. So the example is a nice blend (fast search, use heavyweight insert only when needed). Suggestion: use the with/as syntax when instantiating searchCursor() ( resources.arcgis.com/en/help/main/10.1/index.html#//…). –  Roland Jan 16 at 16:02
    
ok thanks, i understand the principles but getting into reality is proving to be very hard, i get endless blank rows at the bottom of my attribute table. I'll persevere. –  Sam Jan 16 at 16:47
    
well, I've tested my code and it works. But I found a typo ! (InsertCurso instead of InsertCursor). Please check the names of your field and the fact that the separator is a ",". I also recommend that you store your result in another shapefile that the original one, but with already existing field names (featureclass and newfeatureclass are of course the full names of your data. ) –  radouxju Jan 16 at 19:36
    
@ radouxju, I had spotted the typo but didn't want to criticise your excellent code, especially considering i'm not fully au fait how to use it yet! I think i'll crack it this morning though. –  Sam Jan 17 at 8:54

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.