Hi
Does anyone have any clever ideas (or something obvious I am missing) to properly copy across gdb attachments in a python script.
So I have a featureclass in an Enterprise GDB with gdb attachments enabled, and I want to export a subset of these features into an fgdb, and bring along all the gdb attachments with it.
If I use a simple copy/paste eg arcpy.Copy_management(), this brings across all 000,000's of the features and the GB's of attachments. I cannot find a way to have arcpy.Copy work on a subset selection?
Using arcpy.CopyFeatures works great for the export of the featureclass, my makeFeatureLayer call is giving me just the subset I need. However arcpy.CopyFeatures does not bring in the gdb attachments. So what I tried is enabling attachments on the copyFeatures new layer, then arcpy.append from the old gdb attach table to the new.
sourceFC = "someegdb.sde\\someSourceFeatureclass" targetFC = "somefgdb.gdb\\someFeatureClassSubset" layerName = "selectionLayer" expression = "foo=999" arcpy.MakeFeatureLayer_management(sourceFC, layerName, expression) arcpy.CopyFeatures_management(layerName,targetFC) arcpy.EnableAttachments(targetFC) attachTempTable = "attachmentSubSelection" attachTargetTable = targetFC+"__attach" where = 'where REL_GLOBALID in (SELECT "GLOBAL_ID" FROM %s %s)' % (sourceFC ,expression) arcpy.MakeTableView_management(sourceFC+"__ATTACH", attachTempTable, where) arcpy.Append_management(attachTempTable, attachTargetTable, "NO_TEST")
The biggest problem I am finding with this approach is the way gdb attachments uses the GLOBAL_ID / REL_GLOBALID. arcpy.CopyFeatures() just recreates all globalId's in the new layer, but arcpy.Append has brought in the old REL_GLOBALID's.
A rather clunky workaround I am using is to use the feature centroid to update the REL_GLOBALID, so I go through the sourceFeatureclass, record the old globalId and old centroid
centroidDict = {} with arcpy.da.SearchCursor(sourceFC,["GLOBAL_ID","SHAPE@XY"],expression) as cursor: for row in cursor: centroidDict[str(row[1])] = row[0] del row, cursor
Then get the new global id based on centroid
newCentroidDict = {} with arcpy.da.SearchCursor(targetFC,["GLOBAL_ID","SHAPE@XY"]) as cursor: for row in cursor: newCentroidDict[centroidDict[str(row[1])]] = row[0] del row,cursor
And finally update the REL_GLOBALID
with arcpy.da.UpdateCursor(attachTargetTable,["REL_GLOBALID"]) as cur: for row in cur: row[0] = newCentroidDict[row[0]] cur.updateRow(row) del row,cur
Whilst this 'works' there is a clear falecy in the approach, where the features share the same centroid.
So, does anyone have any other ideas, ones that mean I am not having to work against/copy entire/alter the enterprise gdb full layer?
Thanks