pythonw.exe crashing

[MarcCusumano]

I had some code working previously that is now causing pythonw.exe to crash. I have an .mxd set up that I am using to loop through features in a layer and export the layout to jpeg. However now the script crashes when it gets to df.zoomtoselectedfeatures. Here is what I have:

import arcpy  Path = os.getcwd() GSP_Join = "C:\\Users\\TDMC0\\AppData\\Local\\Temp\\scratch.gdb\\GSP_Join"  mxd = arcpy.mapping.MapDocument(Path + r"\GasLeaks2013.mxd") df = arcpy.mapping.ListDataFrames(mxd, "Layers")[0] Layer = arcpy.mapping.ListLayers(mxd, "", df)[0] Layer2 = arcpy.mapping.ListLayers(mxd, "", df)[1] Layer3 = arcpy.mapping.ListLayers(mxd, "", df)[2]  # Create a search cursor try:      rows = arcpy.SearchCursor(GSP_Join, "", "", "CONN_OBJ_ID_NUM", "CONN_OBJ_ID_NUM")  except Exception as e:      print e  # Iterate through the Implausible Reads Feature, getting and zooming to respective extents and exporting layout to JPEG. currentState = " " count = arcpy.GetCount_management(GSP_Join) progress = 1  for row in rows:       try:                           if currentState != row.CONN_OBJ_ID_NUM:                print "\rExporting Data Driven Pages to JPEG...%s of %s" %(progress, count)                currentState = row.CONN_OBJ_ID_NUM                where_clause = "\"CONN_OBJ_ID_NUM\"=" + "'" + currentState + "'"                arcpy.SelectLayerByAttribute_management(Layer, "NEW_SELECTION",  where_clause)                df.zoomToSelectedFeatures()                df.scale = 5500                arcpy.RefreshActiveView()                arcpy.SelectLayerByAttribute_management (Layer, "CLEAR_SELECTION", "")                Layer.definitionQuery = where_clause                Layer2.definitionQuery = where_clause                Layer3.definitionQuery = where_clause                arcpy.mapping.ExportToJPEG(mxd, r"E:\GasLeaksProject\2012\JPEG\CONN_OBJ_ID_NUM" + str(currentState) + ".jpeg", resolution=300, jpeg_quality=70)                Layer.definitionQuery = ""                Layer2.definitionQuery = ""                Layer3.definitionQuery = ""                progress = progress + 1                      except Exception as e:           print e

When I run this I get a Windows OS message saying "pythonw.exe has crashed". I narrowed it down to the line df.zoomToSelectedFeatures() (if I comment out this line the code runs but obviously does not output what I want). Again this was working previously (about six months ago) I must have tested it a hundred times. All of the data sources are working in the mxd, and I tried re-saving the mxd and pointing to the new copy but no luck. I also printed my where_clause and it looks good. It must be a problem with my "list data frames / list layers code". Any help would be appreciated.

[MathewCoyle]

Since you are going through a cursor anyway you can just get the extent of the row shape object instead of using a selection. Unless there is some particular reason you need to select it. I'm also assuming your selections are unique to each row since you call them CONN_OBJ_ID_NUM.

for row in cursor:
    df.extent = row.Shape.extent

[MarcCusumano]

So pythonw.exe is still crashing, I don't think it has anything to do with what I previously thought. I I comment that out, it will get through 14 or so images and then crash. Can anyone see anything wrong with my code / how to optimize it? Thanks.

[MathewCoyle]

If you watch the process and related threads does it get over 2GB? Do you have any AV program running?

[MarcCusumano]

I found out that I cannot even pan to the features in the map document. The rest of the script - the part before the loop that exports the jpeg images - performs spatial selections and queries against point features to detect clusters of gas leaks and writes them to a gdb. The layers in the saved map document point to these new features. When I try to pan to the 24th record it crashes ArcMap.

[ChrisSnyder]

Is there something wonkey with the 24th feature in your GSP_Join layer (null/corupt geometry)?

Can you do it manually and have it work correctly?

[MarcCusumano]

I'm currently testing the process on separate machines - one running 10.0 and my main workstation running 10.1. I'll let you know what I find.

[MarcCusumano · Accepted solution]

Solved this by creating an entirely new mxd. The original was not "corrupt" but there was something causing an issue. Thanks for the help.