Why are some items missing from intellisense?

[ChrisHills]

When creating python scripts, I find that there items missing from the intellisense popups in the Python window. I also see that the same items are missing when examining an object with dir().

For example:-

cursor = arcpy.SearchCursor("My Feature Class")
row = cursor.next()
dir(row)
['__class__', '__cmp__', '__delattr__', '__dict__', '__doc__', '__format__', '__getattr__', '__getattribute__', '__hash__', '__init__', '__module__', '__new__', '__passthrough_to_ao__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__str__', '__subclasshook__', '__weakref__', '_arc_object', '_go', 'getValue', 'isNull', 'setNull', 'setValue']

dir(row.SHAPE)
['JSON', 'WKB', 'WKT', '__add__', '__class__', '__cmp__', '__delattr__', '__dict__', '__doc__', '__format__', '__from_scripting_arc_object__', '__geo_interface__', '__getattribute__', '__hash__', '__init__', '__iter__', '__module__', '__new__', '__or__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__str__', '__sub__', '__subclasshook__', '__type_mapping__', '__type_string__', '__weakref__', '__xor__', '_arc_object', '_fromGeoJson', '_go', '_passthrough', 'area', 'boundary', 'buffer', 'centroid', 'clip', 'contains', 'convexHull', 'crosses', 'cut', 'difference', 'disjoint', 'distanceTo', 'equals', 'extent', 'firstPoint', 'getArea', 'getLength', 'getPart', 'hullRectangle', 'intersect', 'isMultipart', 'labelPoint', 'lastPoint', 'length', 'length3D', 'measureOnLine', 'overlaps', 'partCount', 'pointCount', 'positionAlongLine', 'projectAs', 'queryPointAndDistance', 'snapToLine', 'spatialReference', 'symmetricDifference', 'touches', 'trueCentroid', 'type', 'union', 'within']

Note that SHAPE is missing from the output of dir(row). In the intellisense popup it shows getValue, isNull, setNull, setValue, but not SHAPE.

This makes it a bit more difficult when writing scripts as I have to remember that the object is missing.

[JasonScheirer · Accepted solution]

Remember that SHAPE is a special case, it's a column name in the row. All of the Row's column name attributes are computed at the very last possible minute. The class has its usual attributes like its getValue method, but then it overrides the __getattr__ method as a last-ditch chance for the class instance to return a value for object.SOMETHING before Python throws an AttributeError.

TL;DR: The row class doesn't have each individual column name as an attribute. It instead uses some Python magic to allow Row.COLUMN to act as syntactic sugar for Row.getValue("COLUMN").

[ChrisHills]

Thanks for the helpful answer!