wrong size of shapes

dion

Hi,
I found size() for shapes return wrong value.
See below code, shapes.size() return 1.
However, it could find 7 if loopped with shapes.each().

import pya

nly = pya.Layout(False)
nly.read('t1.oas')
top_cell = nly.cell(0)
ss = top_cell.shapes(0)
print(top_cell.basic_name(), ss.size())

slist = [x for x in ss.each()]
print(top_cell.basic_name(), len(slist))

code results:

top 1
top 7

Is it a bug?

Thanks,
Dion

Matthias

Hi Dion,

I don't have your file, but there is a specific issue with OASIS and shapes: OASIS allows representation of shapes as shape arrays. That means a single object represents multiple effective shapes.

"size" is a quick method that does not do expansion of the shape arrays. So a single shape array is one object, but may represent multiple shapes when they are delivered individually with "each". This explains the difference.

The size count is correct if

• The file format is GDS or any other format except OASIS
• The layout is created in editable mode (True for the argument to the Layout constructor)

I agree this is a bit unintuitive. Internally, shape arrays can either be treated as single or multiple objects, but I did not want to overload the API with such details. Basically "size" is borrowed from the internal API with the array perspective.

I can change the "size" behaviour to reflect the real number of shapes after expansion with "each". This will make "size" a bit slower, but I don't think this will be noticeable in a Python script.

Kind regards,

Matthias

dion

Hi Matthias,
You are correct.
I reviewed the oasis case again and found the polygons in this case are same.
So oasis use displacement record to compress it.

I think it's efficient to represent multiple shapes as oasis.
So I have another question.
Could I get the oasis shapes displacement or array in python script?

Thanks,
Dion

Matthias

Hi Dion,

no sorry, there is no way to get the real number of shapes yet. If you like, you could file a bug on GitHub https://github.com/klayoutmatthias/klayout/issues/152. Please add whatever you like to mention there.

Thanks,

Matthias