Layer Isolation and cells Boolean operation about gdspy

[heitzmann]

@egholami I'm not sure I understand what you're trying to do, but probably cell.get_polygons(by_spec=True) should help.

If you can post a minimal example, it might be easier to help you.

[egholamy]

This is the Trim method that I add to Cell class:**
def Trim (self,
mash):
poly_dic = self.get_polygons(True)
self.elements = []
for ld in poly_dic.keys():
if ld[0]==mash:
self.add(PolygonSet( poly_dic[ld],ld[0], verbose=False))
return self

It is working for Cell layer isolation and for the Boolean cell Operation I am using Get_polygons in the following way but it is very slow:

layli=['EBL','Trenches','Mesa']
for i in range(1,4):
globals()['sepLayer%s'%i]=Ref_cell.copy('{}'.format(layli[i-1]), deep_copy=True)
globals()['sepLayer%s'%i].Trim(mash=i)
globals()['sepLayer%s'%i]=gdspy.fast_boolean(globals()['sepLayer%s'%i].get_polygons(False, None),Iqub0m,'not',layer=i)
re_cellmodified.add(globals()['sepLayer%s'%i])
for one of my layers that have a lot of points and elements it is very slow but it is working
**Now I have tried what you suggested **

for i in range(1,4):
globals()['sepLayer%s'%i]=flatten.copy('{}'.format(layli[i-1]), deep_copy=True)
globals()['sepLayer%s'%i].Trim(mash=i)
globals()['sepLayer%s'%i]=gdspy.fast_boolean(globals()['sepLayer%s'%i].get_polygons(by_spec=True),Iqub0m,'not',layer=i)
re_cellmodified.add(globals()['sepLayer%s'%i])
print(i)

It is giving me this error!

Traceback (most recent call last):

File "C:\Users\ehgh\AppData\Local\Continuum\anaconda3\lib\site-packages\gdspy_init_.py", line 4274, in fast_boolean
result = clipper.clip(polyA, polyB, operation, 1 / precision)
TypeError: 'numpy.int32' object does not support indexing

Is there any way to make Boolean Cell operation fast?

[heitzmann]

@egholami get_polygons(by_spec=True) returns a dictionary, so you're a missing the key (similar to what you do with poly_dic. ALso, you probably don't need to use flatten() or copy().

In any case, boolean operations are slow. The only way around it is to use fewer points... I find people often overshoot the precision they need in curved geometries. Another option would be to try to avoid using boolean operations at all by drawing your geometry some other way.

Unrelated to the issue: it's quite dangerous to be abusing globals() like that. Global variables are in general a bad design choice, let alone using globals. You'd be better of using your own dictionary.

[egholamy]

Thanks a lot! It was a pseudo code and you are right that using the globals() is a bad design choice I would keep it in mind. Using the way that you have explained there would be no need for hard copy.
Unfortunately we deal with some special kind of text and we could not use regular text so I need my own designed text with opening. Because for some chemistry reasons some letters like "O" or "B" could not be used in closed form and need some opening.