Selecting the face of a small object on top of the stack

[ervandarnell]

I'm just getting started on cadquery as an upgrade to openscad. I want to create this cleanly:

It's trivial like so:

cq.Workplane("XY")
.box(2,2,2).faces(">X").workplane()
.box(1,1,1,centered=(1,1,0)).faces("<Z").workplane()
.box(5,5,1, centered=(1,1,0))

But the problem here is the box222 is first in the chain. This won't work for a bigger design. The box551 is a platform to build multiple things on, thus it really needs to be first in the chain:

.box(5,5,1, centered=(1,1,0)).faces(">Z").workplane().center(1,1)
.box(2,2,2, centered=(1,1,0)).faces(">X").workplane()
.box(1,1,1, centered=(1,1,0))

But this does not work, the .faces(">X") applies to the whole object, not the box222 on the top. I get this:

Ultimately, I want something like the following with multiple objects built on the platform:

.box(5,5,1, centered=(1,1,0)).faces(">Z").workplane().tag("platform").center(1,1)
.invoke(add_canon())
.workplaneFromTag("platform").center(-1,-1).invoke(a_different_canon()).

I don't see any way to do this without just calculating positions and putting hard numbers into .transformed().

Seems like this should work (with a call to .end()) but it doesn't:

.box(5,5,1, centered=(1,1,0)).faces(">Z").workplane().center(1,1)
.box(2,2,2, centered=(1,1,0)).end().faces(">X").workplane()
.box(1,1,1, centered=(1,1,0))

Also notice I'm not building on top of the box222 but in front of it, so faces(">Z") is definitely not the right answer.

Is there some useful selector or tag I'm overlooking? Thanks.

[adam-urbanczyk]

With the current API I'd do it like so:

import cadquery as cq

res = (
        cq.Workplane()
        .box(5,5,1, centered=(1,1,0))
        .faces(">Z")
        .workplane()
        .center(-1,0)
        .box(2,2,2, centered=(1,1,0))
        .faces(">Z").siblings('Edge').faces('>X').workplane(centerOption='CenterOfMass')
        .box(1,1,1, centered=(1,1,0))
)

[ervandarnell]

Thanks for the prompt reply. You've taught me something valuable here about the siblings() call, which I had not realized. This does solve my immediate problem.

However, it doesn't really solve the general problem. What I am after is to abstract the idea of building a "canon" so I can call it twice and build this:

If I directly try to use your idea the faces("">Z") refers to something different on the second call. This:

def forward_cube6():
def canon(): return (lambda wp:
wp.box(20,20,20,centered=False)
.faces(">Z").siblings('Edge').faces('>X')
.workplane(centerOption="CenterOfMass")
.box(10,10,10, centered=(1,1,0))
)
canon1 = ( cq.Workplane()
.box(60,60,1, centered=False)
.faces(">Z")
.center(5,5).invoke(canon())
)
canon2 = (canon1
.faces(">Z").center(34,34).invoke(canon())
)
return canon2

After some fiddling, this almost works, putting both canons on the platform relative to the top of the platform (the key part being: .faces("<Z").siblings('Edge').siblings('Edge').faces(">Z") )

def forward_cube6():
def canon(): return ( lambda wp:
wp.box(20,20,20,centered=False)
.faces(">Z").siblings('Edge').faces('>X')
.workplane(centerOption="CenterOfMass")
.box(10,10,10, centered=(1,1,0))
)
platform = base.box(60,60,1, centered=False) # .tag("platform")
canon1 = platform.faces(">Z").center(5,34).invoke(canon())
canon2 = (canon1
.faces("<Z").siblings('Edge').siblings('Edge').faces(">Z")
.workplane()
.center(5,-30).invoke(canon())
)
return canon2

though I am not exactly sure why it works (the second canon is x-y relative to the first canon, and it seems like .faces(">Z").siblings('Edge').faces('>X') should be ambiguous as it could be either the top of canon1 or canon2).

But all of this only defers the problem by finding some circuitous path that works this time, but won't work in general.

What I really need is an explicit tagging of the top of the platform:

def forward_cube6():
def canon(): return ( lambda wp:
wp.box(20,20,20,centered=False)
.faces(">Z").siblings('Edge').faces('>X')
.workplane(centerOption="CenterOfMass")
.box(10,10,10, centered=(1,1,0))
)
platform = base.box(60,60,1, centered=False).tag("platform")
canon1 = platform.faces(">Z").center(5,34).invoke(canon())
canon2 = (canon1
.workplaneFromTagged("platform")
.center(5,-30).invoke(canon())
)
return canon2

but this doesn't work because workplaneFromTagged 'forgets' everything above the tag() call.

I can force it as follows:

....
return canon1 + canon2

returning two objects on top of each other. Is that standard practice?

Even better would be:

def canon(): return ( lambda wp:
wp.box(20,20,20,centered=False)
.end().faces(">Z")
.box(10,10,10, centered=(1,1,0))
)

where end() returns the box222 on top of the stack, the one just constructed, but that does not seem to work either?

Yes, a long message. Thanks for the hint above even if you don't want to plow through this.

[adam-urbanczyk]

You should define a shape of the cannon, add the desired locations and use each/eachpoint. AFAICT invoke is not for you. See also the second example here: https://cadquery.readthedocs.io/en/latest/examples.html#extruding-until-a-given-face.

[ervandarnell]

Thanks for the further suggestion. I tried something similar previously and I don't quite see how it works. The problem is that it seems to need computed coordinates. In the skyscraper example:

skyscrapers_locations = [(-16, 1), (-8, 0), (7, 0.2), (17, -1.2)]

How do I make one thing relative to another?

For instance, if I want this:

and I try to do this with the skyscrapers' approach of multiple initial workplanes, it doesn't seem to work:

def dut(): 
  box1=cq.Workplane().box(1,2,1, centered=(1,1,0))
  platform=cq.Workplane().box(5,5,1,centered=(1,1,1))
  platform = platform.faces(">Z").transformed(offset=(2,2,1)).union(box1)
  return platform

The faces(..).transformed(offset...)) is just ignored:

The boxes on top are just proxies for some more complex function that I want to build in multiple places so I really want that to be a function I can call.

This moves box2, but not relative to faces(">X"), so no help either:
` platform = platform.faces(">X").transformed(offset=(-2,-2,1)).union(box2.translate((2,2,0))

Let me try to restate this. I can always compute the exact location of every object and do
cq.workplane().box(....).translate(...computed point 1...) cq.workplane().box(....).translate(...computed point 2...)
cq.workplane().box(....).translate(...computed point 3...)`

but I am trying to avoid that and use the built-in fluent API for sub-assemblies.

[adam-urbanczyk]

Could you define your use case concisely, preferably with a single (formatted) minimal example?

In the meantime, AFAICT you want to add a shape on multiple locations. The canonical way (using the Workplane API) of doing so would be more or less this:

import cadquery as cq
from cadquery.vis import show

canon = (
    cq.Workplane()
    .box(2,2,2, centered=(1,1,0))
    .faces('>X').workplane(centerOption='CenterOfMass')
    .box(1,1,1, centered=(1,1,0))
)
    
result = (
    cq.Workplane()
    .box(10,10,0.5)
    .faces('>Z').workplane()
    .pushPoints([(0, 0), (3, 2)])
    .eachpoint(canon, combine=True)
)

show(result)

[ervandarnell]

Thank you for bearing with me until we got to a solution!

For me, the answer here is that faces().eachPoint() moves the build point where the next cq.Workspace() starts. I was using union() which ignores the build point and resets to 0.

The other key point is that pushPoints only moves within a workplane(). I still don't understand exactly which things work like this and which don't, but I have a working example to start experimenting with.