help Improving autocad lisp routine
Various things I could suggest, and a few questions as a start....
(foreach) applying to a list of only one item is silly -- just use the item directly. And it's the Offset distance in a command that uses 0,0,0 as the to-which-side pick, but that wouldn't necessarily always be outside any given outline. I would do something like this, as one way of specifying a to-which-side point that is absolutely guaranteed to be outside:
(command "_.offset" 1 ent (polar (getvar 'extmax) (/ pi 4) 10) "")
The (_pac) function apparently builds a Window-Polygon selection outline by finding points every 100th of the way around an outlining object. That seems appropriate if the object might be a Circle or an Ellipse or a Polyline containing arc segments or a Spline. But in your sample drawing everything is Polylines with only line segments. If that's always the case, their vertices can be used directly, much more simply. And the (ssget) filter list can be greatly simplified. Would that always be the situation? There are numerous routines out there that select everything within a Polyline boundary using its vertices, which works well if it's all line segments.
As for "i would ideally like to select the line and the offset line will be automatically selected," after an Offset command, the result will be (entlast), so it certainly can be automatically selected by the routine.
As for "you are able to click on empty space and the lisp will still continue on to the second step," it can check whether you picked something before continuing, in a few ways. One would be to give up if you didn't:
(if (setq ent (car (entsel "\nSelect polyline to offset: ")))
(progn ; then
.... do its thing ....
)
(progn ; else
(alert "Ya gotta pick something!")
(quit)
)
)
It could also check for the kind of thing picked, etc.
Another would be to ask again, until you do:
(while
(not (setq ent (car (entsel "\nSelect polyline to offset: "))))
)
which could also add a check on the kind of thing picked, etc.
More later, maybe....
@john.uhden wrote:
.... I would probably use the approach of first determining the general direction of the polyline and then using s positive offset for to the right and a negative offset for to the left.
Yes, but the positive-vs.-negative offset distance approach [instead of the pick-which-side approach] needs the (vla-offset) function, which requires VLA-object conversion first. And it takes a substantial amount of code to determine whether a closed Polyline's or Spline's drawn order is CW or CCW -- there are multiple threads that get into that.
But given the VLA-object conversion [which sometimes one also wants for other reasons anyway], another approach I have used to get a result on a particular side of a closed object is to use (vla-offset) with a positive distance, compare the areas of the original and the result, and if it went the wrong way, eliminate the result and do it again with the negative distance.
Do you ever have outlines of such a shape [for example the yellow here] that an Offset command could result in more than one object [red]?
That would throw things off, since you would have no control over which is (entlast). [And in this case, if there was something inside that red triangle, regardless of which red result is used, it would be selected even though it's not within the area I assume you want.]
... and another question....
The AutoLisp routine allows the selection of LINE objects. Are the outlines you want ever composed of Lines [as AutoCAD uses the word]? Those are independent of each other, and even if they make up a closed shape, and you actually pick all those that make it up, Offset wouldn't be able to produce anything you could use for selection of contained objects. If they are always either closed Polylines, closed Splines, full Ellipses, or Circles, yes, but Lines [or Arcs, though they're not allowed in your code so far] would be problematic.
If I understand how it's supposed to work well enough, in the one that Fails, the Offsetting should result in the yellow outline here, and the _pac function should be using the green Points [I used DIVIDE at 100] to form a WP selection [I turned it to reduce space consumed here]:
Looking closely at the upper left corner:
It's using those green points for a Window Polygon, which would follow that white diagonal, and the magenta outline isn't fully inside that. If "FAILS" means it doesn't find the magenta outline in selection, that's probably why. It would fail only when the geometric circumstances are right [or, should I say, wrong], which I suspect will happen more often on larger outlines, because of the proportion of 1-unit Offset distance to WP-defining point spacing -- i.e. the green Points get farther apart, increasing the chances of this happening.
That's why using vertices rather than hundredths of the way around would be better, but then you'd get similar failures in selection when there are arc segments.
But then, if it's all starting from selecting that magenta outline, that could simply be added to what the WP selection finds, and it wouldn't matter if that selection missed it.
@stevie_alas wrote:
.... how would the program know to include what we first selected as our outline
For that part, if I interpret correctly [untested], I would probably structure it a little differently, but for the least intervention, I think [I hope] adding this line:
....
(repeat (setq i2 (sslength temp)) (ssadd (ssname temp (setq i2 (1- i2))) add))
)
(ssadd (ssname ss i) add)
)
(sssetfirst nil add)
(ssget "_I")
....
should do it, adding the boundary object(s) into the selection set of things found by the WP selections. Do the failing ones work, at least in terms of including the boundaries, with that added?
@stevie_alas wrote:
.... what would a vertex selection look like ....
One way to do that with a Polyline is to extract all the 10-code entries in its entity data and strip the 10's off, to apply a list of vertex locations in WP selection:
(if
(and
(setq pl (car (entsel "\nClosed Polyline boundary: ")))
(= (cdr (assoc 0 (setq pldata (entget pl)))) "LWPOLYLINE")
(vlax-curve-isClosed pl)
); and
(progn ; then
(setq ss (ssget "_WP" (mapcar 'cdr (vl-remove-if-not '(lambda (x) (= (car x) 10)) pldata))))
(ssadd pl ss); add the Polyline itself, just in case
); progn
); ifBut if Circles are in the picture, they need to be handled entirely differently [no vertices]. They would most likely need to use something like the 100-points approach.
this unfortunately did not work. i get the same result as before. the boundary is not picked up. if i did this correctly that is
@Kent1Cooper has provided some great code, one thing that may help add multiple vertices only where there is an arc. Again need to sort of over do it so get small chords, representing the arcs. Then total number of vertices will be less. It would be nice if could do a ssget WP OBJ.
Something like use divide then check is a point not an end or start point of a pline,segment, when the segment is a bulge ie ARC keep points. Still looking.
Lee-mac has two possible starting programs add a point to a pline, and for a pline get each segment properties. Using bulge details could work out new points.
It is not code I have, will try to find. Sure something is out there. Added to my to do list.
