Fillet two edges of polyline the closest to a clicked point
@T2ioTD wrote:
.... if one first made polylines with hard corners, then (with a proper lisp) he can just click a point near two edges and the routine would fillet them to the radius of a circle passing by (or close?) to the point being clicked.
....
FilletPlineVert.lsp with its FPV command, available here, may work for you. You pick at a vertex, or at the midpoint of an intermediate segment [such as to re-Fillet a corner at a different radius] -- it sets END and MID Osnap modes for you to make that easy. It works on any Polyline, not just the latest-drawn. It's not clear to me whether your idea is that the Circle for the radius is already drawn, but FPV doesn't involve that. I suppose it could be altered to ask you to pick a Circle or Arc from which to get the radius, if you really want it to work that way. See additional comments at the top of and within the file.
@T2ioTD wrote:
.... the radius of the fillet would be calculated by the lisp as the distance from clicked point to the closest vertex of the polyline.
I'd be interested in an illustration, since it seems to me the results can be so different depending on the geometry [red are the clicked points]:
@john.uhden wrote:
Um, I missed finding the pt_p function, though I think we know what it would be, like maybe...
(defun pt_p (p) (or (= (mapcar 'type p) '(REAL REAL REAL)) (= (mapcar 'type p) '(REAL REAL)) ) ) ...
I think that needs two modifications:
The (=) function is for numbers and strings, not lists. That should be (equal).
If the variable contains something entirely inappropriate, such as a string or entity name, that doesn't just return nil, but causes an error.
This gets around both of those:
(defun pt_p (p)
(and
(listp p)
(or
(equal (mapcar 'type p) '(REAL REAL REAL))
(equal (mapcar 'type p) '(REAL REAL))
); or
); and
); defun
@john.uhden wrote:
....
(vl-every 'numberp pt) would work too, but it could include inetgers.
(defun realp (#)(equal (type #) 'REAL)) could work too, as in
(vl-every 'realp pt)
which would work for both 2D and 3D points.
Those would return T for a list containing only one number, or a list of twenty numbers -- they would need to be combined with testing for the length of the list if it's supposed to be a point [2 or 3 real numbers only].
Every time I see a significant amount of code repeated, I figure there's probably a way to simplify. Just playing around, here's one way, that uses those repeated tests only once, and does one test to determine whether it's either 2 or 3 numbers long:
(defun 3dp (p); returns XYZ list from XY or XYZ point list, otherwise nil
(if (and p (listp p) (< 1 (length p) 4) (vl-every 'numberp p))
(append p (nth (length p) '(0 0 (0.0) nil)))
)
)
It doesn't process in a different way depending on whether it's a 2-coordinate or 3-coordinate point list. That difference just determines whether what it appends is 0.0 [to make an XY list into an XYZ list] or nothing [to leave an XYZ list alone]. Come to think of it, that (append) line could be a little shorter:
(append p (if (= (length p) 2) '(0.0) nil))
or shorter still, the version that does process in a different way, but has to repeat the p to feed it out if unappended:
(if (= (length p) 2) (append p '(0.0)) p)
