Convert texts that are close to each other into Mtext
@pendean wrote:
Have you tried TXT2MTXT command ....
That's not going to do it:
They're looking to convert adjacent pairs of Text objects into separate two-line Mtext objects, but by overall collective selection.
I suspect it's going to require stepping through the Text objects and evaluating how far they are from all other Text objects, with some kind of criteria about how close is close enough. @oli-123 , are they always pairs, or might there sometimes be more than 2 close together that you want converted into one Mtext object?
And even finding the correct pairings or groupings, TXT2MTXT does odd things if the rotation is not zero:
@oli-123 wrote:
.... the numbers I want to convert into Mtext are always in pairs, so there won't any with 3 texts.
....
Give this a try [it worked for me in your sample drawing]:
(defun C:2T2M ; = 2 nearby Text objects {to} two-line Mtext
(/ ss1 T1 minpt maxpt LL UR ss2 T2 data1 data2 txt1 txt2)
(if (setq ss1 (ssget "_:L" '((0 . "TEXT"))))
(while (> (sslength ss1) 1); still at least 2 Text objects remaining
(setq T1 (ssname ss1 0))
(vla-getboundingbox (vlax-ename->vla-object T1) 'minpt 'maxpt)
(setq
LL (vlax-safearray->list minpt)
UR (vlax-safearray->list maxpt)
); setq
(if
(and ; finds 2 Text objects [including T1] in vicinity
(setq ss2
(ssget
"_C"
(polar LL (angle UR LL) (distance LL UR))
(polar UR (angle LL UR) (distance LL UR))
'((0 . "TEXT"))
); ssget
); setq
(= (sslength ss2) 2)
); and
(progn ; then
(setq T2 (ssname (ssdel T1 ss2) 0))
(vla-getboundingbox (vlax-ename->vla-object T2) 'minpt 'maxpt)
(setq ; collective bounding box of both
LL (mapcar 'min (vlax-safearray->list minpt) LL)
UR (mapcar 'max (vlax-safearray->list maxpt) UR)
data1 (entget T1)
data2 (entget T2)
txt1 (cdr (assoc 1 data1))
txt2 (cdr (assoc 1 data2))
); setq
(command "_.ucs" "_object" T1)
(if (minusp (cadr (trans (cdr (assoc 10 data2)) 0 1))); T2 "under" T1
(setq txtA txt1 txtB txt2); then
(setq txtA txt2 txtB txt1); else
); if
(command
"_.ucs" "_previous"
"_.mtext"
"_non" LL
"_height" (cdr (assoc 40 data1))
"_rotation" (cdr (assoc 50 data1))
"_style" (cdr (assoc 7 data1))
"_justify" "_mc" "_non" UR
(strcat txtA "\\P" txtB) ""
"_.matchprop" T1 (entlast) "" ; Layer, color etc. override(s)
"_.erase" T1 T2 ""
); command
(ssdel T1 ss1) (ssdel T2 ss1)
); progn
(ssdel T1 ss1); else [no other nearby]
); if
); while
); if
(princ)
); defun
It can very slightly shift the positions, because its line spacing is as defined in the font used, which can be different from the spacing between the original Text objects [which is not the same everywhere in the sample drawing], and also because the bounding box is affected even by the particular text content and the rotation.
It could have unexpected results if arrangement is ever non-standard, such as if the pairs do not match in angle, which could mean it might read which is on top differently from what you would expect. Or if they don't match in other properties -- it uses the one called T1 for all the characteristics [Style, Rotation, Height, Layer and any property overrides], but that could be either the upper or the lower one in a given pair, depending on the method of selection and/or the drawing order.
It could use an *error* handler to ensure setting the UCS back, and command-echo suppression, and the other typical things, but first see whether it does what you want.
That may be possible. If the situation is similar to the OP's, but with groups of 3 instead of 2, then the "reaching out" around a given Text object in the overall selection would need to reach farther out to find the associated others, in case the one it's reaching out from is not the middle one in a group of 3. I would wonder whether some sets might be close enough together that more would be found than intended, and would need to think about how to process only those within a certain kind of relationship. Do you have a [small] sample drawing showing likely arrangements?
A complex task may be able to be done as the purple text is on a different layer than the other text. So maybe a step by step approach using ssget "f" once nothing is found both directions then stop. The text must be approx vertical aligned not a mix of angles maybe version 5.
Dinner time now maybe Kent will jump in if he thinks my idea will work, happy to hear of any other way to approach.
@kwankitfu wrote:
Here is the sample drawing. in fact the data given were usually in grouping of 2 ~ 5 text.
Clarify....
The groupings are not of 2 to 5 Text objects as described above and in the drawing, but of one Text object [on the UNIT-ADDRESS Layer] and one or more Mtext objects [on the CPNO-... Layer]. Do you, in fact, want both kinds combined into the Mtext results? It would seem necessary for two of your circled groups containing only two, one of each kind.
And is it acceptable that they will afterwards, of course, be on one Layer -- the distinction between the Layers of the source objects will be lost?
@Sea-Haven 's suggestion of using Fence selection has the possible drawback that if the Fence line happens to pass through a gap between characters, the Text/Mtext object will not be found. So I think a bounding-box-based selection would be better, likely a Crossing window built to extend above and below that of a UNIT-ADDRESS Text object to find the nearby CPNO-... Mtext objects. Are they always "safely" separated from other groups as in the drawing? And are they always at zero rotation?
@Sea-Haven wrote:
Removed did not work with more testing.
Removed what? Removed it from what?
@radu_609 wrote:
.... convert 3 TEXTs within a radius of 10 units to a MTEXT ....
Define "within a radius." Do I assume correctly that you mean within that kind of distance from each other, without an actual Circle drawn to find Text inside of? If so, do you mean entirely within a virtual Circle like that [I'm having trouble thinking of how a routine could determine that], or with their insertion points all within no more than 20 units of each other regardless of the actual extent of the Text objects:
or .... ?
And what do you have in mind for selection of them? More than one such grouping User-selected collectively? Routine finds them all without User input? Only one such grouping, User-selected with the routine only checking whether they qualify as "close to each other"?
