Plugin search help -- combining lines

[TIG]

Weld makes the selected edges into a bound set of edges in a 'curve' - they are still there with ends.

Thomthom's EdgeTools can heal colinear lines.........

[agamemnus]

Hmm, that works, but I have to specifically select those edges (can't just select basic polys/edges in whole group, either) and then use "colinear from start to end"... (it also repositions the texture in some fashion.) Won't help with split edges that I didn't know existed.

[TIG]

If this relates to the PM about a 'rectangle with 5 edges after a joined edge is deleted'... simply make sure nothing is selected, activate the Move Tool, hover over the rogue vertex location [use a Style showing EndPoints so its easier to spot them] then move the vertex [NOT the edge] so the short part is 'shrunk' to zero length - you might need Alt to get auto-fold to kick in in some cases... Now you have lost an unnecessary edge and vertex, and the rectangle has 4 sides as it ought.

[Builder Boy]

If the lines are all going the same direction, then you can use this:

---

repair_broken_lines.rb

[agamemnus]

Thanks Builder Boy. That hit the spot.

[agamemnus]

A problem with that script is that it does not seem to always dig inside groups/components correctly and misses some lines.

Edit: tried this.... doesn't work. I have a feeling I am missing something.

Edit 2: Used "if (e.class != Sketchup::Face)" instead of
"if (e.class == Sketchup::Group)"... why does one work and not the other? (the == group one doesn't seem to hit?) Also... any better solutions here? This is the best I could cobble up..

require 'sketchup.rb'
# (c) Carlo Roosen 2004

def repair_broken_lines (eparent)
  total = 0
  trunk = Array.new
  ss = eparent.entities
  #search all edges in selection set
  for e in ss
    if (e.kind_of?(Sketchup;;Edge))
      p=e.parent
      #take both vertices 
      for v in e.vertices 
        #take all edges connected to v (including e)
        es=v.edges
        #when two edges are connected
        if (es.length==2)
          vec1 = es[0].line[1]
          vec2 = es[1].line[1]
          #and both are in the same direction
          if (vec1.parallel?(vec2))
            #then the two lines can be repaired
            #this is done by generating a new line from the vertex to a random point
            l = p.entities.add_line (v.position, Geom;;Point3d.new(3.45,6.78,9.12))
            #and put that line in a trunk
            trunk.push(l)
          end
        end
      end
	else
	 if (e.class != Sketchup;;Face)
	  total += repair_broken_lines (e) 
	 end
	end
  end

  #erase all lines in trunk
  total += trunk.length
  for l in trunk
    l.erase!
  end
  return total
end

def repair_broken_lines_recurse

 total = 0;
 m = Sketchup.active_model;
 
 # Define everything as one undo operation.
 m.start_operation "Repair Broken Lines"
 total = repair_broken_lines (m)
 UI.messagebox ("#{total} lines repaired")
 m.commit_operation
end


selection=Sketchup.active_model.selection
if( not file_loaded?("repair_broken_lines.rb") )
    UI.add_context_menu_handler do |context_menu| 
        if selection.count>1
           context_menu.add_item("Repair Broken Lines") { repair_broken_lines_recurse }
        end
    end
end

UI.menu("Plugins").add_item("Repair Broken Lines") { repair_broken_lines_recurse }

file_loaded ("repair_broken_lines.rb")

[TIG]

Here's my working version...

### Original idea (c) Carlo Roosen 2004
### Modified (c) TIG 2011 
### Now also repairs edges inside groups/instances in selection...
###
require 'sketchup.rb'
###
class Repair_broken_lines
 def initialize(ents=nil)
  model=Sketchup.active_model
  lines_to_go=[]
  if not ents
    $total_repair_broken_lines=0
    ss=model.selection.to_a
    begin
      model.start_operation("Repair Broken Lines",true)
    rescue
      model.start_operation("Repair Broken Lines")
    end
  else
    ss=ents
  end#if
  #check all edges in selection set OR entities passed as argument
  ss.each{ |e|
    if e.kind_of?(Sketchup;;Edge) #take both vertices 
      e.vertices.each{ |v| #take all edges connected to v (including e)
        es=v.edges #when two edges are connected
        if es.length==2
          vec1=es[0].line[1]
          vec2=es[1].line[1] #and both are in the same direction
          if vec1.parallel?(vec2) #then the two lines can be repaired
            #make a new line from that vertex to a random point
            pt=Geom;;Point3d.new(1+rand, 1+rand, 1+rand)
            nline=e.parent.entities.add_line(v.position, pt)
            lines_to_go << nline if nline and nline.valid?
          end#if
        end#if
      }#end v
    elsif e.kind_of?(Sketchup;;Group)
      Repair_broken_lines.new(e.entities.to_a)
    elsif e.kind_of?(Sketchup;;ComponentInstance)
      Repair_broken_lines.new(e.definition.entities.to_a)
    end#if
  }#end e
  if not $total_repair_broken_lines
    $total_repair_broken_lines=lines_to_go.length
  else
    $total_repair_broken_lines=$total_repair_broken_lines + lines_to_go.length
  end#if
  lines_to_go.each{ |e| e.erase! if e.valid? }#erase all lines_to_go
  if not ents
    model.commit_operation 
    UI.messagebox("#{$total_repair_broken_lines} Lines Repaired.")
  end#if
 end#def initialize
end#class
###
if not file_loaded?(File.basename(__FILE__))
  if $submenu4 # TIG menus
    $submenu4.add_item("Repair Broken Lines [Selection]"){ Repair_broken_lines.new() }
  else
    UI.menu("Plugins").add_item("Repair Broken Lines [Selection]"){ Repair_broken_lines.new() }
  end
end
file_loaded(File.basename(__FILE__))
###

[agamemnus]

Thanks. Looks like you made total_repair_broken_lines a global. That makes more sense than to create a useless stack as I did. I didn't want to create a stack either but I didn't want to create a global, and I am not 100% familiar with Ruby/am a bit forgetful. I did some searching now though and I see that @@ is class scope, so how about @@ instead of $?

If the code stays here people won't be able to find it very easily. How about re-releasing this plugin, as well as that "default face" snippet you made earlier for me, in the plugins forum?

Also: another problem I noticed is that it tries to repair lines that aren't repairable, like curve endings and single lines. (attached)

    ### Original idea (c) Carlo Roosen 2004
    ### Modified (c) TIG 2011
    ### Now also repairs edges inside groups/instances in selection...
    ###
    require 'sketchup.rb'
    ###
    class Repair_broken_lines
    def initialize(ents=nil)
      model=Sketchup.active_model
      lines_to_go=[]
      if not ents
        @@total_repair_broken_lines=0
        ss=model.selection.to_a
        begin
          model.start_operation("Repair Broken Lines",true)
        rescue
          model.start_operation("Repair Broken Lines")
        end
      else
        ss=ents
      end#if
      #check all edges in selection set OR entities passed as argument
      ss.each{ |e|
        if e.kind_of?(Sketchup;;Edge) #take both vertices
          e.vertices.each{ |v| #take all edges connected to v (including e)
            es=v.edges #when two edges are connected
            if es.length==2
              vec1=es[0].line[1]
              vec2=es[1].line[1] #and both are in the same direction
              if vec1.parallel?(vec2) #then the two lines can be repaired
                #make a new line from that vertex to a random point
                pt=Geom;;Point3d.new(1+rand, 1+rand, 1+rand)
                nline=e.parent.entities.add_line(v.position, pt)
                lines_to_go << nline if nline and nline.valid?
              end#if
            end#if
          }#end v
        elsif e.kind_of?(Sketchup;;Group)
          Repair_broken_lines.new(e.entities.to_a)
        elsif e.kind_of?(Sketchup;;ComponentInstance)
          Repair_broken_lines.new(e.definition.entities.to_a)
        end#if
      }#end e
      if not @@total_repair_broken_lines
        @@total_repair_broken_lines=lines_to_go.length
      else
        @@total_repair_broken_lines=@@total_repair_broken_lines + lines_to_go.length
      end#if
      lines_to_go.each{ |e| e.erase! if e.valid? }#erase all lines_to_go
      if not ents
        model.commit_operation
        UI.messagebox("#{@@total_repair_broken_lines} Lines Repaired.")
      end#if
    end#def initialize
    end#class
    ###
    if not file_loaded?(File.basename(__FILE__))
      if $submenu4 # TIG menus
        $submenu4.add_item("Repair Broken Lines [Selection]"){ Repair_broken_lines.new() }
      else
        UI.menu("Plugins").add_item("Repair Broken Lines [Selection]"){ Repair_broken_lines.new() }
      end
    end
    file_loaded(File.basename(__FILE__))
    ###

---

circle.skp

[TIG]

You are right about setting the variable to 'class' using @@ rather than 'global' with $ - it's much safer - I was churning that out rather too quickly...
The 'trick' to force broken edges into one only works on plain edges.
With an edge that's part of a curve it can't... for you next step why not see if you can find all edges that are co-linear [it already does that and if one is part of a curve then remember all of the curves edges in an array and then explode the curve, immediately remake the curve leaving out the one edge you want to heal. Then it will heal back into its co-linear mate... It just needs a test in the iterated loops...
When 'we' get it 'perfect' then 'we' could issue a joint 3-way update...

[Rich O Brien]

You mean 4-way

[TIG]

@unknownuser said:

You mean 4-way

Like "Additional semi-humorous material supplied by Rich O'Brien."

[Rich O Brien]

If you mean that 'Disconbobulator' isn't a good name for a plugin then so be it. It may not be a impressive as 'Quadrilateralizer' but its close!

[TIG]

@unknownuser said:

If you mean that 'Disconbobulator' isn't a good name for a plugin then so be it. It may not be a impressive as 'Quadrilateralizer' but its close!

I wanted something to explain what it did in the name. RickW's 'Windowizer' makes Windows, my 'Latticeizer' makes Lattices and 'DeBabelizer' translates text [Babelfish/TowerOfBabel etc]... When you make triangles out of a face you 'Triangulate' it BUT when you make a face into quadrilaterals you don't say you 'Quadrilateralate' it [that was an early alternative name!]; an '-izer' ending suggest 'doing' so 'Quadrilateralizer' it became... 'Quadrilateralizer' gets 635 hits on Google, 235 of them when it's combined with 'TIG' - strangely most are Chinese

[Gaieus]

Well, at least (here, on the forums) we do not have to pronounce that quadri... plugin.

Back in Mountain View in 2008, I even had issues with the Protractor tool (certainly Rich will chime in with some ideas about drinking less the night before and such I am sure)

[agamemnus]

@tig said:

You are right about setting the variable to 'class' using @@ rather than 'global' with $ - it's much safer - I was churning that out rather too quickly...
The 'trick' to force broken edges into one only works on plain edges.
With an edge that's part of a curve it can't... for you next step why not see if you can find all edges that are co-linear [it already does that and if one is part of a curve then remember all of the curves edges in an array and then explode the curve, immediately remake the curve leaving out the one edge you want to heal. Then it will heal back into its co-linear mate... It just needs a test in the iterated loops...
When 'we' get it 'perfect' then 'we' could issue a joint 3-way update...

Ummm, maybe later. Doesn't need to be perfect...

Anywayz ... here is another problem: a set of connected lines were 2 of them are said to be repaired, but never are. I attached the model. I think the tolerance is too low perhaps?

---

aline.skp

[TIG]

The edges in the 'line' are not co-linear, the first two edges' vectors differ noticeably and the rest diverge even more... Vector3d(0.0104132, 0.999946, 0) and Vector3d(0.0101495, 0.999948, 0)... so they won't be joined by Sketchup into one edge no matter how hard you try - but you could of course 'weld' them into a 'curve'......

[agamemnus]

@tig said:

The edges in the 'line' are not co-linear, the first two edges' vectors differ noticeably and the rest diverge even more... Vector3d(0.0104132, 0.999946, 0) and Vector3d(0.0101495, 0.999948, 0)... so they won't be joined by Sketchup into one edge no matter how hard you try - but you could of course 'weld' them into a 'curve'......

Yeah, I see...but.... the plugin says they are... or thinks they are...

[TIG]

@agamemnus said:

@tig said:

The edges in the 'line' are not co-linear, the first two edges' vectors differ noticeably and the rest diverge even more... Vector3d(0.0104132, 0.999946, 0) and Vector3d(0.0101495, 0.999948, 0)... so they won't be joined by Sketchup into one edge no matter how hard you try - but you could of course 'weld' them into a 'curve'......

Yeah, I see...but.... the plugin says they are... or thinks they are...
I expect that's because the 'parallel?' test for vectors returns true but Sketchup itself still won't join the lines as they are not co-linear...
We could replace the test with something else as the two vectors are clearly not equal - perhaps comparing them as arrays - so v1.to_a==v2.to_a returns false [correctly] even when v1.parallel?(v2) returns true !!

[agamemnus]

@tig said:

I expect that's because the 'parallel?' test for vectors returns true but Sketchup itself still won't join the lines as they are not co-linear...
We could replace the test with something else as the two vectors are clearly not equal - perhaps comparing them as arrays - so v1.to_a==v2.to_a returns false [correctly] even when v1.parallel?(v2) returns true !!

I see... so it's really a Sketchup bug. "parallel" is built-in, (right?), but it doesn't have the same precision as the main program, which is a decently-sized oversight.

In terms of comparing them as arrays: that's probably slower, but it might have the same comparison as the main program. (I don't know)

I'm not entirely sure how Sketchup does precision anyhow... floats (32 or sometimes64 bit numbers with decimal points)? Really big integers masquerading as floats? Some weird combination? There's a big difference... two string representations of a float will often look different but actually have the same float value. I suspect something like that is happening here.

[agamemnus]

Bump!

I'm going to forget that TIG's code/solution existed in 2 months and re-post my question...