[Plugin] Material_Maintenance v2.2 - 2013-01-13

[driven]

Yes, I did it manually to start and when the results appeared to go backwards I deleted those and used SU to install the .rbz
exactly the same results

BTW. if you can put a 20px margin around your html, I can attach web Inspector to look at the js, I need a crack to get in...

[driven]

!!!=> fromUIHandler: parameter string = 73-0|__vz|refreshMaterials|true|false|false p1 = >>73-0<< p1 = >>__vz<< p1 = >>refreshMaterials<< p1 = >>true<< p1 = >>false<< p1 = >>false<< calling function >>refreshMaterials<< String Time = 6.0e-05 Total Entities = 1
I changed the js and the ruby ¶ into |

[driven]

next error... !!!=> fromUIHandler: parameter string = 73-0|__vE|findComponents|·147713020 p1 = >>73-0<< p1 = >>__vE<< p1 = >>findComponents<< p1 = >>·147713020<< calling function >>findComponents<< Error: #<ArgumentError: invalid value for Integer: "·147713020">

[Myhand]

@driven said:

next error... !!!=> fromUIHandler: parameter string = 73-0|__vE|findComponents|·147713020 p1 = >>73-0<< p1 = >>__vE<< p1 = >>findComponents<< p1 = >>·147713020<< calling function >>findComponents<< Error: #<ArgumentError: invalid value for Integer: "·147713020">

Good work! This one is because the "·" char (183) char is used as an array token, and is also passed as a double byte char. So you would need to change that also.

BTW, I have just seen that the debug.rbz I attached earlier was not built properly. So it might well be that the fix I put in earlier will work, but I cannot get the build to work yet.

Appologies for wasting your time with testing that version!

[thomthom]

@myhand said:

@thomthom said:

SketchUp return strings in UTF-8. Anything above the ASCII character range will be multi-byte characters - which is easily mangled by Ruby 1.8's string manipulation methods as it treats 'characters' as single bytes.

But my characters are well within the ASCII range...? i.e. they still fit within a byte. from which value does it convert them to multi byte?

You seem to be using the ANSI range - you use bytes 182 and 183 which is not ASCII.
ASCII is just 0-127 - while ANSI is a sub-set that extends that with 128-255.

UTF-8 matches the ASCII character encoding - but does not match ANSI. In UTF-8 bytes 128-255 are control characters. So when you feed a UTF-8 system (like SketchUp) ANSI strings you get messed up characters.

¶ encoded in UTF-8 is [194, 182] - which is why when you split by just 182 you get the extra 194 byte ( Â).

[thomthom]

When I send parameters back from WebDialogs I've used combo characters like || or ||| which I know is very unlikely to appear in the data I send. (In the context of what I've done I've been sure they'd not appear.)

In a system where there is user input you'd want to create your own escape sequence where you pick a character as array separator and ensure to escape any occurrence in the data - which then is un-escaped when you receive it on the other end.

[TIG]

Another way is to restrict what a user can enter - e.g.
<input type="text" id="project_name" value="" style="font-size:7pt; width:300px;" onKeyup="(this.value=this.value.replace(/[^-_0-9A-Za-z]/,''));" ... >
This example limits the user's INPUT for the 'project_name' to letters, numbers, _, - etc... Then you can be sure that any 'separator' like '|' won't get replicated in the entered text... Obviously you need to change the regexp to suit your needs...

[Myhand]

@tig said:

Another way is to restrict what a user can enter - e.g.
<input type="text" id="project_name" value="" style="font-size:7pt; width:300px;" onKeyup="(this.value=this.value.replace(/[^-_0-9A-Za-z]/,''));" ... >
This example limits the user's INPUT for the 'project_name' to letters, numbers, _, - etc... Then you can be sure that any 'separator' like '|' won't get replicated in the entered text... Obviously you need to change the regexp to suit your needs...

yes I agree you always want to constrain at the input. In this case though the input in question is the Sketchup material editor, so I unfortunately have no control over that input.

[Myhand]

@thomthom said:

When I send parameters back from WebDialogs I've used combo characters like || or ||| which I know is very unlikely to appear in the data I send. (In the context of what I've done I've been sure they'd not appear.)

In a system where there is user input you'd want to create your own escape sequence where you pick a character as array separator and ensure to escape any occurrence in the data - which then is un-escaped when you receive it on the other end.

Escaping is the best way, and also what I am using in my Simple Jason serializer. I'll probably just implement the de-serialiser and then this whole problem will go away.

Your short cut of using ||| etc. is also a good option and I will probably use that in the short term if my latest attempt to fix the problem does not work.

[Myhand]

John, CMD, do you mind testing the attached version please?

The previous debug version did not change the JS code as I thought it did. This one does. It is the last try with the current control chars, and if this does not work I will create a patch with a multiple chars in the standard ascii range.

Again this requires a uninstall of the old plugin, restart of Sketchup and a install of the new debug version.

Again please send me the output.

Debug version

[driven]

same problem p1 = >>73-0Â<<

[thomthom]

You replaced teh character with String.fromCharCode(183) ?
And from what Driven says, that didn´t work either.

Why don´t you just split the param string with the proper UTF-8 sequence? You are after all dealing with UTF-8 within SketchUp - and also from the WebDialog. So to ensure you avoid encoding problems, which you have right now, just consistently treat everything as UTF-8. If you go back to where you had ¶ in your JS all you need to do is split the param string by the byte by the UTF-8 sequency 194, 182.

Trying to force it through a different encoding, in this case ANSI, is just asking for problems.

[thomthom]

Btw, did you see my comment at the end of the prevous page regarding ASCII vs ANSI vs UTF-8? (Not sure if you missed it, as my last comment will make less sense if you missed it.)
http://sketchucation.com/forums/viewtopic.php?f=323&t=48933&start=45#p442858

[Myhand]

Thank you everyone for your help and good advice. Here is a version with noth of the protocol chars in the normal ascii range. I have chosen 30, and 31 the record and unit separator characters. They are not printable so there is a small chance there might again be problems on the MAC I guess, but this works fine on Windows.

John, CMD do you mind trying again please? Debug is still on, so if this works I will publish a version without the tracing code.

Debug version

[driven]

the good news is

# SU returns  =>
 RUBY_VERSION = 1.8.5
(eval);3119; warning; parenthesize argument(s) for future version
(eval);3142; warning; parenthesize argument(s) for future version
key = Units
value = 0.0
key = StampOffset
value = 12.0
key = GridSpacingX
value = 120.0
key = SmooveRadius
value = 360.0
key = GridSpacingY
value = 120.0
!!!=> fromUIHandler; parameter string = 73-0__vzrefreshMaterialstruefalsefalse
p1 = >>73-0<<
p1 = >>__vz<<
p1 = >>refreshMaterials<<
p1 = >>true<<
p1 = >>false<<
p1 = >>false<<
calling function >>refreshMaterials<<
String Time = 5.9e-05
Total Entities = 1

# SU returns  =>
 RUBY_VERSION = 1.8.7
key = Units
value = 0.0
!!!=> fromUIHandler; parameter string = 73-0__vzrefreshMaterialstruefalsefalse
p1 = >>73-0<<
p1 = >>__vz<<
p1 = >>refreshMaterials<<
p1 = >>true<<
p1 = >>false<<
p1 = >>false<<
calling function >>refreshMaterials<<
String Time = 6.4e-05

seems to work with either RUBY_VERSION, but the verbosity level is higher in 1.8.5
john

[cmd]

@myhand said:

John, CMD, do you mind testing the attached version please?

The previous debug version did not change the JS code as I thought it did. This one does. It is the last try with the current control chars, and if this does not work I will create a patch with a multiple chars in the standard ascii range.

Again this requires a uninstall of the old plugin, restart of Sketchup and a install of the new debug version.

Again please send me the output.

Myhand,

Attached is the output immediately after launching the Material Maintenance plugin.

CMD

---

MM_Ruby output.txt

[TIG]

The text file when opened in 'Notepad++.exe' as 'UTF-8 without BOM' shows this: BUT when the same text is pasted into here it displays differently...
!!!=> fromUIHandler: parameter string = 73-0__vzrefreshMaterialstruefalsefalse Error: #<NoMethodError: private methodputs' called for "73-0":String>`
Presumably the 'RS' record-separator character [ASCII-030] is non-printing...
Not sure how 'puts' seems broken
Are you parsing the string using split and the 'RS' as at splitter ?
You should then be able to 'puts' the array elements ??

[Myhand]

@tig said:

The text file when opened in 'Notepad++.exe' as 'UTF-8 without BOM' shows this: [attachment=0:2y0lvhlv]<!-- ia0 -->Capture.PNG<!-- ia0 -->[/attachment:2y0lvhlv] BUT when the same text is pasted into here it displays differently...
!!!=> fromUIHandler: parameter string = 73-0__vzrefreshMaterialstruefalsefalse Error: #<NoMethodError: private methodputs' called for "73-0":String>`
Presumably the 'RS' record-separator character [ASCII-030] is non-printing...
Not sure how 'puts' seems broken
Are you parsing the string using split and the 'RS' as at splitter ?
You should then be able to 'puts' the array elements ??

Yes I am splitting the string with the 'RS' char. Very strange error message indeed. The code snippet where it fails is:

` puts "!!!=> fromUIHandler: parameter string = #{parameterString}";

  params = unmarshalParams(parameterString);
  
  params.each { |p|
    p.puts "p1 = >>#{p}<<";   // This is line 109
  }`

Again this works fine on Windows, and appears to now also work for driven on his MAC so at a bit of a loss here.

One other strange thing with this is that the output should look like this:

!!!=> fromUIHandler; parameter string = 75-0__materialsRefreshedrefreshMaterialstruefalsefalse
p1 = >>75-0<<
p1 = >>__materialsRefreshed<<
p1 = >>refreshMaterials<<
p1 = >>true<<
p1 = >>false<<
p1 = >>false<<
calling function >>refreshMaterials<<

i.e. the second parameter should be "__materialsRefreshed" not "__vz" which it appears to be for both the MAC implementations. Strictly speaking this should not work even for driven.

CMD, are you sure you have done a clean uninstall and install?

[TIG]

But params.each{|p|puts "p1 = >>#{p}<<"}
should be the way ?
NOT p.puts

[Myhand]

@tig said:

But params.each{|p|puts "p1 = >>#{p}<<"}
should be the way ?
NOT p.puts

thanks TIG!

Still strange it worked on windows and driven's MAC and that the second parameter comes looks wrong but as long as it works.

I am making a new build without the trace code which I will upload soon.