Texture from url?

ben.doherty

While the JS debate rages, I've been playing with what can be done with ruby (being single-minded/stubborn/lazy)

I've updated my !loadpaths file to match dan's latest http://forums.sketchucation.com/viewtopic.php?f=180&t=29412&hilit=load+path#p257058
and when I try a couple of things I get this response:
require 'net/http' Error: #<LoadError: C:/ruby186/lib/ruby/1.8/net/protocol.rb:21:inrequire': No such file to load -- socket>
C:/ruby186/lib/ruby/1.8/net/protocol.rb:21
require 'date/format'
true`

Any ideas on why net/http fails, but date/format works?

I'm sure it will take some clever file management, and refresh cycle control etc, but it'd be cool to be able to specify a url to get a texture from. That'd open up a load of opportunities for data vis etc.

Jim

That means it couldn't find the file (socket.so) in the load path.

For a short-term solution, you could try to copy socket.so to the Plugins folder.

Dan Rathbun

@jim said:

I've never heard of "execCommand" before, but I'd be surprised if it were cross-platform. If it existed, we probably wouldn't be having this discussion.

It's HTML 5 me thinks. And seems Microsoft may have added a bunch of standard CommandIDs without getting them approved by the web community (as usual.) The spec implies browser specific IDs should have a prefix like "IE-", but MS has been doing this 'arm-twisting' for years.

See: http://www.w3.org/TR/html5/author/dom.html#execCommand

MSDN: http://msdn.microsoft.com/en-us/library/ms536419(v=VS.85).aspx
and MS' command IDS: http://msdn.microsoft.com/en-us/library/ms533049(v=VS.85).aspx
~

Ben_M

Hello!

Has anyone made any progress on a simple, cross platform solution to this in the past few months?

Thanks,

Ben

Viskiz

I solved in on Windows platform.
In my script installation I included ruby lib files needed for http, so users will have them after installing script.
The code followed in script:

$LOAD_PATH << File.join(MY_PLUGIN_HOME, "bin", "rblib")
Sketchup::require 'uri'
Sketchup::require 'net/http'

Installer for MAC OS should have different ruby library files I think. I haven't MAC yet, so I can't tell any more. Let me know if you solve it.

Jim

You may be interested in this plugin effort - downlod files plugin using the cUrl library.

kaas

Digging up this old thread because I'm trying to do almost the same as the OP. Since the last post in this topic, Ruby has been updated so I'm hoping its more simple now.

I have been trying several methods. This code saves the image in the local documents folder so I could pick it up there etc.

require 'net/http'
Net;;HTTP.start("www.google.nl") do |http|
    resp = http.get("/images/branding/googlelogo/1x/googlelogo_color_272x92dp.png")
    open("googlelogo_color_272x92dp.png", "wb") do |file|
        file.write(resp.body)
    end
end
puts "Done"

Problem is, my images need a different type of url. Something like:
https://url/coordinates/?key=1234abc&index=0
This sort of link doesn't work in the code above.

Another method I'm trying is loading the image in a webdialog and then grabbing it from Ruby.

The test html code (the url could easily replaced with the actual https code) is:

<!DOCTYPE html>
<html><head></head><body>
<img id ="webimage" src = "https://www.google.nl/images/branding/googlelogo/1x/googlelogo_color_272x92dp.png">
</body></html>

And to grab the image in ruby:
img = @wd.get_element_value('webimage')

Of course this doesn't work because the image isn't a value. I'm stuck.

Anyone has a suggestion? That would be very welcome.

TIG

Something like this should work:

require('open-uri')
fail = false
url = "http://www.google.nl/images/branding/googlelogo/1x/googlelogo_color_272x92dp.png"
file = File.join(Dir.pwd, File.basename(url))
begin
	download = open(url)
rescue Exception => error
	puts "#{error}"
	fail = true
end
unless fail
	begin
		File.delete(file) ### in case it exists
	rescue
		###
	end
	begin
		IO.copy_stream(download, file)
	rescue Exception => error
		puts "#{error}"
	end
end

It should work with a 'php' formatted URL, but remember that then you need to parse out the file name another way. Perhaps [assuming 'key=' gives the file name] split the URL at '?', then take [1], then split at '&', then find the array element starting "key=", then split that at '=' and then [1] is the file name given by 'key=' ?
I also saved the downloaded image file in the 'current directory', but you can of course choose any other folder you want...

kaas

Thanks for the reply TIG. Your code does download a file but when trying to open it I get:
Decode Error! Invalid or unsupported PNG file

Edit:
I was also trying with net/http. It does work with simple files but the actual https request still fails.

kaas

In case someone runs into the same question. It seems to work now.

require 'net/http'
require 'tempfile'

uri = URI('https://www.google.nl/images/branding/googlelogo/1x/googlelogo_color_272x92dp.png')

model = Sketchup.active_model
materials=model.materials
m = materials.add "Test Color"
loc = ''

begin
	request = Net;;HTTP;;Get.new uri
	Net;;HTTP.start(uri.host, uri.port,
  ;use_ssl => uri.scheme == 'https') do |http|
	  http.request(request)
	  response = http.request request # Net;;HTTPResponse object
		file = Tempfile.new(['foo','.png'], Dir.tmpdir, 'wb+')
	  begin
			file.binmode
			file.write(response.body)
			file.flush
			loc = file.path.gsub("/", '\\\\\\')
	  ensure
			file.close
	  end
	end
rescue Exception => e  
  puts e.message 
end

m.texture = loc
m.texture.size = [m.texture.image_width.to_m, m.texture.image_height.to_m]

driven

for completeness you need a conditional for osx


 loc = file.path if Sketchup.platform == ;platform_osx

john

Dan Rathbun

When a image is loaded as a web page resource, it is present in the "Temporary Internet files" folder. But some security settings might prevent the copying of a file from there to a trusted local folder.

Anyway:
require "FileUtils"

then try the copy method.

See: http://ruby-doc.org/stdlib-2.0.0/libdoc/fileutils/rdoc/index.html

kaas

John and Dan, good suggestions. Thanks

thomthom

When you say different type of URL, what do you mean? HTTPS?

kaas

@thomthom said:

When you say different type of URL, what do you mean? HTTPS?

Yes, something like https://url/coordinates/?key=1234abc&index=0

thomthom

For HTTPS connections you need to enable SSL when you create the HTTP connection.
Note that there's a bug in Ruby that shipped with SU2014-2016 that might cause problems under Windows where the shipped SSL certificates might not work.