[X3D-Ecosystem] RawKee Update - Implemented a companion export script for x3d.py

Wed Mar 5 10:12:32 PST 2025

I pasted my JSON text into the code extreme tool… but I don’t see a button to pus to execute the validation… am I missing something?

How do I get it to perform the validation?

From: John Carlson <yottzumm at gmail.com>
Sent: Wednesday, March 5, 2025 12:05 PM
To: Bergstrom, Aaron <aaron.bergstrom at und.edu>
Cc: X3D Ecosystem public discussion <x3d-ecosystem at web3d.org>
Subject: Re: [X3D-Ecosystem] RawKee Update - Implemented a companion export script for x3d.py

My suggestion is to validate your code here:

https://coderextreme.net/X3DJSONLD/src/main/html/validator.html

before using jsonlint.  Jsonlint is used for figuring out location of syntax errors.  X3D is more nuanced, and I’m not sure that Sunrize completely tests everything completely.  Vince pointed out that Sunrize may ignore invalid fields, similar to HTML5.

Good luck,

John

On Wed, Mar 5, 2025 at 10:57 AM Bergstrom, Aaron <aaron.bergstrom at und.edu<mailto:aaron.bergstrom at und.edu>> wrote:
John,

Within the RawKee plugin, I’ve implemented the export functions in two stages.

  *   Stage One - the Maya data gets stored in the x3d.py Scene.
  *   Stage Two – the X3D contents get saved to disk.

I did this because I have future plans to offer additional functionality before the information is written to disk.

While I haven’t yet fully written the code to export the HAnimHumanoid node, I’ve run some tests. In my testing of HAnim data storage during Stage One, I store the Maya joint nodes as DEF’d HAnimJoint nodes in the ‘skeleton’ field first before I add them to the “joints” field as USE’d HAnimJoint nodes.

And then at the time of export, during Stage Two (at the very end of the field sorting function in the RKSceneTraversal.py script), I flip the position of the ‘joints’ and ‘skeleton’ fields in the array in which I hold the names of the fields for the node that is about to be fed to the ‘processSortedNode()’ method – which actually begins the process of writing the node to disk.

The Stage Two field flipping code is fairly simple, so I put it here – where ‘nType’ is the type of X3D node that is about to exported and ‘mNodeList’ is the list of fields of that X3D node that contain multiple X3D nodes. The ‘HAnimHumanoid’ node type is the only node type in the script that receives special treatment. The script is as follows:

if nType == "HAnimHumanoid":
jIdx = -1
sIdx = -1
for idx in range(len(mNodeList)):
if mNodeList[idx] == "joints":
jIdx = idx
elif mNodeList[idx] == "skeleton":
sIdx = idx
if sIdx != -1 and jIdx != -1:
mNodeList[jIdx] = "skeleton"
mNodeList[sIdx] = "joints"

While I haven’t finished the code for HAnimHumanioid export during Stage One, the method described above aligns with how I exported HAnim content in the old C++ version of RawKee, so I am fairly confident that it will work as intended.

I have verified that the JSON encoding produced by the RKSceneTraversal.py script is valid JSON by pasting the code into the online JSON Validator found here:
https://jsonlint.com/

I have verified that the scene graph is valid by loading the resulting x3dj file into Sunrize.

So as far as I can tell, the JSON code is correct, both for JSON validity and X3D scene graph structure… at least for the node types that RawKee is currently able to export.

The ‘containerField’ field value is added as the custom Python attribute ‘_RK__containerField’ to X3D node during Stage One, and is queried during Stage Two when exporting to the XML encoding. ‘x3d.py’ has not be altered in any way.

For VMRL and JSON encodings, the ‘_RK__containerField’ attribute is ignored.

The missing ‘mapping’ field of TextureTransform is handled in a similar manner. During Stage One, where I should be able to do something like:
TextureTransform.mapping = “somestring”

I instead do:
TextureTransform._RK__mapping = “somestring”

And then during Stage Two I query ‘_RK__mapping’ when it appears in the field list.

The field list for each node is produced by querying the Python dictionary for that node using the following code:
nDict = vars(node)
for key, value in nDict.items():
              …some more code…

-- All of that said, I am a bit surprised at how well the RKSceneTraversal.py script works. I started out hard coding the export for each node type, and when I realized that was going to take me forever and ever and ever to implement, I decided there had to be a way I could implement it generically.

I expected it to take me several weeks of troubleshooting to get it right, but surprisingly, after I changed my approach, I was able to get it to export correctly after only two days of work and in less than 700 lines of code.

It relies fairly heavily on x3d.py, so I would not be able to get it to work with such a small amount of code without x3d.py. But from what I can tell, at this point it looks to me like it should work with every X3D node type in x3d.py.

Also, I had a real hard time figuring out the JSON encoding, but Sunrize does a very good job of this. So studied x3dj files created by Sunrize and did my best to mimic this structure in the RawKee JSON export. Doing so went surprisingly well. Sunrize has been key to helping work through many exporter development problems, and verify that my exported encoding actually functions correctly. I think without it, I would be lost.

But I do make a lot of coding errors, so it’s quite possible that there are use cases where my code fails. If anyone has the time and inclination, I would appreciate it if others could take a look at the RKSceneTraversal.py script and provide additional feedback. It can be found here:
https://github.com/und-dream-lab/rawkee/blob/main/rawkee/RKSceneTraversal.py

Thanks,

Aaron

From: John Carlson <yottzumm at gmail.com<mailto:yottzumm at gmail.com>>
Sent: Wednesday, March 5, 2025 4:03 AM
To: X3D Ecosystem public discussion <x3d-ecosystem at web3d.org<mailto:x3d-ecosystem at web3d.org>>
Cc: Bergstrom, Aaron <aaron.bergstrom at und.edu<mailto:aaron.bergstrom at und.edu>>
Subject: Re: [X3D-Ecosystem] RawKee Update - Implemented a companion export script for x3d.py

1.    JSON export was completely broken.
I tried fixing this in x3d.py and failed.  I have worked on python validators for JSON export, but it’s slow in Python.  Validating the scenegraph isn’t enough.
1.    The skeleton and joint export order of the HAnimHumanoid node fields using x3d.py would theoretically dump all the DEF’d joints into the ‘joints’ field, and thus only produce one USE’d root joint in the 'skeleton field. Though I haven’t implemented HAnimHumanoid yet, I re-ordered the field export so that the ‘skeleton’ field will get writen to the file before the ‘joints’ field.
You’ll also want to verify skin and skinCoord field order, plus sites and segments field order with skeleton.
1.    ‘containerField’ was not implemented for XML export, which caused the textures to not work because they weren’t allocated to the proper texture field within the Material node.
Andreas has a good patch for that.
1.    The ‘mapping’ field was missing from the TextureTransform object, so it wouldn’t export.

Good catch.

Also,  how did you solve the DEF/USE ordering issue?

I’ll post what i have for JSON elsewhere.

John
I will do a bit of a search for validation.
On Wed, Mar 5, 2025 at 3:39 AM Bergstrom, Aaron via X3D-Ecosystem <x3d-ecosystem at web3d.org<mailto:x3d-ecosystem at web3d.org>> wrote:
I was having issues with some of the export functionality of x3d.py.

I implemented a companion script. And wrote about it on the Castle Game Engine Website:
https://forum.castle-engine.io/t/rawkee-x3d-exporter-plugin-for-maya-repost/1804/3

Aaron
--
X3D-Ecosystem mailing list
X3D-Ecosystem at web3d.org<mailto:X3D-Ecosystem at web3d.org>
http://web3d.org/mailman/listinfo/x3d-ecosystem_web3d.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://web3d.org/pipermail/x3d-ecosystem_web3d.org/attachments/20250305/b7cd3564/attachment-0001.html>