1) Identify from previous pset value whether prop was an enum (by checking if previous value is a list), instead of searching through prop templates
2) Compare previous and new enum values - if they match, skip editing IFC.
1) If object had multiple representations for the same context, then switch simply wouldn't work as it would always pick up the first representation it could find. Now it prioritizes the one user is selecting in UI.
2) If some object is selected and it also has multiple representations for the same context, then switch representation would switch it to any representation in that context it first finds. As user doesn't provide representation explicitly in that case, now it will stick to the already active representation on selected object if it matches the context, to avoid accident switches.
Notcied by working with example from #5824
- The patch was failed when executed from Bonsai since self.src wasn't provided, fixed now.
- Added a temporary hack to diplsay file selector for this patch.
- small refactor.
It was only checking if it's existing path but wasn't checking whether it's a file, therefore directories with suffixes were slipping in and then failing during `IfcHeaderExtractor(filepath)`.
Mentioned in logs from #5820
Turned out, `area.spaces` also contain all other spaces this area used (e.g. it used be view3d before it became a nodegraph) leading to errors. Simplifying it with `area.spaces.active`.
the way it was intended to originally but absolute paths of ifc and ids got in the way
Replaced TemporaryDirectory with mkdtemp as sometimes .html report doesn't open fast enough and directory gets deleted. Also need to keep the directory to make sure .ods is not removed.
The titleblock is ifc relative, not .blend/cwd relative.
e.g. if you open Blender in C:\ and then would open .ifc from c:\path\to\project.ifc, it would try to create layouts folder in C:\layouts instead of c:\path\to\layouts