When I use python to batch merge models in Blender using a hierarchical structure, the software always crashes

Question

class TestFunction(bpy.types.Operator):
    """TestFunction"""
    bl_idname = "object.test_function"
    bl_label = "Test Function"
    bl_options = {'REGISTER', 'UNDO'}
    
    def merge_hierarchy(obj):
        """
        递归地合并层级结构中的所有子对象
        """
        # 如果对象没有子对象，直接返回
        if not obj.children:
            return
    
        # 找到所有非EMPTY的子对象
        objects_to_merge = [child for child in obj.children if child.type != 'EMPTY']
    
        # 如果有多个非EMPTY的子对象，合并它们
        if len(objects_to_merge) > 1:
            # 选择所有要合并的对象
            for child in objects_to_merge:
                child.select_set(True)
            # 设置第一个对象为活动对象
            bpy.context.view_layer.objects.active = objects_to_merge[0]
            # 执行合并操作
            bpy.ops.object.join()
            # 重新命名合并后的对象为父节点的名称
            merged_object = bpy.context.active_object
            merged_object.name = obj.name
            # 清除选择
            bpy.ops.object.select_all(action='DESELECT')
    
        # 递归处理每个子对象
        for child in obj.children:
            TestFunction.merge_hierarchy(child)
    
    def execute(self, context):
        
        """
        合并所有场景中的层级结构
        """
        # 获取所有顶层对象（没有父对象的对象）
        top_level_objects = [obj for obj in bpy.data.objects if obj.parent is None]
    
        for obj in top_level_objects:
            TestFunction.merge_hierarchy(obj)
        return {'FINISHED'}

Whenever I run the aforementioned code to process models in an engineering file, the software consistently crashes after a while. The scene contains about 6,000 models. I'm using Blender version 3.4.1, which should be paired with Python version 3.10.8. However, when I select multiple objects in Blender and right-click to merge them, the software completes the operation swiftly. Why does it crash when these actions are turned into code for batch execution? I monitored the CPU usage during code execution and noticed a peak utilization rate of 80%. The rate typically fluctuated between 20% and 45%, with significant variations. Despite this, the software remained stable. However, after continued operation, it unexpectedly crashed without prior indications.(The running computer CPU is i9-13900ks) There are many models under the empty object node I selected on the right. I want to merge all models under each hierarchical structure into a single one, ensuring only one object remains under each node.

Answer 1

I'm not sure, but though it seems functional, it could be made simpler, which on a large scale could mean saving a lot of computing power and time.

That being said, it seems that all the objects you need to merge are always child of an empty with a specific name pattern BuildNodes_, so perhaps you could greatly simplify the algorithm:

1. create list of all objects matching this name pattern and with >1 children
2. for each:
   1. deselect all
   2. select children
   3. set first child as active
   4. join.

import bpy

parents = [obj for obj in bpy.context.scene.objects if obj.name.startswith('BuildNodes_') and len(obj.children) > 1]

for parent in parents:
    bpy.ops.object.select_all(action='DESELECT')
    for child in parent.children:
        child.select_set(True)
    bpy.context.view_layer.objects.active = parent.children[0]
    bpy.ops.object.join()

To be tested on a large scale.