natural sorting (object names)

[enzyme69]

Once again I encountered a situation where Object sorting become an issue because of no padding on the names. Usually after using Viewer BMesh node and then using Objects In node to get the objects back, the list sorting become a bit wrong.

So I created a bridge that sort the object by natural sorting.

SN LITE as below:

"""
in obnames s
out obsorted s
"""

# natural sorting for name without zero padding

import re
import bpy

# INPUT: Name of objects to be sorted

#mylist = ['apple_10', 'apple_1', 'apple_12', 'apple_123', 'apple_2']

def natural_sort(l): 
    convert = lambda text: int(text) if text.isdigit() else text.lower() 
    alphanum_key = lambda key: [ convert(c) for c in re.split('([0-9]+)', key) ] 
    return sorted(l, key = alphanum_key)

obnamesorted = natural_sort(obnames)

obsorted = []

for ob in obnamesorted:
    obsorted.append(bpy.data.objects[ob])

[enzyme69]

As a result the list is back to original sorting.

Object ID Sorting can probably do the job as well with custom index value, but that is for future user to figure out themselves.

Basically what I am doing:

• I have bunch of list, containing XYZ and Color. All in order
• XYZ become position of objects, generated object for each position
• the generated object name padding is sorted unexpectedly
• the SNL above should resort it
• then I can color the object based on the original list data

[enzyme69]

Oh yeah btw I tried using Exec Mod Node, but I cannot "import re" for that node. Not sure why. @Kosvor2

[zeffii]

but you already can sort by object's index (using object['idx'] as key ) --- If the objects are generated by a Sverchok Object node (like BMviewer etc...)

in ExecNodeMod. just this one line.

extend(list(sorted(V1, key=lambda obj: obj['idx'])))

[zeffii]

NameError: name 're' is not defined

we can/should fix that. It's a matter of adding locals() to the exec( ) context. Similar to this change

in ExecNodeMod, you can do this too:

def natural_sort(obj):
    import re
    return int(re.search('\d+', obj.name).group())

extend(sorted(V1, key=natural_sort))

[zeffii]

see

https://github.com/nortikin/sverchok/blob/69058d106528a671a74394751b12ed6911dabe1c/nodes/script/multi_exec.py#L192-L199

we can change it to

- exec('\n'.join([j.line for j in self.dynamic_strings])) 
+ exec('\n'.join([j.line for j in self.dynamic_strings]), locals()) 

and imports will work (in main body of the lines of text).

[zeffii]

i've just updated Sverchok. you can now write

from re import search

natural_sort = lambda obj: int(search('\d+', obj.name).group())
extend(sorted(V1, key=natural_sort))

in ExecNode Mod.

[nortikin]

lovely!