Tag based node filtering search (demo implementation)

[zeffii]

@Ribel78 mentions an alternative node finder approach, using tags to filter down from many to few

luckily we have much of the needed bookkeeping already implemented, the one thing left is to convert a tag cloud into a workable filter interface. Lately we added the node_classes lookup, which stores data like

meaning we can extract the menu /category and class name split by capitalized letter..

[zeffii]

import bpy
import sverchok
import re

def remove_if_present(set_name, item):
    if item in set_name:
        set_name.remove(item)

for node in sverchok.utils.node_classes.items():
    Node = lambda: None
    Node.ref = node[1]
    Node.name = node[0]
    tags = re.findall('[A-Z0-9]*[a-z0-9]*', Node.name)
    tags_set = set(list(tags))
    remove_if_present(tags_set, 'Sv')
    remove_if_present(tags_set, '')
    Node.tags = tags_set
    print(Node.name, Node.tags)

that based purely on class name, but there's also the category we can extract..

[zeffii]

import bpy
import sverchok
import re
from sverchok.utils.logging import info

def remove_if_present(set_name, item):
    if item in set_name:
        set_name.remove(item)

def get_tags_from_nodename(Node):
    tags = re.findall('[A-Z0-9]*[a-z0-9]*', Node.name)
    tags_set = set(list(tags))
    remove_if_present(tags_set, 'Sv')
    remove_if_present(tags_set, '')
    return tags_set

def get_tags_from_path(Node):
    path = Node.ref.__module__.split('.')[2].title()
    return set([path])

node_tags = {}
for node in sverchok.utils.node_classes.items():
    Node = lambda: None
    Node.ref = node[1]
    Node.name = node[0]

    tags_set = get_tags_from_nodename(Node)
    tags_set.update(get_tags_from_path(Node))

    Node.tags = tags_set
    # print(Node.tags)
    node_tags[Node.name] = Node.tags

print(node_tags)

[zeffii]

we can filter the word Node too..

[portnov]

MkN I think should be filtered out too.

[zeffii]

import bpy
import sverchok
import re
from sverchok.utils.logging import info

def remove_if_present(set_name, item=None, pattern=None):
    if not pattern:
        if item in set_name:
            set_name.remove(item)
    else:
        items_to_remove = set()

        for named_item in set_name:
            if named_item.lower().startswith(pattern):
                items_to_remove.add(named_item)

        for named_item in items_to_remove:
            set_name.remove(named_item)

def get_tags_from_nodename(Node):
    tags = re.findall('[A-Z0-9]*[a-z0-9]*', Node.name)
    tags_set = set(list(tags))
    remove_if_present(tags_set, 'Sv')
    remove_if_present(tags_set, '')
    remove_if_present(tags_set, 'Node')
    remove_if_present(tags_set, pattern='mk')

    return tags_set

def get_tags_from_path(Node):
    path = Node.ref.__module__.split('.')[2].title()
    return set([path])

node_tags = {}
for node in sverchok.utils.node_classes.items():
    Node = lambda: None
    Node.ref = node[1]
    Node.name = node[0]

    tags_set = get_tags_from_nodename(Node)
    tags_set.update(get_tags_from_path(Node))

    Node.tags = tags_set
    # print(Node.tags)
    node_tags[Node.name] = Node.tags

print(node_tags)

[zeffii]

so now all that's let to do is generate

• a massive set from all values of node_tags, and generate a CollectionProperty of booleans..
• populate a menu with those node_tags keys that contain all "toggled" boolean properties

[portnov]

Tag systems are usually used for searching. I.e. for situation when I want a box and I search for "box". This actually works now with usual blender's Space or our Ctrl-Space searchboxes. Where searchboxes are not good, is questions like "what do we have here?" ("explorability"). And I'm not sure if tagging system will solve this question better.

[zeffii]

i also don't think this is needed.. just want to see what it looks like.

[zeffii]

lol....

['Adaptive', 'Along', 'Alt', 'Analyzer', 'Angles', 'Apply', 'Arc', 'Area', 'Array', 'Art', 'Asset', 'Attractor', 'Axis', 'BBox', 'BMOps', 'BMVerts', 'BMinput', 'BMoutput', 'BMto', 'Basic', 'Bend', 'Bevel', 'Bisect', 'Bmesh', 'Bmesh_Nodes', 'Boom', 'Box', 'Bricks', 'Bvh', 'CSGBoolean', 'Cache', 'Cast', 'Centers', 'Circle', 'Client', 'Color', 'Colors', 'Components', 'Connect', 'Contour', 'Convert', 'Converter', 'Convex', 'Cross', 'Curve', 'Cylinder', 'DNode', 'Data', 'Debug', 'Decompose', 'Deform', 'Del', 'Delaunay', 'Delete', 'Dissolve', 'Distance', 'Doubles', 'Drop', 'Dupli', 'Duplicate', 'Dview', 'Easing', 'Edge', 'Edges', 'Edgs', 'Edit', 'Element', 'Elman1', 'Empty', 'Euler', 'Eval', 'Evaluate', 'Exec', 'Exp', 'Exponential', 'Extrude', 'FLNode', 'Fibonacci', 'Field', 'Fills', 'Filter', 'Flip', 'Float', 'Float2', 'Formula', 'Formula2', 'Fractal', 'Frame', 'Func', 'GText', 'Gen', 'Generative', 'Generator', 'Generators_Extended', 'Generic', 'Get', 'Grease', 'Grid', 'Group', 'Heavy', 'Hexa', 'Hilbert', 'Hilbert3d', 'Hole', 'Homogenous', 'Hull', 'Icosphere', 'Image', 'In', 'Index', 'Info', 'Input', 'Inputs', 'Inset', 'Inside', 'Instancer', 'Instances', 'Int', 'Integer', 'Interpolation', 'Intersect', 'Item2', 'Iterate', 'Join', 'KDTree', 'Knot', 'LNode', 'Lacunarity', 'Lathe', 'Layout', 'Length', 'Lerp', 'Levels', 'Limited', 'Line', 'List', 'List_Main', 'List_Masks', 'List_Mutators', 'List_Struct', 'Lite', 'Logic', 'Loose', 'Map', 'Mask', 'Match', 'Math', 'Matrix', 'Mesh', 'Metaball', 'Mirror', 'Mix', 'Mod', 'Modifier', 'Modifier_Change', 'Modifier_Make', 'Monad', 'Move', 'Multi', 'NGon', 'Network', 'Neuro', 'New', 'Node2', 'Noise', 'Normal', 'Normals', 'Note', 'Num', 'Number', 'Numbers', 'Numpy', 'OBJRay', 'Obj', 'Object', 'Object_Nodes', 'Objects', 'Objs', 'Offset', 'On', 'Out', 'Outputs', 'Overlap', 'PPNode', 'Particles', 'Path', 'Pencil', 'Pipe', 'Plane', 'Point', 'Polar', 'Pols', 'Pols2', 'Polygon', 'Polyline', 'Print', 'Profile', 'Prop', 'Properties', 'Proportional', 'Props', 'Random', 'Randomize', 'Range', 'Raycaster', 'Recalc', 'Region', 'Remote', 'Remove', 'Repeater', 'Reverse', 'Rewire', 'Ring', 'Rnd', 'Rotation', 'Rounded', 'SCNRay', 'Scalar', 'Scale', 'Scene', 'Script', 'Sculpt', 'Section', 'Select', 'Separate', 'Set', 'Shape', 'Shear', 'Shift', 'Shuffle', 'Similar', 'Simple', 'Skin', 'Slice', 'Smooth', 'Solidify', 'Sort', 'Special', 'Sphere', 'Spiral', 'Spline', 'Split', 'Stethoscope', 'Stripes', 'Strokes', 'Subdivide', 'Sum', 'Surface', 'Suzanne', 'Sv3', 'Switch', 'Text', 'Texture', 'To', 'Torus', 'Track', 'Transform', 'Transforms', 'Triangles', 'Triangulate', 'Triangulation2', 'Tube', 'Turbulence', 'Type', 'UVtexture', 'Udp', 'Value', 'Vector', 'Vectors', 'Vert', 'Vertex', 'Vertices', 'Verts', 'Viewer', 'Viz', 'Volume', 'Voronoi2', 'Wafel', 'Wifi', 'Wireframe', 'Zip', 'sv']

[zeffii]

yeah, now i definitely don't see a practical use for this..

[zeffii]

further processing,

• if a key name exists, and a key names, then the pural could be dropped/renamed
• if a key ends in a number, and the key without the number or with a different number appears, then rename all to the sans-number version.

print(node_tags)
all_tags = set()
for tags in node_tags.values():
    all_tags.update(tags)

print(sorted(all_tags))

[zeffii]

initial list could be presented in order of Frequency, or Alphabetical

[zeffii]

This is still waaay to many to start an initial elastic search... by clicking on a word. https://gist.github.com/zeffii/d1db56a487aef0a12b0e61e1ddb95919

There is no UI box that can send a callback to the layout system while you're typing.. meh.. too much work man.