Find objects by property

Traverse, filter, and index Speckle model data — walls, doors, rooms, and QA checks in .NET

Goal

Find walls, doors, rooms, and other elements in a received model graph — and build indexes for QA checks, schedules, and quantity reports.

What you will build

Queries that count elements by category, check required properties, find duplicate applicationId values, and group quantities by level.

When to use this

After Receive2 on connector-published data — before CSV export, version comparison, or sending analysis results back. Read BIM data patterns first if property names are unfamiliar.

Recommended approach

The recommended approach for most filtering tasks is Flatten() on the root object, then filter on DataObject.properties. Use GraphTraversal with DefaultTraversal only when you need parent context or connector-style walk rules.

Prerequisites

• Work with model objects
• Build your first model analysis tool — load a model first

On DataObject instances, semantic fields live in properties: dataObject.properties.GetValueOrDefault("category"). Connector-produced BIM data often uses proxy collections — see Working with Proxies.

Complete example

Complete example

using Speckle.Objects.Data;
using Speckle.Sdk.Models;
using Speckle.Sdk.Models.Extensions;

static string? Category(Base obj) =>
    obj is DataObject d
        ? d.properties.GetValueOrDefault("category")?.ToString()
        : obj["category"] as string;

// Count walls, doors, rooms
var walls = root.Flatten().Where(o => Category(o) == "Walls").ToList();
var doors = root.Flatten().Where(o => Category(o)?.Contains("Door") == true).ToList();
var rooms = root.Flatten().Where(o => Category(o)?.Contains("Room") == true).ToList();

Console.WriteLine($"Walls: {walls.Count}, Doors: {doors.Count}, Rooms: {rooms.Count}");

// Quantities by category
var byCategory = root.Flatten()
    .GroupBy(o => Category(o) ?? "Unknown")
    .ToDictionary(g => g.Key, g => g.Count());

// Quantities by level
var byLevel = root.Flatten()
    .OfType<DataObject>()
    .GroupBy(d => d.properties.GetValueOrDefault("level")?.ToString() ?? "Unknown")
    .ToDictionary(g => g.Key, g => g.Count());

// Required property check
var required = new[] { "category", "level" };
var missing = root.Flatten()
    .OfType<DataObject>()
    .Where(d => required.Any(k => !d.properties.ContainsKey(k)))
    .ToList();

// Duplicate applicationId check
var duplicateIds = root.Flatten()
    .Where(o => o.applicationId is not null)
    .GroupBy(o => o.applicationId!)
    .Where(g => g.Count() > 1)
    .Select(g => g.Key)
    .ToList();

How it works

Flatten vs GraphTraversal

Flatten returns every Base in the graph — the right default for filtering and indexing:

Example

foreach (var obj in root.Flatten())
    Console.WriteLine(obj.TryGetName() ?? "unnamed");

GraphTraversal adds context (parent, property name) and custom walk rules:

Example

using Speckle.Sdk.Models.GraphTraversal;

var traversal = DefaultTraversal.CreateTraversalFunc();
foreach (TraversalContext ctx in traversal.Traverse(root))
{
    var obj = ctx.Current;
    var parentName = ctx.Parent?.Current.TryGetName() ?? "root";
    Console.WriteLine($"{obj.TryGetName()} (from {parentName}.{ctx.PropName})");
}

DefaultTraversal.CreateTraversalFunc() is what Rhino, Grasshopper, Revit, AutoCAD, Tekla, and Navisworks use for receive. Prefer it unless you need to walk different properties.

Building indexes for repeated lookups

Each Flatten() is O(n). Build a dictionary once for multiple lookups:

Example

var byCategory = root.Flatten()
    .GroupBy(Category)
    .ToDictionary(g => g.Key ?? "Unknown", g => g.ToList());

var walls = byCategory.GetValueOrDefault("Walls", new());

An applicationId index resolves proxies:

Example

var byAppId = root.Flatten()
    .Where(o => o.applicationId is not null)
    .ToDictionary(o => o.applicationId!);

Common mistakes

Mistake	Fix
Hand-written recursion on large graphs	Use Flatten or GraphTraversal — they guard against cycles
Rebuilding an index inside a loop	Build the dictionary once outside your query loop
Reading obj["category"] on all connector data	Use DataObject.properties
new GraphTraversal() with no rules	Always pass at least one rule, or use DefaultTraversal
Duplicate results from displayValue	Exclude displayValue in custom traversal rules when extracting BIM properties separately

Next steps

Export model data to CSV

Door and room schedules

Working with Proxies

Resolve levels and groups

Compare two model versions

Diff between publishes