For Developers

Read Inventor and STEP metadata from your own .NET code with the zero-dependency InventorMeta library.

The desktop app and the command line are both thin front-ends over a small core library, InventorMeta. It's a net10.0 library with zero external dependencies - drop it into your own app, service or Inventor add-in and read Inventor and STEP files directly.

The packaged viewer ships on the Microsoft Store; to build any of the three projects yourself, read on.

Build

# core library + CLI (cross-platform)
dotnet build src/InventorMeta.Cli -c Release

# the WinUI desktop viewer (Windows only)
dotnet build src/InventorMeta.App -c Release -r win-x64

ExtrabbitCode.Inventor.MetaReader.slnx
└─ src/
   ├─ InventorMeta/        core library (the parser)        - net10.0
   ├─ InventorMeta.Cli/    command-line tool ("invmeta")    - net10.0
   └─ InventorMeta.App/    WinUI 3 desktop viewer           - net10.0-windows

Open a document

Everything hangs off one class, InventorDocument. Parsing happens in the constructor; the result is a read-only snapshot.

using ExtrabbitCode.Inventor.MetaReader;

var doc = new InventorDocument(@"C:\path\SamplePart.ipt");

doc.FileName;       // "SamplePart.ipt"
doc.DocumentType;   // "Inventor Part (.ipt)"
doc.Kind;           // DocKind.Part | Assembly | Drawing | Presentation | Unknown

Kind comes from the file's root identifier, so it's correct even if the extension is wrong. To guard against non-Inventor files first:

if (!CompoundFile.LooksLikeCompoundFile(path)) return;

Read properties

// the common iProperties, resolved for you
foreach (var kv in doc.Summary)
    Console.WriteLine($"{kv.Key,-20} {kv.Value}");

string partNumber = doc.Summary["Part Number"];

// or every property in every set
foreach (var p in doc.Properties)
    Console.WriteLine($"[{p.Set}] {p.Name} = {p.Display}");

Mass / Volume / Surface Area / Density are cached values and can be stale until Inventor recomputes mass properties - check the Valid Mass Props flag before trusting them. See How it works.

References, model states, thumbnail, JSON

foreach (var r in doc.References) Console.WriteLine(r);

if (doc.HasModelStates)
    foreach (var s in doc.ModelStateDetails)
        Console.WriteLine($"{s.Name}{(s.IsActive ? " (active)" : "")}");

if (doc.Thumbnail != null)
    File.WriteAllBytes("preview." + doc.ThumbnailExt, doc.Thumbnail);

string json = doc.ToJson();

STEP files

STEP files (.stp / .step) go through the same class - the constructor sniffs the format, so you don't have to care which kind of file you were handed:

var doc = new InventorDocument(@"C:\path\bracket.step");

doc.IsStep;         // true
doc.DocumentType;   // "STEP model (.stp/.step)"

// the ISO-10303 header and a geometry summary arrive as regular property sets
foreach (var p in doc.Properties)          // [STEP Header] Author = ..., [STEP Geometry Summary] ...
    Console.WriteLine($"[{p.Set}] {p.Name} = {p.Display}");

For lower-level access, StepFile.Read returns the parsed header and per-entity counts directly, and StepFile.LooksLikeStepFile(path) is the cheap format check (also exposed as InventorDocument.IsStepFile).

The raw container

For lower-level access, CompoundFile parses the OLE Compound File container directly - enumerate the directory, read any stream:

using var cf = new CompoundFile(path);
Console.WriteLine($"CFB v{cf.MajorVersion}, {cf.SectorSize}-byte sectors");
byte[] ufrx = cf.ReadStream("/UFRxDoc");

To understand what these streams and property sets are, read How it works.