| 1 | using System.Security.Cryptography; |
| 2 | using System.Text; |
| 3 | |
| 4 | namespace HybridCodebaseIndex.Core.Embeddings; |
| 5 | |
| 6 | /// <summary> |
| 7 | /// Deterministic, dependency-free embedding for pipeline wiring. |
| 8 | /// Not semantic; use a real provider for meaning. |
| 9 | /// </summary> |
| 10 | internal sealed class DummyEmbeddingProvider : IEmbeddingProvider |
| 11 | { |
| 12 | public int Dimension { get; } |
| 13 | |
| 14 | public DummyEmbeddingProvider(int dimension = 64) |
| 15 | { |
| 16 | Dimension = Math.Clamp(dimension, 16, 256); |
| 17 | } |
| 18 | |
| 19 | public ValueTask<float[]> EmbedAsync(string text, CancellationToken cancellationToken) |
| 20 | { |
| 21 | cancellationToken.ThrowIfCancellationRequested(); |
| 22 | |
| 23 | // Hash -> expand into floats in [-1..1], then L2-normalize. |
| 24 | var bytes = Encoding.UTF8.GetBytes(text ?? ""); |
| 25 | var hash = SHA256.HashData(bytes); |
| 26 | |
| 27 | var v = new float[Dimension]; |
| 28 | var j = 0; |
| 29 | while (j < Dimension) |
| 30 | { |
| 31 | // Mix hash with counter |
| 32 | var tmp = new byte[hash.Length + 4]; |
| 33 | hash.CopyTo(tmp.AsSpan(0, hash.Length)); |
| 34 | BitConverter.TryWriteBytes(tmp.AsSpan(hash.Length, 4), j); |
| 35 | var h2 = SHA256.HashData(tmp); |
| 36 | |
| 37 | for (var i = 0; i < h2.Length && j < Dimension; i += 2, j++) |
| 38 | { |
| 39 | var u16 = (ushort)(h2[i] | (h2[Math.Min(i + 1, h2.Length - 1)] << 8)); |
| 40 | v[j] = (u16 / 32767.5f) - 1f; |
| 41 | } |
| 42 | } |
| 43 | |
| 44 | NormalizeInPlace(v); |
| 45 | return ValueTask.FromResult(v); |
| 46 | } |
| 47 | |
| 48 | private static void NormalizeInPlace(float[] v) |
| 49 | { |
| 50 | double sum = 0; |
| 51 | foreach (var x in v) |
| 52 | sum += x * x; |
| 53 | var norm = Math.Sqrt(sum); |
| 54 | if (norm <= 1e-12) |
| 55 | return; |
| 56 | var inv = (float)(1.0 / norm); |
| 57 | for (var i = 0; i < v.Length; i++) |
| 58 | v[i] *= inv; |
| 59 | } |
| 60 | } |
| 61 | |
| 62 | |