|
| 1 | +// Copyright (c) Microsoft. All rights reserved. |
| 2 | + |
| 3 | +using System; |
| 4 | +using Microsoft.AspNetCore.Builder; |
| 5 | +using Microsoft.Extensions.Configuration; |
| 6 | +using Microsoft.Extensions.DependencyInjection; |
| 7 | +using Microsoft.SemanticKernel.AI.Embeddings; |
| 8 | +using Microsoft.SemanticMemory.Core.AI.AzureOpenAI; |
| 9 | +using Microsoft.SemanticMemory.Core.AI.OpenAI; |
| 10 | +using Microsoft.SemanticMemory.Core.AppBuilders; |
| 11 | +using Microsoft.SemanticMemory.Core.Configuration; |
| 12 | +using Microsoft.SemanticMemory.Core.ContentStorage.AzureBlobs; |
| 13 | +using Microsoft.SemanticMemory.Core.ContentStorage.FileSystemStorage; |
| 14 | +using Microsoft.SemanticMemory.Core.Handlers; |
| 15 | +using Microsoft.SemanticMemory.Core.MemoryStorage.AzureCognitiveSearch; |
| 16 | +using Microsoft.SemanticMemory.Core.MemoryStorage; |
| 17 | +using Microsoft.SemanticMemory.Core.Pipeline.Queue; |
| 18 | +using Microsoft.SemanticMemory.Core.Pipeline.Queue.AzureQueues; |
| 19 | +using Microsoft.SemanticMemory.Core.Pipeline.Queue.FileBasedQueues; |
| 20 | +using Microsoft.SemanticMemory.Core.Pipeline.Queue.RabbitMq; |
| 21 | +using Microsoft.SemanticMemory.Core.Pipeline; |
| 22 | +using Microsoft.SemanticKernel.Connectors.AI.OpenAI.TextEmbedding; |
| 23 | + |
| 24 | +namespace SemanticMemory.Service; |
| 25 | + |
| 26 | +/// <summary> |
| 27 | +/// Flexible dependency injection using dependencies defined in appsettings.json |
| 28 | +/// </summary> |
| 29 | +public static class Builder |
| 30 | +{ |
| 31 | + private const string ConfigRoot = "SemanticMemory"; |
| 32 | + |
| 33 | + public static WebApplicationBuilder CreateBuilder(out SemanticMemoryConfig config) |
| 34 | + { |
| 35 | + WebApplicationBuilder builder = WebApplication.CreateBuilder(); |
| 36 | + config = builder.Configuration.GetSection(ConfigRoot).Get<SemanticMemoryConfig>() |
| 37 | + ?? throw new ConfigurationException("Configuration is null"); |
| 38 | + |
| 39 | + builder.Services.AddSingleton<SemanticMemoryConfig>(config); |
| 40 | + builder.Services.AddSingleton<IMimeTypeDetection, MimeTypesDetection>(); |
| 41 | + builder.Services.AddSingleton<IPipelineOrchestrator, DistributedPipelineOrchestrator>(); |
| 42 | + builder.Services.AddSingleton<DistributedPipelineOrchestrator, DistributedPipelineOrchestrator>(); |
| 43 | + |
| 44 | + ConfigureContentStorage(builder, config); |
| 45 | + ConfigurePipelineHandlers(builder, config); |
| 46 | + ConfigureQueueSystem(builder, config); |
| 47 | + ConfigureEmbeddingGenerator(builder, config); |
| 48 | + ConfigureEmbeddingStorage(builder, config); |
| 49 | + |
| 50 | + return builder; |
| 51 | + } |
| 52 | + |
| 53 | + // Service where documents and temporary files are stored |
| 54 | + private static void ConfigureContentStorage(WebApplicationBuilder builder, SemanticMemoryConfig config) |
| 55 | + { |
| 56 | + switch (config.ContentStorageType) |
| 57 | + { |
| 58 | + case string x when x.Equals("AzureBlobs", StringComparison.OrdinalIgnoreCase): |
| 59 | + builder.Services.AddAzureBlobAsContentStorage(builder.Configuration |
| 60 | + .GetSection(ConfigRoot).GetSection("Services").GetSection("AzureBlobs") |
| 61 | + .Get<AzureBlobConfig>()!); |
| 62 | + break; |
| 63 | + |
| 64 | + case string x when x.Equals("FileSystemContentStorage", StringComparison.OrdinalIgnoreCase): |
| 65 | + builder.Services.AddFileSystemAsContentStorage(builder.Configuration |
| 66 | + .GetSection(ConfigRoot).GetSection("Services").GetSection("FileSystemContentStorage") |
| 67 | + .Get<FileSystemConfig>()!); |
| 68 | + break; |
| 69 | + |
| 70 | + default: |
| 71 | + throw new NotSupportedException($"Unknown/unsupported {config.ContentStorageType} content storage"); |
| 72 | + } |
| 73 | + } |
| 74 | + |
| 75 | + // Register pipeline handlers as hosted services |
| 76 | + private static void ConfigurePipelineHandlers(WebApplicationBuilder builder, SemanticMemoryConfig config) |
| 77 | + { |
| 78 | + builder.Services.AddHandlerAsHostedService<TextExtractionHandler>("extract"); |
| 79 | + builder.Services.AddHandlerAsHostedService<TextPartitioningHandler>("partition"); |
| 80 | + builder.Services.AddHandlerAsHostedService<GenerateEmbeddingsHandler>("gen_embeddings"); |
| 81 | + builder.Services.AddHandlerAsHostedService<SaveEmbeddingsHandler>("save_embeddings"); |
| 82 | + } |
| 83 | + |
| 84 | + // Orchestration dependencies, ie. which queueing system to use |
| 85 | + private static void ConfigureQueueSystem(WebApplicationBuilder builder, SemanticMemoryConfig config) |
| 86 | + { |
| 87 | + switch (config.DataIngestion.DistributedOrchestration.QueueType) |
| 88 | + { |
| 89 | + case string y when y.Equals("AzureQueue", StringComparison.OrdinalIgnoreCase): |
| 90 | + builder.Services.AddAzureQueue(builder.Configuration |
| 91 | + .GetSection(ConfigRoot).GetSection("Services").GetSection("AzureQueue") |
| 92 | + .Get<AzureQueueConfig>()!); |
| 93 | + break; |
| 94 | + |
| 95 | + case string y when y.Equals("RabbitMQ", StringComparison.OrdinalIgnoreCase): |
| 96 | + builder.Services.AddRabbitMq(builder.Configuration |
| 97 | + .GetSection(ConfigRoot).GetSection("Services").GetSection("RabbitMq") |
| 98 | + .Get<RabbitMqConfig>()!); |
| 99 | + break; |
| 100 | + |
| 101 | + case string y when y.Equals("FileBasedQueue", StringComparison.OrdinalIgnoreCase): |
| 102 | + builder.Services.AddFileBasedQueue(builder.Configuration |
| 103 | + .GetSection(ConfigRoot).GetSection("Services").GetSection("FileBasedQueue") |
| 104 | + .Get<FileBasedQueueConfig>()!); |
| 105 | + break; |
| 106 | + |
| 107 | + default: |
| 108 | + throw new NotSupportedException($"Unknown/unsupported {config.DataIngestion.DistributedOrchestration.QueueType} queue type"); |
| 109 | + } |
| 110 | + } |
| 111 | + |
| 112 | + // List of embedding generators to use (multiple generators allowed during ingestion) |
| 113 | + private static void ConfigureEmbeddingGenerator(WebApplicationBuilder builder, SemanticMemoryConfig config) |
| 114 | + { |
| 115 | + var embeddingGenerationServices = new TypeCollection<ITextEmbeddingGeneration>(); |
| 116 | + builder.Services.AddSingleton(embeddingGenerationServices); |
| 117 | + foreach (var type in config.DataIngestion.EmbeddingGeneratorTypes) |
| 118 | + { |
| 119 | + switch (type) |
| 120 | + { |
| 121 | + case string x when x.Equals("AzureOpenAI", StringComparison.OrdinalIgnoreCase): |
| 122 | + case string y when y.Equals("AzureOpenAIEmbedding", StringComparison.OrdinalIgnoreCase): |
| 123 | + embeddingGenerationServices.Add<AzureTextEmbeddingGeneration>(); |
| 124 | + builder.Services.AddAzureOpenAIEmbeddingGeneration(builder.Configuration |
| 125 | + .GetSection(ConfigRoot).GetSection("Services").GetSection("AzureOpenAIEmbedding") |
| 126 | + .Get<AzureOpenAIConfig>()!); |
| 127 | + break; |
| 128 | + |
| 129 | + case string x when x.Equals("OpenAI", StringComparison.OrdinalIgnoreCase): |
| 130 | + embeddingGenerationServices.Add<OpenAITextEmbeddingGeneration>(); |
| 131 | + builder.Services.AddOpenAITextEmbeddingGeneration(builder.Configuration |
| 132 | + .GetSection(ConfigRoot).GetSection("Services").GetSection("OpenAI") |
| 133 | + .Get<OpenAIConfig>()!); |
| 134 | + break; |
| 135 | + |
| 136 | + default: |
| 137 | + throw new NotSupportedException($"Unknown/unsupported {type} text generator"); |
| 138 | + } |
| 139 | + } |
| 140 | + } |
| 141 | + |
| 142 | + // List of Vector DB list where to store embeddings (multiple DBs allowed during ingestion) |
| 143 | + private static void ConfigureEmbeddingStorage(WebApplicationBuilder builder, SemanticMemoryConfig config) |
| 144 | + { |
| 145 | + var vectorDbServices = new TypeCollection<ISemanticMemoryVectorDb>(); |
| 146 | + builder.Services.AddSingleton(vectorDbServices); |
| 147 | + foreach (var type in config.DataIngestion.VectorDbTypes) |
| 148 | + { |
| 149 | + switch (type) |
| 150 | + { |
| 151 | + case string x when x.Equals("AzureCognitiveSearch", StringComparison.OrdinalIgnoreCase): |
| 152 | + vectorDbServices.Add<AzureCognitiveSearchMemory>(); |
| 153 | + builder.Services.AddAzureCognitiveSearchAsVectorDb(builder.Configuration |
| 154 | + .GetSection(ConfigRoot).GetSection("Services").GetSection("AzureCognitiveSearch") |
| 155 | + .Get<AzureCognitiveSearchConfig>()!); |
| 156 | + break; |
| 157 | + |
| 158 | + default: |
| 159 | + throw new NotSupportedException($"Unknown/unsupported {type} vector DB"); |
| 160 | + } |
| 161 | + } |
| 162 | + } |
| 163 | +} |
0 commit comments