src/PdfEncoder.Infrastructure/Llm/StructuredExtractor.cs - master - afanasevn/PdfEncoder

src/PdfEncoder.Infrastructure/Llm/StructuredExtractor.cs

59 строк2 KB

IBS\NAfanasev

Remote LLM base functional

01 июл 2026, 15:54

01 июл 2026, 15:5405aa80e

using System.Text.Json;
using Microsoft.Extensions.Options;
using PdfEncoder.Application.Abstractions;
using PdfEncoder.Application.Models;
using PdfEncoder.Application.Options;
using PdfEncoder.Domain.Models;

namespace PdfEncoder.Infrastructure.Llm;

/// <summary>
/// LLM structured extraction в generic schema.
/// </summary>
public sealed class StructuredExtractor(
    ILlmClient llmClient,
    PromptTemplateProvider promptProvider,
    IOptions<LlmModelOptions> options) : IStructuredExtractor
{
    private static readonly JsonSerializerOptions ContextJsonOptions = new()
    {
        PropertyNamingPolicy = JsonNamingPolicy.CamelCase,
        WriteIndented = false
    };

    /// <inheritdoc />
    public async Task<StructuredExtractionResult> ExtractAsync(
        DocumentContext context,
        string detectedDocumentType,
        Guid documentJobId,
        CancellationToken ct)
    {
        var contextJson = JsonSerializer.Serialize(context, ContextJsonOptions);
        var userPrompt = PromptTemplateProvider.Format(
            promptProvider.GetTemplate("extract-user.txt"),
            new Dictionary<string, string>
            {
                ["detectedType"] = detectedDocumentType,
                ["documentContextJson"] = contextJson
            });

        var completion = await llmClient.CompleteStructuredAsync<LlmExtractResponseDto>(
            new LlmStructuredRequest
            {
                Model = options.Value.ExtractionModel,
                SystemPrompt = promptProvider.GetTemplate("extract-system.txt"),
                UserPrompt = userPrompt,
                StepName = "llm-extract",
                DocumentJobId = documentJobId,
                PreferNativeJsonObject = options.Value.UseNativeJsonObject
            },
            ct);

        return new StructuredExtractionResult
        {
            Fields = completion.Data.Fields ?? [],
            Tables = completion.Data.Tables ?? [],
            Warnings = completion.Data.Warnings ?? []
        };
    }
}