src/PdfEncoder.Infrastructure/Pdf/PdfEmbeddedImageCoverageCalculator.cs - master - afanasevn/PdfEncoder

src/PdfEncoder.Infrastructure/Pdf/PdfEmbeddedImageCoverageCalculator.cs

68 строк2 KB

IBS\NAfanasev

Parse results functional update

07 июл 2026, 00:06

07 июл 2026, 00:068e97443

using PdfEncoder.Application.Abstractions;
using PdfEncoder.Application.Options;
using UglyToad.PdfPig.Content;

namespace PdfEncoder.Infrastructure.Pdf;

/// <summary>
/// Расчёт доли площади страницы, занятой embedded images.
/// </summary>
public static class PdfEmbeddedImageCoverageCalculator
{
    /// <summary>
    /// Возвращает признак наличия images и coverage ratio для страницы.
    /// </summary>
    /// <param name="page">Страница PdfPig.</param>
    public static PageImageMetrics Analyze(Page page)
    {
        var images = page.GetImages().ToList();
        if (images.Count == 0)
        {
            return new PageImageMetrics(HasEmbeddedImages: false, CoverageRatio: 0);
        }

        var pageArea = page.Width * page.Height;
        if (pageArea <= 0)
        {
            return new PageImageMetrics(HasEmbeddedImages: true, CoverageRatio: 0);
        }

        var imageArea = images.Sum(CalculateBoundsArea);
        var ratio = Math.Clamp(imageArea / pageArea, 0, 1);
        return new PageImageMetrics(HasEmbeddedImages: true, CoverageRatio: ratio);
    }

    /// <summary>
    /// True, если страницу нужно обрабатывать full-page OCR по порогам coverage.
    /// </summary>
    /// <param name="page">Классификация страницы.</param>
    /// <param name="options">Пороги из <see cref="PipelineOptions"/>.</param>
    public static bool RequiresFullPageOcr(
        PageClassification page,
        PipelineOptions options)
    {
        if (page.ImageCoverageRatio >= options.FullPageOcrImageCoverageRatio)
        {
            return true;
        }

        return page.HasTextLayer
            && page.HasEmbeddedImages
            && page.ImageCoverageRatio >= options.HybridPageOcrImageCoverageRatio;
    }

    private static double CalculateBoundsArea(IPdfImage image)
    {
        var bounds = image.BoundingBox;
        return Math.Max(0, bounds.Width) * Math.Max(0, bounds.Height);
    }
}

/// <summary>
/// Метрики embedded images на странице.
/// </summary>
/// <param name="HasEmbeddedImages">Есть ли images на странице.</param>
/// <param name="CoverageRatio">Доля площади страницы (0..1).</param>
public readonly record struct PageImageMetrics(
    bool HasEmbeddedImages,
    double CoverageRatio);