src/PdfEncoder.Infrastructure/Configuration/TessdataPathResolver.cs - master - afanasevn/PdfEncoder

src/PdfEncoder.Infrastructure/Configuration/TessdataPathResolver.cs
104 строки3 KB
IBS\NAfanasev
Parse results functional update
07 июл 2026, 00:06
07 июл 2026, 00:068e97443
namespace PdfEncoder.Infrastructure.Configuration;

/// <summary>
/// Разрешает путь к tessdata относительно корня репозитория (для отладки из bin/ и VS).
/// </summary>
public static class TessdataPathResolver
{
    private const string RusDataFile = "rus.traineddata";
    private const string EngDataFile = "eng.traineddata";
    private static string? _cachedDataPath;

    /// <summary>
    /// Возвращает абсолютный путь к tessdata с rus+eng traineddata.
    /// </summary>
    /// <param name="configuredPath">Значение из <c>Tesseract__DataPath</c>.</param>
    public static string Resolve(string configuredPath)
    {
        if (_cachedDataPath is not null)
        {
            return _cachedDataPath;
        }

        var resolved = TryResolveConfiguredPath(configuredPath)
            ?? FindRepoTessdata()
            ?? Path.GetFullPath(configuredPath);

        _cachedDataPath = resolved;
        return resolved;
    }

    /// <summary>
    /// True, если найден каталог tessdata с rus и eng traineddata.
    /// </summary>
    public static bool IsAvailable(string configuredPath = "./tessdata")
    {
        var path = TryResolveConfiguredPath(configuredPath) ?? FindRepoTessdata();
        return path is not null;
    }

    private static string? TryResolveConfiguredPath(string configuredPath)
    {
        if (string.IsNullOrWhiteSpace(configuredPath))
        {
            return null;
        }

        if (Path.IsPathRooted(configuredPath) && HasRequiredDataFiles(configuredPath))
        {
            return configuredPath;
        }

        var fromCurrentDirectory = Path.GetFullPath(Path.Combine(Directory.GetCurrentDirectory(), configuredPath));
        if (HasRequiredDataFiles(fromCurrentDirectory))
        {
            return fromCurrentDirectory;
        }

        var fromBaseDirectory = Path.GetFullPath(Path.Combine(AppContext.BaseDirectory, configuredPath));
        if (HasRequiredDataFiles(fromBaseDirectory))
        {
            return fromBaseDirectory;
        }

        return null;
    }

    private static string? FindRepoTessdata()
    {
        foreach (var startPath in new[] { Directory.GetCurrentDirectory(), AppContext.BaseDirectory })
        {
            var directory = new DirectoryInfo(startPath);
            while (directory is not null)
            {
                var tessdataPath = Path.Combine(directory.FullName, "tessdata");
                if (HasRequiredDataFiles(tessdataPath))
                {
                    return tessdataPath;
                }

                if (IsRepositoryRoot(directory))
                {
                    break;
                }

                directory = directory.Parent;
            }
        }

        return null;
    }

    private static bool IsRepositoryRoot(DirectoryInfo directory)
    {
        return File.Exists(Path.Combine(directory.FullName, "global.json"))
            || File.Exists(Path.Combine(directory.FullName, "PdfEncoder.slnx"));
    }

    private static bool HasRequiredDataFiles(string tessdataPath)
    {
        return Directory.Exists(tessdataPath)
            && File.Exists(Path.Combine(tessdataPath, RusDataFile))
            && File.Exists(Path.Combine(tessdataPath, EngDataFile));
    }
}