CROWLER/Services/PdfStorageService.cs

using System.IO;
using System.Linq;
using CRAWLER.Configuration;
using Microsoft.Extensions.Configuration;

namespace CRAWLER.Services;

internal sealed class PdfStorageService
{
    private readonly KtoPoveritClient _client;
    private readonly string _rootPath;

    public PdfStorageService(IConfiguration configuration, KtoPoveritClient client)
    {
        _client = client;
        var options = configuration.GetSection("Crawler").Get<CrawlerOptions>()
                      ?? throw new InvalidOperationException("Раздел Crawler не найден в appsettings.json.");
        _rootPath = Environment.ExpandEnvironmentVariables(options.PdfStoragePath);
        Directory.CreateDirectory(_rootPath);
    }

    public async Task<string> DownloadAsync(string sourceUrl, string registryNumber, string title, CancellationToken cancellationToken)
    {
        var bytes = await _client.GetBytesAsync(sourceUrl, cancellationToken);
        var fullPath = BuildTargetPath(registryNumber, title, sourceUrl);
        await File.WriteAllBytesAsync(fullPath, bytes, cancellationToken);
        return fullPath;
    }

    public async Task<string> CopyFromLocalAsync(string sourcePath, string registryNumber, string title, CancellationToken cancellationToken)
    {
        var fullPath = BuildTargetPath(registryNumber, title, sourcePath);

        await using var sourceStream = File.Open(sourcePath, FileMode.Open, FileAccess.Read, FileShare.Read);
        await using var targetStream = File.Create(fullPath);
        await sourceStream.CopyToAsync(targetStream, cancellationToken);
        return fullPath;
    }

    public void TryDelete(string path)
    {
        try
        {
            if (!string.IsNullOrWhiteSpace(path) && File.Exists(path))
            {
                File.Delete(path);
            }
        }
        catch
        {
        }
    }

    private string BuildTargetPath(string registryNumber, string title, string sourceIdentity)
    {
        var safeFolder = MakeSafePathSegment(string.IsNullOrWhiteSpace(registryNumber) ? "manual" : registryNumber);
        var folder = Path.Combine(_rootPath, safeFolder);
        Directory.CreateDirectory(folder);

        var baseName = MakeSafePathSegment(string.IsNullOrWhiteSpace(title) ? Path.GetFileNameWithoutExtension(sourceIdentity) : title);
        if (string.IsNullOrWhiteSpace(baseName))
        {
            baseName = "document";
        }

        var fullPath = Path.Combine(folder, baseName + ".pdf");
        if (!File.Exists(fullPath))
        {
            return fullPath;
        }

        var counter = 2;
        while (true)
        {
            var candidate = Path.Combine(folder, $"{baseName}-{counter}.pdf");
            if (!File.Exists(candidate))
            {
                return candidate;
            }

            counter++;
        }
    }

    private static string MakeSafePathSegment(string value)
    {
        var invalid = Path.GetInvalidFileNameChars();
        var cleaned = new string((value ?? string.Empty).Select(ch => invalid.Contains(ch) ? '_' : ch).ToArray()).Trim();
        return string.IsNullOrWhiteSpace(cleaned) ? "file" : cleaned;
    }
}