LMEMappeCatastaliProApp/Services/FiducialPointService.cs

using System.Net;
using System.Net.Http.Headers;
using System.Text.RegularExpressions;

namespace LMEMappeCatastaliApp.Services;

public sealed partial class FiducialPointService
{
    private const string ServiceBaseUrl = "https://www1.agenziaentrate.gov.it/servizi/Monografie/";
    private readonly HttpClient _httpClient = new() { Timeout = TimeSpan.FromSeconds(30) };

    public FiducialPointService()
    {
        _httpClient.DefaultRequestHeaders.UserAgent.ParseAdd("LMEMappeCatastali/1.0 (Android)");
        _httpClient.DefaultRequestHeaders.Accept.Add(new MediaTypeWithQualityHeaderValue("text/html"));
    }

    public async Task<IReadOnlyList<FiducialAreaOption>> GetProvincesAsync(
        CancellationToken cancellationToken = default)
    {
        var html = await GetHtmlAsync("ricerca.php", cancellationToken);
        return ParseOptions(html, "provincia");
    }

    public async Task<IReadOnlyList<FiducialAreaOption>> GetMunicipalitiesAsync(
        string provinceCode,
        CancellationToken cancellationToken = default)
    {
        var code = NormalizeCode(provinceCode);
        var html = await GetHtmlAsync($"ricerca.php?level=1&prov={Uri.EscapeDataString(code)}", cancellationToken);
        return ParseOptions(html, "co");
    }

    public async Task<IReadOnlyList<FiducialAreaOption>> GetSheetsAsync(
        string provinceCode,
        string municipalityCode,
        CancellationToken cancellationToken = default)
    {
        var province = NormalizeCode(provinceCode);
        var municipality = NormalizeCode(municipalityCode);
        var path = $"ricerca.php?level=2&prov={Uri.EscapeDataString(province)}";
        var html = await PostFormAsync(path, new Dictionary<string, string> { ["co"] = municipality }, cancellationToken);
        return ParseOptions(html, "foglio");
    }

    public async Task<IReadOnlyList<FiducialPointEntry>> GetFiducialsAsync(
        string provinceName,
        string municipalityName,
        string municipalityCode,
        string sheet,
        CancellationToken cancellationToken = default)
    {
        var code = NormalizeCode(municipalityCode);
        var normalizedSheet = sheet.Trim();
        var query = string.Join("&",
        [
            $"provincia={Uri.EscapeDataString(provinceName.Trim())}",
            $"comune={Uri.EscapeDataString(municipalityName.Trim())}",
            $"co={Uri.EscapeDataString(code)}"
        ]);
        var html = await PostFormAsync(
            $"risultato.php?{query}",
            new Dictionary<string, string> { ["foglio"] = normalizedSheet },
            cancellationToken);

        return DownloadLinkRegex().Matches(html)
            .Select(match =>
            {
                var relativeUrl = WebUtility.HtmlDecode(match.Groups["url"].Value);
                var fileName = WebUtility.HtmlDecode(match.Groups["name"].Value);
                return new FiducialPointEntry(
                    FormatIdentifier(fileName),
                    new Uri(new Uri(ServiceBaseUrl), relativeUrl).AbsoluteUri);
            })
            .DistinctBy(item => item.Identifier, StringComparer.OrdinalIgnoreCase)
            .OrderBy(item => item.Identifier, StringComparer.OrdinalIgnoreCase)
            .ToArray();
    }

    private async Task<string> GetHtmlAsync(string relativeUrl, CancellationToken cancellationToken)
    {
        using var response = await _httpClient.GetAsync(ServiceBaseUrl + relativeUrl, cancellationToken);
        response.EnsureSuccessStatusCode();
        return await response.Content.ReadAsStringAsync(cancellationToken);
    }

    private async Task<string> PostFormAsync(
        string relativeUrl,
        IReadOnlyDictionary<string, string> values,
        CancellationToken cancellationToken)
    {
        using var content = new FormUrlEncodedContent(values);
        using var response = await _httpClient.PostAsync(ServiceBaseUrl + relativeUrl, content, cancellationToken);
        response.EnsureSuccessStatusCode();
        return await response.Content.ReadAsStringAsync(cancellationToken);
    }

    private static IReadOnlyList<FiducialAreaOption> ParseOptions(string html, string selectId)
    {
        var select = SelectRegex().Matches(html)
            .LastOrDefault(match => string.Equals(match.Groups["id"].Value, selectId, StringComparison.OrdinalIgnoreCase));
        if (select is null)
        {
            return [];
        }

        return OptionRegex().Matches(select.Groups["body"].Value)
            .Select(match => new FiducialAreaOption(
                WebUtility.HtmlDecode(match.Groups["value"].Value).Trim(),
                WebUtility.HtmlDecode(StripTagsRegex().Replace(match.Groups["label"].Value, string.Empty)).Trim()))
            .Where(item => !string.IsNullOrWhiteSpace(item.Code) && !string.IsNullOrWhiteSpace(item.Name))
            .DistinctBy(item => item.Code, StringComparer.OrdinalIgnoreCase)
            .ToArray();
    }

    private static string NormalizeCode(string value)
    {
        var normalized = value.Trim().ToUpperInvariant();
        if (normalized.Length is < 2 or > 4 || normalized.Any(character => !char.IsLetterOrDigit(character)))
        {
            throw new ArgumentException("Codice territoriale non valido.", nameof(value));
        }

        return normalized;
    }

    private static string FormatIdentifier(string fileName)
    {
        var parts = fileName.Split('-', StringSplitOptions.RemoveEmptyEntries | StringSplitOptions.TrimEntries);
        return parts.Length == 3 ? $"PF{parts[2]}/{parts[1]}/{parts[0]}" : fileName;
    }

    [GeneratedRegex("<select\\b[^>]*(?:id|name)=[\\\"'](?<id>[^\\\"']+)[\\\"'][^>]*>(?<body>[\\s\\S]*?)</select>", RegexOptions.IgnoreCase, matchTimeoutMilliseconds: 2000)]
    private static partial Regex SelectRegex();

    [GeneratedRegex("<option\\b[^>]*value=[\\\"'](?<value>[^\\\"']*)[\\\"'][^>]*>(?<label>[\\s\\S]*?)</option>", RegexOptions.IgnoreCase, matchTimeoutMilliseconds: 2000)]
    private static partial Regex OptionRegex();

    [GeneratedRegex("href=[\\\"'](?<url>download\\.php\\?[^\\\"']*namefile=(?<name>[A-Z0-9-]+))[^\\\"']*[\\\"']", RegexOptions.IgnoreCase, matchTimeoutMilliseconds: 2000)]
    private static partial Regex DownloadLinkRegex();

    [GeneratedRegex("<[^>]+>", RegexOptions.None, 2000)]
    private static partial Regex StripTagsRegex();
}

public sealed record FiducialAreaOption(string Code, string Name);

public sealed record FiducialPointEntry(string Identifier, string MonographUrl);