LehrerApp/LehrerApp.Core/Importing/StudentMasterDataCsvImportHandler.cs

using System.Globalization;
using System.Text;
using LehrerApp.Core.Models;

namespace LehrerApp.Core.Importing;

/// <summary>
/// Liest den tabulatorgetrennten Stammdatenexport mit Spalten wie „longName“,
/// „foreName“, „birthDate“ und „klasse.name“. Die Datei trägt historisch die
/// Endung .csv, obwohl als Trennzeichen ein Tabulator verwendet wird.
/// </summary>
public sealed class StudentMasterDataCsvImportHandler : IImportHandler<ImportedStudent>
{
    private static readonly string[] SignatureHeaders =
        ["longName", "foreName", "gender", "birthDate", "klasse.name", "externKey"];

    private static readonly UTF8Encoding StrictUtf8 = new(
        encoderShouldEmitUTF8Identifier: false,
        throwOnInvalidBytes: true);

    public ImportFormatId FormatId => StudentImportFormats.MasterDataCsv;
    public string DisplayName => "Schüler-Stammdatenliste";
    public IReadOnlyCollection<string> SupportedExtensions { get; } = [".csv"];

    public ValueTask<ImportDetection> DetectAsync(
        ImportFile file,
        CancellationToken cancellationToken = default)
    {
        cancellationToken.ThrowIfCancellationRequested();
        string text;
        try
        {
            text = StrictUtf8.GetString(file.Content.Span);
        }
        catch (DecoderFallbackException)
        {
            return ValueTask.FromResult(ImportDetection.NoMatch("Die Datei ist kein gültiges UTF-8."));
        }

        var firstLineEnd = text.IndexOfAny(['\r', '\n']);
        var firstLine = (firstLineEnd >= 0 ? text[..firstLineEnd] : text).TrimStart('\uFEFF');
        var headers = firstLine.Split('\t')
            .Select(header => header.Trim())
            .ToHashSet(StringComparer.OrdinalIgnoreCase);
        if (!SignatureHeaders.All(headers.Contains))
            return ValueTask.FromResult(ImportDetection.NoMatch("Die erwarteten Stammdatenspalten fehlen."));

        return ValueTask.FromResult(ImportDetection.Match(
            file.Extension.Equals(".csv", StringComparison.OrdinalIgnoreCase) ? 100 : 95,
            "Die charakteristischen Stammdatenspalten wurden gefunden."));
    }

    public ValueTask<ImportParseResult<ImportedStudent>> ParseAsync(
        ImportFile file,
        CancellationToken cancellationToken = default)
    {
        cancellationToken.ThrowIfCancellationRequested();
        var messages = new List<ImportMessage>();
        string text;
        try
        {
            text = StrictUtf8.GetString(file.Content.Span);
        }
        catch (DecoderFallbackException)
        {
            messages.Add(new ImportMessage(
                ImportMessageSeverity.Error,
                "master-data.encoding",
                "Die Stammdatenliste ist nicht als gültiges UTF-8 gespeichert."));
            return ValueTask.FromResult(new ImportParseResult<ImportedStudent>([], messages));
        }

        if (!DelimitedTextParser.TryParse(text, '\t', out var records, out var parseError))
        {
            messages.Add(new ImportMessage(
                ImportMessageSeverity.Error,
                "master-data.structure",
                parseError ?? "Die Stammdatenliste ist syntaktisch ungültig."));
            return ValueTask.FromResult(new ImportParseResult<ImportedStudent>([], messages));
        }

        if (records.Count == 0)
        {
            messages.Add(new ImportMessage(
                ImportMessageSeverity.Error,
                "master-data.empty",
                "Die Stammdatenliste ist leer."));
            return ValueTask.FromResult(new ImportParseResult<ImportedStudent>([], messages));
        }

        var headers = records[0].Fields
            .Select((header, index) => (Name: header.Trim().TrimStart('\uFEFF'), Index: index))
            .ToList();
        var duplicateHeader = headers
            .GroupBy(header => header.Name, StringComparer.OrdinalIgnoreCase)
            .FirstOrDefault(group => group.Count() > 1);
        if (duplicateHeader is not null)
            messages.Add(new ImportMessage(
                ImportMessageSeverity.Error,
                "master-data.header.duplicate",
                $"Die Spalte „{duplicateHeader.Key}“ kommt mehrfach vor.",
                "Kopfzeile"));

        var columns = headers
            .GroupBy(header => header.Name, StringComparer.OrdinalIgnoreCase)
            .ToDictionary(group => group.Key, group => group.First().Index, StringComparer.OrdinalIgnoreCase);
        foreach (var required in SignatureHeaders.Where(required => !columns.ContainsKey(required)))
            messages.Add(new ImportMessage(
                ImportMessageSeverity.Error,
                "master-data.header.missing",
                $"Die erforderliche Spalte „{required}“ fehlt.",
                "Kopfzeile"));
        if (messages.Any(message => message.Severity == ImportMessageSeverity.Error))
            return ValueTask.FromResult(new ImportParseResult<ImportedStudent>([], messages));

        var students = new List<ImportedStudent>();
        foreach (var record in records.Skip(1))
        {
            cancellationToken.ThrowIfCancellationRequested();
            if (record.Fields.All(string.IsNullOrWhiteSpace)) continue;
            var source = $"Zeile {record.LineNumber}";
            if (record.Fields.Count != headers.Count)
            {
                messages.Add(new ImportMessage(
                    ImportMessageSeverity.Error,
                    "master-data.columns",
                    $"Die Zeile enthält {record.Fields.Count} statt {headers.Count} Spalten.",
                    source));
                continue;
            }

            var gender = ParseGender(Value(record, columns, "gender"), source, messages);
            var dateOfBirth = ParseDate(Value(record, columns, "birthDate"), source, messages);
            var externalId = FirstNotEmpty(
                Value(record, columns, "externKey"),
                Value(record, columns, "id"));
            students.Add(new ImportedStudent
            {
                FirstName = Value(record, columns, "foreName"),
                LastName = Value(record, columns, "longName"),
                Gender = gender,
                DateOfBirth = dateOfBirth,
                ExternalId = externalId,
                GroupName = NullIfWhiteSpace(Value(record, columns, "klasse.name")),
                Contact = new ImportedStudentContact
                {
                    Email = NullIfWhiteSpace(Value(record, columns, "address.email")),
                    MobilePhone = NullIfWhiteSpace(Value(record, columns, "address.mobile")),
                    Phone = NullIfWhiteSpace(Value(record, columns, "address.phone")),
                    City = NullIfWhiteSpace(Value(record, columns, "address.city")),
                    PostalCode = NullIfWhiteSpace(Value(record, columns, "address.postCode")),
                    Street = NullIfWhiteSpace(Value(record, columns, "address.street")),
                },
                SourceReference = source,
            });
        }

        return ValueTask.FromResult(new ImportParseResult<ImportedStudent>(students, messages));
    }

    private static Gender? ParseGender(
        string value,
        string source,
        ICollection<ImportMessage> messages)
    {
        if (string.IsNullOrWhiteSpace(value)) return null;
        return value.Trim().ToLowerInvariant() switch
        {
            "m" or "männlich" or "maennlich" => Gender.M,
            "w" or "weiblich" => Gender.W,
            "d" or "divers" => Gender.D,
            _ => UnknownGender(value, source, messages),
        };
    }

    private static Gender? UnknownGender(
        string value,
        string source,
        ICollection<ImportMessage> messages)
    {
        messages.Add(new ImportMessage(
            ImportMessageSeverity.Warning,
            "master-data.gender.unknown",
            $"Das Geschlecht „{value.Trim()}“ ist unbekannt und wird nicht übernommen.",
            source));
        return null;
    }

    private static DateOnly? ParseDate(
        string value,
        string source,
        ICollection<ImportMessage> messages)
    {
        if (string.IsNullOrWhiteSpace(value)) return null;
        if (DateOnly.TryParseExact(
                value.Trim(),
                "dd.MM.yyyy",
                CultureInfo.InvariantCulture,
                DateTimeStyles.None,
                out var date))
            return date;

        messages.Add(new ImportMessage(
            ImportMessageSeverity.Error,
            "master-data.birth-date.invalid",
            $"Das Geburtsdatum „{value.Trim()}“ besitzt nicht das erwartete Format TT.MM.JJJJ.",
            source));
        return null;
    }

    private static string Value(
        DelimitedRecord record,
        IReadOnlyDictionary<string, int> columns,
        string column) =>
        columns.TryGetValue(column, out var index) && index < record.Fields.Count
            ? record.Fields[index].Trim()
            : "";

    private static string? FirstNotEmpty(params string[] values) =>
        values.Select(NullIfWhiteSpace).FirstOrDefault(value => value is not null);

    private static string? NullIfWhiteSpace(string? value) =>
        string.IsNullOrWhiteSpace(value) ? null : value.Trim();

}