Data-Coupler/DataConnection/DB/FoxPro/FoxProReader.cs

using System.Text;
using System.Text.RegularExpressions;
using DbfDataReader;
using DataConnection.Interfaces;

namespace DataConnection.DB.FoxPro;

/// <summary>
/// Helper di basso livello per leggere database Visual FoxPro / dBase in modo
/// completamente managed (nessun provider OLE DB/ODBC richiesto, funziona a 64-bit).
///
/// Legge direttamente i file <c>.dbf</c> (tabelle), <c>.fpt</c> (campi memo) e, quando
/// la sorgente è un container <c>.dbc</c>, ne legge il catalogo per ricavare l'elenco
/// ufficiale delle tabelle. La libreria sottostante è <c>DbfDataReader</c> (MIT).
///
/// Sola lettura: il coupler usa FoxPro come sorgente dati.
/// </summary>
public static class FoxProReader
{
    static FoxProReader()
    {
        // Le code page non-Unicode (es. 1252, 1250, 850) non sono disponibili di default
        // su .NET Core/5+: serve registrare il provider.
        Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
    }

    /// <summary>Code page di default (Windows-1252, ANSI Europa occidentale).</summary>
    public const int DefaultCodePage = 1252;

    // Mappa byte "language driver" dell'header DBF -> code page più comuni.
    private static readonly Dictionary<byte, int> LanguageDriverCodePages = new()
    {
        [0x01] = 437,   // U.S. MS-DOS
        [0x02] = 850,   // International MS-DOS
        [0x03] = 1252,  // Windows ANSI (più comune nei dati VFP)
        [0x08] = 865,   // Danish OEM
        [0x09] = 437,   // Dutch OEM
        [0x0A] = 850,   // Dutch OEM*
        [0x64] = 852,   // Eastern European MS-DOS
        [0x65] = 866,   // Russian MS-DOS
        [0x66] = 865,   // Nordic MS-DOS
        [0x67] = 861,   // Icelandic MS-DOS
        [0x6A] = 737,   // Greek MS-DOS
        [0x6B] = 857,   // Turkish MS-DOS
        [0x57] = 1252,  // ANSI
        [0xC8] = 1250,  // Eastern European Windows
        [0xC9] = 1251,  // Russian Windows
        [0xCA] = 1254,  // Turkish Windows
        [0xCB] = 1253,  // Greek Windows
    };

    /// <summary>
    /// Interpreta la "connection string" FoxPro (un percorso, eventualmente con opzioni)
    /// e risolve cartella, eventuale <c>.dbc</c>, encoding e gestione record cancellati.
    ///
    /// Formati accettati:
    /// <list type="bullet">
    /// <item>percorso nudo a un file <c>.dbc</c>: <c>C:\dati\Data.dbc</c></item>
    /// <item>percorso nudo a una cartella di <c>.dbf</c>: <c>C:\dati</c></item>
    /// <item>stringa con chiavi: <c>Data Source=C:\dati\Data.dbc;CodePage=1252;IncludeDeleted=false</c></item>
    /// <item>stringa stile OLE DB VFP: <c>Provider=vfpoledb;Data Source=C:\dati\Data.dbc;...</c> (Data Source viene estratto)</item>
    /// </list>
    /// </summary>
    public static FoxProConnectionInfo Resolve(string connectionString)
    {
        if (string.IsNullOrWhiteSpace(connectionString))
            throw new ArgumentException("Percorso del database FoxPro non specificato.", nameof(connectionString));

        string? path = null;
        int? codePage = null;
        bool includeDeleted = false;

        // Se contiene almeno una coppia chiave=valore, interpretiamo come connection string;
        // altrimenti l'intera stringa è il percorso.
        if (connectionString.Contains('='))
        {
            foreach (var segment in connectionString.Split(';', StringSplitOptions.RemoveEmptyEntries | StringSplitOptions.TrimEntries))
            {
                var idx = segment.IndexOf('=');
                if (idx < 0)
                {
                    // Segmento senza '=': potrebbe essere un percorso nudo mescolato — usalo se non abbiamo già un path
                    path ??= segment.Trim();
                    continue;
                }

                var key = segment[..idx].Trim();
                var value = segment[(idx + 1)..].Trim().Trim('"', '\'');

                switch (key.Replace(" ", "").ToLowerInvariant())
                {
                    case "datasource":
                    case "source":
                    case "data":
                        path = value;
                        break;
                    case "codepage":
                    case "encoding":
                        if (int.TryParse(value, out var cp))
                            codePage = cp;
                        else
                            codePage = TryGetCodePageByName(value);
                        break;
                    case "includedeleted":
                    case "deleted":
                        includeDeleted = value.Equals("true", StringComparison.OrdinalIgnoreCase)
                                         || value.Equals("on", StringComparison.OrdinalIgnoreCase)
                                         || value == "1";
                        break;
                    // "Provider", "Collating Sequence", ecc. sono ignorati (compatibilità OLE DB)
                }
            }
        }
        else
        {
            path = connectionString.Trim();
        }

        path = path?.Trim().Trim('"', '\'');
        if (string.IsNullOrWhiteSpace(path))
            throw new ArgumentException("Impossibile determinare il percorso del database FoxPro dalla stringa fornita.", nameof(connectionString));

        string folder;
        string? dbcPath = null;

        if (path.EndsWith(".dbc", StringComparison.OrdinalIgnoreCase))
        {
            dbcPath = path;
            folder = Path.GetDirectoryName(path) ?? path;
        }
        else
        {
            folder = path.TrimEnd('\\', '/');
            // Se la cartella contiene un singolo .dbc e nessuna ambiguità, lo usiamo come container.
            if (Directory.Exists(folder))
            {
                var dbcs = Directory.GetFiles(folder, "*.dbc");
                if (dbcs.Length == 1)
                    dbcPath = dbcs[0];
            }
        }

        var encoding = ResolveEncoding(codePage, dbcPath, folder);

        return new FoxProConnectionInfo
        {
            Folder = folder,
            DbcPath = dbcPath,
            Encoding = encoding,
            IncludeDeleted = includeDeleted
        };
    }

    /// <summary>Verifica che la sorgente sia raggiungibile (cartella esistente, eventuale .dbc presente).</summary>
    public static void EnsureReachable(FoxProConnectionInfo info)
    {
        if (info.IsContainer && !File.Exists(info.DbcPath!))
            throw new FileNotFoundException($"File database container (.dbc) non trovato: {info.DbcPath}");

        if (!Directory.Exists(info.Folder))
            throw new DirectoryNotFoundException($"Cartella del database FoxPro non trovata: {info.Folder}");
    }

    /// <summary>
    /// Elenco delle tabelle: dal catalogo del <c>.dbc</c> (OBJECTTYPE='Table') se container,
    /// altrimenti enumerando i file <c>.dbf</c> nella cartella. Vengono restituite solo le
    /// tabelle effettivamente apribili (file <c>.dbf</c> presente).
    /// </summary>
    public static List<string> GetTableNames(FoxProConnectionInfo info)
    {
        EnsureReachable(info);

        var names = new List<string>();
        var seen = new HashSet<string>(StringComparer.OrdinalIgnoreCase);

        if (info.IsContainer)
        {
            foreach (var name in ReadTableNamesFromContainer(info))
            {
                if (seen.Add(name) && File.Exists(ResolveTableFilePath(info.Folder, name)))
                    names.Add(name);
            }
        }
        else
        {
            foreach (var file in Directory.EnumerateFiles(info.Folder, "*.dbf", SearchOption.TopDirectoryOnly))
            {
                var name = Path.GetFileNameWithoutExtension(file);
                if (seen.Add(name))
                    names.Add(name);
            }
        }

        names.Sort(StringComparer.OrdinalIgnoreCase);
        return names;
    }

    /// <summary>Schema (colonne) di una singola tabella.</summary>
    public static List<DbColumnInfo> GetTableColumns(FoxProConnectionInfo info, string tableName)
    {
        var filePath = ResolveTableFilePath(info.Folder, tableName);
        if (!File.Exists(filePath))
            throw new FileNotFoundException($"Tabella FoxPro non trovata: {filePath}");

        using var table = new DbfTable(filePath, info.Encoding);
        var columns = new List<DbColumnInfo>();
        foreach (var col in table.Columns)
        {
            columns.Add(new DbColumnInfo
            {
                Name = col.ColumnName.Trim(),
                DataType = MapColumnType(col),
                IsNullable = true,        // VFP: nullability non esposta in modo affidabile dall'header
                IsPrimaryKey = false      // le tabelle .dbf non hanno una PK affidabile; la chiave si sceglie nel coupler
            });
        }
        return columns;
    }

    /// <summary>
    /// Legge i record di una tabella come dizionari nome-colonna -> valore.
    /// I valori null sono rappresentati con <see cref="DBNull.Value"/> (coerente con gli altri manager).
    /// Streaming: non carica l'intera tabella in memoria.
    /// </summary>
    public static IEnumerable<Dictionary<string, object>> ReadRecords(
        FoxProConnectionInfo info, string tableName, int? maxRows = null)
    {
        var filePath = ResolveTableFilePath(info.Folder, tableName);
        if (!File.Exists(filePath))
            throw new FileNotFoundException($"Tabella FoxPro non trovata: {filePath}");

        var options = new DbfDataReaderOptions
        {
            Encoding = info.Encoding,
            SkipDeletedRecords = !info.IncludeDeleted
        };

        using var reader = new DbfDataReader.DbfDataReader(filePath, options);

        // Nomi colonna (trimmati) per ordinale.
        var names = reader.DbfTable.Columns.Select(c => c.ColumnName.Trim()).ToArray();

        long emitted = 0;
        while (reader.Read())
        {
            var row = new Dictionary<string, object>(names.Length, StringComparer.OrdinalIgnoreCase);
            for (int i = 0; i < names.Length; i++)
            {
                row[names[i]] = reader.IsDBNull(i) ? DBNull.Value : reader.GetValue(i);
            }
            yield return row;

            emitted++;
            if (maxRows.HasValue && emitted >= maxRows.Value)
                yield break;
        }
    }

    /// <summary>Numero di record (header) di una tabella, senza scorrere i dati.</summary>
    public static long GetRecordCount(FoxProConnectionInfo info, string tableName)
    {
        var filePath = ResolveTableFilePath(info.Folder, tableName);
        if (!File.Exists(filePath))
            return 0;
        using var table = new DbfTable(filePath, info.Encoding);
        return table.Header.RecordCount;
    }

    /// <summary>
    /// Estrae il nome tabella (e un eventuale <c>TOP n</c>) da una semplice query SELECT.
    /// Gestisce anche le query "wrappate" generate da CreateLimitedQuery
    /// (<c>SELECT TOP n * FROM (SELECT ... FROM tabella) AS subquery</c>): viene preso
    /// l'ultimo identificatore dopo FROM, cioè la tabella reale più interna.
    /// </summary>
    public static (string Table, int? Top)? ParseSimpleSelect(string sql)
    {
        if (string.IsNullOrWhiteSpace(sql))
            return null;

        var fromMatches = Regex.Matches(sql, @"\bFROM\s+\[?([A-Za-z_][\w]*)\]?", RegexOptions.IgnoreCase);
        if (fromMatches.Count == 0)
            return null;

        var table = fromMatches[^1].Groups[1].Value;

        int? top = null;
        var topMatch = Regex.Match(sql, @"\bTOP\s+(\d+)", RegexOptions.IgnoreCase);
        if (topMatch.Success && int.TryParse(topMatch.Groups[1].Value, out var t))
            top = t;

        return (table, top);
    }

    /// <summary>Risolve il path del file .dbf per una tabella, gestendo il case-insensitive di Windows.</summary>
    public static string ResolveTableFilePath(string folder, string tableName)
    {
        var name = tableName.Trim().Trim('[', ']', '"', '\'');
        if (name.EndsWith(".dbf", StringComparison.OrdinalIgnoreCase))
            name = name[..^4];

        var direct = Path.Combine(folder, name + ".dbf");
        if (File.Exists(direct))
            return direct;

        // Ricerca case-insensitive (i cataloghi .dbc spesso usano un case diverso dai file).
        if (Directory.Exists(folder))
        {
            foreach (var file in Directory.EnumerateFiles(folder, "*.dbf", SearchOption.TopDirectoryOnly))
            {
                if (string.Equals(Path.GetFileNameWithoutExtension(file), name, StringComparison.OrdinalIgnoreCase))
                    return file;
            }
        }

        return direct; // non trovato: il chiamante gestirà il FileNotFound
    }

    private static IEnumerable<string> ReadTableNamesFromContainer(FoxProConnectionInfo info)
    {
        // Il file .dbc è esso stesso una tabella DBF: il suo catalogo contiene righe con
        // OBJECTTYPE in {Database, Table, Index, Field, Relation, View}. Ci interessano i 'Table'.
        var options = new DbfDataReaderOptions
        {
            Encoding = info.Encoding,
            SkipDeletedRecords = true
        };

        using var reader = new DbfDataReader.DbfDataReader(info.DbcPath!, options);

        int typeIdx = -1, nameIdx = -1;
        for (int i = 0; i < reader.DbfTable.Columns.Count; i++)
        {
            var col = reader.DbfTable.Columns[i].ColumnName.Trim();
            if (col.Equals("OBJECTTYPE", StringComparison.OrdinalIgnoreCase)) typeIdx = i;
            else if (col.Equals("OBJECTNAME", StringComparison.OrdinalIgnoreCase)) nameIdx = i;
        }

        if (typeIdx < 0 || nameIdx < 0)
            yield break; // .dbc non nel formato atteso

        while (reader.Read())
        {
            var type = reader.IsDBNull(typeIdx) ? null : reader.GetValue(typeIdx)?.ToString()?.Trim();
            if (!string.Equals(type, "Table", StringComparison.OrdinalIgnoreCase))
                continue;

            var name = reader.IsDBNull(nameIdx) ? null : reader.GetValue(nameIdx)?.ToString()?.Trim();
            if (!string.IsNullOrEmpty(name))
                yield return name;
        }
    }

    private static string MapColumnType(DbfColumn col)
    {
        return col.ColumnType switch
        {
            DbfColumnType.Character => $"Character({col.Length})",
            DbfColumnType.WideCharacter => $"Varchar({col.Length})",
            DbfColumnType.Memo => "Memo",
            DbfColumnType.Number => col.DecimalCount > 0
                ? $"Numeric({col.Length},{col.DecimalCount})"
                : $"Numeric({col.Length})",
            DbfColumnType.Float => $"Float({col.Length},{col.DecimalCount})",
            DbfColumnType.Double => "Double",
            DbfColumnType.SignedLong => "Integer",
            DbfColumnType.Currency => "Currency",
            DbfColumnType.Date => "Date",
            DbfColumnType.DateTime => "DateTime",
            DbfColumnType.Boolean => "Logical",
            DbfColumnType.General => "General",
            _ => col.ColumnType.ToString()
        };
    }

    private static Encoding ResolveEncoding(int? overrideCodePage, string? dbcPath, string folder)
    {
        // 1) Override esplicito dell'utente
        if (overrideCodePage.HasValue)
        {
            var enc = TryGetEncoding(overrideCodePage.Value);
            if (enc != null) return enc;
        }

        // 2) Auto-detect dal byte "language driver" dell'header (dbc o prima tabella della cartella)
        try
        {
            string? sampleDbf = dbcPath;
            if (sampleDbf == null && Directory.Exists(folder))
                sampleDbf = Directory.EnumerateFiles(folder, "*.dbf", SearchOption.TopDirectoryOnly).FirstOrDefault();

            if (sampleDbf != null && File.Exists(sampleDbf))
            {
                using var fs = File.OpenRead(sampleDbf);
                var header = new DbfHeader(fs);
                if (LanguageDriverCodePages.TryGetValue(header.LanguageDriver, out var cp))
                {
                    var enc = TryGetEncoding(cp);
                    if (enc != null) return enc;
                }
            }
        }
        catch
        {
            // ignora: si usa il default
        }

        // 3) Default
        return Encoding.GetEncoding(DefaultCodePage);
    }

    private static Encoding? TryGetEncoding(int codePage)
    {
        try { return Encoding.GetEncoding(codePage); }
        catch { return null; }
    }

    private static int? TryGetCodePageByName(string name)
    {
        try { return Encoding.GetEncoding(name).CodePage; }
        catch { return null; }
    }
}