ReportParser.cs

using System.Globalization;

namespace Instrument.Core;

/// <summary>
/// 儀器報表解析器。
///
/// 真實痛點不在「解析文字檔」，而在**同一個物理量，各廠商叫不同名字**：
/// Agilent 寫 "Calibration ID"、Waters 寫 "Calib. File"、舊機台寫 "CalID"。
/// 所以欄位對映表才是這個元件的核心，解析邏輯本身很單純。
/// 這張表就是「來源端資料規格」的具體形式。
/// </summary>
public sealed class ReportParser
{
    /// <summary>欄位別名 → 正規欄位名。跨廠商整合的核心資產。</summary>
    private static readonly Dictionary<string, string> Aliases = new(StringComparer.OrdinalIgnoreCase)
    {
        ["instrument s/n"] = "serial",
        ["instrument sn"] = "serial",
        ["serial number"] = "serial",
        ["device id"] = "serial",

        ["sample id"] = "sample",
        ["sample name"] = "sample",
        ["sample"] = "sample",

        ["acquired"] = "measured_at",
        ["acquisition date"] = "measured_at",
        ["run date"] = "measured_at",
        ["date/time"] = "measured_at",

        ["calibration id"] = "cal_id",
        ["calib. file"] = "cal_id",
        ["calid"] = "cal_id",
        ["calibrated"] = "cal_date",
        ["calibration date"] = "cal_date",
        ["standard"] = "cal_standard",
        ["calibration standard"] = "cal_standard",

        ["column"] = "column",
        ["solvent"] = "solvent",
        ["mobile phase"] = "solvent",

        ["mn"] = "mn",
        ["mw"] = "mw",
        ["pdi"] = "pdi",
        ["polydispersity"] = "pdi",
        ["mw/mn"] = "pdi",

        ["tg (c)"] = "tg",
        ["tg"] = "tg",
        ["glass transition"] = "tg",
        ["tm (c)"] = "tm",
        ["tm"] = "tm",
        ["melting point"] = "tm",

        ["heating rate"] = "scan_rate",
        ["scan rate"] = "scan_rate",
        ["ramp"] = "scan_rate",

        ["sample mass (mg)"] = "mass",
        ["sample weight"] = "mass",
        ["mass"] = "mass",
    };

    /// <summary>從檔案解析。內部委派給 <see cref="ParseContent"/>，讓通訊來源共用同一套解析邏輯。</summary>
    public ParseResult Parse(string filePath)
    {
        string content;
        try
        {
            content = File.ReadAllText(filePath);
        }
        catch (Exception ex)
        {
            return ParseResult.Failed(filePath, $"讀檔失敗：{ex.Message}");
        }
        return ParseContent(content, Path.GetFileName(filePath));
    }

    /// <summary>
    /// 從純文字內容解析。通訊層（TCP／串列埠／SDK）收到的是位元組串流而非檔案，
    /// 所以解析必須與「怎麼拿到資料」解耦。
    /// </summary>
    /// <param name="origin">來源識別（檔名／連線端點／NodeId），出問題時用來追溯。</param>
    public ParseResult ParseContent(string content, string origin)
    {
        var lines = content.Split(['\r', '\n'], StringSplitOptions.RemoveEmptyEntries);
        var fields = new Dictionary<string, string>(StringComparer.OrdinalIgnoreCase);
        var unknownKeys = new List<string>();

        foreach (var raw in lines)
        {
            var line = raw.Trim();
            if (line.Length == 0 || line.StartsWith('#') || line.StartsWith("---")) continue;

            var idx = line.IndexOf(':');
            if (idx <= 0) continue;

            var key = line[..idx].Trim();
            var value = line[(idx + 1)..].Trim();
            if (value.Length == 0) continue;

            if (Aliases.TryGetValue(key, out var canonical))
                fields[canonical] = value;
            else
                unknownKeys.Add(key);          // 未知欄位要浮出來，不能靜默吞掉
        }

        var type = DetectType(fields);
        if (type is null)
            return ParseResult.Failed(origin, "無法判定儀器種類（既無 GPC 也無 DSC 的特徵欄位）");

        var measurement = new Measurement
        {
            SampleId = fields.GetValueOrDefault("sample", ""),
            InstrumentSerial = fields.GetValueOrDefault("serial", ""),
            Instrument = type.Value,
            MeasuredAt = ParseDate(fields.GetValueOrDefault("measured_at")) ?? default,
            SourceFile = origin,

            Mn = ParseNum(fields.GetValueOrDefault("mn")),
            Mw = ParseNum(fields.GetValueOrDefault("mw")),
            ReportedPdi = ParseNum(fields.GetValueOrDefault("pdi")),
            CalibrationId = fields.GetValueOrDefault("cal_id"),
            CalibratedAt = ParseDate(fields.GetValueOrDefault("cal_date")),
            CalibrationStandard = fields.GetValueOrDefault("cal_standard"),
            Column = fields.GetValueOrDefault("column"),
            Solvent = fields.GetValueOrDefault("solvent"),

            TgCelsius = ParseNum(fields.GetValueOrDefault("tg")),
            TmCelsius = ParseNum(fields.GetValueOrDefault("tm")),
            ScanRateCPerMin = ParseNum(fields.GetValueOrDefault("scan_rate")),
            SampleMassMg = ParseNum(fields.GetValueOrDefault("mass")),
        };

        return ParseResult.Ok(measurement, unknownKeys);
    }

    private static InstrumentType? DetectType(Dictionary<string, string> f)
    {
        if (f.ContainsKey("mn") || f.ContainsKey("mw") || f.ContainsKey("cal_id"))
            return InstrumentType.Gpc;
        if (f.ContainsKey("tg") || f.ContainsKey("tm") || f.ContainsKey("scan_rate"))
            return InstrumentType.Dsc;
        return null;
    }

    /// <summary>數值可能帶千分位、單位後綴或前後空白，一律容忍。</summary>
    private static double? ParseNum(string? s)
    {
        if (string.IsNullOrWhiteSpace(s)) return null;
        var cleaned = new string(s.Where(c => char.IsDigit(c) || c is '.' or '-' or '+' or 'e' or 'E').ToArray());
        return double.TryParse(cleaned, NumberStyles.Float, CultureInfo.InvariantCulture, out var v)
            ? v : null;
    }

    private static DateTime? ParseDate(string? s)
    {
        if (string.IsNullOrWhiteSpace(s)) return null;
        string[] formats =
        [
            "yyyy-MM-dd HH:mm:ss", "yyyy-MM-dd HH:mm", "yyyy-MM-dd",
            "yyyy/MM/dd HH:mm:ss", "yyyy/MM/dd", "dd/MM/yyyy HH:mm", "MM/dd/yyyy HH:mm",
        ];
        return DateTime.TryParseExact(s, formats, CultureInfo.InvariantCulture,
                   DateTimeStyles.None, out var d)
            ? d
            : DateTime.TryParse(s, CultureInfo.InvariantCulture, DateTimeStyles.None, out var g)
                ? g : null;
    }
}

public sealed record ParseResult(
    bool Success, Measurement? Measurement, string? Error,
    IReadOnlyList<string> UnknownKeys, string? FilePath = null)
{
    public static ParseResult Ok(Measurement m, IReadOnlyList<string> unknown) =>
        new(true, m, null, unknown);

    public static ParseResult Failed(string filePath, string error) =>
        new(false, null, error, [], filePath);
}