hamkadr/src/JobsMedical.Web/Services/ListingParser.cs

using System.Text.RegularExpressions;
using JobsMedical.Web.Models;

namespace JobsMedical.Web.Services;

/// <summary>Structured guess extracted from a raw channel post. All fields are best-effort.</summary>
public class ParsedListing
{
    public ListingKind Kind { get; set; } = ListingKind.Shift;
    public string? RoleName { get; set; }
    public ShiftType? ShiftType { get; set; }
    public EmploymentType? EmploymentType { get; set; }
    public long? PayAmount { get; set; }        // shift pay or single salary figure
    public int? SharePercent { get; set; }      // profit-share % (درصدی / سهم درآمد)
    public bool PayNegotiable { get; set; }
    public Gender Gender { get; set; } = Gender.Any;   // جنسیت مورد نیاز
    public string? CityName { get; set; }
    public string? DistrictName { get; set; }
    public string? Phone { get; set; }
    public List<string> Notes { get; set; } = new();  // what was/wasn't detected (shown to admin)
}

/// <summary>
/// Turns a messy Persian channel/Divar post into a structured listing guess. This is the
/// Stage-1 implementation: transparent keyword + regex heuristics, no AI dependency (important
/// since LLM APIs are blocked from Iran). A future LlmListingParser can implement the same
/// interface and be swapped in via DI without touching the admin queue.
/// </summary>
public interface IListingParser
{
    ParsedListing Parse(string rawText, IEnumerable<string> knownRoles,
                        IEnumerable<string> knownCities, IEnumerable<string> knownDistricts);
}

public class HeuristicListingParser : IListingParser
{
    public ParsedListing Parse(string raw, IEnumerable<string> knownRoles,
        IEnumerable<string> knownCities, IEnumerable<string> knownDistricts)
    {
        var p = new ParsedListing();
        var text = Normalize(raw);

        // --- Kind: shift vs hiring ---
        bool jobSignals = ContainsAny(text, "استخدام", "جذب", "دعوت به همکاری", "تمام وقت", "تمام‌وقت", "قرارداد", "ماهانه", "حقوق ثابت");
        bool shiftSignals = ContainsAny(text, "شیفت", "آنکال", "انکال", "نوبت", "کشیک");
        p.Kind = (jobSignals && !shiftSignals) ? ListingKind.Job : ListingKind.Shift;
        p.Notes.Add(p.Kind == ListingKind.Job ? "نوع: استخدام (تشخیص خودکار)" : "نوع: شیفت (تشخیص خودکار)");

        // --- Role (longest match first so «پزشک متخصص» beats «پزشک») ---
        foreach (var role in knownRoles.OrderByDescending(r => r.Length))
        {
            if (text.Contains(Normalize(role))) { p.RoleName = role; break; }
        }
        if (p.RoleName is null && ContainsAny(text, "پزشک", "دکتر")) p.RoleName = "پزشک عمومی";
        p.Notes.Add(p.RoleName is null ? "نقش: تشخیص داده نشد" : $"نقش: {p.RoleName}");

        // --- Shift type ---
        if (ContainsAny(text, "آنکال", "انکال")) p.ShiftType = Models.ShiftType.OnCall;
        else if (text.Contains("شب")) p.ShiftType = Models.ShiftType.Night;
        else if (text.Contains("عصر")) p.ShiftType = Models.ShiftType.Evening;
        else if (ContainsAny(text, "صبح", "روز")) p.ShiftType = Models.ShiftType.Day;

        // --- Employment type ---
        if (ContainsAny(text, "پاره وقت", "پاره‌وقت", "پارت تایم")) p.EmploymentType = Models.EmploymentType.PartTime;
        else if (text.Contains("طرح")) p.EmploymentType = Models.EmploymentType.Plan;
        else if (text.Contains("قرارداد")) p.EmploymentType = Models.EmploymentType.Contract;
        else if (ContainsAny(text, "تمام وقت", "تمام‌وقت")) p.EmploymentType = Models.EmploymentType.FullTime;

        // --- Gender requirement ---
        if (ContainsAny(text, "خانم", "خانوم", "بانو", "زن ", "مامای")) p.Gender = Gender.Female;
        else if (ContainsAny(text, "آقا", "اقا", "مرد ", "مرد،", "پسر")) p.Gender = Gender.Male;
        if (p.Gender != Gender.Any)
            p.Notes.Add($"جنسیت: {(p.Gender == Gender.Female ? "خانم" : "آقا")}");

        // --- City / district ---
        p.CityName = knownCities.FirstOrDefault(c => text.Contains(Normalize(c)));
        p.DistrictName = knownDistricts.OrderByDescending(d => d.Length)
            .FirstOrDefault(d => text.Contains(Normalize(d)));

        // --- Profit share (درصدی / سهم) ---
        var latinForShare = ToLatinDigits(text);
        var share = Regex.Match(latinForShare, @"(\d{1,3})\s*(?:٪|%|درصد)");
        if (!share.Success) share = Regex.Match(latinForShare, @"(?:٪|%)\s*(\d{1,3})");
        if (share.Success && int.TryParse(share.Groups[1].Value, out var pct) && pct is > 0 and <= 100)
        { p.SharePercent = pct; p.Notes.Add($"سهم درآمد: {pct}٪"); }
        else if (ContainsAny(text, "درصدی", "سهم درآمد", "شراکت", "پورسانت"))
        { p.Notes.Add("پرداخت درصدی/سهمی (درصد نامشخص)"); }

        // --- Fixed pay ---
        if (ContainsAny(text, "توافقی", "توافق")) { p.PayNegotiable = true; p.Notes.Add("حقوق: توافقی"); }
        else
        {
            var amount = ExtractAmount(text);
            if (amount is not null) { p.PayAmount = amount; p.Notes.Add($"حقوق تخمینی: {amount:#,0} تومان"); }
            else if (p.SharePercent is null) p.Notes.Add("حقوق: تشخیص داده نشد");
        }

        // --- Phone ---
        var phone = Regex.Match(ToLatinDigits(text), @"0?9\d{9}");
        if (phone.Success) p.Phone = phone.Value;

        return p;
    }

    /// <summary>Pull a Toman figure out of free text, handling «میلیون» and Persian digits.</summary>
    private static long? ExtractAmount(string text)
    {
        var latin = ToLatinDigits(text);
        // e.g. "۲ میلیون" / "2.5 میلیون"
        var million = Regex.Match(latin, @"(\d+(?:[.,]\d+)?)\s*میلیون");
        if (million.Success && double.TryParse(million.Groups[1].Value.Replace(",", "."),
                System.Globalization.NumberStyles.Any, System.Globalization.CultureInfo.InvariantCulture, out var m))
            return (long)(m * 1_000_000);

        // Otherwise the largest plain number that looks like money (>= 6 digits after removing separators).
        long best = 0;
        foreach (Match num in Regex.Matches(latin, @"[\d٬,،.]{6,}"))
        {
            var digits = Regex.Replace(num.Value, @"[^\d]", "");
            if (digits.Length >= 6 && long.TryParse(digits, out var v) && v > best) best = v;
        }
        return best > 0 ? best : null;
    }

    private static string Normalize(string s) => s
        .Replace('ي', 'ی').Replace('ك', 'ک').Replace('‌', ' ').Trim();

    private static bool ContainsAny(string text, params string[] needles)
        => needles.Any(n => text.Contains(n));

    private static string ToLatinDigits(string s)
    {
        var chars = s.ToCharArray();
        for (var i = 0; i < chars.Length; i++)
        {
            if (chars[i] >= '۰' && chars[i] <= '۹') chars[i] = (char)('0' + (chars[i] - '۰'));
            else if (chars[i] >= '٠' && chars[i] <= '٩') chars[i] = (char)('0' + (chars[i] - '٠'));
        }
        return new string(chars);
    }
}