hamkadr/src/JobsMedical.Web/Services/Scraping/FacilityMatcher.cs

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

namespace JobsMedical.Web.Services.Scraping;

/// <summary>
/// Persian-aware fuzzy matching for facility names, so the same hospital written slightly
/// differently — spacing, ي/ك vs ی/ک, ZWNJ, with or without «بیمارستان» — resolves to one
/// record instead of creating a duplicate. Used by both the manual review/publish flow and
/// the auto-publish ingestion pipeline.
/// </summary>
public static class FacilityMatcher
{
    // Generic type words stripped to compare the distinctive core of a name.
    private static readonly string[] TypeWords =
    {
        "بیمارستان", "زایشگاه", "پلی کلینیک", "پلیکلینیک", "درمانگاه", "کلینیک",
        "مرکز درمانی", "مرکز جراحی", "مجتمع پزشکی", "مجتمع درمانی", "مرکز", "مجتمع",
        "آزمایشگاه", "داروخانه", "آسایشگاه", "مطب", "تخصصی", "فوق تخصصی", "فوقتخصصی", "عمومی", "دکتر", "دی کلینیک",
        // Generic descriptors — never the distinctive part of a name. Stripping them stops false
        // merges like «درمانگاه شبانه‌روزی اسفند» → «پلی کلینیک شبانه روزی» (they share «شبانه روزی»).
        "شبانه روزی", "شبانه‌روزی", "خیریه", "دولتی", "خصوصی", "۲۴ ساعته", "24 ساعته", "تامین اجتماعی",
    };

    /// <summary>Lower-cased, Arabic→Persian folded, punctuation-stripped, whitespace-collapsed.</summary>
    public static string Normalize(string? s)
    {
        if (string.IsNullOrWhiteSpace(s)) return "";
        var t = s.Replace('ي', 'ی').Replace('ك', 'ک').Replace('ۀ', 'ه').Replace('ة', 'ه')
                 .Replace('أ', 'ا').Replace('إ', 'ا').Replace('آ', 'ا').Replace('ئ', 'ی')
                 .Replace('‌', ' ').ToLowerInvariant();
        var sb = new StringBuilder(t.Length);
        foreach (var ch in t)
            sb.Append(char.IsLetterOrDigit(ch) || ch == ' ' ? ch : ' ');
        return Regex.Replace(sb.ToString(), @"\s+", " ").Trim();
    }

    /// <summary>Normalized name with generic type words removed — the distinctive part.</summary>
    public static string Core(string? s)
    {
        var n = Normalize(s);
        if (n.Length == 0) return "";
        foreach (var w in TypeWords)
        {
            var nw = Normalize(w);
            if (nw.Length == 0) continue;
            n = Regex.Replace(n, $@"(?<![\p{{L}}\p{{N}}]){Regex.Escape(nw)}(?![\p{{L}}\p{{N}}])", " ");
        }
        return Regex.Replace(n, @"\s+", " ").Trim();
    }

    // Filler/verb/locator tokens that are never a real facility name — the parser sweeps these in
    // when an ad has no named facility («بیمارستان هستم», «مطب نیازمندیم سه», «کلینیک های فقط منطقه»).
    private static readonly string[] JunkCoreWords =
    {
        "هستم", "هستیم", "هستش", "میشوم", "میشم", "بشوم", "میباشد", "باشد", "میباشم",
        "نیازمندیم", "نیازمند", "نیازمندم", "داریم", "دارم", "میخواهیم", "میخوام",
        "حتی", "تعدادی", "فقط", "منطقه", "واقع", "های", "مبتدی", "محترم", "خوب",
        "سه", "دو", "یک", "چند", "این", "آن", "همکار", "نیرو",
    };

    // Crawl-source names that must never appear as a public facility («مرکز درمانی (از مدجابز)»),
    // plus the shared placeholder text.
    private static readonly string[] SourceMarkers =
    {
        "مدجابز", "مدجاب", "از تلگرام", "از دیوار", "از بله", "از کانال", "ثبت نشده", "نامشخص",
    };

    /// <summary>
    /// True when a name is NOT a usable facility name: a bare type word («بیمارستان»), a name whose
    /// distinctive core is only filler/verb tokens («بیمارستان هستم» → «هستم»), or a leaked crawl
    /// source / placeholder («... از مدجابز», «نامشخص»). Such an ad has no real named facility and
    /// should fall back to the shared placeholder instead of forging a fake one.
    /// </summary>
    public static bool IsJunkName(string? name)
    {
        var normalized = Normalize(name);
        if (normalized.Length == 0) return true;
        if (SourceMarkers.Any(m => normalized.Contains(Normalize(m)))) return true;

        var core = Core(name);
        if (core.Length == 0) return true;   // bare type word only («بیمارستان»، «کلینیک»)

        var tokens = core.Split(' ', StringSplitOptions.RemoveEmptyEntries);
        return tokens.All(t => t.Length <= 1 || JunkCoreWords.Contains(t));
    }

    /// <summary>True when two names almost certainly denote the same facility.</summary>
    public static bool IsSame(string? a, string? b)
    {
        var na = Normalize(a);
        var nb = Normalize(b);
        if (na.Length == 0 || nb.Length == 0) return false;
        if (na == nb) return true;

        var ca = Core(a);
        var cb = Core(b);
        if (ca.Length >= 2 && ca == cb) return true;
        // one core fully contains the other (e.g. «میلاد» vs «میلاد ۱»)
        if (ca.Length >= 3 && cb.Length >= 3 && (ca.Contains(cb) || cb.Contains(ca))) return true;

        // edit-distance similarity on the most informative basis
        var (x, y) = ca.Length >= 3 && cb.Length >= 3 ? (ca, cb) : (na, nb);
        return Similarity(x, y) >= 0.86;
    }

    /// <summary>
    /// Best existing facility for <paramref name="name"/>: same-city exact match first, then
    /// any-city exact, then same-city fuzzy, then any-city fuzzy. Null when nothing matches.
    /// </summary>
    public static Facility? FindBest(IEnumerable<Facility> facilities, string? name, int? cityId)
    {
        if (string.IsNullOrWhiteSpace(name)) return null;
        var list = facilities as IList<Facility> ?? facilities.ToList();
        var target = Normalize(name);

        return list.FirstOrDefault(f => cityId.HasValue && f.CityId == cityId && Normalize(f.Name) == target)
            ?? list.FirstOrDefault(f => Normalize(f.Name) == target)
            ?? list.FirstOrDefault(f => cityId.HasValue && f.CityId == cityId && IsSame(f.Name, name))
            ?? list.FirstOrDefault(f => IsSame(f.Name, name));
    }

    private static double Similarity(string a, string b)
    {
        if (a == b) return 1;
        var max = Math.Max(a.Length, b.Length);
        return max == 0 ? 1 : 1.0 - (double)Levenshtein(a, b) / max;
    }

    private static int Levenshtein(string a, string b)
    {
        var dp = new int[b.Length + 1];
        for (var j = 0; j <= b.Length; j++) dp[j] = j;
        for (var i = 1; i <= a.Length; i++)
        {
            var prev = dp[0];
            dp[0] = i;
            for (var j = 1; j <= b.Length; j++)
            {
                var tmp = dp[j];
                dp[j] = Math.Min(Math.Min(dp[j] + 1, dp[j - 1] + 1), prev + (a[i - 1] == b[j - 1] ? 0 : 1));
                prev = tmp;
            }
        }
        return dp[b.Length];
    }
}