Match crawled listings to existing facilities (fuzzy) before creating new
CI/CD / CI · dotnet build (push) Successful in 1m28s
CI/CD / Deploy · hamkadr (push) Successful in 2m24s

When publishing a scraped listing we now look for a facility we already
have that is exactly or closely the same, and only create a new one when
there is no match — avoiding duplicates like «بیمارستان میلاد» vs «میلاد».

- ListingParser: extract a facility name (keyword + distinctive words) from
  the post and surface it in the parser notes.
- FacilityMatcher: Persian-aware normalization (ي/ك, ZWNJ, punctuation),
  type-word stripping for a "core" name, contains + Levenshtein similarity,
  and FindBest (same-city exact → any-city exact → same-city fuzzy → fuzzy).
- Review (manual publish): auto-select a matching facility or prefill the
  new-facility name; resolve-or-create uses fuzzy match; dropdown preselects.
- IngestionService (auto-publish): reuse FacilityMatcher against a run-wide
  facility list (grows as new ones are created) instead of exact-name only.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
soroush.asadi
2026-06-08 07:14:48 +03:30
parent a2fc70ae57
commit e6a796ab27
5 changed files with 191 additions and 10 deletions
@@ -16,6 +16,7 @@ public class ParsedListing
public Gender Gender { get; set; } = Gender.Any; // جنسیت مورد نیاز
public string? CityName { get; set; }
public string? DistrictName { get; set; }
public string? FacilityName { get; set; } // hospital/clinic name guessed from the text
public string? Phone { get; set; }
public List<string> Notes { get; set; } = new(); // what was/wasn't detected (shown to admin)
}
@@ -107,6 +108,10 @@ public class HeuristicListingParser : IListingParser
else if (p.SharePercent is null) p.Notes.Add("حقوق: تشخیص داده نشد");
}
// --- Facility name (بیمارستان/درمانگاه/کلینیک ... + the distinctive name) ---
p.FacilityName = ExtractFacilityName(text);
if (p.FacilityName is not null) p.Notes.Add($"مرکز: {p.FacilityName}");
// --- Phone ---
var phone = Regex.Match(ToLatinDigits(text), @"0?9\d{9}");
if (phone.Success) p.Phone = phone.Value;
@@ -114,6 +119,48 @@ public class HeuristicListingParser : IListingParser
return p;
}
// Words that introduce a facility name, longest/most-specific first.
private static readonly string[] FacilityKeywords =
{
"بیمارستان", "زایشگاه", "پلی کلینیک", "پلیکلینیک", "درمانگاه", "کلینیک",
"مرکز درمانی", "مرکز جراحی", "مجتمع پزشکی", "مجتمع درمانی", "مرکز", "مجتمع",
"آزمایشگاه", "مطب", "خانه سالمندان", "سرای سالمندان",
};
// Words that clearly aren't part of a facility's name — stop collecting here.
private static readonly string[] NameStops =
{
"جهت", "برای", "به", "با", "در", "از", "که", "نیاز", "نیازمند", "استخدام", "جذب",
"دعوت", "همکاری", "واقع", "آدرس", "تلفن", "شماره", "شیفت", "ساعت", "حقوق", "روز",
"شب", "صبح", "عصر", "می", "ها", "این", "یک", "محترم",
};
/// <summary>Best-effort hospital/clinic name: a facility keyword plus up to three name words.</summary>
private static string? ExtractFacilityName(string text)
{
foreach (var kw in FacilityKeywords)
{
var idx = text.IndexOf(kw, StringComparison.Ordinal);
if (idx < 0) continue;
var after = text[(idx + kw.Length)..];
var words = after.Split(
new[] { ' ', '\n', '\r', '\t', '،', ',', '.', '؛', ':', '(', ')', '-', '/', '«', '»', '"' },
StringSplitOptions.RemoveEmptyEntries);
var picked = new List<string>();
foreach (var w in words)
{
if (NameStops.Contains(w)) break;
if (Regex.IsMatch(w, @"\d")) break; // numbers/phones aren't names
if (w.Length == 1) break; // stray letters
picked.Add(w);
if (picked.Count >= 3) break;
}
if (picked.Count == 0) continue; // bare keyword (e.g. just «بیمارستان») isn't useful
return (kw + " " + string.Join(" ", picked)).Trim();
}
return null;
}
/// <summary>Pull a Toman figure out of free text, handling «میلیون» and Persian digits.</summary>
private static long? ExtractAmount(string text)
{