[WIP] Bio Update

Elwig/Helpers/Normalizer.cs

@@ -62,7 +62,7 @@ namespace Elwig.Helpers {
 
         private static readonly HashSet<string> StreetGivenNames = [
             "Adalbert", "Adam", "Adele", "Adelheid", "Adolf", "Adolph", "Adrian", "Albert", "Albrecht", "Alexander", "Alfons", "Alfred", "Alois", "Aloisia",
-            "Andreas", "Andrea", "Andre", "Anna", "Anton", "Arnold", "Arthur", "August", "Augustin", "Augustine", "Agnes", "Angelika", "Ada", "Adelheid", "Anni",
+            "Andreas", "Andrea", "Andre", "Anna", "Anton", "Arnold", "Arthur", "August", "Augustin", "Augustine", "Agnes", "Angelika", "Ada", "Adelheid", "Anni", "Anita",
             "Barbara", "Beate", "Beatrix", "Benedikt", "Bernhard", "Berta", "Bertha", "Berthold", "Brüder", "Bruno", "Brigitte",
             "Camillo", "Carl", "Caspar", "Catharina", "Christian", "Christine", "Christof", "Christoph", "Clara", "Clemens", "Conrad", "Claus", "Camillo", "Claudia",
             "Daniel", "Dietmar", "Dietrich",
@@ -93,6 +93,7 @@ namespace Elwig.Helpers {
             "Zacharias"];
 
         private static readonly HashSet<string> LocalityNames = [
+            "Groß-Enzersdorf",
             "Groß Burgstall", "Klein Eberharts", "Klein-Engersdorf", "Groß-Enzersdorf", "Klein-Erlauf",
             "Groß Gerungs", "Klein Gundholz", "Klein-Harras", "Klein-Heinrichschlag", "Groß-Höbarten",
             "Klein-Jetzelsdorf", "Klein-Mariazell", "Groß Meinharts", "Klein-Meiseldorf", "Groß-Neusiedl", "Klein-Neusiedl", "Klein-Pöchlarn",
@@ -101,24 +102,64 @@ namespace Elwig.Helpers {
             "Klein-Ulrichschlag", "Klein Weinberg", "Klein Wetzles", "Klein-Wien",
         ];
 
-        private static bool IsGeographical(string word) {
-            if (!word.EndsWith("er") || StreetLastNames.Contains(word) || word == "Weißer")
+        private static bool IsGivenName(this string word) {
+            return StreetGivenNames.Contains(word);
+        }
+
+        private static bool IsFamilyName(this string word) {
+            if (StreetLastNames.Contains(word))
+                return true;
+            return false;
+        }
+
+        private static bool IsPersonalName(this string word) {
+            return word.IsGivenName() || word.IsFamilyName();
+        }
+
+        private static bool IsGeographicalDerivativeName(this string word) {
+            if (!word.EndsWith("er") || word.IsPersonalName() || word == "Weißer")
                 return false;
-            // FIXME
-            return word.EndsWith("erer") || word == "Wiener" || word == "Brünner" || word == "Znaimer" || word == "Triester" || word == "Prager" || word == "Berliner" ||
-                word == "St. Pöltner" ||
-                word == "Gruber" || word == "Sulzer" || word == "Tullner" || word == "Harraser" || word == "Dörfleser" || word == "Neusiedler" ||
-                word.EndsWith("thaler") || word.EndsWith("inger") || word.EndsWith("dorfer") || word.EndsWith("berger") || word.EndsWith("burger") ||
-                word.EndsWith("tzer") || word.EndsWith("brunner") || word.EndsWith("rther") || word.EndsWith("bacher") || word.EndsWith("städter") ||
-                word.EndsWith("auer") || word.EndsWith("winkler") || word.EndsWith("kirchner") || word.EndsWith("felder") || word.EndsWith("frieder") ||
-                word.EndsWith("hauser") || word.EndsWith("ßer") || word.EndsWith("stättner") || word.EndsWith("dner") ||
-                word.EndsWith("tzner") || word.EndsWith("einer") || word.EndsWith("heimer") || word.EndsWith("brücker") || word.EndsWith("ziger") ||
-                word.EndsWith("zeller") || word.EndsWith("gerner") || word.EndsWith("oner") || word.EndsWith("aaer");
+            return word.StartsWith("St. ") || word.StartsWith("Ober ") || word.StartsWith("Unter ") || word.StartsWith("Maria") ||
+                word.EndsWith("sdorfer") || word.EndsWith("ndorfer") || word.EndsWith("ldorfer") ||
+                word.EndsWith("felder") ||
+                word.EndsWith("sthaler") || word.EndsWith("nthaler") || word.EndsWith("lthaler") ||
+                word.EndsWith("dauer") || word.EndsWith("lauer") || word.EndsWith("mauer") || word.EndsWith("nauer") || word.EndsWith("ßauer") ||
+                word.EndsWith("lbacher") || word.EndsWith("nbacher") || word.EndsWith("ßbacher") ||
+                word.EndsWith("ringer") || word.EndsWith("singer") ||
+                word.EndsWith("sbrunner") || word.EndsWith("nbrunner") || word.EndsWith("lbrunner") || word.EndsWith("abrunner") || word.EndsWith("ingbrunner") ||
+                word.EndsWith("sburger") || word.EndsWith("nburger") || word.EndsWith("mburger") || word.EndsWith("lburger") || word.EndsWith("neuburger") ||
+                word.EndsWith("berger") ||
+                word.EndsWith("winkler") ||
+                word.EndsWith("lsteiner") || word.EndsWith("nsteiner") ||
+                word.EndsWith("städter") ||
+                word.EndsWith("stättner") || word.EndsWith("stettner") ||
+                word.EndsWith("brücker") ||
+                word.EndsWith("further") || word.EndsWith("furter") ||
+                word.EndsWith("aaer") ||
+                word.EndsWith("seer") || word.EndsWith("leer") ||
+                word.EndsWith("tzner") ||
+                word.EndsWith("zeller") ||
+                word.EndsWith("rther") ||
+                word.EndsWith("sdner") ||
+                word.EndsWith("gerner") || word.EndsWith("berner") ||
+                word.EndsWith("frieder") ||
+                word.EndsWith("litzer") || word.EndsWith("mitzer") || word.EndsWith("eutzer") ||
+                word.EndsWith("loner") ||
+                word.EndsWith("skirchner") || word.EndsWith("nkirchner") || word.EndsWith("ingkirchner") ||
+                word.EndsWith("ziger") ||
+                word.EndsWith("ließer") ||
+                word.EndsWith("gerer") ||
+                word.EndsWith("kreuzer") ||
+                word == "Wiener" || word == "Brünner" || word == "Znaimer" || word == "Triester" || word == "Prager" || word == "Berliner" || word == "Basler" ||
+                word == "Retzer" || word == "Tullner" || word == "Badner" ||
+                word == "Gruber" || word == "Sulzer" || word == "Harraser" || word == "Dörfleser" || word == "Neusiedler" ||
+                word == "Dornauer" || word == "Siebenhauser" || word == "Alser" ||
+                word == "Bockenheimer";
         }
 
         public static string NormalizeAddress(string address, string? locality = null) {
             var idx = address.IndexOf(',');
-            if (idx >= 0) address = address.Substring(0, idx);
+            if (idx >= 0) address = address[..idx];
             var (name, nr) = Utils.SplitAddress(address);
             return (NormalizeStreetName(name, locality) + " " + nr).Trim();
         }
@@ -157,11 +198,11 @@ namespace Elwig.Helpers {
                 .Replace("|Kalk|Deutz|", "|Kalk-Deutz|").Replace("|Kalk|Deutzer|", "|Kalk-Deutzer|")
                 .Replace("|Hohe|Wand|", "|Hohe Wand|").Replace("|Hohen|Wand|", "|Hohen Wand|").Replace("|Hohewand|", "|Hohe Wand|")
                 .Replace("|Alten|Markt|", "|Altenmarkt|")
-                .Replace("|Kirch|Hof|", "|Kirchhof|")
+                .Replace("|Kirch|Hof|", "|Kirchhof|").Replace("|Bisc|Hof|", "|Bischof|")
                 .Replace("|Wiener|Neustadt|", "|Wiener Neustadt|").Replace("|Wiener|Neustädter|", "|Wiener Neustädter|")
                 .Replace("|St.|Pölten|", "|St. Pölten|").Replace("|St.|Pöltner|", "|St. Pöltner|")
                 .Replace("|Gross|", "|Groß|")
-                .Replace("|Nr|", "|Nr.|");
+                .Replace("|Nr|", "|Nr.|").Replace("|Und|", "|und|");
             words = [.. temp.Trim('|').Split('|')];
 
             if (words.Count > 1 && words[0] == "Groß" || words[0] == "Klein" && !StreetTypes.Contains(words[1])) {
@@ -193,11 +234,11 @@ namespace Elwig.Helpers {
                 var isGeographical = false;
                 var isPerson = false;
 
-                var geo = words.FindLastIndex(IsGeographical);
+                var geo = words.FindLastIndex(IsGeographicalDerivativeName);
                 if (geo >= 0 && geo < words.Count - 2 && words[^2] != "Haupt" && words[^2] != "Bundes" && words[^2] != "Keller") {
                     isPerson = true;
                     isGeographical = true;
-                } else if (words.Any(w => w.Contains(' ') && !IsGeographical(w))) {
+                } else if (words.Any(w => w.Contains(' ') && !w.IsGeographicalDerivativeName())) {
                     isPerson = true;
                     isGeographical = true;
                 } else if (StreetNumbers.Contains(words[0])) {
@@ -209,7 +250,7 @@ namespace Elwig.Helpers {
                 }
 
                 if (words.Count(w => w.Contains('.')) >= 2 || words.Contains("van") || words.Contains("von") || words.Contains("St.") ||
-                    StreetPersonTitles.Contains(words[0]) || words.Any(StreetGivenNames.Contains))
+                    StreetPersonTitles.Contains(words[0]) || words.Any(w => w.IsGivenName()))
                 {
                     isPerson = true;
                 }
@@ -219,7 +260,7 @@ namespace Elwig.Helpers {
                 }
 
                 if (!isPerson && words[^2].EndsWith("er") && !StreetLastNames.Contains(words[^2])) {
-                    isGeographical = IsGeographical(words[^2]);
+                    isGeographical = IsGeographicalDerivativeName(words[^2]);
                 } else if (words.Count >= 3 && words.Contains("Bad")) {
                     isPerson = true;
                     isGeographical = true;