[Schmitzm-commits] r2195 - trunk/schmitzm-adresses/src/main/java/de/schmitzm/adresses

[scm-commit at wald.intevation.org]

Author: alfonx
Date: 2013-01-23 21:59:23 +0100 (Wed, 23 Jan 2013)
New Revision: 2195

Modified:
   trunk/schmitzm-adresses/src/main/java/de/schmitzm/adresses/AddrStringUtil.java
Log:
Schablonen gemastert

Modified: trunk/schmitzm-adresses/src/main/java/de/schmitzm/adresses/AddrStringUtil.java
===================================================================
--- trunk/schmitzm-adresses/src/main/java/de/schmitzm/adresses/AddrStringUtil.java	2013-01-23 19:47:40 UTC (rev 2194)
+++ trunk/schmitzm-adresses/src/main/java/de/schmitzm/adresses/AddrStringUtil.java	2013-01-23 20:59:23 UTC (rev 2195)
@@ -187,18 +187,26 @@
 
 		String[] s = splitInternal1(stringStrHnZusatz);
 
+		String[] nervigeZusaetzeAnFalscherStelle = new String[] { 
+//				"(?i)^(.*\\d.*)(\\d\\.\\d\\d)(.*)$",
+				"(?i)^(.*?)\\s*((?:\\d+\\.?)?\\s*(?:reihe|og\\s*(?:rechts|links|zi\\.?|zimmer|li|re|hinten|vorne|)+\\s*(?:\\d+\\.?)?))+\\s*+(.*?)$",
+		};
+
 		if (s[0] != null) {
 			RegexCache rC = RegexCache.getInstance();
-			Matcher matcher = rC.getMatcher("^(.*?)\\s*+(\\d+\\.\\s*reihe)\\s*+(.*?)$", s[0]);
-			if (matcher.find()) {
-				s[0] = matcher.group(1);
-				if (matcher.group(3).length() > 0)
-					s[0] += " " + matcher.group(3);
-				if (s[2] == null)
-					s[2] = "";
-				if (s[2].length() > 0)
-					s[2] += " ";
-				s[2] += matcher.group(2);
+			for (String nZ : nervigeZusaetzeAnFalscherStelle) {
+				Matcher matcher = rC.getMatcher(nZ, s[0]);
+				if (matcher.find()) {
+					s[0] = matcher.group(1);
+					if (matcher.group(3).length() > 0)
+						s[0] += " " + matcher.group(3);
+					// Reihen-Information wegschmeissen
+					// if (s[2] == null)
+					// s[2] = "";
+					// if (s[2].length() > 0)
+					// s[2] += " ";
+					// s[2] += matcher.group(2);
+				}
 			}
 		}
 
@@ -241,6 +249,9 @@
 		String zusammen = origStrasse + " " + (origHn != null ? " " + origHn : "")
 				+ (origZusatz != null ? " " + origZusatz : "");
 		zusammen = zusammen.replaceAll("3. Juni", "Dritten Juni");
+		zusammen = zusammen.replaceAll("(?i)^(.*\\d.*)\\b((?:zi?\\.?|zimmer)\\s*\\d+)(.*)$", "$1 $3");
+		zusammen = zusammen.replaceAll("(?i)^(.*\\d.*)\\b(we\\s*\\d\\.\\d\\d)(.*)$", "$1 $3");
+		zusammen = zusammen.replaceAll("(?i)^(.*\\d.*)(\\d\\.\\d\\d)(.*)$", "$1 $3");
 
 		zusammen = trimQuotes(zusammen);
 		zusammen = removeKommataAfterStreetname(zusammen);
@@ -675,6 +686,9 @@
 
 	static public String saeubereStrassenname(final String orginal) {
 		String strasse = LangUtil.removeTags(orginal).toLowerCase();
+		
+		// HN. am Ende
+		strasse = RegexCache.getInstance().replaceAll(strasse, "(\\d)\\s*\\.$", "$1");
 
 		// Forststr. , 3
 		strasse = RegexCache.getInstance().replaceAll(strasse, "(\\D\\D\\D)\\s*[,](\\s*\\d+\\D*)$", "$1 $2");