Further fix right quotation mark parsing

diff --git a/cdmlib-commons/src/main/java/eu/etaxonomy/cdm/common/UTF8.java b/cdmlib-commons/src/main/java/eu/etaxonomy/cdm/common/UTF8.java
index e9b65ebe632d81ce318323ccfa564f3a84da3920..b8cc0845642f2174ae09c30ec165b4242cba056a 100644 (file)
--- a/cdmlib-commons/src/main/java/eu/etaxonomy/cdm/common/UTF8.java
+++ b/cdmlib-commons/src/main/java/eu/etaxonomy/cdm/common/UTF8.java
@@ -1,35 +1,37 @@
-/**\r
- * \r
- */\r
-package eu.etaxonomy.cdm.common;\r
-\r
-/**\r
- * This class is a constant holder for commonly used UTF-8 characters.\r
- *  \r
- * @author a.mueller\r
- * @since 19.06.2013\r
- */\r
-public enum UTF8 {\r
-       \r
-       \r
-       EN_DASH("\u2013"),   // https://de.wikipedia.org/wiki/Halbgeviertstrich\r
-       SPATIUM("\u202F"),   //very short non-breaking space\r
-       EN_DASH_SPATIUM("\u202F\u2013\u202F"),\r
-       HYBRID ("\u00D7"),   // hybrid sign\r
-       SHARP_S("\u00DF"),\r
-       NO_BREAK_SPACE("\u00A0"),\r
-       POLISH_L("\u0142"),\r
-       SMALL_A_ACUTE("\u00E1")\r
-       ;\r
-\r
-       private String value;\r
-       \r
-       private UTF8(String value) {\r
-               this.value = value;\r
-       }\r
-       \r
-       public String toString(){\r
-               return value;\r
-       }\r
-       \r
-}\r
+/**
+ *
+ */
+package eu.etaxonomy.cdm.common;
+
+/**
+ * This class is a constant holder for commonly used UTF-8 characters.
+ *
+ * @author a.mueller
+ * @since 19.06.2013
+ */
+public enum UTF8 {
+
+
+       EN_DASH("\u2013"),   // https://de.wikipedia.org/wiki/Halbgeviertstrich
+       SPATIUM("\u202F"),   //very short non-breaking space
+       EN_DASH_SPATIUM("\u202F\u2013\u202F"),
+       HYBRID ("\u00D7"),   // hybrid sign
+       SHARP_S("\u00DF"),
+       NO_BREAK_SPACE("\u00A0"),
+       POLISH_L("\u0142"),
+       SMALL_A_ACUTE("\u00E1"),
+       RIGHT_SINGLE_QUOT("\u2019") // Right single quotation mark
+       ;
+
+       private String value;
+
+       private UTF8(String value) {
+               this.value = value;
+       }
+
+       @Override
+    public String toString(){
+               return value;
+       }
+
+}

diff --git a/cdmlib-model/src/main/java/eu/etaxonomy/cdm/strategy/parser/NonViralNameParserImplRegExBase.java b/cdmlib-model/src/main/java/eu/etaxonomy/cdm/strategy/parser/NonViralNameParserImplRegExBase.java
index 599afcd25d4817dbf53d61efa9dc55bf1476e184..abb7171a070d878d2ad62ab513d5ce803f9b6e9d 100644 (file)
--- a/cdmlib-model/src/main/java/eu/etaxonomy/cdm/strategy/parser/NonViralNameParserImplRegExBase.java
+++ b/cdmlib-model/src/main/java/eu/etaxonomy/cdm/strategy/parser/NonViralNameParserImplRegExBase.java
@@ -84,7 +84,8 @@ public abstract class NonViralNameParserImplRegExBase  {
 
 
     //AuthorString
-       protected static String authorPart = "(" + "(O[’']|d[’']|D[’']|L[’']|'t|ten\\s||le\\s|zur\\s)?" + "(" + capital2charDotWord + "|DC.)" + "('" + nonCapitalDotWord + ")?" + "|[vV][ao]n(\\sder)?|da|du|de(n|l|\\sla)?)" ;
+    protected static String qm = "[" + UTF8.RIGHT_SINGLE_QUOT + "']";
+    protected static String authorPart = "(" + "([OdDL]"+qm+"|[’']t|ten\\s||le\\s|zur\\s)?" + "(" + capital2charDotWord + "|DC.)" + "('" + nonCapitalDotWord + ")?" + "|[vV][ao]n(\\sder)?|da|du|de(n|l|\\sla)?)" ;
     protected static String author = "(" + authorPart + "(" + fWs + "|-)" + ")+" + "(f\\.|fil\\.|secundus)?" ;
     protected static String finalTeamSplitter = "(" + fWs + "(&)" + fWs + "|" + oWs + "et" + oWs + ")";
     protected static String notFinalTeamSplitter = "(?:" + fWs + "," + fWs + "|" + finalTeamSplitter + ")";