museumsvictoria#157: treat 0xa0 NO-BREAK-SPACE; and tidy-up

scroix · Apr 1, 2024 · b8916be · b8916be
1 parent 85723e1
commit b8916be
Show file tree

Hide file tree

Showing 3 changed files with 14 additions and 17 deletions.
diff --git a/nodel-framework/src/main/java/org/nodel/SimpleName.java b/nodel-framework/src/main/java/org/nodel/SimpleName.java
@@ -373,13 +373,13 @@ public static String flatten(String name, char[] passthrough) {
                     sb.append(Character.toLowerCase(c));
             }
 
-            // spaces and most ASCII control codes
+            // skip spaces and most ASCII control codes
             else if (c <= 32) {
                 continue;
             }
 
-            // else ASCII letters or digits and everything else (extended ASCII and Unicode)
-            else if (c > 127 || Character.isLetterOrDigit(c)) {
+            // else incl. letters or digits and everything else non-7bit-ASCII that is not a space
+            else if (Character.isLetterOrDigit(c) || (c > 127 && !Character.isSpaceChar(c))) {
                 String flatDialect = s_diacritic_tolowerascii.get(c);
                 if (flatDialect != null)
                     sb.append(flatDialect);
@@ -413,16 +413,16 @@ public static void flattenChar(char c, StringBuilder sb) {
         if (c <= 32)
             return;
 
-        // include ASCII letters or digits and everything else
-        if (c > 127 || Character.isLetterOrDigit(c)) {
+        // include letters or digits and everything else non-7-bit-ASCII that is not a space
+        if (Character.isLetterOrDigit(c) || (c > 127 && !Character.isSpaceChar(c))) {
             String flatDialect = s_diacritic_tolowerascii.get(c);
             if (flatDialect != null)
                 sb.append(flatDialect);
             else
                 sb.append(Character.toLowerCase(c));
         }
 
-        // ...skip the remaining ASCII symbols 
+        // ...skip remaining
     }
 
 } // (class)
diff --git a/nodel-framework/src/main/java/org/nodel/Strings.java b/nodel-framework/src/main/java/org/nodel/Strings.java
@@ -44,11 +44,11 @@ public static boolean isBlank(String value) {
         if (len == 0)
             return true;
 
-        // ...and then for any non common-whitespace
+        // ...and then for any non common-whitespace, incl. 0xa0 NO-BREAK SPACE
         for (int a = 0; a < len; a++) {
             char c = value.charAt(a);
 
-            if (c != ' ' && c != '\t' && c != '\r' && c != '\n')
+            if (c != ' ' && c != '\t' && c != '\r' && c != '\n' && !Character.isSpaceChar(c))
                 return false;
         }
 

diff --git a/nodel-framework/src/main/java/org/nodel/core/Nodel.java b/nodel-framework/src/main/java/org/nodel/core/Nodel.java
@@ -25,9 +25,6 @@
 import org.nodel.discovery.AutoDNS;
 import org.nodel.io.UnexpectedIOException;
 
-// TODO: use one of these to implement character normalisation:
-// http://stackoverflow.com/questions/1008802/converting-symbols-accent-letters-to-english-alphabet
-
 public class Nodel {
 
     private final static String VERSION = "2.2.1";
@@ -112,9 +109,9 @@ else if (removeCommentsOnly)
             else if (Character.isLetterOrDigit(c))
                 sb.append(c);
 
-            else if (c > 127)
+            else if (c > 127 && !Character.isSpaceChar(c))
                 // every other extended ASCII and Unicode character
-                // is significant
+                // is significant (except space characters e.g. \u00A0 NO-BREAK SPACE)
                 sb.append(c);
 
             // store last char for comments
@@ -474,19 +471,19 @@ private static String formatAgent() {
         sb.append("nodel/").append(VERSION);
 
         String javaRuntime = System.getProperty("java.runtime.version");
-        if (!Strings.isNullOrEmpty(javaRuntime))
+        if (!Strings.isBlank(javaRuntime))
             sb.append(" java/").append(javaRuntime.replace(' ', '_'));
 
         String vendor = System.getProperty("java.vm.vendor");
-        if (!Strings.isNullOrEmpty(vendor))
+        if (!Strings.isBlank(vendor))
             sb.append(' ').append(vendor.replace(' ', '_'));        
 
         String os = System.getProperty("os.name");
-        if (!Strings.isNullOrEmpty(os))
+        if (!Strings.isBlank(os))
             sb.append(' ').append(os.replace(' ', '_'));
 
         String arch = System.getProperty("sun.arch.data.model");
-        if (!Strings.isNullOrEmpty(arch))
+        if (!Strings.isBlank(arch))
             sb.append(" arch").append(arch.replace(' ', '_'));
 
         return sb.toString();