- added parsing of numeric html entities for crawler

- fixed a bug in search response

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4843 6c8d7289-2bf4-0310-a012-ef5d649a1542

htroot/yacy/user/ysearch.java

@@ -107,7 +107,7 @@ public class ysearch {
         }
         if (sb.facilityDB != null) try { sb.facilityDB.update("zeitgeist", querystring, post); } catch (Exception e) {}
 
-        int itemsPerPage = Math.max((authenticated) ? 1000 : 10, post.getInt("maximumRecords", post.getInt("count", 10))); // SRU syntax with old property as alternative
+        int itemsPerPage = Math.min((authenticated) ? 1000 : 10, post.getInt("maximumRecords", post.getInt("count", 10))); // SRU syntax with old property as alternative
         int offset = post.getInt("startRecord", post.getInt("offset", 0));
         
         boolean global = (post == null) ? true : post.get("resource", "global").equals("global");

htroot/yacysearch.java

@@ -138,7 +138,7 @@ public class yacysearch {
         }
         if (sb.facilityDB != null) try { sb.facilityDB.update("zeitgeist", querystring, post); } catch (Exception e) {}
 
-        int itemsPerPage = Math.max((authenticated) ? 1000 : 10, post.getInt("maximumRecords", post.getInt("count", 10))); // SRU syntax with old property as alternative
+        int itemsPerPage = Math.min((authenticated) ? 1000 : 10, post.getInt("maximumRecords", post.getInt("count", 10))); // SRU syntax with old property as alternative
         int offset = post.getInt("startRecord", post.getInt("offset", 0));
         
         boolean global = (post == null) ? true : post.get("resource", "global").equals("global");

source/de/anomic/htmlFilter/htmlFilterAbstractScraper.java

@@ -411,9 +411,17 @@ public abstract class htmlFilterAbstractScraper implements htmlFilterScraper {
     }
 
     private static char[] transscript(char[] code) {
-        String t = (String) trans.get(new String(code)); 
+        if (code[1] == '#') {
-        if (t == null) return new char[0];
+            if (code[2] == 'x' || code[2] == 'X') {
-        return t.toCharArray();
+                return new char[] {(char) Integer.parseInt((new String(code)).substring(3, code.length - 1), 16)};
+            } else {
+                return new char[] {(char) Integer.parseInt((new String(code)).substring(2, code.length - 1))};
+            }
+        } else {
+            String t = (String) trans.get(new String(code)); 
+            if (t == null) return new char[0];
+            return t.toCharArray();
+        }
     }
 
     protected static serverCharBuffer transscriptAll(serverCharBuffer bb) {