always try to guess the size of a StringBuilder to prevent too many memory re-allocations

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@7572 6c8d7289-2bf4-0310-a012-ef5d649a1542

htroot/api/util/getpageinfo_p.java

@@ -85,10 +85,11 @@ public class getpageinfo_p {
                     prop.putXML("lang", (languages == null) ? "unknown" : languages.iterator().next());
                     
                     // get links and put them into a semicolon-separated list
-                    StringBuilder links = new StringBuilder();
-                    StringBuilder filter = new StringBuilder();
+                    Set<MultiProtocolURI> uris = scraper.getAnchors().keySet();
+                    StringBuilder links = new StringBuilder(uris.size() * 80);
+                    StringBuilder filter = new StringBuilder(uris.size() * 40);
                     count = 0;
-                    for (MultiProtocolURI uri: scraper.getAnchors().keySet()) {
+                    for (MultiProtocolURI uri: uris) {
                         links.append(';').append(uri.toNormalform(true, false));
                         filter.append('|').append(uri.getProtocol()).append("://").append(uri.getHost()).append(".*");
                         prop.putXML("links_" + count + "_link", uri.toNormalform(true, false));

htroot/yacysearchitem.java

@@ -160,9 +160,9 @@ public class yacysearchitem {
             prop.putHTML("content_creator", result.creator());// author
             prop.putHTML("content_subject", result.subject());
             final Set<String>[] query = theQuery.queryWords();
-            final StringBuilder s = new StringBuilder();
+            final StringBuilder s = new StringBuilder(query[0].size() * 20);
             for (final String t: query[0]) {
-                s.append("+").append(t);
+                s.append('+').append(t);
             }
             final String words = (s.length() > 0) ? s.substring(1) : "";
             prop.putHTML("content_words", words);

source/de/anomic/data/ListManager.java

@@ -123,13 +123,13 @@ public class ListManager {
      * @return String with elements from set separated by comma.
      */
     public static String collection2string(final Collection<String> col){
-        final StringBuilder str = new StringBuilder();
+        final StringBuilder str = new StringBuilder(col.size() * 40);
         
         if (col != null && !col.isEmpty()) {
             final Iterator<String> it = col.iterator();
             str.append(it.next());
             while(it.hasNext()) {
-                str.append(",").append(it.next());
+                str.append(',').append(it.next());
             }
         }
         

source/de/anomic/data/URLLicense.java

@@ -56,7 +56,7 @@ public class URLLicense {
     
     public String aquireLicense(final DigestURI url) {
         // generate license key
-        StringBuilder stringBuilder = new StringBuilder();
+        StringBuilder stringBuilder = new StringBuilder(keylen * 2);
         if (url == null) return stringBuilder.toString();
         while (stringBuilder.length() < keylen) stringBuilder.append(Integer.toHexString(random.nextInt()));
         String license = stringBuilder.substring(0, keylen);

source/de/anomic/search/QueryParams.java

@@ -472,7 +472,7 @@ public final class QueryParams {
             final String ext, final int page, final QueryParams theQuery,
             String newQueryString, final String originalUrlMask, final String nav) {
 
-        final StringBuilder sb = new StringBuilder();
+        final StringBuilder sb = new StringBuilder(120);
         sb.append("/yacysearch.");
         sb.append(ext);
         sb.append("?query=");

source/de/anomic/search/RankingProfile.java

@@ -27,7 +27,6 @@
 package de.anomic.search;
 
 import java.util.HashMap;
-import java.util.Iterator;
 import java.util.Map;
 
 import net.yacy.kelondro.logging.Log;
@@ -241,15 +240,13 @@ public class RankingProfile {
     }
     
     public String toExternalURLGet(final String prefix) {
-        final Iterator<Map.Entry<String, String>> i = toExternalMap("").entrySet().iterator();
-        Map.Entry<String, String> entry;
-        final StringBuilder ext = new StringBuilder();
-        while (i.hasNext()) {
-            entry = i.next();
-            ext.append("&");
+    	final Map<String, String> emap = toExternalMap("");
+        final StringBuilder ext = new StringBuilder(emap.size() * 40);
+        for (Map.Entry<String, String> entry: emap.entrySet()) {
+            ext.append('&');
             ext.append(prefix);
             ext.append(entry.getKey());
-            ext.append("=");
+            ext.append('=');
             ext.append(entry.getValue());
         }
         return ext.toString();

source/de/anomic/search/TextSnippet.java

@@ -350,8 +350,8 @@ public class TextSnippet implements Comparable<TextSnippet>, Comparator<TextSnip
         //especially p. 123 and p. 390/391 (in the German version of the 2nd edition)
 
         StringBuilder theWord = new StringBuilder(word);
-        StringBuilder prefix = new StringBuilder();
-        StringBuilder postfix = new StringBuilder();
+        StringBuilder prefix = new StringBuilder(40);
+        StringBuilder postfix = new StringBuilder(40);
         int len = 0;
 
         // cut off prefix if it contains of non-characters or non-numbers
@@ -372,7 +372,7 @@ public class TextSnippet implements Comparable<TextSnippet>, Comparator<TextSnip
             StringBuilder out = null;
             String temp = "";
             for(int k=0; k < theWord.length(); k++) {
-                out = new StringBuilder();
+                out = new StringBuilder(80);
                 //is character a special character?
                 if(p4.matcher(theWord.substring(k,k+1)).find()) {
                     if (UTF8.String(Word.word2hash(temp)).equals(UTF8.String(h))) temp = "<b>" + CharacterCoding.unicode2html(temp, false) + "</b>";

source/de/anomic/server/serverObjects.java

@@ -373,7 +373,7 @@ public class serverObjects extends HashMap<String, String> implements Cloneable
      */
     public String toString() {
         if (this.size() == 0) return "";
-        StringBuilder param = new StringBuilder();
+        StringBuilder param = new StringBuilder(this.size() * 40);
         for (Map.Entry<String, String> entry: this.entrySet()) {
             param.append(MultiProtocolURI.escape(entry.getKey()));
             param.append('=');

source/de/anomic/yacy/yacyPeerActions.java

@@ -261,7 +261,7 @@ public class yacyPeerActions {
         try {
             final long mins = millis / 60000;
             
-            final StringBuilder uptime = new StringBuilder();
+            final StringBuilder uptime = new StringBuilder(40);
             
             final int uptimeDays  = (int) (Math.floor(mins/1440.0));
             final int uptimeHours = (int) (Math.floor(mins/60.0)%24);
@@ -271,7 +271,7 @@ public class yacyPeerActions {
                   .append(((uptimeDays == 1)?" day ":" days "))
                   .append((uptimeHours < 10)?"0":"")
                   .append(uptimeHours)
-                  .append(":")
+                  .append(':')
                   .append((uptimeMins < 10)?"0":"")
                   .append(uptimeMins);            
             

source/de/anomic/yacy/yacySearch.java

@@ -131,7 +131,7 @@ public class yacySearch extends Thread {
     }
     
     public static String set2string(final HandleSet hashes) {
-        StringBuilder wh = new StringBuilder();
+        StringBuilder wh = new StringBuilder(hashes.size() * 12);
         final Iterator<byte[]> iter = hashes.iterator();
         while (iter.hasNext()) { wh.append(UTF8.String(iter.next())); }
         return wh.toString();

source/net/yacy/YaCySearchClient.java

@@ -70,7 +70,7 @@ public class YaCySearchClient {
         public SearchResult() throws IOException {
             URL url;
             Document doc;
-            String u = new StringBuilder().append("http://")
+            String u = new StringBuilder(120).append("http://")
                     .append(host)
                     .append(":")
                     .append(port)
@@ -89,7 +89,7 @@ public class YaCySearchClient {
         
         @Override
         public String toString() {
-            StringBuilder sb = new StringBuilder();
+            StringBuilder sb = new StringBuilder(this.size() * 80 + 1);
             for (RSSEntry entry: this) sb.append(entry.toString());
             return sb.toString();
         }
@@ -111,7 +111,7 @@ public class YaCySearchClient {
 
         @Override
         public String toString() {
-            return new StringBuilder().append("Title      : ")
+            return new StringBuilder(80).append("Title      : ")
                     .append(title)
                     .append("\nLink       : ")
                     .append(link)

source/net/yacy/cora/document/JSONArray.java

@@ -284,7 +284,7 @@ public class JSONArray {
      */
     private String join(String separator) throws JSONException {
         int len = length();
-        StringBuilder sb = new StringBuilder();
+        StringBuilder sb = new StringBuilder(len * 20);
 
         for (int i = 0; i < len; i += 1) {
             if (i > 0) {

source/net/yacy/cora/document/MultiProtocolURI.java

@@ -298,18 +298,18 @@ public class MultiProtocolURI implements Serializable, Comparable<MultiProtocolU
         // handle international domains
         if (!Punycode.isBasic(host)) try {
             final String[] domainParts = patternDot.split(host, 0);
-            StringBuilder buffer = new StringBuilder();
+            StringBuilder buffer = new StringBuilder(80);
             // encode each domain-part separately
-            for(int i=0; i<domainParts.length; i++) {
-            final String part = domainParts[i];
-            if(!Punycode.isBasic(part)) {
-                    buffer.append("xn--").append(Punycode.encode(part));
-            } else {
-                buffer.append(part);
-            }
-            if(i != domainParts.length-1) {
-                buffer.append('.');
-            }
+            for(int i = 0; i < domainParts.length; i++) {
+	            final String part = domainParts[i];
+	            if (!Punycode.isBasic(part)) {
+	                buffer.append("xn--").append(Punycode.encode(part));
+	            } else {
+	                buffer.append(part);
+	            }
+	            if (i != domainParts.length-1) {
+	                buffer.append('.');
+	            }
             }
             host = buffer.toString();
         } catch (final PunycodeException e) {}

source/net/yacy/cora/document/Punycode.java

@@ -45,16 +45,16 @@ public class Punycode {
     int n = INITIAL_N;
     int delta = 0;
     int bias = INITIAL_BIAS;
-    final StringBuilder output = new StringBuilder();
+    final StringBuilder output = new StringBuilder(input.length() + 1);
 
     // Copy all basic code points to the output
     int b = 0;
     for (int i = 0; i < input.length(); i++) {
-      final char c = input.charAt(i);
-      if (isBasic(c)) {
-	output.append(c);
-	b++;
-      }
+        final char c = input.charAt(i);
+        if (isBasic(c)) {
+            output.append(c);
+            b++;
+        }
     }
 
     // Append delimiter
@@ -133,20 +133,20 @@ public class Punycode {
     int n = INITIAL_N;
     int i = 0;
     int bias = INITIAL_BIAS;
-    final StringBuilder output = new StringBuilder();
 
     int d = input.lastIndexOf(DELIMITER);
+    final StringBuilder output = new StringBuilder(d + 1);
     if (d > 0) {
-      for (int j = 0; j < d; j++) {
-	final char c = input.charAt(j);
-	if (!isBasic(c)) {
-	  throw new PunycodeException(PunycodeException.BAD_INPUT);
-	}
-	output.append(c);
-      }
-      d++;
+        for (int j = 0; j < d; j++) {
+            final char c = input.charAt(j);
+            if (!isBasic(c)) {
+                throw new PunycodeException(PunycodeException.BAD_INPUT);
+            }
+            output.append(c);
+        }
+        d++;
     } else {
-      d = 0;
+        d = 0;
     }
 
     while (d < input.length()) {

source/net/yacy/cora/document/RSSReader.java

@@ -46,7 +46,7 @@ public class RSSReader extends DefaultHandler {
     
     private RSSReader(int maxsize) {
         theChannel = new RSSFeed(maxsize);
-        buffer = new StringBuilder();
+        buffer = new StringBuilder(300);
         item = null;
         parsingChannel = false;
         parsingImage = false;

source/net/yacy/document/Document.java

@@ -644,11 +644,11 @@ dc_rights
         
         long docTextLength = 0;
         final ByteBuffer         content       = new ByteBuffer();
-        final StringBuilder      authors       = new StringBuilder();
-        final StringBuilder      publishers    = new StringBuilder();
-        final StringBuilder      subjects      = new StringBuilder();
-        final StringBuilder      title         = new StringBuilder();
-        final StringBuilder      description   = new StringBuilder();
+        final StringBuilder      authors       = new StringBuilder(80);
+        final StringBuilder      publishers    = new StringBuilder(80);
+        final StringBuilder      subjects      = new StringBuilder(80);
+        final StringBuilder      title         = new StringBuilder(80);
+        final StringBuilder      description   = new StringBuilder(80);
         final LinkedList<String> sectionTitles = new LinkedList<String>();
 
         final Map<MultiProtocolURI, String> anchors = new HashMap<MultiProtocolURI, String>();

source/net/yacy/document/SnippetExtractor.java

@@ -165,6 +165,7 @@ public class SnippetExtractor {
                 // so cut it here at both ends at once
                 assert maxpos >= minpos;
                 final int newlen = Math.max(10, maxpos - minpos + 10);
+                assert maxLength >= newlen: "maxLength = " + maxLength + ", newlen = " + newlen;
                 final int around = (maxLength - newlen) / 2;
                 assert minpos - around < sentence.length() : "maxpos = " + maxpos + ", minpos = " + minpos + ", around = " + around + ", sentence.length() = " + sentence.length() + ", maxLength = " + maxLength + ", newlen = " + newlen; //maxpos = 435, minpos = 17, around = -124, sentence.length() = 44
                 sentence = "[..] " + sentence.substring(minpos - around, ((maxpos + around) > sentence.length()) ? sentence.length() : (maxpos + around)).trim() + " [..]";

source/net/yacy/document/content/SurrogateReader.java

@@ -58,7 +58,7 @@ public class SurrogateReader extends DefaultHandler implements Runnable {
     private final InputStream stream;
     
     public SurrogateReader(final InputStream stream, int queueSize) throws IOException {
-        this.buffer = new StringBuilder();
+        this.buffer = new StringBuilder(300);
         this.parsingValue = false;
         this.surrogate = null;
         this.elementName = null;

source/net/yacy/document/parser/docParser.java

@@ -64,12 +64,12 @@ public class docParser extends AbstractParser implements Parser {
             throw new Parser.Failure("error in docParser, WordTextExtractorFactory: " + e.getMessage(), location);
         }
 
-        final StringBuilder contents = new StringBuilder();
+        final StringBuilder contents = new StringBuilder(80);
         try {
             contents.append(extractor.getText().trim());
-            contents.append(" ");
+            contents.append(' ');
             contents.append(extractor.getHeaderText());
-            contents.append(" ");
+            contents.append(' ');
             contents.append(extractor.getFooterText());
         } catch (Exception e) {
             throw new Parser.Failure("error in docParser, getText: " + e.getMessage(), location);

source/net/yacy/document/parser/psParser.java

@@ -66,7 +66,7 @@ public class psParser extends AbstractParser implements Parser {
     private boolean testForPs2Ascii() {
         try {
             String procOutputLine = null;
-            final StringBuilder procOutput = new StringBuilder();
+            final StringBuilder procOutput = new StringBuilder(80);
             
             final Process ps2asciiProc = Runtime.getRuntime().exec(new String[]{"ps2ascii", "--version"});
             final BufferedReader stdOut = new BufferedReader(new InputStreamReader(ps2asciiProc.getInputStream()));

source/net/yacy/document/parser/sitemapParser.java

@@ -175,7 +175,7 @@ public class sitemapParser extends AbstractParser implements Parser {
         }
         @Override
         public String toString() {
-            final StringBuilder sb = new StringBuilder();
+            final StringBuilder sb = new StringBuilder(300);
             for (final URLEntry entry: this) {
                 sb.append(entry.toString());
             }

source/net/yacy/document/parser/torrentParser.java

@@ -72,7 +72,7 @@ public class torrentParser extends AbstractParser implements Parser {
         final String comment = (commento == null) ? "" : UTF8.String(commento.getString());
         //Date creation = new Date(map.get("creation date").getInteger());
         final BObject infoo = map.get("info");
-        final StringBuilder filenames = new StringBuilder();
+        final StringBuilder filenames = new StringBuilder(80);
         String title = "";
         if (infoo != null) {
             final Map<String, BObject> info = infoo.getMap();

source/net/yacy/kelondro/blob/Tables.java

@@ -459,7 +459,7 @@ public class Tables {
         
         @Override
         public String toString() {
-            StringBuilder sb = new StringBuilder();
+            StringBuilder sb = new StringBuilder(this.size() * 40);
             sb.append('{');
             for (Map.Entry<String, byte[]> entry: this.entrySet()) {
                 sb.append(entry.getKey()).append('=').append(UTF8.String(entry.getValue())).append(", ");

source/net/yacy/kelondro/index/Column.java

@@ -194,7 +194,7 @@ public final class Column {
 
     @Override
     public final String toString() {
-        final StringBuilder s = new StringBuilder();
+        final StringBuilder s = new StringBuilder(20);
         switch (celltype) {
         case celltype_undefined:
             s.append(nickname);

source/net/yacy/kelondro/index/Row.java

@@ -136,7 +136,7 @@ public final class Row {
     
     @Override
     public final String toString() {
-        final StringBuilder s = new StringBuilder();
+        final StringBuilder s = new StringBuilder(80);
         s.append(row[0].toString());
         for (int i = 1; i < row.length; i++) {
             s.append(", ");

source/net/yacy/kelondro/index/RowCollection.java

@@ -1018,7 +1018,7 @@ public class RowCollection implements Iterable<Row.Entry>, Cloneable {
     }
     
     public synchronized String toString() {
-        final StringBuilder s = new StringBuilder();
+        final StringBuilder s = new StringBuilder(80);
         final Iterator<Row.Entry> i = iterator();
         if (i.hasNext()) s.append(i.next().toString());
         while (i.hasNext()) s.append(", " + (i.next()).toString());

source/net/yacy/kelondro/logging/MiniLogFormatter.java

@@ -32,7 +32,7 @@ import java.util.logging.SimpleFormatter;
 
 public final class MiniLogFormatter extends SimpleFormatter {
 
-    private final StringBuilder buffer = new StringBuilder();
+    private final StringBuilder buffer = new StringBuilder(80);
 
     public MiniLogFormatter() {
         super();

source/net/yacy/kelondro/util/BDecoder.java

@@ -147,7 +147,7 @@ public class BDecoder {
         }
         @Override
         public String toString() {
-            StringBuilder s = new StringBuilder();
+            StringBuilder s = new StringBuilder(l.size() * 40 + 1);
             s.append("[");
             for (BObject o: l) s.append(o.toString()).append(",");
             s.setLength(s.length() - 1);
@@ -176,11 +176,11 @@ public class BDecoder {
         }
         @Override
         public String toString() {
-            StringBuilder s = new StringBuilder();
-            s.append("{");
-            for (Map.Entry<String, BObject> e: m.entrySet()) s.append(e.getKey()).append(":").append(e.getValue().toString()).append(","); 
+            StringBuilder s = new StringBuilder(m.size() * 40 + 1);
+            s.append('{');
+            for (Map.Entry<String, BObject> e: m.entrySet()) s.append(e.getKey()).append(':').append(e.getValue().toString()).append(','); 
             s.setLength(s.length() - 1);
-            s.append("}");
+            s.append('}');
             return s.toString();
         }
         public void toStream(OutputStream os) throws IOException {

source/net/yacy/kelondro/util/FileUtils.java

@@ -602,7 +602,7 @@ public final class FileUtils {
      * @return String representation of the file content.
      */
     public static String getListString(final File listFile, final boolean withcomments){
-        final StringBuilder temp = new StringBuilder();
+        final StringBuilder temp = new StringBuilder(300);
         
         BufferedReader br = null;        
         try{
@@ -726,7 +726,7 @@ public final class FileUtils {
      * @return returns <code>true</code> if successful, <code>false</code> otherwise
      */
     public static boolean writeList(final File listFile, final String[] list){
-        final StringBuilder out = new StringBuilder();
+        final StringBuilder out = new StringBuilder(list.length * 40 + 1);
         for(int i=0;i < list.length; i++){
             out.append(list[i]).append(CR).append(LF);
         }

source/net/yacy/kelondro/util/MapTools.java

@@ -109,8 +109,8 @@ public final class MapTools {
     }
     
     public static String set2string(final Set<String> s, final String separator, final boolean braces) {
-        final StringBuilder buf = new StringBuilder();
-        if (braces) buf.append("{");
+        final StringBuilder buf = new StringBuilder(s.size() * 40 + 1);
+        if (braces) buf.append('{');
         final Iterator<String> i = s.iterator();
         boolean hasNext = i.hasNext();
         while (hasNext) {
@@ -118,7 +118,7 @@ public final class MapTools {
             hasNext = i.hasNext();
             if (hasNext) buf.append(separator);
         }
-        if (braces) buf.append("}");
+        if (braces) buf.append('}');
         return new String(buf);
     }
 

source/net/yacy/kelondro/workflow/WorkflowProcessor.java

@@ -218,7 +218,7 @@ public class WorkflowProcessor<J extends WorkflowJob> {
     }
     
     public String getChilds() {
-        StringBuilder s = new StringBuilder();
+        StringBuilder s = new StringBuilder(this.childs.length * 40 + 1);
         for (int i = 0; i < this.childs.length; i++) {
             s.append(this.childs[i]);
             s.append(' ');