From d7cbf4cdd41d86ad582340556ebcfc95639dad20 Mon Sep 17 00:00:00 2001
From: orbiter <orbiter@6c8d7289-2bf4-0310-a012-ef5d649a1542>
Date: Fri, 17 Apr 2009 13:47:06 +0000
Subject: [PATCH] more performance hacks: less overhead in word hash
 computation

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@5825 6c8d7289-2bf4-0310-a012-ef5d649a1542
---
 source/de/anomic/http/httpClient.java         |  5 +--
 .../de/anomic/kelondro/order/Base64Order.java | 37 +++++++++++++++++++
 source/de/anomic/plasma/parser/Word.java      |  2 +-
 3 files changed, 40 insertions(+), 4 deletions(-)

diff --git a/source/de/anomic/http/httpClient.java b/source/de/anomic/http/httpClient.java
index c02050645..14b9c8322 100644
--- a/source/de/anomic/http/httpClient.java
+++ b/source/de/anomic/http/httpClient.java
@@ -554,9 +554,8 @@ public class httpClient {
                     Log.logWarning("HTTPC", "Proxy authentication contains invalid characters, trying anyway");
                 }
                 final String remoteProxyPwd = hostProxyConfig.getProxyPwd();
-                final String credentials = Base64Order.standardCoder.encodeString(remoteProxyUser.replace(":",
-                                                                                                                  "") +
-                        ":" + remoteProxyPwd);
+                final String credentials = Base64Order.standardCoder.encodeString(
+                                        remoteProxyUser.replace(":", "") + ":" + remoteProxyPwd);
                 method.setRequestHeader(httpRequestHeader.PROXY_AUTHORIZATION, "Basic " + credentials);
             }
         }
diff --git a/source/de/anomic/kelondro/order/Base64Order.java b/source/de/anomic/kelondro/order/Base64Order.java
index b2ff31bea..0c7fdf7b9 100644
--- a/source/de/anomic/kelondro/order/Base64Order.java
+++ b/source/de/anomic/kelondro/order/Base64Order.java
@@ -154,6 +154,15 @@ public class Base64Order extends AbstractOrder<byte[]> implements ByteOrder, Cod
         return new String(s);
     }
 
+    public final byte[] encodeLongSubstr(long c, int length) {
+        final byte[] s = new byte[length];
+        while (length > 0) {
+            s[--length] = (byte) alpha[(byte) (c & 0x3F)];
+            c >>= 6;
+        }
+        return s;
+    }
+
     public final void encodeLong(long c, final byte[] b, final int offset, int length) {
         assert offset + length <= b.length;
         while (length > 0) {
@@ -213,6 +222,34 @@ public class Base64Order extends AbstractOrder<byte[]> implements ByteOrder, Cod
         //assert lene == out.length() : "lene = " + lene + ", out.len = " + out.length();
         return new String(out);
     }
+    
+    public final byte[] encodeSubstring(final byte[] in, int sublen) {
+        if (in.length == 0) return null;
+        byte[] out = new byte[sublen];
+        int writepos = 0;
+        int pos = 0;
+        long l;
+        while (in.length - pos >= 3 && writepos < sublen) {
+            l = ((((0XffL & in[pos]) << 8) + (0XffL & in[pos + 1])) << 8) + (0XffL & in[pos + 2]);
+            pos += 3;
+            System.arraycopy(encodeLongSubstr(l, 4), 0, out, writepos, 4);
+            writepos += 4;
+        }
+        // now there may be remaining bytes
+        if (in.length % 3 != 0 && writepos < sublen) {
+            if (in.length % 3 == 2) {
+                System.arraycopy(encodeLong((((0XffL & in[pos]) << 8) + (0XffL & in[pos + 1])) << 8, 4), 0, out, writepos, 3);
+                writepos += 3;
+            } else {
+                System.arraycopy(encodeLong((((0XffL & in[pos])) << 8) << 8, 4).substring(0, 2), 0, out, writepos, 2);
+                writepos += 2;
+            }
+        }
+                                
+        if (rfc1113compliant) while (writepos % 4 > 0 && writepos < sublen) out[writepos] = '=';
+        //assert encode(in).substring(0, sublen).equals(new String(out));
+        return out;
+    }
 
     public final String decodeString(final String in, final String info) {
         try {
diff --git a/source/de/anomic/plasma/parser/Word.java b/source/de/anomic/plasma/parser/Word.java
index 66bb017c0..bbbebbf51 100644
--- a/source/de/anomic/plasma/parser/Word.java
+++ b/source/de/anomic/plasma/parser/Word.java
@@ -84,7 +84,7 @@ public class Word {
     public static final byte[] word2hash(final String word) {
         byte[] h = hashCache.get(word);
         if (h != null) return h;
-        h = Base64Order.enhancedCoder.encode(Digest.encodeMD5Raw(word.toLowerCase(Locale.ENGLISH))).substring(0, yacySeedDB.commonHashLength).getBytes();
+        h = Base64Order.enhancedCoder.encodeSubstring(Digest.encodeMD5Raw(word.toLowerCase(Locale.ENGLISH)), yacySeedDB.commonHashLength);
         hashCache.put(word, h); // prevent expensive MD5 computation and encoding
         return h;
     }