From e2c86a8ebaa94e7d4e262b78c765cb3a8a6918ba Mon Sep 17 00:00:00 2001
From: Michael Peter Christen <mc@yacy.net>
Date: Sun, 29 Oct 2023 12:49:08 +0100
Subject: [PATCH] added a ZIM cluster pointer cache

---
 source/org/openzim/ZIMFile.java   | 19 +++++++++++++++----
 source/org/openzim/ZIMReader.java |  5 +----
 2 files changed, 16 insertions(+), 8 deletions(-)

diff --git a/source/org/openzim/ZIMFile.java b/source/org/openzim/ZIMFile.java
index dd209b5e9..2dcfb2208 100644
--- a/source/org/openzim/ZIMFile.java
+++ b/source/org/openzim/ZIMFile.java
@@ -33,8 +33,9 @@ import java.util.List;
  *         Proof-Reading, unclustering, refactoring,
  *         naming adoption to https://wiki.openzim.org/wiki/ZIM_file_format,
  *         change of Exception handling, 
- *         extension to more attributes as defined in spec (bugfix for mime type loading)
- *         int/long bugfix (did reading of long values with int variables, causing negative offsets)
+ *         extension to more attributes as defined in spec (bugfix for mime type loading),
+ *         int/long bugfix (did reading of long values with int variables, causing negative offsets),
+ *         added url pointer, title pointer and cluster pointer caches
  */
 public class ZIMFile extends File {
 
@@ -45,10 +46,10 @@ public class ZIMFile extends File {
     public final int  header_majorVersion;
     public final int  header_minorVersion;
     public final int  header_entryCount;
-    public final int  header_clusterCount;
+    private final int  header_clusterCount;
     public final long header_urlPtrPos;
     public final long header_titlePtrPos;
-    public final long header_clusterPtrPos;
+    private final long header_clusterPtrPos;
     public final long header_mimeListPos;
     public final int  header_mainPage;
     public final int  header_layoutPage;
@@ -61,6 +62,7 @@ public class ZIMFile extends File {
     private final String[] mimeTypeList;
     private final byte[]   urlPtrListBlob;
     private final byte[]   titlePtrListBlob;
+    private final byte[]   clusterPtrListBlob;
 
     public ZIMFile(final String path) throws IOException {
         super(path);
@@ -120,6 +122,11 @@ public class ZIMFile extends File {
         this.titlePtrListBlob = new byte[this.header_entryCount * 4];
         mReader.seek(this.header_titlePtrPos);
         RandomAccessFileZIMInputStream.readFully(mReader, this.titlePtrListBlob);
+
+        // Initialize the Cluster Pointer List
+        this.clusterPtrListBlob = new byte[this.header_clusterCount * 8];
+        mReader.seek(this.header_clusterPtrPos);
+        RandomAccessFileZIMInputStream.readFully(mReader, this.clusterPtrListBlob);
     }
 
     public final String getMimeType(int idx) {
@@ -133,4 +140,8 @@ public class ZIMFile extends File {
     public final int getTitlePtr(final int idx) {
         return RandomAccessFileZIMInputStream.toFourLittleEndianInteger(this.titlePtrListBlob, idx * 4);
     }
+
+    public final long geClusterPtr(final int idx) {
+        return RandomAccessFileZIMInputStream.toEightLittleEndianLong(this.clusterPtrListBlob, idx * 8);
+    }
 }
diff --git a/source/org/openzim/ZIMReader.java b/source/org/openzim/ZIMReader.java
index 8d773d473..427b53072 100644
--- a/source/org/openzim/ZIMReader.java
+++ b/source/org/openzim/ZIMReader.java
@@ -215,11 +215,8 @@ public class ZIMReader {
         // This is now an article, so thus we can cast to ArticleEntry
         final ArticleEntry article = (ArticleEntry) directoryInfo;
 
-        // Move to the cluster entry in the clusterPtrPos
-        this.mFile.mReader.seek(this.mFile.header_clusterPtrPos + article.cluster_number * 8L);
-
         // Read the location of the cluster
-        final long clusterPos = this.mFile.mReader.readEightLittleEndianBytesLong();
+        final long clusterPos = this.mFile.geClusterPtr(article.cluster_number);
 
         // Move to the cluster
         this.mFile.mReader.seek(clusterPos);