Add a check kind to check clusters' content.

openzim · Jun 10, 2024 · a5d1c2f · a5d1c2f
1 parent 643b023
commit a5d1c2f
Show file tree

Hide file tree

Showing 3 changed files with 17 additions and 0 deletions.
diff --git a/include/zim/zim.h b/include/zim/zim.h
@@ -114,6 +114,11 @@ namespace zim
      */
     CLUSTER_PTRS,
 
+    /**
+     * Checks that offsets inside each clusters are valid.
+     */
+    CLUSTERS_OFFSETS,
+
     /**
      * Checks that mime-type values in dirents are valid.
      */

diff --git a/src/fileimpl.cpp b/src/fileimpl.cpp
@@ -625,6 +625,7 @@ class Grouping
       case IntegrityCheck::DIRENT_ORDER: return FileImpl::checkDirentOrder();
       case IntegrityCheck::TITLE_INDEX: return FileImpl::checkTitleIndex();
       case IntegrityCheck::CLUSTER_PTRS: return FileImpl::checkClusterPtrs();
+      case IntegrityCheck::CLUSTERS_OFFSETS: return FileImpl::checkClusters();
       case IntegrityCheck::DIRENT_MIMETYPES: return FileImpl::checkDirentMimeTypes();
       case IntegrityCheck::COUNT: ASSERT("shouldn't have reached here", ==, "");
     }
@@ -676,6 +677,16 @@ class Grouping
     return true;
   }
 
+  bool FileImpl::checkClusters() {
+    const cluster_index_type clusterCount = getCountClusters().v;
+    for ( cluster_index_type i = 0; i < clusterCount; ++i )
+    {
+      // For a read of each clusters (which will throw ZimFileFormatError in case of error)
+      readCluster(cluster_index_t(i));
+    }
+    return true;
+  }
+
   bool FileImpl::checkClusterPtrs() {
     const cluster_index_type clusterCount = getCountClusters().v;
     const offset_t validClusterRangeStart(80); // XXX: really???

diff --git a/src/fileimpl.h b/src/fileimpl.h
@@ -168,6 +168,7 @@ namespace zim
       bool checkDirentOrder();
       bool checkTitleIndex();
       bool checkClusterPtrs();
+      bool checkClusters();
       bool checkDirentMimeTypes();
   };