Merge branch '#1988_FixRegexTask'

sepinf-inc · Nov 23, 2023 · fd55e3f · fd55e3f
2 parents 5ddf1aa + 3b90bb2
commit fd55e3f
Show file tree

Hide file tree

Showing 2 changed files with 31 additions and 5 deletions.
diff --git a/iped-engine/src/main/java/iped/engine/hashdb/HashDBDataSource.java b/iped-engine/src/main/java/iped/engine/hashdb/HashDBDataSource.java
@@ -34,10 +34,10 @@ public class HashDBDataSource {
 
     private static final String ledFileLength = "fileLength";
     private static final String ledFileExt = "fileExt";
-    private static final String ledMd5_512 = "md5_512";
-    private static final String ledMd5_64k = "md5_64k";
+    public static final String ledMd5_512 = "md5_512";
+    public static final String ledMd5_64k = "md5_64k";
 
-    private static final String photoDna = "photoDna";
+    public static final String photoDna = "photoDna";
     private static final int photoDnaBase64Len = 192;
 
     private static final String propertySet = "set";

diff --git a/iped-engine/src/main/java/iped/engine/task/regex/RegexTask.java b/iped-engine/src/main/java/iped/engine/task/regex/RegexTask.java
@@ -40,8 +40,14 @@
 import iped.engine.config.RegexTaskConfig;
 import iped.engine.config.RegexTaskConfig.RegexEntry;
 import iped.engine.data.Item;
+import iped.engine.hashdb.HashDBDataSource;
 import iped.engine.lucene.analysis.FastASCIIFoldingFilter;
 import iped.engine.task.AbstractTask;
+import iped.engine.task.HashDBLookupTask;
+import iped.engine.task.HashTask;
+import iped.engine.task.PhotoDNALookup;
+import iped.engine.task.index.IndexItem;
+import iped.properties.ExtraProperties;
 
 public class RegexTask extends AbstractTask {
 
@@ -61,6 +67,26 @@ public class RegexTask extends AbstractTask {
 
     private static FSTConfiguration fastSerializer = FSTConfiguration.createDefaultConfiguration();
 
+    private static final Set<String> ignoredKeys = new HashSet<String>();
+    static {
+        // Ignore these keys when reading item's properties to be searched (issue #1988)
+        ignoredKeys.add(HashTask.HASH.EDONKEY.toString());
+        ignoredKeys.add(HashTask.HASH.MD5.toString());
+        ignoredKeys.add(HashTask.HASH.SHA1.toString());
+        ignoredKeys.add(HashTask.HASH.SHA256.toString());
+        ignoredKeys.add(HashTask.HASH.SHA512.toString());
+        ignoredKeys.add(IndexItem.TRACK_ID);
+        ignoredKeys.add(IndexItem.PARENT_TRACK_ID);
+        ignoredKeys.add(IndexItem.CONTAINER_TRACK_ID);
+        ignoredKeys.add(IndexItem.EVIDENCE_UUID);
+        ignoredKeys.add(ExtraProperties.GLOBAL_ID);
+        ignoredKeys.add(HashDBLookupTask.ATTRIBUTES_PREFIX + HashDBDataSource.ledMd5_512);
+        ignoredKeys.add(HashDBLookupTask.ATTRIBUTES_PREFIX + HashDBDataSource.ledMd5_64k);
+        ignoredKeys.add(HashDBLookupTask.ATTRIBUTES_PREFIX + HashDBDataSource.photoDna);
+        ignoredKeys.add(PhotoDNALookup.PHOTO_DNA_HIT_PREFIX + HashTask.HASH.MD5.name());
+        ignoredKeys.add(PhotoDNALookup.PHOTO_DNA_NEAREAST_HASH);
+    }
+
     private char[] cbuf = new char[1 << 20];
 
     private static RegexValidator regexValidator;
@@ -273,9 +299,9 @@ protected void process(IItem item) throws Exception {
     private Reader getExtraAttributeReader(IItem item) {
         StringBuilder sb = new StringBuilder();
         for (String key : item.getExtraAttributeMap().keySet().toArray(new String[0])) {
-            if (!key.startsWith(REGEX_PREFIX)) {
+            if (!key.startsWith(REGEX_PREFIX) && !ignoredKeys.contains(key)) {
                 Object val = item.getExtraAttribute(key);
-                sb.append(key).append(": ").append(val.toString());
+                sb.append(key).append(": ").append(val.toString()).append('\n');
             }
         }
         return new StringReader(sb.toString());