close #163: Infinite loop parsing PDF

torakiki · Sep 25, 2023 · 2af172b · 2af172b
1 parent 92ad538
commit 2af172b
Showing 1 changed file with 13 additions and 3 deletions.
diff --git a/src/main/java/org/sejda/sambox/input/XrefParser.java b/src/main/java/org/sejda/sambox/input/XrefParser.java
@@ -92,6 +92,8 @@ void onEntryFound(XrefEntry entry)
             }
         };
     }
+
+    private HashSet<Long> catalogSeenOffsets = new HashSet<>();
 
     /**
      * parse the xref using the given parser.
@@ -139,9 +141,14 @@ else if (line.contains(COSName.CATALOG.getName()))
                                 long position = parser.position();
                                 try
                                 {
-                                    // we do our best to make sure we have a catalog even in corrupted docs
-                                    LOG.debug("Parsing potential Catalog at {}", lastObjectOffset);
-                                    onPotentialCatalog();
+                                    if(catalogSeenOffsets.contains(lastObjectOffset)){
+                                        LOG.debug("Already parsed potential Catalog at {}, skipping", lastObjectOffset);
+                                    } else {
+                                        catalogSeenOffsets.add(lastObjectOffset);
+                                        // we do our best to make sure we have a catalog even in corrupted docs
+                                        LOG.debug("Parsing potential Catalog at {}", lastObjectOffset);
+                                        onPotentialCatalog();    
+                                    }
                                 }
                                 catch (IOException e)
                                 {
@@ -200,6 +207,9 @@ private void onPotentialCatalog() throws IOException
             }
             trailer.setFallbackScanStatus(xrefScanStatus.name());
         }
+
+        // cleanup
+        catalogSeenOffsets.clear();
     }
 
     /**