Sanitize trailer dictionary after reading malformed file

cstpdk · Jul 18, 2019 · bb4efff · bb4efff
1 parent 36dd7f1
commit bb4efff
Showing 1 changed file with 14 additions and 1 deletion.
diff --git a/pdfread.ml b/pdfread.ml
@@ -1772,14 +1772,27 @@ let read_malformed_pdf upw opw i =
             raise (Pdf.PDFError (Pdf.input_pdferror i "Malformed /Root entry"))
       in
         i.Pdfio.seek_in 0;
+        (* Fix Size entry and remove Prev, XRefStm, Filter, Index, W, Type,
+        and DecodeParms *)
+        let trailerdict' =
+          Pdf.Dictionary
+            (add "/Size" (Pdf.Integer (length objects))
+              (remove "/W"
+                (remove "/Type"
+                  (remove "/Index"
+                    (remove "/Prev"
+                      (remove "/XRefStm"
+                        (remove "/Filter"
+                          (remove "/DecodeParms" trailerdict))))))))
+        in
         let was_linearized = is_linearized i in
           Printf.eprintf "Malformed PDF reconstruction succeeded!\n";
           flush stderr;
           {Pdf.major = major;
            Pdf.minor = minor;
            Pdf.root = root;
            Pdf.objects = Pdf.objects_of_list None objects;
-           Pdf.trailerdict = Pdf.Dictionary trailerdict;
+           Pdf.trailerdict = trailerdict';
            Pdf.was_linearized = was_linearized;
            Pdf.saved_encryption = None}