parser.rb in hexapdf-0.14.0

- old
+ new

@@ -265,10 +265,31 @@
       trailer = @tokenizer.next_object
       unless trailer.kind_of?(Hash)
         raise_malformed("Trailer is #{trailer.class} instead of dictionary ", pos: @tokenizer.pos)
       end
 
+      unless trailer[:Prev] || xref.max_oid == 0 || xref.entry?(0)
+        first_entry = xref[xref.oids[0]]
+        test_entry = xref[xref.oids[-1]]
+        @tokenizer.pos = test_entry.pos + @header_offset
+        test_oid = @tokenizer.next_token
+        first_oid = first_entry.oid
+
+        force_failure = !first_entry.free? || first_entry.gen != 65535 ||
+          !test_oid.kind_of?(Integer) || xref.oids[-1] - test_oid != first_oid
+        maybe_raise("Main cross-reference section has invalid numbering",
+                    pos: offset + @header_offset, force: force_failure)
+
+        new_xref = XRefSection.new
+        xref.oids.each do |oid|
+          entry = xref[oid]
+          entry.oid -= first_oid
+          new_xref.send(:[]=, entry.oid, entry.gen, entry)
+        end
+        xref = new_xref
+      end
+
       [xref, trailer]
     end
 
     # Returns the offset of the main cross-reference section/stream.
     #
@@ -357,11 +378,12 @@
       @document.config['parser.on_correctable_error'].call(@document, msg, @tokenizer.pos)
 
       xref = XRefSection.new
       @tokenizer.pos = 0
       while true
+        @tokenizer.skip_whitespace
         pos = @tokenizer.pos
-        @tokenizer.scan_until(/(\n|\r\n?)+|\z/)
+        @tokenizer.scan_until(/(\n|\r\n?)+/)
         next_new_line_pos = @tokenizer.pos
         @tokenizer.pos = pos
 
         token = @tokenizer.next_token rescue nil
         if token.kind_of?(Integer)