parser.rb in combine_pdf-0.2.34

- old
+ new

@@ -118,13 +118,10 @@
             @parsed << stream_data.shift
           end
         end
       end
 
-      # Strings were unified, we can let them go..
-      @strings_dictionary.clear
-
       # serialize_objects_and_references.catalog_pages
 
       # Benchmark.bm do |bm|
       # 	bm.report("serialize") {1000.times {serialize_objects_and_references} }
       # 	bm.report("serialize - old") {1000.times {old_serialize_objects_and_references} }
@@ -133,10 +130,13 @@
 
       serialize_objects_and_references
 
       catalog_pages
 
+      # Strings were unified, we can let them go..
+      @strings_dictionary.clear
+
       # collect any missing objects from the forms_data
       unless @forms_object.nil? || @forms_object.empty?
         @forms_object[:related_objects] = (@parsed.select { |o| o[:FT] }).map! { |o| { is_reference_only: true, referenced_object: o } }
         @forms_object[:related_objects].delete @forms_object
       end
@@ -220,10 +220,12 @@
             out << out.pop.merge(indirect_generation_number: out.pop, indirect_reference_id: out.pop)
           else
             out << { indirect_without_dictionary: out.pop, indirect_generation_number: out.pop, indirect_reference_id: out.pop }
           end
           fresh = true
+          # fix wkhtmltopdf use of PDF 1.1 Dest using symbols instead of strings
+          out.last[:Dest] = unify_string(out.last[:Dest].to_s) if out.last[:Dest] && out.last[:Dest].is_a?(Symbol)
         # puts "!!!!!!!!! Error with :indirect_reference_id\n\nObject #{out.last}  :indirect_reference_id = #{out.last[:indirect_reference_id]}" unless out.last[:indirect_reference_id].is_a?(Integer)
         ##########################################
         ## parse a Hex String
         ##########################################
         elsif str = @scanner.scan(/<[0-9a-fA-F]*>/)
@@ -392,10 +394,12 @@
             if out.last.is_a? Hash
               out << out.pop.merge(indirect_generation_number: out.pop, indirect_reference_id: out.pop)
             else
               out << { indirect_without_dictionary: out.pop, indirect_generation_number: out.pop, indirect_reference_id: out.pop }
             end
+            # fix wkhtmltopdf use of PDF 1.1 Dest using symbols instead of strings
+            out.last[:Dest] = unify_string(out.last[:Dest].to_s) if out.last[:Dest] && out.last[:Dest].is_a?(Symbol)
             warn "'endobj' keyword was missing for Object ID: #{out.last[:indirect_reference_id]}, trying to auto-fix issue, but might fail."
 
             out << keep.pop
             out << keep.pop
           end
@@ -487,9 +491,14 @@
             catalog_pages(catalogs[:Kids], inheritance_hash.dup) unless catalogs[:Kids].nil?
           when :Catalog
             @forms_object.update((catalogs[:AcroForm][:referenced_object] || catalogs[:AcroForm]), &self.class.method(:hash_update_proc_for_new)) if catalogs[:AcroForm]
             @names_object.update((catalogs[:Names][:referenced_object] || catalogs[:Names]), &self.class.method(:hash_update_proc_for_new)) if catalogs[:Names]
             @outlines_object.update((catalogs[:Outlines][:referenced_object] || catalogs[:Outlines]), &self.class.method(:hash_update_proc_for_new)) if catalogs[:Outlines]
+            if catalogs[:Dests] # convert PDF 1.1 Dests to PDF 1.2+ Dests
+              dests_arry = (@names_object[:Dests] ||= {})
+              dests_arry = ((dests_arry[:referenced_object] || dests_arry)[:Names] ||= [])
+              ((catalogs[:Dests][:referenced_object] || catalogs[:Dests])[:referenced_object] || (catalogs[:Dests][:referenced_object] || catalogs[:Dests])).each {|k,v| next if CombinePDF::PDF::PRIVATE_HASH_KEYS.include?(k); dests_arry << unify_string(k.to_s); dests_arry << v; }
+            end
             catalog_pages(catalogs[:Pages], inheritance_hash.dup) unless catalogs[:Pages].nil?
           end
         end
       end
       self