[WIP]: SystemStackError: stack level too deep when parsing specific PDF files with catalogs[:Resources]

[tagliala]

Refers to #214, but it is not the same issue

I can't replicate #214 with the same use case because I'm getting malformed PDF file? error

Unfortunately I can't share the file, but this is the error

SystemStackError: 7657 -> 37
     +->> 3812 cycles of 2 lines:
     | ~/dev/combine_pdf/lib/combine_pdf/parser.rb:737:in `merge'
     | ~/combine_pdf/lib/combine_pdf/parser.rb:737:in `block in <class:PDFParser>'
     +-<<

It is happening here:

combine_pdf/lib/combine_pdf/parser.rb

Line 547 in fe11372

(inheritance_hash[:Resources][:referenced_object] || inheritance_hash[:Resources]).update((catalogs[:Resources][:referenced_object] || catalogs[:Resources]), &HASH_UPDATE_PROC_FOR_OLD)

I've tested against some open PRs, several versions of combine_pdf, and checked existing issues, unfortunately there's no solution at the moment

I'm still investigating

Standalone reproducible test case (can't provide the file, sorry)

#!/usr/bin/env ruby

# frozen_string_literal: true

FILE = ARGV[0]

if FILE.nil?
  puts "Usage: #{File.basename($0)} path.to.pdf"
  exit 1
elsif !File.exist?(FILE)
  puts "File '#{FILE}' not found"
  exit 1
end

require 'bundler/inline'

gemfile(true) do
  source 'https://rubygems.org'

  gem 'combine_pdf'
  gem 'minitest'
end

require 'combine_pdf'
require 'minitest/autorun'

class BugTest < Minitest::Test
  def test_combine_pdf_parse
    assert CombinePDF.load(FILE)
  end
end

> puts catalogs[:Resources][:referenced_object].keys.inspect

Keys:
[:Font, :XObject, :ExtGState, :ProcSet, :indirect_generation_number, :indirect_reference_id]

Problem in key: `:XObject`

Keys (:XObject):
[:Im19, :Im22, :Im23, :Im26, :Im37, :Im47, :Im48, :Tr20, :Tr24, :Tr52]

Problem in keys:
[:Tr20, :Tr24, :Tr52]

Keys (:Tr20)
[:is_reference_only, :referenced_object]

Problem in key: :referenced_object

Keys (:referenced_object)
[:Type, :Subtype, :BBox, :Resources, :Group, :Length, :Filter, :raw_stream_content, :indirect_generation_number, :indirect_reference_id]

Problem in key:
:Resources

Keys (:Resources)
[:is_reference_only, :referenced_object]

Problem in key:
:referenced_object

Keys: (:referenced_object)
[:Font, :XObject, :ExtGState, :ProcSet, :indirect_generation_number, :indirect_reference_id]

Problem in key:
:XObject

Keys (:XObject):
[:Im19, :Im22, :Im23, :Im26, :Im37, :Im47, :Im48, :Tr20, :Tr24, :Tr52]

Same as above? Recursive Hash?

[tagliala]

Question: Is "HASH_UPDATE_PROC_FOR_OLD" a custom implementation of a reverse_deep_merge algorithm?