From 44cb8af0aa0a368c3c1186d03b85666c3638962c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Mikl=C3=B3s=20Fazekas?= <mfazekas@szemafor.com>
Date: Tue, 3 May 2022 12:43:35 +0200
Subject: [PATCH 1/2] HP Scan invalid Length workaround

---
 lib/combine_pdf/parser.rb | 18 +++++++++++++++++-
 1 file changed, 17 insertions(+), 1 deletion(-)

diff --git a/lib/combine_pdf/parser.rb b/lib/combine_pdf/parser.rb
index 2689750..1ae3116 100644
--- a/lib/combine_pdf/parser.rb
+++ b/lib/combine_pdf/parser.rb
@@ -34,6 +34,7 @@ class PDFParser
     attr_reader :info_object, :root_object, :names_object, :forms_object, :outlines_object, :metadata
 
     attr_reader :allow_optional_content, :raise_on_encrypted
+    attr_reader :relaxed
     # when creating a parser, it is important to set the data (String) we wish to parse.
     #
     # <b>the data is required and it is not possible to set the data at a later stage</b>
@@ -59,6 +60,7 @@ def initialize(string, options = {})
       @scanner = nil
       @allow_optional_content = options[:allow_optional_content]
       @raise_on_encrypted = options[:raise_on_encrypted]
+      @relaxed = options[:relaxed]
     end
 
     # parse the data in the new parser (the data already set through the initialize / new method)
@@ -363,7 +365,21 @@ def _parse_
           # advance by the publshed stream length (if any)
           old_pos = @scanner.pos
           if(out.last.is_a?(Hash) && out.last[:Length].is_a?(Integer) && out.last[:Length] > 2)
-            @scanner.pos += out.last[:Length] - 2
+            begin
+              @scanner.pos += out.last[:Length] - 2
+            rescue RangeError => error
+              raise error unless @relaxed
+              oldpos = @scanner.pos
+              skipped = @scanner.skip_until(/endstream/)
+              if skipped
+                len = skipped - 'endstream'.length
+                warn "CombinePDF parser: invalid length: #{out.last[:Length]} for object: #{out.last} should be: #{len}"
+                @scanner.pos = oldpos
+                @scanner.pos += len
+              else
+                raise ParsingError, "Parsing Error: PDF file error - a stream object with invalid length of #{out.last[:Length]} for object #{out.last} and no endstream found, to work around it"
+              end
+            end
           end
 
           # the following was dicarded because some PDF files didn't have an EOL marker as required

From c450a0559eedc35f792623503061ff29bbbc2c52 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Mikl=C3=B3s=20Fazekas?= <mfazekas@szemafor.com>
Date: Wed, 15 Nov 2023 10:25:13 +0100
Subject: [PATCH 2/2] Use check stream length to check and correct stream
 lengths if needed

---
 lib/combine_pdf/parser.rb | 43 +++++++++++++++++++++++++--------------
 1 file changed, 28 insertions(+), 15 deletions(-)

diff --git a/lib/combine_pdf/parser.rb b/lib/combine_pdf/parser.rb
index 1ae3116..99340c2 100644
--- a/lib/combine_pdf/parser.rb
+++ b/lib/combine_pdf/parser.rb
@@ -34,7 +34,7 @@ class PDFParser
     attr_reader :info_object, :root_object, :names_object, :forms_object, :outlines_object, :metadata
 
     attr_reader :allow_optional_content, :raise_on_encrypted
-    attr_reader :relaxed
+    attr_reader :check_stream_length
     # when creating a parser, it is important to set the data (String) we wish to parse.
     #
     # <b>the data is required and it is not possible to set the data at a later stage</b>
@@ -60,7 +60,7 @@ def initialize(string, options = {})
       @scanner = nil
       @allow_optional_content = options[:allow_optional_content]
       @raise_on_encrypted = options[:raise_on_encrypted]
-      @relaxed = options[:relaxed]
+      @check_stream_length = options[:check_stream_length]
     end
 
     # parse the data in the new parser (the data already set through the initialize / new method)
@@ -365,20 +365,10 @@ def _parse_
           # advance by the publshed stream length (if any)
           old_pos = @scanner.pos
           if(out.last.is_a?(Hash) && out.last[:Length].is_a?(Integer) && out.last[:Length] > 2)
-            begin
+            if @check_stream_length
+              advance_pos_with_length_check(@scanner, out.last[:Length], out.last)
+            else
               @scanner.pos += out.last[:Length] - 2
-            rescue RangeError => error
-              raise error unless @relaxed
-              oldpos = @scanner.pos
-              skipped = @scanner.skip_until(/endstream/)
-              if skipped
-                len = skipped - 'endstream'.length
-                warn "CombinePDF parser: invalid length: #{out.last[:Length]} for object: #{out.last} should be: #{len}"
-                @scanner.pos = oldpos
-                @scanner.pos += len
-              else
-                raise ParsingError, "Parsing Error: PDF file error - a stream object with invalid length of #{out.last[:Length]} for object #{out.last} and no endstream found, to work around it"
-              end
             end
           end
 
@@ -523,6 +513,29 @@ def _parse_
 
     protected
 
+    def advance_pos_with_length_check(scanner, length, obj)
+      endstream = 'endstream'
+      orig_pos = scanner.pos
+      if scanner.rest_size > length
+        scanner.pos += length
+        if scanner.check(endstream)
+          scanner.pos -= 2
+          return
+        end
+        warn "Invalid length no #{endstream} found - object: #{obj}!"
+      else
+        warn "Invalid length in stream points out of the file - object: #{obj}!"
+      end
+      scanner.pos = orig_pos
+      skipped = scanner.skip_until(/endstream/)
+      if skipped
+        correct_len = skipped - endstream.length
+        scanner.pos -= endstream.length + 2
+      else
+        raise ParsingError, "Parsing Error: PDF file error - a stream object with invalid length of #{length} for object #{out} and no #{endstream} found after, to work around it"
+      end
+    end
+
     # resets cataloging and pages
     def catalog_pages(catalogs = nil, inheritance_hash = {})
       unless catalogs