hexapdf 0.11.9 → 0.13.0

data/lib/hexapdf/layout/text_box.rb

@@ -4,7 +4,7 @@
 # This file is part of HexaPDF.
 #
 # HexaPDF - A Versatile PDF Creation and Manipulation Library For Ruby
-# Copyright (C) 2014-2019 Thomas Leitner
+# Copyright (C) 2014-2020 Thomas Leitner
 #
 # HexaPDF is free software: you can redistribute it and/or modify it
 # under the terms of the GNU Affero General Public License version 3 as

data/lib/hexapdf/layout/text_fragment.rb

@@ -4,7 +4,7 @@
 # This file is part of HexaPDF.
 #
 # HexaPDF - A Versatile PDF Creation and Manipulation Library For Ruby
-# Copyright (C) 2014-2019 Thomas Leitner
+# Copyright (C) 2014-2020 Thomas Leitner
 #
 # HexaPDF is free software: you can redistribute it and/or modify it
 # under the terms of the GNU Affero General Public License version 3 as

data/lib/hexapdf/layout/text_layouter.rb

@@ -4,7 +4,7 @@
 # This file is part of HexaPDF.
 #
 # HexaPDF - A Versatile PDF Creation and Manipulation Library For Ruby
-# Copyright (C) 2014-2019 Thomas Leitner
+# Copyright (C) 2014-2020 Thomas Leitner
 #
 # HexaPDF is free software: you can redistribute it and/or modify it
 # under the terms of the GNU Affero General Public License version 3 as

data/lib/hexapdf/layout/text_shaper.rb

@@ -4,7 +4,7 @@
 # This file is part of HexaPDF.
 #
 # HexaPDF - A Versatile PDF Creation and Manipulation Library For Ruby
-# Copyright (C) 2014-2019 Thomas Leitner
+# Copyright (C) 2014-2020 Thomas Leitner
 #
 # HexaPDF is free software: you can redistribute it and/or modify it
 # under the terms of the GNU Affero General Public License version 3 as
@@ -68,9 +68,10 @@ module HexaPDF
           text_fragment.clear_cache
         end
         if text_fragment.style.font_features[:kern] && font.wrapped_font.features.include?(:kern)
-          if font.font_type == :TrueType
+          case font.font_type
+          when :TrueType
             process_true_type_kerning(text_fragment)
-          elsif font.font_type == :Type1
+          when :Type1
             process_type1_kerning(text_fragment)
           end
           text_fragment.clear_cache

data/lib/hexapdf/layout/width_from_polygon.rb

@@ -4,7 +4,7 @@
 # This file is part of HexaPDF.
 #
 # HexaPDF - A Versatile PDF Creation and Manipulation Library For Ruby
-# Copyright (C) 2014-2019 Thomas Leitner
+# Copyright (C) 2014-2020 Thomas Leitner
 #
 # HexaPDF is free software: you can redistribute it and/or modify it
 # under the terms of the GNU Affero General Public License version 3 as

data/lib/hexapdf/name_tree_node.rb

@@ -4,7 +4,7 @@
 # This file is part of HexaPDF.
 #
 # HexaPDF - A Versatile PDF Creation and Manipulation Library For Ruby
-# Copyright (C) 2014-2019 Thomas Leitner
+# Copyright (C) 2014-2020 Thomas Leitner
 #
 # HexaPDF is free software: you can redistribute it and/or modify it
 # under the terms of the GNU Affero General Public License version 3 as

data/lib/hexapdf/number_tree_node.rb

@@ -4,7 +4,7 @@
 # This file is part of HexaPDF.
 #
 # HexaPDF - A Versatile PDF Creation and Manipulation Library For Ruby
-# Copyright (C) 2014-2019 Thomas Leitner
+# Copyright (C) 2014-2020 Thomas Leitner
 #
 # HexaPDF is free software: you can redistribute it and/or modify it
 # under the terms of the GNU Affero General Public License version 3 as

data/lib/hexapdf/object.rb

@@ -4,7 +4,7 @@
 # This file is part of HexaPDF.
 #
 # HexaPDF - A Versatile PDF Creation and Manipulation Library For Ruby
-# Copyright (C) 2014-2019 Thomas Leitner
+# Copyright (C) 2014-2020 Thomas Leitner
 #
 # HexaPDF is free software: you can redistribute it and/or modify it
 # under the terms of the GNU Affero General Public License version 3 as
@@ -122,9 +122,6 @@ module HexaPDF
 
     include Comparable
 
-    # A list of classes whose objects cannot be duplicated.
-    NOT_DUPLICATABLE_CLASSES = [NilClass, FalseClass, TrueClass, Symbol, Integer, Float].freeze
-
     # :call-seq:
     #   HexaPDF::Object.deep_copy(object)    -> copy
     #
@@ -132,15 +129,13 @@ module HexaPDF
     def self.deep_copy(object)
       case object
       when Hash
-        object.each_with_object({}) {|(key, val), memo| memo[key] = deep_copy(val) }
+        object.transform_values {|value| deep_copy(value) }
       when Array
         object.map {|o| deep_copy(o) }
       when HexaPDF::Object
         (object.indirect? || object.must_be_indirect? ? object : deep_copy(object.value))
       when HexaPDF::Reference
         object
-      when *NOT_DUPLICATABLE_CLASSES
-        object
       else
         object.dup
       end
@@ -251,29 +246,31 @@ module HexaPDF
     end
 
     # :call-seq:
-    #   obj.validate(auto_correct: true)                               -> true or false
-    #   obj.validate(auto_correct: true) {|msg, correctable| block }   -> true or false
+    #   obj.validate(auto_correct: true)                                    -> true or false
+    #   obj.validate(auto_correct: true) {|msg, correctable, obj| block }   -> true or false
     #
-    # Validates the object and, optionally, corrects problems when the option +auto_correct+ is set.
-    # The validation routine itself has to be implemented in the #perform_validation method - see
-    # its documentation for more information.
+    # Validates the object, optionally corrects problems when the option +auto_correct+ is set and
+    # returns +true+ if the object is deemed valid and +false+ otherwise.
     #
     # If a block is given, it is called on validation problems with a problem description and
-    # whether the problem is correctable.
+    # whether the problem is automatically correctable. The third argument to the block is usually
+    # this object but may be another object if during auto-correction a new object was created and
+    # validated.
     #
-    # Returns +true+ if the object is deemed valid and +false+ otherwise.
+    # The validation routine itself has to be implemented in the #perform_validation method - see
+    # its documentation for more information.
     #
     # *Note*: Even if the return value is +true+ there may be problems since HexaPDF doesn't
     # currently implement the full PDF spec. However, if the return value is +false+, there is
     # certainly a problem!
     def validate(auto_correct: true)
-      catch do |catch_tag|
-        perform_validation do |msg, correctable|
-          yield(msg, correctable) if block_given?
-          throw(catch_tag, false) unless auto_correct && correctable
-        end
-        true
+      result = true
+      perform_validation do |msg, correctable, object|
+        yield(msg, correctable, object || self) if block_given?
+        result = false unless correctable
+        return false unless auto_correct
       end
+      result
     end
 
     # Makes a deep copy of the source PDF object and resets the object identifier.
@@ -339,17 +336,25 @@ module HexaPDF
     # are also performed!
     #
     # When the validation routine finds that the object is invalid, it has to yield a problem
-    # description and whether the problem can be corrected. After yielding, the problem has to be
-    # corrected which poses no problem because the #validate method makes sure that the yield only
-    # returns if the problem is actually correctable and if it should be corrected.
+    # description and whether the problem can be corrected. An optional third argument may contain
+    # the object that gets validated if it is different from this object (may happen when
+    # auto-correction is used).
+    #
+    # After yielding, the problem has to be corrected if it is correctable. If it is not correctable
+    # and not correcting would lead to exceptions the method has to return early.
     #
-    # Here is a sample validation routine for stream objects:
+    # Here is a sample validation routine for a dictionary object type:
     #
     #   def perform_validation
     #     super
-    #     unless value.kind_of?(Hash)
-    #       yield("A stream object needs a Hash as value")
-    #       self.value = {}
+    #
+    #     if value[:SomeKey].length != 7
+    #       yield("Length of /SomeKey is invalid")
+    #       # No need to return early here because following check doesn't rely on /SomeKey
+    #     end
+    #
+    #     if value[:OtherKey] % 2 == 0
+    #       yield("/OtherKey needs to contain an odd number of elements")
     #     end
     #   end
     def perform_validation(&block)

data/lib/hexapdf/parser.rb

@@ -4,7 +4,7 @@
 # This file is part of HexaPDF.
 #
 # HexaPDF - A Versatile PDF Creation and Manipulation Library For Ruby
-# Copyright (C) 2014-2019 Thomas Leitner
+# Copyright (C) 2014-2020 Thomas Leitner
 #
 # HexaPDF is free software: you can redistribute it and/or modify it
 # under the terms of the GNU Affero General Public License version 3 as
@@ -59,6 +59,7 @@ module HexaPDF
       @tokenizer = Tokenizer.new(io)
       @document = document
       @object_stream_data = {}
+      @reconstructed_revision = nil
       retrieve_pdf_header_offset_and_version
     end
 
@@ -86,6 +87,8 @@ module HexaPDF
       end
 
       @document.wrap(obj, oid: oid, gen: gen, stream: stream)
+    rescue HexaPDF::MalformedPDFError
+      reconstructed_revision.object(xref_entry)
     end
 
     # Parses the indirect object at the specified offset.
@@ -235,14 +238,14 @@ module HexaPDF
         @tokenizer.skip_whitespace
         start.upto(start + number_of_entries - 1) do |oid|
           pos, gen, type = @tokenizer.next_xref_entry do |matched_size|
-            maybe_raise("Invalid cross-reference subsection entry", pos: @tokenizer.pos,
-                        force: matched_size == 20)
+            maybe_raise("Invalid cross-reference entry", pos: @tokenizer.pos,
+                        force: !matched_size)
           end
           if xref.entry?(oid)
             next
           elsif type == 'n'
             if pos == 0 || gen > 65535
-              maybe_raise("Invalid in use cross-reference entry in cross-reference section",
+              maybe_raise("Invalid in use cross-reference entry",
                           pos: @tokenizer.pos)
               xref.add_free_entry(oid, gen)
             else
@@ -281,7 +284,8 @@ module HexaPDF
       @io.seek(0, IO::SEEK_END)
       step_size = 1024
       pos = @io.pos
-      eof_not_found = startxref_missing = false
+      eof_not_found = pos == 0
+      startxref_missing = false
 
       while pos != 0
         @io.pos = [pos - step_size, 0].max
@@ -312,6 +316,11 @@ module HexaPDF
       @startxref_offset = lines[eof_index - 1].to_i
     end
 
+    # Returns the reconstructed revision.
+    def reconstructed_revision
+      @reconstructed_revision ||= reconstruct_revision
+    end
+
     # Returns the PDF version number that is stored in the file header.
     #
     # See: PDF1.7 s7.5.2
@@ -333,10 +342,64 @@ module HexaPDF
     # See: PDF1.7 s7.5.2, ADB1.7 sH.3-3.4.1
     def retrieve_pdf_header_offset_and_version
       @io.seek(0)
-      @header_offset = @io.read(1024).index(/%PDF-(\d\.\d)/) || 0
+      @header_offset = (@io.read(1024) || '').index(/%PDF-(\d\.\d)/) || 0
       @header_version = $1
     end
 
+    # Tries to reconstruct the PDF document's main cross-reference table by serially parsing the
+    # file and returning a Revision object for loading the found objects.
+    #
+    # If the file contains multiple cross-reference sections, all objects will be put into a single
+    # cross-reference table, later objects overwriting prior ones.
+    def reconstruct_revision
+      raise unless @document.config['parser.try_xref_reconstruction']
+      msg = "#{$!} - trying cross-reference table reconstruction"
+      @document.config['parser.on_correctable_error'].call(@document, msg, @tokenizer.pos)
+
+      xref = XRefSection.new
+      @tokenizer.pos = 0
+      while true
+        pos = @tokenizer.pos
+        @tokenizer.scan_until(/(\n|\r\n?)+|\z/)
+        next_new_line_pos = @tokenizer.pos
+        @tokenizer.pos = pos
+
+        token = @tokenizer.next_token rescue nil
+        if token.kind_of?(Integer)
+          gen = @tokenizer.next_token rescue nil
+          tok = @tokenizer.next_token rescue nil
+          if @tokenizer.pos > next_new_line_pos
+            @tokenizer.pos = next_new_line_pos
+          elsif gen.kind_of?(Integer) && tok.kind_of?(Tokenizer::Token) && tok == 'obj'
+            xref.add_in_use_entry(token, gen, pos)
+            @tokenizer.scan_until(/(?:\n|\r\n?)endobj\b/)
+          end
+        elsif token.kind_of?(Tokenizer::Token) && token == 'trailer'
+          obj = @tokenizer.next_object rescue nil
+          # Use last trailer found in case of multiple revisions but use first trailer in case of
+          # linearized file.
+          trailer = obj if obj.kind_of?(Hash) && (obj.key?(:Prev) || trailer.nil?)
+        elsif token == Tokenizer::NO_MORE_TOKENS
+          break
+        else
+          @tokenizer.pos = next_new_line_pos
+        end
+      end
+
+      trailer&.delete(:Prev) # no need for this and may wreak havoc
+      if !trailer || trailer.empty?
+        raise_malformed("Could not reconstruct malformed PDF because trailer was not found", pos: 0)
+      end
+
+      loader = lambda do |xref_entry|
+        obj, oid, gen, stream = parse_indirect_object(xref_entry.pos)
+        @document.wrap(obj, oid: oid, gen: gen, stream: stream)
+      end
+
+      Revision.new(@document.wrap(trailer, type: :XXTrailer), xref_section: xref,
+                   loader: loader)
+    end
+
     # Raises a HexaPDF::MalformedPDFError with the given message and source position.
     def raise_malformed(msg, pos: nil)
       raise HexaPDF::MalformedPDFError.new(msg, pos: pos)

data/lib/hexapdf/pdf_array.rb

@@ -4,7 +4,7 @@
 # This file is part of HexaPDF.
 #
 # HexaPDF - A Versatile PDF Creation and Manipulation Library For Ruby
-# Copyright (C) 2014-2019 Thomas Leitner
+# Copyright (C) 2014-2020 Thomas Leitner
 #
 # HexaPDF is free software: you can redistribute it and/or modify it
 # under the terms of the GNU Affero General Public License version 3 as
@@ -83,7 +83,7 @@ module HexaPDF
     # subclasses) and the given data has not (including subclasses), the data is stored inside the
     # HexaPDF::Object.
     def []=(index, data)
-      if value[index].class == HexaPDF::Object && !data.kind_of?(HexaPDF::Object) &&
+      if value[index].instance_of?(HexaPDF::Object) && !data.kind_of?(HexaPDF::Object) &&
           !data.kind_of?(HexaPDF::Reference)
         value[index].value = data
       else
@@ -113,6 +113,13 @@ module HexaPDF
       value.delete_at(index)
     end
 
+    # Deletes all values from the PDFArray that are equal to the given object.
+    #
+    # Returns the last deleted item, or +nil+ if no matching item is found.
+    def delete(object)
+      value.delete(object)
+    end
+
     # :call-seq:
     #   array.slice!(index)             -> obj or nil
     #   array.slice!(start, length)     -> new_array or nil
@@ -196,7 +203,7 @@ module HexaPDF
         data = document.deref(data)
         value[index] = data if index
       end
-      if data.class == HexaPDF::Object || (data.kind_of?(HexaPDF::Object) && data.value.nil?)
+      if data.instance_of?(HexaPDF::Object) || (data.kind_of?(HexaPDF::Object) && data.value.nil?)
         data = data.value
       end
       data

data/lib/hexapdf/rectangle.rb

@@ -4,7 +4,7 @@
 # This file is part of HexaPDF.
 #
 # HexaPDF - A Versatile PDF Creation and Manipulation Library For Ruby
-# Copyright (C) 2014-2019 Thomas Leitner
+# Copyright (C) 2014-2020 Thomas Leitner
 #
 # HexaPDF is free software: you can redistribute it and/or modify it
 # under the terms of the GNU Affero General Public License version 3 as
@@ -59,31 +59,61 @@ module HexaPDF
       self[0]
     end
 
+    # Sets the x-coordinate of the bottom-left corner to the given value.
+    def left=(x)
+      value[0] = x
+    end
+
     # Returns the x-coordinate of the top-right corner.
     def right
       self[2]
     end
 
+    # Sets the x-coordinate of the top-right corner to the given value.
+    def right=(x)
+      value[2] = x
+    end
+
     # Returns the y-coordinate of the bottom-left corner.
     def bottom
       self[1]
     end
 
+    # Sets the y-coordinate of the bottom-left corner to the given value.
+    def bottom=(y)
+      value[1] = y
+    end
+
     # Returns the y-coordinate of the top-right corner.
     def top
       self[3]
     end
 
+    # Sets the y-coordinate of the top-right corner to the given value.
+    def top=(y)
+      value[3] = y
+    end
+
     # Returns the width of the rectangle.
     def width
       self[2] - self[0]
     end
 
+    # Sets the width of the rectangle to the given value.
+    def width=(val)
+      self[2] = self[0] + val
+    end
+
     # Returns the height of the rectangle.
     def height
       self[3] - self[1]
     end
 
+    # Sets the height of the rectangle to the given value.
+    def height=(val)
+      self[3] = self[1] + val
+    end
+
     # Compares this rectangle to +other+ like in Object#== but also allows comparison to simple
     # arrays if the rectangle is a direct object.
     def ==(other)

data/lib/hexapdf/reference.rb

@@ -4,7 +4,7 @@
 # This file is part of HexaPDF.
 #
 # HexaPDF - A Versatile PDF Creation and Manipulation Library For Ruby
-# Copyright (C) 2014-2019 Thomas Leitner
+# Copyright (C) 2014-2020 Thomas Leitner
 #
 # HexaPDF is free software: you can redistribute it and/or modify it
 # under the terms of the GNU Affero General Public License version 3 as

data/lib/hexapdf/revision.rb

@@ -4,7 +4,7 @@
 # This file is part of HexaPDF.
 #
 # HexaPDF - A Versatile PDF Creation and Manipulation Library For Ruby
-# Copyright (C) 2014-2019 Thomas Leitner
+# Copyright (C) 2014-2020 Thomas Leitner
 #
 # HexaPDF is free software: you can redistribute it and/or modify it
 # under the terms of the GNU Affero General Public License version 3 as
@@ -172,6 +172,7 @@ module HexaPDF
 
       obj = object(ref_or_oid)
       obj.data.value = nil
+      obj.document = nil
       if mark_as_free
         add_without_check(HexaPDF::Object.new(nil, oid: obj.oid, gen: obj.gen))
       else