pdf-reader 2.7.0 → 2.9.2

data/lib/pdf/reader/object_hash.rb

@@ -48,7 +48,11 @@ class PDF::Reader
       @trailer     = @xref.trailer
       @cache       = opts[:cache] || PDF::Reader::ObjectCache.new
       @sec_handler = NullSecurityHandler.new
-      @sec_handler = build_security_handler(opts)
+      @sec_handler = SecurityHandlerFactory.build(
+        deref(trailer[:Encrypt]),
+        deref(trailer[:ID]),
+        opts[:password]
+      )
     end
 
     # returns the type of object a ref points to
@@ -92,6 +96,218 @@ class PDF::Reader
     end
     alias :deref :object
 
+    # If key is a PDF::Reader::Reference object, lookup the corresponding
+    # object in the PDF and return it. Otherwise return key untouched.
+    #
+    # Guaranteed to only return an Array or nil. If the dereference results in
+    # any other type then a MalformedPDFError exception will raise. Useful when
+    # expecting an Array and no other type will do.
+    def deref_array(key)
+      obj = deref(key)
+
+      return obj if obj.nil?
+
+      obj.tap { |obj|
+        raise MalformedPDFError, "expected object to be an Array or nil" if !obj.is_a?(Array)
+      }
+    end
+
+    # If key is a PDF::Reader::Reference object, lookup the corresponding
+    # object in the PDF and return it. Otherwise return key untouched.
+    #
+    # Guaranteed to only return an Array of Numerics or nil. If the dereference results in
+    # any other type then a MalformedPDFError exception will raise. Useful when
+    # expecting an Array and no other type will do.
+    #
+    # Some effort to cast array elements to a number is made for any non-numeric elements.
+    def deref_array_of_numbers(key)
+      arr = deref(key)
+
+      return arr if arr.nil?
+
+      raise MalformedPDFError, "expected object to be an Array" unless arr.is_a?(Array)
+
+      arr.map { |item|
+        if item.is_a?(Numeric)
+          item
+        elsif item.respond_to?(:to_f)
+          item.to_f
+        elsif item.respond_to?(:to_i)
+          item.to_i
+        else
+          raise MalformedPDFError, "expected object to be a number"
+        end
+      }
+    end
+
+    # If key is a PDF::Reader::Reference object, lookup the corresponding
+    # object in the PDF and return it. Otherwise return key untouched.
+    #
+    # Guaranteed to only return a Hash or nil. If the dereference results in
+    # any other type then a MalformedPDFError exception will raise. Useful when
+    # expecting an Array and no other type will do.
+    def deref_hash(key)
+      obj = deref(key)
+
+      return obj if obj.nil?
+
+      obj.tap { |obj|
+        raise MalformedPDFError, "expected object to be a Hash or nil" if !obj.is_a?(Hash)
+      }
+    end
+
+    # If key is a PDF::Reader::Reference object, lookup the corresponding
+    # object in the PDF and return it. Otherwise return key untouched.
+    #
+    # Guaranteed to only return a PDF name (Symbol) or nil. If the dereference results in
+    # any other type then a MalformedPDFError exception will raise. Useful when
+    # expecting an Array and no other type will do.
+    #
+    # Some effort to cast to a symbol is made when the reference points to a non-symbol.
+    def deref_name(key)
+      obj = deref(key)
+
+      return obj if obj.nil?
+
+      if !obj.is_a?(Symbol)
+        if obj.respond_to?(:to_sym)
+          obj = obj.to_sym
+        else
+          raise MalformedPDFError, "expected object to be a Name"
+        end
+      end
+
+      obj
+    end
+
+    # If key is a PDF::Reader::Reference object, lookup the corresponding
+    # object in the PDF and return it. Otherwise return key untouched.
+    #
+    # Guaranteed to only return an Integer or nil. If the dereference results in
+    # any other type then a MalformedPDFError exception will raise. Useful when
+    # expecting an Array and no other type will do.
+    #
+    # Some effort to cast to an int is made when the reference points to a non-integer.
+    def deref_integer(key)
+      obj = deref(key)
+
+      return obj if obj.nil?
+
+      if !obj.is_a?(Integer)
+        if obj.respond_to?(:to_i)
+          obj = obj.to_i
+        else
+          raise MalformedPDFError, "expected object to be an Integer"
+        end
+      end
+
+      obj
+    end
+
+    # If key is a PDF::Reader::Reference object, lookup the corresponding
+    # object in the PDF and return it. Otherwise return key untouched.
+    #
+    # Guaranteed to only return a Numeric or nil. If the dereference results in
+    # any other type then a MalformedPDFError exception will raise. Useful when
+    # expecting an Array and no other type will do.
+    #
+    # Some effort to cast to a number is made when the reference points to a non-number.
+    def deref_number(key)
+      obj = deref(key)
+
+      return obj if obj.nil?
+
+      if !obj.is_a?(Numeric)
+        if obj.respond_to?(:to_f)
+          obj = obj.to_f
+        elsif obj.respond_to?(:to_i)
+          obj.to_i
+        else
+          raise MalformedPDFError, "expected object to be a number"
+        end
+      end
+
+      obj
+    end
+
+    # If key is a PDF::Reader::Reference object, lookup the corresponding
+    # object in the PDF and return it. Otherwise return key untouched.
+    #
+    # Guaranteed to only return a PDF::Reader::Stream or nil. If the dereference results in
+    # any other type then a MalformedPDFError exception will raise. Useful when
+    # expecting a stream and no other type will do.
+    def deref_stream(key)
+      obj = deref(key)
+
+      return obj if obj.nil?
+
+      obj.tap { |obj|
+        if !obj.is_a?(PDF::Reader::Stream)
+          raise MalformedPDFError, "expected object to be an Array or nil"
+        end
+      }
+    end
+
+    # If key is a PDF::Reader::Reference object, lookup the corresponding
+    # object in the PDF and return it. Otherwise return key untouched.
+    #
+    # Guaranteed to only return a String or nil. If the dereference results in
+    # any other type then a MalformedPDFError exception will raise. Useful when
+    # expecting a string and no other type will do.
+    #
+    # Some effort to cast to a string is made when the reference points to a non-string.
+    def deref_string(key)
+      obj = deref(key)
+
+      return obj if obj.nil?
+
+      if !obj.is_a?(String)
+        if obj.respond_to?(:to_s)
+          obj = obj.to_s
+        else
+          raise MalformedPDFError, "expected object to be a string"
+        end
+      end
+
+      obj
+    end
+
+    # If key is a PDF::Reader::Reference object, lookup the corresponding
+    # object in the PDF and return it. Otherwise return key untouched.
+    #
+    # Guaranteed to only return a PDF Name (symbol), Array or nil. If the dereference results in
+    # any other type then a MalformedPDFError exception will raise. Useful when
+    # expecting a Name or Array and no other type will do.
+    def deref_name_or_array(key)
+      obj = deref(key)
+
+      return obj if obj.nil?
+
+      obj.tap { |obj|
+        if !obj.is_a?(Symbol) && !obj.is_a?(Array)
+          raise MalformedPDFError, "expected object to be an Array or Name"
+        end
+      }
+    end
+
+    # If key is a PDF::Reader::Reference object, lookup the corresponding
+    # object in the PDF and return it. Otherwise return key untouched.
+    #
+    # Guaranteed to only return a PDF::Reader::Stream, Array or nil. If the dereference results in
+    # any other type then a MalformedPDFError exception will raise. Useful when
+    # expecting a stream or Array and no other type will do.
+    def deref_stream_or_array(key)
+      obj = deref(key)
+
+      return obj if obj.nil?
+
+      obj.tap { |obj|
+        if !obj.is_a?(PDF::Reader::Stream) && !obj.is_a?(Array)
+          raise MalformedPDFError, "expected object to be an Array or Stream"
+        end
+      }
+    end
+
     # Recursively dereferences the object refered to be +key+. If +key+ is not
     # a PDF::Reader::Reference, the key is returned unchanged.
     #
@@ -99,6 +315,22 @@ class PDF::Reader
       deref_internal!(key, {})
     end
 
+    def deref_array!(key)
+      deref!(key).tap { |obj|
+        if !obj.nil? && !obj.is_a?(Array)
+          raise MalformedPDFError, "expected object (#{obj.inspect}) to be an Array or nil"
+        end
+      }
+    end
+
+    def deref_hash!(key)
+      deref!(key).tap { |obj|
+        if !obj.nil? && !obj.is_a?(Hash)
+          raise MalformedPDFError, "expected object (#{obj.inspect}) to be a Hash or nil"
+        end
+      }
+    end
+
     # Access an object from the PDF. key can be an int or a PDF::Reader::Reference
     # object.
     #
@@ -233,7 +465,10 @@ class PDF::Reader
     #
     def page_references
       root  = fetch(trailer[:Root])
-      @page_references ||= get_page_objects(root[:Pages]).flatten
+      @page_references ||= begin
+                             pages_root = deref_hash(root[:Pages]) || {}
+                             get_page_objects(pages_root)
+                           end
     end
 
     def encrypted?
@@ -299,36 +534,6 @@ class PDF::Reader
       end
     end
 
-    def build_security_handler(opts = {})
-      encrypt = deref(trailer[:Encrypt])
-      if NullSecurityHandler.supports?(encrypt)
-        NullSecurityHandler.new
-      elsif StandardSecurityHandler.supports?(encrypt)
-        encmeta = !encrypt.has_key?(:EncryptMetadata) || encrypt[:EncryptMetadata].to_s == "true"
-        StandardSecurityHandler.new(
-          key_length: (encrypt[:Length] || 40).to_i,
-          revision: encrypt[:R],
-          owner_key: encrypt[:O],
-          user_key: encrypt[:U],
-          permissions: encrypt[:P].to_i,
-          encrypted_metadata: encmeta,
-          file_id: (deref(trailer[:ID]) || []).first,
-          password: opts[:password],
-          cfm: encrypt.fetch(:CF, {}).fetch(encrypt[:StmF], {}).fetch(:CFM, nil)
-        )
-      elsif StandardSecurityHandlerV5.supports?(encrypt)
-        StandardSecurityHandlerV5.new(
-            O: encrypt[:O],
-            U: encrypt[:U],
-            OE: encrypt[:OE],
-            UE: encrypt[:UE],
-            password: opts[:password]
-        )
-      else
-        UnimplementedSecurityHandler.new
-      end
-    end
-
     def decrypt(ref, obj)
       case obj
       when PDF::Reader::Stream then
@@ -362,19 +567,21 @@ class PDF::Reader
       @object_stream ||= {}
     end
 
-    # returns a nested array of object references for all pages in this object store.
+    # returns an array of object references for all pages in this object store. The ordering of
+    # the Array is significant and matches the page ordering of the document
     #
-    def get_page_objects(ref)
-      obj = deref(ref)
-
-      unless obj.kind_of?(::Hash)
-        raise MalformedPDFError, "Dereferenced page object must be a dict"
-      end
-
-      if obj[:Type] == :Page
-        ref
-      elsif obj[:Kids]
-        deref(obj[:Kids]).map { |kid| get_page_objects(kid) }
+    def get_page_objects(obj)
+      derefed_obj = deref_hash(obj)
+
+      if derefed_obj[:Type] == :Page
+        [obj]
+      elsif derefed_obj[:Kids]
+        kids = deref_array(derefed_obj[:Kids]) || []
+        kids.map { |kid|
+          get_page_objects(kid)
+        }.flatten
+      else
+        raise MalformedPDFError, "Expected Page or Pages object"
       end
     end
 

data/lib/pdf/reader/page.rb

@@ -14,7 +14,7 @@ module PDF
     # objects accessor to help walk the page dictionary in any useful way.
     #
     class Page
-      include ResourceMethods
+      extend Forwardable
 
       # lowlevel hash-like access to all objects in the underlying PDF
       attr_reader :objects
@@ -27,6 +27,15 @@ module PDF
       # operations
       attr_reader :cache
 
+      def_delegators :resources, :color_spaces
+      def_delegators :resources, :fonts
+      def_delegators :resources, :graphic_states
+      def_delegators :resources, :patterns
+      def_delegators :resources, :procedure_sets
+      def_delegators :resources, :properties
+      def_delegators :resources, :shadings
+      def_delegators :resources, :xobjects
+
       # creates a new page wrapper.
       #
       # * objects - an ObjectHash instance that wraps a PDF file
@@ -34,7 +43,7 @@ module PDF
       #
       def initialize(objects, pagenum, options = {})
         @objects, @pagenum = objects, pagenum
-        @page_object = objects.deref(objects.page_references[pagenum - 1])
+        @page_object = objects.deref_hash(objects.page_references[pagenum - 1])
         @cache       = options[:cache] || {}
 
         unless @page_object.is_a?(::Hash)
@@ -60,7 +69,7 @@ module PDF
       def attributes
         @attributes ||= {}.tap { |hash|
           page_with_ancestors.reverse.each do |obj|
-            hash.merge!(@objects.deref(obj))
+            hash.merge!(@objects.deref_hash(obj) || {})
           end
         }
         # This shouldn't be necesary, but some non compliant PDFs leave MediaBox
@@ -101,13 +110,24 @@ module PDF
       # returns the plain text content of this page encoded as UTF-8. Any
       # characters that can't be translated will be returned as a ▯
       #
-      def text
+      def text(opts = {})
         receiver = PageTextReceiver.new
         walk(receiver)
-        receiver.content
+        runs = receiver.runs(opts)
+
+        # rectangles[:MediaBox] can never be nil, but I have no easy way to tell sorbet that atm
+        mediabox = rectangles[:MediaBox] || Rectangle.new(0, 0, 0, 0)
+
+        PageLayout.new(runs, mediabox).to_s
       end
       alias :to_s :text
 
+      def runs(opts = {})
+        receiver = PageTextReceiver.new
+        walk(receiver)
+        receiver.runs(opts)
+      end
+
       # processes the raw content stream for this page in sequential order and
       # passes callbacks to the receiver objects.
       #
@@ -132,6 +152,9 @@ module PDF
       # the program in the correct order and calls out to your implementation.
       #
       def walk(*receivers)
+        receivers = receivers.map { |receiver|
+          ValidatingReceiver.new(receiver)
+        }
         callback(receivers, :page=, [self])
         content_stream(receivers, raw_content)
       end
@@ -140,10 +163,10 @@ module PDF
       # see here unless you're a PDF nerd like me.
       #
       def raw_content
-        contents = objects.deref(@page_object[:Contents])
+        contents = objects.deref_stream_or_array(@page_object[:Contents])
         [contents].flatten.compact.map { |obj|
-          objects.deref(obj)
-        }.map { |obj|
+          objects.deref_stream(obj)
+        }.compact.map { |obj|
           obj.unfiltered_data
         }.join(" ")
       end
@@ -174,17 +197,22 @@ module PDF
       # values are defaulted according to section 7.7.3.3 of the PDF Spec 1.7
       #
       def rectangles
-        mediabox = objects.deref!(attributes[:MediaBox])
-        cropbox = objects.deref!(attributes[:Cropbox]) || mediabox
-        bleedbox = objects.deref!(attributes[:BleedBox]) || cropbox
-        trimbox = objects.deref!(attributes[:TrimBox]) || cropbox
-        artbox = objects.deref!(attributes[:ArtBox]) || cropbox
-
-        mediarect = Rectangle.new(*mediabox)
-        croprect = Rectangle.new(*cropbox)
-        bleedrect = Rectangle.new(*bleedbox)
-        trimrect = Rectangle.new(*trimbox)
-        artrect = Rectangle.new(*artbox)
+        # attributes[:MediaBox] can never be nil, but I have no easy way to tell sorbet that atm
+        mediabox = objects.deref_array_of_numbers(attributes[:MediaBox]) || []
+        cropbox = objects.deref_array_of_numbers(attributes[:CropBox]) || mediabox
+        bleedbox = objects.deref_array_of_numbers(attributes[:BleedBox]) || cropbox
+        trimbox = objects.deref_array_of_numbers(attributes[:TrimBox]) || cropbox
+        artbox = objects.deref_array_of_numbers(attributes[:ArtBox]) || cropbox
+
+        begin
+          mediarect = Rectangle.from_array(mediabox)
+          croprect = Rectangle.from_array(cropbox)
+          bleedrect = Rectangle.from_array(bleedbox)
+          trimrect = Rectangle.from_array(trimbox)
+          artrect = Rectangle.from_array(artbox)
+        rescue ArgumentError => e
+          raise MalformedPDFError, e.message
+        end
 
         if rotate > 0
           mediarect.apply_rotation(rotate)
@@ -206,14 +234,14 @@ module PDF
       private
 
       def root
-       @root ||= objects.deref(@objects.trailer[:Root])
+        @root ||= objects.deref_hash(@objects.trailer[:Root]) || {}
       end
 
       # Returns the resources that accompany this page. Includes
       # resources inherited from parents.
       #
       def resources
-        @resources ||= @objects.deref(attributes[:Resources]) || {}
+        @resources ||= Resources.new(@objects, @objects.deref_hash(attributes[:Resources]) || {})
       end
 
       def content_stream(receivers, instructions)
@@ -249,7 +277,8 @@ module PDF
         if origin.nil?
           []
         else
-          obj = objects.deref(origin)
+          obj = objects.deref_hash(origin)
+          PDF::Reader::Error.validate_not_nil_as_malformed(obj, "parent")
           [ select_inheritable(obj) ] + ancestors(obj[:Parent])
         end
       end

data/lib/pdf/reader/page_layout.rb

@@ -21,10 +21,8 @@ class PDF::Reader
       # PDF::Reader::Rectangle at some point
       PDF::Reader::Error.validate_not_nil(mediabox, "mediabox")
 
-      runs = ZeroWidthRunsFilter.exclude_zero_width_runs(runs)
-      runs = OverlappingRunsFilter.exclude_redundant_runs(runs)
-      @mediabox = mediabox
-      @runs = merge_runs(runs)
+      @mediabox = process_mediabox(mediabox)
+      @runs = runs
       @mean_font_size   = mean(@runs.map(&:font_size)) || DEFAULT_FONT_SIZE
       @mean_font_size = DEFAULT_FONT_SIZE if @mean_font_size == 0
       @median_glyph_width = median(@runs.map(&:mean_character_width)) || 0
@@ -51,13 +49,11 @@ class PDF::Reader
     private
 
     def page_width
-      # TODO once @mediabox is a Rectangle, this can be just `@mediabox.width`
-      (@mediabox[2].to_f - @mediabox[0].to_f).abs
+      @mediabox.width
     end
 
     def page_height
-      # TODO once @mediabox is a Rectangle, this can be just `@mediabox.height`
-      (@mediabox[3].to_f - @mediabox[1].to_f).abs
+      @mediabox.height
     end
 
     # given an array of strings, return a new array with empty rows from the
@@ -109,30 +105,20 @@ class PDF::Reader
       end
     end
 
-    # take a collection of TextRun objects and merge any that are in close
-    # proximity
-    def merge_runs(runs)
-      runs.group_by { |char|
-        char.y.to_i
-      }.map { |y, chars|
-        group_chars_into_runs(chars.sort)
-      }.flatten.sort
+    def local_string_insert(haystack, needle, index)
+      haystack[Range.new(index, index + needle.length - 1)] = String.new(needle)
     end
 
-    def group_chars_into_runs(chars)
-      chars.each_with_object([]) do |char, runs|
-        if runs.empty?
-          runs << char
-        elsif runs.last.mergable?(char)
-          runs[-1] = runs.last + char
-        else
-          runs << char
-        end
+    def process_mediabox(mediabox)
+      if mediabox.is_a?(Array)
+        msg = "Passing the mediabox to PageLayout as an Array is deprecated," +
+          " please use a Rectangle instead"
+        $stderr.puts msg
+        PDF::Reader::Rectangle.from_array(mediabox)
+      else
+        mediabox
       end
     end
 
-    def local_string_insert(haystack, needle, index)
-      haystack[Range.new(index, index + needle.length - 1)] = String.new(needle)
-    end
   end
 end

data/lib/pdf/reader/page_state.rb

@@ -384,7 +384,7 @@ class PDF::Reader
       #
       def build_fonts(raw_fonts)
         wrapped_fonts = raw_fonts.map { |label, font|
-          [label, PDF::Reader::Font.new(@objects, @objects.deref(font))]
+          [label, PDF::Reader::Font.new(@objects, @objects.deref_hash(font) || {})]
         }
 
         ::Hash[wrapped_fonts]

data/lib/pdf/reader/page_text_receiver.rb

@@ -47,9 +47,32 @@ module PDF
         @characters = []
       end
 
+      def runs(opts = {})
+        runs = @characters
+
+        if rect = opts.fetch(:rect, @page.rectangles[:CropBox])
+          runs = BoundingRectangleRunsFilter.runs_within_rect(runs, rect)
+        end
+
+        if opts.fetch(:skip_zero_width, true)
+          runs = ZeroWidthRunsFilter.exclude_zero_width_runs(runs)
+        end
+
+        if opts.fetch(:skip_overlapping, true)
+          runs = OverlappingRunsFilter.exclude_redundant_runs(runs)
+        end
+
+        if opts.fetch(:merge, true)
+          runs = merge_runs(runs)
+        end
+
+        runs
+      end
+
+      # deprecated
       def content
-        mediabox = @page.rectangles[:MediaBox].to_a
-        PageLayout.new(@characters, mediabox).to_s
+        mediabox = @page.rectangles[:MediaBox]
+        PageLayout.new(runs, mediabox).to_s
       end
 
       #####################################################
@@ -64,8 +87,10 @@ module PDF
         params.each do |arg|
           if arg.is_a?(String)
             internal_show_text(arg)
-          else
+          elsif arg.is_a?(Numeric)
             @state.process_glyph_displacement(0, arg, false)
+          else
+            # skip it
           end
         end
       end
@@ -96,6 +121,7 @@ module PDF
       private
 
       def internal_show_text(string)
+        PDF::Reader::Error.validate_type_as_malformed(string, "string", String)
         if @state.current_font.nil?
           raise PDF::Reader::MalformedPDFError, "current font is invalid"
         end
@@ -109,7 +135,7 @@ module PDF
 
           # apply to glyph displacment for the current glyph so the next
           # glyph will appear in the correct position
-          glyph_width = @state.current_font.glyph_width(glyph_code) / 1000.0
+          glyph_width = @state.current_font.glyph_width_in_text_space(glyph_code)
           th = 1
           scaled_glyph_width = glyph_width * @state.font_size * th
           unless utf8_chars == SPACE
@@ -119,12 +145,6 @@ module PDF
         end
       end
 
-      # TODO: revist this. It rotates the co-ordinates to the right direction, but I don't
-      #       think it sets the correct x,y values. We get away with it because we don't
-      #       return the text with co-ordinates, only the full text arranged in a string.
-      #
-      #       We should provide an API for extracting the text with positioning data and spec
-      #       that. I suspect the co-ords might be wrong for rotated pages
       def apply_rotation(x, y)
         if @page.rotate == 90
           tmp = x
@@ -141,6 +161,28 @@ module PDF
         return x, y
       end
 
+      # take a collection of TextRun objects and merge any that are in close
+      # proximity
+      def merge_runs(runs)
+        runs.group_by { |char|
+          char.y.to_i
+        }.map { |y, chars|
+          group_chars_into_runs(chars.sort)
+        }.flatten.sort
+      end
+
+      def group_chars_into_runs(chars)
+        chars.each_with_object([]) do |char, runs|
+          if runs.empty?
+            runs << char
+          elsif runs.last.mergable?(char)
+            runs[-1] = runs.last + char
+          else
+            runs << char
+          end
+        end
+      end
+
     end
   end
 end