pdfrb 0.3.0 → 0.7.1

data/lib/pdfrb/model/type/media_offset.rb

@@ -0,0 +1,106 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Media Offset Time (s13.3.6.5). Time-based media offset.
+      class MediaOffsetTime < Pdfrb::Model::Cos::Dictionary
+        def subtype; self[:S]&.to_sym; end
+        def timespan; self[:T]; end
+
+        def time_offset
+          return nil unless timespan
+
+          timespan.is_a?(Pdfrb::Model::Cos::Dictionary) ? timespan[:V] : timespan
+        end
+      end
+
+      # Media Offset Frame (s13.3.6.5). Frame-based media offset.
+      class MediaOffsetFrame < Pdfrb::Model::Cos::Dictionary
+        def subtype; self[:S]&.to_sym; end
+        def frame_number; self[:F]; end
+        def time; self[:T]; end
+
+        def has_frame?
+          !!frame_number
+        end
+      end
+
+      # Media Offset Marker (s13.3.6.5). Marker-based media offset.
+      class MediaOffsetMarker < Pdfrb::Model::Cos::Dictionary
+        def subtype; self[:S]&.to_sym; end
+        def marker_name; self[:M]; end
+      end
+
+      # Media Player Info (s13.3.5.5). Identifies which players can
+      # render the rendition.
+      class MediaPlayerInfo < Pdfrb::Model::Cos::Dictionary
+        def player_identifier; self[:PID]; end
+        def must_honor; self[:MH]; end
+        def best_effort; self[:BE]; end
+
+        def has_constraints?
+          !!must_honor || !!best_effort
+        end
+
+        def resolved_player_identifier
+          ref = player_identifier
+          return nil unless ref && document
+
+          document.object(ref)
+        end
+      end
+
+      # Media Players (s13.3.5.5). Player info lists.
+      class MediaPlayers < Pdfrb::Model::Cos::Dictionary
+        def must_use; self[:MU]; end
+        def alternate; self[:A]; end
+        def never_use; self[:NU]; end
+
+        def must_use_count
+          return 0 unless must_use
+
+          arr = must_use.is_a?(Pdfrb::Model::PdfArray) ? must_use.to_a : must_use
+          arr.is_a?(Array) ? arr.size : 0
+        end
+
+        def never_use_count
+          return 0 unless never_use
+
+          arr = never_use.is_a?(Pdfrb::Model::PdfArray) ? never_use.to_a : never_use
+          arr.is_a?(Array) ? arr.size : 0
+        end
+      end
+
+      # Media Screen Parameters (s13.3.5.4). Controls how media is
+      # displayed on the screen annotation.
+      class MediaScreenParameters < Pdfrb::Model::Cos::Dictionary
+        def must_honor; self[:MH]; end
+        def best_effort; self[:BE]; end
+        def fit; self[:FIT]; end
+
+        def fit_meet?; fit == :Meet; end
+        def fit_slice?; fit == :Slice; end
+        def fit_fill?; fit == :Fill; end
+        def fit_hidden?; fit == :Hidden; end
+        def fit_tiled?; fit == :Tiled; end
+      end
+
+      # Media Criteria (s13.3.5.5). Constraints when selecting players.
+      class MediaCriteria < Pdfrb::Model::Cos::Dictionary
+        def must_honor; self[:MH]; end
+        def best_effort; self[:BE]; end
+        def bit_rate; self[:Rate]; end
+        def bandwidth; self[:RateBS]; end
+        def cpu; self[:RateCpu]; end
+        def ram; self[:RateRam]; end
+        def language; self[:Lang]; end
+        def mime_type; self[:MimeType]; end
+
+        def has_bit_rate_constraint?
+          !!bit_rate
+        end
+      end
+    end
+  end
+end

data/lib/pdfrb/model/type/metadata.rb

@@ -9,9 +9,33 @@ module Pdfrb
         arlington_object "Metadata"
         register_type :Metadata
 
-        # Per PDF 2.0 App Note 003, /Metadata must NOT appear on
-        # Document Part Metadata (DPM) dicts. Validate callers don't
-        # try to attach one there.
+        def type; self[:Type]; end
+        def subtype; self[:Subtype]; end
+        def filter; self[:Filter]; end
+        def associated_files; self[:AF]; end
+
+        def xml?
+          subtype&.to_sym == :XML
+        end
+
+        def xmp?
+          return false unless xml? && stream
+
+          data = stream.to_s
+          data.include?("<x:xmpmeta") || data.include?("xmlns:x=")
+        end
+
+        def xmp_packet?
+          return false unless stream
+
+          stream.to_s.include?("<?xpacket")
+        end
+
+        def xmp_string
+          return nil unless stream
+
+          stream.to_s.force_encoding(Encoding::UTF_8)
+        end
       end
     end
   end

data/lib/pdfrb/model/type/misc_helpers.rb

@@ -0,0 +1,75 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Border array [horizontal_radius vertical_radius width_dash
+      # ...]  Deprecated; use BorderStyling instead (s7.7.3.3).
+      # Encapsulated as a thin struct.
+      class BorderArray < Pdfrb::Model::PdfArray
+        def horizontal_radius
+          self[0] || 0
+        end
+
+        def vertical_radius
+          self[1] || 0
+        end
+
+        def border_width
+          self[2] || 1
+        end
+
+        def horizontal_style
+          (self[3]&.to_s || :S).to_sym
+        end
+      end
+
+      # Lab Range Array [a_min a_max b_min b_max]. Used inside the
+      # Lab color space dict (s8.6.3.4).
+      class LabRangeArray < Pdfrb::Model::PdfArray
+        def a_min; self[0]; end
+        def a_max; self[1]; end
+        def b_min; self[2]; end
+        def b_max; self[3]; end
+
+        def default?
+          a_min == -100 && a_max == 100 && b_min == -100 && b_max == 100
+        end
+      end
+
+      # Dest Output Profile Ref (s7.9.6, optional indirect reference).
+      # Used by PDF PostScript/XMP reference forms.
+      class DestOutputProfileRef < Pdfrb::Model::Reference
+        def resolved
+          document&.object(self)
+        end
+      end
+
+      # Output Intents array (s14.11.5). Container for the Catalog
+      # /OutputIntents array — convenience wrapper that just forwards
+      # to the type.
+      class OutputIntentsContainer < Pdfrb::Model::Cos::Dictionary
+        def intent_type; self[:S]; end
+
+        def pdfx?
+          intent_type&.to_sym == :GTS_PDFX
+        end
+
+        def pdfa?
+          intent_type&.to_sym == :GTS_PDFA
+        end
+      end
+
+      # CMapStream (s9.7.5.3). CMap stream definition embedded in a PDF.
+      class CMapStream < Pdfrb::Model::Cos::Stream
+        def type; self[:Type]; end
+        def base_font; self[:BaseFont]; end
+        def encoding; self[:Encoding]; end
+
+        def cmap_data
+          decoded_stream&.force_encoding(Encoding::BINARY)
+        end
+      end
+    end
+  end
+end

data/lib/pdfrb/model/type/movie.rb

@@ -0,0 +1,28 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Movie dictionary (s13.4). Deprecated since PDF 2.0 but still
+      # appears in legacy PDFs. Use Screen + Rendition actions instead.
+      class Movie < Pdfrb::Model::Cos::Dictionary
+        def type; self[:Type]; end
+        def file_spec; self[:F]; end
+        def aspect; self[:Aspect]; end
+        def rotate; self[:Rotate] || 0; end
+        def poster; self[:Poster]; end
+
+        def has_poster?
+          !!poster
+        end
+
+        def resolved_file_spec
+          ref = file_spec
+          return nil unless ref && document
+
+          document.object(ref)
+        end
+      end
+    end
+  end
+end

data/lib/pdfrb/model/type/names.rb

@@ -8,6 +8,57 @@ module Pdfrb
       # /EmbeddedFiles, /AlternatePresentations, /Renditions.
       class Names < Pdfrb::Model::Cos::Dictionary
         arlington_object "Name"
+        register_type :Names
+
+        def destinations; self[:Dests]; end
+        def ap; self[:AP]; end
+        def javascript; self[:JavaScript]; end
+        def pages; self[:Pages]; end
+        def templates; self[:Templates]; end
+        def urls; self[:URLS]; end
+        def embedded_files; self[:EmbeddedFiles]; end
+        def alternate_presentations; self[:AlternatePresentations]; end
+        def renditions; self[:Renditions]; end
+
+        def has_embedded_files?
+          !!embedded_files
+        end
+
+        def has_javascript?
+          !!javascript
+        end
+
+        def has_destinations?
+          !!destinations
+        end
+
+        def each_embedded_file(&block)
+          return enum_for(:each_embedded_file) unless block
+          return unless embedded_files && document
+
+          walk_name_tree(embedded_files, &block)
+        end
+
+        private
+
+        def walk_name_tree(node_ref, &block)
+          node = if node_ref.is_a?(Pdfrb::Model::Reference) && document
+                   document.object(node_ref)
+                 else
+                   node_ref
+                 end
+          return unless node
+
+          if node[:Names]
+            arr = node[:Names]
+            arr = arr.to_a if arr.is_a?(Pdfrb::Model::PdfArray)
+            arr.each_slice(2, &block)
+          elsif node[:Kids]
+            arr = node[:Kids]
+            arr = arr.to_a if arr.is_a?(Pdfrb::Model::PdfArray)
+            arr.each { |kid| walk_name_tree(kid, &block) }
+          end
+        end
       end
     end
   end

data/lib/pdfrb/model/type/namespace.rb

@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Namespace dictionary (s14.7.6.1, PDF 2.0). Namespaces for
+      # structure element types in tagged PDF documents.
+      class Namespace < Cos::Dictionary
+        register_type :Namespace
+
+        def type; self[:Type]; end
+        def namespace_name; self[:NS]; end
+        def role_map_ns; self[:RoleMapNS]; end
+
+        def has_role_map?
+          !!role_map_ns
+        end
+
+        def mapped_role(custom_name)
+          return nil unless role_map_ns.is_a?(Hash)
+
+          role_map_ns[custom_name.to_sym] || role_map_ns[custom_name.to_s]
+        end
+      end
+    end
+  end
+end

data/lib/pdfrb/model/type/object_reference.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Object reference dict (s14.7.6). Used in structure trees to
+      # refer to a content item without including it directly.
+      class ObjectReference < Cos::Dictionary
+        register_type :ObjRef
+
+        def type; self[:Type]; end
+        def page; self[:Pg]; end
+        def object; self[:Obj]; end
+
+        def resolved_page
+          ref = page
+          return nil unless ref && document
+
+          document.object(ref)
+        end
+
+        def resolved_object
+          ref = object
+          return nil unless ref && document
+
+          document.object(ref)
+        end
+      end
+    end
+  end
+end

data/lib/pdfrb/model/type/object_stream.rb

@@ -8,6 +8,44 @@ module Pdfrb
       class ObjectStream < Pdfrb::Model::Cos::Stream
         arlington_object "ObjectStream"
         register_type :ObjStm
+
+        def type; self[:Type]; end
+        def number_of_objects; self[:N]; end
+        def first_byte_offset; self[:First]; end
+        def extends; self[:Extends]; end
+        def filter; self[:Filter]; end
+
+        def has_parent?
+          !!extends
+        end
+
+        def compressed_count
+          number_of_objects || 0
+        end
+
+        def each_object_reference
+          return enum_for(:each_object_reference) unless block_given?
+          return unless number_of_objects && first_byte_offset
+
+          stream_data = decoded_stream
+          return unless stream_data
+
+          offset = 0
+          number_of_objects.times do
+            space_idx = stream_data.index(" ", offset)
+            return if space_idx.nil?
+
+            oid = stream_data[offset...space_idx].to_i
+
+            newline_idx = stream_data.index(/\s/, space_idx + 1)
+            return if newline_idx.nil?
+
+            byte_offset = stream_data[(space_idx + 1)...newline_idx].to_i
+
+            yield oid, byte_offset + first_byte_offset
+            offset = newline_idx + 1
+          end
+        end
       end
     end
   end

data/lib/pdfrb/model/type/opt_content_ext.rb

@@ -0,0 +1,94 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Optional Content Usage Application dict (s8.11.4.1). Decides
+      # which OCGs to enable when the user views/prints/exports the
+      # document.
+      class OptContentUsageApplication < Pdfrb::Model::Cos::Dictionary
+        def event; self[:Event]&.to_sym; end
+        def ocgs; self[:OCGs]; end
+        def category; self[:Category]; end
+
+        def view_event?; event == :View; end
+        def print_event?; event == :Print; end
+        def export_event?; event == :Export; end
+
+        def each_ocg
+          return enum_for(:each_ocg) unless block_given?
+          return unless ocgs && document
+
+          arr = ocgs.is_a?(Pdfrb::Model::PdfArray) ? ocgs.to_a : ocgs
+          arr.each do |ref|
+            obj = ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+            yield obj if obj
+          end
+        end
+      end
+
+      # Optional Content Creator Info dict (s8.11.2.2). Attribution
+      # for who created a layer.
+      class OptContentCreatorInfo < Pdfrb::Model::Cos::Dictionary
+        def creator; self[:Creator]; end
+        def subtitle; self[:Subtitle]; end
+        def creator_app_version; self[:CreatorAppVersion]; end
+
+        def has_creator?
+          !!creator
+        end
+      end
+
+      # Optional Content Export dict (s8.11.2.3). Specific OCG toggle
+      # for the export workflow.
+      class OptContentExport < Pdfrb::Model::Cos::Dictionary
+        def export_ocgs; self[:ExportOCGs]; end
+        def intent; self[:Intent]&.to_sym; end
+
+        def visible_only?
+          intent == :Visible
+        end
+
+        def all_export?
+          intent == :All
+        end
+      end
+
+      # Optional Content Print dict (s8.11.2.4). Similar to Export but
+      # for the printing workflow.
+      class OptContentPrint < Pdfrb::Model::Cos::Dictionary
+        def print_ocgs; self[:PrintOCGs]; end
+        def print_state; self[:PrintState]&.to_sym; end
+
+        def on_printed?; print_state == :On; end
+        def off_printed?; print_state == :Off; end
+        def leave_unchanged?; print_state == :Unchanged; end
+      end
+
+      # Optional Content View dict (s8.11.2.5). Per-view OCG visibility.
+      class OptContentView < Pdfrb::Model::Cos::Dictionary
+        def view_name; self[:Name]; end
+        def view_state_name; self[:ViewState]; end
+      end
+
+      # Optional Content Language dict (s8.11.2.6). Maps OCG to natural
+      # language for preference-based visibility.
+      class OptContentLanguage < Pdfrb::Model::Cos::Dictionary
+        def language_string; self[:Lang]; end
+        def ocgs; self[:OCGs]; end
+        def preferred; self[:Preferred] || 0; end
+
+        def each_ocg
+          return enum_for(:each_ocg) unless block_given?
+          return unless ocgs && document
+
+          arr = ocgs.is_a?(Pdfrb::Model::PdfArray) ? ocgs.to_a : ocgs
+          arr.each do |ref|
+            obj = ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+            yield obj if obj
+          end
+        end
+      end
+    end
+  end
+end

data/lib/pdfrb/model/type/opt_content_page_element.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Opt Content Page Element (s8.11). Marks page elements for
+      # automatic optional-content group creation.
+      class OptContentPageElement < Pdfrb::Model::Cos::Dictionary
+        def subtype; self[:Subtype]&.to_sym; end
+        def ocgs; self[:OCGs]; end
+
+        def has_ocgs?
+          !!ocgs && (!ocgs.is_a?(Array) || !ocgs.empty?)
+        end
+      end
+    end
+  end
+end

data/lib/pdfrb/model/type/optional_content_config.rb

@@ -0,0 +1,56 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Optional Content Configuration (s8.11.4.1). Per-view configuration
+      # of OCG visibility, intent, and locking.
+      class OptionalContentConfiguration < Cos::Dictionary
+        register_type :OCConfig
+
+        def type; self[:Type]; end
+        def name; self[:Name]; end
+        def creator; self[:Creator]; end
+        def base_state; (self[:BaseState] || :ON).to_sym; end
+        def on; self[:ON]; end
+        def off; self[:OFF]; end
+        def intent; self[:Intent]; end
+        def list_mode; self[:ListMode]&.to_sym || :AllPages; end
+        def rb_groups; self[:RBGroups]; end
+        def locked; self[:Locked]; end
+
+        def all_on?; base_state == :ON; end
+        def all_off?; base_state == :OFF; end
+        def unchanged?; base_state == :Unchanged; end
+
+        def has_intent?
+          !!intent
+        end
+
+        def view_intent?
+          return true unless intent
+
+          arr = intent.is_a?(Pdfrb::Model::PdfArray) ? intent.to_a : intent
+          arr = [arr] unless arr.is_a?(Array)
+          arr.map(&:to_sym).include?(:View)
+        end
+
+        def print_intent?
+          return false unless intent
+
+          arr = intent.is_a?(Pdfrb::Model::PdfArray) ? intent.to_a : intent
+          arr = [arr] unless arr.is_a?(Array)
+          arr.map(&:to_sym).include?(:Print)
+        end
+
+        def locked_groups?
+          !!locked
+        end
+
+        def radio_button_groups?
+          !!rb_groups
+        end
+      end
+    end
+  end
+end

data/lib/pdfrb/model/type/optional_content_group.rb

@@ -7,18 +7,43 @@ module Pdfrb
       class OptionalContentGroup < Pdfrb::Model::Cos::Dictionary
         arlington_object "OptContentGroup"
         register_type :OCG
-      end
 
-      # Optional Content Membership (s8.11.4.2). Boolean combination
-      # of OCG visibility states.
-      class OptionalContentMembership < Pdfrb::Model::Cos::Dictionary
-        arlington_object "OptContentMembership"
-        register_type :OCMD
-      end
+        def type; self[:Type]; end
+        def name; self[:Name]; end
+        def intent; self[:Intent]; end
+        def usage; self[:Usage]; end
+
+        def view_intent?
+          return true unless intent
+
+          arr = intent.is_a?(Pdfrb::Model::PdfArray) ? intent.to_a : intent
+          arr = [arr] unless arr.is_a?(Array)
+          arr.map(&:to_sym).include?(:View)
+        end
+
+        def print_intent?
+          return false unless intent
+
+          arr = intent.is_a?(Pdfrb::Model::PdfArray) ? intent.to_a : intent
+          arr = [arr] unless arr.is_a?(Array)
+          arr.map(&:to_sym).include?(:Print)
+        end
+
+        def export_intent?
+          return false unless intent
+
+          arr = intent.is_a?(Pdfrb::Model::PdfArray) ? intent.to_a : intent
+          arr = [arr] unless arr.is_a?(Array)
+          arr.map(&:to_sym).include?(:Export)
+        end
+
+        def default_visible?
+          usage && usage[:View] && usage[:View][:ViewState]&.to_sym == :ON
+        end
 
-      # Optional Content Properties (s8.11.1). Catalog /OCProperties.
-      class OptionalContentProperties < Pdfrb::Model::Cos::Dictionary
-        arlington_object "OptContentProperties"
+        def has_usage?
+          !!usage
+        end
       end
     end
   end

data/lib/pdfrb/model/type/optional_content_membership.rb

@@ -3,10 +3,44 @@
 module Pdfrb
   module Model
     module Type
-      # Alias for OptionalContentMembership — keeps the autoload path
-      # distinct from the group class.
+      # Optional Content Membership (s8.11.4.2). Boolean combination
+      # of OCG visibility states.
       class OptionalContentMembership < Pdfrb::Model::Cos::Dictionary
         arlington_object "OptContentMembership"
+        register_type :OCMD
+
+        def ocfgs; self[:OCGs]; end
+        def policy; self[:P]; end
+        def expression; self[:VE]; end
+
+        def all_on_policy?
+          (policy || 1) == 1
+        end
+
+        def any_on_policy?
+          policy == 2
+        end
+
+        def any_off_policy?
+          policy == 3
+        end
+
+        def has_expression?
+          !!expression
+        end
+
+        def each_ocg
+          return enum_for(:each_ocg) unless block_given?
+          return unless ocfgs && document
+
+          arr = ocfgs.is_a?(Pdfrb::Model::Reference) ? document.object(ocfgs) : ocfgs
+          return unless arr.is_a?(Array) || arr.is_a?(Pdfrb::Model::PdfArray)
+
+          arr.each do |ref|
+            obj = ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+            yield obj if obj
+          end
+        end
       end
     end
   end