pdfrb 0.1.0

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

@@ -0,0 +1,50 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Root of the document object hierarchy (s7.7.2). Linked from
+      # the trailer's /Root. Fields: /Type, /Version, /Extensions,
+      # /Pages, /PageLabels, /Names, /Dests, /ViewerPreferences,
+      # /PageLayout, /PageMode, /Outlines, /Threads, /OpenAction,
+      # /AA, /URI, /AcroForm, /Metadata, /StructTreeRoot, /MarkInfo,
+      # /Lang, /SpiderInfo, /OutputIntents, /PieceInfo, /OCProperties,
+      # /Perms, /Legal, /Requirements, /Collection, /NeedsRendering,
+      # /DSS, /AF, /DPartRoot.
+      class Catalog < Pdfrb::Model::Cos::Dictionary
+        arlington_object "Catalog"
+        register_type :Catalog
+
+        # Returns the /Pages tree root, auto-resolving the reference.
+        def pages
+          self[:Pages]
+        end
+
+        # Returns the /AcroForm dict (or nil).
+        def acro_form
+          self[:AcroForm]
+        end
+
+        def outlines
+          self[:Outlines]
+        end
+
+        def names
+          self[:Names]
+        end
+
+        def metadata
+          self[:Metadata]
+        end
+
+        def struct_tree_root
+          self[:StructTreeRoot]
+        end
+
+        def open_action
+          self[:OpenAction]
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # CID-keyed font (s9.7.3). Inner font of a /Type0 composite.
+      # /Subtype /CIDFontType0 or /CIDFontType2.
+      class CIDFont < Pdfrb::Model::Cos::Dictionary
+        arlington_object "FontCIDType0"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Embedded file stream (s7.11.3). The actual payload of a
+      # /Filespec /EF entry.
+      class EmbeddedFile < Pdfrb::Model::Cos::Stream
+        arlington_object "EmbeddedFileStream"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Public-key security handler encryption dict (s7.6.5.1).
+      class EncryptionPublicKey < Pdfrb::Model::Cos::Dictionary
+        arlington_object "EncryptionPublicKey"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Standard security handler encryption dict (s7.6.3.4).
+      class EncryptionStandard < Pdfrb::Model::Cos::Dictionary
+        arlington_object "EncryptionStandard"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # File specification (s7.11.2). /FS, /F, /UF, /EF, /Desc, etc.
+      class FileSpecification < Pdfrb::Model::Cos::Dictionary
+        arlington_object "FileSpecification"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # File trailer dict (s7.5.5). Root of the cross-reference chain.
+      # Fields: /Size, /Prev, /Root, /Encrypt, /Info, /ID, /XRefStm,
+      # /AuthCode, /AdditionalStreams, /DocChecksum.
+      class FileTrailer < Pdfrb::Model::Cos::Dictionary
+        arlington_object "FileTrailer"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Base Font dict (s9.6.2). Concrete subclasses: FontType1,
+      # FontTrueType, FontType0, CIDFont, FontType3.
+      class Font < Pdfrb::Model::Cos::Dictionary
+        arlington_object "FontMap"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Font descriptor (s9.8). Carries metrics + embedded font file.
+      class FontDescriptor < Pdfrb::Model::Cos::Dictionary
+        arlington_object "FontDescriptorType1"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # TrueType font (s9.6.3). /Subtype /TrueType.
+      class FontTrueType < Pdfrb::Model::Cos::Dictionary
+        arlington_object "FontTrueType"
+      end
+    end
+  end
+end

data/lib/pdfrb/model/type/font_type0.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Type0 (composite) font (s9.7). Outer font that wraps a CIDFont.
+      class FontType0 < Pdfrb::Model::Cos::Dictionary
+        arlington_object "FontType0"
+      end
+    end
+  end
+end

data/lib/pdfrb/model/type/font_type1.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Type1 font (s9.6.2). /Subtype /Type1.
+      class FontType1 < Pdfrb::Model::Cos::Dictionary
+        arlington_object "FontType1"
+      end
+    end
+  end
+end

data/lib/pdfrb/model/type/font_type3.rb

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Type 3 glyph font (s9.6.5). /Subtype /Type3, /CharProcs,
+      # /Resources, /FontBBox, /FontMatrix, /Encoding, /Widths.
+      class FontType3 < Pdfrb::Model::Cos::Dictionary
+        arlington_object "FontType3"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Named graphics-state parameter (s8.4.3) referenced by `gs`
+      # operator. Fields: /Type, /LW, /LC, /LJ, /ML, /D, /RI, /SA,
+      # /BM, /SM, /CA, /ca, /AIS, /OP, /op, /OPM, /TK, /Font, /TR,
+      # /TR2, /HTO, /HTP, /HT, /FL, /UseBlackPtComp, /BG, /BG2,
+      # /UCR, /UCR2, /SMask, /HUk.
+      class GraphicsStateParameter < Pdfrb::Model::Cos::Dictionary
+        arlington_object "GraphicsStateParameter"
+        register_type :ExtGState
+      end
+    end
+  end
+end

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

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Document info dict (s7.9.3). Title/Author/Subject/Keywords/
+      # Creator/Producer/CreationDate/ModDate/Trapped.
+      # Pre-PDF-2.0; superseded by /Metadata XMP in modern docs but
+      # still widely used.
+      class Info < Pdfrb::Model::Cos::Dictionary
+        arlington_object "DocInfo"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # AcroForm interactive form root (s12.7.2). Catalog /AcroForm.
+      class InteractiveForm < Pdfrb::Model::Cos::Dictionary
+        arlington_object "InteractiveForm"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Metadata stream (s14.3.2, PDF 2.0 App Note 003). XMP payload.
+      # Has /Type /Metadata, /Subtype /XML.
+      class Metadata < Pdfrb::Model::Cos::Stream
+        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.
+      end
+    end
+  end
+end

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

@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Names dictionary (s7.9.6) — top-level name-tree roots for
+      # /Dests, /AP, /JavaScript, /Pages, /Templates, /URLS,
+      # /EmbeddedFiles, /AlternatePresentations, /Renditions.
+      class Names < Pdfrb::Model::Cos::Dictionary
+        arlington_object "Name"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Compressed object stream (s7.5.7). /Type /ObjStm, /N count,
+      # /First byte offset of first object, /Extends parent ObjStm.
+      class ObjectStream < Pdfrb::Model::Cos::Stream
+        arlington_object "ObjectStream"
+        register_type :ObjStm
+      end
+    end
+  end
+end

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

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Optional Content Group (s8.11.2). Layer toggle for PDF layers.
+      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
+
+      # Optional Content Properties (s8.11.1). Catalog /OCProperties.
+      class OptionalContentProperties < Pdfrb::Model::Cos::Dictionary
+        arlington_object "OptContentProperties"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Alias for OptionalContentMembership — keeps the autoload path
+      # distinct from the group class.
+      class OptionalContentMembership < Pdfrb::Model::Cos::Dictionary
+        arlington_object "OptContentMembership"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Alias for OptionalContentProperties — autoload path.
+      class OptionalContentProperties < Pdfrb::Model::Cos::Dictionary
+        arlington_object "OptContentProperties"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,20 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Outline (bookmark) root (s12.3.3). Linked from Catalog /Outlines.
+      class Outline < Pdfrb::Model::Cos::Dictionary
+        arlington_object "Outline"
+        register_type :Outlines
+      end
+
+      # One bookmark entry (s12.3.3). Forms a doubly-linked list with
+      # /Next, /Prev, /First, /Last, /Parent; carries /Title, /A or
+      # /Dest for the navigation target.
+      class OutlineItem < Pdfrb::Model::Cos::Dictionary
+        arlington_object "OutlineItem"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Single bookmark entry. Lives in its own file so the autoload
+      # path matches the TSV-derived class name.
+      class OutlineItem < Pdfrb::Model::Cos::Dictionary
+        arlington_object "OutlineItem"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Output intent (s14.11.5) — ICC profile + condition metadata.
+      class OutputIntent < Pdfrb::Model::Cos::Dictionary
+        arlington_object "OutputIntents"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,86 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Page object (s7.7.3.3). Fields: /Type, /Parent, /LastModified,
+      # /Resources, /MediaBox, /CropBox, /BleedBox, /TrimBox, /ArtBox,
+      # /BoxColorInfo, /Contents, /Rotate, /Group, /Thumb, /B, /Dur,
+      # /Trans, /Annots, /AA, /Metadata, /PieceInfo, /StructParents,
+      # /ID, /PZ, /SeparationInfo, /Tabs, /TemplateInstantiated,
+      # /PresSteps, /UserUnit, /VP, /AF, /OutputIntents, /DPart.
+      class Page < Pdfrb::Model::Cos::Dictionary
+        arlington_object "PageObject"
+        register_type :Page
+
+        # Returns a +Content::Canvas+ bound to this page's /Contents
+        # stream. Auto-creates a Contents stream if the page has
+        # none. Memoised per page instance.
+        def canvas
+          @canvas ||= begin
+            ref = value[:Contents]
+            stream =
+              if ref.nil?
+                new_stream = document.add({}, type: Pdfrb::Model::Cos::Stream)
+                new_stream.stream = ""
+                value[:Contents] = Pdfrb::Model::Reference.new(new_stream.oid, 0)
+                new_stream
+              else
+                ref.is_a?(Pdfrb::Model::Reference) ? document.object(ref) : ref
+              end
+            Pdfrb::Content::Canvas.new(stream, document: document)
+          end
+        end
+
+        # Resolve and concatenate the page's content-stream bytes.
+        # /Contents is either a single Stream or an array of Streams.
+        def decoded_content
+          contents = value[:Contents]
+          return "".b if contents.nil?
+
+          streams = contents.is_a?(::Array) ? contents : [contents]
+          streams.each_with_object(+"".b) do |ref_or_stream, buf|
+            stream = ref_or_stream.is_a?(Pdfrb::Model::Reference) ?
+                       document&.object(ref_or_stream) : ref_or_stream
+            next unless stream.is_a?(Pdfrb::Model::Cos::Stream)
+
+            buf << stream.decoded_stream
+          end
+        end
+
+        # Resolve inheritable /MediaBox by walking up the page tree.
+        def media_box
+          inheritable(:MediaBox)
+        end
+
+        def crop_box
+          inheritable(:CropBox) || media_box
+        end
+
+        def resources
+          inheritable(:Resources)
+        end
+
+        def rotate
+          inheritable(:Rotate) || 0
+        end
+
+        private
+
+        def inheritable(key)
+          cur = self
+          while cur
+            val = cur.value[key]
+            return val unless val.nil?
+
+            parent_ref = cur.value[:Parent]
+            return nil unless parent_ref.is_a?(Pdfrb::Model::Reference)
+            return nil unless document
+
+            cur = document.object(parent_ref)
+          end
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,46 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Intermediate page-tree node (s7.7.3.2). Holds /Kids (array of
+      # PageTreeNode or Page) + /Count + inheritable /MediaBox,
+      # /CropBox, /Resources, /Rotate. The root uses the same dict
+      # shape and is conventionally distinguished via PageTreeNodeRoot
+      # for type-dispatch purposes.
+      class PageTreeNode < Pdfrb::Model::Cos::Dictionary
+        arlington_object "PageTreeNode"
+        register_type :Pages
+
+        # Walk all leaf PageObjects under this node.
+        def each_page(&block)
+          return enum_for(:each_page) unless block_given?
+
+          kids = self[:Kids] || []
+          kids.each do |kid_ref|
+            kid = kid_ref.is_a?(Pdfrb::Model::Reference) ?
+                    document&.object(kid_ref) : kid_ref
+            next unless kid
+
+            case kid[:Type]
+            when :Pages then kid.each_page(&block)
+            when :Page then yield kid
+            end
+          end
+        end
+
+        def page_count
+          self[:Count] || 0
+        end
+      end
+
+      # The root of the page tree — same shape as PageTreeNode but
+      # linked directly from the Catalog's /Pages. Tracked separately
+      # so type-dispatch can distinguish root from interior nodes
+      # (some validation rules apply only to the root).
+      class PageTreeNodeRoot < PageTreeNode
+        arlington_object "PageTreeNodeRoot"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,53 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Page-resources dict (s7.8). Named lookups for fonts, xobjects,
+      # colors, patterns, shadings, extgstate, properties.
+      class Resources < Pdfrb::Model::Cos::Dictionary
+        arlington_object "Resource"
+
+        def font(name)
+          lookup_named(:Font, name)
+        end
+
+        def xobject(name)
+          lookup_named(:XObject, name)
+        end
+
+        def ext_gstate(name)
+          lookup_named(:ExtGState, name)
+        end
+
+        def color_space(name)
+          lookup_named(:ColorSpace, name)
+        end
+
+        def pattern(name)
+          lookup_named(:Pattern, name)
+        end
+
+        def shading(name)
+          lookup_named(:Shading, name)
+        end
+
+        def properties(name)
+          lookup_named(:Properties, name)
+        end
+
+        private
+
+        def lookup_named(key, name)
+          sub = self[key]
+          return nil unless sub.is_a?(::Hash) || sub.is_a?(Pdfrb::Model::Cos::Dictionary)
+
+          ref = sub.value ? sub.value[name] : sub[name]
+          return ref unless ref.is_a?(Pdfrb::Model::Reference) && document
+
+          document.object(ref)
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Structure element (s14.7.4). One node in the structure tree.
+      # Holds /S (structure type), /P (parent), /K (kids), /Pg,
+      # /A, /C. Per PDF/UA Tech Note 001, /ActualText may appear
+      # for /Figure elements.
+      class StructElem < Pdfrb::Model::Cos::Dictionary
+        arlington_object "StructElem"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Structure tree root (s14.7.2). Catalog /StructTreeRoot.
+      # Holds /Type, /K, /ParentTree, /ParentTreeNextKey, /RoleMap,
+      # /ClassMap.
+      class StructTreeRoot < Pdfrb::Model::Cos::Dictionary
+        arlington_object "StructTreeRoot"
+        register_type :StructTreeRoot
+      end
+    end
+  end
+end

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

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Form XObject (s8.10). /Type /XObject, /Subtype /Form,
+      # /BBox, /Matrix, /Resources, /Group, /Filter, etc.
+      class XObjectForm < Pdfrb::Model::Cos::Stream
+        arlington_object "XObjectFormType1"
+      end
+    end
+  end
+end

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

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module Pdfrb
+  module Model
+    module Type
+      # Image XObject (s8.9). /Type /XObject, /Subtype /Image,
+      # /Width, /Height, /ColorSpace, /BitsPerComponent, etc.
+      class XObjectImage < Pdfrb::Model::Cos::Stream
+        arlington_object "XObjectImage"
+      end
+    end
+  end
+end