origami 2.0.0 → 2.0.1

data/lib/origami/page.rb

@@ -29,9 +29,7 @@ module Origami
         # Pass the Page object if a block is present.
         #
         def append_page(page = Page.new)
-            unless self.Catalog and self.Catalog.Pages and self.Catalog.Pages.is_a?(PageTreeNode)
-                raise InvalidPDFError, "Invalid page tree"
-            end
+            init_page_tree
 
             treeroot = self.Catalog.Pages
 
@@ -55,9 +53,7 @@ module Origami
         # Pass the Page object if a block is present.
         #
         def insert_page(index, page = Page.new)
-            unless self.Catalog and self.Catalog.Pages and self.Catalog.Pages.is_a?(PageTreeNode)
-                raise InvalidPageTreeError, "Invalid page tree"
-            end
+            init_page_tree
 
             # Page from another document must be exported.
             page = page.export if page.document and page.document != self
@@ -73,9 +69,7 @@ module Origami
         # Returns an Enumerator of Page
         #
         def pages
-            unless self.Catalog and self.Catalog.Pages and self.Catalog.Pages.is_a?(PageTreeNode)
-                raise InvalidPageTreeError, "Invalid page tree"
-            end
+            init_page_tree
 
             self.Catalog.Pages.pages
         end
@@ -84,9 +78,7 @@ module Origami
         # Iterate through each page, returns self.
         #
         def each_page(&b)
-            unless self.Catalog and self.Catalog.Pages and self.Catalog.Pages.is_a?(PageTreeNode)
-                raise InvalidPageTreeError, "Invalid page tree"
-            end
+            init_page_tree
 
             self.Catalog.Pages.each_page(&b)
         end
@@ -95,9 +87,7 @@ module Origami
         # Get the n-th Page object.
         #
         def get_page(n)
-            unless self.Catalog and self.Catalog.Pages and self.Catalog.Pages.is_a?(PageTreeNode)
-                raise InvalidPageTreeError, "Invalid page tree"
-            end
+            init_page_tree
 
             self.Catalog.Pages.get_page(n)
         end
@@ -115,6 +105,19 @@ module Origami
         def each_named_page(&b)
             each_name(Names::PAGES, &b)
         end
+
+        private
+
+        def init_page_tree #:nodoc:
+            unless self.Catalog.key?(:Pages)
+                self.Catalog.Pages = PageTreeNode.new
+                return
+            end
+
+            unless self.Catalog.Pages.is_a?(PageTreeNode)
+                raise InvalidPageTreeError, "Root page node is not a PageTreeNode"
+            end
+        end
     end
 
     module ResourcesHolder
@@ -151,11 +154,12 @@ module Origami
         # If _name_ is not specified, a new name will be automatically generated.
         #
         def add_resource(type, rsrc, name = nil)
-            if name.nil? and existing = self.resources(type).key(rsrc)
-                return existing
+            if name.nil?
+                rsrc_name = self.resources(type).key(rsrc)
+                return rsrc_name if rsrc_name
             end
 
-            name = new_id(type) unless name
+            name ||= new_id(type)
             target = self.is_a?(Resources) ? self : (self.Resources ||= Resources.new)
 
             rsrc_dict = (target[type] and target[type].solve) || (target[type] = Dictionary.new)
@@ -234,14 +238,6 @@ module Origami
 
             Name.new(prefix + n)
         end
-
-        def new_extgstate_id; new_id(Resources::EXTGSTATE) end
-        def new_colorspace_id; new_id(Resources::COLORSPACE) end
-        def new_pattern_id; new_id(Resources::PATTERN) end
-        def new_shading_id; new_id(Resources::SHADING) end
-        def new_xobject_id; new_id(Resources::XOBJECT) end
-        def new_font_id; new_name(Resources::FONT) end
-        def new_properties_id; new_name(Resources::PROPERTIES) end
     end
 
     #
@@ -293,7 +289,7 @@ module Origami
             self.Count = 0
             self.Kids = []
 
-            super(hash, parser)
+            super
 
             set_indirect(true)
         end
@@ -421,7 +417,7 @@ module Origami
             raise IndexError, "Page not found"
         end
 
-        def << (pageset)
+        def <<(pageset)
             pageset = [pageset] unless pageset.is_a?(::Array)
             unless pageset.all? {|item| item.is_a?(Page) or item.is_a?(PageTreeNode) }
                 raise TypeError, "Cannot add anything but Page and PageTreeNode to this node"

data/lib/origami/parser.rb

@@ -67,8 +67,6 @@ module Origami
             @options.update(options)
             @logger = @options[:logger]
             @data = nil
-
-            ::String.disable_colorization(false) if @options[:colorize_log]
         end
 
         def pos
@@ -107,21 +105,8 @@ module Origami
                 obj = Object.parse(@data, self)
                 return if obj.nil?
 
-                if Origami::OPTIONS[:enable_type_propagation] and @deferred_casts.key?(obj.reference)
-                    types = @deferred_casts[obj.reference]
-                    types = [ types ] unless types.is_a?(::Array)
-
-                    # Promote object if a compatible type is found.
-                    if cast_type = types.find{|type| type < obj.class}
-                        obj = obj.cast_to(cast_type, self)
-                    end
-                end
-
-                trace "Read #{obj.type} object#{
-                    if obj.class != obj.native_type
-                        " (" + obj.native_type.to_s.split('::').last + ")"
-                    end
-                }, #{obj.reference}"
+                obj = try_object_promotion(obj)
+                trace "Read #{obj.type} object, #{obj.reference}"
 
                 @options[:callback].call(obj)
                 obj
@@ -205,54 +190,52 @@ module Origami
 
         private
 
-        def error(str = "") #:nodoc:
-            if @options[:colorize_log]
-                @logger.puts "[error] #{str}".red
-            else
-                @logger.puts "[error] #{str}"
-            end
-        end
+        #
+        # Attempt to promote an object using the deferred casts.
+        #
+        def try_object_promotion(obj)
+            return obj unless Origami::OPTIONS[:enable_type_propagation] and @deferred_casts.key?(obj.reference)
 
-        def warn(str = "") #:nodoc:
-            return unless @options[:verbosity] >= VERBOSE_INFO
+            types = @deferred_casts[obj.reference]
+            types = [ types ] unless types.is_a?(::Array)
 
-            if @options[:colorize_log]
-                @logger.puts "[info ] Warning: #{str}".yellow
+            # Promote object if a compatible type is found.
+            cast_type = types.find {|type| type < obj.class }
+            if cast_type
+                obj = obj.cast_to(cast_type, self)
             else
-                @logger.puts "[info ] Warning: #{str}"
+                obj
             end
         end
 
-        def info(str = "") #:nodoc:
-            return unless @options[:verbosity] >= VERBOSE_INFO
+        def error(msg = "") #:nodoc:
+            log(VERBOSE_QUIET, 'error', :red, msg.red)
+        end
 
-            if @options[:colorize_log]
-                @logger.print "[info ] ".green
-                @logger.puts str
-            else
-                @logger.puts "[info ] #{str}"
-            end
+        def warn(msg = "") #:nodoc:
+            log(VERBOSE_INFO, 'warn ', :yellow, msg.yellow)
         end
 
-        def debug(str = "") #:nodoc:
-            return unless @options[:verbosity] >= VERBOSE_DEBUG
+        def info(msg = "") #:nodoc:
+            log(VERBOSE_INFO, 'info ', :green, msg)
+        end
 
-            if @options[:colorize_log]
-                @logger.print "[debug] ".magenta
-                @logger.puts str
-            else
-                @logger.puts "[debug] #{str}"
-            end
+        def debug(msg = "") #:nodoc:
+            log(VERBOSE_DEBUG, 'debug', :magenta, msg)
+        end
+
+        def trace(msg = "") #:nodoc:
+            log(VERBOSE_TRACE, 'trace', :cyan, msg)
         end
 
-        def trace(str = "") #:nodoc:
-            return unless @options[:verbosity] >= VERBOSE_TRACE
+        def log(level, prefix, color, message) #:nodoc:
+            return unless @options[:verbosity] >= level
 
             if @options[:colorize_log]
-                @logger.print "[trace] ".cyan
-                @logger.puts str
+                @logger.print "[#{prefix}] ".colorize(color)
+                @logger.puts message
             else
-                @logger.puts "[trace] #{str}"
+                @logger.puts "[#{prefix}] #{message}"
             end
         end
 
@@ -266,7 +249,7 @@ module Origami
                 current_state.each_pair do |ref, type|
                     type = [ type ] unless type.is_a?(::Array)
                     type.each do |hint|
-                        break if document.cast_object(ref, hint, self)
+                        break if document.cast_object(ref, hint)
                     end
                 end
             end

data/lib/origami/parsers/fdf.rb

@@ -27,7 +27,7 @@ module Origami
             def parse(stream) #:nodoc:
                 super(stream)
 
-                fdf = FDF.new
+                fdf = FDF.new(self)
                 fdf.header = FDF::Header.parse(@data)
                 @options[:callback].call(fdf.header)
 
@@ -43,7 +43,7 @@ module Origami
                     trailer = fdf.revisions.first.trailer
 
                     if trailer[:Root].is_a?(Reference)
-                        fdf.cast_object(trailer[:Root], FDF::Catalog, self)
+                        fdf.cast_object(trailer[:Root], FDF::Catalog)
                     end
 
                     propagate_types(fdf)

data/lib/origami/parsers/pdf.rb

@@ -27,13 +27,14 @@ module Origami
             def initialize(params = {})
                 options =
                 {
-                  password: '',                 # Default password being tried when opening a protected document.
-                  prompt_password: lambda do    # Callback procedure to prompt password when document is encrypted.
-                      require 'io/console'
-                      STDERR.print "Password: "
-                      STDIN.noecho(&:gets).chomp
-                  end,
-                  force: false              # Force PDF header detection
+                    decrypt: true,                # Attempt to decrypt to document if encrypted (recommended).
+                    password: '',                 # Default password being tried when opening a protected document.
+                    prompt_password: lambda do    # Callback procedure to prompt password when document is encrypted.
+                        require 'io/console'
+                        STDERR.print "Password: "
+                        STDIN.noecho(&:gets).chomp
+                    end,
+                    force: false                  # Force PDF header detection
                 }.update(params)
 
                 super(options)
@@ -62,6 +63,8 @@ module Origami
             end
 
             def parse_finalize(pdf) #:nodoc:
+                cast_trailer_objects(pdf)
+
                 warn "This file has been linearized." if pdf.linearized?
 
                 propagate_types(pdf) if Origami::OPTIONS[:enable_type_propagation]
@@ -72,23 +75,43 @@ module Origami
                 if pdf.encrypted?
                     warn "This document contains encrypted data!"
 
-                    passwd = @options[:password]
-                    begin
-                        pdf.decrypt(passwd)
-                    rescue EncryptionInvalidPasswordError
-                        if passwd.empty?
-                            passwd = @options[:prompt_password].call
-                            retry unless passwd.empty?
-                        end
-
-                        raise
-                    end
+                    decrypt_document(pdf) if @options[:decrypt]
                 end
 
                 warn "This document has been signed!" if pdf.signed?
 
                 pdf
             end
+
+            def cast_trailer_objects(pdf) #:nodoc:
+                trailer = pdf.trailer
+
+                if trailer[:Root].is_a?(Reference)
+                    pdf.cast_object(trailer[:Root], Catalog)
+                end
+
+                if trailer[:Info].is_a?(Reference)
+                    pdf.cast_object(trailer[:Info], Metadata)
+                end
+
+                if trailer[:Encrypt].is_a?(Reference)
+                    pdf.cast_object(trailer[:Encrypt], Encryption::Standard::Dictionary)
+                end
+            end
+
+            def decrypt_document(pdf) #:nodoc:
+                passwd = @options[:password]
+                begin
+                    pdf.decrypt(passwd)
+                rescue EncryptionInvalidPasswordError
+                    if passwd.empty?
+                        passwd = @options[:prompt_password].call
+                        retry unless passwd.empty?
+                    end
+
+                    raise
+                end
+            end
         end
     end
 

data/lib/origami/parsers/pdf/lazy.rb

@@ -35,20 +35,9 @@ module Origami
                 pdf = parse_initialize
                 revisions = []
 
-                # Set the scanner position at the end.
-                @data.terminate
-
-                # Locate the startxref token.
-                until @data.match?(/#{Trailer::XREF_TOKEN}/)
-                    raise ParsingError, "No xref token found" if @data.pos == 0
-                    @data.pos -= 1
-                end
-
-                # Extract the offset of the last xref section.
-                trailer = Trailer.parse(@data, self)
-                raise ParsingError, "Cannot locate xref section" if trailer.startxref.zero?
+                # Locate the last xref offset at the end of the file.
+                xref_offset = locate_last_xref_offset
 
-                xref_offset = trailer.startxref
                 while xref_offset and xref_offset != 0
 
                     # Create a new revision based on the xref section offset.
@@ -78,6 +67,29 @@ module Origami
 
             private
 
+            #
+            # The document is scanned starting from the end, by locating the last startxref token.
+            #
+            def locate_last_xref_offset
+                # Set the scanner position at the end.
+                @data.terminate
+
+                # Locate the startxref token.
+                until @data.match?(/#{Trailer::XREF_TOKEN}/)
+                    raise ParsingError, "No xref token found" if @data.pos == 0
+                    @data.pos -= 1
+                end
+
+                # Extract the offset of the last xref section.
+                trailer = Trailer.parse(@data, self)
+                raise ParsingError, "Cannot locate xref section" if trailer.startxref.zero?
+
+                trailer.startxref
+            end
+
+            # 
+            # In the LazyParser, the revisions are parsed by jumping through the cross-references (table or streams).
+            #
             def parse_revision(pdf, offset)
                 raise ParsingError, "Invalid xref offset" if offset < 0 or offset >= @data.string.size
 
@@ -88,48 +100,73 @@ module Origami
 
                 # Regular xref section.
                 if @data.match?(/#{XRef::Section::TOKEN}/)
-                    xreftable = parse_xreftable
-                    raise ParsingError, "Cannot parse xref section" if xreftable.nil?
-
-                    revision.xreftable = xreftable
-                    revision.trailer = parse_trailer
-
-                    # Handle hybrid cross-references.
-                    if revision.trailer[:XRefStm].is_a?(Integer)
-                        begin
-                            offset = revision.trailer[:XRefStm].to_i
-                            xrefstm = parse_object(offset)
-
-                            if xrefstm.is_a?(XRefStream)
-                                revision.xrefstm = xrefstm
-                            else
-                                warn "Invalid xref stream at offset #{offset}"
-                            end
-
-                        rescue
-                            warn "Cannot parse xref stream at offset #{offset}"
-                        end
-                    end
+                    parse_revision_from_xreftable(revision)
 
                 # The xrefs are stored in a stream.
                 else
-                    xrefstm = parse_object
-                    raise ParsingError, "Invalid xref stream" unless xrefstm.is_a?(XRefStream)
+                    parse_revision_from_xrefstm(revision)
+                end
 
-                    revision.xrefstm = xrefstm
+                revision
+            end
 
-                    # Search for the trailer.
-                    if @data.skip_until Regexp.union(Trailer::XREF_TOKEN, *Trailer::TOKENS)
-                        @data.pos -= @data.matched_size
+            #
+            # Assume the current pointer is at the xreftable of the revision.
+            # We are expecting:
+            #   - a regular xref table, starting with xref
+            #   - a revision trailer
+            #
+            # The trailer may hold a XRefStm entry in case of hybrid references.
+            #
+            def parse_revision_from_xreftable(revision)
+                xreftable = parse_xreftable
+                raise ParsingError, "Cannot parse xref section" if xreftable.nil?
+
+                revision.xreftable = xreftable
+                revision.trailer = parse_trailer
+
+                # Handle hybrid cross-references.
+                if revision.trailer[:XRefStm].is_a?(Integer)
+                    begin
+                        offset = revision.trailer[:XRefStm].to_i
+                        xrefstm = parse_object(offset)
+
+                        if xrefstm.is_a?(XRefStream)
+                            revision.xrefstm = xrefstm
+                        else
+                            warn "Invalid xref stream at offset #{offset}"
+                        end
 
-                        revision.trailer = parse_trailer
-                    else
-                        warn "No trailer found."
-                        revision.trailer = Trailer.new
+                    rescue
+                        warn "Cannot parse xref stream at offset #{offset}"
                     end
                 end
+            end
 
-                revision
+            #
+            # Assume the current pointer is at the xref stream of the revision.
+            #
+            # The XRefStream should normally be at the end of the revision.
+            # We scan after the object for a trailer token.
+            # 
+            # The revision is allowed not to have a trailer, and the stream
+            # dictionary will be used as the trailer dictionary in that case.
+            #
+            def parse_revision_from_xrefstm(revision)
+                xrefstm = parse_object
+                raise ParsingError, "Invalid xref stream" unless xrefstm.is_a?(XRefStream)
+
+                revision.xrefstm = xrefstm
+
+                # Search for the trailer.
+                if @data.skip_until Regexp.union(Trailer::XREF_TOKEN, *Trailer::TOKENS)
+                    @data.pos -= @data.matched_size
+
+                    revision.trailer = parse_trailer
+                else
+                    warn "No trailer found."
+                    revision.trailer = Trailer.new
+                end
             end
         end
     end