lib/tiff.rb in exifr-0.10.7 vs lib/tiff.rb in exifr-0.10.8

@@ -1,6 +1,6 @@
-# Copyright (c) 2007, 2008 - R.W. van 't Veer
+# Copyright (c) 2007, 2008, 2009 - R.W. van 't Veer
 
 require 'rational'
 
 module EXIFR
   # = TIFF decoder
@@ -300,26 +300,14 @@
     })
 
     # Names for all recognized TIFF fields.
     TAGS = ([TAG_MAPPING.keys, TAG_MAPPING.values.map{|v|v.values}].flatten.uniq - IFD_TAGS).map{|v|v.to_s}
 
-    # +file+ is a filename or an IO object.
+    # +file+ is a filename or an IO object.  Hint: use StringIO when working with slurped data like blobs.
     def initialize(file)
-      data = file.respond_to?(:read) ? file.read : File.open(file, 'rb') { |io| io.read }
+      data = Data.new(file)
 
-      class << data
-        attr_accessor :short, :long
-        def readshort(pos); self[pos..(pos + 1)].unpack(@short)[0]; end
-        def readlong(pos); self[pos..(pos + 3)].unpack(@long)[0]; end
-      end
-
-      case data[0..1]
-      when 'II'; data.short, data.long = 'v', 'V'
-      when 'MM'; data.short, data.long = 'n', 'N'
-      else; raise 'no II or MM marker found'
-      end
-
       @ifds = [IFD.new(data)]
       while ifd = @ifds.last.next
         break if @ifds.find{|i| i.offset == ifd.offset}
         @ifds << ifd
       end
@@ -462,12 +450,13 @@
     class Field # :nodoc:
       attr_reader :tag, :offset, :value
 
       def initialize(data, pos)
         @tag, count, @offset = data.readshort(pos), data.readlong(pos + 4), data.readlong(pos + 8)
+        @type = data.readshort(pos + 2)
 
-        case data.readshort(pos + 2)
+        case @type
         when 1, 6 # byte, signed byte
           # TODO handle signed bytes
           len, pack = count, proc { |d| d }
         when 2 # ascii
           len, pack = count, proc { |d| d.strip }
@@ -475,10 +464,18 @@
           # TODO handle signed
           len, pack = count * 2, proc { |d| d.unpack(data.short + '*') }
         when 4, 9 # long, signed long
           # TODO handle signed
           len, pack = count * 4, proc { |d| d.unpack(data.long + '*') }
+        when 7 # undefined
+          # UserComment 
+          if @tag == 0x9286
+            len, pack = count, proc { |d| d.strip }
+            len -= 8 # reduce to account for first 8 bytes
+            start = len > 4 ? @offset + 8 : (pos + 8) # UserComment first 8-bytes is char code
+            @value = [pack[data[start..(start + len - 1)]]].flatten
+          end
         when 5, 10
           len, pack = count * 8, proc do |d|
             r = []
             d.unpack(data.long + '*').each_with_index do |v,i|
               i % 2 == 0 ? r << [v] : r.last << v
@@ -491,13 +488,60 @@
               end
             end
           end
         end
 
-        if len && pack
+        if len && pack && @type != 7
           start = len > 4 ? @offset : (pos + 8)
           @value = [pack[data[start..(start + len - 1)]]].flatten
         end
+      end
+    end
+
+    class Data #:nodoc:
+      attr_reader :short, :long
+
+      def initialize(file)
+        @file = file.respond_to?(:read) ? file : File.open(file, 'rb')
+        @buffer = ''
+        @pos = 0
+
+        case self[0..1]
+        when 'II'; @short, @long = 'v', 'V'
+        when 'MM'; @short, @long = 'n', 'N'
+        else; raise 'no II or MM marker found'
+        end
+      end
+
+      def [](pos)
+        unless pos.respond_to?(:begin) && pos.respond_to?(:end)
+          pos = pos..pos
+        end
+
+        if pos.begin < @pos || pos.end >= @pos + @buffer.size
+          read_for(pos)
+        end
+
+        @buffer[(pos.begin - @pos)..(pos.end - @pos)]
+      end
+
+      def readshort(pos)
+        self[pos..(pos + 1)].unpack(@short)[0]
+      end
+
+      def readlong(pos)
+        self[pos..(pos + 3)].unpack(@long)[0]
+      end
+
+      def size
+        @file.seek(0, IO::SEEK_END)
+        @file.pos
+      end
+
+    private
+      def read_for(pos)
+        @file.seek(@pos = pos.begin)
+        @buffer = @file.read([pos.end - pos.begin, 4096].max)
       end
     end
   end
 end