podoff.rb in podoff-0.9.0

- old
+ new
@@ -24,8 +24,253 @@
 #++
 
 
 module Podoff
 
-  VERSION = '0.0.1'
+  VERSION = '0.9.0'
+
+  def self.load(path)
+
+    Podoff::Document.new(
+      File.open(path, 'r:iso8859-1') { |f| f.read })
+  end
+
+  class Document
+
+    attr_reader :header
+    attr_reader :objs
+    attr_reader :footer
+
+    def initialize(s)
+
+      fail ArgumentError.new('not a PDF file') \
+        unless s.match(/\A%PDF-\d+\.\d+\n/)
+
+      @header = []
+      #
+      @objs = {}
+      cur = nil
+      #
+      @footer = nil
+
+      s.split("\n").each do |l|
+
+        if @footer
+          @footer << l
+        elsif m = /^(\d+ \d+) obj\b/.match(l)
+          cur = (@objs[m[1]] = Obj.new(self, m[1]))
+          cur << l
+        elsif m = /^xref\b/.match(l)
+          @footer = []
+          @footer << l
+        elsif cur
+          cur << l
+        else
+          @header << l
+        end
+      end
+    end
+
+    def fonts; @objs.values.select(&:is_font?); end
+    def pages; @objs.values.select(&:is_page?); end
+
+    def page(i)
+
+      i < 1 ? nil : @objs.values.find { |o| o.page_number == i }
+    end
+
+    def dup
+
+      d0 = self
+
+      d = d0.class.allocate
+
+      d.instance_eval do
+        @header = d0.header.dup
+        @footer = d0.footer.dup
+        @objs = d0.objs.values.inject({}) { |h, v| h[v.ref] = v.dup(d); h }
+      end
+
+      d
+    end
+
+    def write(path)
+
+      File.open(path, 'wb') do |f|
+
+        @header.each { |l| f.print(l); f.print("\n") }
+
+        @objs.values.each do |o|
+          o.lines.each { |l| f.print(l); f.print("\n") }
+        end
+
+        @footer.each { |l| f.print(l); f.print("\n") }
+      end
+    end
+  end
+
+  class Obj
+
+    attr_reader :document
+    attr_reader :ref
+    attr_reader :lines
+
+    def initialize(doc, ref)
+
+      @document = doc
+      @ref = ref
+      @lines = []
+    end
+
+    def <<(l)
+
+      @lines << l
+    end
+
+    def lookup(k)
+
+      @lines.each do |l|
+
+        m = l.match(/^\/#{k} (.*)$/)
+        return m[1] if m
+      end
+
+      nil
+    end
+
+    def index(o, start=0)
+
+      @lines[start..-1].each_with_index do |l, i|
+
+        if o.is_a?(String)
+          return start + i if l == o
+        else
+          return start + i if l.match(o)
+        end
+      end
+
+      nil
+    end
+
+    def type
+
+      t = lookup('Type')
+      t ? t[1..-1] : nil
+    end
+
+    def page_number
+
+      r = lookup('pdftk_PageNum')
+      r ? r.to_i : nil
+    end
+
+    def is_page?
+
+      page_number != nil
+    end
+
+    def is_font?
+
+      type() == 'Font'
+    end
+
+    def parent
+
+      # /Parent 2 0 R
+
+      r = lookup('Parent')
+
+      r ? r[0..-2].strip : nil
+    end
+
+    def kids
+
+      # /Kids [1 0 R 16 0 R 33 0 R]
+
+      r = lookup('Kids')
+      (r || '').split(/[\[\]R]/).collect(&:strip).reject(&:empty?)
+    end
+
+    def contents
+
+      r = lookup('Contents')
+      r ? r[0..-2].strip : nil
+    end
+
+    def font_names
+
+      @lines.inject(nil) do |names, l|
+
+        if names
+          return names if l == '>>'
+          if m = l.match(/\/([^ ]+) /); names << m[1]; end
+        elsif l.match(/\/Font\s*$/)
+          names = []
+        end
+
+        names
+      end
+
+      []
+    end
+
+    def dup(new_doc)
+
+      o0 = self
+      o = o0.class.new(new_doc, @ref)
+      o.instance_eval { @lines = o0.lines.dup }
+
+      o
+    end
+
+    def find(opts={}, &block)
+
+      return self if block.call(self)
+
+      [ *kids, contents ].compact.each do |k|
+        o = @document.objs[k]
+        return o if o && block.call(o)
+      end
+
+      nil
+    end
+
+    def crop_box
+
+      r = lookup('CropBox') || lookup('MediaBox')
+
+      r ? r.strip[1..-2].split(' ').collect(&:strip).collect(&:to_f) : nil
+    end
+
+    def crop_dims
+
+      x, y, w, h = crop_box
+
+      x ? [ w - x, h - y ] : nil
+    end
+
+    def prepend_text(x, y, text, opts={})
+
+      o = find { |o| o.index('BT') }
+      fail ArgumentError.new('found no BT in the tree') unless o
+
+      font = opts[:font] || o.font_names.first || 'TT0'
+      size = opts[:size] || 10
+      comm = opts[:comment]
+
+      i = o.index('BT')
+      bt = []
+      bt << 'BT'
+      bt << "#{x} #{y} Td"
+      bt << "/#{font} #{size} Tf"
+      bt << "(#{text})Tj"
+      bt << 'ET'
+      bt << " % #{comm}" if comm
+      bt = bt.join(' ')
+
+      o.lines.insert(i, bt)
+
+      o
+    end
+  end
 end
 