Add support for reading ZIP files utilising AES encryption

lib/zip.rb

@@ -22,6 +22,7 @@
 require 'zip/pass_thru_compressor'
 require 'zip/pass_thru_decompressor'
 require 'zip/inflater'
+require 'zip/decrypter'
 require 'zip/deflater'
 require 'zip/streamable_stream'
 require 'zip/streamable_directory'

lib/zip/constants.rb

@@ -11,6 +11,8 @@ module Zip
   VERSION_NEEDED_TO_EXTRACT              = 20
   VERSION_NEEDED_TO_EXTRACT_ZIP64        = 45
 
+  GP_FLAGS_DESCRIPTOR_PRESENT = 0x08
+
   FILE_TYPE_FILE    = 010
   FILE_TYPE_DIR     = 004
   FILE_TYPE_SYMLINK = 012

lib/zip/decompressor.rb

@@ -1,5 +1,7 @@
 module Zip
   class Decompressor  #:nodoc:all
+    attr_accessor :input_stream
+
     CHUNK_SIZE = 32768
     def initialize(input_stream)
       super()

lib/zip/decrypter.rb

@@ -0,0 +1,95 @@
+require 'openssl'
+require 'stringio'
+
+module Zip
+  class Decrypter < Decompressor #:nodoc:all
+    VERIFIER_LENGTH = 2
+    BLOCK_SIZE = 16
+    AUTHENTICATION_CODE_LENGTH = 10
+
+    attr_writer :password
+
+    def initialize(input_stream, encryption_strength, entry_size, decompressor)
+      super(input_stream)
+
+      @data_length = entry_size - AUTHENTICATION_CODE_LENGTH
+      @decompressor = decompressor
+      @decompressor.input_stream = StringIO.new
+      @encryption_strength = encryption_strength
+      @prepared = false
+    end
+
+    def sysread(number_of_bytes = nil, buf = '')
+      prepare_aes unless @prepared
+
+      amount_to_read = @data_length
+      raise RuntimeError, "Incorrect entry size given, can't proceed" if amount_to_read <= 0
+
+      counter = 1
+      while amount_to_read > 0
+        set_iv(counter)
+
+        encrypted = @input_stream.read([BLOCK_SIZE, amount_to_read].min)
+        # Add the decrypted data to the IO object the decompressor interacts with
+        @decompressor.input_stream.write(@cipher.update(encrypted))
+
+        amount_to_read -= BLOCK_SIZE
+        counter += 1
+      end
+
+      # TODO: Check Authentication value
+      @input_stream.read(AUTHENTICATION_CODE_LENGTH)
+
+      @decompressor.input_stream.rewind
+      @decompressor.sysread
+    end
+
+    def input_finished?
+      @decompressor.input_finished?
+    end
+
+    alias :eof :input_finished?
+    alias :eof? :input_finished?
+
+    private
+
+    def prepare_aes
+      raise RuntimeError, "No password given" if @password.nil?
+      n = @encryption_strength + 1
+
+      headers = {
+        bits: 64 * n,
+        key_length: 8 * n,
+        mac_length: 8 * n,
+        salt_length: 4 * n
+      }
+
+      raise RuntimeError, "AES-#{headers[:bits]} is not supported." unless [0x01, 0x02, 0x03].include? @encryption_strength
+
+      @cipher = OpenSSL::Cipher::AES.new(headers[:bits], :CTR)
+      @cipher.decrypt
+
+      salt = @input_stream.read(headers[:salt_length])
+      verification = @input_stream.read(VERIFIER_LENGTH)
+      # The first few bytes are AES setup. Ensure we don't read beyond the end of the data during sysread
+      @data_length -= (headers[:salt_length] + VERIFIER_LENGTH)
+
+      key = OpenSSL::PKCS5.pbkdf2_hmac_sha1(
+        @password,
+        salt,
+        1000,
+        headers[:key_length] + headers[:mac_length] + VERIFIER_LENGTH
+      )
+
+      raise RuntimeError, "Incorrect password" unless key[-2..-1] == verification
+      @cipher.key = key
+
+      @prepared = true
+    end
+
+    def set_iv(counter)
+      # Reverse engineered this value from Zip4j's AES support.
+      @cipher.iv = [counter].pack("Vx12")
+    end
+  end
+end

lib/zip/entry.rb

@@ -2,6 +2,7 @@ module Zip
   class Entry
     STORED   = 0
     DEFLATED = 8
+    ENCRYPTED = 99
     # Language encoding flag (EFS) bit
     EFS = 0b100000000000
 
@@ -141,7 +142,10 @@ def cdir_header_size #:nodoc:all
     end
 
     def next_header_offset #:nodoc:all
-      local_entry_offset + self.compressed_size
+      # FIXME. the data descriptor can be only 12 bytes long, if the signature isn't present
+      # I need to add code to deal with this!
+      # I also have no idea why I have to subtract 8 bytes for AES  zips. It works, sooooo...
+      local_entry_offset + self.compressed_size + (@data_descriptor_present ? 16 : 0) - (@extra.member?('AES') ? 8 : 0)
     end
 
     # Extracts entry to file dest_path (defaults to @name).
@@ -248,6 +252,34 @@ def read_local_entry(io) #:nodoc:all
       end
       parse_zip64_extra(true)
       @local_header_size = calculate_local_header_size
+
+      # If the "data descriptor present" bit is set in the general flags
+      # we need to read the uncompressed size and CRC from the data descriptor
+      # otherwise they'll remain as 0
+      @data_descriptor_present = (@gp_flags & ::Zip::GP_FLAGS_DESCRIPTOR_PRESENT != 0)
+      if @data_descriptor_present
+        pos = io.tell
+        # We need to seek forwards until we find the data descriptor signature
+        # (504B0708) or the next record's local file header signature (504B0304)
+        # and scan back a few
+        last_four = []
+
+        loop do
+          last_four.push(io.read(1))
+          last_four = last_four[-4..-1] if last_four.length > 4
+
+          case last_four
+          when %W{\x50 \x4B \x07 \x08}
+            break
+          when %W{\x50 \x4B \x03 \x04}
+            io.seek(-12,IO::SEEK_CUR )
+            break
+          end
+        end
+
+        @crc, @compressed_size, @size = io.read(12).unpack("VVV")
+        io.seek(pos)
+      end
     end
 
     def pack_local_entry

lib/zip/extra_field.rb

@@ -94,6 +94,7 @@ def local_size
 require 'zip/extra_field/unix'
 require 'zip/extra_field/zip64'
 require 'zip/extra_field/zip64_placeholder'
+require 'zip/extra_field/aes'
 
 # Copyright (C) 2002, 2003 Thomas Sondergaard
 # rubyzip is free software; you can redistribute it and/or

lib/zip/extra_field/aes.rb

@@ -0,0 +1,31 @@
+module Zip
+  # Info-ZIP Extra for AES encryption
+  class ExtraField::AES < ExtraField::Generic
+    attr_reader :data_size, :vendor_version, :vendor_id, :encryption_strength, :compression_method
+    HEADER_ID = "\x01\x99".force_encoding("ASCII-8BIT")
+    register_map
+
+    def initialize(binstr = nil)
+      @data_size = nil
+      @vendor_version = nil
+      @vendor_id = nil
+      @encryption_strength = nil
+      @compression_method = nil
+      binstr and merge(binstr)
+    end
+
+    def merge(binstr)
+      return if binstr.empty?
+      _, @data_size, @vendor_version, @vendor_id, @encryption_strength, @compression_method = binstr.to_s.unpack("vvva2Cv")
+    end
+
+    def pack_for_local
+      return '' unless @data_size && @vendor_version && @vendor_id && @encryption_strength && @compression_method
+      [0x01, 0x99, @data_size, @vendor_version, @vendor_id, @encryption_strength, @compression_method].pack("vvvvM2Cv")
+    end
+
+    def pack_for_c_dir
+      pack_for_local
+    end
+  end
+end

lib/zip/file.rb

@@ -88,6 +88,10 @@ def initialize(file_name, create = nil, buffer = false, options = {})
       @restore_times       = options[:restore_times]        || true
     end
 
+    def password=(password)
+      @password = password
+    end
+
     class << self
       # Same as #new. If a block is passed the ZipFile object is passed
       # to the block and is automatically closed afterwards just as with
@@ -220,7 +224,16 @@ def split(zip_file_name, segment_size = MAX_SEGMENT_SIZE, delete_zip_file = true
     # the stream object is passed to the block and the stream is automatically
     # closed afterwards just as with ruby's builtin File.open method.
     def get_input_stream(entry, &aProc)
-      get_entry(entry).get_input_stream(&aProc)
+
+      if aProc
+        get_entry(entry).get_input_stream do |zis|
+          zis.password = @password if zis.respond_to? :password= # Make sure that tempfile does not call :password=
+          aProc.call(zis)
+        end
+      else
+        get_entry(entry).get_input_stream
+      end
+
     end
 
     # Returns an output stream to the specified entry. If entry is not an instance

lib/zip/input_stream.rb

@@ -129,17 +129,39 @@ def open_entry
       @current_entry
     end
 
-    def get_decompressor
-      case
-      when @current_entry.nil?
+    def get_decompressor(compression_method = nil)
+      compression_method ||= @current_entry.compression_method unless @current_entry.nil?
+      case compression_method
+      when nil
         ::Zip::NullDecompressor
-      when @current_entry.compression_method == ::Zip::Entry::STORED
+      when ::Zip::Entry::STORED
         ::Zip::PassThruDecompressor.new(@archive_io, @current_entry.size)
-      when @current_entry.compression_method == ::Zip::Entry::DEFLATED
+      when ::Zip::Entry::DEFLATED
         ::Zip::Inflater.new(@archive_io)
+      when ::Zip::Entry::ENCRYPTED
+        if @current_entry.extra["AES"].vendor_id != "AE"
+          raise ZipCompressionMethodError, "The #{@current_entry.extra["AES"].vendor_id} encryption method is not supported"
+        end
+
+        unless [1,2].include? @current_entry.extra["AES"].vendor_version
+          raise ZipCompressionMethodError, "Only AES-1 and AES-2 style encryption is supported"
+        end
+
+        if @current_entry.extra["AES"].compression_method == ::Zip::Entry::ENCRYPTED
+          # This would create infinite recursion.
+          raise ZipCompressionMethodError, "This zip file is malformed"
+        end
+
+        ::Zip::Decrypter.new(
+          @archive_io,
+          @current_entry.extra["AES"].encryption_strength,
+          @current_entry.compressed_size,
+          get_decompressor(@current_entry.extra["AES"].compression_method)
+        )
+
       else
         raise ::Zip::CompressionMethodError,
-              "Unsupported compression method #{@current_entry.compression_method}"
+              "Unsupported compression method #{compression_method}"
       end
     end
 

lib/zip/ioextras/abstract_input_stream.rb

@@ -16,8 +16,10 @@ def initialize
 
       attr_accessor :lineno
       attr_reader :pos
+      attr_writer :password
 
       def read(number_of_bytes = nil, buf = '')
+        @decompressor.password = @password if !@password.nil? and @decompressor.respond_to? :password=
         tbuf = if @output_buffer.bytesize > 0
                  if number_of_bytes <= @output_buffer.bytesize
                    @output_buffer.slice!(0, number_of_bytes)