python-pickle 0.1.0 → 0.2.0

data/lib/python/pickle/protocol3.rb

@@ -5,13 +5,19 @@ require 'python/pickle/instructions/short_bin_bytes'
 module Python
   module Pickle
     class Protocol3 < Protocol2
-      # Opcodes for Pickle protocol version 2.
+      # The `BINBYTES` opcode.
       #
-      # @see http://formats.kaitai.io/python_pickle/ruby.html
-      OPCODES = Protocol2::OPCODES + Set[
-        66, # BINBYTES
-        67  # SHORT_BINBYTES
-      ]
+      # @since 0.2.0
+      #
+      # @see https://github.com/python/cpython/blob/v3.10.9/Lib/pickle.py#L175
+      BINBYTES = 66
+
+      # The `SHORT_BINBYTES` opcode.
+      #
+      # @since 0.2.0
+      #
+      # @see https://github.com/python/cpython/blob/v3.10.9/Lib/pickle.py#L176
+      SHORT_BINBYTES = 67
 
       #
       # Reads an instruction from the pickle stream.
@@ -27,137 +33,100 @@ module Python
         #
         # Protocol 0 instructions
         #
-        when 40 # MARK
-          Instructions::MARK
-        when 46 # STOP
-          Instructions::STOP
-        when 48 # POP
-          Instructions::POP
-        when 49 # POP_MARK
-          Instructions::POP_MARK
-        when 50 # DUP
-          Instructions::DUP
-        when 70 # FLOAT
-          Instructions::Float.new(read_float)
-        when 73 # INT
-          Instructions::Int.new(read_int)
-        when 76 # LONG
-          Instructions::Long.new(read_long)
-        when 78 # NONE
-          Instructions::NONE
-        when 82 # REDUCE
-          Instructions::REDUCE
-        when 83 # STRING
-          Instructions::String.new(read_string)
-        when 86 # UNICODE
-          Instructions::String.new(read_unicode_string)
-        when 97 # APPEND
-          Instructions::APPEND
-        when 98 # BUILD
-          Instructions::BUILD
-        when 99 # GLOBAL
-          Instructions::Global.new(read_nl_string,read_nl_string)
-        when 100 # DICT
-          Instructions::DICT
-        when 103 # GET
-          Instructions::Get.new(read_int)
-        when 108 # LIST
-          Instructions::LIST
-        when 112 # PUT
-          Instructions::Put.new(read_int)
-        when 115 # SETITEM
-          Instructions::SETITEM
-        when 116 # TUPLE
-          Instructions::TUPLE
+        when MARK      then Instructions::MARK
+        when STOP      then Instructions::STOP
+        when POP       then Instructions::POP
+        when POP_MARK  then Instructions::POP_MARK
+        when DUP       then Instructions::DUP
+        when FLOAT     then read_float_instruction
+        when INT       then read_int_instruction
+        when LONG      then read_long_instruction
+        when NONE      then Instructions::NONE
+        when REDUCE    then Instructions::REDUCE
+        when STRING    then read_string_instruction
+        when UNICODE   then read_unicode_instruction
+        when APPEND    then Instructions::APPEND
+        when BUILD     then Instructions::BUILD
+        when GLOBAL    then read_global_instruction
+        when DICT      then Instructions::DICT
+        when GET       then read_get_instruction
+        when LIST      then Instructions::LIST
+        when PUT       then read_put_instruction
+        when SETITEM   then Instructions::SETITEM
+        when TUPLE     then Instructions::TUPLE
+        when INST      then read_inst_instruction
+        when OBJ       then Instructions::OBJ
+        when PERSID    then read_persid_instruction
+        when BINPERSID then Instructions::BINPERSID
         #
         # Protocol 1 instructions
         #
-        when 41  # EMPTY_TUPLE
-          Instructions::EMPTY_TUPLE
-        when 71  # BINFLOAT
-          Instructions::BinFloat.new(read_float64_be)
-        when 75  # BININT1
-          Instructions::BinInt1.new(read_uint8)
-        when 84  # BINSTRING
-          length = read_uint32_le
-          string = @io.read(length)
-
-          Instructions::BinString.new(length,string)
-        when 85  # SHORT_BINSTRING
-          length = read_uint8
-          string = @io.read(length)
-
-          Instructions::ShortBinString.new(length,string)
-        when 88  # BINUNICODE
-          length = read_uint32_le
-          string = @io.read(length).force_encoding(Encoding::UTF_8)
-
-          Instructions::BinUnicode.new(length,string)
-        when 93  # EMPTY_LIST
-          Instructions::EMPTY_LIST
-        when 101 # APPENDS
-          Instructions::APPENDS
-        when 104 # BINGET
-          Instructions::BinGet.new(read_uint8)
-        when 106 # LONG_BINGET
-          Instructions::LongBinGet.new(read_uint32_le)
-        when 113 # BINPUT
-          Instructions::BinPut.new(read_uint8)
-        when 117 # SETITEMS
-          Instructions::SETITEMS
-        when 125 # EMPTY_DICT
-          Instructions::EMPTY_DICT
+        when EMPTY_TUPLE     then Instructions::EMPTY_TUPLE
+        when BINFLOAT        then read_binfloat_instruction
+        when BININT1         then read_binint1_instruction
+        when BINSTRING       then read_binstring_instruction
+        when SHORT_BINSTRING then read_short_binstring_instruction
+        when BINUNICODE      then read_binunicode_instruction
+        when EMPTY_LIST      then Instructions::EMPTY_LIST
+        when APPENDS         then Instructions::APPENDS
+        when BINGET          then read_binget_instruction
+        when LONG_BINGET     then read_long_binget_instruction
+        when BINPUT          then read_binput_instruction
+        when SETITEMS        then Instructions::SETITEMS
+        when EMPTY_DICT      then Instructions::EMPTY_DICT
         #
         # Protocol 2 instructions
         #
-        when 128 # PROT
-          Instructions::Proto.new(read_uint8)
-        when 129 # NEWOBJ
-          Instructions::NEWOBJ
-        when 130 # EXT1
-          Instructions::Ext1.new(read_uint8)
-        when 131 # EXT2
-          Instructions::Ext2.new(read_uint16_le)
-        when 132 # EXT4
-          Instructions::Ext4.new(read_uint32_le)
-        when 133 # TUPLE1
-          Instructions::TUPLE1
-        when 134 # TUPLE2
-          Instructions::TUPLE2
-        when 135 # TUPLE3
-          Instructions::TUPLE3
-        when 136 # NEWTRUE
-          Instructions::NEWTRUE
-        when 137 # NEWFALSE
-          Instructions::NEWFALSE
-        when 138 # LONG1
-          length = read_uint8
-          long   = read_int_le(length)
-
-          Instructions::Long1.new(length,long)
-        when 139 # LONG4
-          length = read_uint32_le
-          long   = read_int_le(length)
-
-          Instructions::Long4.new(length,long)
+        when PROTO    then read_proto_instruction
+        when NEWOBJ   then Instructions::NEWOBJ
+        when EXT1     then read_ext1_instruction
+        when EXT2     then read_ext2_instruction
+        when EXT4     then read_ext4_instruction
+        when TUPLE1   then Instructions::TUPLE1
+        when TUPLE2   then Instructions::TUPLE2
+        when TUPLE3   then Instructions::TUPLE3
+        when NEWTRUE  then Instructions::NEWTRUE
+        when NEWFALSE then Instructions::NEWFALSE
+        when LONG1    then read_long1_instruction
+        when LONG4    then read_long4_instruction
         #
         # Protocol 3 instructions
         #
-        when 66 # BINBYTES
-          length = read_uint32_le
-          bytes  = @io.read(length)
-
-          Instructions::BinBytes.new(length,bytes)
-        when 67 # SHORT_BINBYTES
-          length = read_uint8
-          bytes  = @io.read(length)
-
-          Instructions::ShortBinBytes.new(length,bytes)
+        when BINBYTES       then read_binbytes_instruction
+        when SHORT_BINBYTES then read_short_binbytes_instruction
         else
           raise(InvalidFormat,"invalid opcode (#{opcode.inspect}) for protocol 3")
         end
       end
 
+      #
+      # Reads a `BINBYTES` instruction.
+      #
+      # @return [Instructions::BinBytes]
+      #
+      # @since 0.2.0
+      #
+      def read_binbytes_instruction
+        length = read_uint32_le
+        bytes  = @io.read(length)
+
+        Instructions::BinBytes.new(length,bytes)
+      end
+
+      #
+      # Reads a `SHORT_BINBYTES` instruction.
+      #
+      # @return [Instructions::ShortBinBytes]
+      #
+      # @since 0.2.0
+      #
+      def read_short_binbytes_instruction
+        length = read_uint8
+        bytes  = @io.read(length)
+
+        Instructions::ShortBinBytes.new(length,bytes)
+      end
+
     end
   end
 end

data/lib/python/pickle/protocol4.rb

@@ -26,21 +26,75 @@ module Python
         @io_stack = []
       end
 
-      # Opcodes for Pickle protocol 4.
-      #
-      # @see https://peps.python.org/pep-3154/
-      OPCODES = Protocol3::OPCODES + Set[
-        140, # SHORT_BINUNICODE
-        141, # BINUNICODE8
-        142, # BINBYTES8
-        143, # EMPTY_SET
-        144, # ADDITEMS
-        145, # FROZENSET
-        146, # NEWOBJ_EX
-        147, # STACK_GLOBAL
-        148, # MEMOIZE
-        149  # FRAME
-      ]
+      # The `SHORT_BINUNICODE` opcode.
+      #
+      # @since 0.2.0
+      #
+      # @see https://github.com/python/cpython/blob/v2.7/Lib/pickle.py#L180
+      SHORT_BINUNICODE = 140
+
+      # The `BINUNICODE8` opcode.
+      #
+      # @since 0.2.0
+      #
+      # @see https://github.com/python/cpython/blob/v2.7/Lib/pickle.py#L181
+      BINUNICODE8 = 141
+
+      # The `BINBYTES` opcode.
+      #
+      # @since 0.2.0
+      #
+      # @see https://github.com/python/cpython/blob/v2.7/Lib/pickle.py#L182
+      BINBYTES8 = 142
+
+      # The `EMPTY_SET` opcode.
+      #
+      # @since 0.2.0
+      #
+      # @see https://github.com/python/cpython/blob/v2.7/Lib/pickle.py#L183
+      EMPTY_SET = 143
+
+      # The `ADDITEMS` opcode.
+      #
+      # @since 0.2.0
+      #
+      # @see https://github.com/python/cpython/blob/v2.7/Lib/pickle.py#L184
+      ADDITEMS = 144
+
+      # The `FROZENSET` opcode.
+      #
+      # @since 0.2.0
+      #
+      # @see https://github.com/python/cpython/blob/v2.7/Lib/pickle.py#L185
+      FROZENSET = 145
+
+      # The `NEWOBJ_EX` opcode.
+      #
+      # @since 0.2.0
+      #
+      # @see https://github.com/python/cpython/blob/v2.7/Lib/pickle.py#L186
+      NEWOBJ_EX = 146
+
+      # The `STACK_GLOBAL` opcode.
+      #
+      # @since 0.2.0
+      #
+      # @see https://github.com/python/cpython/blob/v2.7/Lib/pickle.py#L187
+      STACK_GLOBAL = 147
+
+      # The `MEMOIZE` opcode.
+      #
+      # @since 0.2.0
+      #
+      # @see https://github.com/python/cpython/blob/v2.7/Lib/pickle.py#L188
+      MEMOIZE = 148
+
+      # The `FRAME` opcode.
+      #
+      # @since 0.2.0
+      #
+      # @see https://github.com/python/cpython/blob/v2.7/Lib/pickle.py#L189
+       FRAME = 149
 
       #
       # Reads an instruction from the pickle stream.
@@ -56,168 +110,80 @@ module Python
         #
         # Protocol 0 instructions
         #
-        when 40 # MARK
-          Instructions::MARK
-        when 46 # STOP
-          Instructions::STOP
-        when 48 # POP
-          Instructions::POP
-        when 49 # POP_MARK
-          Instructions::POP_MARK
-        when 50 # DUP
-          Instructions::DUP
-        when 70 # FLOAT
-          Instructions::Float.new(read_float)
-        when 73 # INT
-          Instructions::Int.new(read_int)
-        when 76 # LONG
-          Instructions::Long.new(read_long)
-        when 78 # NONE
-          Instructions::NONE
-        when 82 # REDUCE
-          Instructions::REDUCE
-        when 83 # STRING
-          Instructions::String.new(read_string)
-        when 86 # UNICODE
-          Instructions::String.new(read_unicode_string)
-        when 97 # APPEND
-          Instructions::APPEND
-        when 98 # BUILD
-          Instructions::BUILD
-        when 99 # GLOBAL
-          Instructions::Global.new(read_nl_string,read_nl_string)
-        when 100 # DICT
-          Instructions::DICT
-        when 103 # GET
-          Instructions::Get.new(read_int)
-        when 108 # LIST
-          Instructions::LIST
-        when 112 # PUT
-          Instructions::Put.new(read_int)
-        when 115 # SETITEM
-          Instructions::SETITEM
-        when 116 # TUPLE
-          Instructions::TUPLE
+        when MARK      then Instructions::MARK
+        when STOP      then Instructions::STOP
+        when POP       then Instructions::POP
+        when POP_MARK  then Instructions::POP_MARK
+        when DUP       then Instructions::DUP
+        when FLOAT     then read_float_instruction
+        when INT       then read_int_instruction
+        when LONG      then read_long_instruction
+        when NONE      then Instructions::NONE
+        when REDUCE    then Instructions::REDUCE
+        when STRING    then read_string_instruction
+        when UNICODE   then read_unicode_instruction
+        when APPEND    then Instructions::APPEND
+        when BUILD     then Instructions::BUILD
+        when GLOBAL    then read_global_instruction
+        when DICT      then Instructions::DICT
+        when GET       then read_get_instruction
+        when LIST      then Instructions::LIST
+        when PUT       then read_put_instruction
+        when SETITEM   then Instructions::SETITEM
+        when TUPLE     then Instructions::TUPLE
+        when INST      then read_inst_instruction
+        when OBJ       then Instructions::OBJ
+        when PERSID    then read_persid_instruction
+        when BINPERSID then Instructions::BINPERSID
         #
         # Protocol 1 instructions
         #
-        when 41  # EMPTY_TUPLE
-          Instructions::EMPTY_TUPLE
-        when 71  # BINFLOAT
-          Instructions::BinFloat.new(read_float64_be)
-        when 75  # BININT1
-          Instructions::BinInt1.new(read_uint8)
-        when 84  # BINSTRING
-          length = read_uint32_le
-          string = @io.read(length)
-
-          Instructions::BinString.new(length,string)
-        when 85  # SHORT_BINSTRING
-          length = read_uint8
-          string = @io.read(length)
-
-          Instructions::ShortBinString.new(length,string)
-        when 88  # BINUNICODE
-          length = read_uint32_le
-          string = @io.read(length).force_encoding(Encoding::UTF_8)
-
-          Instructions::BinUnicode.new(length,string)
-        when 93  # EMPTY_LIST
-          Instructions::EMPTY_LIST
-        when 101 # APPENDS
-          Instructions::APPENDS
-        when 104 # BINGET
-          Instructions::BinGet.new(read_uint8)
-        when 106 # LONG_BINGET
-          Instructions::LongBinGet.new(read_uint32_le)
-        when 113 # BINPUT
-          Instructions::BinPut.new(read_uint8)
-        when 117 # SETITEMS
-          Instructions::SETITEMS
-        when 125 # EMPTY_DICT
-          Instructions::EMPTY_DICT
+        when EMPTY_TUPLE     then Instructions::EMPTY_TUPLE
+        when BINFLOAT        then read_binfloat_instruction
+        when BININT1         then read_binint1_instruction
+        when BINSTRING       then read_binstring_instruction
+        when SHORT_BINSTRING then read_short_binstring_instruction
+        when BINUNICODE      then read_binunicode_instruction
+        when EMPTY_LIST      then Instructions::EMPTY_LIST
+        when APPENDS         then Instructions::APPENDS
+        when BINGET          then read_binget_instruction
+        when LONG_BINGET     then read_long_binget_instruction
+        when BINPUT          then read_binput_instruction
+        when SETITEMS        then Instructions::SETITEMS
+        when EMPTY_DICT      then Instructions::EMPTY_DICT
         #
         # Protocol 2 instructions
         #
-        when 128 # PROT
-          Instructions::Proto.new(read_uint8)
-        when 129 # NEWOBJ
-          Instructions::NEWOBJ
-        when 130 # EXT1
-          Instructions::Ext1.new(read_uint8)
-        when 131 # EXT2
-          Instructions::Ext2.new(read_uint16_le)
-        when 132 # EXT4
-          Instructions::Ext4.new(read_uint32_le)
-        when 133 # TUPLE1
-          Instructions::TUPLE1
-        when 134 # TUPLE2
-          Instructions::TUPLE2
-        when 135 # TUPLE3
-          Instructions::TUPLE3
-        when 136 # NEWTRUE
-          Instructions::NEWTRUE
-        when 137 # NEWFALSE
-          Instructions::NEWFALSE
-        when 138 # LONG1
-          length = read_uint8
-          long   = read_int_le(length)
-
-          Instructions::Long1.new(length,long)
-        when 139 # LONG4
-          length = read_uint32_le
-          long   = read_int_le(length)
-
-          Instructions::Long4.new(length,long)
+        when PROTO    then read_proto_instruction
+        when NEWOBJ   then Instructions::NEWOBJ
+        when EXT1     then read_ext1_instruction
+        when EXT2     then read_ext2_instruction
+        when EXT4     then read_ext4_instruction
+        when TUPLE1   then Instructions::TUPLE1
+        when TUPLE2   then Instructions::TUPLE2
+        when TUPLE3   then Instructions::TUPLE3
+        when NEWTRUE  then Instructions::NEWTRUE
+        when NEWFALSE then Instructions::NEWFALSE
+        when LONG1    then read_long1_instruction
+        when LONG4    then read_long4_instruction
         #
         # Protocol 3 instructions
         #
-        when 66 # BINBYTES
-          length = read_uint32_le
-          bytes  = @io.read(length)
-
-          Instructions::BinBytes.new(length,bytes)
-        when 67 # SHORT_BINBYTES
-          length = read_uint8
-          bytes  = @io.read(length)
-
-          Instructions::ShortBinBytes.new(length,bytes)
+        when BINBYTES       then read_binbytes_instruction
+        when SHORT_BINBYTES then read_short_binbytes_instruction
         #
         # Protocol 4 instructions
         #
-        when 140 # SHORT_BINUNICODE
-          length = read_uint8
-          string = read_utf8_string(length)
-
-          Instructions::ShortBinUnicode.new(length,string)
-        when 141 # BINUNICODE8
-          length = read_uint64_le
-          string = read_utf8_string(length)
-
-          Instructions::BinUnicode8.new(length,string)
-        when 142 # BINBYTES8
-          length = read_uint64_le
-          bytes  = @io.read(length)
-
-          Instructions::BinBytes8.new(length,bytes)
-        when 143 # EMPTY_SET
-          Instructions::EMPTY_SET
-        when 144 # ADDITEMS
-          Instructions::ADDITEMS
-        when 145 # FROZENSET
-          Instructions::FROZENSET
-        when 146 # NEWOBJ_EX
-          Instructions::NEWOBJ_EX
-        when 147 # STACK_GLOBAL
-          Instructions::STACK_GLOBAL
-        when 148 # MEMOIZE
-          Instructions::MEMOIZE
-        when 149 # FRAME
-          length = read_uint64_le
-
-          enter_frame(read_frame(length))
-
-          Instructions::Frame.new(length)
+        when SHORT_BINUNICODE then read_short_binunicode_instruction
+        when BINUNICODE8      then read_binunicode8_instruction
+        when BINBYTES8        then read_binbytes8_instruction
+        when EMPTY_SET        then Instructions::EMPTY_SET
+        when ADDITEMS         then Instructions::ADDITEMS
+        when FROZENSET        then Instructions::FROZENSET
+        when NEWOBJ_EX        then Instructions::NEWOBJ_EX
+        when STACK_GLOBAL     then Instructions::STACK_GLOBAL
+        when MEMOIZE          then Instructions::MEMOIZE
+        when FRAME            then read_frame_instruction
         else
           raise(InvalidFormat,"invalid opcode (#{opcode.inspect}) for protocol 4")
         end
@@ -280,6 +246,63 @@ module Python
         @io = @io_stack.pop
       end
 
+      #
+      # Reads a `SHORT_BINUNICODE` instruction.
+      #
+      # @return [Instructions::ShortBinUnicode]
+      #
+      # @since 0.2.0
+      #
+      def read_short_binunicode_instruction
+        length = read_uint8
+        string = read_utf8_string(length)
+
+        Instructions::ShortBinUnicode.new(length,string)
+      end
+
+      #
+      # Reads a `BINUNICODE8` instruction.
+      #
+      # @return [Instructions::BinUnicode8]
+      #
+      # @since 0.2.0
+      #
+      def read_binunicode8_instruction
+        length = read_uint64_le
+        string = read_utf8_string(length)
+
+        Instructions::BinUnicode8.new(length,string)
+      end
+
+      #
+      # Reads a `BINBYTES8` instruction.
+      #
+      # @return [Instructions::BinBytes8]
+      #
+      # @since 0.2.0
+      #
+      def read_binbytes8_instruction
+        length = read_uint64_le
+        bytes  = @io.read(length)
+
+        Instructions::BinBytes8.new(length,bytes)
+      end
+
+      #
+      # Reads a `FRAME` instruction.
+      #
+      # @return [Instructions::Frame]
+      #
+      # @since 0.2.0
+      #
+      def read_frame_instruction
+        length = read_uint64_le
+
+        enter_frame(read_frame(length))
+
+        Instructions::Frame.new(length)
+      end
+
     end
   end
 end