Doc/libmarshal.tex

   1 \section{Built-in Module \sectcode{marshal}}
   2
   3 \bimodindex{marshal}
   4 This module contains functions that can read and write Python
   5 values in a binary format.  The format is specific to Python, but
   6 independent of machine architecture issues (e.g., you can write a
   7 Python value to a file on a PC, transport the file to a Sun, and read
   8 it back there).  Details of the format are undocumented on purpose;
   9 it may change between Python versions (although it rarely does).%
  10 \footnote{The name of this module stems from a bit of terminology used
  11 by the designers of Modula-3 (amongst others), who use the term
  12 ``marshalling'' for shipping of data around in a self-contained form.
  13 Strictly speaking, ``to marshal'' means to convert some data from
  14 internal to external form (in an RPC buffer for instance) and
  15 ``unmarshalling'' for the reverse process.}
  16
  17 This is not a general ``persistency'' module.  For general persistency
  18 and transfer of Python objects through RPC calls, see the modules
  19 \code{pickle} and \code{shelve}.  The \code{marshal} module exists
  20 mainly to support reading and writing the ``pseudo-compiled'' code for
  21 Python modules of \samp{.pyc} files.
  22 \stmodindex{pickle}
  23 \stmodindex{shelve}
  24 \obindex{code}
  25
  26 Not all Python object types are supported; in general, only objects
  27 whose value is independent from a particular invocation of Python can
  28 be written and read by this module.  The following types are supported:
  29 \code{None}, integers, long integers, floating point numbers,
  30 strings, tuples, lists, dictionaries, and code objects, where it
  31 should be understood that tuples, lists and dictionaries are only
  32 supported as long as the values contained therein are themselves
  33 supported; and recursive lists and dictionaries should not be written
  34 (they will cause infinite loops).
  35
  36 {\bf Caveat:} On machines where C's \code{long int} type has more than
  37 32 bits (such as the DEC Alpha), it
  38 is possible to create plain Python integers that are longer than 32
  39 bits.  Since the current \code{marshal} module uses 32 bits to
  40 transfer plain Python integers, such values are silently truncated.
  41 This particularly affects the use of very long integer literals in
  42 Python modules --- these will be accepted by the parser on such
  43 machines, but will be silently be truncated when the module is read
  44 from the \code{.pyc} instead.%
  45 \footnote{A solution would be to refuse such literals in the parser,
  46 since they are inherently non-portable.  Another solution would be to
  47 let the \code{marshal} module raise an exception when an integer value
  48 would be truncated.  At least one of these solutions will be
  49 implemented in a future version.}
  50
  51 There are functions that read/write files as well as functions
  52 operating on strings.
  53
  54 The module defines these functions:
  55
  56 \renewcommand{\indexsubitem}{(in module marshal)}
  57
  58 \begin{funcdesc}{dump}{value\, file}
  59   Write the value on the open file.  The value must be a supported
  60   type.  The file must be an open file object such as
  61   \code{sys.stdout} or returned by \code{open()} or
  62   \code{posix.popen()}.
  63
  64   If the value has an unsupported type, garbage is written which cannot
  65   be read back by \code{load()}.
  66 \end{funcdesc}
  67
  68 \begin{funcdesc}{load}{file}
  69   Read one value from the open file and return it.  If no valid value
  70   is read, raise \code{EOFError}, \code{ValueError} or
  71   \code{TypeError}.  The file must be an open file object.
  72 \end{funcdesc}
  73
  74 \begin{funcdesc}{dumps}{value}
  75   Return the string that would be written to a file by
  76   \code{dump(value, file)}.  The value must be a supported type.
  77 \end{funcdesc}
  78
  79 \begin{funcdesc}{loads}{string}
  80   Convert the string to a value.  If no valid value is found, raise
  81   \code{EOFError}, \code{ValueError} or \code{TypeError}.  Extra
  82   characters in the string are ignored.
  83 \end{funcdesc}