[Subversion] / BytecodeAssembler / README.txt  

Diff of /BytecodeAssembler/README.txt

Parent Directory | Revision Log

version 2205, Tue Jul 4 22:21:25 2006 UTC version 2620, Mon Aug 2 07:06:28 2010 UTC
Line 9 
Line 9 
 bytecode instead of on these mechanical issues.  bytecode instead of on these mechanical issues.
   
 In addition to a low-level opcode-oriented API for directly generating specific  In addition to a low-level opcode-oriented API for directly generating specific
 bytecodes, this module also offers an extensible mini-AST framework for  Python bytecodes, this module also offers an extensible mini-AST framework for
 generating code from high-level specifications.  This framework does most of  generating code from high-level specifications.  This framework does most of
 the work needed to transform tree-like structures into linear bytecode  the work needed to transform tree-like structures into linear bytecode
 instructions, and includes the ability to do compile-time constant folding.  instructions, and includes the ability to do compile-time constant folding.
   
   Please see the `BytecodeAssembler reference manual`_ for more details.
   
   .. _BytecodeAssembler reference manual: http://peak.telecommunity.com/DevCenter/BytecodeAssembler#toc
   
   Changes since version 0.5:
   
   * Fix incorrect stack size calculation for ``MAKE_CLOSURE`` on Python 2.5+
   
   Changes since version 0.3:
   
   * New node types:
   
     * ``For(iterable, assign, body)`` -- define a "for" loop over `iterable`
   
     * ``UnpackSequence(nodes)`` -- unpacks a sequence that's ``len(nodes)`` long,
       and then generates the given nodes.
   
     * ``LocalAssign(name)`` -- issues a ``STORE_FAST``, ``STORE_DEREF`` or
       ``STORE_LOCAL`` as appropriate for the given name.
   
     * ``Function(body, name='<lambda>', args=(), var=None, kw=None, defaults=())``
       -- creates a nested function from `body` and puts it on the stack.
   
     * ``If(cond, then_, else_=Pass)`` -- "if" statement analogue
   
     * ``ListComp(body)`` and ``LCAppend(value)`` -- implement list comprehensions
   
     * ``YieldStmt(value)`` -- generates a ``YIELD_VALUE`` (plus a ``POP_TOP`` in
       Python 2.5+)
   
   * ``Code`` objects are now iterable, yielding ``(offset, op, arg)`` triples,
     where `op` is numeric and `arg` is either numeric or ``None``.
   
   * ``Code`` objects' ``.code()`` method can now take a "parent" ``Code`` object,
     to link the child code's free variables to cell variables in the parent.
   
   * Added ``Code.from_spec()`` classmethod, that initializes a code object from a
     name and argument spec.
   
   * ``Code`` objects now have a ``.nested(name, args, var, kw)`` method, that
     creates a child code object with the same ``co_filename`` and the supplied
     name/arg spec.
   
   * Fixed incorrect stack tracking for the ``FOR_ITER`` and ``YIELD_VALUE``
     opcodes
   
   * Ensure that ``CO_GENERATOR`` flag is set if ``YIELD_VALUE`` opcode is used
   
   * Change tests so that Python 2.3's broken line number handling in ``dis.dis``
     and constant-folding optimizer don't generate spurious failures in this
     package's test suite.
   
   
   Changes since version 0.2:
   
   * Added ``Suite``, ``TryExcept``, and ``TryFinally`` node types
   
   * Added a ``Getattr`` node type that does static or dynamic attribute access
     and constant folding
   
   * Fixed ``code.from_function()`` not copying the ``co_filename`` attribute when
     ``copy_lineno`` was specified.
   
   * The ``repr()`` of AST nodes doesn't include a trailing comma for 1-argument
     node types any more.
   
   * Added a ``Pass`` symbol that generates no code, a ``Compare()`` node type
     that does n-way comparisons, and ``And()`` and ``Or()`` node types for doing
     logical operations.
   
   * The ``COMPARE_OP()`` method now accepts operator strings like ``"<="``,
     ``"not in"``, ``"exception match"``, and so on, as well as numeric opcodes.
     See the standard library's ``opcode`` module for a complete list of the
     strings accepted (in the ``cmp_op`` tuple).  ``"<>"`` is also accepted as an
     alias for ``"!="``.
   
   * Added code to verify that forward jump offsets don't exceed a 64KB span, and
     support absolute backward jumps to locations >64KB.
   
 Changes since version 0.1:  Changes since version 0.1:
   
   * Constant handling has been fixed so that it doesn't confuse equal values of
     differing types (e.g. ``1.0`` and ``True``), or equal unhashable objects
     (e.g. two empty lists).
   
   * Removed ``nil``, ``ast_curry()`` and ``folding_curry()``, replacing them with
     the ``nodetype()`` decorator and ``fold_args()``; please see the docs for
     more details.
   
 * Added stack tracking across jumps, globally verifying stack level prediction  * Added stack tracking across jumps, globally verifying stack level prediction
   consistency and rejecting dead code.    consistency and automatically rejecting attempts to generate dead code.  It
     should now be virtually impossible to accidentally generate bytecode that can
     crash the interpreter.  (If you find a way, let me know!)
   
 Changes since version 0.0.1:  Changes since version 0.0.1:
   
Line 40 
Line 129 
 * Various bug fixes  * Various bug fixes
   
 There are a few features that aren't tested yet, and not all opcodes may be  There are a few features that aren't tested yet, and not all opcodes may be
 fully supported.  Notably, the following features are still NOT reliably  fully supported.  Also note the following limitations:
 supported yet:  
   
 * Wide jump addressing (for generated bytecode>64K in size)  * Jumps to as-yet-undefined labels cannot span a distance greater than 65,535
     bytes.
   
 * The ``dis()`` module in Python 2.3 has a bug that makes it show incorrect  * The ``dis()`` function in Python 2.3 has a bug that makes it show incorrect
   line numbers when the difference between two adjacent line numbers is    line numbers when the difference between two adjacent line numbers is
   greater than 255.  This causes two shallow failures in the current test    greater than 255.  (To work around this, the test_suite uses a later version
   suite when it's run under Python 2.3.    of ``dis()``, but do note that it may affect your own tests if you use
     ``dis()`` with Python 2.3 and use widely separated line numbers.)
   
 If you find any other issues, please let me know.  If you find any other issues, please let me know.
   
Line 58 
Line 148 
 Questions and discussion regarding this software should be directed to the  Questions and discussion regarding this software should be directed to the
 `PEAK Mailing List <http://www.eby-sarna.com/mailman/listinfo/peak>`_.  `PEAK Mailing List <http://www.eby-sarna.com/mailman/listinfo/peak>`_.
   
   .. _toc:
 .. contents:: **Table of Contents**  .. contents:: **Table of Contents**
   
   
Line 108 
Line 199 
     >>> f()      >>> f()
     42      42
   
   Finally, code objects are also iterable, yielding ``(offset, opcode, arg)``
   tuples, where `arg` is ``None`` for opcodes with no arguments, and an integer
   otherwise::
   
       >>> import peak.util.assembler as op
       >>> list(c) == [
       ...     (0, op.LOAD_CONST, 1),
       ...     (3, op.RETURN_VALUE, None)
       ... ]
       True
   
   This can be useful for testing or otherwise inspecting code you've generated.
   
   
 Opcodes and Arguments  Opcodes and Arguments
 =====================  =====================
Line 253 
Line 357 
     >>> c = Code()      >>> c = Code()
     >>> c.co_cellvars = ('a','b')      >>> c.co_cellvars = ('a','b')
   
       >>> import sys
     >>> c.LOAD_CLOSURE('a')      >>> c.LOAD_CLOSURE('a')
     >>> c.LOAD_CLOSURE('b')      >>> c.LOAD_CLOSURE('b')
       >>> if sys.version>='2.5':
       ...     c.BUILD_TUPLE(2) # In Python 2.5+, free vars must be in a tuple
     >>> c.LOAD_CONST(None)  # in real code, this'd be a Python code constant      >>> c.LOAD_CONST(None)  # in real code, this'd be a Python code constant
     >>> c.MAKE_CLOSURE(0,2) # no defaults, 2 free vars in the new function      >>> c.MAKE_CLOSURE(0,2) # no defaults, 2 free vars in the new function
   
       >>> c.stack_size         # This will be 1, no matter what Python version
       1
   
   The ``COMPARE_OP`` method takes an argument which can be a valid comparison
   integer constant, or a string containing a Python operator, e.g.::
   
       >>> c = Code()
       >>> c.LOAD_CONST(1)
       >>> c.LOAD_CONST(2)
       >>> c.COMPARE_OP('not in')
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (1)
                     3 LOAD_CONST               2 (2)
                     6 COMPARE_OP               7 (not in)
   
   The full list of valid operator strings can be found in the standard library's
   ``opcode`` module.  ``"<>"`` is also accepted as an alias for ``"!="``::
   
       >>> c.LOAD_CONST(3)
       >>> c.COMPARE_OP('<>')
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (1)
                     3 LOAD_CONST               2 (2)
                     6 COMPARE_OP               7 (not in)
                     9 LOAD_CONST               3 (3)
                    12 COMPARE_OP               3 (!=)
   
   
 High-Level Code Generation  High-Level Code Generation
 ==========================  ==========================
Line 291 
Line 425 
                  21 LOAD_CONST               0 (None)                   21 LOAD_CONST               0 (None)
                  24 LOAD_CONST               8 (<code object <lambda> at ...>)                   24 LOAD_CONST               8 (<code object <lambda> at ...>)
   
   Note that although some values of different types may compare equal to each
   other, ``Code`` objects will not substitute a value of a different type than
   the one you requested::
   
       >>> c = Code()
       >>> c(1, True, 1.0, 1L)     # equal, but different types
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (1)
                     3 LOAD_CONST               2 (True)
                     6 LOAD_CONST               3 (1.0)
                     9 LOAD_CONST               4 (1L)
   
 Simple Containers  Simple Containers
 -----------------  -----------------
Line 335 
Line 480 
 a constant, rather than generating code to recreate the tuple using a series of  a constant, rather than generating code to recreate the tuple using a series of
 ``LOAD_CONST`` operations followed by a ``BUILD_TUPLE``.  ``LOAD_CONST`` operations followed by a ``BUILD_TUPLE``.
   
   If the value wrapped in a ``Const`` is not hashable, it is compared by identity
   rather than value.  This prevents equal mutable values from being reused by
   accident, e.g. if you plan to mutate the "constant" values later::
   
       >>> c = Code()
       >>> c(Const([]), Const([]))     # equal, but not the same object!
       >>> dis(c.code())
         0           0 LOAD_CONST               1 ([])
                     3 LOAD_CONST               2 ([])
   
   Thus, although ``Const`` objects hash and compare based on equality for
   hashable types::
   
       >>> hash(Const(3)) == hash(3)
       True
       >>> Const(3)==Const(3)
       True
   
   They hash and compare based on object identity for non-hashable types::
   
       >>> c = Const([])
       >>> hash(c) == hash(id(c.value))
       True
       >>> c == Const(c.value)     # compares equal if same object
       True
       >>> c == Const([])          # but is not equal to a merely equal object
       False
   
   
   ``Suite`` and ``Pass``
   ----------------------
   
   On occasion, it's helpful to be able to group a sequence of opcodes,
   expressions, or statements together, to be passed as an argument to other node
   types.  The ``Suite`` node type accomplishes this::
   
       >>> from peak.util.assembler import Suite, Pass
   
       >>> c = Code()
       >>> c.return_(Suite([Const(42), Code.DUP_TOP, Code.POP_TOP]))
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (42)
                     3 DUP_TOP
                     4 POP_TOP
                     5 RETURN_VALUE
   
   And ``Pass`` is a shortcut for an empty ``Suite``, that generates nothing::
   
       >>> Suite([])
       Pass
   
       >>> c = Code()
       >>> c(Pass)
       >>> c.return_(None)
       >>> dis(c.code())
         0           0 LOAD_CONST               0 (None)
                     3 RETURN_VALUE
   
   
 Local and Global Names  Local and Global Names
 ----------------------  ----------------------
Line 350 
Line 553 
       0           0 LOAD_FAST                0 (x)        0           0 LOAD_FAST                0 (x)
                   3 LOAD_GLOBAL              0 (y)                    3 LOAD_GLOBAL              0 (y)
   
   
 As with simple constants and ``Const`` wrappers, these objects can be used to  As with simple constants and ``Const`` wrappers, these objects can be used to
 construct more complex expressions, like ``{a:(b,c)}``::  construct more complex expressions, like ``{a:(b,c)}``::
   
Line 366 
Line 568 
                  16 ROT_THREE                   16 ROT_THREE
                  17 STORE_SUBSCR                   17 STORE_SUBSCR
   
   The ``LocalAssign`` node type takes a name, and stores a value in a local
   variable::
   
       >>> from peak.util.assembler import LocalAssign
       >>> c = Code()
       >>> c(42, LocalAssign('x'))
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (42)
                     3 STORE_FAST               0 (x)
   
 If the code object is not using "fast locals" (i.e. ``CO_OPTIMIZED`` isn't  If the code object is not using "fast locals" (i.e. ``CO_OPTIMIZED`` isn't
 set), local variables will be dereferenced using ``LOAD_NAME`` instead of  set), local variables will be referenced using ``LOAD_NAME`` and ``STORE_NAME``
 ``LOAD_FAST``, and if the referenced local name is a "cell" or "free"  instead of ``LOAD_FAST`` and ``STORE_FAST``, and if the referenced local name
 variable, ``LOAD_DEREF`` is used instead::  is a "cell" or "free" variable, ``LOAD_DEREF`` and ``STORE_DEREF`` are used
   instead::
   
     >>> from peak.util.assembler import CO_OPTIMIZED      >>> from peak.util.assembler import CO_OPTIMIZED
     >>> c = Code()      >>> c = Code()
Line 377 
Line 590 
     >>> c.co_cellvars = ('y',)      >>> c.co_cellvars = ('y',)
     >>> c.co_freevars = ('z',)      >>> c.co_freevars = ('z',)
     >>> c( Local('x'), Local('y'), Local('z') )      >>> c( Local('x'), Local('y'), Local('z') )
       >>> c( LocalAssign('x'), LocalAssign('y'), LocalAssign('z') )
     >>> dis(c.code())      >>> dis(c.code())
       0           0 LOAD_NAME                0 (x)        0           0 LOAD_NAME                0 (x)
                   3 LOAD_DEREF               0 (y)                    3 LOAD_DEREF               0 (y)
                   6 LOAD_DEREF               1 (z)                    6 LOAD_DEREF               1 (z)
                     9 STORE_NAME               0 (x)
                    12 STORE_DEREF              0 (y)
                    15 STORE_DEREF              1 (z)
   
   
   Obtaining Attributes
   --------------------
   
   The ``Getattr`` node type takes an expression and an attribute name.  The
   attribute name can be a constant string, in which case a ``LOAD_ATTR`` opcode
   is used, and constant folding is done if possible::
   
       >>> from peak.util.assembler import Getattr
   
       >>> c = Code()
       >>> c(Getattr(Local('x'), '__class__'))
       >>> dis(c.code())
         0           0 LOAD_FAST                0 (x)
                     3 LOAD_ATTR                0 (__class__)
   
   
       >>> Getattr(Const(object), '__class__') # const expression, const result
       Const(<type 'type'>)
   
   Or the attribute name can be an expression, in which case a ``getattr()`` call
   is compiled instead::
   
       >>> c = Code()
       >>> c(Getattr(Local('x'), Local('y')))
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (<built-in function getattr>)
                     3 LOAD_FAST                0 (x)
                     6 LOAD_FAST                1 (y)
                     9 CALL_FUNCTION            2
   
   
 Calling Functions and Methods  Calling Functions and Methods
Line 462 
Line 710 
                   3 RETURN_VALUE                    3 RETURN_VALUE
   
   
   ``If`` Conditions
   -----------------
   
   The ``If()`` node type generates conditional code, roughly equivalent to a
   Python if/else statement::
   
       >>> from peak.util.assembler import If
       >>> c = Code()
       >>> c( If(Local('a'), Return(42), Return(55)) )
       >>> dis(c.code())
         0           0 LOAD_FAST                0 (a)
                     3 JUMP_IF_FALSE            5 (to 11)
                     6 POP_TOP
                     7 LOAD_CONST               1 (42)
                    10 RETURN_VALUE
               >>   11 POP_TOP
                    12 LOAD_CONST               2 (55)
                    15 RETURN_VALUE
   
   However, it can also be used like a Python 2.5+ conditional expression
   (regardless of the targeted Python version)::
   
       >>> c = Code()
       >>> c( Return(If(Local('a'), 42, 55)) )
       >>> dis(c.code())
         0           0 LOAD_FAST                0 (a)
                     3 JUMP_IF_FALSE            7 (to 13)
                     6 POP_TOP
                     7 LOAD_CONST               1 (42)
                    10 JUMP_FORWARD             4 (to 17)
               >>   13 POP_TOP
                    14 LOAD_CONST               2 (55)
               >>   17 RETURN_VALUE
   
   
   Note that ``If()`` does *not* do constant-folding on its condition; even if the
   condition is a constant, it will be tested at runtime.  This avoids issues with
   using mutable constants, e.g.::
   
       >>> c = Code()
       >>> c(If(Const([]), 42, 55))
       >>> dis(c.code())
         0           0 LOAD_CONST               1 ([])
                     3 JUMP_IF_FALSE            7 (to 13)
                     6 POP_TOP
                     7 LOAD_CONST               2 (42)
                    10 JUMP_FORWARD             4 (to 17)
               >>   13 POP_TOP
                    14 LOAD_CONST               3 (55)
   
   
 Labels and Jump Targets  Labels and Jump Targets
 -----------------------  -----------------------
   
Line 510 
Line 809 
     AssertionError: Label previously defined      AssertionError: Label previously defined
   
   
   N-Way Comparisons
   -----------------
   
   You can generate N-way comparisons using the ``Compare()`` node type::
   
       >>> from peak.util.assembler import Compare
   
       >>> c = Code()
       >>> c(Compare(Local('a'), [('<', Local('b'))]))
       >>> dis(c.code())
         0           0 LOAD_FAST                0 (a)
                     3 LOAD_FAST                1 (b)
                     6 COMPARE_OP               0 (<)
   
   3-way comparisons generate code that's a bit more complex.  Here's a three-way
   comparison (``a<b<c``)::
   
       >>> c = Code()
       >>> c.return_(Compare(Local('a'), [('<', Local('b')), ('<', Local('c'))]))
       >>> dis(c.code())
         0           0 LOAD_FAST                0 (a)
                     3 LOAD_FAST                1 (b)
                     6 DUP_TOP
                     7 ROT_THREE
                     8 COMPARE_OP               0 (<)
                    11 JUMP_IF_FALSE           10 (to 24)
                    14 POP_TOP
                    15 LOAD_FAST                2 (c)
                    18 COMPARE_OP               0 (<)
                    21 JUMP_FORWARD             2 (to 26)
               >>   24 ROT_TWO
                    25 POP_TOP
               >>   26 RETURN_VALUE
   
   And a four-way (``a<b>c!=d``)::
   
       >>> c = Code()
       >>> c.return_(
       ...     Compare( Local('a'), [
       ...         ('<', Local('b')), ('>', Local('c')), ('!=', Local('d'))
       ...     ])
       ... )
       >>> dis(c.code())
         0           0 LOAD_FAST                0 (a)
                     3 LOAD_FAST                1 (b)
                     6 DUP_TOP
                     7 ROT_THREE
                     8 COMPARE_OP               0 (<)
                    11 JUMP_IF_FALSE           22 (to 36)
                    14 POP_TOP
                    15 LOAD_FAST                2 (c)
                    18 DUP_TOP
                    19 ROT_THREE
                    20 COMPARE_OP               4 (>)
                    23 JUMP_IF_FALSE           10 (to 36)
                    26 POP_TOP
                    27 LOAD_FAST                3 (d)
                    30 COMPARE_OP               3 (!=)
                    33 JUMP_FORWARD             2 (to 38)
               >>   36 ROT_TWO
                    37 POP_TOP
               >>   38 RETURN_VALUE
   
   
   Sequence Unpacking
   ------------------
   
   The ``UnpackSequence`` node type takes a sequence of code generation targets,
   and generates an ``UNPACK_SEQUENCE`` of the correct length, followed by the
   targets::
   
       >>> from peak.util.assembler import UnpackSequence
       >>> c = Code()
       >>> c((1,2), UnpackSequence([LocalAssign('x'), LocalAssign('y')]))
       >>> dis(c.code())   # x, y = 1, 2
         0           0 LOAD_CONST               1 (1)
                     3 LOAD_CONST               2 (2)
                     6 BUILD_TUPLE              2
                     9 UNPACK_SEQUENCE          2
                    12 STORE_FAST               0 (x)
                    15 STORE_FAST               1 (y)
   
   
   Yield Statements
   ----------------
   
   The ``YieldStmt`` node type generates the necessary opcode(s) for a ``yield``
   statement, based on the target Python version.  (In Python 2.5+, a ``POP_TOP``
   must be generated after a ``YIELD_VALUE`` in order to create a yield statement,
   as opposed to a yield expression.)  It also sets the code flags needed to make
   the resulting code object a generator::
   
       >>> from peak.util.assembler import YieldStmt
       >>> c = Code()
       >>> c(YieldStmt(1), YieldStmt(2), Return(None))
       >>> list(eval(c.code()))
       [1, 2]
   
   
   
 Constant Detection and Folding  Constant Detection and Folding
 ==============================  ==============================
   
Line 535 
Line 934 
     >>> const_value(Local('x'))      >>> const_value(Local('x'))
     Traceback (most recent call last):      Traceback (most recent call last):
       ...        ...
     NotAConstant: <bound method str.Local of 'x'>      NotAConstant: Local('x')
   
 Tuples of constants are recursively replaced by constant tuples::  Tuples of constants are recursively replaced by constant tuples::
   
Line 550 
Line 949 
     >>> const_value( (1,Global('y')) )      >>> const_value( (1,Global('y')) )
     Traceback (most recent call last):      Traceback (most recent call last):
       ...        ...
     NotAConstant: <bound method str.Global of 'y'>      NotAConstant: Global('y')
   
 As do any types not previously described here::  As do any types not previously described here::
   
Line 576 
Line 975 
 ``Const`` node instead of a ``Call`` node::  ``Const`` node instead of a ``Call`` node::
   
     >>> Call( Const(type), [1] )      >>> Call( Const(type), [1] )
     <bound method type.Const of <type 'int'>>      Const(<type 'int'>)
   
 Thus, you can also take the ``const_value()`` of such calls::  Thus, you can also take the ``const_value()`` of such calls::
   
Line 587 
Line 986 
 passed in to another ``Call``::  passed in to another ``Call``::
   
     >>> Call(Const(type), [Call( Const(dict), [], [('x',27)] )])      >>> Call(Const(type), [Call( Const(dict), [], [('x',27)] )])
     <bound method type.Const of <type 'dict'>>      Const(<type 'dict'>)
   
 Notice that this folding takes place eagerly, during AST construction.  If you  Notice that this folding takes place eagerly, during AST construction.  If you
 want to implement delayed folding after constant propagation or variable  want to implement delayed folding after constant propagation or variable
Line 625 
Line 1024 
 ``globals()``, in other words.  ``globals()``, in other words.
   
   
   Logical And/Or
   --------------
   
   You can evaluate logical and/or expressions using the ``And`` and ``Or`` node
   types::
   
       >>> from peak.util.assembler import And, Or
   
       >>> c = Code()
       >>> c.return_( And([Local('x'), Local('y')]) )
       >>> dis(c.code())
         0           0 LOAD_FAST                0 (x)
                     3 JUMP_IF_FALSE            4 (to 10)
                     6 POP_TOP
                     7 LOAD_FAST                1 (y)
               >>   10 RETURN_VALUE
   
       >>> c = Code()
       >>> c.return_( Or([Local('x'), Local('y')]) )
       >>> dis(c.code())
         0           0 LOAD_FAST                0 (x)
                     3 JUMP_IF_TRUE             4 (to 10)
                     6 POP_TOP
                     7 LOAD_FAST                1 (y)
               >>   10 RETURN_VALUE
   
   
   True or false constants are folded automatically, avoiding code generation
   for intermediate values that will never be used in the result::
   
       >>> c = Code()
       >>> c.return_( And([1, 2, Local('y')]) )
       >>> dis(c.code())
         0           0 LOAD_FAST                0 (y)
                     3 RETURN_VALUE
   
       >>> c = Code()
       >>> c.return_( And([1, 2, Local('y'), 0]) )
       >>> dis(c.code())
         0           0 LOAD_FAST                0 (y)
                     3 JUMP_IF_FALSE            4 (to 10)
                     6 POP_TOP
                     7 LOAD_CONST               1 (0)
               >>   10 RETURN_VALUE
   
       >>> c = Code()
       >>> c.return_( Or([1, 2, Local('y')]) )
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (1)
                     3 RETURN_VALUE
   
       >>> c = Code()
       >>> c.return_( Or([False, Local('y'), 3]) )
       >>> dis(c.code())
         0           0 LOAD_FAST                0 (y)
                     3 JUMP_IF_TRUE             4 (to 10)
                     6 POP_TOP
                     7 LOAD_CONST               1 (3)
               >>   10 RETURN_VALUE
   
   
 Custom Code Generation  Custom Code Generation
 ======================  ======================
   
Line 646 
Line 1106 
 As you can see, the ``Code.DUP_TOP()`` is called on the code instance, causing  As you can see, the ``Code.DUP_TOP()`` is called on the code instance, causing
 a ``DUP_TOP`` opcode to be output.  This is sometimes a handy trick for  a ``DUP_TOP`` opcode to be output.  This is sometimes a handy trick for
 accessing values that are already on the stack.  More commonly, however, you'll  accessing values that are already on the stack.  More commonly, however, you'll
 want to implement more sophisticated callables, perhaps something like::  want to implement more sophisticated callables.
   
   To make it easy to create diverse target types, a ``nodetype()`` decorator is
   provided::
   
     >>> from peak.util.assembler import ast_curry      >>> from peak.util.assembler import nodetype
   
   It allows you to create code generation target types using functions.  Your
   function should take one or more arguments, with a ``code=None`` optional
   argument in the last position.  It should check whether ``code is None`` when
   called, and if so, return a tuple of the preceding arguments.  If ``code``
   is not ``None``, then it should do whatever code generating tasks are required.
   For example::
   
     >>> def TryFinally(block1, block2, code=None):      >>> def TryFinally(block1, block2, code=None):
     ...     if code is None:      ...     if code is None:
     ...         return ast_curry(TryFinally, block1, block2)      ...         return block1, block2
     ...     code(      ...     code(
     ...         Code.SETUP_FINALLY,      ...         Code.SETUP_FINALLY,
     ...             block1,      ...             block1,
Line 660 
Line 1130 
     ...             block2,      ...             block2,
     ...         Code.END_FINALLY      ...         Code.END_FINALLY
     ...     )      ...     )
       >>> TryFinally = nodetype()(TryFinally)
   
   Note: although the nodetype() generator can be used above the function
   definition in either Python 2.3 or 2.4, it cannot be done in a doctest under
   Python 2.3, so this document doesn't attempt to demonstrate that.  Under
   2.4, you would do something like this::
   
       @nodetype()
       def TryFinally(...):
   
   and code that needs to also work under 2.3 should do something like this::
   
       nodetype()
       def TryFinally(...):
   
   But to keep the examples here working with doctest, we'll be doing our
   ``nodetype()`` calls after the end of the function definitions, e.g.::
   
     >>> def ExprStmt(value, code=None):      >>> def ExprStmt(value, code=None):
     ...     if code is None:      ...     if code is None:
     ...         return ast_curry(ExprStmt, value)      ...         return value,
     ...     code( value, Code.POP_TOP )      ...     code( value, Code.POP_TOP )
       >>> ExprStmt = nodetype()(ExprStmt)
   
     >>> c = Code()      >>> c = Code()
     >>> c( TryFinally(ExprStmt(1), ExprStmt(2)) )      >>> c( TryFinally(ExprStmt(1), ExprStmt(2)) )
Line 678 
Line 1166 
                  14 POP_TOP                   14 POP_TOP
                  15 END_FINALLY                   15 END_FINALLY
   
   The ``nodetype()`` decorator is virtually identical to the ``struct()``
   decorator in the DecoratorTools package, except that it does not support
   ``*args``, does not create a field for the ``code`` argument, and generates a
   ``__call__()`` method that reinvokes the wrapped function to do the actual
   code generation.
   
   Among the benefits of this decorator are:
   
 The ``ast_curry()`` utility function returns an ``instancemethod`` chain that  * It gives your node types a great debugging format::
 binds the given arguments to the given function, creating a hashable and  
 comparable data structure -- a trivial sort of "AST node".  Just follow the  
 code pattern above, using a ``code=None`` final argument, and returning a  
 curried version of the function if ``code is None``.  Otherwise, your function  
 should simply do whatever is needed to "generate" the arguments.  
   
 (This is exactly the same pattern that ``peak.util.assembler`` uses internally  
 to implement ``Const``, ``Call``, ``Local``, and other wrapper functions.)  
   
 The ``ast_curry()`` utility function isn't quite perfect; due to a quirk of the  
 ``instancemethod`` type, it can't save arguments whose value is ``None``: if  
 you pass a ``None`` argument to ``ast_curry()``, it will be replaced with a  
 special ``nil`` object that tests as false, and generates a ``None`` constant  
 when code is generated for it.  If your function accepts any arguments that  
 might have a value of ``None``, you must correctly handle the cases where you  
 receive a value of ``nil`` (found in ``peak.util.assembler``) instead of  
 ``None``.  
   
 However, if you can use ``ast_curry()`` to generate your AST nodes, you will  
 have objects that are hashable and comparable by default, as long as none of  
 your child nodes are unhashable or incomparable.  This can be useful for  
 algorithms that require comparing AST subtrees, such as common subexpression  
 elimination.  
   
       >>> tf = TryFinally(ExprStmt(1), ExprStmt(2))
       >>> tf
       TryFinally(ExprStmt(1), ExprStmt(2))
   
   * It makes named fields accessible::
   
       >>> tf.block1
       ExprStmt(1)
   
       >>> tf.block2
       ExprStmt(2)
   
   * Hashing and comparison work as expected (handy for algorithms that require
     comparing or caching AST subtrees, such as common subexpression
     elimination)::
   
       >>> ExprStmt(1) == ExprStmt(1)
       True
       >>> ExprStmt(1) == ExprStmt(2)
       False
   
   
   Please see the `struct decorator documentation`_ for info on how to customize
   node types further.
   
   .. _struct decorator documentation: http://peak.telecommunity.com/DevCenter/DecoratorTools#the-struct-decorator
   
   Note: hashing only works if all the values you return in your argument tuple
   are hashable, so you should try to convert them if possible.  For example, if
   an argument accepts any sequence, you should probably convert it to a tuple
   before returning it.  Most of the examples in this document, and the node types
   supplied by ``peak.util.assembler`` itself do this.
   
   
 Constant Folding in Custom Targets  Constant Folding in Custom Targets
Line 711 
Line 1215 
   
 If you want to incorporate constant-folding into your AST nodes, you can do  If you want to incorporate constant-folding into your AST nodes, you can do
 so by checking for constant values and folding them at either construction  so by checking for constant values and folding them at either construction
 or code generation time.  For example, this ``And`` node type folds constants  or code generation time.  For example, this ``And`` node type (a simpler
 during code generation, by not generating unnecessary branches when it can  version of the one included in ``peak.util.assembler``) folds constants during
   code generation, by not generating unnecessary branches when it can
 prove which way a branch will go::  prove which way a branch will go::
   
     >>> from peak.util.assembler import NotAConstant      >>> from peak.util.assembler import NotAConstant
   
     >>> def And(values, code=None):      >>> def And(values, code=None):
     ...     if code is None:      ...     if code is None:
     ...         return ast_curry(And, tuple(values))      ...         return tuple(values),
     ...     end = Label()      ...     end = Label()
     ...     for value in values[:-1]:      ...     for value in values[:-1]:
     ...         try:      ...         try:
Line 730 
Line 1235 
     ...         else:       # and false constants end the chain right away      ...         else:       # and false constants end the chain right away
     ...             return code(value, end)      ...             return code(value, end)
     ...     code(values[-1], end)      ...     code(values[-1], end)
       >>> And = nodetype()(And)
   
     >>> c = Code()      >>> c = Code()
     >>> c.return_( And([1, 2]) )      >>> c.return_( And([1, 2]) )
Line 754 
Line 1260 
   
 The above example only folds constants at code generation time, however.  You  The above example only folds constants at code generation time, however.  You
 can also do constant folding at AST construction time, using the  can also do constant folding at AST construction time, using the
 ``folding_curry()`` function.  For example::  ``fold_args()`` function.  For example::
   
     >>> from peak.util.assembler import folding_curry      >>> from peak.util.assembler import fold_args
   
     >>> def Getattr(ob, name, code=None):      >>> def Getattr(ob, name, code=None):
     ...     try:      ...     try:
Line 764 
Line 1270 
     ...     except NotAConstant:      ...     except NotAConstant:
     ...         return Call(Const(getattr), [ob, name])      ...         return Call(Const(getattr), [ob, name])
     ...     if code is None:      ...     if code is None:
     ...         return folding_curry(Getattr, ob, name)      ...         return fold_args(Getattr, ob, name)
     ...     code(ob)      ...     code(ob)
     ...     code.LOAD_ATTR(name)      ...     code.LOAD_ATTR(name)
       >>> Getattr = nodetype()(Getattr)
   
     >>> const_value(Getattr(1, '__class__'))      >>> const_value(Getattr(1, '__class__'))
     <type 'int'>      <type 'int'>
   
 The ``folding_curry()`` function is essentially the same as ``ast_curry()``,  The ``fold_args()`` function tries to evaluate the node immediately, if all of
 unless all of the arguments it's given are recognized as constants.  In that  its arguments are constants, by creating a temporary ``Code`` object, and
 case, ``folding_curry()`` will create a temporary ``Code`` object, and run the  running the supplied function against it, then doing an ``eval()`` on the
 curried function against it, doing an ``eval()`` on the generated code and  generated code and wrapping the result in a ``Const``.  However, if any of the
 wrapping the result in a ``Const``.  arguments are non-constant, the original arguments (less the function) are
   returned. This causes a normal node instance to be created instead of a
   ``Const``.
   
 This isn't a very *fast* way of doing partial evaluation, but it makes it  This isn't a very *fast* way of doing partial evaluation, but it makes it
 really easy to define new code generation targets without writing custom  really easy to define new code generation targets without writing custom
 constant-folding code for each one.  Just use ``folding_curry()`` instead of  constant-folding code for each one.  Just ``return fold_args(ThisType, *args)``
 ``ast_curry()`` if you want your node constructor to be able to do eager  instead of ``return args``, if you want your node constructor to be able to do
 evaluation.  If you need to, you can check your parameters in order to decide  eager evaluation.  If you need to, you can check your parameters in order to
 whether to call ``ast_curry()`` or ``folding_curry()``; this is in fact how  decide whether to call ``fold_args()`` or not; this is in fact how ``Call``
 ``Call`` implements its ``fold`` argument and the suppression of folding when  implements its ``fold`` argument and the suppression of folding when
 the call has no arguments.  the call has no arguments.
   
   (By the way, this same ``Getattr`` node type is also available
   
   
 Setting the Code's Calling Signature  Setting the Code's Calling Signature
 ====================================  ====================================
Line 804 
Line 1315 
   
     >>> import inspect      >>> import inspect
   
     >>> inspect.getargspec(f1)      >>> tuple(inspect.getargspec(f1))
     (['a', 'b'], 'c', 'd', None)      (['a', 'b'], 'c', 'd', None)
   
     >>> inspect.getargspec(f2)      >>> tuple(inspect.getargspec(f2))
     (['a', 'b'], 'c', 'd', None)      (['a', 'b'], 'c', 'd', None)
   
 Note that these constructors do not copy any actual *code* from the code  Note that these constructors do not copy any actual *code* from the code
Line 819 
Line 1330 
     >>> c1 = Code.from_function(f1, copy_lineno=True)      >>> c1 = Code.from_function(f1, copy_lineno=True)
     >>> c1.co_firstlineno      >>> c1.co_firstlineno
     1      1
       >>> c1.co_filename is f1.func_code.co_filename
       True
   
 If you create a ``Code`` instance from a function that has nested positional  If you create a ``Code`` instance from a function that has nested positional
 arguments, the returned code object will include a prologue to unpack the  arguments, the returned code object will include a prologue to unpack the
Line 844 
Line 1357 
 unpacking process, and is designed so that the ``inspect`` module will  unpacking process, and is designed so that the ``inspect`` module will
 recognize it as an argument unpacking prologue::  recognize it as an argument unpacking prologue::
   
     >>> inspect.getargspec(f3)      >>> tuple(inspect.getargspec(f3))
     (['a', ['b', 'c'], ['d', ['e', 'f']]], None, None, None)      (['a', ['b', 'c'], ['d', ['e', 'f']]], None, None, None)
   
     >>> inspect.getargspec(f4)      >>> tuple(inspect.getargspec(f4))
     (['a', ['b', 'c'], ['d', ['e', 'f']]], None, None, None)      (['a', ['b', 'c'], ['d', ['e', 'f']]], None, None, None)
   
   You can also use the ``from_spec(name='<lambda>', args=(), var=None, kw=None)``
   classmethod to explicitly set a name and argument spec for a new code object::
   
       >>> c = Code.from_spec('a', ('b', ('c','d'), 'e'), 'f', 'g')
       >>> c.co_name
       'a'
   
       >>> c.co_varnames
       ['b', '.1', 'e', 'f', 'g', 'c', 'd']
   
       >>> c.co_argcount
       3
   
       >>> tuple(inspect.getargs(c.code()))
       (['b', ['c', 'd'], 'e'], 'f', 'g')
   
   
 Code Attributes  Code Attributes
 ===============  ===============
Line 874 
Line 1403 
     42      42
   
     >>> import inspect      >>> import inspect
     >>> inspect.getargspec(f)      >>> tuple(inspect.getargspec(f))
     (['a', 'b', 'c'], None, None, None)      (['a', 'b', 'c'], None, None, None)
   
 Although Python code objects want ``co_varnames`` to be a tuple, ``Code``  Although Python code objects want ``co_varnames`` to be a tuple, ``Code``
Line 1027 
Line 1556 
 code that might be unreachable.  For example, consider this ``If``  code that might be unreachable.  For example, consider this ``If``
 implementation::  implementation::
   
     >>> def Pass(code=None):  
     ...     if code is None:  
     ...         return Pass  
   
     >>> def If(cond, then, else_=Pass, code=None):      >>> def If(cond, then, else_=Pass, code=None):
     ...     if code is None:      ...     if code is None:
     ...         return ast_curry(If,cond,then,else_)      ...         return cond, then, else_
     ...     else_clause = Label()      ...     else_clause = Label()
     ...     end_if = Label()      ...     end_if = Label()
     ...     code(cond, else_clause.JUMP_IF_FALSE, Code.POP_TOP, then)      ...     code(cond, else_clause.JUMP_IF_FALSE, Code.POP_TOP, then)
     ...     code(end_if.JUMP_FORWARD, else_clause, Code.POP_TOP, else_)      ...     code(end_if.JUMP_FORWARD, else_clause, Code.POP_TOP, else_)
     ...     code(end_if)      ...     code(end_if)
       >>> If = nodetype()(If)
   
 It works okay if there's no dead code::  It works okay if there's no dead code::
   
     >>> c = Code()      >>> c = Code()
     >>> c( If(23, 42, 55) )      >>> c( If(Local('a'), 42, 55) )
     >>> dis(c.code())      >>> dis(c.code())
       0           0 LOAD_CONST               1 (23)        0           0 LOAD_FAST                0 (a)
                   3 JUMP_IF_FALSE            7 (to 13)                    3 JUMP_IF_FALSE            7 (to 13)
                   6 POP_TOP                    6 POP_TOP
                   7 LOAD_CONST               2 (42)                    7 LOAD_CONST               1 (42)
                  10 JUMP_FORWARD             4 (to 17)                   10 JUMP_FORWARD             4 (to 17)
             >>   13 POP_TOP              >>   13 POP_TOP
                  14 LOAD_CONST               3 (55)                   14 LOAD_CONST               2 (55)
   
 But it breaks if you end the "then" block with a return::  But it breaks if you end the "then" block with a return::
   
Line 1065 
Line 1591 
   
     >>> def If(cond, then, else_=Pass, code=None):      >>> def If(cond, then, else_=Pass, code=None):
     ...     if code is None:      ...     if code is None:
     ...         return ast_curry(If,cond,then,else_)      ...         return cond, then, else_
     ...     else_clause = Label()      ...     else_clause = Label()
     ...     end_if = Label()      ...     end_if = Label()
     ...     code(cond, else_clause.JUMP_IF_FALSE, Code.POP_TOP, then)      ...     code(cond, else_clause.JUMP_IF_FALSE, Code.POP_TOP, then)
     ...     if code.stack_size is not None:      ...     if code.stack_size is not None:
     ...         end_if.JUMP_FORWARD(code)      ...         end_if.JUMP_FORWARD(code)
     ...     code(else_clause, Code.POP_TOP, else_, end_if)      ...     code(else_clause, Code.POP_TOP, else_, end_if)
       >>> If = nodetype()(If)
   
 As you can see, the dead code is now eliminated::  As you can see, the dead code is now eliminated::
   
     >>> c = Code()      >>> c = Code()
     >>> c( If(23, Return(42), 55) )      >>> c( If(Local('a'), Return(42), 55) )
     >>> dis(c.code())      >>> dis(c.code())
       0           0 LOAD_CONST               1 (23)        0           0 LOAD_FAST                0 (a)
                   3 JUMP_IF_FALSE            5 (to 11)                    3 JUMP_IF_FALSE            5 (to 11)
                   6 POP_TOP                    6 POP_TOP
                   7 LOAD_CONST               2 (42)                    7 LOAD_CONST               1 (42)
                  10 RETURN_VALUE                   10 RETURN_VALUE
             >>   11 POP_TOP              >>   11 POP_TOP
                  12 LOAD_CONST               3 (55)                   12 LOAD_CONST               2 (55)
   
   
 Blocks, Loops, and Exception Handling  Blocks, Loops, and Exception Handling
Line 1209 
Line 1736 
             >>   10 LOAD_CONST               0 (None)              >>   10 LOAD_CONST               0 (None)
                  13 RETURN_VALUE                   13 RETURN_VALUE
   
 Labels have a ``POP_BLOCK`` attribute that you can pass in when generating  (Labels have a ``POP_BLOCK`` attribute that you can pass in when generating
 code.  code.)
   
   And, for generating typical try/except blocks, you can use the ``TryExcept``
   node type, which takes a body, a sequence of exception-type/handler pairs,
   and an optional "else" clause::
   
       >>> from peak.util.assembler import TryExcept
       >>> c = Code()
       >>> c.return_(
       ...     TryExcept(
       ...         Return(1),                                      # body
       ...         [(Const(KeyError),2), (Const(TypeError),3)],    # handlers
       ...         Return(4)                                       # else clause
       ...     )
       ... )
   
       >>> dis(c.code())
         0           0 SETUP_EXCEPT             8 (to 11)
                     3 LOAD_CONST               1 (1)
                     6 RETURN_VALUE
                     7 POP_BLOCK
                     8 JUMP_FORWARD            43 (to 54)
               >>   11 DUP_TOP
                    12 LOAD_CONST               2 (<...exceptions.KeyError...>)
                    15 COMPARE_OP              10 (exception match)
                    18 JUMP_IF_FALSE           10 (to 31)
                    21 POP_TOP
                    22 POP_TOP
                    23 POP_TOP
                    24 POP_TOP
                    25 LOAD_CONST               3 (2)
                    28 JUMP_FORWARD            27 (to 58)
               >>   31 POP_TOP
                    32 DUP_TOP
                    33 LOAD_CONST               4 (<...exceptions.TypeError...>)
                    36 COMPARE_OP              10 (exception match)
                    39 JUMP_IF_FALSE           10 (to 52)
                    42 POP_TOP
                    43 POP_TOP
                    44 POP_TOP
                    45 POP_TOP
                    46 LOAD_CONST               5 (3)
                    49 JUMP_FORWARD             6 (to 58)
               >>   52 POP_TOP
                    53 END_FINALLY
               >>   54 LOAD_CONST               6 (4)
                    57 RETURN_VALUE
               >>   58 RETURN_VALUE
   
   
 Try/Finally Blocks  Try/Finally Blocks
Line 1247 
Line 1821 
 adjusts the maximum expected stack size to accomodate up to three values being  adjusts the maximum expected stack size to accomodate up to three values being
 put on the stack by the Python interpreter for exception handling.  put on the stack by the Python interpreter for exception handling.
   
   For your convenience, the ``TryFinally`` node type can also be used to generate
   try/finally blocks::
   
       >>> from peak.util.assembler import TryFinally
       >>> c = Code()
       >>> c( TryFinally(ExprStmt(1), ExprStmt(2)) )
       >>> dis(c.code())
         0           0 SETUP_FINALLY            8 (to 11)
                     3 LOAD_CONST               1 (1)
                     6 POP_TOP
                     7 POP_BLOCK
                     8 LOAD_CONST               0 (None)
               >>   11 LOAD_CONST               2 (2)
                    14 POP_TOP
                    15 END_FINALLY
   
   
 Loops  Loops
 -----  -----
Line 1358 
Line 1948 
             >>   19 END_FINALLY              >>   19 END_FINALLY
                  20 POP_BLOCK                   20 POP_BLOCK
   
   ``for`` Loops
   -------------
   
   There is a ``For()`` node type available for generating simple loops (without
   break/continue support).  It takes an iterable expression, an assignment
   clause, and a loop body::
   
       >>> from peak.util.assembler import For
       >>> y = Call(Const(range), (3,))
       >>> x = LocalAssign('x')
       >>> body = Suite([Local('x'), Code.PRINT_EXPR])
   
       >>> c = Code()
       >>> c(For(y, x, body))  # for x in range(3): print x
       >>> c.return_()
       >>> dis(c.code())
         0           0 LOAD_CONST               1 ([0, 1, 2])
                     3 GET_ITER
               >>    4 FOR_ITER                10 (to 17)
                     7 STORE_FAST               0 (x)
                    10 LOAD_FAST                0 (x)
                    13 PRINT_EXPR
                    14 JUMP_ABSOLUTE            4
               >>   17 LOAD_CONST               0 (None)
                    20 RETURN_VALUE
   
   The arguments are given in execution order: first the "in" value of the loop,
   then the assignment to a loop variable, and finally the body of the loop.  The
   distinction between the assignment and body, however, is only for clarity and
   convenience (to avoid needing to glue the assignment to the body with a
   ``Suite``).  If you already have a suite or only need one node for the entire
   loop body, you can do the same thing with only two arguments::
   
       >>> c = Code()
       >>> c(For(y, Code.PRINT_EXPR))
       >>> c.return_()
       >>> dis(c.code())
         0           0 LOAD_CONST               1 ([0, 1, 2])
                     3 GET_ITER
               >>    4 FOR_ITER                 4 (to 11)
                     7 PRINT_EXPR
                     8 JUMP_ABSOLUTE            4
               >>   11 LOAD_CONST               0 (None)
                    14 RETURN_VALUE
   
   Notice, by the way, that ``For()`` does NOT set up a loop block for you, so if
   you want to be able to use break and continue, you'll need to wrap the loop in
   a labelled SETUP_LOOP/POP_BLOCK pair, as described in the preceding sections.
   
   
   List Comprehensions
   -------------------
   
   In order to generate correct list comprehension code for the target Python
   version, you must use the ``ListComp()`` and ``LCAppend()`` node types.  This
   is because Python versions 2.4 and up store the list being built in a temporary
   variable, and use a special ``LIST_APPEND`` opcode to append values, while 2.3
   stores the list's ``append()`` method in the temporary variable, and calls it
   to append values.
   
   The ``ListComp()`` node wraps a code body (usually a ``For()`` loop) and
   manages the creation and destruction of a temporary variable (e.g. ``_[1]``,
   ``_[2]``, etc.).  The ``LCAppend()`` node type wraps a value or expression to
   be appended to the innermost active ``ListComp()`` in progress::
   
       >>> from peak.util.assembler import ListComp, LCAppend
       >>> c = Code()
       >>> simple = ListComp(For(y, x, LCAppend(Local('x'))))
       >>> c.return_(simple)
       >>> eval(c.code())
       [0, 1, 2]
   
       >>> c = Code()
       >>> c.return_(ListComp(For(y, x, LCAppend(simple))))
       >>> eval(c.code())
       [[0, 1, 2], [0, 1, 2], [0, 1, 2]]
   
   
   Closures and Nested Functions
   =============================
   
   Free and Cell Variables
   -----------------------
   
   To implement closures and nested scopes, your code objects must use "free" or
   "cell" variables in place of regular "fast locals".  A "free" variable is one
   that is defined in an outer scope, and a "cell" variable is one that's defined
   in the current scope, but will also be used by nested functions.
   
   The simplest way to set up free or cell variables is to use a code object's
   ``makefree(names)`` and ``makecells(names)`` methods::
   
       >>> c = Code()
       >>> c.co_cellvars
       ()
       >>> c.co_freevars
       ()
   
       >>> c.makefree(['x', 'y'])
       >>> c.makecells(['z'])
   
       >>> c.co_cellvars
       ('z',)
       >>> c.co_freevars
       ('x', 'y')
   
   When a name has been defined as a free or cell variable, the ``_DEREF`` opcode
   variants are used to generate ``Local()`` and ``LocalAssign()`` nodes::
   
       >>> c((Local('x'), Local('y')), LocalAssign('z'))
       >>> dis(c.code())
         0           0 LOAD_DEREF               1 (x)
                     3 LOAD_DEREF               2 (y)
                     6 BUILD_TUPLE              2
                     9 STORE_DEREF              0 (z)
   
   If you have already written code in a code object that operates on the relevant
   locals, the code is retroactively patched to use the ``_DEREF`` opcodes::
   
       >>> c = Code()
       >>> c((Local('x'), Local('y')), LocalAssign('z'))
       >>> dis(c.code())
         0           0 LOAD_FAST                0 (x)
                     3 LOAD_FAST                1 (y)
                     6 BUILD_TUPLE              2
                     9 STORE_FAST               2 (z)
   
       >>> c.makefree(['x', 'y'])
       >>> c.makecells(['z'])
   
       >>> dis(c.code())
         0           0 LOAD_DEREF               1 (x)
                     3 LOAD_DEREF               2 (y)
                     6 BUILD_TUPLE              2
                     9 STORE_DEREF              0 (z)
   
   This means that you can defer the decision of which locals are free/cell
   variables until the code is ready to be generated.  In fact, by passing in
   a "parent" code object to the ``.code()`` method, you can get BytecodeAssembler
   to automatically call ``makefree()`` and ``makecells()`` for the correct
   variable names in the child and parent code objects, as we'll see in the next
   section.
   
   
   Nested Code Objects
   -------------------
   
   To create a code object for use in a nested scope, you can use the parent code
   object's ``nested()`` method.  It works just like the ``from_spec()``
   classmethod, except that the ``co_filename`` of the parent is copied to the
   child::
   
       >>> p = Code()
       >>> p.co_filename = 'testname'
   
       >>> c = p.nested('sub', ['a','b'], 'c', 'd')
   
       >>> c.co_name
       'sub'
   
       >>> c.co_filename
       'testname'
   
       >>> tuple(inspect.getargs(c.code(p)))
       (['a', 'b'], 'c', 'd')
   
   Notice that you must pass the parent code object to the child's ``.code()``
   method to ensure that free/cell variables are properly set up.  When the
   ``code()`` method is given another code object as a parameter, it automatically
   converts any locally-read (but not written) to "free" variables in the child
   code, and ensures that those same variables become "cell" variables in the
   supplied parent code object::
   
       >>> p.LOAD_CONST(42)
       >>> p(LocalAssign('a'))
       >>> dis(p.code())
         0           0 LOAD_CONST               1 (42)
                     3 STORE_FAST               0 (a)
   
       >>> c = p.nested()
       >>> c(Local('a'))
   
       >>> dis(c.code(p))
         0           0 LOAD_DEREF               0 (a)
   
       >>> dis(p.code())
         0           0 LOAD_CONST               1 (42)
                     3 STORE_DEREF              0 (a)
   
   Notice that the ``STORE_FAST`` in the parent code object was automatically
   patched to a ``STORE_DEREF``, with an updated offset if applicable.  Any
   future use of ``Local('a')`` or ``LocalAssign('a')`` in the parent or child
   code objects will now refer to the free/cell variable, rather than the "local"
   variable::
   
       >>> p(Local('a'))
       >>> dis(p.code())
         0           0 LOAD_CONST               1 (42)
                     3 STORE_DEREF              0 (a)
                     6 LOAD_DEREF               0 (a)
   
       >>> c(LocalAssign('a'))
       >>> dis(c.code(p))
         0           0 LOAD_DEREF               0 (a)
                     3 STORE_DEREF              0 (a)
   
   
   ``Function()``
   --------------
   
   The ``Function(body, name='<lambda>', args=(), var=None, kw=None, defaults=())``
   node type creates a function object from the specified body and the optional
   name, argument specs, and defaults.  The ``Function()`` node generates code to
   create the function object with the appropriate defaults and closure (if
   applicable), and any needed free/cell variables are automatically set up in the
   parent and child code objects.  The newly generated function will be on top of
   the stack at the end of the generated code::
   
       >>> from peak.util.assembler import Function
       >>> c = Code()
       >>> c.co_filename = '<string>'
       >>> c.return_(Function(Return(Local('a')), 'f', ['a'], defaults=[42]))
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (42)
                     3 LOAD_CONST               2 (<... f ..., file "<string>", line -1>)
                     6 MAKE_FUNCTION            1
                     9 RETURN_VALUE
   
   Now that we've generated the code for a function returning a function, let's
   run it, to get the function we defined::
   
       >>> f = eval(c.code())
       >>> f
       <function f at ...>
   
       >>> tuple(inspect.getargspec(f))
       (['a'], None, None, (42,))
   
       >>> f()
       42
   
       >>> f(99)
       99
   
   Now let's create a doubly nested function, with some extras::
   
       >>> c = Code()
       >>> c.co_filename = '<string>'
       >>> c.return_(
       ...     Function(Return(Function(Return(Local('a')))),
       ...     'f', ['a', 'b'], 'c', 'd', [99, 66])
       ... )
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (99)
                     3 LOAD_CONST               2 (66)
                     6 LOAD_CONST               3 (<... f ..., file "<string>", line -1>)
                     9 MAKE_FUNCTION            2
                    12 RETURN_VALUE
   
       >>> f = eval(c.code())
       >>> f
       <function f at ...>
   
       >>> tuple(inspect.getargspec(f))
       (['a', 'b'], 'c', 'd', (99, 66))
   
       >>> dis(f)
         0           0 LOAD_CLOSURE             0 (a)
                     ... LOAD_CONST               1 (<... <lambda> ..., file "<string>", line -1>)
                     ... MAKE_CLOSURE             0
                     ... RETURN_VALUE
   
       >>> dis(f())
         0           0 LOAD_DEREF               0 (a)
                     3 RETURN_VALUE
   
       >>> f(42)()
       42
   
       >>> f()()
       99
   
   As you can see, ``Function()`` not only takes care of setting up free/cell
   variables in all the relevant scopes, it also chooses whether to use
   ``MAKE_FUNCTION`` or ``MAKE_CLOSURE``, and generates code for the defaults.
   
   (Note, by the way, that the `defaults` argument should be a sequence of
   generatable expressions; in the examples here, we used numbers, but they could
   have been arbitrary expression nodes.)
   
   
 ----------------------  ----------------------
 Internals and Doctests  Internals and Doctests
Line 1380 
Line 2260 
     >>> simple_code(1,1).co_stacksize      >>> simple_code(1,1).co_stacksize
     1      1
   
     >>> dis(simple_code(13,414))    # FAILURE EXPECTED IN PYTHON 2.3      >>> dis(simple_code(13,414))
      13           0 LOAD_CONST               0 (None)       13           0 LOAD_CONST               0 (None)
     414           3 RETURN_VALUE      414           3 RETURN_VALUE
   
Line 1393 
Line 2273 
     >>> simple_code(13,14,100).co_stacksize      >>> simple_code(13,14,100).co_stacksize
     100      100
   
     >>> dis(simple_code(13,572,120))    # FAILURE EXPECTED IN Python 2.3      >>> dis(simple_code(13,572,120))
      13           0 LOAD_CONST               0 (None)       13           0 LOAD_CONST               0 (None)
                   3 LOAD_CONST               0 (None)                    3 LOAD_CONST               0 (None)
     ...      ...
Line 1452 
Line 2332 
                   3 LOAD_ATTR                1 (bar)                    3 LOAD_ATTR                1 (bar)
                   6 DELETE_FAST              0 (baz)                    6 DELETE_FAST              0 (baz)
   
   Code iteration::
   
       >>> c.DUP_TOP()
       >>> c.return_(Code.POP_TOP)
       >>> list(c) == [
       ...     (0, op.LOAD_GLOBAL, 0),
       ...     (3, op.LOAD_ATTR, 1),
       ...     (6, op.DELETE_FAST, 0),
       ...     (9, op.DUP_TOP, None),
       ...     (10, op.POP_TOP, None),
       ...     (11, op.RETURN_VALUE, None)
       ... ]
       True
   
   Code patching::
   
       >>> c = Code()
       >>> c.LOAD_CONST(42)
       >>> c.STORE_FAST('x')
       >>> c.LOAD_FAST('x')
       >>> c.DELETE_FAST('x')
       >>> c.RETURN_VALUE()
   
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (42)
                     3 STORE_FAST               0 (x)
                     6 LOAD_FAST                0 (x)
                     9 DELETE_FAST              0 (x)
                    12 RETURN_VALUE
   
   
       >>> c.co_varnames
       ['x']
       >>> c.co_varnames.append('y')
   
       >>> c._patch(
       ...     {op.LOAD_FAST:  op.LOAD_FAST,
       ...      op.STORE_FAST: op.STORE_FAST,
       ...      op.DELETE_FAST: op.DELETE_FAST},
       ...     {0: 1}
       ... )
   
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (42)
                     3 STORE_FAST               1 (y)
                     6 LOAD_FAST                1 (y)
                     9 DELETE_FAST              1 (y)
                    12 RETURN_VALUE
   
       >>> c._patch({op.RETURN_VALUE: op.POP_TOP})
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (42)
                     3 STORE_FAST               1 (y)
                     6 LOAD_FAST                1 (y)
                     9 DELETE_FAST              1 (y)
                    12 POP_TOP
   
   Converting locals to free/cell vars::
   
       >>> c = Code()
       >>> c.LOAD_CONST(42)
       >>> c.STORE_FAST('x')
       >>> c.LOAD_FAST('x')
   
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (42)
                     3 STORE_FAST               0 (x)
                     6 LOAD_FAST                0 (x)
   
       >>> c.co_freevars = 'y', 'x'
       >>> c.co_cellvars = 'z',
   
       >>> c._locals_to_cells()
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (42)
                     3 STORE_DEREF              2 (x)
                     6 LOAD_DEREF               2 (x)
   
       >>> c.DELETE_FAST('x')
       >>> c._locals_to_cells()
       Traceback (most recent call last):
         ...
       AssertionError: Can't delete local 'x' used in nested scope
   
       >>> c = Code()
       >>> c.LOAD_CONST(42)
       >>> c.STORE_FAST('x')
       >>> c.LOAD_FAST('x')
   
       >>> c.co_freevars
       ()
       >>> c.makefree(['x'])
       >>> c.co_freevars
       ('x',)
   
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (42)
                     3 STORE_DEREF              0 (x)
                     6 LOAD_DEREF               0 (x)
   
       >>> c = Code()
       >>> c.LOAD_CONST(42)
       >>> c.STORE_FAST('x')
       >>> c.LOAD_FAST('x')
       >>> c.makecells(['x'])
       >>> c.co_freevars
       ()
       >>> c.co_cellvars
       ('x',)
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (42)
                     3 STORE_DEREF              0 (x)
                     6 LOAD_DEREF               0 (x)
   
       >>> c = Code()
       >>> c.LOAD_CONST(42)
       >>> c.STORE_FAST('x')
       >>> c.LOAD_FAST('x')
       >>> c.makefree('x')
       >>> c.makecells(['y'])
       >>> c.co_freevars
       ('x',)
       >>> c.co_cellvars
       ('y',)
       >>> dis(c.code())
         0           0 LOAD_CONST               1 (42)
                     3 STORE_DEREF              1 (x)
                     6 LOAD_DEREF               1 (x)
   
       >>> c = Code()
       >>> c.co_flags &= ~op.CO_OPTIMIZED
       >>> c.makecells(['q'])
       Traceback (most recent call last):
         ...
       AssertionError: Can't use cellvars in unoptimized scope
   
   
   
   Auto-free promotion with code parent:
   
       >>> p = Code()
       >>> c = Code()
       >>> c.LOAD_FAST('x')
       >>> dis(c.code(p))
         0           0 LOAD_DEREF               0 (x)
       >>> p.co_cellvars
       ('x',)
   
       >>> p = Code()
       >>> c = Code.from_function(lambda x,y,z=2: None)
       >>> c.LOAD_FAST('x')
       >>> c.LOAD_FAST('y')
       >>> c.LOAD_FAST('z')
   
       >>> dis(c.code(p))
         0           0 LOAD_FAST                0 (x)
                     3 LOAD_FAST                1 (y)
                     6 LOAD_FAST                2 (z)
       >>> p.co_cellvars
       ()
   
       >>> c.LOAD_FAST('q')
       >>> dis(c.code(p))
         0           0 LOAD_FAST                0 (x)
                     3 LOAD_FAST                1 (y)
                     6 LOAD_FAST                2 (z)
                     9 LOAD_DEREF               0 (q)
       >>> p.co_cellvars
       ('q',)
   
       >>> p = Code()
       >>> c = Code.from_function(lambda x,*y,**z: None)
       >>> c.LOAD_FAST('q')
       >>> c.LOAD_FAST('x')
       >>> c.LOAD_FAST('y')
       >>> c.LOAD_FAST('z')
       >>> dis(c.code(p))
         0           0 LOAD_DEREF               0 (q)
                     3 LOAD_FAST                0 (x)
                     6 LOAD_FAST                1 (y)
                     9 LOAD_FAST                2 (z)
       >>> p.co_cellvars
       ('q',)
   
       >>> p = Code()
       >>> c = Code.from_function(lambda x,*y: None)
       >>> c.LOAD_FAST('x')
       >>> c.LOAD_FAST('y')
       >>> c.LOAD_FAST('z')
       >>> dis(c.code(p))
         0           0 LOAD_FAST                0 (x)
                     3 LOAD_FAST                1 (y)
                     6 LOAD_DEREF               0 (z)
       >>> p.co_cellvars
       ('z',)
   
       >>> p = Code()
       >>> c = Code.from_function(lambda x,**y: None)
       >>> c.LOAD_FAST('x')
       >>> c.LOAD_FAST('y')
       >>> c.LOAD_FAST('z')
       >>> dis(c.code(p))
         0           0 LOAD_FAST                0 (x)
                     3 LOAD_FAST                1 (y)
                     6 LOAD_DEREF               0 (z)
       >>> p.co_cellvars
       ('z',)
   
   
 Stack tracking on jumps::  Stack tracking on jumps::
   
Line 1486 
Line 2574 
       ...        ...
     AssertionError: Stack level mismatch: actual=1 expected=0      AssertionError: Stack level mismatch: actual=1 expected=0
   
       >>> from peak.util.assembler import For
       >>> c = Code()
       >>> c(For((), Code.POP_TOP, Pass))
       >>> c.return_()
       >>> dis(c.code())
         0           0 BUILD_TUPLE              0
                     3 GET_ITER
               >>    4 FOR_ITER                 4 (to 11)
                     7 POP_TOP
                     8 JUMP_ABSOLUTE            4
               >>   11 LOAD_CONST               0 (None)
                    14 RETURN_VALUE
   
       >>> c.stack_history
       [0, 1, 1, 1, 1, 2, 2, 2, 1, None, None, 0, 1, 1, 1]
   
   
   Yield value::
   
       >>> import sys
       >>> from peak.util.assembler import CO_GENERATOR
       >>> c = Code()
       >>> c.co_flags & CO_GENERATOR
       0
       >>> c(42, Code.YIELD_VALUE)
       >>> c.stack_size == int(sys.version>='2.5')
       True
       >>> (c.co_flags & CO_GENERATOR) == CO_GENERATOR
       True
   
   
   
Line 1646 
Line 2762 
       ...        ...
     AssertionError: Stack underflow      AssertionError: Stack underflow
   
     >>> c.LOAD_CONST(1)      >>> c = Code()
     >>> c.LOAD_CONST(2) # simulate being a function      >>> c.LOAD_CONST(1) # closure
     >>> c.MAKE_CLOSURE(1, 0)      >>> if sys.version>='2.5': c.BUILD_TUPLE(1)
       >>> c.LOAD_CONST(2) # default
       >>> c.LOAD_CONST(3) # simulate being a function
       >>> c.MAKE_CLOSURE(1, 1)
     >>> c.stack_size      >>> c.stack_size
     1      1
   
     >>> c = Code()      >>> c = Code()
     >>> c.LOAD_CONST(1)      >>> c.LOAD_CONST(1)
     >>> c.LOAD_CONST(2)      >>> c.LOAD_CONST(2)
       >>> if sys.version>='2.5': c.BUILD_TUPLE(2)
     >>> c.LOAD_CONST(3) # simulate being a function      >>> c.LOAD_CONST(3) # simulate being a function
     >>> c.MAKE_CLOSURE(1, 1)      >>> c.MAKE_CLOSURE(0, 2)
     >>> c.stack_size      >>> c.stack_size
     1      1
   
   
   
 Labels and backpatching forward references::  Labels and backpatching forward references::
   
     >>> c = Code()      >>> c = Code()
Line 1741 
Line 2862 
       0           0 LOAD_CONST               1 ({'x': 1})        0           0 LOAD_CONST               1 ({'x': 1})
                   3 RETURN_VALUE                    3 RETURN_VALUE
   
   Try/Except stack level tracking::
   
       >>> def class_or_type_of(expr):
       ...     return Suite([expr, TryExcept(
       ...         Suite([Getattr(Code.DUP_TOP, '__class__'), Code.ROT_TWO]),
       ...         [(Const(AttributeError), Call(Const(type), (Code.ROT_TWO,)))]
       ...     )])
   
       >>> def type_or_class(x): pass
       >>> c = Code.from_function(type_or_class)
       >>> c.return_(class_or_type_of(Local('x')))
       >>> dis(c.code())
         0           0 LOAD_FAST                0 (x)
                     3 SETUP_EXCEPT             9 (to 15)
                     6 DUP_TOP
                     7 LOAD_ATTR                0 (__class__)
                    10 ROT_TWO
                    11 POP_BLOCK
                    12 JUMP_FORWARD            26 (to 41)
               >>   15 DUP_TOP
                    16 LOAD_CONST               1 (<...exceptions.AttributeError...>)
                    19 COMPARE_OP              10 (exception match)
                    22 JUMP_IF_FALSE           14 (to 39)
                    25 POP_TOP
                    26 POP_TOP
                    27 POP_TOP
                    28 POP_TOP
                    29 LOAD_CONST               2 (<type 'type'>)
                    32 ROT_TWO
                    33 CALL_FUNCTION            1
                    36 JUMP_FORWARD             2 (to 41)
               >>   39 POP_TOP
                    40 END_FINALLY
               >>   41 RETURN_VALUE
   
       >>> type_or_class.func_code = c.code()
       >>> type_or_class(23)
       <type 'int'>
   
   
   
   
   
 Demo: "Computed Goto"/"Switch Statement"  Demo: "Computed Goto"/"Switch Statement"
Line 1751 
Line 2913 
   
     >>> from peak.util.assembler import LOAD_CONST, POP_BLOCK      >>> from peak.util.assembler import LOAD_CONST, POP_BLOCK
   
     >>> def Pass(code=None):  
     ...     if code is None:  
     ...         return Pass  
   
     >>> import sys      >>> import sys
     >>> WHY_CONTINUE = {'2.3':5, '2.4':32, '2.5':32}[sys.version[:3]]      >>> WHY_CONTINUE = {'2.3':5}.get(sys.version[:3], 32)
   
     >>> def Switch(expr, cases, default=Pass, code=None):      >>> def Switch(expr, cases, default=Pass, code=None):
     ...     if code is None:      ...     if code is None:
     ...         return ast_curry(Switch, expr, tuple(cases), default)      ...         return expr, tuple(cases), default
     ...      ...
     ...     d = {}      ...     d = {}
     ...     else_block  = Label()      ...     else_block  = Label()
Line 1789 
Line 2947 
     ...             Code.POP_BLOCK,      ...             Code.POP_BLOCK,
     ...         end_switch      ...         end_switch
     ...     )      ...     )
       >>> Switch = nodetype()(Switch)
   
     >>> c = Code()      >>> c = Code()
     >>> c.co_argcount=1      >>> c.co_argcount=1
Line 1826 
Line 2985 
 TODO  TODO
 ====  ====
   
 * AST introspection  
     * ast_type(node): called function, Const, or node.__class__  
       * tuples are Const if their contents are; no other types are Const  
     * ast_children(node): tuple of argument values for curried types, const value,  
       or empty tuple.  If node is a tuple, the value must be flattened.  
     * is_const(node): ast_type(node) is Const  
   
 * Inline builtins (getattr, operator.getitem, etc.) to opcodes  
     * Getattr/Op/Unary("symbol", arg1 [, arg2]) node types -> Call() if folding  
     * Call() translates functions back to Ops if inlining  
   
 * Pretty printing and short-naming of ASTs  
   
 * Test NAME vs. FAST operators flag checks/sets  * Test NAME vs. FAST operators flag checks/sets
   
 * Test code flags generation/cloning  * Test code flags generation/cloning
   
   * Exhaustive tests of all opcodes' stack history effects
   
   * Test wide jumps and wide argument generation in general

Generate output suitable for use with a patch program
Legend:
Removed from v.2205  
changed lines
  Added in v.2620
cvs-admin@eby-sarna.com

Powered by ViewCVS 1.0-dev

ViewCVS and CVS Help