options / options / options.py


import sys

from stuf import orderedstuf
from chainstuf import chainstuf

    from options.nulltype import NullType
except ImportError:
    from nulltype import NullType
    from options.funclike import *
except ImportError:
    from funclike import *
# Do not understand why we need these try/except and old-style
# relative import fallbacks. But if they're not there, tests
# fail. With them there, everything works great. So... ?

Unset      = NullType('Unset')
Prohibited = NullType('Prohibited')

def attrs(m, first=[], underscores=False):
    Given a mapping m, return a string listing its values in a
    key=value format. Items with underscores are, by default, not
    listed. If you want some things listed first, include them in
    the list first.
    keys = first[:]
    for k in m.keys():
        if not underscores and k.startswith('_'):
        if k not in first:
    return ', '.join([ "{0}={1}".format(k, repr(m[k])) for k in keys ])

class Options(orderedstuf):
    Options handler.
    def __init__(self, *args, **kwargs):
        orderedstuf.__init__(self, *args, **kwargs)
        self._magic = {}
    def __repr__(self):
        return "{0}({1})".format(self.__class__.__name__, attrs(self))
    def set(self, **kwargs):
        # To set values at the base options level, create a temporary next level,
        # which will have the magical option interpretation. Then copy the resulting
        # values here. Do in original ordering.
        oc = OptionsChain(self, kwargs)
        for key in self.keys():
            self[key] = oc[key] 
    def push(self, kwargs):
        Create the next layer down. Intended for instances to call during
        return OptionsChain(self, kwargs)
        # NB the implicit addflat() call is gone

    def add(self, **kwargs):
        Create the next layer down. Like ``push()``, but accepts full kwargs
        not just a dict. Intended for subclasses to call when defining their
        own class options. Not for instances to call.
        child = OptionsChain(self, {})
        for key, value in kwargs.items():
            child[key] = value
        return child
    def magic(self, **kwargs):
        Set some options as having 'magical' update properties. In a sense, this
        is like Python ``properties`` that have a setter.  NB no magical
        processing is done to the base Options. These are assumed to have whatever
        adjustments are needed when they are originally set.
        self.setdefault('_magic', {})
        for k, v in kwargs.items():
            self._magic[k] = real_func(v)
    def magical(self, key):
        Instance based decorator, specifying a function in the using module
        as a magical function. Note that the magical methods will be called
        with a self of None. 
        def my_decorator(func):
            self._magic[key] = func
            return func
        return my_decorator
        # not sure why we take such care to get the order right for Options, because the
        # dict handed to Options.__init__ is not in order!
        # should there be a leftovers_ok option that raises an error on push()
        # if there are leftovers?    

class OptionsChain(chainstuf):
    def __init__(self, bottom, kwargs):
        Create an OptionsChain, pushing one level down.
        chainstuf.__init__(self, bottom)
        processed = self._process(bottom, kwargs)
        self.maps = [ processed, bottom ]
    def _magicalized(self, key, value):
        Get the magically processed value for a single key value pair.
        If there is no magical processing to be done, just returns value.
        magicfn = self._magic.get(key, None)
        if magicfn is None:
            return value
        argcount = func_code(magicfn).co_argcount
        if argcount == 1:
            return magicfn(value)
        elif argcount == 2:
            return magicfn(value, self)
        elif argcount == 3:
            return magicfn(None, value, self)
            raise ValueError('magic function should have 1-3 arguments, not {0}'.format(argcount))

    def _process(self, base, kwargs):
        Given kwargs, removes any key:value pairs corresponding to this set of
        options. Those pairs are interpreted according to'paramater
        interpretation magic' if needed, then returned as dict. Any key:value
        pairs remaining in kwargs are not options related to this class, and may
        be used for other purposes.

        opts = {}
        for key, value in kwargs.items():
            if key in base:
                opts[key] = self._magicalized(key, value)
        # NB base identical to self when called from set(), but not when called
        # from __init__()
        # empty kwargs of 'taken' options
        for key in opts:
            del kwargs[key]

        return opts
    def __repr__(self):
        Get repr() of OptionsChain. Dig down to find earliest ancestor, which
        contains the right ordering of keys.
        grandpa = self.maps[-1]
        n_layers  = len(self.maps)
        while not isinstance(grandpa, Options):
            n_layers += len(grandpa.maps) - 1
            grandpa = grandpa.maps[-1]
        guts = attrs(self, first=list(grandpa.keys()), underscores=True)
        return "{0}({1} layers: {2})".format(self.__class__.__name__, n_layers, guts)
    def push(self, kwargs):
        Create the next layer down. Intended for instances to call
        individual method invocations, where ``self.options`` is already
        an OptionChain. 
        return OptionsChain(self, kwargs)
        # NB Implicit addflat() has been removed

    def addflat(self, args, keys):
        Sometimes kwargs aren't the most elegant way to provide options. In those
        cases, this routine helps map flat args to kwargs. Provide the actual args,
        followed by keys in the order in which they should be consumed. There can
        be more keys than args, but not the other way around. Returns a list of
        keys used (from which one can also determine if a key was not used).
        if len(args) > len(keys):
            raise ValueError('More args than keys not allowed')
        additional = dict(zip(keys, args))
        keys_used = list(additional.keys())
        return keys_used 

    def add(self, **kwargs):
        Create the next layer down. Like ``push()``, but accepts full kwargs
        not just a dict. Intended for subclasses to call when defining their
        own class options. Not for instances to call.
        child = OptionsChain(self, {})
        for key, value in kwargs.items():
            child[key] = value
        return child
    def set(self, **kwargs):
        newopts = self._process(self, kwargs)
        for k, v in newopts.items():
            if v is Unset:
                del self.maps[0][k]
            elif self[k] is Prohibited:
                raise KeyError("changes to '{0}' are prohibited".format(k))
                self.maps[0][k] = v
    # TBD Need to examine all the places we need to check for Prohibited
    # when setting values - eg, what about in Options
    # Do we want to define a more specific exception for Prohibited?

    # TBD when using Prohibited values, __getattr__ and __getitem_ should raise
    # an exception when being accessed extenrally

    #  TBD Handle the unsetting of magic in subclasses

    #def __setattr__(self, name, value):
    #    print "OptionsChain.__setattr__() name:", name, "value:", value
    #    if name in self:
    #        print "    in self"
    #        if value is Unset and name in self.maps[0]:
    #            # only unset the very top level
    #            del self.maps[0][name]
    #        else:
    #            self[name] = self._magicalized(name, value)
    #    else:
    #        print "    not in self; punt to superclass"
    #        chainstuf.__setattr__(self, name, value)

    # could possibly extend set() magic to setattr (but would have to be
    # careful of recursions). Current draft doesn't work - infinite recursion
class OptionsContext(object):
    Context manager so that modules that use Options can easily implement
    a `with x.settings(...):` capability. In x's class:
    def settings(self, **kwargs):
        return OptionsContext(self, kwargs)

    def __init__(self, caller, kwargs):
        When `with x.method(*args, **kwargs)` is called, it creates an OptionsContext
        passing in its **kwargs. 
        self.caller = caller        
        if 'opts' in kwargs:
            newopts = OptionsChain(caller.options, kwargs['opts'])
            newopts.maps.insert(0, caller._process(newopts, kwargs))
            newopts = OptionsChain(caller.options, kwargs)
        caller.options = newopts

    def __enter__(self):
        Called when the `with` is about to be 'entered'. Whatever this returns
        will be the value of `x` if the `as x` construction is used. Not generally
        needed for option setting, but might be needed in a subclass.
        return self.caller
    def __exit__(self, exc_type, exc_value, traceback):
        Called when leaving the `with`. Reset caller's options to what they were
        before we entered.
        self.caller.options = self.caller.options.maps[-1]

    # Using a with statement and OptionsContext can effectively reduce the
    # thread safety of an object, even to NIL. This is because the object is
    # modified for an indeterminate period. It would be possible to improve
    # thread safety with an with..as construction, if what was returned by as
    # were a proxy that didn't modify the original object.  Another approach
    # might be to provide per-thread options, with _get_options() looking
    # options up in a tid-indexed hash, and set() operations creating a copy
    # (copy-on-write, per thread, essentially).
    # with module quoter have found some interesting use cases like abbreviation
    # args (MultiSetter), and have further explored how subclasses will integrate
    # with the delegation-based options. Have discovered that some subclasses will
    # want flat args, prohibit superclass args. Also, some args could potentially
    # allow setting at instantiation time but not call/use time.
    # Would it make sense to support magicalized class setting -- for subclasses?
    # even if it would, how to accomplish neatly? probably would require a
    # property like object that lets you assign the magic along with the initial value
    # next step: integrate setter from testprop.py
    # consider adding _for key pointing to object that options are for
    # also providing access to whole dict for arguments like chars
    # and clean up HTMLQuoter using the improved functions & access
    # quoter might provide the usecase for a getter, too - in that suffix is == prefix
    # if suffix itself not given
    # instead of Magic() or Setter() maybe Property() or Prop()
Tip: Filter by directory path e.g. /media app.js to search for public/media/app.js.
Tip: Use camelCasing e.g. ProjME to search for ProjectModifiedEvent.java.
Tip: Filter by extension type e.g. /repo .js to search for all .js files in the /repo directory.
Tip: Separate your search with spaces e.g. /ssh pom.xml to search for src/ssh/pom.xml.
Tip: Use ↑ and ↓ arrow keys to navigate and return to view the file.
Tip: You can also navigate files with Ctrl+j (next) and Ctrl+k (previous) and view the file with Ctrl+o.
Tip: You can also navigate files with Alt+j (next) and Alt+k (previous) and view the file with Alt+o.