Document powerline.lint.spec.Spec object

This commit is contained in:
ZyX 2014-09-20 00:12:00 +04:00
parent b781c72348
commit e9c9b6519f
1 changed files with 352 additions and 1 deletions

View File

@ -12,6 +12,51 @@ from powerline.lint.selfcheck import havemarks
class Spec(object):
'''Class that describes some JSON value
In powerline it is only used to describe JSON values stored in powerline
configuration.
:param dict keys:
Dictionary that maps keys that may be present in the given JSON
dictionary to their descriptions. If this parameter is not empty it
implies that described value has dictionary type. Non-dictionary types
must be described using ``Spec()``: without arguments.
.. note::
Methods that create the specifications return ``self``, so calls to them
may be chained: ``Spec().type(unicode).re('^\w+$')``. This does not
apply to functions that *apply* specification like :py:meth`Spec.match`.
.. note::
Methods starting with ``check_`` return two values: first determines
whether caller should proceed on running other checks, second
determines whether there were any problems (i.e. whether error was
reported). One should not call these methods directly: there is
:py:meth:`Spec.match` method for checking values.
.. note::
In ``check_`` and ``match`` methods specifications are identified by
their indexes for the purpose of simplyfying :py:meth:`Spec.copy`
method.
Some common parameters:
``data``:
Whatever data supplied by the first caller for checker functions. Is not
processed by :py:class:`Spec` methods in any fashion.
``context``:
:py:class:`powerline.lint.context.Context` instance, describes context
of the value. :py:class:`Spec` methods only use its ``.key`` methods for
error messages.
``echoerr``:
Callable that should be used to echo errors. Is supposed to take four
optional keyword arguments: ``problem``, ``problem_mark``, ``context``,
``context_mark``.
``value``:
Checked value.
'''
def __init__(self, **keys):
self.specs = []
self.keys = {}
@ -24,6 +69,13 @@ class Spec(object):
self.update(**keys)
def update(self, **keys):
'''Describe additional keys that may be present in given JSON value
If called with some keyword arguments implies that described value is
a dictionary. If called without keyword parameters it is no-op.
:return: self.
'''
for k, v in keys.items():
self.keys[k] = len(self.specs)
self.specs.append(v)
@ -33,6 +85,14 @@ class Spec(object):
return self
def copy(self, copied=None):
'''Deep copy the spec
:param dict copied:
Internal dictionary used for storing already copied values. This
parameter should not be used.
:return: New :py:class:`Spec` object that is a deep copy of ``self``.
'''
copied = copied or {}
try:
return copied[id(self)]
@ -42,6 +102,15 @@ class Spec(object):
return self.__class__()._update(self.__dict__, copied)
def _update(self, d, copied):
'''Helper for the :py:meth:`Spec.copy` function
Populates new instance with values taken from the old one.
:param dict d:
``__dict__`` of the old instance.
:param dict copied:
Storage for already copied values.
'''
self.__dict__.update(d)
self.keys = copy(self.keys)
self.checks = copy(self.checks)
@ -50,6 +119,30 @@ class Spec(object):
return self
def unknown_spec(self, keyfunc, spec):
'''Define specification for non-static keys
This method should be used if key names cannot be determined at runtime
or if a number of keys share identical spec (in order to not repeat it).
:py:meth:`Spec.match` method processes dictionary in the given order:
* First it tries to use specifications provided at the initialization or
by the :py:meth:`Spec.update` method.
* If no specification for given key was provided it processes
specifications from ``keyfunc`` argument in order they were supplied.
Once some key matches specification supplied second ``spec`` argument
is used to determine correctness of the value.
:param Spec keyfunc:
:py:class:`Spec` instance or a regular function that returns two
values (the same :py:meth:`Spec.match` returns). This argument is
used to match keys that were not provided at initialization or via
:py:meth:`Spec.update`.
:param Spec spec:
:py:class:`Spec` instance that will be used to check keys matched by
``keyfunc``.
:return: self.
'''
if isinstance(keyfunc, Spec):
self.specs.append(keyfunc)
keyfunc = len(self.specs) - 1
@ -58,10 +151,31 @@ class Spec(object):
return self
def unknown_msg(self, msgfunc):
'''Define message which will be used when unknown key was found
Unknown is a key that was not provided at the initialization and via
:py:meth:`Spec.update` and did not match any ``keyfunc`` proided via
:py:meth:`Spec.unknown_spec`.
:param msgfunc:
Function that takes that unknown key as an argument and returns the
message text. Text will appear at the top (start of the sentence).
:return: self.
'''
self.ufailmsg = msgfunc
return self
def context_message(self, msg):
'''Define message that describes context
:param str msg:
Message that describes context. Is written using the
:py:meth:`str.format` syntax and is expected to display keyword
parameter ``key``.
:return: self.
'''
self.cmsg = msg
for spec in self.specs:
if not spec.cmsg:
@ -69,6 +183,15 @@ class Spec(object):
return self
def check_type(self, value, context_mark, data, context, echoerr, types):
'''Check that given value matches given type(s)
:param tuple types:
List of accepted types. Since :py:class:`Spec` is supposed to
describe JSON values only ``dict``, ``list``, ``unicode``, ``bool``,
``float`` and ``NoneType`` types make any sense.
:return: proceed, hadproblem.
'''
havemarks(value)
if type(value.value) not in types:
echoerr(
@ -85,6 +208,34 @@ class Spec(object):
return True, False
def check_func(self, value, context_mark, data, context, echoerr, func, msg_func):
'''Check value using given function
:param function func:
Callable that should accept four positional parameters:
#. checked value,
#. ``data`` parameter with arbitrary data (supplied by top-level
caller),
#. current context and
#. function used for echoing errors.
This callable should return three values:
#. determines whether ``check_func`` caller should proceed
calling other checks,
#. determines whether ``check_func`` should echo error on its own
(it should be set to False if ``func`` echoes error itself) and
#. determines whether function has found some errors in the checked
value.
:param function msg_func:
Callable that takes checked value as the only positional parameter
and returns a string that describes the problem. Only useful for
small checker functions since it is ignored when second returned
value is false.
:return: proceed, hadproblem.
'''
havemarks(value)
proceed, echo, hadproblem = func(value, data, context, echoerr)
if echo and hadproblem:
@ -95,6 +246,19 @@ class Spec(object):
return proceed, hadproblem
def check_list(self, value, context_mark, data, context, echoerr, item_func, msg_func):
'''Check that each value in the list matches given specification
:param function item_func:
Callable like ``func`` from :py:meth:`Spec.check_func`. Unlike
``func`` this callable is called for each value in the list and may
be a :py:class:`Spec` object index.
:param func msg_func:
Callable like ``msg_func`` from :py:meth:`Spec.check_func`. Should
accept one problematic item and is not used for :py:class:`Spec`
object indicies in ``item_func`` method.
:return: proceed, hadproblem.
'''
havemarks(value)
i = 0
hadproblem = False
@ -124,6 +288,17 @@ class Spec(object):
return True, hadproblem
def check_either(self, value, context_mark, data, context, echoerr, start, end):
'''Check that given value matches one of the given specifications
:param int start:
First specification index.
:param int end:
Specification index that is greater by 1 then last specification
index.
This method does not give an error if any specification from
``self.specs[start:end]`` is matched by the given value.
'''
havemarks(value)
new_echoerr = DelayedEchoErr(echoerr)
@ -140,6 +315,17 @@ class Spec(object):
return False, hadproblem
def check_tuple(self, value, context_mark, data, context, echoerr, start, end):
'''Check that given value is a list with items matching specifications
:param int start:
First specification index.
:param int end:
Specification index that is greater by 1 then last specification
index.
This method checks that each item in the value list matches
specification with index ``start + item_number``.
'''
havemarks(value)
hadproblem = False
for (i, item, spec) in zip(itertools.count(), value, self.specs[start:end]):
@ -157,6 +343,15 @@ class Spec(object):
return True, hadproblem
def type(self, *args):
'''Describe value that has one of the types given in arguments
:param args:
List of accepted types. Since :py:class:`Spec` is supposed to
describe JSON values only ``dict``, ``list``, ``unicode``, ``bool``,
``float`` and ``NoneType`` types make any sense.
:return: self.
'''
self.checks.append(('check_type', args))
return self
@ -177,6 +372,20 @@ class Spec(object):
}
def len(self, comparison, cint, msg_func=None):
'''Describe value that has given length
:param str comparison:
Type of the comparison. Valid values: ``le``, ``lt``, ``ge``,
``gt``, ``eq``.
:param int cint:
Integer with which length is compared.
:param function msg_func:
Function that should accept checked value and return message that
describes the problem with this value. Default value will emit
something like length of ["foo", "bar"] is not greater then 10.
:return: self.
'''
cmp_func = self.cmp_funcs[comparison]
msg_func = (
msg_func
@ -191,6 +400,25 @@ class Spec(object):
return self
def cmp(self, comparison, cint, msg_func=None):
'''Describe value that is a number or string that has given property
:param str comparison:
Type of the comparison. Valid values: ``le``, ``lt``, ``ge``,
``gt``, ``eq``. This argument will restrict the number or string to
emit True on the given comparison.
:param cint:
Number or string with which value is compared. Type of this
parameter affects required type of the checked value: ``str`` and
``unicode`` types imply ``unicode`` values, ``float`` type implies
that value can be either ``int`` or ``float``, ``int`` type implies
``int`` value and for any other type the behavior is undefined.
:param function msg_func:
Function that should accept checked value and return message that
describes the problem with this value. Default value will emit
something like 10 is not greater then 10.
:return: self.
'''
if type(cint) is str:
self.type(unicode)
elif type(cint) is float:
@ -207,6 +435,14 @@ class Spec(object):
return self
def unsigned(self, msg_func=None):
'''Describe unsigned integer value
:param function msg_func:
Function that should accept checked value and return message that
describes the problem with this value.
:return: self.
'''
self.type(int)
self.checks.append((
'check_func',
@ -216,6 +452,20 @@ class Spec(object):
return self
def list(self, item_func, msg_func=None):
'''Describe list with any number of elements, each matching given spec
:param item_func:
:py:class:`Spec` instance or a callable. Check out
:py:meth:`Spec.check_list` documentation for more details. Note that
in :py:meth:`Spec.check_list` description :py:class:`Spec` instance
is replaced with its index in ``self.specs``.
:param function msg_func:
Function that should accept checked value and return message that
describes the problem with this value. Default value will emit just
failed check, which is rather indescriptive.
:return: self.
'''
self.type(list)
if isinstance(item_func, Spec):
self.specs.append(item_func)
@ -224,6 +474,17 @@ class Spec(object):
return self
def tuple(self, *specs):
'''Describe list with the given number of elements, each matching corresponding spec
:param (Spec,) specs:
List of specifications. Last element(s) in this list may be
optional. Each element in this list describes element with the same
index in the checked value. Check out :py:meth:`Spec.check_tuple`
for more details, but note that there list of specifications is
replaced with start and end indicies in ``self.specs``.
:return: self.
'''
self.type(list)
max_len = len(specs)
@ -246,10 +507,25 @@ class Spec(object):
return self
def func(self, func, msg_func=None):
'''Describe value that is checked by the given function
Check out :py:meth:`Spec.check_func` documentation for more details.
'''
self.checks.append(('check_func', func, msg_func or (lambda value: 'failed check')))
return self
def re(self, regex, msg_func=None):
'''Describe value that is a string that matches given regular expression
:param str regex:
Regular expression that should be matched by the value.
:param function msg_func:
Function that should accept checked value and return message that
describes the problem with this value. Default value will emit
something like String "xyz" does not match "[a-f]+".
:return: self.
'''
self.type(unicode)
compiled = re.compile(regex)
msg_func = msg_func or (lambda value: 'String "{0}" does not match "{1}"'.format(value, regex))
@ -261,6 +537,15 @@ class Spec(object):
return self
def ident(self, msg_func=None):
'''Describe value that is an identifier like ``foo:bar`` or ``foo``
:param function msg_func:
Function that should accept checked value and return message that
describes the problem with this value. Default value will emit
something like String "xyz" is not an identifier.
:return: self.
'''
msg_func = (
msg_func
or (lambda value: 'String "{0}" is not an alphanumeric/underscore colon-separated identifier'.format(value))
@ -268,6 +553,17 @@ class Spec(object):
return self.re('^\w+(?::\w+)?$', msg_func)
def oneof(self, collection, msg_func=None):
'''Describe value that is equal to one of the value in the collection
:param set collection:
A collection of possible values.
:param function msg_func:
Function that should accept checked value and return message that
describes the problem with this value. Default value will emit
something like "xyz" must be one of {"abc", "def", "ghi"}.
:return: self.
'''
msg_func = msg_func or (lambda value: '"{0}" must be one of {1!r}'.format(value, list(collection)))
self.checks.append((
'check_func',
@ -277,6 +573,19 @@ class Spec(object):
return self
def error(self, msg):
'''Describe value that must not be there
Useful for giving more descriptive errors for some specific keys then
just found unknown key: shutdown_event or for forbidding certain
values when :py:meth:`Spec.unknown_spec` was used.
:param str msg:
Message given for the offending value. It is formatted using
:py:meth:`str.format` with the only positional parameter which is
the value itself.
:return: self.
'''
self.checks.append((
'check_func',
(lambda *args: (True, True, True)),
@ -285,20 +594,54 @@ class Spec(object):
return self
def either(self, *specs):
'''Describes value that matches one of the given specs
Check out :py:meth:`Spec.check_either` method documentation for more
details, but note that there a list of specs was replaced by start and
end indicies in ``self.specs``.
:return: self.
'''
start = len(self.specs)
self.specs.extend(specs)
self.checks.append(('check_either', start, len(self.specs)))
return self
def optional(self):
'''Mark value as optional
Only useful for key specs in :py:meth:`Spec.__init__` and
:py:meth:`Spec.update` and some last supplied to :py:meth:`Spec.tuple`.
:return: self.
'''
self.isoptional = True
return self
def required(self):
'''Mark value as required
Only useful for key specs in :py:meth:`Spec.__init__` and
:py:meth:`Spec.update` and some last supplied to :py:meth:`Spec.tuple`.
.. note::
Value is required by default. This method is only useful for
altering existing specification (or rather its copy).
:return: self.
'''
self.isoptional = False
return self
def match_checks(self, *args):
'''Process checks registered for the given value
Processes only top-level checks: key specifications given using at the
initialization or via :py:meth:`Spec.unknown_spec` are processed by
:py:meth:`Spec.match`.
:return: proceed, hadproblem.
'''
hadproblem = False
for check in self.checks:
proceed, chadproblem = getattr(self, check[0])(*(args + check[1:]))
@ -309,6 +652,10 @@ class Spec(object):
return True, hadproblem
def match(self, value, context_mark=None, data=None, context=(), echoerr=echoerr):
'''Check that given value matches this specification
:return: proceed, hadproblem.
'''
havemarks(value)
proceed, hadproblem = self.match_checks(value, context_mark, data, context, echoerr)
if proceed:
@ -367,7 +714,11 @@ class Spec(object):
return True, hadproblem
def __getitem__(self, key):
'''Get specification for the given key
'''
return self.specs[self.keys[key]]
def __setitem__(self, key, value):
return self.update(**{key: value})
'''Set specification for the given key
'''
self.update(**{key: value})