Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Tree: 4a4ef4f7ac
Fetching contributors…

Cannot retrieve contributors at this time

292 lines (264 sloc) 11.397 kB
import re
import traceback
from zope.interface import implementer
from pyramid.interfaces import IActionInfo
from pyramid.compat import (
from pyramid.exceptions import ConfigurationError
from pyramid.traversal import (
from hashlib import md5
MAX_ORDER = 1 << 30
DEFAULT_PHASH = md5().hexdigest()
class ActionInfo(object):
def __init__(self, file, line, function, src):
self.file = file
self.line = line
self.function = function
self.src = src
def __str__(self):
srclines = self.src.split('\n')
src = '\n'.join(' %s' % x for x in srclines)
return 'Line %s of file %s:\n%s' % (self.line, self.file, src)
def action_method(wrapped):
""" Wrapper to provide the right conflict info report data when a method
that calls Configurator.action calls another that does the same"""
def wrapper(self, *arg, **kw):
if self._ainfo is None:
self._ainfo = []
info = kw.pop('_info', None)
if is_nonstr_iter(info) and len(info) == 4:
# _info permitted as extract_stack tuple
info = ActionInfo(*info)
if info is None:
f = traceback.extract_stack(limit=3)
info = ActionInfo(*f[-2])
except: # pragma: no cover
info = ActionInfo(None, 0, '', '')
result = wrapped(self, *arg, **kw)
return result
wrapper.__name__ = wrapped.__name__
wrapper.__doc__ = wrapped.__doc__
wrapper.__docobj__ = wrapped # for sphinx
return wrapper
def make_predicates(xhr=None, request_method=None, path_info=None,
request_param=None, match_param=None, header=None,
accept=None, containment=None, request_type=None,
traverse=None, custom=()):
# ----------
# Given an argument list, a predicate list is computed.
# Predicates are added to a predicate list in (presumed)
# computation expense order. All predicates associated with a
# view or route must evaluate true for the view or route to
# "match" during a request. Elsewhere in the code, we evaluate
# predicates using a generator expression. The fastest predicate
# should be evaluated first, then the next fastest, and so on, as
# if one returns false, the remainder of the predicates won't need
# to be evaluated.
# While we compute predicates, we also compute a predicate hash
# (aka phash) that can be used by a caller to identify identical
# predicate lists.
# --------
# A "order" is computed for the predicate list. An order is
# a scoring.
# Each predicate is associated with a weight value, which is a
# multiple of 2. The weight of a predicate symbolizes the
# relative potential "importance" of the predicate to all other
# predicates. A larger weight indicates greater importance.
# All weights for a given predicate list are bitwise ORed together
# to create a "score"; this score is then subtracted from
# MAX_ORDER and divided by an integer representing the number of
# predicates+1 to determine the order.
# The order represents the ordering in which a "multiview" ( a
# collection of views that share the same context/request/name
# triad but differ in other ways via predicates) will attempt to
# call its set of views. Views with lower orders will be tried
# first. The intent is to a) ensure that views with more
# predicates are always evaluated before views with fewer
# predicates and b) to ensure a stable call ordering of views that
# share the same number of predicates. Views which do not have
# any predicates get an order of MAX_ORDER, meaning that they will
# be tried very last.
predicates = []
weights = []
h = md5()
if xhr:
def xhr_predicate(context, request):
return request.is_xhr
xhr_predicate.__text__ = "xhr = True"
weights.append(1 << 1)
h.update(bytes_('xhr:%r' % bool(xhr)))
if request_method is not None:
if not is_nonstr_iter(request_method):
request_method = (request_method,)
request_method = sorted(request_method)
def request_method_predicate(context, request):
return request.method in request_method
text = "request method = %s" % repr(request_method)
request_method_predicate.__text__ = text
weights.append(1 << 2)
for m in request_method:
h.update(bytes_('request_method:%r' % m))
if path_info is not None:
path_info_val = re.compile(path_info)
except re.error as why:
raise ConfigurationError(why.args[0])
def path_info_predicate(context, request):
return path_info_val.match(request.path_info) is not None
text = "path_info = %s"
path_info_predicate.__text__ = text % path_info
weights.append(1 << 3)
h.update(bytes_('path_info:%r' % path_info))
if request_param is not None:
request_param_val = None
if '=' in request_param:
request_param, request_param_val = request_param.split('=', 1)
if request_param_val is None:
text = "request_param %s" % request_param
text = "request_param %s = %s" % (request_param, request_param_val)
def request_param_predicate(context, request):
if request_param_val is None:
return request_param in request.params
return request.params.get(request_param) == request_param_val
request_param_predicate.__text__ = text
weights.append(1 << 4)
bytes_('request_param:%r=%r' % (request_param, request_param_val)))
if header is not None:
header_name = header
header_val = None
if ':' in header:
header_name, header_val = header.split(':', 1)
header_val = re.compile(header_val)
except re.error as why:
raise ConfigurationError(why.args[0])
if header_val is None:
text = "header %s" % header_name
text = "header %s = %s" % (header_name, header_val)
def header_predicate(context, request):
if header_val is None:
return header_name in request.headers
val = request.headers.get(header_name)
if val is None:
return False
return header_val.match(val) is not None
header_predicate.__text__ = text
weights.append(1 << 5)
h.update(bytes_('header:%r=%r' % (header_name, header_val)))
if accept is not None:
def accept_predicate(context, request):
return accept in request.accept
accept_predicate.__text__ = "accept = %s" % accept
weights.append(1 << 6)
h.update(bytes_('accept:%r' % accept))
if containment is not None:
def containment_predicate(context, request):
return find_interface(context, containment) is not None
containment_predicate.__text__ = "containment = %s" % containment
weights.append(1 << 7)
h.update(bytes_('containment:%r' % hash(containment)))
if request_type is not None:
def request_type_predicate(context, request):
return request_type.providedBy(request)
text = "request_type = %s"
request_type_predicate.__text__ = text % request_type
weights.append(1 << 8)
h.update(bytes_('request_type:%r' % hash(request_type)))
if match_param is not None:
if isinstance(match_param, string_types):
match_param, match_param_val = match_param.split('=', 1)
match_param = {match_param: match_param_val}
text = "match_param %s" % match_param
def match_param_predicate(context, request):
for k, v in match_param.items():
if request.matchdict.get(k) != v:
return False
return True
match_param_predicate.__text__ = text
weights.append(1 << 9)
h.update(bytes_('match_param:%r' % match_param))
if custom:
for num, predicate in enumerate(custom):
if getattr(predicate, '__text__', None) is None:
text = '<unknown custom predicate>'
predicate.__text__ = text
except AttributeError:
# if this happens the predicate is probably a classmethod
if hasattr(predicate, '__func__'):
predicate.__func__.__text__ = text
else: # pragma: no cover ; 2.5 doesn't have __func__
predicate.im_func.__text__ = text
# using hash() here rather than id() is intentional: we
# want to allow custom predicates that are part of
# frameworks to be able to define custom __hash__
# functions for custom predicates, so that the hash output
# of predicate instances which are "logically the same"
# may compare equal.
h.update(bytes_('custom%s:%r' % (num, hash(predicate))))
weights.append(1 << 10)
if traverse is not None:
# ``traverse`` can only be used as a *route* "predicate"; it
# adds 'traverse' to the matchdict if it's specified in the
# routing args. This causes the ResourceTreeTraverser to use
# the resolved traverse pattern as the traversal path.
from pyramid.urldispatch import _compile_route
_, tgenerate = _compile_route(traverse)
def traverse_predicate(context, request):
if 'traverse' in context:
return True
m = context['match']
tvalue = tgenerate(m)
m['traverse'] = traversal_path_info(tvalue)
return True
# This isn't actually a predicate, it's just a infodict
# modifier that injects ``traverse`` into the matchdict. As a
# result, the ``traverse_predicate`` function above always
# returns True, and we don't need to update the hash or attach
# a weight to it
score = 0
for bit in weights:
score = score | bit
order = (MAX_ORDER - score) / (len(predicates) + 1)
phash = h.hexdigest()
return order, predicates, phash
def as_sorted_tuple(val):
if not is_nonstr_iter(val):
val = (val,)
val = tuple(sorted(val))
return val
Jump to Line
Something went wrong with that request. Please try again.