1
0
mirror of https://github.com/django/django.git synced 2025-10-24 06:06:09 +00:00

Deborgified the app cache.

Improved Andrew's hack to create temporary app caches to handle
migrations. Now the main app cache has a "master" flag set to True
(which is a non-default keyword argument, thus unlikely to be used by
mistake). Other app cache instances have "master" set to False.

The only sanctioned way to access the app cache is by importing
django.core.apps.app_cache.

If you were instanciating an app cache and relying on the Borg pattern,
you'll have to refactor your code.
This commit is contained in:
Aymeric Augustin
2013-12-17 17:47:19 +01:00
parent 6e895f9e06
commit 0242c56fd8
12 changed files with 71 additions and 87 deletions

View File

@@ -23,47 +23,41 @@ class UnavailableApp(Exception):
pass
def _initialize():
class AppCache(object):
"""
Returns a dictionary to be used as the initial value of the
[shared] state of the app cache.
A cache that stores installed applications and their models. Used to
provide reverse-relations and for app introspection.
"""
return dict(
def __init__(self, master=False):
# Only one master of the app-cache may exist at a given time, and it
# shall be the app_cache variable defined at the end of this module.
if master and hasattr(sys.modules[__name__], 'app_cache'):
raise RuntimeError("You may create only one master app cache.")
# When master is set to False, the app cache isn't populated from
# INSTALLED_APPS and ignores the only_installed arguments to
# get_model[s].
self.master = master
# Mapping of labels to AppConfig instances for installed apps.
app_configs=OrderedDict(),
self.app_configs=OrderedDict()
# Pending lookups for lazy relations
pending_lookups={},
self.pending_lookups={}
# Set of app names. Allows restricting the set of installed apps.
# Used by TransactionTestCase.available_apps for performance reasons.
available_apps=None,
self.available_apps=None
# -- Everything below here is only used when populating the cache --
loads_installed=True,
loaded=False,
handled=set(),
postponed=[],
nesting_level=0,
_get_models_cache={},
)
self.loaded=False
self.handled=set()
self.postponed=[]
self.nesting_level=0
self._get_models_cache={}
class BaseAppCache(object):
"""
A cache that stores installed applications and their models. Used to
provide reverse-relations and for app introspection (e.g. admin).
This provides the base (non-Borg) AppCache class - the AppCache
subclass adds borg-like behaviour for the few cases where it's needed.
"""
def __init__(self):
self.__dict__ = _initialize()
# This stops populate loading from INSTALLED_APPS and ignores the
# only_installed arguments to get_model[s]
self.loads_installed = False
def populate(self):
"""
Fill in all the cache information. This method is threadsafe, in the
@@ -72,7 +66,7 @@ class BaseAppCache(object):
"""
if self.loaded:
return
if not self.loads_installed:
if not self.master:
self.loaded = True
return
# Note that we want to use the import lock here - the app loading is
@@ -217,7 +211,7 @@ class BaseAppCache(object):
included in the list of models. However, if you specify
include_swapped, they will be.
"""
if not self.loads_installed:
if not self.master:
only_installed = False
cache_key = (app_mod, include_auto_created, include_deferred, only_installed, include_swapped)
model_list = None
@@ -271,7 +265,7 @@ class BaseAppCache(object):
Raises UnavailableApp when set_available_apps() in in effect and
doesn't include app_label.
"""
if not self.loads_installed:
if not self.master:
only_installed = False
if seed_cache:
self.populate()
@@ -399,18 +393,4 @@ class BaseAppCache(object):
self.register_model(app_label, model)
class AppCache(BaseAppCache):
"""
A cache that stores installed applications and their models. Used to
provide reverse-relations and for app introspection (e.g. admin).
Borg version of the BaseAppCache class.
"""
__shared_state = _initialize()
def __init__(self):
self.__dict__ = self.__shared_state
app_cache = AppCache()
app_cache = AppCache(master=True)

View File

@@ -1,4 +1,4 @@
from django.core.apps.cache import BaseAppCache
from django.core.apps.cache import AppCache
from django.db.backends.schema import BaseDatabaseSchemaEditor
from django.db.models.fields.related import ManyToManyField
@@ -39,7 +39,7 @@ class DatabaseSchemaEditor(BaseDatabaseSchemaEditor):
del body[field.name]
del mapping[field.column]
# Work inside a new AppCache
app_cache = BaseAppCache()
app_cache = AppCache()
# Construct a new model for the new state
meta_contents = {
'app_label': model._meta.app_label,

View File

@@ -1,4 +1,4 @@
from django.core.apps.cache import BaseAppCache
from django.core.apps.cache import AppCache
from django.db import models
from django.utils.timezone import now
@@ -22,7 +22,7 @@ class MigrationRecorder(object):
applied = models.DateTimeField(default=now)
class Meta:
app_cache = BaseAppCache()
app_cache = AppCache()
app_label = "migrations"
db_table = "django_migrations"

View File

@@ -1,4 +1,4 @@
from django.core.apps.cache import BaseAppCache
from django.core.apps.cache import AppCache
from django.db import models
from django.db.models.options import DEFAULT_NAMES, normalize_unique_together
from django.utils import six
@@ -32,7 +32,7 @@ class ProjectState(object):
def render(self):
"Turns the project state into actual models in a new AppCache"
if self.app_cache is None:
self.app_cache = BaseAppCache()
self.app_cache = AppCache()
# We keep trying to render the models in a loop, ignoring invalid
# base errors, until the size of the unrendered models doesn't
# decrease by at least one, meaning there's a base dependency loop/