X-Git-Url: https://git.openstreetmap.org./osqa.git/blobdiff_plain/0ba16baba0615dd405486c7d87f943d71518375c..04aadd147b0d494c043caf9702684592652a26ab:/forum/models/base.py diff --git a/forum/models/base.py b/forum/models/base.py index dfdf989..2cfe7db 100644 --- a/forum/models/base.py +++ b/forum/models/base.py @@ -1,5 +1,9 @@ import datetime import re +try: + from hashlib import md5 +except: + from md5 import new as md5 from urllib import quote_plus, urlencode from django.db import models, IntegrityError, connection, transaction from django.utils.http import urlquote as django_urlquote @@ -12,52 +16,69 @@ from django.template.defaultfilters import slugify from django.db.models.signals import post_delete, post_save, pre_save, pre_delete from django.utils.translation import ugettext as _ from django.utils.safestring import mark_safe +from django.utils.encoding import force_unicode from django.contrib.sitemaps import ping_google import django.dispatch from forum import settings import logging +if not hasattr(cache, 'get_many'): + #put django 1.2 code here + pass + class LazyQueryList(object): def __init__(self, model, items): - self.model = model self.items = items + self.model = model def __getitem__(self, k): - return self.model.objects.get(id=self.items[k]) + return self.model.objects.get(id=self.items[k][0]) def __iter__(self): for id in self.items: - yield self.model.objects.get(id=id) + yield self.model.objects.get(id=id[0]) def __len__(self): return len(self.items) +class ToFetch(unicode): + pass + class CachedQuerySet(models.query.QuerySet): + def lazy(self): - if len(self.query.extra) == 0: - return LazyQueryList(self.model, list(self.values_list('id', flat=True))) + if not len(self.query.aggregates): + values_list = ['id'] + + if len(self.query.extra): + extra_keys = self.query.extra.keys() + values_list += extra_keys + + return LazyQueryList(self.model, list(self.values_list(*values_list))) else: return self def obj_from_datadict(self, datadict): obj = self.model() obj.__dict__.update(datadict) + + if hasattr(obj, '_state'): + obj._state.db = 'default' + return obj + def _base_clone(self): + return self._clone(klass=models.query.QuerySet) + def get(self, *args, **kwargs): - try: - pk = [v for (k,v) in kwargs.items() if k in ('pk', 'pk__exact', 'id', 'id__exact' - ) or k.endswith('_ptr__pk') or k.endswith('_ptr__id')][0] - except: - pk = None + key = self.model.infer_cache_key(kwargs) - if pk is not None: - key = self.model.cache_key(pk) + if key is not None: obj = cache.get(key) if obj is None: - obj = super(CachedQuerySet, self).get(*args, **kwargs) + obj = self._base_clone().get(*args, **kwargs) obj.cache() else: obj = self.obj_from_datadict(obj) @@ -65,7 +86,98 @@ class CachedQuerySet(models.query.QuerySet): return obj - return super(CachedQuerySet, self).get(*args, **kwargs) + return self._base_clone().get(*args, **kwargs) + + def _fetch_from_query_cache(self, key): + invalidation_key = self.model._get_cache_query_invalidation_key() + cached_result = cache.get_many([invalidation_key, key]) + + if not invalidation_key in cached_result: + self.model._set_query_cache_invalidation_timestamp() + return None + + if (key in cached_result) and(cached_result[invalidation_key] < cached_result[key][0]): + return cached_result[key][1] + + return None + + def count(self): + cache_key = self.model._generate_cache_key("CNT:%s" % self._get_query_hash()) + result = self._fetch_from_query_cache(cache_key) + + if result is not None: + return result + + result = super(CachedQuerySet, self).count() + cache.set(cache_key, (datetime.datetime.now(), result), 60 * 60) + return result + + def iterator(self): + cache_key = self.model._generate_cache_key("QUERY:%s" % self._get_query_hash()) + on_cache_query_attr = self.model.value_to_list_on_cache_query() + + to_return = None + to_cache = {} + + with_aggregates = len(self.query.aggregates) > 0 + key_list = self._fetch_from_query_cache(cache_key) + + if key_list is None: + if not with_aggregates: + values_list = [on_cache_query_attr] + + if len(self.query.extra): + values_list += self.query.extra.keys() + + key_list = [v[0] for v in self.values_list(*values_list)] + to_cache[cache_key] = (datetime.datetime.now(), key_list) + else: + to_return = list(super(CachedQuerySet, self).iterator()) + to_cache[cache_key] = (datetime.datetime.now(), [ + (row.__dict__[on_cache_query_attr], dict([(k, row.__dict__[k]) for k in self.query.aggregates.keys()])) + for row in to_return]) + elif with_aggregates: + tmp = key_list + key_list = [k[0] for k in tmp] + with_aggregates = [k[1] for k in tmp] + del tmp + + if (not to_return) and key_list: + row_keys = [self.model.infer_cache_key({on_cache_query_attr: attr}) for attr in key_list] + cached = cache.get_many(row_keys) + + to_return = [ + (ck in cached) and self.obj_from_datadict(cached[ck]) or ToFetch(force_unicode(key_list[i])) for i, ck in enumerate(row_keys) + ] + + if len(cached) != len(row_keys): + to_fetch = [unicode(tr) for tr in to_return if isinstance(tr, ToFetch)] + + fetched = dict([(force_unicode(r.__dict__[on_cache_query_attr]), r) for r in + models.query.QuerySet(self.model).filter(**{"%s__in" % on_cache_query_attr: to_fetch})]) + + to_return = [(isinstance(tr, ToFetch) and fetched[unicode(tr)] or tr) for tr in to_return] + to_cache.update(dict([(self.model.infer_cache_key({on_cache_query_attr: attr}), r._as_dict()) for attr, r in fetched.items()])) + + if with_aggregates: + for i, r in enumerate(to_return): + r.__dict__.update(with_aggregates[i]) + + + if len(to_cache): + cache.set_many(to_cache, 60 * 60) + + if to_return: + for row in to_return: + if hasattr(row, 'leaf'): + yield row.leaf + else: + yield row + + def _get_query_hash(self): + return md5(unicode(self.query).encode("utf-8")).hexdigest() + + class CachedManager(models.Manager): use_for_related_fields = True @@ -81,9 +193,9 @@ class CachedManager(models.Manager): class DenormalizedField(object): - def __init__(self, manager, **kwargs): + def __init__(self, manager, *args, **kwargs): self.manager = manager - self.filter = kwargs + self.filter = (args, kwargs) def setup_class(self, cls, name): dict_name = '_%s_dencache_' % name @@ -92,7 +204,7 @@ class DenormalizedField(object): val = inst.__dict__.get(dict_name, None) if val is None: - val = getattr(inst, self.manager).filter(**self.filter).count() + val = getattr(inst, self.manager).filter(*self.filter[0], **self.filter[1]).count() inst.__dict__[dict_name] = val inst.cache() @@ -135,14 +247,13 @@ class BaseModel(models.Model): def __init__(self, *args, **kwargs): super(BaseModel, self).__init__(*args, **kwargs) - self.reset_original_state() - - @classmethod - def cache_key(cls, pk): - return '%s:%s:%s' % (settings.APP_URL, cls.__name__, pk) + self.reset_original_state(kwargs.keys()) - def reset_original_state(self): + def reset_original_state(self, reset_fields=None): self._original_state = self._as_dict() + + if reset_fields: + self._original_state.update(dict([(f, None) for f in reset_fields])) def get_dirty_fields(self): return [f.name for f in self._meta.fields if self._original_state[f.attname] != self.__dict__[f.attname]] @@ -157,10 +268,13 @@ class BaseModel(models.Model): (f.name, getattr(self, f.name)) for f in self._meta.fields if self._original_state[f.attname] != self.__dict__[f.attname] ]) - def save(self, *args, **kwargs): + def save(self, full_save=False, *args, **kwargs): put_back = [k for k, v in self.__dict__.items() if isinstance(v, models.expressions.ExpressionNode)] - if self.id: + if hasattr(self, '_state'): + self._state.db = 'default' + + if self.id and not full_save: self.__class__.objects.filter(id=self.id).update(**self._get_update_kwargs()) else: super(BaseModel, self).save() @@ -175,16 +289,53 @@ class BaseModel(models.Model): self.uncache() self.reset_original_state() + self._set_query_cache_invalidation_timestamp() self.cache() + @classmethod + def _get_cache_query_invalidation_key(cls): + return cls._generate_cache_key("INV_TS") + + @classmethod + def _set_query_cache_invalidation_timestamp(cls): + cache.set(cls._get_cache_query_invalidation_key(), datetime.datetime.now(), 60 * 60 * 24) + + for base in filter(lambda c: issubclass(c, BaseModel) and (not c is BaseModel), cls.__bases__): + base._set_query_cache_invalidation_timestamp() + + @classmethod + def _generate_cache_key(cls, key, group=None): + if group is None: + group = cls.__name__ + + return '%s:%s:%s' % (settings.APP_URL, group, key) + + def cache_key(self): + return self._generate_cache_key(self.id) + + @classmethod + def value_to_list_on_cache_query(cls): + return 'id' + + @classmethod + def infer_cache_key(cls, querydict): + try: + pk = [v for (k,v) in querydict.items() if k in ('pk', 'pk__exact', 'id', 'id__exact' + ) or k.endswith('_ptr__pk') or k.endswith('_ptr__id')][0] + + return cls._generate_cache_key(pk) + except: + return None + def cache(self): - cache.set(self.cache_key(self.id), self._as_dict(), 60 * 60) + cache.set(self.cache_key(), self._as_dict(), 60 * 60) def uncache(self): - cache.delete(self.cache_key(self.id)) + cache.delete(self.cache_key()) def delete(self): self.uncache() + self._set_query_cache_invalidation_timestamp() super(BaseModel, self).delete()