Removed trailing_pathsep support from TreeManager.get_with_path. Set nodes to have...
[philo.git] / models / base.py
index 27c45dc..43de3b0 100644 (file)
@@ -1,23 +1,30 @@
+from django import forms
 from django.db import models
 from django.contrib.contenttypes.models import ContentType
 from django.contrib.contenttypes import generic
-from django.utils import simplejson as json
 from django.core.exceptions import ObjectDoesNotExist
+from django.core.validators import RegexValidator
+from django.utils import simplejson as json
+from django.utils.encoding import smart_str
 from philo.exceptions import AncestorDoesNotExist
-from philo.utils import ContentTypeRegistryLimiter
+from philo.models.fields import JSONField
+from philo.utils import ContentTypeRegistryLimiter, ContentTypeSubclassLimiter
 from philo.signals import entity_class_prepared
+from philo.validators import json_validator
 from UserDict import DictMixin
+from mptt.models import MPTTModel, MPTTModelBase, MPTTOptions
 
 
 class Tag(models.Model):
-       name = models.CharField(max_length=250)
-       slug = models.SlugField(unique=True)
+       name = models.CharField(max_length=255)
+       slug = models.SlugField(max_length=255, unique=True)
        
        def __unicode__(self):
                return self.name
        
        class Meta:
                app_label = 'philo'
+               ordering = ('name',)
 
 
 class Titled(models.Model):
@@ -31,59 +38,203 @@ class Titled(models.Model):
                abstract = True
 
 
-class Attribute(models.Model):
-       entity_content_type = models.ForeignKey(ContentType, verbose_name='Entity type')
-       entity_object_id = models.PositiveIntegerField(verbose_name='Entity ID')
-       entity = generic.GenericForeignKey('entity_content_type', 'entity_object_id')
-       key = models.CharField(max_length=255)
-       json_value = models.TextField(verbose_name='Value (JSON)', help_text='This value must be valid JSON.')
+value_content_type_limiter = ContentTypeRegistryLimiter()
+
+
+def register_value_model(model):
+       value_content_type_limiter.register_class(model)
+
+
+register_value_model(Tag)
+
+
+def unregister_value_model(model):
+       value_content_type_limiter.unregister_class(model)
+
+
+class AttributeValue(models.Model):
+       attribute_set = generic.GenericRelation('Attribute', content_type_field='value_content_type', object_id_field='value_object_id')
        
-       def get_value(self):
-               return json.loads(self.json_value)
+       @property
+       def attribute(self):
+               return self.attribute_set.all()[0]
        
        def set_value(self, value):
-               self.json_value = json.dumps(value)
+               raise NotImplementedError
        
-       def delete_value(self):
-               self.json_value = json.dumps(None)
+       def value_formfields(self, **kwargs):
+               """Define any formfields that would be used to construct an instance of this value."""
+               raise NotImplementedError
        
-       value = property(get_value, set_value, delete_value)
+       def construct_instance(self, **kwargs):
+               """Apply cleaned data from the formfields generated by valid_formfields to oneself."""
+               raise NotImplementedError
        
        def __unicode__(self):
-               return u'"%s": %s' % (self.key, self.value)
+               return unicode(self.value)
        
        class Meta:
-               app_label = 'philo'
-               unique_together = ('key', 'entity_content_type', 'entity_object_id')
+               abstract = True
 
 
-value_content_type_limiter = ContentTypeRegistryLimiter()
+attribute_value_limiter = ContentTypeSubclassLimiter(AttributeValue)
 
 
-def register_value_model(model):
-       value_content_type_limiter.register_class(model)
+class JSONValue(AttributeValue):
+       value = JSONField(verbose_name='Value (JSON)', help_text='This value must be valid JSON.', default='null')
+       
+       def __unicode__(self):
+               return smart_str(self.value)
+       
+       def value_formfields(self):
+               kwargs = {'initial': self.value_json}
+               field = self._meta.get_field('value')
+               return {field.name: field.formfield(**kwargs)}
+       
+       def construct_instance(self, **kwargs):
+               field_name = self._meta.get_field('value').name
+               self.set_value(kwargs.pop(field_name, None))
+       
+       def set_value(self, value):
+               self.value = value
+       
+       class Meta:
+               app_label = 'philo'
 
 
-def unregister_value_model(model):
-       value_content_type_limiter.unregister_class(model)
+class ForeignKeyValue(AttributeValue):
+       content_type = models.ForeignKey(ContentType, limit_choices_to=value_content_type_limiter, verbose_name='Value type', null=True, blank=True)
+       object_id = models.PositiveIntegerField(verbose_name='Value ID', null=True, blank=True)
+       value = generic.GenericForeignKey()
+       
+       def value_formfields(self):
+               field = self._meta.get_field('content_type')
+               fields = {field.name: field.formfield(initial=getattr(self.content_type, 'pk', None))}
+               
+               if self.content_type:
+                       kwargs = {
+                               'initial': self.object_id,
+                               'required': False,
+                               'queryset': self.content_type.model_class()._default_manager.all()
+                       }
+                       fields['value'] = forms.ModelChoiceField(**kwargs)
+               return fields
+       
+       def construct_instance(self, **kwargs):
+               field_name = self._meta.get_field('content_type').name
+               ct = kwargs.pop(field_name, None)
+               if ct is None or ct != self.content_type:
+                       self.object_id = None
+                       self.content_type = ct
+               else:
+                       value = kwargs.pop('value', None)
+                       self.set_value(value)
+                       if value is None:
+                               self.content_type = ct
+       
+       def set_value(self, value):
+               self.value = value
+       
+       class Meta:
+               app_label = 'philo'
 
 
+class ManyToManyValue(AttributeValue):
+       content_type = models.ForeignKey(ContentType, limit_choices_to=value_content_type_limiter, verbose_name='Value type', null=True, blank=True)
+       values = models.ManyToManyField(ForeignKeyValue, blank=True, null=True)
+       
+       def get_object_ids(self):
+               return self.values.values_list('object_id', flat=True)
+       object_ids = property(get_object_ids)
+       
+       def set_value(self, value):
+               # Value must be a queryset. Watch out for ModelMultipleChoiceField;
+               # it returns its value as a list if empty.
+               
+               self.content_type = ContentType.objects.get_for_model(value.model)
+               
+               # Before we can fiddle with the many-to-many to foreignkeyvalues, we need
+               # a pk.
+               if self.pk is None:
+                       self.save()
+               
+               object_ids = value.values_list('id', flat=True)
+               
+               # These lines shouldn't be necessary; however, if object_ids is an EmptyQuerySet,
+               # the code (specifically the object_id__in query) won't work without them. Unclear why...
+               # TODO: is this still the case?
+               if not object_ids:
+                       self.values.all().delete()
+               else:
+                       self.values.exclude(object_id__in=object_ids, content_type=self.content_type).delete()
+                       
+                       current_ids = self.object_ids
+                       
+                       for object_id in object_ids:
+                               if object_id in current_ids:
+                                       continue
+                               self.values.create(content_type=self.content_type, object_id=object_id)
+       
+       def get_value(self):
+               if self.content_type is None:
+                       return None
+               
+               # HACK to be safely explicit until http://code.djangoproject.com/ticket/15145 is resolved
+               object_ids = self.object_ids
+               manager = self.content_type.model_class()._default_manager
+               if not object_ids:
+                       return manager.none()
+               return manager.filter(id__in=self.object_ids)
+       
+       value = property(get_value, set_value)
+       
+       def value_formfields(self):
+               field = self._meta.get_field('content_type')
+               fields = {field.name: field.formfield(initial=getattr(self.content_type, 'pk', None))}
+               
+               if self.content_type:
+                       kwargs = {
+                               'initial': self.object_ids,
+                               'required': False,
+                               'queryset': self.content_type.model_class()._default_manager.all()
+                       }
+                       fields['value'] = forms.ModelMultipleChoiceField(**kwargs)
+               return fields
+       
+       def construct_instance(self, **kwargs):
+               field_name = self._meta.get_field('content_type').name
+               ct = kwargs.pop(field_name, None)
+               if ct is None or ct != self.content_type:
+                       self.values.clear()
+                       self.content_type = ct
+               else:
+                       value = kwargs.get('value', None)
+                       if not value:
+                               value = self.content_type.model_class()._default_manager.none()
+                       self.set_value(value)
+       construct_instance.alters_data = True
+       
+       class Meta:
+               app_label = 'philo'
+
 
-class Relationship(models.Model):
-       entity_content_type = models.ForeignKey(ContentType, related_name='relationship_entity_set', verbose_name='Entity type')
+class Attribute(models.Model):
+       entity_content_type = models.ForeignKey(ContentType, related_name='attribute_entity_set', verbose_name='Entity type')
        entity_object_id = models.PositiveIntegerField(verbose_name='Entity ID')
        entity = generic.GenericForeignKey('entity_content_type', 'entity_object_id')
-       key = models.CharField(max_length=255)
-       value_content_type = models.ForeignKey(ContentType, related_name='relationship_value_set', limit_choices_to=value_content_type_limiter, verbose_name='Value type', null=True, blank=True)
+       
+       value_content_type = models.ForeignKey(ContentType, related_name='attribute_value_set', limit_choices_to=attribute_value_limiter, verbose_name='Value type', null=True, blank=True)
        value_object_id = models.PositiveIntegerField(verbose_name='Value ID', null=True, blank=True)
        value = generic.GenericForeignKey('value_content_type', 'value_object_id')
        
+       key = models.CharField(max_length=255, validators=[RegexValidator("\w+")], help_text="Must contain one or more alphanumeric characters or underscores.")
+       
        def __unicode__(self):
                return u'"%s": %s' % (self.key, self.value)
        
        class Meta:
                app_label = 'philo'
-               unique_together = ('key', 'entity_content_type', 'entity_object_id')
+               unique_together = (('key', 'entity_content_type', 'entity_object_id'), ('value_content_type', 'value_object_id'))
 
 
 class QuerySetMapper(object, DictMixin):
@@ -93,11 +244,15 @@ class QuerySetMapper(object, DictMixin):
        
        def __getitem__(self, key):
                try:
-                       return self.queryset.get(key__exact=key).value
+                       value = self.queryset.get(key__exact=key).value
                except ObjectDoesNotExist:
                        if self.passthrough is not None:
                                return self.passthrough.__getitem__(key)
                        raise KeyError
+               else:
+                       if value is not None:
+                               return value.value
+                       return value
        
        def keys(self):
                keys = set(self.queryset.values_list('key', flat=True).distinct())
@@ -131,72 +286,10 @@ class Entity(models.Model):
        __metaclass__ = EntityBase
        
        attribute_set = generic.GenericRelation(Attribute, content_type_field='entity_content_type', object_id_field='entity_object_id')
-       relationship_set = generic.GenericRelation(Relationship, content_type_field='entity_content_type', object_id_field='entity_object_id')
        
        @property
        def attributes(self):
-               return QuerySetMapper(self.attribute_set)
-       
-       @property
-       def relationships(self):
-               return QuerySetMapper(self.relationship_set)
-       
-       @property
-       def _added_attribute_registry(self):
-               if not hasattr(self, '_real_added_attribute_registry'):
-                       self._real_added_attribute_registry = {}
-               return self._real_added_attribute_registry
-       
-       @property
-       def _removed_attribute_registry(self):
-               if not hasattr(self, '_real_removed_attribute_registry'):
-                       self._real_removed_attribute_registry = []
-               return self._real_removed_attribute_registry
-       
-       @property
-       def _added_relationship_registry(self):
-               if not hasattr(self, '_real_added_relationship_registry'):
-                       self._real_added_relationship_registry = {}
-               return self._real_added_relationship_registry
-       
-       @property
-       def _removed_relationship_registry(self):
-               if not hasattr(self, '_real_removed_relationship_registry'):
-                       self._real_removed_relationship_registry = []
-               return self._real_removed_relationship_registry
-       
-       def save(self, *args, **kwargs):
-               super(Entity, self).save(*args, **kwargs)
-               
-               for key in self._removed_attribute_registry:
-                       self.attribute_set.filter(key__exact=key).delete()
-               del self._removed_attribute_registry[:]
-               
-               for key, value in self._added_attribute_registry.items():
-                       try:
-                               attribute = self.attribute_set.get(key__exact=key)
-                       except Attribute.DoesNotExist:
-                               attribute = Attribute()
-                               attribute.entity = self
-                               attribute.key = key
-                       attribute.value = value
-                       attribute.save()
-               self._added_attribute_registry.clear()
-               
-               for key in self._removed_relationship_registry:
-                       self.relationship_set.filter(key__exact=key).delete()
-               del self._removed_relationship_registry[:]
-               
-               for key, value in self._added_relationship_registry.items():
-                       try:
-                               relationship = self.relationship_set.get(key__exact=key)
-                       except Relationship.DoesNotExist:
-                               relationship = Relationship()
-                               relationship.entity = self
-                               relationship.key = key
-                       relationship.value = value
-                       relationship.save()
-               self._added_relationship_registry.clear()
+               return QuerySetMapper(self.attribute_set.all())
        
        class Meta:
                abstract = True
@@ -205,69 +298,131 @@ class Entity(models.Model):
 class TreeManager(models.Manager):
        use_for_related_fields = True
        
-       def roots(self):
-               return self.filter(parent__isnull=True)
-       
-       def get_with_path(self, path, root=None, absolute_result=True, pathsep='/'):
+       def get_with_path(self, path, root=None, absolute_result=True, pathsep='/', field='slug'):
                """
-               Returns the object with the path, or None if there is no object with that path,
-               unless absolute_result is set to False, in which case it returns a tuple containing
-               the deepest object found along the path, and the remainder of the path after that
-               object as a string (or None in the case that there is no remaining path).
+               Returns the object with the path, unless absolute_result is set to False, in which
+               case it returns a tuple containing the deepest object found along the path, and the
+               remainder of the path after that object as a string (or None if there is no remaining
+               path). Raises a DoesNotExist exception if no object is found with the given path.
+               
+               If the path you're searching for is known to exist, it is always faster to use
+               absolute_result=True - unless the path depth is over ~40, in which case the high cost
+               of the absolute query makes a binary search (i.e. non-absolute) faster.
                """
-               slugs = path.split(pathsep)
-               obj = root
-               remaining_slugs = list(slugs)
-               remainder = None
-               for slug in slugs:
-                       remaining_slugs.remove(slug)
-                       if slug: # ignore blank slugs, handles for multiple consecutive pathseps
+               # Note: SQLite allows max of 64 tables in one join. That means the binary search will
+               # only work on paths with a max depth of 127 and the absolute fetch will only work
+               # to a max depth of (surprise!) 63. Although this could be handled, chances are your
+               # tree structure won't be that deep.
+               segments = path.split(pathsep)
+               
+               # Clean out blank segments. Handles multiple consecutive pathseps.
+               while True:
+                       try:
+                               segments.remove('')
+                       except ValueError:
+                               break
+               
+               # Special-case a lack of segments. No queries necessary.
+               if not segments:
+                       if root is not None:
+                               if absolute_result:
+                                       return root
+                               return root, None
+                       else:
+                               raise self.model.DoesNotExist('%s matching query does not exist.' % self.model._meta.object_name)
+               
+               def make_query_kwargs(segments, root):
+                       kwargs = {}
+                       prefix = ""
+                       revsegs = list(segments)
+                       revsegs.reverse()
+                       
+                       for segment in revsegs:
+                               kwargs["%s%s__exact" % (prefix, field)] = segment
+                               prefix += "parent__"
+                       
+                       if prefix:
+                               kwargs[prefix[:-2]] = root
+                       
+                       return kwargs
+               
+               def find_obj(segments, depth, deepest_found=None):
+                       if deepest_found is None:
+                               deepest_level = 0
+                       elif root is None:
+                               deepest_level = deepest_found.get_level() + 1
+                       else:
+                               deepest_level = deepest_found.get_level() - root.get_level()
+                       try:
+                               obj = self.get(**make_query_kwargs(segments[deepest_level:depth], deepest_found or root))
+                       except self.model.DoesNotExist:
+                               if not deepest_level and depth > 1:
+                                       # make sure there's a root node...
+                                       depth = 1
+                               else:
+                                       # Try finding one with half the path since the deepest find.
+                                       depth = (deepest_level + depth)/2
+                               
+                               if deepest_level == depth:
+                                       # This should happen if nothing is found with any part of the given path.
+                                       if root is not None and deepest_found is None:
+                                               return root, pathsep.join(segments)
+                                       raise
+                               
+                               return find_obj(segments, depth, deepest_found)
+                       else:
+                               # Yay! Found one!
+                               if root is None:
+                                       deepest_level = obj.get_level() + 1
+                               else:
+                                       deepest_level = obj.get_level() - root.get_level()
+                               
+                               # Could there be a deeper one?
+                               if obj.is_leaf_node():
+                                       return obj, pathsep.join(segments[deepest_level:]) or None
+                               
+                               depth += (len(segments) - depth)/2 or len(segments) - depth
+                               
+                               if depth > deepest_level + obj.get_descendant_count():
+                                       depth = deepest_level + obj.get_descendant_count()
+                               
+                               if deepest_level == depth:
+                                       return obj, pathsep.join(segments[deepest_level:]) or None
+                               
                                try:
-                                       obj = self.get(slug__exact=slug, parent__exact=obj)
+                                       return find_obj(segments, depth, obj)
                                except self.model.DoesNotExist:
-                                       if absolute_result:
-                                               obj = None
-                                       remaining_slugs.insert(0, slug)
-                                       remainder = pathsep.join(remaining_slugs)
-                                       break
-               if obj:
-                       if absolute_result:
-                               return obj
-                       else:
-                               return (obj, remainder)
-               raise self.model.DoesNotExist('%s matching query does not exist.' % self.model._meta.object_name)
+                                       # Then this was the deepest.
+                                       return obj, pathsep.join(segments[deepest_level:])
+               
+               if absolute_result:
+                       return self.get(**make_query_kwargs(segments, root))
+               
+               # Try a modified binary search algorithm. Feed the root in so that query complexity
+               # can be reduced. It might be possible to weight the search towards the beginning
+               # of the path, since short paths are more likely, but how far forward? It would
+               # need to shift depending on len(segments) - perhaps logarithmically?
+               return find_obj(segments, len(segments)/2 or len(segments))
 
 
-class TreeModel(models.Model):
+class TreeModel(MPTTModel):
        objects = TreeManager()
        parent = models.ForeignKey('self', related_name='children', null=True, blank=True)
        slug = models.SlugField(max_length=255)
        
-       def has_ancestor(self, ancestor):
-               parent = self
-               while parent:
-                       if parent == ancestor:
-                               return True
-                       parent = parent.parent
-               return False
-       
        def get_path(self, root=None, pathsep='/', field='slug'):
+               if root == self:
+                       return ''
+               
+               if root is not None and not self.is_descendant_of(root):
+                       raise AncestorDoesNotExist(root)
+               
+               qs = self.get_ancestors(include_self=True)
+               
                if root is not None:
-                       if not self.has_ancestor(root):
-                               raise AncestorDoesNotExist(root)
-                       path = ''
-                       parent = self
-                       while parent and parent != root:
-                               path = getattr(parent, field, '?') + pathsep + path
-                               parent = parent.parent
-                       return path
-               else:
-                       path = getattr(self, field, '?')
-                       parent = self.parent
-                       while parent and parent != root:
-                               path = getattr(parent, field, '?') + pathsep + path
-                               parent = parent.parent
-                       return path
+                       qs = qs.filter(**{'%s__gt' % self._mptt_meta.level_attr: root.get_level()})
+               
+               return pathsep.join([getattr(parent, field, '?') for parent in qs])
        path = property(get_path)
        
        def __unicode__(self):
@@ -278,18 +433,22 @@ class TreeModel(models.Model):
                abstract = True
 
 
+class TreeEntityBase(MPTTModelBase, EntityBase):
+       def __new__(meta, name, bases, attrs):
+               attrs['_mptt_meta'] = MPTTOptions(attrs.pop('MPTTMeta', None))
+               cls = EntityBase.__new__(meta, name, bases, attrs)
+               
+               return meta.register(cls)
+
+
 class TreeEntity(Entity, TreeModel):
+       __metaclass__ = TreeEntityBase
+       
        @property
        def attributes(self):
                if self.parent:
-                       return QuerySetMapper(self.attribute_set, passthrough=self.parent.attributes)
+                       return QuerySetMapper(self.attribute_set.all(), passthrough=self.parent.attributes)
                return super(TreeEntity, self).attributes
        
-       @property
-       def relationships(self):
-               if self.parent:
-                       return QuerySetMapper(self.relationship_set, passthrough=self.parent.relationships)
-               return super(TreeEntity, self).relationships
-       
        class Meta:
                abstract = True
\ No newline at end of file