Merge branch 'master' of github.com:MongoEngine/mongoengine into fix_baselist_marked_changed_bug

This commit is contained in:
Bastien Gérard
2018-12-15 20:36:42 +01:00
57 changed files with 2051 additions and 1062 deletions

View File

@@ -3,10 +3,10 @@ from mongoengine.errors import NotRegistered
__all__ = ('UPDATE_OPERATORS', 'get_document', '_document_registry')
UPDATE_OPERATORS = set(['set', 'unset', 'inc', 'dec', 'mul',
'pop', 'push', 'push_all', 'pull',
'pull_all', 'add_to_set', 'set_on_insert',
'min', 'max', 'rename'])
UPDATE_OPERATORS = {'set', 'unset', 'inc', 'dec', 'mul',
'pop', 'push', 'push_all', 'pull',
'pull_all', 'add_to_set', 'set_on_insert',
'min', 'max', 'rename'}
_document_registry = {}
@@ -19,7 +19,7 @@ def get_document(name):
# Possible old style name
single_end = name.split('.')[-1]
compound_end = '.%s' % single_end
possible_match = [k for k in _document_registry.keys()
possible_match = [k for k in _document_registry
if k.endswith(compound_end) or k == single_end]
if len(possible_match) == 1:
doc = _document_registry.get(possible_match.pop(), None)

View File

@@ -35,10 +35,9 @@ class BaseDict(dict):
_name = None
def __init__(self, dict_items, instance, name):
Document = _import_class('Document')
EmbeddedDocument = _import_class('EmbeddedDocument')
BaseDocument = _import_class('BaseDocument')
if isinstance(instance, (Document, EmbeddedDocument)):
if isinstance(instance, BaseDocument):
self._instance = weakref.proxy(instance)
self._name = name
super(BaseDict, self).__init__(dict_items)
@@ -56,11 +55,11 @@ class BaseDict(dict):
EmbeddedDocument = _import_class('EmbeddedDocument')
if isinstance(value, EmbeddedDocument) and value._instance is None:
value._instance = self._instance
elif not isinstance(value, BaseDict) and isinstance(value, dict):
elif isinstance(value, dict) and not isinstance(value, BaseDict):
value = BaseDict(value, None, '%s.%s' % (self._name, key))
super(BaseDict, self).__setitem__(key, value)
value._instance = self._instance
elif not isinstance(value, BaseList) and isinstance(value, list):
elif isinstance(value, list) and not isinstance(value, BaseList):
value = BaseList(value, None, '%s.%s' % (self._name, key))
super(BaseDict, self).__setitem__(key, value)
value._instance = self._instance
@@ -100,10 +99,9 @@ class BaseList(list):
_name = None
def __init__(self, list_items, instance, name):
Document = _import_class('Document')
EmbeddedDocument = _import_class('EmbeddedDocument')
BaseDocument = _import_class('BaseDocument')
if isinstance(instance, (Document, EmbeddedDocument)):
if isinstance(instance, BaseDocument):
self._instance = weakref.proxy(instance)
self._name = name
super(BaseList, self).__init__(list_items)
@@ -119,12 +117,12 @@ class BaseList(list):
EmbeddedDocument = _import_class('EmbeddedDocument')
if isinstance(value, EmbeddedDocument) and value._instance is None:
value._instance = self._instance
elif not isinstance(value, BaseDict) and isinstance(value, dict):
elif isinstance(value, dict) and not isinstance(value, BaseDict):
# Replace dict by BaseDict
value = BaseDict(value, None, '%s.%s' % (self._name, key))
super(BaseList, self).__setitem__(key, value)
value._instance = self._instance
elif not isinstance(value, BaseList) and isinstance(value, list):
elif isinstance(value, list) and not isinstance(value, BaseList):
# Replace list by BaseList
value = BaseList(value, None, '%s.%s' % (self._name, key))
super(BaseList, self).__setitem__(key, value)
@@ -218,6 +216,9 @@ class EmbeddedDocumentList(BaseList):
Filters the list by only including embedded documents with the
given keyword arguments.
This method only supports simple comparison (e.g: .filter(name='John Doe'))
and does not support operators like __gte, __lte, __icontains like queryset.filter does
:param kwargs: The keyword arguments corresponding to the fields to
filter on. *Multiple arguments are treated as if they are ANDed
together.*
@@ -358,7 +359,7 @@ class EmbeddedDocumentList(BaseList):
class StrictDict(object):
__slots__ = ()
_special_fields = set(['get', 'pop', 'iteritems', 'items', 'keys', 'create'])
_special_fields = {'get', 'pop', 'iteritems', 'items', 'keys', 'create'}
_classes = {}
def __init__(self, **kwargs):

View File

@@ -1,11 +1,8 @@
import copy
import numbers
from collections import Hashable
from functools import partial
from bson import ObjectId, json_util
from bson.dbref import DBRef
from bson.son import SON
from bson import DBRef, ObjectId, SON, json_util
import pymongo
import six
@@ -19,6 +16,7 @@ from mongoengine.base.fields import ComplexBaseField
from mongoengine.common import _import_class
from mongoengine.errors import (FieldDoesNotExist, InvalidDocumentError,
LookUpError, OperationError, ValidationError)
from mongoengine.python_support import Hashable
__all__ = ('BaseDocument', 'NON_FIELD_ERRORS')
@@ -302,7 +300,7 @@ class BaseDocument(object):
data['_cls'] = self._class_name
# only root fields ['test1.a', 'test2'] => ['test1', 'test2']
root_fields = set([f.split('.')[0] for f in fields])
root_fields = {f.split('.')[0] for f in fields}
for field_name in self:
if root_fields and field_name not in root_fields:
@@ -404,7 +402,15 @@ class BaseDocument(object):
@classmethod
def from_json(cls, json_data, created=False):
"""Converts json data to an unsaved document instance"""
"""Converts json data to a Document instance
:param json_data: The json data to load into the Document
:param created: If True, the document will be considered as a brand new document
If False and an id is provided, it will consider that the data being
loaded corresponds to what's already in the database (This has an impact of subsequent call to .save())
If False and no id is provided, it will consider the data as a new document
(default ``False``)
"""
return cls._from_son(json_util.loads(json_data), created=created)
def __expand_dynamic_values(self, name, value):
@@ -495,7 +501,13 @@ class BaseDocument(object):
self._changed_fields = []
def _nestable_types_changed_fields(self, changed_fields, key, data, inspected):
def _nestable_types_changed_fields(self, changed_fields, base_key, data):
"""Inspect nested data for changed fields
:param changed_fields: Previously collected changed fields
:param base_key: The base key that must be used to prepend changes to this data
:param data: data to inspect for changes
"""
# Loop list / dict fields as they contain documents
# Determine the iterator to use
if not hasattr(data, 'items'):
@@ -503,68 +515,60 @@ class BaseDocument(object):
else:
iterator = data.iteritems()
for index, value in iterator:
list_key = '%s%s.' % (key, index)
for index_or_key, value in iterator:
item_key = '%s%s.' % (base_key, index_or_key)
# don't check anything lower if this key is already marked
# as changed.
if list_key[:-1] in changed_fields:
if item_key[:-1] in changed_fields:
continue
if hasattr(value, '_get_changed_fields'):
changed = value._get_changed_fields(inspected)
changed_fields += ['%s%s' % (list_key, k)
for k in changed if k]
changed = value._get_changed_fields()
changed_fields += ['%s%s' % (item_key, k) for k in changed if k]
elif isinstance(value, (list, tuple, dict)):
self._nestable_types_changed_fields(
changed_fields, list_key, value, inspected)
changed_fields, item_key, value)
def _get_changed_fields(self, inspected=None):
def _get_changed_fields(self):
"""Return a list of all fields that have explicitly been changed.
"""
EmbeddedDocument = _import_class('EmbeddedDocument')
DynamicEmbeddedDocument = _import_class('DynamicEmbeddedDocument')
ReferenceField = _import_class('ReferenceField')
GenericReferenceField = _import_class('GenericReferenceField')
SortedListField = _import_class('SortedListField')
changed_fields = []
changed_fields += getattr(self, '_changed_fields', [])
inspected = inspected or set()
if hasattr(self, 'id') and isinstance(self.id, Hashable):
if self.id in inspected:
return changed_fields
inspected.add(self.id)
for field_name in self._fields_ordered:
db_field_name = self._db_field_map.get(field_name, field_name)
key = '%s.' % db_field_name
data = self._data.get(field_name, None)
field = self._fields.get(field_name)
if hasattr(data, 'id'):
if data.id in inspected:
continue
if isinstance(field, ReferenceField):
if db_field_name in changed_fields:
# Whole field already marked as changed, no need to go further
continue
elif (
isinstance(data, (EmbeddedDocument, DynamicEmbeddedDocument)) and
db_field_name not in changed_fields
):
if isinstance(field, ReferenceField): # Don't follow referenced documents
continue
if isinstance(data, EmbeddedDocument):
# Find all embedded fields that have been changed
changed = data._get_changed_fields(inspected)
changed = data._get_changed_fields()
changed_fields += ['%s%s' % (key, k) for k in changed if k]
elif (isinstance(data, (list, tuple, dict)) and
db_field_name not in changed_fields):
elif isinstance(data, (list, tuple, dict)):
if (hasattr(field, 'field') and
isinstance(field.field, ReferenceField)):
isinstance(field.field, (ReferenceField, GenericReferenceField))):
continue
elif isinstance(field, SortedListField) and field._ordering:
# if ordering is affected whole list is changed
if any(map(lambda d: field._ordering in d._changed_fields, data)):
if any(field._ordering in d._changed_fields for d in data):
changed_fields.append(db_field_name)
continue
self._nestable_types_changed_fields(
changed_fields, key, data, inspected)
changed_fields, key, data)
return changed_fields
def _delta(self):
@@ -576,7 +580,6 @@ class BaseDocument(object):
set_fields = self._get_changed_fields()
unset_data = {}
parts = []
if hasattr(self, '_changed_fields'):
set_data = {}
# Fetch each set item from its path
@@ -586,15 +589,13 @@ class BaseDocument(object):
new_path = []
for p in parts:
if isinstance(d, (ObjectId, DBRef)):
# Don't dig in the references
break
elif isinstance(d, list) and p.lstrip('-').isdigit():
if p[0] == '-':
p = str(len(d) + int(p))
try:
d = d[int(p)]
except IndexError:
d = None
elif isinstance(d, list) and p.isdigit():
# An item of a list (identified by its index) is updated
d = d[int(p)]
elif hasattr(d, 'get'):
# dict-like (dict, embedded document)
d = d.get(p)
new_path.append(p)
path = '.'.join(new_path)
@@ -606,26 +607,26 @@ class BaseDocument(object):
# Determine if any changed items were actually unset.
for path, value in set_data.items():
if value or isinstance(value, (numbers.Number, bool)):
if value or isinstance(value, (numbers.Number, bool)): # Account for 0 and True that are truthy
continue
# If we've set a value that ain't the default value don't unset it.
default = None
parts = path.split('.')
if (self._dynamic and len(parts) and parts[0] in
self._dynamic_fields):
del set_data[path]
unset_data[path] = 1
continue
elif path in self._fields:
# If we've set a value that ain't the default value don't unset it.
default = None
if path in self._fields:
default = self._fields[path].default
else: # Perform a full lookup for lists / embedded lookups
d = self
parts = path.split('.')
db_field_name = parts.pop()
for p in parts:
if isinstance(d, list) and p.lstrip('-').isdigit():
if p[0] == '-':
p = str(len(d) + int(p))
if isinstance(d, list) and p.isdigit():
d = d[int(p)]
elif (hasattr(d, '__getattribute__') and
not isinstance(d, dict)):
@@ -643,10 +644,9 @@ class BaseDocument(object):
default = None
if default is not None:
if callable(default):
default = default()
default = default() if callable(default) else default
if default != value:
if value != default:
continue
del set_data[path]
@@ -692,7 +692,7 @@ class BaseDocument(object):
fields = cls._fields
if not _auto_dereference:
fields = copy.copy(fields)
fields = copy.deepcopy(fields)
for field_name, field in fields.iteritems():
field._auto_dereference = _auto_dereference
@@ -1083,6 +1083,6 @@ class BaseDocument(object):
sep = getattr(field, 'display_sep', ' ')
values = value if field.__class__.__name__ in ('ListField', 'SortedListField') else [value]
return sep.join([
dict(field.choices).get(val, val)
six.text_type(dict(field.choices).get(val, val))
for val in values or []])
return value

View File

@@ -55,7 +55,7 @@ class BaseField(object):
field. Generally this is deprecated in favour of the
`FIELD.validate` method
:param choices: (optional) The valid choices
:param null: (optional) Is the field value can be null. If no and there is a default value
:param null: (optional) If the field value can be null. If no and there is a default value
then the default value is set
:param sparse: (optional) `sparse=True` combined with `unique=True` and `required=False`
means that uniqueness won't be enforced for `None` values
@@ -130,7 +130,6 @@ class BaseField(object):
def __set__(self, instance, value):
"""Descriptor for assigning a value to a field in a document.
"""
# If setting to None and there is a default
# Then set the value to the default value
if value is None:
@@ -267,13 +266,15 @@ class ComplexBaseField(BaseField):
ReferenceField = _import_class('ReferenceField')
GenericReferenceField = _import_class('GenericReferenceField')
EmbeddedDocumentListField = _import_class('EmbeddedDocumentListField')
dereference = (self._auto_dereference and
auto_dereference = instance._fields[self.name]._auto_dereference
dereference = (auto_dereference and
(self.field is None or isinstance(self.field,
(GenericReferenceField, ReferenceField))))
_dereference = _import_class('DeReference')()
self._auto_dereference = instance._fields[self.name]._auto_dereference
if instance._initialised and dereference and instance._data.get(self.name):
instance._data[self.name] = _dereference(
instance._data.get(self.name), max_depth=1, instance=instance,
@@ -294,7 +295,7 @@ class ComplexBaseField(BaseField):
value = BaseDict(value, instance, self.name)
instance._data[self.name] = value
if (self._auto_dereference and instance._initialised and
if (auto_dereference and instance._initialised and
isinstance(value, (BaseList, BaseDict)) and
not value._dereferenced):
value = _dereference(
@@ -313,11 +314,16 @@ class ComplexBaseField(BaseField):
if hasattr(value, 'to_python'):
return value.to_python()
BaseDocument = _import_class('BaseDocument')
if isinstance(value, BaseDocument):
# Something is wrong, return the value as it is
return value
is_list = False
if not hasattr(value, 'items'):
try:
is_list = True
value = {k: v for k, v in enumerate(value)}
value = {idx: v for idx, v in enumerate(value)}
except TypeError: # Not iterable return the value
return value
@@ -502,7 +508,7 @@ class GeoJsonBaseField(BaseField):
def validate(self, value):
"""Validate the GeoJson object based on its type."""
if isinstance(value, dict):
if set(value.keys()) == set(['type', 'coordinates']):
if set(value.keys()) == {'type', 'coordinates'}:
if value['type'] != self._type:
self.error('%s type must be "%s"' %
(self._name, self._type))

View File

@@ -18,14 +18,14 @@ class DocumentMetaclass(type):
"""Metaclass for all documents."""
# TODO lower complexity of this method
def __new__(cls, name, bases, attrs):
flattened_bases = cls._get_bases(bases)
super_new = super(DocumentMetaclass, cls).__new__
def __new__(mcs, name, bases, attrs):
flattened_bases = mcs._get_bases(bases)
super_new = super(DocumentMetaclass, mcs).__new__
# If a base class just call super
metaclass = attrs.get('my_metaclass')
if metaclass and issubclass(metaclass, DocumentMetaclass):
return super_new(cls, name, bases, attrs)
return super_new(mcs, name, bases, attrs)
attrs['_is_document'] = attrs.get('_is_document', False)
attrs['_cached_reference_fields'] = []
@@ -121,7 +121,8 @@ class DocumentMetaclass(type):
# inheritance of classes where inheritance is set to False
allow_inheritance = base._meta.get('allow_inheritance')
if not allow_inheritance and not base._meta.get('abstract'):
raise ValueError('Document %s may not be subclassed' %
raise ValueError('Document %s may not be subclassed. '
'To enable inheritance, use the "allow_inheritance" meta attribute.' %
base.__name__)
# Get superclasses from last base superclass
@@ -138,7 +139,7 @@ class DocumentMetaclass(type):
attrs['_types'] = attrs['_subclasses'] # TODO depreciate _types
# Create the new_class
new_class = super_new(cls, name, bases, attrs)
new_class = super_new(mcs, name, bases, attrs)
# Set _subclasses
for base in document_bases:
@@ -147,7 +148,7 @@ class DocumentMetaclass(type):
base._types = base._subclasses # TODO depreciate _types
(Document, EmbeddedDocument, DictField,
CachedReferenceField) = cls._import_classes()
CachedReferenceField) = mcs._import_classes()
if issubclass(new_class, Document):
new_class._collection = None
@@ -219,29 +220,26 @@ class DocumentMetaclass(type):
return new_class
def add_to_class(self, name, value):
setattr(self, name, value)
@classmethod
def _get_bases(cls, bases):
def _get_bases(mcs, bases):
if isinstance(bases, BasesTuple):
return bases
seen = []
bases = cls.__get_bases(bases)
bases = mcs.__get_bases(bases)
unique_bases = (b for b in bases if not (b in seen or seen.append(b)))
return BasesTuple(unique_bases)
@classmethod
def __get_bases(cls, bases):
def __get_bases(mcs, bases):
for base in bases:
if base is object:
continue
yield base
for child_base in cls.__get_bases(base.__bases__):
for child_base in mcs.__get_bases(base.__bases__):
yield child_base
@classmethod
def _import_classes(cls):
def _import_classes(mcs):
Document = _import_class('Document')
EmbeddedDocument = _import_class('EmbeddedDocument')
DictField = _import_class('DictField')
@@ -254,9 +252,9 @@ class TopLevelDocumentMetaclass(DocumentMetaclass):
collection in the database.
"""
def __new__(cls, name, bases, attrs):
flattened_bases = cls._get_bases(bases)
super_new = super(TopLevelDocumentMetaclass, cls).__new__
def __new__(mcs, name, bases, attrs):
flattened_bases = mcs._get_bases(bases)
super_new = super(TopLevelDocumentMetaclass, mcs).__new__
# Set default _meta data if base class, otherwise get user defined meta
if attrs.get('my_metaclass') == TopLevelDocumentMetaclass:
@@ -319,7 +317,7 @@ class TopLevelDocumentMetaclass(DocumentMetaclass):
not parent_doc_cls._meta.get('abstract', False)):
msg = 'Abstract document cannot have non-abstract base'
raise ValueError(msg)
return super_new(cls, name, bases, attrs)
return super_new(mcs, name, bases, attrs)
# Merge base class metas.
# Uses a special MetaDict that handles various merging rules
@@ -360,7 +358,7 @@ class TopLevelDocumentMetaclass(DocumentMetaclass):
attrs['_meta'] = meta
# Call super and get the new class
new_class = super_new(cls, name, bases, attrs)
new_class = super_new(mcs, name, bases, attrs)
meta = new_class._meta
@@ -394,7 +392,7 @@ class TopLevelDocumentMetaclass(DocumentMetaclass):
'_auto_id_field', False)
if not new_class._meta.get('id_field'):
# After 0.10, find not existing names, instead of overwriting
id_name, id_db_name = cls.get_auto_id_names(new_class)
id_name, id_db_name = mcs.get_auto_id_names(new_class)
new_class._auto_id_field = True
new_class._meta['id_field'] = id_name
new_class._fields[id_name] = ObjectIdField(db_field=id_db_name)
@@ -419,7 +417,7 @@ class TopLevelDocumentMetaclass(DocumentMetaclass):
return new_class
@classmethod
def get_auto_id_names(cls, new_class):
def get_auto_id_names(mcs, new_class):
id_name, id_db_name = ('id', '_id')
if id_name not in new_class._fields and \
id_db_name not in (v.db_field for v in new_class._fields.values()):

22
mongoengine/base/utils.py Normal file
View File

@@ -0,0 +1,22 @@
import re
class LazyRegexCompiler(object):
"""Descriptor to allow lazy compilation of regex"""
def __init__(self, pattern, flags=0):
self._pattern = pattern
self._flags = flags
self._compiled_regex = None
@property
def compiled_regex(self):
if self._compiled_regex is None:
self._compiled_regex = re.compile(self._pattern, self._flags)
return self._compiled_regex
def __get__(self, instance, owner):
return self.compiled_regex
def __set__(self, instance, value):
raise AttributeError("Can not set attribute LazyRegexCompiler")