Added initial cut at serialization framework, along with some basic t…

…ests and a stab at some docs. This is all a bit rough right now, so expect some bumps.

git-svn-id: http://code.djangoproject.com/svn/django/trunk@3225 bcc190cf-cafb-0310-a4f2-bffc1f526a37

django/core/serializers/__init__.py

@@ -0,0 +1,76 @@
+"""
+Interfaces for serializing Django objects.
+
+Usage::
+
+    >>> from django.core import serializers
+    >>> json = serializers.serialize("json", some_query_set)
+    >>> objects = list(serializers.deserialize("json", json))
+        
+To add your own serializers, use the SERIALIZATION_MODULES setting::
+
+    SERIALIZATION_MODULES = {
+        "csv" : "path.to.csv.serializer",
+        "txt" : "path.to.txt.serializer",
+    }
+
+"""
+
+from django.conf import settings
+
+# Built-in serializers
+BUILTIN_SERIALIZERS = {
+    "xml" : "django.core.serializers.xml_serializer",
+}
+
+_serializers = {}
+
+def register_serializer(format, serializer_module):
+    """Register a new serializer by passing in a module name."""
+    module = __import__(serializer_module, '', '', [''])
+    _serializers[format] = module
+
+def unregister_serializer(format):
+    """Unregister a given serializer"""
+    del _serializers[format]
+
+def get_serializer(format):
+    if not _serializers:
+        _load_serializers()
+    return _serializers[format].Serializer
+
+def get_deserializer(format):
+    if not _serializers:
+        _load_serializers()
+    return _serializers[format].Deserializer
+
+def serialize(format, queryset, **options):
+    """
+    Serialize a queryset (or any iterator that returns database objects) using
+    a certain serializer.
+    """
+    s = get_serializer(format)()
+    s.serialize(queryset, **options)
+    return s.getvalue()
+
+def deserialize(format, stream_or_string):
+    """
+    Deserialize a stream or a string. Returns an iterator that yields ``(obj,
+    m2m_relation_dict)``, where ``obj`` is a instantiated -- but *unsaved* --
+    object, and ``m2m_relation_dict`` is a dictionary of ``{m2m_field_name :
+    list_of_related_objects}``.
+    """
+    d = get_deserializer(format)
+    return d(stream_or_string)
+
+def _load_serializers():
+    """
+    Register built-in and settings-defined serializers. This is done lazily so
+    that user code has a chance to (e.g.) set up custom settings without
+    needing to be careful of import order.
+    """
+    for format in BUILTIN_SERIALIZERS:
+        register_serializer(format, BUILTIN_SERIALIZERS[format])
+    if hasattr(settings, "SERIALIZATION_MODULES"):
+        for format in settings.SERIALIZATION_MODULES:
+            register_serializer(format, settings.SERIALIZATION_MODULES[format])

django/core/serializers/base.py

@@ -0,0 +1,159 @@
+"""
+Module for abstract serializer/unserializer base classes.
+"""
+
+try:
+    from cStringIO import StringIO
+except ImportError:
+    from StringIO import StringIO
+from django.db import models
+
+class SerializationError(Exception):
+    """Something bad happened during serialization."""
+    pass
+
+class DeserializationError(Exception):
+    """Something bad happened during deserialization."""
+    pass
+
+class Serializer(object):
+    """
+    Abstract serializer base class.
+    """
+
+    def serialize(self, queryset, **options):
+        """
+        Serialize a queryset.
+        """
+        self.options = options
+
+        self.stream = options.get("stream", StringIO())
+
+        self.start_serialization()
+        for obj in queryset:
+            self.start_object(obj)
+            for field in obj._meta.fields:
+                if field.rel is None:
+                    self.handle_field(obj, field)
+                else:
+                    self.handle_fk_field(obj, field)
+            for field in obj._meta.many_to_many:
+                self.handle_m2m_field(obj, field)
+            self.end_object(obj)
+        self.end_serialization()
+        return self.getvalue()
+
+    def get_string_value(self, obj, field):
+        """
+        Convert a field's value to a string.
+        """
+        if isinstance(field, models.DateTimeField):
+            value = getattr(obj, field.name).strftime("%Y-%m-%d %H:%M:%S")
+        elif isinstance(field, models.FileField):
+            value = getattr(obj, "get_%s_url" % field.name, lambda: None)()
+        else:
+            value = field.flatten_data(follow=None, obj=obj).get(field.name, "")
+        return str(value)
+
+    def start_serialization(self):
+        """
+        Called when serializing of the queryset starts.
+        """
+        raise NotImplementedError
+
+    def end_serialization(self):
+        """
+        Called when serializing of the queryset ends.
+        """
+        pass
+
+    def start_object(self, obj):
+        """
+        Called when serializing of an object starts.
+        """
+        raise NotImplementedError
+
+    def end_object(self, obj):
+        """
+        Called when serializing of an object ends.
+        """
+        pass
+
+    def handle_field(self, obj, field):
+        """
+        Called to handle each individual (non-relational) field on an object.
+        """
+        raise NotImplementedError
+
+    def handle_fk_field(self, obj, field):
+        """
+        Called to handle a ForeignKey field.
+        """
+        raise NotImplementedError
+
+    def handle_m2m_field(self, obj, field):
+        """
+        Called to handle a ManyToManyField.
+        """
+        raise NotImplementedError
+
+    def getvalue(self):
+        """
+        Return the fully serialized queryset.
+        """
+        return self.stream.getvalue()
+
+class Deserializer(object):
+    """
+    Abstract base deserializer class.
+    """
+
+    def __init__(self, stream_or_string, **options):
+        """
+        Init this serializer given a stream or a string
+        """
+        self.options = options
+        if isinstance(stream_or_string, basestring):
+            self.stream = StringIO(stream_or_string)
+        else:
+            self.stream = stream_or_string
+        # hack to make sure that the models have all been loaded before
+        # deserialization starts (otherwise subclass calls to get_model()
+        # and friends might fail...)
+        models.get_apps()
+
+    def __iter__(self):
+        return self
+
+    def next(self):
+        """Iteration iterface -- return the next item in the stream"""
+        raise NotImplementedError
+
+class DeserializedObject(object):
+    """
+    A deserialzed model.
+    
+    Basically a container for holding the pre-saved deserialized data along
+    with the many-to-many data saved with the object.
+    
+    Call ``save()`` to save the object (with the many-to-many data) to the
+    database; call ``save(save_m2m=False)`` to save just the object fields
+    (and not touch the many-to-many stuff.)
+    """
+
+    def __init__(self, obj, m2m_data=None):
+        self.object = obj
+        self.m2m_data = m2m_data
+
+    def __repr__(self):
+        return "<DeserializedObject: %s>" % str(self.object)
+
+    def save(self, save_m2m=True):
+        self.object.save()
+        if self.m2m_data and save_m2m:
+            for accessor_name, object_list in self.m2m_data.items():
+                setattr(self.object, accessor_name, object_list)
+
+        # prevent a second (possibly accidental) call to save() from saving 
+        # the m2m data twice.
+        self.m2m_data = None