Prepare for 1.1.0 (#30)

* refactor the view inspection process to be more modular and allow recursive customization
* add operation_id argument to @swagger_auto_
* add inspections for min/max validators
* add support for URLPathVersioning and NamespaceVersioning
* integrate with djangorestframework-camel-case
* fix bugs, improve tests and documentation
This commit is contained in:
Cristi Vîjdea
2017-12-26 22:50:59 +01:00
committed by GitHub
parent 73adc49b2c
commit c89f96fcb0
52 changed files with 2167 additions and 854 deletions
+26 -1
View File
@@ -2,6 +2,26 @@ from django.conf import settings
from rest_framework.settings import perform_import
SWAGGER_DEFAULTS = {
'DEFAULT_AUTO_SCHEMA_CLASS': 'drf_yasg.inspectors.SwaggerAutoSchema',
'DEFAULT_FIELD_INSPECTORS': [
'drf_yasg.inspectors.CamelCaseJSONFilter',
'drf_yasg.inspectors.ReferencingSerializerInspector',
'drf_yasg.inspectors.RelatedFieldInspector',
'drf_yasg.inspectors.ChoiceFieldInspector',
'drf_yasg.inspectors.FileFieldInspector',
'drf_yasg.inspectors.DictFieldInspector',
'drf_yasg.inspectors.SimpleFieldInspector',
'drf_yasg.inspectors.StringDefaultFieldInspector',
],
'DEFAULT_FILTER_INSPECTORS': [
'drf_yasg.inspectors.CoreAPICompatInspector',
],
'DEFAULT_PAGINATOR_INSPECTORS': [
'drf_yasg.inspectors.DjangoRestResponsePagination',
'drf_yasg.inspectors.CoreAPICompatInspector',
],
'USE_SESSION_AUTH': True,
'SECURITY_DEFINITIONS': {
'basic': {
@@ -28,7 +48,12 @@ REDOC_DEFAULTS = {
'PATH_IN_MIDDLE': False,
}
IMPORT_STRINGS = []
IMPORT_STRINGS = [
'DEFAULT_AUTO_SCHEMA_CLASS',
'DEFAULT_FIELD_INSPECTORS',
'DEFAULT_FILTER_INSPECTORS',
'DEFAULT_PAGINATOR_INSPECTORS',
]
class AppSettings(object):
+23 -2
View File
@@ -98,6 +98,9 @@ class OpenAPICodecJson(_OpenAPICodec):
return json.dumps(spec)
YAML_MAP_TAG = u'tag:yaml.org,2002:map'
class SaneYamlDumper(yaml.SafeDumper):
"""YamlDumper class usable for dumping ``OrderedDict`` and list instances in a standard way."""
@@ -122,7 +125,7 @@ class SaneYamlDumper(yaml.SafeDumper):
To use yaml.safe_dump(), you need the following.
"""
tag = u'tag:yaml.org,2002:map'
tag = YAML_MAP_TAG
value = []
node = yaml.MappingNode(tag, value, flow_style=flow_style)
if dump.alias_key is not None:
@@ -158,7 +161,7 @@ def yaml_sane_dump(data, binary):
* list elements are indented into their parents
* YAML references/aliases are disabled
:param dict data: the data to be serializers
:param dict data: the data to be dumped
:param bool binary: True to return a utf-8 encoded binary object, False to return a string
:return: the serialized YAML
:rtype: str,bytes
@@ -166,6 +169,24 @@ def yaml_sane_dump(data, binary):
return yaml.dump(data, Dumper=SaneYamlDumper, default_flow_style=False, encoding='utf-8' if binary else None)
class SaneYamlLoader(yaml.SafeLoader):
def construct_odict(self, node, deep=False):
self.flatten_mapping(node)
return OrderedDict(self.construct_pairs(node))
SaneYamlLoader.add_constructor(YAML_MAP_TAG, SaneYamlLoader.construct_odict)
def yaml_sane_load(stream):
"""Load the given YAML stream while preserving the input order for mapping items.
:param stream: YAML stream (can be a string or a file-like object)
:rtype: OrderedDict
"""
return yaml.load(stream, Loader=SaneYamlLoader)
class OpenAPICodecYaml(_OpenAPICodec):
media_type = 'application/yaml'
+120 -27
View File
@@ -2,12 +2,15 @@ import re
from collections import defaultdict, OrderedDict
import uritemplate
from django.utils.encoding import force_text
from rest_framework import versioning
from rest_framework.schemas.generators import SchemaGenerator, EndpointEnumerator as _EndpointEnumerator
from rest_framework.schemas.inspectors import get_pk_description
from . import openapi
from .inspectors import SwaggerAutoSchema
from .app_settings import swagger_settings
from .inspectors.field import get_queryset_field, get_basic_type_info
from .openapi import ReferenceResolver
from .utils import inspect_model_field, get_model_field
PATH_PARAMETER_RE = re.compile(r'{(?P<parameter>\w+)}')
@@ -52,7 +55,7 @@ class EndpointEnumerator(_EndpointEnumerator):
class OpenAPISchemaGenerator(object):
"""
This class iterates over all registered API endpoints and returns an appropriate OpenAPI 2.0 compliant schema.
Method implementations shamelessly stolen and adapted from rest_framework SchemaGenerator.
Method implementations shamelessly stolen and adapted from rest-framework ``SchemaGenerator``.
"""
endpoint_enumerator_class = EndpointEnumerator
@@ -70,10 +73,14 @@ class OpenAPISchemaGenerator(object):
self.info = info
self.version = version
def get_schema(self, request=None, public=False):
"""Generate an :class:`.Swagger` representing the API schema.
@property
def url(self):
return self._gen.url
:param rest_framework.request.Request request: the request used for filtering
def get_schema(self, request=None, public=False):
"""Generate a :class:`.Swagger` object representing the API schema.
:param Request request: the request used for filtering
accesible endpoints and finding the spec URI
:param bool public: if True, all endpoints are included regardless of access through `request`
@@ -81,10 +88,11 @@ class OpenAPISchemaGenerator(object):
:rtype: openapi.Swagger
"""
endpoints = self.get_endpoints(request)
endpoints = self.replace_version(endpoints, request)
components = ReferenceResolver(openapi.SCHEMA_DEFINITIONS)
paths = self.get_paths(endpoints, components, public)
paths = self.get_paths(endpoints, components, request, public)
url = self._gen.url
url = self.url
if not url and request is not None:
url = request.build_absolute_uri()
@@ -102,16 +110,40 @@ class OpenAPISchemaGenerator(object):
:return: the view instance
"""
view = self._gen.create_view(callback, method, request)
overrides = getattr(callback, 'swagger_auto_schema', None)
overrides = getattr(callback, '_swagger_auto_schema', None)
if overrides is not None:
# decorated function based view must have its decorator information passed on to the re-instantiated view
for method, _ in overrides.items():
view_method = getattr(view, method, None)
if view_method is not None: # pragma: no cover
setattr(view_method.__func__, 'swagger_auto_schema', overrides)
setattr(view_method.__func__, '_swagger_auto_schema', overrides)
return view
def get_endpoints(self, request=None):
def replace_version(self, endpoints, request):
"""If ``request.version`` is not ``None``, replace the version parameter in the path of any endpoints using
``URLPathVersioning`` as a versioning class.
:param dict endpoints: endpoints as returned by :meth:`.get_endpoints`
:param Request request: the request made against the schema view
:return: endpoints with modified paths
"""
version = getattr(request, 'version', None)
if version is None:
return endpoints
new_endpoints = {}
for path, endpoint in endpoints.items():
view_cls = endpoint[0]
versioning_class = getattr(view_cls, 'versioning_class', None)
version_param = getattr(versioning_class, 'version_param', 'version')
if versioning_class is not None and issubclass(versioning_class, versioning.URLPathVersioning):
path = path.replace('{%s}' % version_param, version)
new_endpoints[path] = endpoint
return new_endpoints
def get_endpoints(self, request):
"""Iterate over all the registered endpoints in the API and return a fake view with the right parameters.
:param rest_framework.request.Request request: request to bind to the endpoint views
@@ -131,9 +163,7 @@ class OpenAPISchemaGenerator(object):
return {path: (view_cls[path], methods) for path, methods in view_paths.items()}
def get_operation_keys(self, subpath, method, view):
"""Return a list of keys that should be used to group an operation within the specification.
::
"""Return a list of keys that should be used to group an operation within the specification. ::
/users/ ("users", "list"), ("users", "create")
/users/{pk}/ ("users", "read"), ("users", "update"), ("users", "delete")
@@ -149,39 +179,94 @@ class OpenAPISchemaGenerator(object):
"""
return self._gen.get_keys(subpath, method, view)
def get_paths(self, endpoints, components, public):
def determine_path_prefix(self, paths):
"""
Given a list of all paths, return the common prefix which should be
discounted when generating a schema structure.
This will be the longest common string that does not include that last
component of the URL, or the last component before a path parameter.
For example: ::
/api/v1/users/
/api/v1/users/{pk}/
The path prefix is ``/api/v1/``.
:param list[str] paths: list of paths
:rtype: str
"""
return self._gen.determine_path_prefix(paths)
def get_paths(self, endpoints, components, request, public):
"""Generate the Swagger Paths for the API from the given endpoints.
:param dict endpoints: endpoints as returned by get_endpoints
:param ReferenceResolver components: resolver/container for Swagger References
:param Request request: the request made against the schema view; can be None
:param bool public: if True, all endpoints are included regardless of access through `request`
:rtype: openapi.Paths
"""
if not endpoints:
return openapi.Paths(paths={})
prefix = self._gen.determine_path_prefix(endpoints.keys())
prefix = self.determine_path_prefix(list(endpoints.keys()))
paths = OrderedDict()
default_schema_cls = SwaggerAutoSchema
for path, (view_cls, methods) in sorted(endpoints.items()):
path_parameters = self.get_path_parameters(path, view_cls)
operations = {}
for method, view in methods:
if not public and not self._gen.has_view_permissions(path, method, view):
continue
operation_keys = self.get_operation_keys(path[len(prefix):], method, view)
overrides = self.get_overrides(view, method)
auto_schema_cls = overrides.get('auto_schema', default_schema_cls)
schema = auto_schema_cls(view, path, method, overrides, components)
operations[method.lower()] = schema.get_operation(operation_keys)
operations[method.lower()] = self.get_operation(view, path, prefix, method, components, request)
if operations:
paths[path] = openapi.PathItem(parameters=path_parameters, **operations)
paths[path] = self.get_path_item(path, view_cls, operations)
return openapi.Paths(paths=paths)
def get_operation(self, view, path, prefix, method, components, request):
"""Get an :class:`.Operation` for the given API endpoint (path, method). This method delegates to
:meth:`~.inspectors.ViewInspector.get_operation` of a :class:`~.inspectors.ViewInspector` determined
according to settings and :func:`@swagger_auto_schema <.swagger_auto_schema>` overrides.
:param view: the view associated with this endpoint
:param str path: the path component of the operation URL
:param str prefix: common path prefix among all endpoints
:param str method: the http method of the operation
:param openapi.ReferenceResolver components: referenceable components
:param Request request: the request made against the schema view; can be None
:rtype: openapi.Operation
"""
operation_keys = self.get_operation_keys(path[len(prefix):], method, view)
overrides = self.get_overrides(view, method)
# the inspector class can be specified, in decreasing order of priorty,
# 1. globaly via DEFAULT_AUTO_SCHEMA_CLASS
view_inspector_cls = swagger_settings.DEFAULT_AUTO_SCHEMA_CLASS
# 2. on the view/viewset class
view_inspector_cls = getattr(view, 'swagger_schema', view_inspector_cls)
# 3. on the swagger_auto_schema decorator
view_inspector_cls = overrides.get('auto_schema', view_inspector_cls)
view_inspector = view_inspector_cls(view, path, method, components, request, overrides)
return view_inspector.get_operation(operation_keys)
def get_path_item(self, path, view_cls, operations):
"""Get a :class:`.PathItem` object that describes the parameters and operations related to a single path in the
API.
:param str path: the path
:param type view_cls: the view that was bound to this path in urlpatterns
:param dict[str,openapi.Operation] operations: operations defined on this path, keyed by lowercase HTTP method
:rtype: openapi.PathItem
"""
path_parameters = self.get_path_parameters(path, view_cls)
return openapi.PathItem(parameters=path_parameters, **operations)
def get_overrides(self, view, method):
"""Get overrides specified for a given operation.
@@ -193,7 +278,7 @@ class OpenAPISchemaGenerator(object):
method = method.lower()
action = getattr(view, 'action', method)
action_method = getattr(view, action, None)
overrides = getattr(action_method, 'swagger_auto_schema', {})
overrides = getattr(action_method, '_swagger_auto_schema', {})
if method in overrides:
overrides = overrides[method]
@@ -212,13 +297,21 @@ class OpenAPISchemaGenerator(object):
model = getattr(getattr(view_cls, 'queryset', None), 'model', None)
for variable in uritemplate.variables(path):
model, model_field = get_model_field(queryset, variable)
attrs = inspect_model_field(model, model_field)
model, model_field = get_queryset_field(queryset, variable)
attrs = get_basic_type_info(model_field) or {'type': openapi.TYPE_STRING}
if hasattr(view_cls, 'lookup_value_regex') and getattr(view_cls, 'lookup_field', None) == variable:
attrs['pattern'] = view_cls.lookup_value_regex
if model_field and model_field.help_text:
description = force_text(model_field.help_text)
elif model_field and model_field.primary_key:
description = get_pk_description(model, model_field)
else:
description = None
field = openapi.Parameter(
name=variable,
description=description,
required=True,
in_=openapi.IN_PATH,
**attrs
+38
View File
@@ -0,0 +1,38 @@
from .base import (
BaseInspector, ViewInspector, FilterInspector, PaginatorInspector,
FieldInspector, SerializerInspector, NotHandled
)
from .field import (
InlineSerializerInspector, ReferencingSerializerInspector, RelatedFieldInspector, SimpleFieldInspector,
FileFieldInspector, ChoiceFieldInspector, DictFieldInspector, StringDefaultFieldInspector,
CamelCaseJSONFilter
)
from .query import (
CoreAPICompatInspector, DjangoRestResponsePagination
)
from .view import SwaggerAutoSchema
from ..app_settings import swagger_settings
# these settings must be accesed only after definig/importing all the classes in this module to avoid ImportErrors
ViewInspector.field_inspectors = swagger_settings.DEFAULT_FIELD_INSPECTORS
ViewInspector.filter_inspectors = swagger_settings.DEFAULT_FILTER_INSPECTORS
ViewInspector.paginator_inspectors = swagger_settings.DEFAULT_PAGINATOR_INSPECTORS
__all__ = [
# base inspectors
'BaseInspector', 'FilterInspector', 'PaginatorInspector', 'FieldInspector', 'SerializerInspector', 'ViewInspector',
# filter and pagination inspectors
'CoreAPICompatInspector', 'DjangoRestResponsePagination',
# field inspectors
'InlineSerializerInspector', 'ReferencingSerializerInspector', 'RelatedFieldInspector', 'SimpleFieldInspector',
'FileFieldInspector', 'ChoiceFieldInspector', 'DictFieldInspector', 'StringDefaultFieldInspector',
'CamelCaseJSONFilter',
# view inspectors
'SwaggerAutoSchema',
# module constants
'NotHandled',
]
+406
View File
@@ -0,0 +1,406 @@
import inspect
import logging
from django.utils.encoding import force_text
from rest_framework import serializers
from rest_framework.utils import json, encoders
from rest_framework.viewsets import GenericViewSet
from .. import openapi
from ..utils import is_list_view
#: Sentinel value that inspectors must return to signal that they do not know how to handle an object
NotHandled = object()
logger = logging.getLogger(__name__)
class BaseInspector(object):
def __init__(self, view, path, method, components, request):
"""
:param view: the view associated with this endpoint
:param str path: the path component of the operation URL
:param str method: the http method of the operation
:param openapi.ReferenceResolver components: referenceable components
:param Request request: the request made against the schema view; can be None
"""
self.view = view
self.path = path
self.method = method
self.components = components
self.request = request
def process_result(self, result, method_name, obj, **kwargs):
"""After an inspector handles an object (i.e. returns a value other than :data:`.NotHandled`), all inspectors
that were probed get the chance to alter the result, in reverse order. The inspector that handled the object
is the first to receive a ``process_result`` call with the object it just returned.
This behaviour is similar to the Django request/response middleware processing.
If this inspector has no post-processing to do, it should just ``return result`` (the default implementation).
:param result: the return value of the winning inspector, or ``None`` if no inspector handled the object
:param str method_name: name of the method that was called on the inspector
:param obj: first argument passed to inspector method
:param kwargs: additional arguments passed to inspector method
:return:
"""
return result
def probe_inspectors(self, inspectors, method_name, obj, initkwargs=None, **kwargs):
"""Probe a list of inspectors with a given object. The first inspector in the list to return a value that
is not :data:`.NotHandled` wins.
:param list[type[BaseInspector]] inspectors: list of inspectors to probe
:param str method_name: name of the target method on the inspector
:param obj: first argument to inspector method
:param dict initkwargs: extra kwargs for instantiating inspector class
:param kwargs: additional arguments to inspector method
:return: the return value of the winning inspector, or ``None`` if no inspector handled the object
"""
initkwargs = initkwargs or {}
tried_inspectors = []
for inspector in inspectors:
assert inspect.isclass(inspector), "inspector must be a class, not an object"
assert issubclass(inspector, BaseInspector), "inspectors must subclass BaseInspector"
inspector = inspector(self.view, self.path, self.method, self.components, self.request, **initkwargs)
tried_inspectors.append(inspector)
method = getattr(inspector, method_name, None)
if method is None:
continue
result = method(obj, **kwargs)
if result is not NotHandled:
break
else: # pragma: no cover
logger.warning("%s ignored because no inspector in %s handled it (operation: %s)",
obj, inspectors, method_name)
result = None
for inspector in reversed(tried_inspectors):
result = inspector.process_result(result, method_name, obj, **kwargs)
return result
class PaginatorInspector(BaseInspector):
"""Base inspector for paginators.
Responisble for determining extra query parameters and response structure added by given paginators.
"""
def get_paginator_parameters(self, paginator):
"""Get the pagination parameters for a single paginator **instance**.
Should return :data:`.NotHandled` if this inspector does not know how to handle the given `paginator`.
:param BasePagination paginator: the paginator
:rtype: list[openapi.Parameter]
"""
return NotHandled
def get_paginated_response(self, paginator, response_schema):
"""Add appropriate paging fields to a response :class:`.Schema`.
Should return :data:`.NotHandled` if this inspector does not know how to handle the given `paginator`.
:param BasePagination paginator: the paginator
:param openapi.Schema response_schema: the response schema that must be paged.
:rtype: openapi.Schema
"""
return NotHandled
class FilterInspector(BaseInspector):
"""Base inspector for filter backends.
Responsible for determining extra query parameters added by given filter backends.
"""
def get_filter_parameters(self, filter_backend):
"""Get the filter parameters for a single filter backend **instance**.
Should return :data:`.NotHandled` if this inspector does not know how to handle the given `filter_backend`.
:param BaseFilterBackend filter_backend: the filter backend
:rtype: list[openapi.Parameter]
"""
return NotHandled
class FieldInspector(BaseInspector):
"""Base inspector for serializers and serializer fields. """
def __init__(self, view, path, method, components, request, field_inspectors):
super(FieldInspector, self).__init__(view, path, method, components, request)
self.field_inspectors = field_inspectors
def field_to_swagger_object(self, field, swagger_object_type, use_references, **kwargs):
"""Convert a drf Serializer or Field instance into a Swagger object.
Should return :data:`.NotHandled` if this inspector does not know how to handle the given `field`.
:param rest_framework.serializers.Field field: the source field
:param type[openapi.SwaggerDict] swagger_object_type: should be one of Schema, Parameter, Items
:param bool use_references: if False, forces all objects to be declared inline
instead of by referencing other components
:param kwargs: extra attributes for constructing the object;
if swagger_object_type is Parameter, ``name`` and ``in_`` should be provided
:return: the swagger object
:rtype: openapi.Parameter,openapi.Items,openapi.Schema,openapi.SchemaRef
"""
return NotHandled
def probe_field_inspectors(self, field, swagger_object_type, use_references, **kwargs):
"""Helper method for recursively probing `field_inspectors` to handle a given field.
All arguments are the same as :meth:`.field_to_swagger_object`.
:rtype: openapi.Parameter,openapi.Items,openapi.Schema,openapi.SchemaRef
"""
return self.probe_inspectors(
self.field_inspectors, 'field_to_swagger_object', field, {'field_inspectors': self.field_inspectors},
swagger_object_type=swagger_object_type, use_references=use_references, **kwargs
)
def _get_partial_types(self, field, swagger_object_type, use_references, **kwargs):
"""Helper method to extract generic information from a field and return a partial constructor for the
appropriate openapi object.
All arguments are the same as :meth:`.field_to_swagger_object`.
The return value is a tuple consisting of:
* a function for constructing objects of `swagger_object_type`; its prototype is: ::
def SwaggerType(existing_object=None, **instance_kwargs):
This function creates an instance of `swagger_object_type`, passing the following attributes to its init,
in order of precedence:
- arguments specified by the ``kwargs`` parameter of :meth:`._get_partial_types`
- ``instance_kwargs`` passed to the constructor function
- ``title``, ``description``, ``required``, ``default`` and ``read_only`` inferred from the field,
where appropriate
If ``existing_object`` is not ``None``, it is updated instead of creating a new object.
* a type that should be used for child objects if `field` is of an array type. This can currently have two
values:
- :class:`.Schema` if `swagger_object_type` is :class:`.Schema`
- :class:`.Items` if `swagger_object_type` is :class:`.Parameter` or :class:`.Items`
:rtype: tuple[callable,(type[openapi.Schema],type[openapi.Items])]
"""
assert swagger_object_type in (openapi.Schema, openapi.Parameter, openapi.Items)
assert not isinstance(field, openapi.SwaggerDict), "passed field is already a SwaggerDict object"
title = force_text(field.label) if field.label else None
title = title if swagger_object_type == openapi.Schema else None # only Schema has title
description = force_text(field.help_text) if field.help_text else None
description = description if swagger_object_type != openapi.Items else None # Items has no description either
def SwaggerType(existing_object=None, **instance_kwargs):
if 'required' not in instance_kwargs and swagger_object_type == openapi.Parameter:
instance_kwargs['required'] = field.required
if 'default' not in instance_kwargs and swagger_object_type != openapi.Items:
default = getattr(field, 'default', serializers.empty)
if default is not serializers.empty:
if callable(default):
try:
if hasattr(default, 'set_context'):
default.set_context(field)
default = default()
except Exception: # pragma: no cover
logger.warning("default for %s is callable but it raised an exception when "
"called; 'default' field will not be added to schema", field, exc_info=True)
default = None
if default is not None:
try:
default = field.to_representation(default)
# JSON roundtrip ensures that the value is valid JSON;
# for example, sets and tuples get transformed into lists
default = json.loads(json.dumps(default, cls=encoders.JSONEncoder))
except Exception: # pragma: no cover
logger.warning("'default' on schema for %s will not be set because "
"to_representation raised an exception", field, exc_info=True)
default = None
if default is not None:
instance_kwargs['default'] = default
if 'read_only' not in instance_kwargs and swagger_object_type == openapi.Schema:
# TODO: read_only is only relevant for schema `properties` - should not be generated in other cases
if field.read_only:
instance_kwargs['read_only'] = True
instance_kwargs.setdefault('title', title)
instance_kwargs.setdefault('description', description)
instance_kwargs.update(kwargs)
if existing_object is not None:
assert isinstance(existing_object, swagger_object_type)
for attr, val in sorted(instance_kwargs.items()):
setattr(existing_object, attr, val)
return existing_object
return swagger_object_type(**instance_kwargs)
# arrays in Schema have Schema elements, arrays in Parameter and Items have Items elements
child_swagger_type = openapi.Schema if swagger_object_type == openapi.Schema else openapi.Items
return SwaggerType, child_swagger_type
class SerializerInspector(FieldInspector):
def get_schema(self, serializer):
"""Convert a DRF Serializer instance to an :class:`.openapi.Schema`.
Should return :data:`.NotHandled` if this inspector does not know how to handle the given `serializer`.
:param serializers.BaseSerializer serializer: the ``Serializer`` instance
:rtype: openapi.Schema
"""
return NotHandled
def get_request_parameters(self, serializer, in_):
"""Convert a DRF serializer into a list of :class:`.Parameter`\ s.
Should return :data:`.NotHandled` if this inspector does not know how to handle the given `serializer`.
:param serializers.BaseSerializer serializer: the ``Serializer`` instance
:param str in_: the location of the parameters, one of the `openapi.IN_*` constants
:rtype: list[openapi.Parameter]
"""
return NotHandled
class ViewInspector(BaseInspector):
body_methods = ('PUT', 'PATCH', 'POST') #: methods that are allowed to have a request body
# real values set in __init__ to prevent import errors
field_inspectors = [] #:
filter_inspectors = [] #:
paginator_inspectors = [] #:
def __init__(self, view, path, method, components, request, overrides):
"""
Inspector class responsible for providing :class:`.Operation` definitions given a view, path and method.
:param dict overrides: manual overrides as passed to :func:`@swagger_auto_schema <.swagger_auto_schema>`
"""
super(ViewInspector, self).__init__(view, path, method, components, request)
self.overrides = overrides
self._prepend_inspector_overrides('field_inspectors')
self._prepend_inspector_overrides('filter_inspectors')
self._prepend_inspector_overrides('paginator_inspectors')
def _prepend_inspector_overrides(self, inspectors):
extra_inspectors = self.overrides.get(inspectors, None)
if extra_inspectors:
default_inspectors = [insp for insp in getattr(self, inspectors) if insp not in extra_inspectors]
setattr(self, inspectors, extra_inspectors + default_inspectors)
def get_operation(self, operation_keys):
"""Get an :class:`.Operation` for the given API endpoint (path, method).
This includes query, body parameters and response schemas.
:param tuple[str] operation_keys: an array of keys describing the hierarchical layout of this view in the API;
e.g. ``('snippets', 'list')``, ``('snippets', 'retrieve')``, etc.
:rtype: openapi.Operation
"""
raise NotImplementedError("ViewInspector must implement get_operation()!")
# methods below provided as default implementations for probing inspectors
def should_filter(self):
"""Determine whether filter backend parameters should be included for this request.
:rtype: bool
"""
if not getattr(self.view, 'filter_backends', None):
return False
if self.method.lower() not in ["get", "delete"]:
return False
if not isinstance(self.view, GenericViewSet):
return True
return is_list_view(self.path, self.method, self.view)
def get_filter_parameters(self):
"""Return the parameters added to the view by its filter backends.
:rtype: list[openapi.Parameter]
"""
if not self.should_filter():
return []
fields = []
for filter_backend in self.view.filter_backends:
fields += self.probe_inspectors(self.filter_inspectors, 'get_filter_parameters', filter_backend()) or []
return fields
def should_page(self):
"""Determine whether paging parameters and structure should be added to this operation's request and response.
:rtype: bool
"""
if not hasattr(self.view, 'paginator'):
return False
if self.view.paginator is None:
return False
if self.method.lower() != 'get':
return False
return is_list_view(self.path, self.method, self.view)
def get_pagination_parameters(self):
"""Return the parameters added to the view by its paginator.
:rtype: list[openapi.Parameter]
"""
if not self.should_page():
return []
return self.probe_inspectors(self.paginator_inspectors, 'get_paginator_parameters', self.view.paginator) or []
def serializer_to_schema(self, serializer):
"""Convert a serializer to an OpenAPI :class:`.Schema`.
:param serializers.BaseSerializer serializer: the ``Serializer`` instance
:returns: the converted :class:`.Schema`, or ``None`` in case of an unknown serializer
:rtype: openapi.Schema,openapi.SchemaRef,None
"""
return self.probe_inspectors(
self.field_inspectors, 'get_schema', serializer, {'field_inspectors': self.field_inspectors}
)
def serializer_to_parameters(self, serializer, in_):
"""Convert a serializer to a possibly empty list of :class:`.Parameter`\ s.
:param serializers.BaseSerializer serializer: the ``Serializer`` instance
:param str in_: the location of the parameters, one of the `openapi.IN_*` constants
:rtype: list[openapi.Parameter]
"""
return self.probe_inspectors(
self.field_inspectors, 'get_request_parameters', serializer, {'field_inspectors': self.field_inspectors},
in_=in_
) or []
def get_paginated_response(self, response_schema):
"""Add appropriate paging fields to a response :class:`.Schema`.
:param openapi.Schema response_schema: the response schema that must be paged.
:returns: the paginated response class:`.Schema`, or ``None`` in case of an unknown pagination scheme
:rtype: openapi.Schema
"""
return self.probe_inspectors(self.paginator_inspectors, 'get_paginated_response',
self.view.paginator, response_schema=response_schema)
+455
View File
@@ -0,0 +1,455 @@
import operator
from collections import OrderedDict
from django.core import validators
from django.db import models
from rest_framework import serializers
from rest_framework.settings import api_settings as rest_framework_settings
from .base import NotHandled, SerializerInspector, FieldInspector
from .. import openapi
from ..errors import SwaggerGenerationError
from ..utils import filter_none
class InlineSerializerInspector(SerializerInspector):
"""Provides serializer conversions using :meth:`.FieldInspector.field_to_swagger_object`."""
#: whether to output :class:`.Schema` definitions inline or into the ``definitions`` section
use_definitions = False
def get_schema(self, serializer):
return self.probe_field_inspectors(serializer, openapi.Schema, self.use_definitions)
def get_request_parameters(self, serializer, in_):
fields = getattr(serializer, 'fields', {})
return [
self.probe_field_inspectors(
value, openapi.Parameter, self.use_definitions,
name=self.get_parameter_name(key), in_=in_
)
for key, value
in fields.items()
]
def get_property_name(self, field_name):
return field_name
def get_parameter_name(self, field_name):
return field_name
def field_to_swagger_object(self, field, swagger_object_type, use_references, **kwargs):
SwaggerType, ChildSwaggerType = self._get_partial_types(field, swagger_object_type, use_references, **kwargs)
if isinstance(field, (serializers.ListSerializer, serializers.ListField)):
child_schema = self.probe_field_inspectors(field.child, ChildSwaggerType, use_references)
return SwaggerType(
type=openapi.TYPE_ARRAY,
items=child_schema,
)
elif isinstance(field, serializers.Serializer):
if swagger_object_type != openapi.Schema:
raise SwaggerGenerationError("cannot instantiate nested serializer as " + swagger_object_type.__name__)
serializer = field
serializer_meta = getattr(serializer, 'Meta', None)
if hasattr(serializer_meta, 'ref_name'):
ref_name = serializer_meta.ref_name
else:
ref_name = type(serializer).__name__
if ref_name.endswith('Serializer'):
ref_name = ref_name[:-len('Serializer')]
def make_schema_definition():
properties = OrderedDict()
required = []
for key, value in serializer.fields.items():
key = self.get_property_name(key)
properties[key] = self.probe_field_inspectors(value, ChildSwaggerType, use_references)
if value.required:
required.append(key)
return SwaggerType(
type=openapi.TYPE_OBJECT,
properties=properties,
required=required or None,
)
if not ref_name or not use_references:
return make_schema_definition()
definitions = self.components.with_scope(openapi.SCHEMA_DEFINITIONS)
definitions.setdefault(ref_name, make_schema_definition)
return openapi.SchemaRef(definitions, ref_name)
return NotHandled
class ReferencingSerializerInspector(InlineSerializerInspector):
use_definitions = True
def get_queryset_field(queryset, field_name):
"""Try to get information about a model and model field from a queryset.
:param queryset: the queryset
:param field_name: target field name
:returns: the model and target field from the queryset as a 2-tuple; both elements can be ``None``
:rtype: tuple
"""
model = getattr(queryset, 'model', None)
model_field = get_model_field(model, field_name)
return model, model_field
def get_model_field(model, field_name):
"""Try to get the given field from a django db model.
:param model: the model
:param field_name: target field name
:return: model field or ``None``
"""
try:
if field_name == 'pk':
return model._meta.pk
else:
return model._meta.get_field(field_name)
except Exception: # pragma: no cover
return None
def get_parent_serializer(field):
"""Get the nearest parent ``Serializer`` instance for the given field.
:return: ``Serializer`` or ``None``
"""
while field is not None:
if isinstance(field, serializers.Serializer):
return field
field = field.parent
return None # pragma: no cover
def get_related_model(model, source):
"""Try to find the other side of a model relationship given the name of a related field.
:param model: one side of the relationship
:param str source: related field name
:return: related model or ``None``
"""
try:
return getattr(model, source).rel.related_model
except Exception: # pragma: no cover
return None
class RelatedFieldInspector(FieldInspector):
"""Provides conversions for ``RelatedField``\ s."""
def field_to_swagger_object(self, field, swagger_object_type, use_references, **kwargs):
SwaggerType, ChildSwaggerType = self._get_partial_types(field, swagger_object_type, use_references, **kwargs)
if isinstance(field, serializers.ManyRelatedField):
child_schema = self.probe_field_inspectors(field.child_relation, ChildSwaggerType, use_references)
return SwaggerType(
type=openapi.TYPE_ARRAY,
items=child_schema,
unique_items=True,
)
if not isinstance(field, serializers.RelatedField):
return NotHandled
field_queryset = getattr(field, 'queryset', None)
if isinstance(field, (serializers.PrimaryKeyRelatedField, serializers.SlugRelatedField)):
if getattr(field, 'pk_field', ''):
# a PrimaryKeyRelatedField can have a `pk_field` attribute which is a
# serializer field that will convert the PK value
result = self.probe_field_inspectors(field.pk_field, swagger_object_type, use_references, **kwargs)
# take the type, format, etc from `pk_field`, and the field-level information
# like title, description, default from the PrimaryKeyRelatedField
return SwaggerType(existing_object=result)
target_field = getattr(field, 'slug_field', 'pk')
if field_queryset is not None:
# if the RelatedField has a queryset, try to get the related model field from there
model, model_field = get_queryset_field(field_queryset, target_field)
else:
# if the RelatedField has no queryset (e.g. read only), try to find the target model
# from the view queryset or ModelSerializer model, if present
view_queryset = getattr(self.view, 'queryset', None)
serializer_meta = getattr(get_parent_serializer(field), 'Meta', None)
this_model = getattr(view_queryset, 'model', None) or getattr(serializer_meta, 'model', None)
source = getattr(field, 'source', '') or field.field_name
model = get_related_model(this_model, source)
model_field = get_model_field(model, target_field)
attrs = get_basic_type_info(model_field) or {'type': openapi.TYPE_STRING}
return SwaggerType(**attrs)
elif isinstance(field, serializers.HyperlinkedRelatedField):
return SwaggerType(type=openapi.TYPE_STRING, format=openapi.FORMAT_URI)
return SwaggerType(type=openapi.TYPE_STRING)
def find_regex(regex_field):
"""Given a ``Field``, look for a ``RegexValidator`` and try to extract its pattern and return it as a string.
:param serializers.Field regex_field: the field instance
:return: the extracted pattern, or ``None``
:rtype: str
"""
regex_validator = None
for validator in regex_field.validators:
if isinstance(validator, validators.RegexValidator):
if regex_validator is not None:
# bail if multiple validators are found - no obvious way to choose
return None # pragma: no cover
regex_validator = validator
# regex_validator.regex should be a compiled re object...
return getattr(getattr(regex_validator, 'regex', None), 'pattern', None)
numeric_fields = (serializers.IntegerField, serializers.FloatField, serializers.DecimalField)
limit_validators = [
# minimum and maximum apply to numbers
(validators.MinValueValidator, numeric_fields, 'minimum', operator.__gt__),
(validators.MaxValueValidator, numeric_fields, 'maximum', operator.__lt__),
# minLength and maxLength apply to strings
(validators.MinLengthValidator, serializers.CharField, 'min_length', operator.__gt__),
(validators.MaxLengthValidator, serializers.CharField, 'max_length', operator.__lt__),
# minItems and maxItems apply to lists
(validators.MinLengthValidator, serializers.ListField, 'min_items', operator.__gt__),
(validators.MaxLengthValidator, serializers.ListField, 'max_items', operator.__lt__),
]
def find_limits(field):
"""Given a ``Field``, look for min/max value/length validators and return appropriate limit validation attributes.
:param serializers.Field field: the field instance
:return: the extracted limits
:rtype: OrderedDict
"""
limits = {}
applicable_limits = [
(validator, attr, improves)
for validator, field_class, attr, improves in limit_validators
if isinstance(field, field_class)
]
for validator in field.validators:
if not hasattr(validator, 'limit_value'):
continue
for validator_class, attr, improves in applicable_limits:
if isinstance(validator, validator_class):
if attr not in limits or improves(validator.limit_value, limits[attr]):
limits[attr] = validator.limit_value
return OrderedDict(sorted(limits.items()))
model_field_to_basic_type = [
(models.AutoField, (openapi.TYPE_INTEGER, None)),
(models.BinaryField, (openapi.TYPE_STRING, openapi.FORMAT_BINARY)),
(models.BooleanField, (openapi.TYPE_BOOLEAN, None)),
(models.NullBooleanField, (openapi.TYPE_BOOLEAN, None)),
(models.DateTimeField, (openapi.TYPE_STRING, openapi.FORMAT_DATETIME)),
(models.DateField, (openapi.TYPE_STRING, openapi.FORMAT_DATE)),
(models.DecimalField, (openapi.TYPE_NUMBER, None)),
(models.DurationField, (openapi.TYPE_INTEGER, None)),
(models.FloatField, (openapi.TYPE_NUMBER, None)),
(models.IntegerField, (openapi.TYPE_INTEGER, None)),
(models.IPAddressField, (openapi.TYPE_STRING, openapi.FORMAT_IPV4)),
(models.GenericIPAddressField, (openapi.TYPE_STRING, openapi.FORMAT_IPV6)),
(models.SlugField, (openapi.TYPE_STRING, openapi.FORMAT_SLUG)),
(models.TextField, (openapi.TYPE_STRING, None)),
(models.TimeField, (openapi.TYPE_STRING, None)),
(models.UUIDField, (openapi.TYPE_STRING, openapi.FORMAT_UUID)),
(models.CharField, (openapi.TYPE_STRING, None)),
]
ip_format = {'ipv4': openapi.FORMAT_IPV4, 'ipv6': openapi.FORMAT_IPV6}
serializer_field_to_basic_type = [
(serializers.EmailField, (openapi.TYPE_STRING, openapi.FORMAT_EMAIL)),
(serializers.SlugField, (openapi.TYPE_STRING, openapi.FORMAT_SLUG)),
(serializers.URLField, (openapi.TYPE_STRING, openapi.FORMAT_URI)),
(serializers.IPAddressField, (openapi.TYPE_STRING, lambda field: ip_format.get(field.protocol, None))),
(serializers.UUIDField, (openapi.TYPE_STRING, openapi.FORMAT_UUID)),
(serializers.RegexField, (openapi.TYPE_STRING, None)),
(serializers.CharField, (openapi.TYPE_STRING, None)),
((serializers.BooleanField, serializers.NullBooleanField), (openapi.TYPE_BOOLEAN, None)),
(serializers.IntegerField, (openapi.TYPE_INTEGER, None)),
((serializers.FloatField, serializers.DecimalField), (openapi.TYPE_NUMBER, None)),
(serializers.DurationField, (openapi.TYPE_NUMBER, None)), # ?
(serializers.DateField, (openapi.TYPE_STRING, openapi.FORMAT_DATE)),
(serializers.DateTimeField, (openapi.TYPE_STRING, openapi.FORMAT_DATETIME)),
(serializers.ModelField, (openapi.TYPE_STRING, None)),
]
basic_type_info = serializer_field_to_basic_type + model_field_to_basic_type
def get_basic_type_info(field):
"""Given a serializer or model ``Field``, return its basic type information - ``type``, ``format``, ``pattern``,
and any applicable min/max limit values.
:param field: the field instance
:return: the extracted attributes as a dictionary, or ``None`` if the field type is not known
:rtype: OrderedDict
"""
if field is None:
return None
for field_class, type_format in basic_type_info:
if isinstance(field, field_class):
swagger_type, format = type_format
if callable(format):
format = format(field)
break
else: # pragma: no cover
return None
pattern = find_regex(field) if format in (None, openapi.FORMAT_SLUG) else None
limits = find_limits(field)
result = OrderedDict([
('type', swagger_type),
('format', format),
('pattern', pattern)
])
result.update(limits)
result = filter_none(result)
return result
class SimpleFieldInspector(FieldInspector):
"""Provides conversions for fields which can be described using just ``type``, ``format``, ``pattern``
and min/max validators.
"""
def field_to_swagger_object(self, field, swagger_object_type, use_references, **kwargs):
type_info = get_basic_type_info(field)
if type_info is None:
return NotHandled
SwaggerType, ChildSwaggerType = self._get_partial_types(field, swagger_object_type, use_references, **kwargs)
return SwaggerType(**type_info)
class ChoiceFieldInspector(FieldInspector):
"""Provides conversions for ``ChoiceField`` and ``MultipleChoiceField``."""
def field_to_swagger_object(self, field, swagger_object_type, use_references, **kwargs):
SwaggerType, ChildSwaggerType = self._get_partial_types(field, swagger_object_type, use_references, **kwargs)
if isinstance(field, serializers.MultipleChoiceField):
return SwaggerType(
type=openapi.TYPE_ARRAY,
items=ChildSwaggerType(
type=openapi.TYPE_STRING,
enum=list(field.choices.keys())
)
)
elif isinstance(field, serializers.ChoiceField):
return SwaggerType(type=openapi.TYPE_STRING, enum=list(field.choices.keys()))
return NotHandled
class FileFieldInspector(FieldInspector):
"""Provides conversions for ``FileField``\ s."""
def field_to_swagger_object(self, field, swagger_object_type, use_references, **kwargs):
SwaggerType, ChildSwaggerType = self._get_partial_types(field, swagger_object_type, use_references, **kwargs)
if isinstance(field, serializers.FileField):
# swagger 2.0 does not support specifics about file fields, so ImageFile gets no special treatment
# OpenAPI 3.0 does support it, so a future implementation could handle this better
err = SwaggerGenerationError("FileField is supported only in a formData Parameter or response Schema")
if swagger_object_type == openapi.Schema:
# FileField.to_representation returns URL or file name
result = SwaggerType(type=openapi.TYPE_STRING, read_only=True)
if getattr(field, 'use_url', rest_framework_settings.UPLOADED_FILES_USE_URL):
result.format = openapi.FORMAT_URI
return result
elif swagger_object_type == openapi.Parameter:
param = SwaggerType(type=openapi.TYPE_FILE)
if param['in'] != openapi.IN_FORM:
raise err # pragma: no cover
return param
else:
raise err # pragma: no cover
return NotHandled
class DictFieldInspector(FieldInspector):
"""Provides conversion for ``DictField``."""
def field_to_swagger_object(self, field, swagger_object_type, use_references, **kwargs):
SwaggerType, ChildSwaggerType = self._get_partial_types(field, swagger_object_type, use_references, **kwargs)
if isinstance(field, serializers.DictField) and swagger_object_type == openapi.Schema:
child_schema = self.probe_field_inspectors(field.child, ChildSwaggerType, use_references)
return SwaggerType(
type=openapi.TYPE_OBJECT,
additional_properties=child_schema
)
return NotHandled
class StringDefaultFieldInspector(FieldInspector):
"""For otherwise unhandled fields, return them as plain :data:`.TYPE_STRING` objects."""
def field_to_swagger_object(self, field, swagger_object_type, use_references, **kwargs): # pragma: no cover
# TODO unhandled fields: TimeField HiddenField JSONField
SwaggerType, ChildSwaggerType = self._get_partial_types(field, swagger_object_type, use_references, **kwargs)
return SwaggerType(type=openapi.TYPE_STRING)
try:
from djangorestframework_camel_case.parser import CamelCaseJSONParser
from djangorestframework_camel_case.render import CamelCaseJSONRenderer
from djangorestframework_camel_case.render import camelize
except ImportError: # pragma: no cover
class CamelCaseJSONFilter(FieldInspector):
pass
else:
def camelize_string(s):
"""Hack to force ``djangorestframework_camel_case`` to camelize a plain string."""
return next(iter(camelize({s: ''})))
def camelize_schema(schema_or_ref, components):
"""Recursively camelize property names for the given schema using ``djangorestframework_camel_case``."""
schema = openapi.resolve_ref(schema_or_ref, components)
if getattr(schema, 'properties', {}):
schema.properties = OrderedDict(
(camelize_string(key), camelize_schema(val, components))
for key, val in schema.properties.items()
)
if getattr(schema, 'required', []):
schema.required = [camelize_string(p) for p in schema.required]
return schema_or_ref
class CamelCaseJSONFilter(FieldInspector):
def is_camel_case(self):
return any(issubclass(parser, CamelCaseJSONParser) for parser in self.view.parser_classes) \
or any(issubclass(renderer, CamelCaseJSONRenderer) for renderer in self.view.renderer_classes)
def process_result(self, result, method_name, obj, **kwargs):
if isinstance(result, openapi.Schema.OR_REF) and self.is_camel_case():
return camelize_schema(result, self.components)
return result
+76
View File
@@ -0,0 +1,76 @@
from collections import OrderedDict
import coreschema
from rest_framework.pagination import CursorPagination, PageNumberPagination, LimitOffsetPagination
from .base import PaginatorInspector, FilterInspector
from .. import openapi
class CoreAPICompatInspector(PaginatorInspector, FilterInspector):
"""Converts ``coreapi.Field``\ s to :class:`.openapi.Parameter`\ s for filters and paginators that implement a
``get_schema_fields`` method.
"""
def get_paginator_parameters(self, paginator):
fields = []
if hasattr(paginator, 'get_schema_fields'):
fields = paginator.get_schema_fields(self.view)
return [self.coreapi_field_to_parameter(field) for field in fields]
def get_filter_parameters(self, filter_backend):
fields = []
if hasattr(filter_backend, 'get_schema_fields'):
fields = filter_backend.get_schema_fields(self.view)
return [self.coreapi_field_to_parameter(field) for field in fields]
def coreapi_field_to_parameter(self, field):
"""Convert an instance of `coreapi.Field` to a swagger :class:`.Parameter` object.
:param coreapi.Field field:
:rtype: openapi.Parameter
"""
location_to_in = {
'query': openapi.IN_QUERY,
'path': openapi.IN_PATH,
'form': openapi.IN_FORM,
'body': openapi.IN_FORM,
}
coreapi_types = {
coreschema.Integer: openapi.TYPE_INTEGER,
coreschema.Number: openapi.TYPE_NUMBER,
coreschema.String: openapi.TYPE_STRING,
coreschema.Boolean: openapi.TYPE_BOOLEAN,
}
return openapi.Parameter(
name=field.name,
in_=location_to_in[field.location],
type=coreapi_types.get(type(field.schema), openapi.TYPE_STRING),
required=field.required,
description=field.schema.description,
)
class DjangoRestResponsePagination(PaginatorInspector):
"""Provides response schema pagination warpping for django-rest-framework's LimitOffsetPagination,
PageNumberPagination and CursorPagination
"""
def get_paginated_response(self, paginator, response_schema):
assert response_schema.type == openapi.TYPE_ARRAY, "array return expected for paged response"
paged_schema = None
if isinstance(paginator, (LimitOffsetPagination, PageNumberPagination, CursorPagination)):
has_count = not isinstance(paginator, CursorPagination)
paged_schema = openapi.Schema(
type=openapi.TYPE_OBJECT,
properties=OrderedDict((
('count', openapi.Schema(type=openapi.TYPE_INTEGER) if has_count else None),
('next', openapi.Schema(type=openapi.TYPE_STRING, format=openapi.FORMAT_URI)),
('previous', openapi.Schema(type=openapi.TYPE_STRING, format=openapi.FORMAT_URI)),
('results', response_schema),
)),
required=['count', 'results']
)
return paged_schema
@@ -1,63 +1,22 @@
import inspect
from collections import OrderedDict
import coreschema
from rest_framework import serializers, status
from rest_framework.request import is_form_media_type
from rest_framework.schemas import AutoSchema
from rest_framework.status import is_success
from rest_framework.viewsets import GenericViewSet
from . import openapi
from .errors import SwaggerGenerationError
from .utils import serializer_field_to_swagger, no_body, is_list_view, param_list_to_odict
from .base import ViewInspector
from .. import openapi
from ..errors import SwaggerGenerationError
from ..utils import force_serializer_instance, no_body, is_list_view, param_list_to_odict, guess_response_status
def force_serializer_instance(serializer):
"""Force `serializer` into a ``Serializer`` instance. If it is not a ``Serializer`` class or instance, raises
an assertion error.
:param serializer: serializer class or instance
:return: serializer instance
"""
if inspect.isclass(serializer):
assert issubclass(serializer, serializers.BaseSerializer), "Serializer required, not %s" % serializer.__name__
return serializer()
assert isinstance(serializer, serializers.BaseSerializer), \
"Serializer class or instance required, not %s" % type(serializer).__name__
return serializer
class SwaggerAutoSchema(object):
body_methods = ('PUT', 'PATCH', 'POST') #: methods allowed to have a request body
def __init__(self, view, path, method, overrides, components):
"""Inspector class responsible for providing :class:`.Operation` definitions given a
:param view: the view associated with this endpoint
:param str path: the path component of the operation URL
:param str method: the http method of the operation
:param dict overrides: manual overrides as passed to :func:`@swagger_auto_schema <.swagger_auto_schema>`
:param openapi.ReferenceResolver components: referenceable components
"""
super(SwaggerAutoSchema, self).__init__()
class SwaggerAutoSchema(ViewInspector):
def __init__(self, view, path, method, components, request, overrides):
super(SwaggerAutoSchema, self).__init__(view, path, method, components, request, overrides)
self._sch = AutoSchema()
self.view = view
self.path = path
self.method = method
self.overrides = overrides
self.components = components
self._sch.view = view
def get_operation(self, operation_keys):
"""Get an :class:`.Operation` for the given API endpoint (path, method).
This includes query, body parameters and response schemas.
:param tuple[str] operation_keys: an array of keys describing the hierarchical layout of this view in the API;
e.g. ``('snippets', 'list')``, ``('snippets', 'retrieve')``, etc.
:rtype: openapi.Operation
"""
consumes = self.get_consumes()
body = self.get_request_body_parameters(consumes)
@@ -66,17 +25,19 @@ class SwaggerAutoSchema(object):
parameters = [param for param in parameters if param is not None]
parameters = self.add_manual_parameters(parameters)
operation_id = self.get_operation_id(operation_keys)
description = self.get_description()
tags = self.get_tags(operation_keys)
responses = self.get_responses()
return openapi.Operation(
operation_id='_'.join(operation_keys),
operation_id=operation_id,
description=description,
responses=responses,
parameters=parameters,
consumes=consumes,
tags=[operation_keys[0]],
tags=tags,
)
def get_request_body_parameters(self, consumes):
@@ -105,7 +66,7 @@ class SwaggerAutoSchema(object):
else:
if schema is None:
schema = self.get_request_body_schema(serializer)
return [self.make_body_parameter(schema)]
return [self.make_body_parameter(schema)] if schema is not None else []
def get_view_serializer(self):
"""Return the serializer as defined by the view's ``get_serializer()`` method.
@@ -192,26 +153,6 @@ class SwaggerAutoSchema(object):
responses=self.get_response_schemas(response_serializers)
)
def get_paged_response_schema(self, response_schema):
"""Add appropriate paging fields to a response :class:`.Schema`.
:param openapi.Schema response_schema: the response schema that must be paged.
:rtype: openapi.Schema
"""
assert response_schema.type == openapi.TYPE_ARRAY, "array return expected for paged response"
paged_schema = openapi.Schema(
type=openapi.TYPE_OBJECT,
properties={
'count': openapi.Schema(type=openapi.TYPE_INTEGER),
'next': openapi.Schema(type=openapi.TYPE_STRING, format=openapi.FORMAT_URI),
'previous': openapi.Schema(type=openapi.TYPE_STRING, format=openapi.FORMAT_URI),
'results': response_schema,
},
required=['count', 'results']
)
return paged_schema
def get_default_responses(self):
"""Get the default responses determined for this view from the request serializer and request method.
@@ -219,28 +160,26 @@ class SwaggerAutoSchema(object):
"""
method = self.method.lower()
default_status = status.HTTP_200_OK
default_status = guess_response_status(method)
default_schema = ''
if method == 'post':
default_status = status.HTTP_201_CREATED
default_schema = self.get_request_serializer() or self.get_view_serializer()
elif method == 'delete':
default_status = status.HTTP_204_NO_CONTENT
elif method in ('get', 'put', 'patch'):
default_schema = self.get_request_serializer() or self.get_view_serializer()
default_schema = default_schema or ''
if any(is_form_media_type(encoding) for encoding in self.get_consumes()):
default_schema = ''
if default_schema and not isinstance(default_schema, openapi.Schema):
default_schema = self.serializer_to_schema(default_schema) or ''
if default_schema:
if not isinstance(default_schema, openapi.Schema):
default_schema = self.serializer_to_schema(default_schema)
if is_list_view(self.path, self.method, self.view) and self.method.lower() == 'get':
default_schema = openapi.Schema(type=openapi.TYPE_ARRAY, items=default_schema)
if self.should_page():
default_schema = self.get_paged_response_schema(default_schema)
default_schema = self.get_paginated_response(default_schema) or default_schema
return {str(default_status): default_schema}
return OrderedDict({str(default_status): default_schema})
def get_response_serializers(self):
"""Return the response codes that this view is expected to return, and the serializer for each response body.
@@ -254,7 +193,7 @@ class SwaggerAutoSchema(object):
manual_responses = self.overrides.get('responses', None) or {}
manual_responses = OrderedDict((str(sc), resp) for sc, resp in manual_responses.items())
responses = {}
responses = OrderedDict()
if not any(is_success(int(sc)) for sc in manual_responses if sc != 'default'):
responses = self.get_default_responses()
@@ -268,7 +207,7 @@ class SwaggerAutoSchema(object):
:return: a dictionary of status code to :class:`.Response` object
:rtype: dict[str, openapi.Response]
"""
responses = {}
responses = OrderedDict()
for sc, serializer in response_serializers.items():
if isinstance(serializer, str):
response = openapi.Response(
@@ -325,84 +264,18 @@ class SwaggerAutoSchema(object):
return natural_parameters + serializer_parameters
def should_filter(self):
"""Determine whether filter backend parameters should be included for this request.
def get_operation_id(self, operation_keys):
"""Return an unique ID for this operation. The ID must be unique across
all :class:`.Operation` objects in the API.
:rtype: bool
:param tuple[str] operation_keys: an array of keys derived from the pathdescribing the hierarchical layout
of this view in the API; e.g. ``('snippets', 'list')``, ``('snippets', 'retrieve')``, etc.
:rtype: str
"""
if not getattr(self.view, 'filter_backends', None):
return False
if self.method.lower() not in ["get", "delete"]:
return False
if not isinstance(self.view, GenericViewSet):
return True
return is_list_view(self.path, self.method, self.view)
def get_filter_backend_parameters(self, filter_backend):
"""Get the filter parameters for a single filter backend **instance**.
:param BaseFilterBackend filter_backend: the filter backend
:rtype: list[openapi.Parameter]
"""
fields = []
if hasattr(filter_backend, 'get_schema_fields'):
fields = filter_backend.get_schema_fields(self.view)
return [self.coreapi_field_to_parameter(field) for field in fields]
def get_filter_parameters(self):
"""Return the parameters added to the view by its filter backends.
:rtype: list[openapi.Parameter]
"""
if not self.should_filter():
return []
fields = []
for filter_backend in self.view.filter_backends:
fields += self.get_filter_backend_parameters(filter_backend())
return fields
def should_page(self):
"""Determine whether paging parameters and structure should be added to this operation's request and response.
:rtype: bool
"""
if not hasattr(self.view, 'paginator'):
return False
if self.view.paginator is None:
return False
if self.method.lower() != 'get':
return False
return is_list_view(self.path, self.method, self.view)
def get_paginator_parameters(self, paginator):
"""Get the pagination parameters for a single paginator **instance**.
:param BasePagination paginator: the paginator
:rtype: list[openapi.Parameter]
"""
fields = []
if hasattr(paginator, 'get_schema_fields'):
fields = paginator.get_schema_fields(self.view)
return [self.coreapi_field_to_parameter(field) for field in fields]
def get_pagination_parameters(self):
"""Return the parameters added to the view by its paginator.
:rtype: list[openapi.Parameter]
"""
if not self.should_page():
return []
return self.get_paginator_parameters(self.view.paginator)
operation_id = self.overrides.get('operation_id', '')
if not operation_id:
operation_id = '_'.join(operation_keys)
return operation_id
def get_description(self):
"""Return an operation description determined as appropriate from the view's method and class docstrings.
@@ -415,6 +288,16 @@ class SwaggerAutoSchema(object):
description = self._sch.get_description(self.path, self.method)
return description
def get_tags(self, operation_keys):
"""Get a list of tags for this operation. Tags determine how operations relate with each other, and in the UI
each tag will show as a group containing the operations that use it.
:param tuple[str] operation_keys: an array of keys derived from the pathdescribing the hierarchical layout
of this view in the API; e.g. ``('snippets', 'list')``, ``('snippets', 'retrieve')``, etc.
:rtype: list[str]
"""
return [operation_keys[0]]
def get_consumes(self):
"""Return the MIME types this endpoint can consume.
@@ -424,62 +307,3 @@ class SwaggerAutoSchema(object):
if all(is_form_media_type(encoding) for encoding in media_types):
return media_types
return media_types[:1]
def serializer_to_schema(self, serializer):
"""Convert a DRF Serializer instance to an :class:`.openapi.Schema`.
:param serializers.BaseSerializer serializer: the ``Serializer`` instance
:rtype: openapi.Schema
"""
definitions = self.components.with_scope(openapi.SCHEMA_DEFINITIONS)
return serializer_field_to_swagger(serializer, openapi.Schema, definitions)
def serializer_to_parameters(self, serializer, in_):
"""Convert a DRF serializer into a list of :class:`.Parameter`\ s using :meth:`.field_to_parameter`
:param serializers.BaseSerializer serializer: the ``Serializer`` instance
:param str in_: the location of the parameters, one of the `openapi.IN_*` constants
:rtype: list[openapi.Parameter]
"""
fields = getattr(serializer, 'fields', {})
return [
self.field_to_parameter(value, key, in_)
for key, value
in fields.items()
]
def field_to_parameter(self, field, name, in_):
"""Convert a DRF serializer Field to a swagger :class:`.Parameter` object.
:param coreapi.Field field:
:param str name: the name of the parameter
:param str in_: the location of the parameter, one of the `openapi.IN_*` constants
:rtype: openapi.Parameter
"""
return serializer_field_to_swagger(field, openapi.Parameter, name=name, in_=in_)
def coreapi_field_to_parameter(self, field):
"""Convert an instance of `coreapi.Field` to a swagger :class:`.Parameter` object.
:param coreapi.Field field:
:rtype: openapi.Parameter
"""
location_to_in = {
'query': openapi.IN_QUERY,
'path': openapi.IN_PATH,
'form': openapi.IN_FORM,
'body': openapi.IN_FORM,
}
coreapi_types = {
coreschema.Integer: openapi.TYPE_INTEGER,
coreschema.Number: openapi.TYPE_NUMBER,
coreschema.String: openapi.TYPE_STRING,
coreschema.Boolean: openapi.TYPE_BOOLEAN,
}
return openapi.Parameter(
name=field.name,
in_=location_to_in[field.location],
type=coreapi_types.get(type(field.schema), openapi.TYPE_STRING),
required=field.required,
description=field.schema.description,
)
+57 -22
View File
@@ -1,9 +1,11 @@
import re
from collections import OrderedDict
from coreapi.compat import urlparse
from future.utils import raise_from
from inflection import camelize
from .utils import filter_none
TYPE_OBJECT = "object" #:
TYPE_STRING = "string" #:
TYPE_NUMBER = "number" #:
@@ -94,8 +96,9 @@ class SwaggerDict(OrderedDict):
raise AttributeError
try:
return self[make_swagger_name(item)]
except KeyError as e:
raise_from(AttributeError("object of class " + type(self).__name__ + " has no attribute " + item), e)
except KeyError:
# raise_from is EXTREMELY slow, replaced with plain raise
raise AttributeError("object of class " + type(self).__name__ + " has no attribute " + item)
def __delattr__(self, item):
if item.startswith('_'):
@@ -230,7 +233,7 @@ class Swagger(SwaggerDict):
self.base_path = '/'
self.paths = paths
self.definitions = definitions
self.definitions = filter_none(definitions)
self._insert_extras__()
@@ -270,13 +273,13 @@ class PathItem(SwaggerDict):
self.patch = patch
self.delete = delete
self.options = options
self.parameters = parameters
self.parameters = filter_none(parameters)
self._insert_extras__()
class Operation(SwaggerDict):
def __init__(self, operation_id, responses, parameters=None, consumes=None,
produces=None, description=None, tags=None, **extra):
produces=None, summary=None, description=None, tags=None, **extra):
"""Information about an API operation (path + http method combination)
:param str operation_id: operation ID, should be unique across all operations
@@ -284,17 +287,19 @@ class Operation(SwaggerDict):
:param list[.Parameter] parameters: parameters accepted
:param list[str] consumes: content types accepted
:param list[str] produces: content types produced
:param str description: operation description
:param str summary: operation summary; should be < 120 characters
:param str description: operation description; can be of any length and supports markdown
:param list[str] tags: operation tags
"""
super(Operation, self).__init__(**extra)
self.operation_id = operation_id
self.summary = summary
self.description = description
self.parameters = [param for param in parameters if param is not None]
self.parameters = filter_none(parameters)
self.responses = responses
self.consumes = consumes
self.produces = produces
self.tags = tags
self.consumes = filter_none(consumes)
self.produces = filter_none(produces)
self.tags = filter_none(tags)
self._insert_extras__()
@@ -352,21 +357,26 @@ class Parameter(SwaggerDict):
class Schema(SwaggerDict):
OR_REF = ()
OR_REF = () #: useful for type-checking, e.g ``isinstance(obj, openapi.Schema.OR_REF)``
def __init__(self, description=None, required=None, type=None, properties=None, additional_properties=None,
format=None, enum=None, pattern=None, items=None, **extra):
def __init__(self, title=None, description=None, type=None, format=None, enum=None, pattern=None, properties=None,
additional_properties=None, required=None, items=None, default=None, read_only=None, **extra):
"""Describes a complex object accepted as parameter or returned as a response.
:param description: schema description
:param list[str] required: list of requried property names
:param str title: schema title
:param str description: schema description
:param str type: value type; required
:param list[.Schema,.SchemaRef] properties: object properties; required if `type` is ``object``
:param bool,.Schema,.SchemaRef additional_properties: allow wildcard properties not listed in `properties`
:param str format: value format, see OpenAPI spec
:param list enum: restrict possible values
:param str pattern: pattern if type is ``string``
:param .Schema,.SchemaRef items: only valid if `type` is ``array``
:param list[.Schema,.SchemaRef] properties: object properties; required if `type` is ``object``
:param bool,.Schema,.SchemaRef additional_properties: allow wildcard properties not listed in `properties`
:param list[str] required: list of requried property names
:param .Schema,.SchemaRef items: type of array items, only valid if `type` is ``array``
:param default: only valid when insider another ``Schema``\ 's ``properties``;
the default value of this property if it is not provided, must conform to the type of this Schema
:param read_only: only valid when insider another ``Schema``\ 's ``properties``;
declares the property as read only - it must only be sent as part of responses, never in requests
"""
super(Schema, self).__init__(**extra)
if required is True or required is False:
@@ -374,19 +384,24 @@ class Schema(SwaggerDict):
raise AssertionError(
"the `requires` attribute of schema must be an array of required properties, not a boolean!")
assert type is not None, "type is required!"
self.title = title
self.description = description
self.required = required
self.required = filter_none(required)
self.type = type
self.properties = properties
self.properties = filter_none(properties)
self.additional_properties = additional_properties
self.format = format
self.enum = enum
self.pattern = pattern
self.items = items
self.read_only = read_only
self.default = default
self._insert_extras__()
class _Ref(SwaggerDict):
ref_name_re = re.compile(r"#/(?P<scope>.+)/(?P<name>[^/]+)$")
def __init__(self, resolver, name, scope, expected_type):
"""Base class for all reference types. A reference object has only one property, ``$ref``, which must be a JSON
reference to a valid object in the specification, e.g. ``#/definitions/Article`` to refer to an article model.
@@ -404,6 +419,15 @@ class _Ref(SwaggerDict):
.format(actual=type(obj).__name__, expected=expected_type.__name__)
self.ref = ref_name
def resolve(self, resolver):
"""Get the object targeted by this reference from the given component resolver.
:param .ReferenceResolver resolver: component resolver which must contain the referneced object
:returns: the target object
"""
ref_match = self.ref_name_re.match(self.ref)
return resolver.get(ref_match.group('name'), scope=ref_match.group('scope'))
def __setitem__(self, key, value, **kwargs):
if key == "$ref":
return super(_Ref, self).__setitem__(key, value, **kwargs)
@@ -427,6 +451,17 @@ class SchemaRef(_Ref):
Schema.OR_REF = (Schema, SchemaRef)
def resolve_ref(ref_or_obj, resolver):
"""Resolve `ref_or_obj` if it is a reference type. Return it unchaged if not.
:param SwaggerDict,_Ref ref_or_obj:
:param resolver: component resolver which must contain the referenced object
"""
if isinstance(ref_or_obj, _Ref):
return ref_or_obj.resolve(resolver)
return ref_or_obj
class Responses(SwaggerDict):
def __init__(self, responses, default=None, **extra):
"""Describes the expected responses of an :class:`.Operation`.
@@ -483,7 +518,7 @@ class ReferenceResolver(object):
self._objects[scope] = OrderedDict()
def with_scope(self, scope):
"""Return a new :class:`.ReferenceResolver` whose scope is defaulted and forced to `scope`.
"""Return a view into this :class:`.ReferenceResolver` whose scope is defaulted and forced to `scope`.
:param str scope: target scope, must be in this resolver's `scopes`
:return: the bound resolver
+2 -2
View File
@@ -14,7 +14,7 @@ class _SpecRenderer(BaseRenderer):
@classmethod
def with_validators(cls, validators):
assert all(vld in VALIDATORS for vld in validators), "allowed validators are" + ", ".join(VALIDATORS)
assert all(vld in VALIDATORS for vld in validators), "allowed validators are " + ", ".join(VALIDATORS)
return type(cls.__name__, (cls,), {'validators': validators})
def render(self, data, media_type=None, renderer_context=None):
@@ -45,7 +45,7 @@ class SwaggerYAMLRenderer(_SpecRenderer):
class _UIRenderer(BaseRenderer):
"""Base class for web UI renderers. Handles loading an passing settings to the appropriate template."""
"""Base class for web UI renderers. Handles loading and passing settings to the appropriate template."""
media_type = 'text/html'
charset = 'utf-8'
template = ''
@@ -166,9 +166,11 @@
layout: "StandaloneLayout",
filter: true,
requestInterceptor: function(request) {
console.log(request);
var headers = request.headers || {};
headers["X-CSRFToken"] = document.querySelector("[name=csrfmiddlewaretoken]").value;
var csrftoken = document.querySelector("[name=csrfmiddlewaretoken]");
if (csrftoken) {
headers["X-CSRFToken"] = csrftoken.value;
}
return request;
}
};
+177 -434
View File
@@ -1,17 +1,9 @@
import inspect
import logging
from collections import OrderedDict
from django.core.validators import RegexValidator
from django.db import models
from django.utils.encoding import force_text
from rest_framework import serializers
from rest_framework import status, serializers
from rest_framework.mixins import RetrieveModelMixin, DestroyModelMixin, UpdateModelMixin
from rest_framework.schemas.inspectors import get_pk_description
from rest_framework.settings import api_settings
from rest_framework.utils import json, encoders
from . import openapi
from .errors import SwaggerGenerationError
logger = logging.getLogger(__name__)
@@ -19,6 +11,141 @@ logger = logging.getLogger(__name__)
no_body = object()
def swagger_auto_schema(method=None, methods=None, auto_schema=None, request_body=None, query_serializer=None,
manual_parameters=None, operation_id=None, operation_description=None, responses=None,
field_inspectors=None, filter_inspectors=None, paginator_inspectors=None,
**extra_overrides):
"""Decorate a view method to customize the :class:`.Operation` object generated from it.
`method` and `methods` are mutually exclusive and must only be present when decorating a view method that accepts
more than one HTTP request method.
The `auto_schema` and `operation_description` arguments take precendence over view- or method-level values.
.. versionchanged:: 1.1
Added the ``extra_overrides`` and ``operatiod_id`` parameters.
.. versionchanged:: 1.1
Added the ``field_inspectors``, ``filter_inspectors`` and ``paginator_inspectors`` parameters.
:param str method: for multi-method views, the http method the options should apply to
:param list[str] methods: for multi-method views, the http methods the options should apply to
:param .inspectors.SwaggerAutoSchema auto_schema: custom class to use for generating the Operation object;
this overrides both the class-level ``swagger_schema`` attribute and the ``DEFAULT_AUTO_SCHEMA_CLASS``
setting
:param .Schema,.SchemaRef,.Serializer request_body: custom request body, or :data:`.no_body`. The value given here
will be used as the ``schema`` property of a :class:`.Parameter` with ``in: 'body'``.
A Schema or SchemaRef is not valid if this request consumes form-data, because ``form`` and ``body`` parameters
are mutually exclusive in an :class:`.Operation`. If you need to set custom ``form`` parameters, you can use
the `manual_parameters` argument.
If a ``Serializer`` class or instance is given, it will be automatically converted into a :class:`.Schema`
used as a ``body`` :class:`.Parameter`, or into a list of ``form`` :class:`.Parameter`\ s, as appropriate.
:param .Serializer query_serializer: if you use a ``Serializer`` to parse query parameters, you can pass it here
and have :class:`.Parameter` objects be generated automatically from it.
If any ``Field`` on the serializer cannot be represented as a ``query`` :class:`.Parameter`
(e.g. nested Serializers, file fields, ...), the schema generation will fail with an error.
Schema generation will also fail if the name of any Field on the `query_serializer` conflicts with parameters
generated by ``filter_backends`` or ``paginator``.
:param list[.Parameter] manual_parameters: a list of manual parameters to override the automatically generated ones
:class:`.Parameter`\ s are identified by their (``name``, ``in``) combination, and any parameters given
here will fully override automatically generated parameters if they collide.
It is an error to supply ``form`` parameters when the request does not consume form-data.
:param str operation_id: operation ID override; the operation ID must be unique accross the whole API
:param str operation_description: operation description override
:param dict[str,(.Schema,.SchemaRef,.Response,str,Serializer)] responses: a dict of documented manual responses
keyed on response status code. If no success (``2xx``) response is given, one will automatically be
generated from the request body and http method. If any ``2xx`` response is given the automatic response is
suppressed.
* if a plain string is given as value, a :class:`.Response` with no body and that string as its description
will be generated
* if a :class:`.Schema`, :class:`.SchemaRef` is given, a :class:`.Response` with the schema as its body and
an empty description will be generated
* a ``Serializer`` class or instance will be converted into a :class:`.Schema` and treated as above
* a :class:`.Response` object will be used as-is; however if its ``schema`` attribute is a ``Serializer``,
it will automatically be converted into a :class:`.Schema`
:param list[.FieldInspector] field_inspectors: extra serializer and field inspectors; these will be tried
before :attr:`.ViewInspector.field_inspectors` on the :class:`.inspectors.SwaggerAutoSchema` instance
:param list[.FilterInspector] filter_inspectors: extra filter inspectors; these will be tried before
:attr:`.ViewInspector.filter_inspectors` on the :class:`.inspectors.SwaggerAutoSchema` instance
:param list[.PaginatorInspector] paginator_inspectors: extra paginator inspectors; these will be tried before
:attr:`.ViewInspector.paginator_inspectors` on the :class:`.inspectors.SwaggerAutoSchema` instance
:param extra_overrides: extra values that will be saved into the ``overrides`` dict; these values will be available
in the handling :class:`.inspectors.SwaggerAutoSchema` instance via ``self.overrides``
"""
def decorator(view_method):
data = {
'auto_schema': auto_schema,
'request_body': request_body,
'query_serializer': query_serializer,
'manual_parameters': manual_parameters,
'operation_id': operation_id,
'operation_description': operation_description,
'responses': responses,
'filter_inspectors': list(filter_inspectors) if filter_inspectors else None,
'paginator_inspectors': list(paginator_inspectors) if paginator_inspectors else None,
'field_inspectors': list(field_inspectors) if field_inspectors else None,
}
data = {k: v for k, v in data.items() if v is not None}
data.update(extra_overrides)
# if the method is a detail_route or list_route, it will have a bind_to_methods attribute
bind_to_methods = getattr(view_method, 'bind_to_methods', [])
# if the method is actually a function based view (@api_view), it will have a 'cls' attribute
view_cls = getattr(view_method, 'cls', None)
http_method_names = getattr(view_cls, 'http_method_names', [])
if bind_to_methods or http_method_names:
# detail_route, list_route or api_view
assert bool(http_method_names) != bool(bind_to_methods), "this should never happen"
available_methods = http_method_names + bind_to_methods
existing_data = getattr(view_method, '_swagger_auto_schema', {})
if http_method_names:
_route = "api_view"
else:
_route = "detail_route" if view_method.detail else "list_route"
_methods = methods
if len(available_methods) > 1:
assert methods or method, \
"on multi-method %s, you must specify swagger_auto_schema on a per-method basis " \
"using one of the `method` or `methods` arguments" % _route
assert bool(methods) != bool(method), "specify either method or methods"
assert not isinstance(methods, str), "`methods` expects to receive a list of methods;" \
" use `method` for a single argument"
if method:
_methods = [method.lower()]
else:
_methods = [mth.lower() for mth in methods]
assert not any(mth in existing_data for mth in _methods), "method defined multiple times"
assert all(mth in available_methods for mth in _methods), "method not bound to %s" % _route
existing_data.update((mth.lower(), data) for mth in _methods)
else:
existing_data[available_methods[0]] = data
view_method._swagger_auto_schema = existing_data
else:
assert method is None and methods is None, \
"the methods argument should only be specified when decorating a detail_route or list_route; you " \
"should also ensure that you put the swagger_auto_schema decorator AFTER (above) the _route decorator"
view_method._swagger_auto_schema = data
return view_method
return decorator
def is_list_view(path, method, view):
"""Check if the given path/method appears to represent a list view (as opposed to a detail/instance view).
@@ -52,431 +179,13 @@ def is_list_view(path, method, view):
return True
def swagger_auto_schema(method=None, methods=None, auto_schema=None, request_body=None, query_serializer=None,
manual_parameters=None, operation_description=None, responses=None):
"""Decorate a view method to customize the :class:`.Operation` object generated from it.
`method` and `methods` are mutually exclusive and must only be present when decorating a view method that accepts
more than one HTTP request method.
The `auto_schema` and `operation_description` arguments take precendence over view- or method-level values.
:param str method: for multi-method views, the http method the options should apply to
:param list[str] methods: for multi-method views, the http methods the options should apply to
:param .SwaggerAutoSchema auto_schema: custom class to use for generating the Operation object
:param .Schema,.SchemaRef,.Serializer request_body: custom request body, or :data:`.no_body`. The value given here
will be used as the ``schema`` property of a :class:`.Parameter` with ``in: 'body'``.
A Schema or SchemaRef is not valid if this request consumes form-data, because ``form`` and ``body`` parameters
are mutually exclusive in an :class:`.Operation`. If you need to set custom ``form`` parameters, you can use
the `manual_parameters` argument.
If a ``Serializer`` class or instance is given, it will be automatically converted into a :class:`.Schema`
used as a ``body`` :class:`.Parameter`, or into a list of ``form`` :class:`.Parameter`\ s, as appropriate.
:param .Serializer query_serializer: if you use a ``Serializer`` to parse query parameters, you can pass it here
and have :class:`.Parameter` objects be generated automatically from it.
If any ``Field`` on the serializer cannot be represented as a ``query`` :class:`.Parameter`
(e.g. nested Serializers, file fields, ...), the schema generation will fail with an error.
Schema generation will also fail if the name of any Field on the `query_serializer` conflicts with parameters
generated by ``filter_backends`` or ``paginator``.
:param list[.Parameter] manual_parameters: a list of manual parameters to override the automatically generated ones
:class:`.Parameter`\ s are identified by their (``name``, ``in``) combination, and any parameters given
here will fully override automatically generated parameters if they collide.
It is an error to supply ``form`` parameters when the request does not consume form-data.
:param str operation_description: operation description override
:param dict[str,(.Schema,.SchemaRef,.Response,str,Serializer)] responses: a dict of documented manual responses
keyed on response status code. If no success (``2xx``) response is given, one will automatically be
generated from the request body and http method. If any ``2xx`` response is given the automatic response is
suppressed.
* if a plain string is given as value, a :class:`.Response` with no body and that string as its description
will be generated
* if a :class:`.Schema`, :class:`.SchemaRef` is given, a :class:`.Response` with the schema as its body and
an empty description will be generated
* a ``Serializer`` class or instance will be converted into a :class:`.Schema` and treated as above
* a :class:`.Response` object will be used as-is; however if its ``schema`` attribute is a ``Serializer``,
it will automatically be converted into a :class:`.Schema`
"""
def decorator(view_method):
data = {
'auto_schema': auto_schema,
'request_body': request_body,
'query_serializer': query_serializer,
'manual_parameters': manual_parameters,
'operation_description': operation_description,
'responses': responses,
}
data = {k: v for k, v in data.items() if v is not None}
# if the method is a detail_route or list_route, it will have a bind_to_methods attribute
bind_to_methods = getattr(view_method, 'bind_to_methods', [])
# if the method is actually a function based view (@api_view), it will have a 'cls' attribute
view_cls = getattr(view_method, 'cls', None)
http_method_names = getattr(view_cls, 'http_method_names', [])
if bind_to_methods or http_method_names:
# detail_route, list_route or api_view
assert bool(http_method_names) != bool(bind_to_methods), "this should never happen"
available_methods = http_method_names + bind_to_methods
existing_data = getattr(view_method, 'swagger_auto_schema', {})
if http_method_names:
_route = "api_view"
else:
_route = "detail_route" if view_method.detail else "list_route"
_methods = methods
if len(available_methods) > 1:
assert methods or method, \
"on multi-method %s, you must specify swagger_auto_schema on a per-method basis " \
"using one of the `method` or `methods` arguments" % _route
assert bool(methods) != bool(method), "specify either method or methods"
assert not isinstance(methods, str), "`methods` expects to receive a list of methods;" \
" use `method` for a single argument"
if method:
_methods = [method.lower()]
else:
_methods = [mth.lower() for mth in methods]
assert not any(mth in existing_data for mth in _methods), "method defined multiple times"
assert all(mth in available_methods for mth in _methods), "method not bound to %s" % _route
existing_data.update((mth.lower(), data) for mth in _methods)
else:
existing_data[available_methods[0]] = data
view_method.swagger_auto_schema = existing_data
else:
assert method is None and methods is None, \
"the methods argument should only be specified when decorating a detail_route or list_route; you " \
"should also ensure that you put the swagger_auto_schema decorator AFTER (above) the _route decorator"
view_method.swagger_auto_schema = data
return view_method
return decorator
def get_model_field(queryset, field_name):
"""Try to get information about a model and model field from a queryset.
:param queryset: the queryset
:param field_name: the target field name
:returns: the model and target field from the queryset as a 2-tuple; both elements can be ``None``
:rtype: tuple
"""
model = getattr(queryset, 'model', None)
try:
model_field = model._meta.get_field(field_name)
except Exception: # pragma: no cover
model_field = None
return model, model_field
model_field_to_swagger_type = [
(models.AutoField, (openapi.TYPE_INTEGER, None)),
(models.BinaryField, (openapi.TYPE_STRING, openapi.FORMAT_BINARY)),
(models.BooleanField, (openapi.TYPE_BOOLEAN, None)),
(models.NullBooleanField, (openapi.TYPE_BOOLEAN, None)),
(models.DateTimeField, (openapi.TYPE_STRING, openapi.FORMAT_DATETIME)),
(models.DateField, (openapi.TYPE_STRING, openapi.FORMAT_DATE)),
(models.DecimalField, (openapi.TYPE_NUMBER, None)),
(models.DurationField, (openapi.TYPE_INTEGER, None)),
(models.FloatField, (openapi.TYPE_NUMBER, None)),
(models.IntegerField, (openapi.TYPE_INTEGER, None)),
(models.IPAddressField, (openapi.TYPE_STRING, openapi.FORMAT_IPV4)),
(models.GenericIPAddressField, (openapi.TYPE_STRING, openapi.FORMAT_IPV6)),
(models.SlugField, (openapi.TYPE_STRING, openapi.FORMAT_SLUG)),
(models.TextField, (openapi.TYPE_STRING, None)),
(models.TimeField, (openapi.TYPE_STRING, None)),
(models.UUIDField, (openapi.TYPE_STRING, openapi.FORMAT_UUID)),
(models.CharField, (openapi.TYPE_STRING, None)),
]
def inspect_model_field(model, model_field):
"""Extract information from a django model field instance.
:param model: the django model
:param model_field: a field on the model
:return: description, type, format and pattern extracted from the model field
:rtype: OrderedDict
"""
if model is not None and model_field is not None:
for model_field_class, tf in model_field_to_swagger_type:
if isinstance(model_field, model_field_class):
swagger_type, format = tf
break
else: # pragma: no cover
swagger_type, format = None, None
if format is None or format == openapi.FORMAT_SLUG:
pattern = find_regex(model_field)
else:
pattern = None
if model_field.help_text:
description = force_text(model_field.help_text)
elif model_field.primary_key:
description = get_pk_description(model, model_field)
else:
description = None
def guess_response_status(method):
if method == 'post':
return status.HTTP_201_CREATED
elif method == 'delete':
return status.HTTP_204_NO_CONTENT
else:
description = None
swagger_type = None
format = None
pattern = None
result = OrderedDict([
('description', description),
('type', swagger_type or openapi.TYPE_STRING),
('format', format),
('pattern', pattern)
])
# TODO: filter none
return result
def serializer_field_to_swagger(field, swagger_object_type, definitions=None, **kwargs):
"""Convert a drf Serializer or Field instance into a Swagger object.
:param rest_framework.serializers.Field field: the source field
:param type[openapi.SwaggerDict] swagger_object_type: should be one of Schema, Parameter, Items
:param .ReferenceResolver definitions: used to serialize Schemas by reference
:param kwargs: extra attributes for constructing the object;
if swagger_object_type is Parameter, ``name`` and ``in_`` should be provided
:return: the swagger object
:rtype: openapi.Parameter, openapi.Items, openapi.Schema
"""
assert swagger_object_type in (openapi.Schema, openapi.Parameter, openapi.Items)
assert not isinstance(field, openapi.SwaggerDict), "passed field is already a SwaggerDict object"
title = force_text(field.label) if field.label else None
title = title if swagger_object_type == openapi.Schema else None # only Schema has title
title = None
description = force_text(field.help_text) if field.help_text else None
description = description if swagger_object_type != openapi.Items else None # Items has no description either
def SwaggerType(existing_object=None, **instance_kwargs):
if swagger_object_type == openapi.Parameter and 'required' not in instance_kwargs:
instance_kwargs['required'] = field.required
if swagger_object_type != openapi.Items and 'default' not in instance_kwargs:
default = getattr(field, 'default', serializers.empty)
if default is not serializers.empty:
if callable(default):
try:
if hasattr(default, 'set_context'):
default.set_context(field)
default = default()
except Exception as e: # pragma: no cover
logger.warning("default for %s is callable but it raised an exception when "
"called; 'default' field will not be added to schema", field, exc_info=True)
default = None
if default is not None:
try:
default = field.to_representation(default)
# JSON roundtrip ensures that the value is valid JSON;
# for example, sets get transformed into lists
default = json.loads(json.dumps(default, cls=encoders.JSONEncoder))
except Exception: # pragma: no cover
logger.warning("'default' on schema for %s will not be set because "
"to_representation raised an exception", field, exc_info=True)
default = None
if default is not None:
instance_kwargs['default'] = default
if swagger_object_type == openapi.Schema and 'read_only' not in instance_kwargs:
if field.read_only:
instance_kwargs['read_only'] = True
instance_kwargs.update(kwargs)
instance_kwargs.pop('title', None)
instance_kwargs.pop('description', None)
if existing_object is not None:
existing_object.title = title
existing_object.description = description
for attr, val in instance_kwargs.items():
setattr(existing_object, attr, val)
return existing_object
return swagger_object_type(title=title, description=description, **instance_kwargs)
# arrays in Schema have Schema elements, arrays in Parameter and Items have Items elements
ChildSwaggerType = openapi.Schema if swagger_object_type == openapi.Schema else openapi.Items
# ------ NESTED
if isinstance(field, (serializers.ListSerializer, serializers.ListField)):
child_schema = serializer_field_to_swagger(field.child, ChildSwaggerType, definitions)
return SwaggerType(
type=openapi.TYPE_ARRAY,
items=child_schema,
)
elif isinstance(field, serializers.Serializer):
if swagger_object_type != openapi.Schema:
raise SwaggerGenerationError("cannot instantiate nested serializer as " + swagger_object_type.__name__)
assert definitions is not None, "ReferenceResolver required when instantiating Schema"
serializer = field
if hasattr(serializer, '__ref_name__'):
ref_name = serializer.__ref_name__
else:
ref_name = type(serializer).__name__
if ref_name.endswith('Serializer'):
ref_name = ref_name[:-len('Serializer')]
def make_schema_definition():
properties = OrderedDict()
required = []
for key, value in serializer.fields.items():
properties[key] = serializer_field_to_swagger(value, ChildSwaggerType, definitions)
if value.required:
required.append(key)
return SwaggerType(
type=openapi.TYPE_OBJECT,
properties=properties,
required=required or None,
)
if not ref_name:
return make_schema_definition()
definitions.setdefault(ref_name, make_schema_definition)
return openapi.SchemaRef(definitions, ref_name)
elif isinstance(field, serializers.ManyRelatedField):
child_schema = serializer_field_to_swagger(field.child_relation, ChildSwaggerType, definitions)
return SwaggerType(
type=openapi.TYPE_ARRAY,
items=child_schema,
unique_items=True, # is this OK?
)
elif isinstance(field, serializers.PrimaryKeyRelatedField):
if field.pk_field:
result = serializer_field_to_swagger(field.pk_field, swagger_object_type, definitions, **kwargs)
return SwaggerType(existing_object=result)
attrs = {'type': openapi.TYPE_STRING}
try:
model = field.queryset.model
pk_field = model._meta.pk
except Exception: # pragma: no cover
logger.warning("an exception was raised when attempting to extract the primary key related to %s; "
"falling back to plain string" % field, exc_info=True)
else:
attrs.update(inspect_model_field(model, pk_field))
return SwaggerType(**attrs)
elif isinstance(field, serializers.HyperlinkedRelatedField):
return SwaggerType(type=openapi.TYPE_STRING, format=openapi.FORMAT_URI)
elif isinstance(field, serializers.SlugRelatedField):
model, model_field = get_model_field(field.queryset, field.slug_field)
attrs = inspect_model_field(model, model_field)
return SwaggerType(**attrs)
elif isinstance(field, serializers.RelatedField):
return SwaggerType(type=openapi.TYPE_STRING)
# ------ CHOICES
elif isinstance(field, serializers.MultipleChoiceField):
return SwaggerType(
type=openapi.TYPE_ARRAY,
items=ChildSwaggerType(
type=openapi.TYPE_STRING,
enum=list(field.choices.keys())
)
)
elif isinstance(field, serializers.ChoiceField):
return SwaggerType(type=openapi.TYPE_STRING, enum=list(field.choices.keys()))
# ------ BOOL
elif isinstance(field, (serializers.BooleanField, serializers.NullBooleanField)):
return SwaggerType(type=openapi.TYPE_BOOLEAN)
# ------ NUMERIC
elif isinstance(field, (serializers.DecimalField, serializers.FloatField)):
# TODO: min_value max_value
return SwaggerType(type=openapi.TYPE_NUMBER)
elif isinstance(field, serializers.IntegerField):
# TODO: min_value max_value
return SwaggerType(type=openapi.TYPE_INTEGER)
elif isinstance(field, serializers.DurationField):
return SwaggerType(type=openapi.TYPE_INTEGER)
# ------ STRING
elif isinstance(field, serializers.EmailField):
return SwaggerType(type=openapi.TYPE_STRING, format=openapi.FORMAT_EMAIL)
elif isinstance(field, serializers.RegexField):
return SwaggerType(type=openapi.TYPE_STRING, pattern=find_regex(field))
elif isinstance(field, serializers.SlugField):
return SwaggerType(type=openapi.TYPE_STRING, format=openapi.FORMAT_SLUG, pattern=find_regex(field))
elif isinstance(field, serializers.URLField):
return SwaggerType(type=openapi.TYPE_STRING, format=openapi.FORMAT_URI)
elif isinstance(field, serializers.IPAddressField):
format = {'ipv4': openapi.FORMAT_IPV4, 'ipv6': openapi.FORMAT_IPV6}.get(field.protocol, None)
return SwaggerType(type=openapi.TYPE_STRING, format=format)
elif isinstance(field, serializers.CharField):
# TODO: min_length max_length (for all CharField subclasses above too)
return SwaggerType(type=openapi.TYPE_STRING)
elif isinstance(field, serializers.UUIDField):
return SwaggerType(type=openapi.TYPE_STRING, format=openapi.FORMAT_UUID)
# ------ DATE & TIME
elif isinstance(field, serializers.DateField):
return SwaggerType(type=openapi.TYPE_STRING, format=openapi.FORMAT_DATE)
elif isinstance(field, serializers.DateTimeField):
return SwaggerType(type=openapi.TYPE_STRING, format=openapi.FORMAT_DATETIME)
# ------ OTHERS
elif isinstance(field, serializers.FileField):
# swagger 2.0 does not support specifics about file fields, so ImageFile gets no special treatment
# OpenAPI 3.0 does support it, so a future implementation could handle this better
err = SwaggerGenerationError("FileField is supported only in a formData Parameter or response Schema")
if swagger_object_type == openapi.Schema:
# FileField.to_representation returns URL or file name
result = SwaggerType(type=openapi.TYPE_STRING, read_only=True)
if getattr(field, 'use_url', api_settings.UPLOADED_FILES_USE_URL):
result.format = openapi.FORMAT_URI
return result
elif swagger_object_type == openapi.Parameter:
param = SwaggerType(type=openapi.TYPE_FILE)
if param['in'] != openapi.IN_FORM:
raise err # pragma: no cover
return param
else:
raise err # pragma: no cover
elif isinstance(field, serializers.DictField) and swagger_object_type == openapi.Schema:
child_schema = serializer_field_to_swagger(field.child, ChildSwaggerType, definitions)
return SwaggerType(
type=openapi.TYPE_OBJECT,
additional_properties=child_schema
)
elif isinstance(field, serializers.ModelField):
return SwaggerType(type=openapi.TYPE_STRING)
# TODO unhandled fields: TimeField HiddenField JSONField
# everything else gets string by default
return SwaggerType(type=openapi.TYPE_STRING)
def find_regex(regex_field):
"""Given a ``Field``, look for a ``RegexValidator`` and try to extract its pattern and return it as a string.
:param serializers.Field regex_field: the field instance
:return: the extracted pattern, or ``None``
:rtype: str
"""
regex_validator = None
for validator in regex_field.validators:
if isinstance(validator, RegexValidator):
if regex_validator is not None:
# bail if multiple validators are found - no obvious way to choose
return None # pragma: no cover
regex_validator = validator
# regex_validator.regex should be a compiled re object...
return getattr(getattr(regex_validator, 'regex', None), 'pattern', None)
return status.HTTP_200_OK
def param_list_to_odict(parameters):
@@ -492,3 +201,37 @@ def param_list_to_odict(parameters):
result = OrderedDict(((param.name, param.in_), param) for param in parameters)
assert len(result) == len(parameters), "duplicate Parameters found"
return result
def filter_none(obj):
"""Remove ``None`` values from tuples, lists or dictionaries. Return other objects as-is.
:param obj:
:return: collection with ``None`` values removed
"""
if obj is None:
return None
new_obj = None
if isinstance(obj, dict):
new_obj = type(obj)((k, v) for k, v in obj.items() if k is not None and v is not None)
if isinstance(obj, (list, tuple)):
new_obj = type(obj)(v for v in obj if v is not None)
if new_obj is not None and len(new_obj) != len(obj):
return new_obj # pragma: no cover
return obj
def force_serializer_instance(serializer):
"""Force `serializer` into a ``Serializer`` instance. If it is not a ``Serializer`` class or instance, raises
an assertion error.
:param serializer: serializer class or instance
:return: serializer instance
"""
if inspect.isclass(serializer):
assert issubclass(serializer, serializers.BaseSerializer), "Serializer required, not %s" % serializer.__name__
return serializer()
assert isinstance(serializer, serializers.BaseSerializer), \
"Serializer class or instance required, not %s" % type(serializer).__name__
return serializer
+14 -5
View File
@@ -82,12 +82,24 @@ def get_schema_view(info, url=None, patterns=None, urlconf=None, public=False, v
renderer_classes = _spec_renderers
def get(self, request, version='', format=None):
generator = self.generator_class(info, version, url, patterns, urlconf)
generator = self.generator_class(info, request.version or version or '', url, patterns, urlconf)
schema = generator.get_schema(request, self.public)
if schema is None:
raise exceptions.PermissionDenied() # pragma: no cover
return Response(schema)
@classmethod
def apply_cache(cls, view, cache_timeout, cache_kwargs):
"""Override this method to customize how caching is applied to the view.
Arguments described in :meth:`.as_cached_view`.
"""
if not cls.public:
view = vary_on_headers('Cookie', 'Authorization')(view)
view = cache_page(cache_timeout, **cache_kwargs)(view)
view = deferred_never_cache(view) # disable in-browser caching
return view
@classmethod
def as_cached_view(cls, cache_timeout=0, cache_kwargs=None, **initkwargs):
"""
@@ -102,10 +114,7 @@ def get_schema_view(info, url=None, patterns=None, urlconf=None, public=False, v
cache_kwargs = cache_kwargs or {}
view = cls.as_view(**initkwargs)
if cache_timeout != 0:
if not public:
view = vary_on_headers('Cookie', 'Authorization')(view)
view = cache_page(cache_timeout, **cache_kwargs)(view)
view = deferred_never_cache(view) # disable in-browser caching
view = cls.apply_cache(view, cache_timeout, cache_kwargs)
elif cache_kwargs:
warnings.warn("cache_kwargs ignored because cache_timeout is 0 (disabled)")
return view