Files
gen/openapi/preprocess_spec.py
2023-10-28 19:32:32 +02:00

593 lines
22 KiB
Python

# Copyright 2016 The Kubernetes Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import print_function
import json
import operator
import os.path
import sys
import argparse
from collections import OrderedDict
import urllib3
# these four constants are shown as part of this example in []:
# "[watch]Pod[List]" is the deprecated version of "[list]Pod?[watch]=True"
WATCH_OP_PREFIX = "watch"
WATCH_OP_SUFFIX = "List"
LIST_OP_PREFIX = "list"
DELETECOLLECTION_OP_PREFIX = "deleteCollection"
WATCH_QUERY_PARAM_NAME = "watch"
ALLOW_WATCH_BOOKMARKS_QUERY_PARAM_NAME = "allowWatchBookmarks"
CUSTOM_OBJECTS_SPEC_PATH = os.path.join(
os.path.dirname(__file__),
'custom_objects_spec.json')
_ops = ['get', 'put', 'post', 'delete', 'options', 'head', 'patch']
class PreprocessingException(Exception):
pass
def _title(s):
if len(s) == 0:
return s
return s[0].upper() + s[1:]
def _to_camel_case(s):
return ''.join(_title(y) for y in s.split("_"))
def apply_func_to_spec_operations(spec, func, *params):
"""Apply func to each operation in the spec.
:param spec: The OpenAPI spec to apply func to.
:param func: the function to apply to the spec's operations. It should be
a func(operation, parent) where operation will be each
operation of the spec and parent would be the parent object of
the given operation.
If the return value of the func is True, then the operation
will be deleted from the spec.
"""
for k, v in spec['paths'].items():
for op in _ops:
if op not in v:
continue
if func(v[op], v, *params):
del v[op]
def _has_property(prop_list, property_name):
for prop in prop_list:
if prop["name"] == property_name:
return True
def remove_watch_operations(op, parent, operation_ids):
op_id = op['operationId']
if not op_id.startswith(WATCH_OP_PREFIX):
return
list_id = (LIST_OP_PREFIX +
op_id.replace(WATCH_OP_SUFFIX, "")[len(WATCH_OP_PREFIX):])
if list_id not in operation_ids:
raise PreprocessingException("Cannot find %s" % list_id)
list_op = operation_ids[list_id]
params = []
if 'parameters' in list_op:
params += list_op['parameters']
if 'parameters' in parent:
params += parent['parameters']
if not _has_property(params, WATCH_QUERY_PARAM_NAME):
raise PreprocessingException("%s has no watch query param" % list_id)
return True
def strip_delete_collection_operation_watch_params(op, parent):
op_id = op['operationId']
if not op_id.startswith(DELETECOLLECTION_OP_PREFIX):
return
params = []
if 'parameters' in op:
for i in range(len(op['parameters'])):
paramName = op['parameters'][i].get('name', None)
if paramName != WATCH_QUERY_PARAM_NAME and paramName != ALLOW_WATCH_BOOKMARKS_QUERY_PARAM_NAME:
params.append(op['parameters'][i])
op['parameters'] = params
return False
def strip_401_response(operation, _):
if 'responses' in operation:
operation['responses'].pop('401', None)
if len(operation['responses']) == 0:
operation['responses']['200'] = { 'description': 'OK' }
def transform_to_csharp_stream_response(operation, _):
if operation.get('operationId', None) == 'readNamespacedPodLog' or operation.get('x-kubernetes-action', None) == 'connect':
operation['responses']['200']["schema"] = {
"type": "object",
"format": "file" ,
}
def transform_to_csharp_consume_json(operation, _):
if operation.get('consumes', None) == ["*/*",] or operation.get('consumes', None) == "*/*":
operation['consumes'] = ["application/json"]
def transform_to_java_consume_json(operation, _):
operation['consumes'] = ["application/json"]
def strip_tags_from_operation_id(operation, _):
operation_id = operation['operationId']
if 'tags' in operation:
for t in operation['tags']:
operation_id = operation_id.replace(_to_camel_case(t), '')
operation['operationId'] = operation_id
def clean_crd_meta(spec):
for k, v in spec['definitions'].items():
if k.endswith('List'):
print("Using built-in v1.ListMeta")
v['properties']['metadata']['$ref'] = '#/definitions/io.k8s.apimachinery.pkg.apis.meta.v1.ListMeta'
v['properties']['metadata'].pop('properties', None)
find_rename_ref_recursive(spec, '#/definitions/io.k8s.apimachinery.pkg.apis.meta.v1.ListMeta', '#/definitions/v1.ListMeta')
find_rename_ref_recursive(spec, '#/definitions/io.k8s.apimachinery.pkg.apis.meta.v1.ObjectMeta', '#/definitions/v1.ObjectMeta')
find_rename_ref_recursive(spec, '#/definitions/io.k8s.apimachinery.pkg.apis.meta.v1.ObjectMeta_v2', '#/definitions/v1.ObjectMeta')
find_rename_ref_recursive(spec, '#/definitions/io.k8s.apimachinery.pkg.apis.meta.v1.Status', '#/definitions/v1.Status')
find_rename_ref_recursive(spec, '#/definitions/io.k8s.apimachinery.pkg.apis.meta.v1.Status_v2', '#/definitions/v1.Status')
find_rename_ref_recursive(spec, '#/definitions/io.k8s.apimachinery.pkg.apis.meta.v1.Patch', '#/definitions/v1.Patch')
find_rename_ref_recursive(spec, '#/definitions/io.k8s.apimachinery.pkg.apis.meta.v1.DeleteOptions', '#/definitions/v1.DeleteOptions')
find_rename_ref_recursive(spec, '#/definitions/io.k8s.apimachinery.pkg.apis.meta.v1.DeleteOptions_v2', '#/definitions/v1.DeleteOptions')
find_rename_ref_recursive(spec, '#/definitions/io.k8s.api.autoscaling.v1.Scale', '#/definitions/v1.Scale')
find_rename_ref_recursive(spec, '#/definitions/io.k8s.api.autoscaling.v1.Scale_v2', '#/definitions/v1.Scale')
def add_custom_objects_spec(spec):
with open(CUSTOM_OBJECTS_SPEC_PATH, 'r') as custom_objects_spec_file:
custom_objects_spec = json.loads(custom_objects_spec_file.read())
for path in custom_objects_spec.keys():
if path not in spec['paths'].keys():
spec['paths'][path] = custom_objects_spec[path]
return spec
def add_codegen_request_body(operation, _):
if 'parameters' in operation and len(operation['parameters']) > 0:
if operation['parameters'][0].get('in') == 'body':
operation['x-codegen-request-body-name'] = 'body'
def drop_paths(spec):
paths = {}
if (os.environ.get('GENERATE_APIS') or False):
group_prefix = os.environ.get('KUBERNETES_CRD_GROUP_PREFIX')
group_prefix_reversed = '.'.join(group_prefix.split('.')[::-1])
for k, v in spec['paths'].items():
if k.startswith('/apis/' + group_prefix_reversed):
print("Adding Custom Resource api path %s" %k)
paths[k] = v
else:
print("Ignoring non Custom Resource api path %s" %k)
spec['paths'] = paths
def fix_paths(spec):
# see https://github.com/kubernetes/kubernetes/issues/117455
paths = spec['paths']
if '/.well-known/openid-configuration/' in paths:
paths['/.well-known/openid-configuration'] = paths['/.well-known/openid-configuration/']
del paths['/.well-known/openid-configuration/']
if '/openid/v1/jwks/' in paths:
paths['/openid/v1/jwks'] = paths['/openid/v1/jwks/']
del paths['/openid/v1/jwks/']
def expand_parameters(spec):
if 'parameters' not in spec:
return
for path in spec['paths'].keys():
if 'parameters' in spec['paths'][path]:
for i in range(len(spec['paths'][path]['parameters'])):
param = spec['paths'][path]['parameters'][i]
if '$ref' in param:
ref = param['$ref']
param = spec['parameters'][ref.split('/')[-1]]
spec['paths'][path]['parameters'][i] = param
for method in spec['paths'][path].keys():
if 'parameters' in spec['paths'][path][method]:
for i in range(len(spec['paths'][path][method]['parameters'])):
param = spec['paths'][path][method]['parameters'][i]
if '$ref' in param:
ref = param['$ref']
param = spec['parameters'][ref.split('/')[-1]]
spec['paths'][path][method]['parameters'][i] = param
del spec['parameters']
def process_swagger(spec, client_language, crd_mode=False):
spec = add_custom_objects_spec(spec)
if crd_mode:
drop_paths(spec)
fix_paths(spec)
expand_parameters(spec)
apply_func_to_spec_operations(spec, strip_tags_from_operation_id)
if client_language == "csharp":
# 401s in the spec block the csharp code generator from throwing on 401
apply_func_to_spec_operations(spec, strip_401_response)
# force to autorest to generate stream
apply_func_to_spec_operations(spec, transform_to_csharp_stream_response)
# force to consume json if */*
apply_func_to_spec_operations(spec, transform_to_csharp_consume_json)
if client_language == "java":
# force to consume json, need this change due to https://github.com/OpenAPITools/openapi-generator/pull/10769
apply_func_to_spec_operations(spec, transform_to_java_consume_json)
apply_func_to_spec_operations(spec, strip_delete_collection_operation_watch_params)
apply_func_to_spec_operations(spec, add_codegen_request_body)
operation_ids = {}
apply_func_to_spec_operations(spec, lambda op, _: operator.setitem(
operation_ids, op['operationId'], op))
try:
apply_func_to_spec_operations(
spec, remove_watch_operations, operation_ids)
except PreprocessingException as e:
print(e)
if crd_mode:
filter_api_group(spec)
remove_model_prefixes(spec, crd_mode)
inline_primitive_models(spec, preserved_primitives_for_language(client_language))
remove_bad_descriptions(spec, bad_description_pattern_for_language(client_language))
if crd_mode:
clean_crd_meta(spec)
add_custom_formatting(spec, format_for_language(client_language))
add_custom_typing(spec, type_for_language(client_language))
remove_models(spec, removed_models_for_language(client_language))
add_openapi_codegen_x_implement_extension(spec, client_language)
return spec
def preserved_primitives_for_language(client_language):
if client_language == "java":
return ["intstr.IntOrString", "resource.Quantity", "v1.Patch"]
elif client_language == "csharp":
return ["intstr.IntOrString", "resource.Quantity", "v1.Patch"]
elif client_language == "haskell-http-client":
return ["intstr.IntOrString", "resource.Quantity"]
elif client_language in ["typescript", "typescript-fetch"]:
return ["intstr.IntOrString", "v1.MicroTime"]
elif client_language == "c":
return ["intstr.IntOrString"]
else:
return []
def bad_description_pattern_for_language(client_language):
if client_language == 'typescript-fetch':
return '*/'
return None
def format_for_language(client_language):
if client_language == "java":
return {"resource.Quantity": "quantity", "v1.Patch": "patch"}
else:
return {}
def type_for_language(client_language):
if client_language == "java":
return {"v1.Patch": { "type": "string"}}
elif client_language == "typescript":
return {"v1.MicroTime": { "type": "string", "format": "date-time-micro" }}
elif client_language == "csharp":
return {
"v1.Patch": { "type": "object", "properties": {"content": { "type": "object"}} },
"resource.Quantity": { "type": "object", "properties": {"value": { "type": "string"}} },
"intstr.IntOrString" : { "type": "object", "properties": {"value": { "type": "string"}} },
}
else:
return {}
def removed_models_for_language(client_language):
if client_language == "haskell-http-client":
return ["intstr.IntOrString", "resource.Quantity"]
else:
return []
def rename_model(spec, old_name, new_name):
if new_name in spec['definitions']:
raise PreprocessingException(
"Cannot rename model %s. new name %s exists." %
(old_name, new_name))
find_rename_ref_recursive(spec,
"#/definitions/" + old_name,
"#/definitions/" + new_name)
spec['definitions'][new_name] = spec['definitions'][old_name]
del spec['definitions'][old_name]
def find_rename_ref_recursive(root, old, new):
if isinstance(root, list):
for r in root:
find_rename_ref_recursive(r, old, new)
if isinstance(root, dict):
if "$ref" in root:
if root["$ref"] == old:
root["$ref"] = new
for k, v in root.items():
find_rename_ref_recursive(v, old, new)
def is_model_deprecated(m):
"""
Check if a mode is deprecated model redirection.
A deprecated mode redirecation has only two members with a
description starts with "Deprecated." string.
"""
if len(m) != 2:
return False
if "$ref" not in m or "description" not in m:
return False
return m["description"].startswith("Deprecated.")
def filter_api_group(spec):
models = {}
for k, v in spec['definitions'].items():
if k.startswith(os.environ.get('KUBERNETES_CRD_GROUP_PREFIX')):
print("Adding Custom Resource by prefix %s" %k)
models[k] = v
elif k.startswith("io.k8s"):
print("Removing builtin Kubernetes Resource %s" %k)
else:
print("Ignoring Custom Resource %s" %k)
spec['definitions'] = models
def remove_deprecated_models(spec):
"""
In kubernetes 1.8 some of the models are renamed. Our remove_model_prefixes
still creates the same model names but there are some models added to
reference old model names to new names. These models broke remove_model_prefixes
and need to be removed.
"""
models = {}
for k, v in spec['definitions'].items():
if is_model_deprecated(v):
print("Removing deprecated model %s" % k)
else:
models[k] = v
spec['definitions'] = models
def remove_model_prefixes(spec, crd_mode=False):
"""Remove full package name from OpenAPI model names.
Starting kubernetes 1.6, all models has full package name. This is
verbose and inconvenient in python client. This function tries to remove
parts of the package name but will make sure there is no conflicting model
names. This will keep most of the model names generated by previous client
but will change some of them.
"""
remove_deprecated_models(spec)
models = {}
for k, v in spec['definitions'].items():
if k.startswith("io.k8s"):
models[k] = {"split_n": 2}
elif crd_mode and k.startswith(os.environ.get('KUBERNETES_CRD_GROUP_PREFIX')):
if os.environ.get('OPENAPI_MODEL_LENGTH') or False:
models[k] = {"split_n": int(os.environ.get('OPENAPI_MODEL_LENGTH'))}
conflict = True
while conflict:
for k, v in models.items():
splits = k.rsplit(".", v["split_n"])
v["removed_prefix"] = splits.pop(0)
v["new_name"] = ".".join(splits)
conflict = False
for k, v in models.items():
for k2, v2 in models.items():
if k != k2 and v["new_name"] == v2["new_name"]:
v["conflict"] = True
v2["conflict"] = True
conflict = True
if conflict:
for k, v in models.items():
if "conflict" in v:
print("Resolving conflict for %s" % k)
v["split_n"] += 1
del v["conflict"]
for k, v in models.items():
if "new_name" not in v:
raise PreprocessingException("Cannot rename model %s" % k)
print("Removing prefix %s from %s...\n" % (v["removed_prefix"], k))
rename_model(spec, k, v["new_name"])
def find_replace_ref_recursive(root, ref_name, replace_map):
if isinstance(root, list):
for r in root:
find_replace_ref_recursive(r, ref_name, replace_map)
if isinstance(root, dict):
if "$ref" in root:
if root["$ref"] == ref_name:
del root["$ref"]
for k, v in replace_map.items():
if k in root:
if k != "description":
raise PreprocessingException(
"Cannot inline model %s because of "
"conflicting key %s." % (ref_name, k))
continue
root[k] = v
for k, v in root.items():
find_replace_ref_recursive(v, ref_name, replace_map)
def remove_models(spec, to_remove_models):
for k in to_remove_models:
print("Removing model `%s " % k)
del spec['definitions'][k]
def inline_primitive_models(spec, excluded_primitives):
to_remove_models = []
for k, v in spec['definitions'].items():
if k in excluded_primitives:
continue
if "properties" not in v:
if k == "intstr.IntOrString":
v["type"] = "object"
if "type" not in v:
v["type"] = "object"
print("Making model `%s` inline as %s..." % (k, v["type"]))
find_replace_ref_recursive(spec, "#/definitions/" + k, v)
to_remove_models.append(k)
for k in to_remove_models:
del spec['definitions'][k]
def remove_bad_descriptions_recursively(model, pattern):
desc = model.get('description', '')
if desc.find(pattern) >= 0:
model['description'] = ''
if 'properties' not in model:
return
for prop in model['properties'].values():
remove_bad_descriptions_recursively(prop, pattern)
def remove_bad_descriptions(spec, pattern):
if not pattern:
return
for model in spec['definitions'].values():
remove_bad_descriptions_recursively(model, pattern)
def add_custom_formatting(spec, custom_formats):
for k, v in spec['definitions'].items():
if k not in custom_formats:
continue
v["format"] = custom_formats[k]
def add_custom_typing(spec, custom_types):
for k, v in spec['definitions'].items():
if k not in custom_types:
continue
v.update(custom_types[k])
def add_openapi_codegen_x_implement_extension(spec, client_language):
if client_language != "java":
return
if os.environ.get('OPENAPI_SKIP_BASE_INTERFACE') or False:
return
for k, v in spec['definitions'].items():
if "x-kubernetes-group-version-kind" not in v:
continue
if k == "v1.Status":
# Status is explicitly exlucded because it's obviously not a list object
# but it has ListMeta.
continue
if "metadata" not in v['properties']:
continue # not a legitimate kubernetes api object
if v["properties"]["metadata"]["$ref"] == "#/definitions/v1.ListMeta":
if "x-implements" not in v:
v["x-implements"] = []
v["x-implements"].append("io.kubernetes.client.common.KubernetesListObject")
elif v["properties"]["metadata"]["$ref"] == "#/definitions/v1.ObjectMeta":
if "x-implements" not in v:
v["x-implements"] = []
v["x-implements"].append("io.kubernetes.client.common.KubernetesObject")
def write_json(filename, object):
with open(filename, 'w') as out:
json.dump(object, out, sort_keys=False, indent=2, separators=(',', ': '), ensure_ascii=True)
def main():
argparser = argparse.ArgumentParser()
argparser.add_argument(
'client_language',
help='Client language to setup spec for'
)
argparser.add_argument(
'kubernetes_branch',
help='Branch/tag of github.com/kubernetes/kubernetes to get spec from'
)
argparser.add_argument(
'output_spec_path',
help='Path to output spec file to'
)
argparser.add_argument(
'username',
help='Optional username if working on forks',
default='kubernetes'
)
argparser.add_argument(
'repository',
help='Optional repository name if working with kubernetes ecosystem projects',
default='kubernetes'
)
args = argparser.parse_args()
unprocessed_spec = args.output_spec_path + ".unprocessed"
in_spec = ""
if os.environ.get("OPENAPI_SKIP_FETCH_SPEC") or False:
with open(unprocessed_spec, 'r') as content:
in_spec = json.load(content, object_pairs_hook=OrderedDict)
else:
pool = urllib3.PoolManager()
spec_url = 'https://raw.githubusercontent.com/%s/%s/' \
'%s/api/openapi-spec/swagger.json' % (args.username,
args.repository,
args.kubernetes_branch)
with pool.request('GET', spec_url, preload_content=False) as response:
if response.status != 200:
print("Error downloading spec file %s. Reason: %s" % (spec_url, response.reason))
return 1
in_spec = json.load(response, object_pairs_hook=OrderedDict)
write_json(unprocessed_spec, in_spec)
# use version from branch/tag name if spec doesn't provide it
if in_spec['info']['version'] == 'unversioned':
in_spec['info']['version'] = args.kubernetes_branch
crd_mode = os.environ.get('KUBERNETES_CRD_MODE') or False
out_spec = process_swagger(in_spec, args.client_language, crd_mode)
write_json(args.output_spec_path, out_spec)
return 0
if __name__ == '__main__':
sys.exit(main())