scalable-pypeline 1.1.0__py2.py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- pypeline/__init__.py +1 -0
- pypeline/celery.py +270 -0
- pypeline/celery_beat.py +254 -0
- pypeline/cli/__init__.py +0 -0
- pypeline/cli/config_server.py +48 -0
- pypeline/cli/core.py +32 -0
- pypeline/cli/deploy.py +138 -0
- pypeline/cloud.py +80 -0
- pypeline/constants.py +139 -0
- pypeline/deploy.py +167 -0
- pypeline/extensions.py +16 -0
- pypeline/flask/__init__.py +28 -0
- pypeline/flask/api/__init__.py +0 -0
- pypeline/flask/api/pipelines.py +245 -0
- pypeline/flask/api/schedules.py +67 -0
- pypeline/flask/api/utils.py +36 -0
- pypeline/flask/decorators.py +92 -0
- pypeline/flask/flask_sermos.py +219 -0
- pypeline/generators.py +196 -0
- pypeline/lib/__init__.py +0 -0
- pypeline/lib/config_server.py +159 -0
- pypeline/logging_config.py +171 -0
- pypeline/pipeline_config_schema.py +197 -0
- pypeline/schedule_config_schema.py +210 -0
- pypeline/sermos_yaml.py +737 -0
- pypeline/utils/__init__.py +0 -0
- pypeline/utils/config_utils.py +327 -0
- pypeline/utils/graph_utils.py +144 -0
- pypeline/utils/module_utils.py +119 -0
- pypeline/utils/task_utils.py +803 -0
- scalable_pypeline-1.1.0.dist-info/LICENSE +177 -0
- scalable_pypeline-1.1.0.dist-info/METADATA +166 -0
- scalable_pypeline-1.1.0.dist-info/RECORD +38 -0
- scalable_pypeline-1.1.0.dist-info/WHEEL +6 -0
- scalable_pypeline-1.1.0.dist-info/entry_points.txt +2 -0
- scalable_pypeline-1.1.0.dist-info/top_level.txt +2 -0
- tests/fixtures/__init__.py +1 -0
- tests/fixtures/s3_fixtures.py +52 -0
pypeline/sermos_yaml.py
ADDED
@@ -0,0 +1,737 @@
|
|
1
|
+
""" Definition of the `sermos.yaml` file. This is only relevant/used for
|
2
|
+
managed deployments through Sermos.ai. If self-hosting, safely disregard this
|
3
|
+
yaml format, no `sermos.yaml` is required for your application.
|
4
|
+
|
5
|
+
If using, a basic file may look like::
|
6
|
+
|
7
|
+
imageConfig:
|
8
|
+
- name: base-image
|
9
|
+
installCommand: sermos-demo-client[core]
|
10
|
+
- name: public-api-image
|
11
|
+
repositoryUrl: myregistry/public-api-image
|
12
|
+
|
13
|
+
environmentVariables:
|
14
|
+
- name: GLOBAL_ENV_VAR
|
15
|
+
value: globally-available-env-var
|
16
|
+
|
17
|
+
serviceConfig:
|
18
|
+
- name: base-worker
|
19
|
+
serviceType: celery-worker
|
20
|
+
imageName: base-image
|
21
|
+
queue: default-queue
|
22
|
+
registeredTasks:
|
23
|
+
- handler: sermos_demo_client.workers.demo_worker.demo_task
|
24
|
+
|
25
|
+
|
26
|
+
imageConfig:
|
27
|
+
- name: base-image
|
28
|
+
installCommand: sermos-demo-client
|
29
|
+
sourceSshUrl: git@gitlab.com:sermos/sermos-demo-client.git
|
30
|
+
baseImage: registry.gitlab.com/sermos/sermos-tools:0.3.0
|
31
|
+
|
32
|
+
serviceConfig:
|
33
|
+
- name: demo-api
|
34
|
+
serviceType: external
|
35
|
+
serviceId: ${SERVICE_ID_API} # Rendered using `sermos deploy` if available in the environment.
|
36
|
+
imageName: base-image
|
37
|
+
command: gunicorn --log-level info -k gevent -b 0.0.0.0:5000 sermos_demo_client.app:create_app()
|
38
|
+
port: 5000
|
39
|
+
replicaCount: 1
|
40
|
+
cpuLimit: 0.5
|
41
|
+
memoryLimit: 0.5
|
42
|
+
environmentVariables:
|
43
|
+
- name: FLASK_SECRET_KEY
|
44
|
+
value: ${FLASK_SECRET_KEY}
|
45
|
+
- name: sermos-worker
|
46
|
+
serviceType: celery-worker
|
47
|
+
serviceId: ${SERVICE_ID_WORKER}
|
48
|
+
imageName: base-image
|
49
|
+
command: celery -A sermos_demo_client.celery worker --without-gossip --without-mingle -c '1' -l INFO --queue default-task-queue
|
50
|
+
replicaCount: 1
|
51
|
+
cpuLimit: 0.5
|
52
|
+
memoryLimit: 0.5
|
53
|
+
registeredTasks:
|
54
|
+
- handler: sermos_demo_client.workers.demo_worker.demo_worker_task
|
55
|
+
- handler: sermos_demo_client.workers.demo_worker.demo_model_task
|
56
|
+
environmentVariables:
|
57
|
+
- name: WORKER_NAME
|
58
|
+
value: sermos-worker
|
59
|
+
|
60
|
+
pipelines:
|
61
|
+
demo-pipeline:
|
62
|
+
name: demo-pipeline
|
63
|
+
description: Demo Pipeline.
|
64
|
+
schemaVersion: 1
|
65
|
+
config:
|
66
|
+
dagAdjacency:
|
67
|
+
node_a:
|
68
|
+
- node_b
|
69
|
+
- node_c
|
70
|
+
metadata:
|
71
|
+
maxRetry: 3
|
72
|
+
maxTtl: 60
|
73
|
+
queue: default-task-queue
|
74
|
+
taskDefinitions:
|
75
|
+
node_a:
|
76
|
+
handler: sermos_demo_client.workers.demo_pipeline.demo_pipeline_node_a
|
77
|
+
node_b:
|
78
|
+
handler: sermos_demo_client.workers.demo_pipeline.demo_pipeline_node_b
|
79
|
+
queue: node-b-queue
|
80
|
+
node_c:
|
81
|
+
handler: sermos_demo_client.workers.demo_pipeline.demo_pipeline_node_c
|
82
|
+
|
83
|
+
scheduledTasks:
|
84
|
+
demo-model-task:
|
85
|
+
name: Demo Model Task
|
86
|
+
enabled: true
|
87
|
+
config:
|
88
|
+
scheduleType: interval
|
89
|
+
task: sermos_demo_client.workers.demo_worker.demo_model_task
|
90
|
+
queue: default-task-queue
|
91
|
+
schedule:
|
92
|
+
every: 60
|
93
|
+
period: seconds
|
94
|
+
schemaVersion: 1
|
95
|
+
|
96
|
+
"""
|
97
|
+
import re
|
98
|
+
import os
|
99
|
+
import logging
|
100
|
+
import pkg_resources
|
101
|
+
import yaml
|
102
|
+
from yaml.loader import SafeLoader
|
103
|
+
from marshmallow import Schema, fields, pre_load, EXCLUDE, INCLUDE,\
|
104
|
+
validates_schema
|
105
|
+
from marshmallow.validate import OneOf
|
106
|
+
from marshmallow.exceptions import ValidationError
|
107
|
+
from pypeline.utils.module_utils import SermosModuleLoader, normalized_pkg_name
|
108
|
+
from pypeline.constants import SERMOS_YAML_PATH, SERMOS_CLIENT_PKG_NAME
|
109
|
+
from pypeline.pipeline_config_schema import BasePipelineSchema
|
110
|
+
from pypeline.schedule_config_schema import BaseScheduleSchema
|
111
|
+
|
112
|
+
logger = logging.getLogger(__name__)
|
113
|
+
|
114
|
+
|
115
|
+
class InvalidPackagePath(Exception):
|
116
|
+
pass
|
117
|
+
|
118
|
+
|
119
|
+
class InvalidSermosConfig(Exception):
|
120
|
+
pass
|
121
|
+
|
122
|
+
|
123
|
+
class MissingSermosConfig(Exception):
|
124
|
+
pass
|
125
|
+
|
126
|
+
|
127
|
+
class InvalidImageConfig(Exception):
|
128
|
+
pass
|
129
|
+
|
130
|
+
|
131
|
+
class ExcludeUnknownSchema(Schema):
|
132
|
+
class Meta:
|
133
|
+
unknown = EXCLUDE
|
134
|
+
|
135
|
+
|
136
|
+
class EnvironmentVariableSchema(ExcludeUnknownSchema):
|
137
|
+
""" A single environment variables (singular)
|
138
|
+
"""
|
139
|
+
name = fields.String(required=True,
|
140
|
+
description="Environment variable name.",
|
141
|
+
example="MY_ENV_VAR")
|
142
|
+
value = fields.String(required=True,
|
143
|
+
description="Environment variable value.",
|
144
|
+
example="my special value")
|
145
|
+
|
146
|
+
|
147
|
+
class EnvironmentVariablesSchema(Schema):
|
148
|
+
""" Multiple environment variables (plural)
|
149
|
+
"""
|
150
|
+
environmentVariables = fields.List(
|
151
|
+
fields.Nested(EnvironmentVariableSchema, required=True),
|
152
|
+
description="List of name/value environment variable pairs available "
|
153
|
+
"to the scope of this service.",
|
154
|
+
required=False)
|
155
|
+
|
156
|
+
|
157
|
+
class ServiceRequestsSchema(Schema):
|
158
|
+
|
159
|
+
replicaCount = fields.Integer(
|
160
|
+
required=False,
|
161
|
+
description="Baseline (min) scale of this service to have available.",
|
162
|
+
default=1,
|
163
|
+
example=1)
|
164
|
+
cpuRequest = fields.Float(
|
165
|
+
required=False,
|
166
|
+
description="Requested CPUs to be available for each replica.",
|
167
|
+
default=0.5,
|
168
|
+
example="0.5")
|
169
|
+
memoryRequest = fields.Float(
|
170
|
+
required=False,
|
171
|
+
description="Requested memory (in GB) to be available for each replica.",
|
172
|
+
default=0.5,
|
173
|
+
example="0.5 (means half of 1 GB)")
|
174
|
+
ephemeralStorageRequest = fields.Float(
|
175
|
+
required=False,
|
176
|
+
description="Requested ephemeral storage (in GB) to "
|
177
|
+
"be available for each replica.",
|
178
|
+
default=8,
|
179
|
+
example="2 (means 2 GB)")
|
180
|
+
cpuLimit = fields.Float(
|
181
|
+
required=False,
|
182
|
+
description="Maximum CPUs to be available for each replica.",
|
183
|
+
default=0.5,
|
184
|
+
example="0.5")
|
185
|
+
memoryLimit = fields.Float(
|
186
|
+
required=False,
|
187
|
+
description="Maximum memory (in GB) to be available for each replica.",
|
188
|
+
default=0.5,
|
189
|
+
example="0.5 (means half of 1 GB)")
|
190
|
+
ephemeralStorageLimit = fields.Float(
|
191
|
+
required=False,
|
192
|
+
description="Maximum ephemeral storage (in GB) to "
|
193
|
+
"be available for each replica.",
|
194
|
+
default=8,
|
195
|
+
example="2 (means 2 GB)")
|
196
|
+
|
197
|
+
|
198
|
+
class NameSchema(Schema):
|
199
|
+
""" Validated name string field.
|
200
|
+
"""
|
201
|
+
name = fields.String(
|
202
|
+
required=True,
|
203
|
+
description="Name for service or image. Must include "
|
204
|
+
"only alphanumeric characters along with `_` and `-`.",
|
205
|
+
example="my-service-name")
|
206
|
+
|
207
|
+
@pre_load
|
208
|
+
def validate_characters(self, item, **kwargs):
|
209
|
+
""" Ensure name field conforms to allowed characters
|
210
|
+
"""
|
211
|
+
valid_chars = r'^[\w\d\-\_]+$'
|
212
|
+
if not bool(re.match(valid_chars, item['name'])):
|
213
|
+
raise ValueError(
|
214
|
+
f"Invalid name: {item['name']}. Only alphanumeric characters "
|
215
|
+
"allowed along with `-` and `_`.")
|
216
|
+
return item
|
217
|
+
|
218
|
+
|
219
|
+
class SermosImageConfigSchema(NameSchema):
|
220
|
+
installCommand = fields.String(
|
221
|
+
required=False,
|
222
|
+
description="The pip install command to use when Sermos is "
|
223
|
+
"responsible for building your image.",
|
224
|
+
example="sermos-client-pkg[core,special_feature]")
|
225
|
+
sourceSshUrl = fields.String(
|
226
|
+
required=False,
|
227
|
+
description="The source code ssh url to use when Sermos is "
|
228
|
+
"responsible for building your image.",
|
229
|
+
example="git@github.com:myorg/sermos-client.git")
|
230
|
+
baseImage = fields.String(
|
231
|
+
required=False,
|
232
|
+
description="The Docker base image to use as the starting point when "
|
233
|
+
"Sermos is responsible for building your image",
|
234
|
+
example="rhoai/sermos:latest")
|
235
|
+
repositoryUrl = fields.String(
|
236
|
+
required=False,
|
237
|
+
description="Deprecated - Use imageUri instead.")
|
238
|
+
repositoryUser = fields.String(
|
239
|
+
required=False,
|
240
|
+
description="Deprecated - Use registryUser instead.",)
|
241
|
+
repositoryPassword = fields.String(
|
242
|
+
required=False,
|
243
|
+
description="Deprecated - Use registryPassword instead.")
|
244
|
+
imageUri = fields.String(
|
245
|
+
required=False,
|
246
|
+
description="The Docker image uri when using an image not built by "
|
247
|
+
"Sermos. Tag is optional, if excluded, `latest` is used.",
|
248
|
+
example="rhoai/custom-image ; rhoai/custom-image:v0.0.0")
|
249
|
+
registryDomain = fields.String(
|
250
|
+
required=False,
|
251
|
+
description="Registry domain for using a pre-built image.",
|
252
|
+
example="index.docker.io")
|
253
|
+
registryUser = fields.String(
|
254
|
+
required=False,
|
255
|
+
description="Optional registry username if provided registryDomain"
|
256
|
+
"is a private registry.",
|
257
|
+
example="rhoai")
|
258
|
+
registryPassword = fields.String(
|
259
|
+
required=False,
|
260
|
+
description="Optional registry password if provided registryDomain "
|
261
|
+
"is a private registry. NOTE: Strongly recommended to use environment "
|
262
|
+
"variable interpolation in your sermos.yaml file, do not commit "
|
263
|
+
"unencrypted secrets to a git repository. "
|
264
|
+
"e.g. repositoryPassword: ${DOCKER_REPOSITORY_PASSWORD}",
|
265
|
+
example="abc123")
|
266
|
+
|
267
|
+
|
268
|
+
class SermosSharedConfigSchema(Schema):
|
269
|
+
""" Attributes shared across internal and external service types
|
270
|
+
"""
|
271
|
+
command = fields.String(
|
272
|
+
required=False,
|
273
|
+
_required=True,
|
274
|
+
description="Command to be run as container CMD.",
|
275
|
+
example="gunicorn -b 0.0.0.0:5000 package.app:create_app()")
|
276
|
+
port = fields.Integer(
|
277
|
+
required=False,
|
278
|
+
_required=True,
|
279
|
+
description="Port (and targetPort) to direct traffic.",
|
280
|
+
example=5000)
|
281
|
+
|
282
|
+
|
283
|
+
class SermosExternalConfigSchema(SermosSharedConfigSchema):
|
284
|
+
""" Attributes required for serviceType: external
|
285
|
+
|
286
|
+
Note: Schema lists these are not required in order for this to be used
|
287
|
+
as a mixin to SermosServiceConfigSchema. The validation is done
|
288
|
+
programmatically based on serviceType.
|
289
|
+
"""
|
290
|
+
serviceId = fields.String(
|
291
|
+
required=False,
|
292
|
+
_required=True,
|
293
|
+
description="The serviceId provided by Sermos. Find in admin console.",
|
294
|
+
example="dry-gorge-8018")
|
295
|
+
|
296
|
+
|
297
|
+
class SermosInternalSchema(SermosSharedConfigSchema):
|
298
|
+
""" Attributes required for serviceType: internal
|
299
|
+
|
300
|
+
Note: Schema lists these are not required in order for this to be used
|
301
|
+
as a mixin to SermosServiceConfigSchema. The validation is done
|
302
|
+
programmatically based on serviceType.
|
303
|
+
"""
|
304
|
+
protocol = fields.String(required=False,
|
305
|
+
_required=True,
|
306
|
+
description="Protocol to use.",
|
307
|
+
example="TCP",
|
308
|
+
validate=OneOf(['TCP', 'UDP']))
|
309
|
+
|
310
|
+
|
311
|
+
class SermosRegisteredTaskDetailConfigSchema(Schema):
|
312
|
+
handler = fields.String(
|
313
|
+
required=True,
|
314
|
+
description="Full path to the Method handles work / pipeline tasks.",
|
315
|
+
example="sermos_customer_client.workers.worker_group.useful_worker")
|
316
|
+
|
317
|
+
event = fields.Raw(
|
318
|
+
required=False,
|
319
|
+
unknown=INCLUDE,
|
320
|
+
description="Arbitrary user data, passed through `event` arg in task.")
|
321
|
+
|
322
|
+
|
323
|
+
class SermosCeleryWorkerConfigSchema(Schema):
|
324
|
+
""" Attributes required for serviceType: celery-worker
|
325
|
+
|
326
|
+
Note: Schema lists these are not required in order for this to be used
|
327
|
+
as a mixin to SermosServiceConfigSchema. The validation is done
|
328
|
+
programmatically based on serviceType.
|
329
|
+
"""
|
330
|
+
command = fields.String(
|
331
|
+
required=False,
|
332
|
+
_required=True,
|
333
|
+
description="Command to be run as container CMD.",
|
334
|
+
example="celery -A mypkg.celery.celery worker --queue my-queue")
|
335
|
+
|
336
|
+
registeredTasks = fields.List(
|
337
|
+
fields.Nested(SermosRegisteredTaskDetailConfigSchema, required=True),
|
338
|
+
required=False,
|
339
|
+
_required=True,
|
340
|
+
description="List of task handlers to register for to your Sermos app."
|
341
|
+
)
|
342
|
+
|
343
|
+
|
344
|
+
# Mapping of serviceType keys to their respective schema
|
345
|
+
service_types = {
|
346
|
+
'external': SermosExternalConfigSchema,
|
347
|
+
'internal': SermosInternalSchema,
|
348
|
+
'celery-worker': SermosCeleryWorkerConfigSchema
|
349
|
+
}
|
350
|
+
|
351
|
+
|
352
|
+
class SermosServiceConfigSchema(ExcludeUnknownSchema, ServiceRequestsSchema,
|
353
|
+
EnvironmentVariablesSchema,
|
354
|
+
SermosExternalConfigSchema,
|
355
|
+
SermosInternalSchema,
|
356
|
+
SermosCeleryWorkerConfigSchema, NameSchema):
|
357
|
+
""" Base service config object definition for workers and internal/external
|
358
|
+
services.
|
359
|
+
"""
|
360
|
+
imageName = fields.String(
|
361
|
+
required=True,
|
362
|
+
description="Specify the name of the base image to use for this "
|
363
|
+
"service.",
|
364
|
+
example="custom-worker-name")
|
365
|
+
|
366
|
+
serviceType = fields.String(required=True,
|
367
|
+
description="Name of the worker.",
|
368
|
+
example="useful_worker",
|
369
|
+
validate=OneOf(service_types.keys()))
|
370
|
+
|
371
|
+
|
372
|
+
class SermosYamlSchema(ExcludeUnknownSchema, EnvironmentVariablesSchema):
|
373
|
+
""" The primary `sermos.yaml` file schema. This defines all available
|
374
|
+
properties in a valid Sermos configuration file.
|
375
|
+
"""
|
376
|
+
|
377
|
+
imageConfig = fields.List(fields.Nested(SermosImageConfigSchema,
|
378
|
+
required=True),
|
379
|
+
required=True,
|
380
|
+
description="List of available base images. At "
|
381
|
+
"least one image must be defined. The 'name' "
|
382
|
+
"of the image is used in each service defined "
|
383
|
+
"in `serviceConfig`")
|
384
|
+
|
385
|
+
serviceConfig = fields.List(
|
386
|
+
fields.Nested(SermosServiceConfigSchema,
|
387
|
+
required=True,
|
388
|
+
description="Core service configuration."),
|
389
|
+
description="List of workers for Sermos to manage.",
|
390
|
+
required=True)
|
391
|
+
|
392
|
+
pipelines = fields.Dict(keys=fields.String(),
|
393
|
+
values=fields.Nested(BasePipelineSchema),
|
394
|
+
description="List of pipelines",
|
395
|
+
required=False)
|
396
|
+
|
397
|
+
scheduledTasks = fields.Dict(keys=fields.String(),
|
398
|
+
values=fields.Nested(BaseScheduleSchema),
|
399
|
+
description="List of scheduled tasks",
|
400
|
+
required=False)
|
401
|
+
|
402
|
+
def validate_errors(self, schema: Schema, value: dict):
|
403
|
+
""" Run Marshmallow validate() and raise if any errors
|
404
|
+
"""
|
405
|
+
schema = schema()
|
406
|
+
errors = schema.validate(value)
|
407
|
+
if len(errors.keys()) > 0:
|
408
|
+
raise ValidationError(errors)
|
409
|
+
|
410
|
+
@validates_schema
|
411
|
+
def validate_schema(self, data, **kwargs):
|
412
|
+
""" Additional validation.
|
413
|
+
|
414
|
+
Nested fields that are not required are not validated by Marshmallow
|
415
|
+
by default. Do a single level down of validation for now.
|
416
|
+
|
417
|
+
Each serviceType has attributes that are required but are listed
|
418
|
+
as not required in the marshmallow schema. Validate here.
|
419
|
+
|
420
|
+
imageConfig can provide *either* an install command for Sermos
|
421
|
+
to use to build the image for customer *or* a Docker repository
|
422
|
+
for Sermos to pull.
|
423
|
+
"""
|
424
|
+
# Vaidate nested
|
425
|
+
key_schema_pairs = (
|
426
|
+
('imageConfig', SermosImageConfigSchema),
|
427
|
+
('environmentVariables', EnvironmentVariableSchema),
|
428
|
+
('serviceConfig', SermosServiceConfigSchema),
|
429
|
+
)
|
430
|
+
for k_s in key_schema_pairs:
|
431
|
+
val = data.get(k_s[0], None)
|
432
|
+
if val is not None:
|
433
|
+
if type(val) == list:
|
434
|
+
for v in val:
|
435
|
+
self.validate_errors(k_s[1], v)
|
436
|
+
else:
|
437
|
+
self.validate_errors(k_s[1], val)
|
438
|
+
|
439
|
+
# Validate the services. We list every service schema field as not
|
440
|
+
# required in order to use them as mixins for a generic service object,
|
441
|
+
# however, they ARE required, so validate here using the custom
|
442
|
+
# metadata property `_required`. Default to value of `required`.
|
443
|
+
service_image_names = []
|
444
|
+
for service in data.get('serviceConfig'):
|
445
|
+
service_type = service['serviceType']
|
446
|
+
schema = service_types[service_type]
|
447
|
+
for field in schema().fields:
|
448
|
+
try:
|
449
|
+
if schema().fields[field].metadata.get(
|
450
|
+
'_required',
|
451
|
+
getattr(schema().fields[field], 'required')):
|
452
|
+
assert field in service
|
453
|
+
except AssertionError:
|
454
|
+
raise ValidationError(
|
455
|
+
f"`{field}` missing in {service_type} definition.")
|
456
|
+
|
457
|
+
service_image_names.append(service['imageName'])
|
458
|
+
|
459
|
+
# Validate imageConfig
|
460
|
+
image_names = []
|
461
|
+
for image in data.get('imageConfig'):
|
462
|
+
image_names.append(image['name'])
|
463
|
+
try:
|
464
|
+
if image.get('installCommand', None) is not None:
|
465
|
+
assert image.get('repositoryUrl', None) is None
|
466
|
+
if image.get('repositoryUrl', None) is not None:
|
467
|
+
assert image.get('installCommand', None) is None
|
468
|
+
except AssertionError:
|
469
|
+
raise InvalidImageConfig(
|
470
|
+
"Each imageConfig may have *either* an installCommand "
|
471
|
+
"*or* a repositoryUrl but not both. Review "
|
472
|
+
f"image `{image['name']}`")
|
473
|
+
|
474
|
+
if image.get('installCommand', None) is not None:
|
475
|
+
if image.get('sourceSshUrl') is None:
|
476
|
+
raise InvalidImageConfig(
|
477
|
+
"Each imageConfig must have a sourceSshUrl if "
|
478
|
+
"installCommand is specified")
|
479
|
+
if image.get('baseImage') is None:
|
480
|
+
raise InvalidImageConfig(
|
481
|
+
"Each imageConfig must have a baseImage if "
|
482
|
+
"installCommand is specified")
|
483
|
+
|
484
|
+
# Verify each imageName referenced in each service exists in imageConfig
|
485
|
+
try:
|
486
|
+
assert set(service_image_names).issubset(image_names)
|
487
|
+
except AssertionError:
|
488
|
+
raise InvalidImageConfig(
|
489
|
+
"Mismatched imageName in at least one "
|
490
|
+
f"service ({service_image_names}) compared to images available "
|
491
|
+
f"in imageConfig ({image_names})")
|
492
|
+
|
493
|
+
# Validate unique pipeline ids
|
494
|
+
if 'pipelines' in data:
|
495
|
+
pipeline_ids = set()
|
496
|
+
for pipeline_id, pipeline_data in data['pipelines'].items():
|
497
|
+
if pipeline_id in pipeline_ids:
|
498
|
+
raise ValidationError("All pipeline ids must be unique!")
|
499
|
+
pipeline_ids.add(pipeline_id)
|
500
|
+
schema_version = pipeline_data['schemaVersion']
|
501
|
+
PipelineSchema = \
|
502
|
+
BasePipelineSchema.get_by_version(schema_version)
|
503
|
+
self.validate_errors(PipelineSchema, pipeline_data)
|
504
|
+
|
505
|
+
# Validate unique scheduled tasks names
|
506
|
+
if 'scheduledTasks' in data:
|
507
|
+
task_ids = set()
|
508
|
+
for task_id, task_data in data['scheduledTasks'].items():
|
509
|
+
if task_id in task_ids:
|
510
|
+
raise ValidationError("All schedule ids must be unique!")
|
511
|
+
task_ids.add(task_id)
|
512
|
+
schema_version = task_data['schemaVersion']
|
513
|
+
TaskSchema = BaseScheduleSchema.get_by_version(schema_version)
|
514
|
+
self.validate_errors(TaskSchema, task_data)
|
515
|
+
|
516
|
+
|
517
|
+
class YamlPatternConstructor():
|
518
|
+
""" Adds a pattern resolver + constructor to PyYaml.
|
519
|
+
|
520
|
+
Typical/deault usage is for parsing environment variables
|
521
|
+
in a yaml file but this can be used for any pattern you provide.
|
522
|
+
|
523
|
+
See: https://pyyaml.org/wiki/PyYAMLDocumentation
|
524
|
+
"""
|
525
|
+
def __init__(self,
|
526
|
+
env_var_pattern: str = None,
|
527
|
+
add_constructor: bool = True):
|
528
|
+
self.env_var_pattern = env_var_pattern
|
529
|
+
if self.env_var_pattern is None:
|
530
|
+
# Default pattern is: ${VAR:default}
|
531
|
+
self.env_var_pattern = r'^\$\{(.*)\}$'
|
532
|
+
self.path_matcher = re.compile(self.env_var_pattern)
|
533
|
+
|
534
|
+
if add_constructor:
|
535
|
+
self.add_constructor()
|
536
|
+
|
537
|
+
def _path_constructor(self, loader, node):
|
538
|
+
""" Extract the matched value, expand env variable,
|
539
|
+
and replace the match
|
540
|
+
|
541
|
+
TODO: Would need to update this (specifically the parsing) if any
|
542
|
+
pattern other than our default (or a highly compatible variation)
|
543
|
+
is provided.
|
544
|
+
"""
|
545
|
+
# Try to match the correct env variable pattern in this node's value
|
546
|
+
# If the value does not match the pattern, return None (which means
|
547
|
+
# this node will not be parsed for ENV variables and instead just
|
548
|
+
# returned as-is).
|
549
|
+
env_var_name = re.match(self.env_var_pattern, node.value)
|
550
|
+
try:
|
551
|
+
env_var_name = env_var_name.group(1)
|
552
|
+
except AttributeError:
|
553
|
+
return None
|
554
|
+
|
555
|
+
# If we get down here, then the 'node.value' matches our specified
|
556
|
+
# pattern, so try to parse. env_var_name is the value inside ${...}.
|
557
|
+
# Split on `:`, which is our delimiter for default values.
|
558
|
+
env_var_name_split = env_var_name.split(':')
|
559
|
+
|
560
|
+
# Attempt to retrieve the environment variable...from the environment
|
561
|
+
env_var = os.environ.get(env_var_name_split[0], None)
|
562
|
+
|
563
|
+
if env_var is None: # Nothing found in environment
|
564
|
+
# If a default was provided (e.g. VAR:default), return that.
|
565
|
+
# We join anything after first element because the default
|
566
|
+
# value might be a URL or something with a colon in it
|
567
|
+
# which would have 'split' above
|
568
|
+
if len(env_var_name_split) > 1:
|
569
|
+
return ":".join(env_var_name_split[1:])
|
570
|
+
return 'unset' # Return 'unset' if not in environ nor default
|
571
|
+
return env_var
|
572
|
+
|
573
|
+
def add_constructor(self):
|
574
|
+
""" Initialize PyYaml with ability to resolve/load environment
|
575
|
+
variables defined in a yaml template when they exist in
|
576
|
+
the environment.
|
577
|
+
|
578
|
+
Add to SafeLoader in addition to standard Loader.
|
579
|
+
"""
|
580
|
+
# Add the `!env_var` tag to any scalar (value) that matches the
|
581
|
+
# pattern self.path_matcher. This allows the template to be much more
|
582
|
+
# intuitive vs needing to add !env_var to the beginning of each value
|
583
|
+
yaml.add_implicit_resolver('!env_var', self.path_matcher)
|
584
|
+
yaml.add_implicit_resolver('!env_var',
|
585
|
+
self.path_matcher,
|
586
|
+
Loader=SafeLoader)
|
587
|
+
|
588
|
+
# Add constructor for the tag `!env_var`, which is a function that
|
589
|
+
# converts a node of a YAML representation graph to a native Python
|
590
|
+
# object.
|
591
|
+
yaml.add_constructor('!env_var', self._path_constructor)
|
592
|
+
yaml.add_constructor('!env_var',
|
593
|
+
self._path_constructor,
|
594
|
+
Loader=SafeLoader)
|
595
|
+
|
596
|
+
|
597
|
+
def parse_config_file(sermos_yaml: str):
|
598
|
+
""" Parse the `sermos.yaml` file when it's been loaded.
|
599
|
+
|
600
|
+
Arguments:
|
601
|
+
sermos_yaml (required): String of loaded sermos.yaml file.
|
602
|
+
"""
|
603
|
+
YamlPatternConstructor() # Add our env variable parser
|
604
|
+
try:
|
605
|
+
sermos_yaml_schema = SermosYamlSchema()
|
606
|
+
# First suss out yaml issues
|
607
|
+
sermos_config = yaml.safe_load(sermos_yaml)
|
608
|
+
# Then schema issues
|
609
|
+
sermos_config = sermos_yaml_schema.load(sermos_config)
|
610
|
+
except ValidationError as e:
|
611
|
+
msg = "Invalid Sermos configuration due to {}"\
|
612
|
+
.format(e.messages)
|
613
|
+
logger.error(msg)
|
614
|
+
raise InvalidSermosConfig(msg)
|
615
|
+
except InvalidImageConfig as e:
|
616
|
+
msg = "Invalid imageConfig configuration due to {}"\
|
617
|
+
.format(e)
|
618
|
+
logger.error(msg)
|
619
|
+
raise InvalidImageConfig(msg)
|
620
|
+
except Exception as e:
|
621
|
+
msg = "Invalid Sermos configuration, likely due to invalid "\
|
622
|
+
"YAML formatting ..."
|
623
|
+
logger.exception("{} {}".format(msg, e))
|
624
|
+
raise InvalidSermosConfig(msg)
|
625
|
+
return sermos_config
|
626
|
+
|
627
|
+
|
628
|
+
def _get_pkg_name(pkg_name: str) -> str:
|
629
|
+
""" Retrieve the normalized package name.
|
630
|
+
"""
|
631
|
+
if pkg_name is None:
|
632
|
+
pkg_name = SERMOS_CLIENT_PKG_NAME # From environment
|
633
|
+
if pkg_name is None:
|
634
|
+
return None
|
635
|
+
return normalized_pkg_name(pkg_name)
|
636
|
+
|
637
|
+
|
638
|
+
def load_sermos_config(pkg_name: str = None,
|
639
|
+
sermos_yaml_filename: str = None,
|
640
|
+
as_dict: bool = True):
|
641
|
+
""" Load and parse the `sermos.yaml` file. Issue usable exceptions for
|
642
|
+
known error modes so bootstrapping can handle appropriately.
|
643
|
+
|
644
|
+
Arguments:
|
645
|
+
pkg_name (required): Directory name for your Python
|
646
|
+
package. e.g. my_package_name . If none provided, will check
|
647
|
+
environment for `SERMOS_CLIENT_PKG_NAME`. If not found,
|
648
|
+
will exit.
|
649
|
+
sermos_yaml_filename (optional): Relative path to find your
|
650
|
+
`sermos.yaml` configuration file. Defaults to `sermos.yaml`
|
651
|
+
which should be found inside your `pkg_name`
|
652
|
+
as_dict (optional): If true (default), return the loaded sermos
|
653
|
+
configuration as a dictionary. If false, return the loaded
|
654
|
+
string value of the yaml file.
|
655
|
+
"""
|
656
|
+
if sermos_yaml_filename is None:
|
657
|
+
sermos_yaml_filename = SERMOS_YAML_PATH
|
658
|
+
|
659
|
+
logger.info(f"Loading `sermos.yaml` from package `{pkg_name}` "
|
660
|
+
f"and file location `{sermos_yaml_filename}` ...")
|
661
|
+
sermos_config = None
|
662
|
+
|
663
|
+
pkg_name = _get_pkg_name(pkg_name)
|
664
|
+
if pkg_name is None: # Nothing to retrieve at this point
|
665
|
+
logger.warning("Unable to retrieve sermos.yaml configuration ...")
|
666
|
+
return sermos_config
|
667
|
+
|
668
|
+
try:
|
669
|
+
sermos_config_path = pkg_resources.resource_filename(
|
670
|
+
pkg_name, sermos_yaml_filename)
|
671
|
+
except Exception as e:
|
672
|
+
msg = "Either pkg_name ({}) or sermos_yaml_filename ({}) is "\
|
673
|
+
"invalid ...".format(pkg_name, sermos_yaml_filename)
|
674
|
+
logger.error("{} ... {}".format(msg, e))
|
675
|
+
raise InvalidPackagePath(e)
|
676
|
+
|
677
|
+
try:
|
678
|
+
with open(sermos_config_path, 'r') as f:
|
679
|
+
sermos_yaml = f.read()
|
680
|
+
sermos_config = parse_config_file(sermos_yaml)
|
681
|
+
except InvalidSermosConfig as e:
|
682
|
+
raise
|
683
|
+
except InvalidImageConfig as e:
|
684
|
+
raise
|
685
|
+
except FileNotFoundError as e:
|
686
|
+
msg = "Sermos config file could not be found at path {} ...".format(
|
687
|
+
sermos_config_path)
|
688
|
+
raise MissingSermosConfig(msg)
|
689
|
+
except Exception as e:
|
690
|
+
raise e
|
691
|
+
if as_dict:
|
692
|
+
return sermos_config
|
693
|
+
return yaml.safe_dump(sermos_config)
|
694
|
+
|
695
|
+
|
696
|
+
def load_client_config_and_version(pkg_name: str = None,
|
697
|
+
sermos_yaml_filename: str = None):
|
698
|
+
""" Load and parse the `sermos.yaml` file and a client package's version.
|
699
|
+
|
700
|
+
Arguments:
|
701
|
+
pkg_name (required): Directory name for your Python
|
702
|
+
package. e.g. my_package_name . If none provided, will check
|
703
|
+
environment for `SERMOS_CLIENT_PKG_NAME`. If not found,
|
704
|
+
will exit.
|
705
|
+
sermos_yaml_filename (optional): Relative path to find your
|
706
|
+
`sermos.yaml` configuration file. Defaults to `sermos.yaml`
|
707
|
+
which should be found inside your `pkg_name`
|
708
|
+
as_dict (optional): If true (default), return the loaded sermos
|
709
|
+
configuration as a dictionary. If false, return the loaded
|
710
|
+
string value of the yaml file.
|
711
|
+
|
712
|
+
For this to work properly, the provided package must be installed in the
|
713
|
+
same environment as this Sermos package and it must have a `__version__`
|
714
|
+
variable inside its `__init__.py` file, e.g. `__version__ = '0.0.0'`
|
715
|
+
"""
|
716
|
+
sermos_config = None
|
717
|
+
client_version = None
|
718
|
+
|
719
|
+
pkg_name = _get_pkg_name(pkg_name)
|
720
|
+
|
721
|
+
try:
|
722
|
+
loader = SermosModuleLoader()
|
723
|
+
pkg = loader.get_module(pkg_name + '.__init__')
|
724
|
+
client_version = getattr(pkg, '__version__', '0.0.0')
|
725
|
+
sermos_config = load_sermos_config(pkg_name, sermos_yaml_filename)
|
726
|
+
except MissingSermosConfig as e:
|
727
|
+
logger.error(e)
|
728
|
+
except InvalidSermosConfig as e:
|
729
|
+
logger.error(e)
|
730
|
+
except InvalidPackagePath as e:
|
731
|
+
logger.error(e)
|
732
|
+
except Exception as e:
|
733
|
+
logger.error("Unable to load client's pkg __version__ or "
|
734
|
+
"{} config file for package: {} ... {}".format(
|
735
|
+
sermos_yaml_filename, pkg_name, e))
|
736
|
+
|
737
|
+
return sermos_config, client_version
|