capo-sagemaker-runtime-http2 0.5.0__tar.gz → 0.7.0__tar.gz
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.
- {capo_sagemaker_runtime_http2-0.5.0/src/capo_sagemaker_runtime_http2.egg-info → capo_sagemaker_runtime_http2-0.7.0}/PKG-INFO +2 -2
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/pyproject.toml +2 -2
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_auth/_sigv4.py +9 -9
- capo_sagemaker_runtime_http2-0.7.0/src/capo_sagemaker_runtime_http2/_aws.py +28 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_operations/amazon_sage_maker_runtime_http2/invoke_endpoint_with_bidirectional_stream.py +13 -10
- capo_sagemaker_runtime_http2-0.7.0/src/capo_sagemaker_runtime_http2/_services/_pipeline.py +522 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_services/async_sage_maker_runtime_http2.py +45 -7
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_services/sage_maker_runtime_http2.py +45 -7
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/errors/_base.py +8 -3
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/errors/input_validation_error.py +11 -3
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/errors/internal_server_error.py +13 -3
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/errors/internal_stream_failure.py +11 -3
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/errors/model_error.py +13 -3
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/errors/model_stream_error.py +13 -3
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/errors/service_unavailable_error.py +11 -3
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/types/invoke_endpoint_with_bidirectional_stream_input.py +10 -1
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/types/invoke_endpoint_with_bidirectional_stream_output.py +10 -1
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/types/request_payload_part.py +8 -1
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/types/request_stream_event.py +6 -3
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/types/response_payload_part.py +8 -1
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/types/response_stream_event.py +8 -5
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0/src/capo_sagemaker_runtime_http2.egg-info}/PKG-INFO +2 -2
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2.egg-info/SOURCES.txt +1 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2.egg-info/requires.txt +1 -1
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/tests/test_sigv4.py +21 -6
- capo_sagemaker_runtime_http2-0.5.0/src/capo_sagemaker_runtime_http2/_services/_pipeline.py +0 -266
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/LICENSE +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/README.md +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/setup.cfg +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/__init__.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_async.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_auth/_identity.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_auth/_providers.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_auth/_signers.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_auth/_zapros_handler.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_body.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_iter.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_pagination.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_protocol/__init__.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_protocol/errors.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_protocol/eventstream.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_protocol/serialize.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_protocol/xml.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_rule_engine/__init__.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_rule_engine/_aws_partition.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_rule_engine/_endpoint_rule_set.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_rule_engine/_endpoint_runtime.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/_services/_aws_config.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/errors/__init__.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/py.typed +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/types/_prelude/blob.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/types/_prelude/timestamp.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2/types/sensitive_blob.py +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2.egg-info/dependency_links.txt +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/src/capo_sagemaker_runtime_http2.egg-info/top_level.txt +0 -0
- {capo_sagemaker_runtime_http2-0.5.0 → capo_sagemaker_runtime_http2-0.7.0}/tests/test_body.py +0 -0
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: capo-sagemaker-runtime-http2
|
|
3
|
-
Version: 0.
|
|
3
|
+
Version: 0.7.0
|
|
4
4
|
Summary: Python SDK for AWS Sagemaker Runtime Http2.
|
|
5
5
|
Classifier: Development Status :: 4 - Beta
|
|
6
6
|
Classifier: Intended Audience :: Developers
|
|
@@ -17,7 +17,7 @@ Classifier: Typing :: Typed
|
|
|
17
17
|
Requires-Python: >=3.10
|
|
18
18
|
Description-Content-Type: text/markdown
|
|
19
19
|
License-File: LICENSE
|
|
20
|
-
Requires-Dist: zapros>=0.
|
|
20
|
+
Requires-Dist: zapros>=0.21.0
|
|
21
21
|
Provides-Extra: anyio
|
|
22
22
|
Requires-Dist: anyio>=3.0; extra == "anyio"
|
|
23
23
|
Provides-Extra: sso
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
[project]
|
|
2
2
|
name = "capo-sagemaker-runtime-http2"
|
|
3
|
-
version = "0.
|
|
3
|
+
version = "0.7.0"
|
|
4
4
|
description = "Python SDK for AWS Sagemaker Runtime Http2."
|
|
5
5
|
readme = "README.md"
|
|
6
6
|
requires-python = ">=3.10"
|
|
@@ -19,7 +19,7 @@ classifiers = [
|
|
|
19
19
|
"Typing :: Typed",
|
|
20
20
|
]
|
|
21
21
|
dependencies = [
|
|
22
|
-
"zapros>=0.
|
|
22
|
+
"zapros>=0.21.0",
|
|
23
23
|
]
|
|
24
24
|
|
|
25
25
|
[project.optional-dependencies]
|
|
@@ -17,7 +17,7 @@ import hmac
|
|
|
17
17
|
import re
|
|
18
18
|
from collections.abc import AsyncIterator, Iterator
|
|
19
19
|
from typing import Any, Literal, TypedDict, cast
|
|
20
|
-
from urllib.parse import quote
|
|
20
|
+
from urllib.parse import quote
|
|
21
21
|
|
|
22
22
|
import zapros
|
|
23
23
|
from pywhatwgurl import URLSearchParams
|
|
@@ -134,11 +134,11 @@ def _remove_dot_segments(path: str) -> str:
|
|
|
134
134
|
def _canonical_path(path: str, *, double_encode: bool, normalize: bool) -> str:
|
|
135
135
|
"""Build CanonicalURI.
|
|
136
136
|
|
|
137
|
-
|
|
138
|
-
normalize dot segments and URI-encode each
|
|
139
|
-
rules-engine ``disableNormalizePath`` /
|
|
140
|
-
turn those steps off — S3-family
|
|
141
|
-
signed exactly as provided.
|
|
137
|
+
``path`` is the path as sent on the wire, so it is already URI-encoded
|
|
138
|
+
once. Standard services normalize dot segments and URI-encode each
|
|
139
|
+
segment a second time. The rules-engine ``disableNormalizePath`` /
|
|
140
|
+
``disableDoubleEncoding`` flags turn those steps off — S3-family
|
|
141
|
+
services set both, so the path is signed exactly as provided.
|
|
142
142
|
"""
|
|
143
143
|
if not path:
|
|
144
144
|
return "/"
|
|
@@ -148,9 +148,9 @@ def _canonical_path(path: str, *, double_encode: bool, normalize: bool) -> str:
|
|
|
148
148
|
path = _remove_dot_segments(path)
|
|
149
149
|
if not double_encode:
|
|
150
150
|
return path
|
|
151
|
-
|
|
152
|
-
|
|
153
|
-
return quote(
|
|
151
|
+
# Never decode first: that would turn a "%2F" inside a segment (an ARN in
|
|
152
|
+
# a path label) into a separator and leave it out of the second encoding.
|
|
153
|
+
return quote(path, safe="/~")
|
|
154
154
|
|
|
155
155
|
|
|
156
156
|
def _canonical_query(query: str) -> str:
|
|
@@ -0,0 +1,28 @@
|
|
|
1
|
+
"""Error codes and HTTP statuses that are retried whether or not the model marks them @retryable."""
|
|
2
|
+
|
|
3
|
+
# Transient: retried with the normal backoff.
|
|
4
|
+
KNOWN_RETRYABLE_STATUSES = frozenset({500, 502, 503, 504})
|
|
5
|
+
KNOWN_RETRYABLE_CODES = frozenset(
|
|
6
|
+
{"PriorRequestNotComplete", "RequestTimeout", "RequestTimeoutException"}
|
|
7
|
+
)
|
|
8
|
+
|
|
9
|
+
# Throttling: retried with the longer throttling backoff.
|
|
10
|
+
KNOWN_THROTTLING_STATUSES = frozenset({429})
|
|
11
|
+
KNOWN_THROTTLING_CODES = frozenset(
|
|
12
|
+
{
|
|
13
|
+
"BandwidthLimitExceeded",
|
|
14
|
+
"EC2ThrottledException",
|
|
15
|
+
"LimitExceededException",
|
|
16
|
+
"PriorRequestNotComplete",
|
|
17
|
+
"ProvisionedThroughputExceededException",
|
|
18
|
+
"RequestLimitExceeded",
|
|
19
|
+
"RequestThrottled",
|
|
20
|
+
"RequestThrottledException",
|
|
21
|
+
"SlowDown",
|
|
22
|
+
"ThrottledException",
|
|
23
|
+
"Throttling",
|
|
24
|
+
"ThrottlingException",
|
|
25
|
+
"TooManyRequestsException",
|
|
26
|
+
"TransactionInProgressException",
|
|
27
|
+
}
|
|
28
|
+
)
|
|
@@ -2,6 +2,7 @@
|
|
|
2
2
|
|
|
3
3
|
from __future__ import annotations
|
|
4
4
|
|
|
5
|
+
import functools
|
|
5
6
|
import json
|
|
6
7
|
from typing import Any, cast
|
|
7
8
|
from urllib.parse import quote
|
|
@@ -47,27 +48,27 @@ def handle_error(response: zapros.Response) -> Never:
|
|
|
47
48
|
match code:
|
|
48
49
|
case "InputValidationError":
|
|
49
50
|
raise capo_sagemaker_runtime_http2.errors.input_validation_error.InputValidationError.from_json(
|
|
50
|
-
data, message
|
|
51
|
+
data, message, response=response
|
|
51
52
|
)
|
|
52
53
|
case "InternalServerError":
|
|
53
54
|
raise capo_sagemaker_runtime_http2.errors.internal_server_error.InternalServerError.from_json(
|
|
54
|
-
data, message
|
|
55
|
+
data, message, response=response
|
|
55
56
|
)
|
|
56
57
|
case "InternalStreamFailure":
|
|
57
58
|
raise capo_sagemaker_runtime_http2.errors.internal_stream_failure.InternalStreamFailure.from_json(
|
|
58
|
-
data, message
|
|
59
|
+
data, message, response=response
|
|
59
60
|
)
|
|
60
61
|
case "ModelError":
|
|
61
62
|
raise capo_sagemaker_runtime_http2.errors.model_error.ModelError.from_json(
|
|
62
|
-
data, message
|
|
63
|
+
data, message, response=response
|
|
63
64
|
)
|
|
64
65
|
case "ModelStreamError":
|
|
65
66
|
raise capo_sagemaker_runtime_http2.errors.model_stream_error.ModelStreamError.from_json(
|
|
66
|
-
data, message
|
|
67
|
+
data, message, response=response
|
|
67
68
|
)
|
|
68
69
|
case "ServiceUnavailableError":
|
|
69
70
|
raise capo_sagemaker_runtime_http2.errors.service_unavailable_error.ServiceUnavailableError.from_json(
|
|
70
|
-
data, message
|
|
71
|
+
data, message, response=response
|
|
71
72
|
)
|
|
72
73
|
case _:
|
|
73
74
|
raise UnknownServiceError(code=code, message=message, response=response)
|
|
@@ -77,8 +78,9 @@ def handle_response(
|
|
|
77
78
|
response: zapros.Response,
|
|
78
79
|
) -> capo_sagemaker_runtime_http2.types.invoke_endpoint_with_bidirectional_stream_output.InvokeEndpointWithBidirectionalStreamOutput:
|
|
79
80
|
_message_decoder = MessageDecoder()
|
|
80
|
-
_union_deser = (
|
|
81
|
-
capo_sagemaker_runtime_http2.types.response_stream_event.deserialize_event_json
|
|
81
|
+
_union_deser = functools.partial(
|
|
82
|
+
capo_sagemaker_runtime_http2.types.response_stream_event.deserialize_event_json,
|
|
83
|
+
response=response,
|
|
82
84
|
)
|
|
83
85
|
_iter = cast(Any, response.iter_bytes())
|
|
84
86
|
out: capo_sagemaker_runtime_http2.types.invoke_endpoint_with_bidirectional_stream_output.InvokeEndpointWithBidirectionalStreamOutput = {
|
|
@@ -95,8 +97,9 @@ async def async_handle_response(
|
|
|
95
97
|
response: zapros.Response,
|
|
96
98
|
) -> capo_sagemaker_runtime_http2.types.invoke_endpoint_with_bidirectional_stream_output.InvokeEndpointWithBidirectionalStreamOutput:
|
|
97
99
|
_message_decoder = MessageDecoder()
|
|
98
|
-
_union_deser = (
|
|
99
|
-
capo_sagemaker_runtime_http2.types.response_stream_event.deserialize_event_json
|
|
100
|
+
_union_deser = functools.partial(
|
|
101
|
+
capo_sagemaker_runtime_http2.types.response_stream_event.deserialize_event_json,
|
|
102
|
+
response=response,
|
|
100
103
|
)
|
|
101
104
|
_iter = cast(Any, response.async_iter_bytes())
|
|
102
105
|
out: capo_sagemaker_runtime_http2.types.invoke_endpoint_with_bidirectional_stream_output.InvokeEndpointWithBidirectionalStreamOutput = {
|
|
@@ -0,0 +1,522 @@
|
|
|
1
|
+
from __future__ import annotations
|
|
2
|
+
|
|
3
|
+
import logging
|
|
4
|
+
import random
|
|
5
|
+
import threading
|
|
6
|
+
import time
|
|
7
|
+
from collections.abc import AsyncIterator, Iterator, Sequence
|
|
8
|
+
from dataclasses import dataclass
|
|
9
|
+
from typing import (
|
|
10
|
+
TYPE_CHECKING,
|
|
11
|
+
Any,
|
|
12
|
+
Awaitable,
|
|
13
|
+
Callable,
|
|
14
|
+
Generic,
|
|
15
|
+
TypeVar,
|
|
16
|
+
cast,
|
|
17
|
+
)
|
|
18
|
+
|
|
19
|
+
from typing_extensions import Never
|
|
20
|
+
from zapros import (
|
|
21
|
+
AsyncClient,
|
|
22
|
+
Client,
|
|
23
|
+
ConnectionError,
|
|
24
|
+
ConnectTimeoutError,
|
|
25
|
+
PoolTimeoutError,
|
|
26
|
+
Response,
|
|
27
|
+
SSLError,
|
|
28
|
+
TimeoutError,
|
|
29
|
+
)
|
|
30
|
+
|
|
31
|
+
from capo_sagemaker_runtime_http2._async import anysleep
|
|
32
|
+
from capo_sagemaker_runtime_http2._aws import (
|
|
33
|
+
KNOWN_RETRYABLE_CODES,
|
|
34
|
+
KNOWN_RETRYABLE_STATUSES,
|
|
35
|
+
KNOWN_THROTTLING_CODES,
|
|
36
|
+
KNOWN_THROTTLING_STATUSES,
|
|
37
|
+
)
|
|
38
|
+
from capo_sagemaker_runtime_http2._body import Body
|
|
39
|
+
from capo_sagemaker_runtime_http2._iter import StaticAnyIterator
|
|
40
|
+
from capo_sagemaker_runtime_http2.errors import ServiceError, UnknownServiceError
|
|
41
|
+
|
|
42
|
+
if TYPE_CHECKING:
|
|
43
|
+
from capo_sagemaker_runtime_http2._auth._identity import Credentials
|
|
44
|
+
from capo_sagemaker_runtime_http2._auth._providers import IdentityProvider
|
|
45
|
+
|
|
46
|
+
logger = logging.getLogger("capo_sagemaker_runtime_http2.operation")
|
|
47
|
+
|
|
48
|
+
TInput = TypeVar("TInput")
|
|
49
|
+
TOutput = TypeVar("TOutput")
|
|
50
|
+
|
|
51
|
+
|
|
52
|
+
class RetryQuota:
|
|
53
|
+
"""Retry tokens shared by all operations of one client: retries spend them, successes refund them."""
|
|
54
|
+
|
|
55
|
+
CAPACITY: int = 500
|
|
56
|
+
TRANSIENT_COST: int = 14
|
|
57
|
+
THROTTLING_COST: int = 5
|
|
58
|
+
NO_RETRY_INCREMENT: int = 1
|
|
59
|
+
|
|
60
|
+
def __init__(self) -> None:
|
|
61
|
+
self._available: int = self.CAPACITY
|
|
62
|
+
|
|
63
|
+
def acquire(self, is_throttling: bool) -> int | None:
|
|
64
|
+
cost = self.THROTTLING_COST if is_throttling else self.TRANSIENT_COST
|
|
65
|
+
if cost > self._available:
|
|
66
|
+
return None
|
|
67
|
+
self._available -= cost
|
|
68
|
+
return cost
|
|
69
|
+
|
|
70
|
+
def release(self, last_retry_cost: int | None) -> None:
|
|
71
|
+
refund = self.NO_RETRY_INCREMENT if last_retry_cost is None else last_retry_cost
|
|
72
|
+
self._available = min(self.CAPACITY, self._available + refund)
|
|
73
|
+
|
|
74
|
+
|
|
75
|
+
class ThreadSafeRetryQuota(RetryQuota):
|
|
76
|
+
"""A ``RetryQuota`` that sync clients can share across threads."""
|
|
77
|
+
|
|
78
|
+
def __init__(self) -> None:
|
|
79
|
+
super().__init__()
|
|
80
|
+
self._lock: threading.Lock = threading.Lock()
|
|
81
|
+
|
|
82
|
+
def acquire(self, is_throttling: bool) -> int | None:
|
|
83
|
+
with self._lock:
|
|
84
|
+
return super().acquire(is_throttling)
|
|
85
|
+
|
|
86
|
+
def release(self, last_retry_cost: int | None) -> None:
|
|
87
|
+
with self._lock:
|
|
88
|
+
super().release(last_retry_cost)
|
|
89
|
+
|
|
90
|
+
|
|
91
|
+
@dataclass
|
|
92
|
+
class OperationOptions:
|
|
93
|
+
client: Client
|
|
94
|
+
retry_quota: RetryQuota
|
|
95
|
+
use_dual_stack: bool | None = None
|
|
96
|
+
use_fips: bool | None = None
|
|
97
|
+
endpoint: str | None = None
|
|
98
|
+
region: str | None = None
|
|
99
|
+
retry_max_attempts: int | None = None
|
|
100
|
+
credentials_provider: IdentityProvider[Credentials] | None = None
|
|
101
|
+
anonymous: bool | None = None
|
|
102
|
+
|
|
103
|
+
|
|
104
|
+
@dataclass
|
|
105
|
+
class AsyncOperationOptions:
|
|
106
|
+
client: AsyncClient
|
|
107
|
+
retry_quota: RetryQuota
|
|
108
|
+
use_dual_stack: bool | None = None
|
|
109
|
+
use_fips: bool | None = None
|
|
110
|
+
endpoint: str | None = None
|
|
111
|
+
region: str | None = None
|
|
112
|
+
retry_max_attempts: int | None = None
|
|
113
|
+
credentials_provider: IdentityProvider[Credentials] | None = None
|
|
114
|
+
anonymous: bool | None = None
|
|
115
|
+
|
|
116
|
+
|
|
117
|
+
@dataclass
|
|
118
|
+
class OperationRequest(Generic[TInput]):
|
|
119
|
+
input: TInput
|
|
120
|
+
options: OperationOptions
|
|
121
|
+
|
|
122
|
+
|
|
123
|
+
@dataclass
|
|
124
|
+
class AsyncOperationRequest(Generic[TInput]):
|
|
125
|
+
input: TInput
|
|
126
|
+
options: AsyncOperationOptions
|
|
127
|
+
|
|
128
|
+
|
|
129
|
+
@dataclass
|
|
130
|
+
class OperationResponse(Generic[TOutput]):
|
|
131
|
+
output: TOutput
|
|
132
|
+
response: Response
|
|
133
|
+
|
|
134
|
+
|
|
135
|
+
@dataclass
|
|
136
|
+
class AsyncOperationResponse(Generic[TOutput]):
|
|
137
|
+
output: TOutput
|
|
138
|
+
response: Response
|
|
139
|
+
|
|
140
|
+
|
|
141
|
+
NextFn = Callable[
|
|
142
|
+
[OperationRequest[TInput]],
|
|
143
|
+
OperationResponse[TOutput],
|
|
144
|
+
]
|
|
145
|
+
|
|
146
|
+
AsyncNextFn = Callable[
|
|
147
|
+
[AsyncOperationRequest[TInput]],
|
|
148
|
+
Awaitable[AsyncOperationResponse[TOutput]],
|
|
149
|
+
]
|
|
150
|
+
|
|
151
|
+
Interceptor = Callable[
|
|
152
|
+
[OperationRequest[TInput], NextFn[TInput, TOutput]],
|
|
153
|
+
OperationResponse[TOutput],
|
|
154
|
+
]
|
|
155
|
+
|
|
156
|
+
AsyncInterceptor = Callable[
|
|
157
|
+
[AsyncOperationRequest[TInput], AsyncNextFn[TInput, TOutput]],
|
|
158
|
+
Awaitable[AsyncOperationResponse[TOutput]],
|
|
159
|
+
]
|
|
160
|
+
|
|
161
|
+
|
|
162
|
+
@dataclass
|
|
163
|
+
class OperationLogging:
|
|
164
|
+
"""What the logging interceptor needs to know about one operation call."""
|
|
165
|
+
|
|
166
|
+
operation: str
|
|
167
|
+
# Return a copy that is safe to log; None when the shape has nothing to hide.
|
|
168
|
+
redact_input: Callable[[Any], dict[str, Any]] | None = None
|
|
169
|
+
redact_output: Callable[[Any], dict[str, Any]] | None = None
|
|
170
|
+
|
|
171
|
+
|
|
172
|
+
def execute_pipeline(
|
|
173
|
+
request: OperationRequest[TInput],
|
|
174
|
+
handler: NextFn[TInput, TOutput],
|
|
175
|
+
interceptors: Sequence[Interceptor[TInput, TOutput]],
|
|
176
|
+
) -> OperationResponse[TOutput]:
|
|
177
|
+
def make_chain(index: int) -> NextFn[TInput, TOutput]:
|
|
178
|
+
if index < len(interceptors):
|
|
179
|
+
interceptor = interceptors[index]
|
|
180
|
+
|
|
181
|
+
def next_fn(req: OperationRequest[TInput]) -> OperationResponse[TOutput]:
|
|
182
|
+
return interceptor(req, make_chain(index + 1))
|
|
183
|
+
|
|
184
|
+
return next_fn
|
|
185
|
+
return handler
|
|
186
|
+
|
|
187
|
+
return make_chain(0)(request)
|
|
188
|
+
|
|
189
|
+
|
|
190
|
+
async def aexecute_pipeline(
|
|
191
|
+
request: AsyncOperationRequest[TInput],
|
|
192
|
+
handler: AsyncNextFn[TInput, TOutput],
|
|
193
|
+
interceptors: Sequence[AsyncInterceptor[TInput, TOutput]],
|
|
194
|
+
) -> AsyncOperationResponse[TOutput]:
|
|
195
|
+
def make_chain(index: int) -> AsyncNextFn[TInput, TOutput]:
|
|
196
|
+
if index < len(interceptors):
|
|
197
|
+
interceptor = interceptors[index]
|
|
198
|
+
|
|
199
|
+
async def next_fn(
|
|
200
|
+
req: AsyncOperationRequest[TInput],
|
|
201
|
+
) -> AsyncOperationResponse[TOutput]:
|
|
202
|
+
return await interceptor(req, make_chain(index + 1))
|
|
203
|
+
|
|
204
|
+
return next_fn
|
|
205
|
+
return handler
|
|
206
|
+
|
|
207
|
+
return await make_chain(0)(request)
|
|
208
|
+
|
|
209
|
+
|
|
210
|
+
def raise_error(response: Response, handle_error: Callable[[Response], Never]) -> Never:
|
|
211
|
+
try:
|
|
212
|
+
handle_error(response)
|
|
213
|
+
except ServiceError as exc:
|
|
214
|
+
_apply_retry_rules(exc, response)
|
|
215
|
+
raise
|
|
216
|
+
except Exception as exc:
|
|
217
|
+
# The error body could not be decoded (an HTML page from a proxy, a
|
|
218
|
+
# truncated read): the status code is all there is to go on.
|
|
219
|
+
error = UnknownServiceError(code=None, message=None, response=response)
|
|
220
|
+
_apply_retry_rules(error, response)
|
|
221
|
+
raise error from exc
|
|
222
|
+
|
|
223
|
+
|
|
224
|
+
def _apply_retry_rules(exc: ServiceError, response: Response) -> None:
|
|
225
|
+
# A retry can be safe even when the error is not marked @retryable: a known
|
|
226
|
+
# throttling or transient status or error code, or a Retry-After header.
|
|
227
|
+
if (
|
|
228
|
+
response.status in KNOWN_THROTTLING_STATUSES
|
|
229
|
+
or exc.code in KNOWN_THROTTLING_CODES
|
|
230
|
+
):
|
|
231
|
+
exc.is_throttling_error = True
|
|
232
|
+
exc.is_retryable = True
|
|
233
|
+
if (
|
|
234
|
+
response.status in KNOWN_RETRYABLE_STATUSES
|
|
235
|
+
or exc.code in KNOWN_RETRYABLE_CODES
|
|
236
|
+
or "retry-after" in response.headers
|
|
237
|
+
):
|
|
238
|
+
exc.is_retryable = True
|
|
239
|
+
|
|
240
|
+
|
|
241
|
+
def _is_retryable(exc: Exception) -> bool:
|
|
242
|
+
if isinstance(exc, ServiceError):
|
|
243
|
+
return exc.is_retryable
|
|
244
|
+
if isinstance(exc, SSLError):
|
|
245
|
+
return False
|
|
246
|
+
if isinstance(exc, (ConnectionError, TimeoutError)):
|
|
247
|
+
return True
|
|
248
|
+
return False
|
|
249
|
+
|
|
250
|
+
|
|
251
|
+
def _is_pre_transmission(exc: Exception) -> bool:
|
|
252
|
+
# True when the request body was certainly not written before the error,
|
|
253
|
+
# so the same stream can be reused as-is. A bare ConnectionError is
|
|
254
|
+
# ambiguous ("cannot be established or is lost"), so it does not qualify.
|
|
255
|
+
return isinstance(exc, (ConnectTimeoutError, PoolTimeoutError))
|
|
256
|
+
|
|
257
|
+
|
|
258
|
+
def _prepare_retry(inp: object) -> bool:
|
|
259
|
+
if not isinstance(inp, dict):
|
|
260
|
+
return True
|
|
261
|
+
for value in inp.values():
|
|
262
|
+
if isinstance(value, Body):
|
|
263
|
+
value = cast(Body[Iterator[bytes]], value)
|
|
264
|
+
if value.rebuild() is None:
|
|
265
|
+
return False
|
|
266
|
+
elif isinstance(value, StaticAnyIterator):
|
|
267
|
+
continue
|
|
268
|
+
elif isinstance(value, (Iterator, AsyncIterator)):
|
|
269
|
+
return False
|
|
270
|
+
return True
|
|
271
|
+
|
|
272
|
+
|
|
273
|
+
async def _aprepare_retry(inp: object) -> bool:
|
|
274
|
+
if not isinstance(inp, dict):
|
|
275
|
+
return True
|
|
276
|
+
for value in inp.values():
|
|
277
|
+
if isinstance(value, Body):
|
|
278
|
+
value = cast(Body[AsyncIterator[bytes]], value)
|
|
279
|
+
if await value.arebuild() is None:
|
|
280
|
+
return False
|
|
281
|
+
elif isinstance(value, StaticAnyIterator):
|
|
282
|
+
continue
|
|
283
|
+
elif isinstance(value, (Iterator, AsyncIterator)):
|
|
284
|
+
return False
|
|
285
|
+
return True
|
|
286
|
+
|
|
287
|
+
|
|
288
|
+
def _retry_delay(attempt: int, is_throttling: bool) -> float:
|
|
289
|
+
base = 1.0 if is_throttling else 0.5
|
|
290
|
+
delay = base * (2.0 ** min(attempt - 1, 10))
|
|
291
|
+
delay = min(20.0, delay)
|
|
292
|
+
return random.uniform(0.0, delay)
|
|
293
|
+
|
|
294
|
+
|
|
295
|
+
def retry() -> Interceptor[TInput, TOutput]:
|
|
296
|
+
def interceptor(
|
|
297
|
+
request: OperationRequest[TInput], next: NextFn[TInput, TOutput]
|
|
298
|
+
) -> OperationResponse[TOutput]:
|
|
299
|
+
max_attempts = request.options.retry_max_attempts or 3
|
|
300
|
+
quota = request.options.retry_quota
|
|
301
|
+
last_retry_cost: int | None = None
|
|
302
|
+
for attempt in range(1, max_attempts + 1):
|
|
303
|
+
try:
|
|
304
|
+
response = next(request)
|
|
305
|
+
except Exception as exc:
|
|
306
|
+
if not _is_retryable(exc) or attempt == max_attempts:
|
|
307
|
+
raise
|
|
308
|
+
if not _is_pre_transmission(exc) and not _prepare_retry(request.input):
|
|
309
|
+
logger.debug(
|
|
310
|
+
"attempt %d/%d failed: %r, not retrying because the request body cannot be replayed",
|
|
311
|
+
attempt,
|
|
312
|
+
max_attempts,
|
|
313
|
+
exc,
|
|
314
|
+
extra={
|
|
315
|
+
"attempt": attempt,
|
|
316
|
+
"max_attempts": max_attempts,
|
|
317
|
+
"error": exc,
|
|
318
|
+
},
|
|
319
|
+
)
|
|
320
|
+
raise
|
|
321
|
+
is_throttling = (
|
|
322
|
+
isinstance(exc, ServiceError) and exc.is_throttling_error
|
|
323
|
+
)
|
|
324
|
+
cost = quota.acquire(is_throttling)
|
|
325
|
+
if cost is None:
|
|
326
|
+
logger.debug(
|
|
327
|
+
"attempt %d/%d failed: %r, not retrying because the retry quota is exhausted",
|
|
328
|
+
attempt,
|
|
329
|
+
max_attempts,
|
|
330
|
+
exc,
|
|
331
|
+
extra={
|
|
332
|
+
"attempt": attempt,
|
|
333
|
+
"max_attempts": max_attempts,
|
|
334
|
+
"error": exc,
|
|
335
|
+
},
|
|
336
|
+
)
|
|
337
|
+
raise
|
|
338
|
+
last_retry_cost = cost
|
|
339
|
+
delay = _retry_delay(attempt, is_throttling)
|
|
340
|
+
logger.debug(
|
|
341
|
+
"attempt %d/%d failed: %r, retrying in %.2fs",
|
|
342
|
+
attempt,
|
|
343
|
+
max_attempts,
|
|
344
|
+
exc,
|
|
345
|
+
delay,
|
|
346
|
+
extra={
|
|
347
|
+
"attempt": attempt,
|
|
348
|
+
"max_attempts": max_attempts,
|
|
349
|
+
"error": exc,
|
|
350
|
+
"delay": delay,
|
|
351
|
+
},
|
|
352
|
+
)
|
|
353
|
+
time.sleep(delay)
|
|
354
|
+
else:
|
|
355
|
+
quota.release(last_retry_cost)
|
|
356
|
+
return response
|
|
357
|
+
raise AssertionError("unreachable")
|
|
358
|
+
|
|
359
|
+
return interceptor
|
|
360
|
+
|
|
361
|
+
|
|
362
|
+
def aretry() -> AsyncInterceptor[TInput, TOutput]:
|
|
363
|
+
async def interceptor(
|
|
364
|
+
request: AsyncOperationRequest[TInput], next: AsyncNextFn[TInput, TOutput]
|
|
365
|
+
) -> AsyncOperationResponse[TOutput]:
|
|
366
|
+
max_attempts = request.options.retry_max_attempts or 3
|
|
367
|
+
quota = request.options.retry_quota
|
|
368
|
+
last_retry_cost: int | None = None
|
|
369
|
+
for attempt in range(1, max_attempts + 1):
|
|
370
|
+
try:
|
|
371
|
+
response = await next(request)
|
|
372
|
+
except Exception as exc:
|
|
373
|
+
if not _is_retryable(exc) or attempt == max_attempts:
|
|
374
|
+
raise
|
|
375
|
+
if not _is_pre_transmission(exc) and not await _aprepare_retry(
|
|
376
|
+
request.input
|
|
377
|
+
):
|
|
378
|
+
logger.debug(
|
|
379
|
+
"attempt %d/%d failed: %r, not retrying because the request body cannot be replayed",
|
|
380
|
+
attempt,
|
|
381
|
+
max_attempts,
|
|
382
|
+
exc,
|
|
383
|
+
extra={
|
|
384
|
+
"attempt": attempt,
|
|
385
|
+
"max_attempts": max_attempts,
|
|
386
|
+
"error": exc,
|
|
387
|
+
},
|
|
388
|
+
)
|
|
389
|
+
raise
|
|
390
|
+
is_throttling = (
|
|
391
|
+
isinstance(exc, ServiceError) and exc.is_throttling_error
|
|
392
|
+
)
|
|
393
|
+
cost = quota.acquire(is_throttling)
|
|
394
|
+
if cost is None:
|
|
395
|
+
logger.debug(
|
|
396
|
+
"attempt %d/%d failed: %r, not retrying because the retry quota is exhausted",
|
|
397
|
+
attempt,
|
|
398
|
+
max_attempts,
|
|
399
|
+
exc,
|
|
400
|
+
extra={
|
|
401
|
+
"attempt": attempt,
|
|
402
|
+
"max_attempts": max_attempts,
|
|
403
|
+
"error": exc,
|
|
404
|
+
},
|
|
405
|
+
)
|
|
406
|
+
raise
|
|
407
|
+
last_retry_cost = cost
|
|
408
|
+
delay = _retry_delay(attempt, is_throttling)
|
|
409
|
+
logger.debug(
|
|
410
|
+
"attempt %d/%d failed: %r, retrying in %.2fs",
|
|
411
|
+
attempt,
|
|
412
|
+
max_attempts,
|
|
413
|
+
exc,
|
|
414
|
+
delay,
|
|
415
|
+
extra={
|
|
416
|
+
"attempt": attempt,
|
|
417
|
+
"max_attempts": max_attempts,
|
|
418
|
+
"error": exc,
|
|
419
|
+
"delay": delay,
|
|
420
|
+
},
|
|
421
|
+
)
|
|
422
|
+
await anysleep(delay)
|
|
423
|
+
else:
|
|
424
|
+
quota.release(last_retry_cost)
|
|
425
|
+
return response
|
|
426
|
+
raise AssertionError("unreachable")
|
|
427
|
+
|
|
428
|
+
return interceptor
|
|
429
|
+
|
|
430
|
+
|
|
431
|
+
def log_operation(operation_logging: OperationLogging) -> Interceptor[TInput, TOutput]:
|
|
432
|
+
operation = operation_logging.operation
|
|
433
|
+
redact_input = operation_logging.redact_input
|
|
434
|
+
redact_output = operation_logging.redact_output
|
|
435
|
+
|
|
436
|
+
def interceptor(
|
|
437
|
+
request: OperationRequest[TInput], next: NextFn[TInput, TOutput]
|
|
438
|
+
) -> OperationResponse[TOutput]:
|
|
439
|
+
debug = logger.isEnabledFor(logging.DEBUG)
|
|
440
|
+
if debug:
|
|
441
|
+
input_ = (
|
|
442
|
+
request.input if redact_input is None else redact_input(request.input)
|
|
443
|
+
)
|
|
444
|
+
logger.debug(
|
|
445
|
+
"%s input: %r",
|
|
446
|
+
operation,
|
|
447
|
+
input_,
|
|
448
|
+
extra={"operation": operation, "input": input_},
|
|
449
|
+
)
|
|
450
|
+
try:
|
|
451
|
+
response = next(request)
|
|
452
|
+
except Exception as exc:
|
|
453
|
+
logger.debug(
|
|
454
|
+
"%s failed: %r",
|
|
455
|
+
operation,
|
|
456
|
+
exc,
|
|
457
|
+
extra={"operation": operation, "error": exc},
|
|
458
|
+
)
|
|
459
|
+
raise
|
|
460
|
+
if debug:
|
|
461
|
+
output = (
|
|
462
|
+
response.output
|
|
463
|
+
if redact_output is None
|
|
464
|
+
else redact_output(response.output)
|
|
465
|
+
)
|
|
466
|
+
logger.debug(
|
|
467
|
+
"%s output: %r",
|
|
468
|
+
operation,
|
|
469
|
+
output,
|
|
470
|
+
extra={"operation": operation, "output": output},
|
|
471
|
+
)
|
|
472
|
+
return response
|
|
473
|
+
|
|
474
|
+
return interceptor
|
|
475
|
+
|
|
476
|
+
|
|
477
|
+
def alog_operation(
|
|
478
|
+
operation_logging: OperationLogging,
|
|
479
|
+
) -> AsyncInterceptor[TInput, TOutput]:
|
|
480
|
+
operation = operation_logging.operation
|
|
481
|
+
redact_input = operation_logging.redact_input
|
|
482
|
+
redact_output = operation_logging.redact_output
|
|
483
|
+
|
|
484
|
+
async def interceptor(
|
|
485
|
+
request: AsyncOperationRequest[TInput], next: AsyncNextFn[TInput, TOutput]
|
|
486
|
+
) -> AsyncOperationResponse[TOutput]:
|
|
487
|
+
debug = logger.isEnabledFor(logging.DEBUG)
|
|
488
|
+
if debug:
|
|
489
|
+
input_ = (
|
|
490
|
+
request.input if redact_input is None else redact_input(request.input)
|
|
491
|
+
)
|
|
492
|
+
logger.debug(
|
|
493
|
+
"%s input: %r",
|
|
494
|
+
operation,
|
|
495
|
+
input_,
|
|
496
|
+
extra={"operation": operation, "input": input_},
|
|
497
|
+
)
|
|
498
|
+
try:
|
|
499
|
+
response = await next(request)
|
|
500
|
+
except Exception as exc:
|
|
501
|
+
logger.debug(
|
|
502
|
+
"%s failed: %r",
|
|
503
|
+
operation,
|
|
504
|
+
exc,
|
|
505
|
+
extra={"operation": operation, "error": exc},
|
|
506
|
+
)
|
|
507
|
+
raise
|
|
508
|
+
if debug:
|
|
509
|
+
output = (
|
|
510
|
+
response.output
|
|
511
|
+
if redact_output is None
|
|
512
|
+
else redact_output(response.output)
|
|
513
|
+
)
|
|
514
|
+
logger.debug(
|
|
515
|
+
"%s output: %r",
|
|
516
|
+
operation,
|
|
517
|
+
output,
|
|
518
|
+
extra={"operation": operation, "output": output},
|
|
519
|
+
)
|
|
520
|
+
return response
|
|
521
|
+
|
|
522
|
+
return interceptor
|