/
rest.py
468 lines (386 loc) · 18 KB
/
rest.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
# -*- coding: utf-8 -*-
# Copyright 2023 Google LLC
#
# 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.
#
import dataclasses
import json # type: ignore
import re
from typing import Any, Callable, Dict, List, Optional, Sequence, Tuple, Union
import warnings
from google.api_core import gapic_v1, path_template, rest_helpers, rest_streaming
from google.api_core import exceptions as core_exceptions
from google.api_core import retry as retries
from google.auth import credentials as ga_credentials # type: ignore
from google.auth.transport.grpc import SslCredentials # type: ignore
from google.auth.transport.requests import AuthorizedSession # type: ignore
from google.protobuf import json_format
import grpc # type: ignore
from requests import __version__ as requests_version
try:
OptionalRetry = Union[retries.Retry, gapic_v1.method._MethodDefault, None]
except AttributeError: # pragma: NO COVER
OptionalRetry = Union[retries.Retry, object, None] # type: ignore
from google.longrunning import operations_pb2 # type: ignore
from google.ai.generativelanguage_v1beta.types import discuss_service
from .base import DEFAULT_CLIENT_INFO as BASE_DEFAULT_CLIENT_INFO
from .base import DiscussServiceTransport
DEFAULT_CLIENT_INFO = gapic_v1.client_info.ClientInfo(
gapic_version=BASE_DEFAULT_CLIENT_INFO.gapic_version,
grpc_version=None,
rest_version=requests_version,
)
class DiscussServiceRestInterceptor:
"""Interceptor for DiscussService.
Interceptors are used to manipulate requests, request metadata, and responses
in arbitrary ways.
Example use cases include:
* Logging
* Verifying requests according to service or custom semantics
* Stripping extraneous information from responses
These use cases and more can be enabled by injecting an
instance of a custom subclass when constructing the DiscussServiceRestTransport.
.. code-block:: python
class MyCustomDiscussServiceInterceptor(DiscussServiceRestInterceptor):
def pre_count_message_tokens(self, request, metadata):
logging.log(f"Received request: {request}")
return request, metadata
def post_count_message_tokens(self, response):
logging.log(f"Received response: {response}")
return response
def pre_generate_message(self, request, metadata):
logging.log(f"Received request: {request}")
return request, metadata
def post_generate_message(self, response):
logging.log(f"Received response: {response}")
return response
transport = DiscussServiceRestTransport(interceptor=MyCustomDiscussServiceInterceptor())
client = DiscussServiceClient(transport=transport)
"""
def pre_count_message_tokens(
self,
request: discuss_service.CountMessageTokensRequest,
metadata: Sequence[Tuple[str, str]],
) -> Tuple[discuss_service.CountMessageTokensRequest, Sequence[Tuple[str, str]]]:
"""Pre-rpc interceptor for count_message_tokens
Override in a subclass to manipulate the request or metadata
before they are sent to the DiscussService server.
"""
return request, metadata
def post_count_message_tokens(
self, response: discuss_service.CountMessageTokensResponse
) -> discuss_service.CountMessageTokensResponse:
"""Post-rpc interceptor for count_message_tokens
Override in a subclass to manipulate the response
after it is returned by the DiscussService server but before
it is returned to user code.
"""
return response
def pre_generate_message(
self,
request: discuss_service.GenerateMessageRequest,
metadata: Sequence[Tuple[str, str]],
) -> Tuple[discuss_service.GenerateMessageRequest, Sequence[Tuple[str, str]]]:
"""Pre-rpc interceptor for generate_message
Override in a subclass to manipulate the request or metadata
before they are sent to the DiscussService server.
"""
return request, metadata
def post_generate_message(
self, response: discuss_service.GenerateMessageResponse
) -> discuss_service.GenerateMessageResponse:
"""Post-rpc interceptor for generate_message
Override in a subclass to manipulate the response
after it is returned by the DiscussService server but before
it is returned to user code.
"""
return response
@dataclasses.dataclass
class DiscussServiceRestStub:
_session: AuthorizedSession
_host: str
_interceptor: DiscussServiceRestInterceptor
class DiscussServiceRestTransport(DiscussServiceTransport):
"""REST backend transport for DiscussService.
An API for using Generative Language Models (GLMs) in dialog
applications.
Also known as large language models (LLMs), this API provides
models that are trained for multi-turn dialog.
This class defines the same methods as the primary client, so the
primary client can load the underlying transport implementation
and call it.
It sends JSON representations of protocol buffers over HTTP/1.1
"""
def __init__(
self,
*,
host: str = "generativelanguage.googleapis.com",
credentials: Optional[ga_credentials.Credentials] = None,
credentials_file: Optional[str] = None,
scopes: Optional[Sequence[str]] = None,
client_cert_source_for_mtls: Optional[Callable[[], Tuple[bytes, bytes]]] = None,
quota_project_id: Optional[str] = None,
client_info: gapic_v1.client_info.ClientInfo = DEFAULT_CLIENT_INFO,
always_use_jwt_access: Optional[bool] = False,
url_scheme: str = "https",
interceptor: Optional[DiscussServiceRestInterceptor] = None,
api_audience: Optional[str] = None,
) -> None:
"""Instantiate the transport.
Args:
host (Optional[str]):
The hostname to connect to (default: 'generativelanguage.googleapis.com').
credentials (Optional[google.auth.credentials.Credentials]): The
authorization credentials to attach to requests. These
credentials identify the application to the service; if none
are specified, the client will attempt to ascertain the
credentials from the environment.
credentials_file (Optional[str]): A file with credentials that can
be loaded with :func:`google.auth.load_credentials_from_file`.
This argument is ignored if ``channel`` is provided.
scopes (Optional(Sequence[str])): A list of scopes. This argument is
ignored if ``channel`` is provided.
client_cert_source_for_mtls (Callable[[], Tuple[bytes, bytes]]): Client
certificate to configure mutual TLS HTTP channel. It is ignored
if ``channel`` is provided.
quota_project_id (Optional[str]): An optional project to use for billing
and quota.
client_info (google.api_core.gapic_v1.client_info.ClientInfo):
The client info used to send a user-agent string along with
API requests. If ``None``, then default info will be used.
Generally, you only need to set this if you are developing
your own client library.
always_use_jwt_access (Optional[bool]): Whether self signed JWT should
be used for service account credentials.
url_scheme: the protocol scheme for the API endpoint. Normally
"https", but for testing or local servers,
"http" can be specified.
"""
# Run the base constructor
# TODO(yon-mg): resolve other ctor params i.e. scopes, quota, etc.
# TODO: When custom host (api_endpoint) is set, `scopes` must *also* be set on the
# credentials object
maybe_url_match = re.match("^(?P<scheme>http(?:s)?://)?(?P<host>.*)$", host)
if maybe_url_match is None:
raise ValueError(
f"Unexpected hostname structure: {host}"
) # pragma: NO COVER
url_match_items = maybe_url_match.groupdict()
host = f"{url_scheme}://{host}" if not url_match_items["scheme"] else host
super().__init__(
host=host,
credentials=credentials,
client_info=client_info,
always_use_jwt_access=always_use_jwt_access,
api_audience=api_audience,
)
self._session = AuthorizedSession(
self._credentials, default_host=self.DEFAULT_HOST
)
if client_cert_source_for_mtls:
self._session.configure_mtls_channel(client_cert_source_for_mtls)
self._interceptor = interceptor or DiscussServiceRestInterceptor()
self._prep_wrapped_messages(client_info)
class _CountMessageTokens(DiscussServiceRestStub):
def __hash__(self):
return hash("CountMessageTokens")
__REQUIRED_FIELDS_DEFAULT_VALUES: Dict[str, Any] = {}
@classmethod
def _get_unset_required_fields(cls, message_dict):
return {
k: v
for k, v in cls.__REQUIRED_FIELDS_DEFAULT_VALUES.items()
if k not in message_dict
}
def __call__(
self,
request: discuss_service.CountMessageTokensRequest,
*,
retry: OptionalRetry = gapic_v1.method.DEFAULT,
timeout: Optional[float] = None,
metadata: Sequence[Tuple[str, str]] = (),
) -> discuss_service.CountMessageTokensResponse:
r"""Call the count message tokens method over HTTP.
Args:
request (~.discuss_service.CountMessageTokensRequest):
The request object. Counts the number of tokens in the ``prompt`` sent to a
model.
Models may tokenize text differently, so each model may
return a different ``token_count``.
retry (google.api_core.retry.Retry): Designation of what errors, if any,
should be retried.
timeout (float): The timeout for this request.
metadata (Sequence[Tuple[str, str]]): Strings which should be
sent along with the request as metadata.
Returns:
~.discuss_service.CountMessageTokensResponse:
A response from ``CountMessageTokens``.
It returns the model's ``token_count`` for the
``prompt``.
"""
http_options: List[Dict[str, str]] = [
{
"method": "post",
"uri": "/v1beta/{model=models/*}:countMessageTokens",
"body": "*",
},
]
request, metadata = self._interceptor.pre_count_message_tokens(
request, metadata
)
pb_request = discuss_service.CountMessageTokensRequest.pb(request)
transcoded_request = path_template.transcode(http_options, pb_request)
# Jsonify the request body
body = json_format.MessageToJson(
transcoded_request["body"], use_integers_for_enums=True
)
uri = transcoded_request["uri"]
method = transcoded_request["method"]
# Jsonify the query params
query_params = json.loads(
json_format.MessageToJson(
transcoded_request["query_params"],
use_integers_for_enums=True,
)
)
query_params.update(self._get_unset_required_fields(query_params))
query_params["$alt"] = "json;enum-encoding=int"
# Send the request
headers = dict(metadata)
headers["Content-Type"] = "application/json"
response = getattr(self._session, method)(
"{host}{uri}".format(host=self._host, uri=uri),
timeout=timeout,
headers=headers,
params=rest_helpers.flatten_query_params(query_params, strict=True),
data=body,
)
# In case of error, raise the appropriate core_exceptions.GoogleAPICallError exception
# subclass.
if response.status_code >= 400:
raise core_exceptions.from_http_response(response)
# Return the response
resp = discuss_service.CountMessageTokensResponse()
pb_resp = discuss_service.CountMessageTokensResponse.pb(resp)
json_format.Parse(response.content, pb_resp, ignore_unknown_fields=True)
resp = self._interceptor.post_count_message_tokens(resp)
return resp
class _GenerateMessage(DiscussServiceRestStub):
def __hash__(self):
return hash("GenerateMessage")
__REQUIRED_FIELDS_DEFAULT_VALUES: Dict[str, Any] = {}
@classmethod
def _get_unset_required_fields(cls, message_dict):
return {
k: v
for k, v in cls.__REQUIRED_FIELDS_DEFAULT_VALUES.items()
if k not in message_dict
}
def __call__(
self,
request: discuss_service.GenerateMessageRequest,
*,
retry: OptionalRetry = gapic_v1.method.DEFAULT,
timeout: Optional[float] = None,
metadata: Sequence[Tuple[str, str]] = (),
) -> discuss_service.GenerateMessageResponse:
r"""Call the generate message method over HTTP.
Args:
request (~.discuss_service.GenerateMessageRequest):
The request object. Request to generate a message
response from the model.
retry (google.api_core.retry.Retry): Designation of what errors, if any,
should be retried.
timeout (float): The timeout for this request.
metadata (Sequence[Tuple[str, str]]): Strings which should be
sent along with the request as metadata.
Returns:
~.discuss_service.GenerateMessageResponse:
The response from the model.
This includes candidate messages and
conversation history in the form of
chronologically-ordered messages.
"""
http_options: List[Dict[str, str]] = [
{
"method": "post",
"uri": "/v1beta/{model=models/*}:generateMessage",
"body": "*",
},
]
request, metadata = self._interceptor.pre_generate_message(
request, metadata
)
pb_request = discuss_service.GenerateMessageRequest.pb(request)
transcoded_request = path_template.transcode(http_options, pb_request)
# Jsonify the request body
body = json_format.MessageToJson(
transcoded_request["body"], use_integers_for_enums=True
)
uri = transcoded_request["uri"]
method = transcoded_request["method"]
# Jsonify the query params
query_params = json.loads(
json_format.MessageToJson(
transcoded_request["query_params"],
use_integers_for_enums=True,
)
)
query_params.update(self._get_unset_required_fields(query_params))
query_params["$alt"] = "json;enum-encoding=int"
# Send the request
headers = dict(metadata)
headers["Content-Type"] = "application/json"
response = getattr(self._session, method)(
"{host}{uri}".format(host=self._host, uri=uri),
timeout=timeout,
headers=headers,
params=rest_helpers.flatten_query_params(query_params, strict=True),
data=body,
)
# In case of error, raise the appropriate core_exceptions.GoogleAPICallError exception
# subclass.
if response.status_code >= 400:
raise core_exceptions.from_http_response(response)
# Return the response
resp = discuss_service.GenerateMessageResponse()
pb_resp = discuss_service.GenerateMessageResponse.pb(resp)
json_format.Parse(response.content, pb_resp, ignore_unknown_fields=True)
resp = self._interceptor.post_generate_message(resp)
return resp
@property
def count_message_tokens(
self,
) -> Callable[
[discuss_service.CountMessageTokensRequest],
discuss_service.CountMessageTokensResponse,
]:
# The return type is fine, but mypy isn't sophisticated enough to determine what's going on here.
# In C++ this would require a dynamic_cast
return self._CountMessageTokens(self._session, self._host, self._interceptor) # type: ignore
@property
def generate_message(
self,
) -> Callable[
[discuss_service.GenerateMessageRequest],
discuss_service.GenerateMessageResponse,
]:
# The return type is fine, but mypy isn't sophisticated enough to determine what's going on here.
# In C++ this would require a dynamic_cast
return self._GenerateMessage(self._session, self._host, self._interceptor) # type: ignore
@property
def kind(self) -> str:
return "rest"
def close(self):
self._session.close()
__all__ = ("DiscussServiceRestTransport",)