File size: 8,610 Bytes
e3278e4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import json
from typing import Any, Dict, List, Optional

import httpx
from openai.types.image import Image

import litellm
from litellm.llms.custom_httpx.http_handler import (
    AsyncHTTPHandler,
    HTTPHandler,
    get_async_httpx_client,
)
from litellm.llms.vertex_ai.gemini.vertex_and_google_ai_studio_gemini import VertexLLM
from litellm.types.utils import ImageResponse


class VertexImageGeneration(VertexLLM):
    def process_image_generation_response(
        self,
        json_response: Dict[str, Any],
        model_response: ImageResponse,
        model: Optional[str] = None,
    ) -> ImageResponse:
        if "predictions" not in json_response:
            raise litellm.InternalServerError(
                message=f"image generation response does not contain 'predictions', got {json_response}",
                llm_provider="vertex_ai",
                model=model,
            )

        predictions = json_response["predictions"]
        response_data: List[Image] = []

        for prediction in predictions:
            bytes_base64_encoded = prediction["bytesBase64Encoded"]
            image_object = Image(b64_json=bytes_base64_encoded)
            response_data.append(image_object)

        model_response.data = response_data
        return model_response

    def image_generation(
        self,
        prompt: str,
        vertex_project: Optional[str],
        vertex_location: Optional[str],
        vertex_credentials: Optional[str],
        model_response: ImageResponse,
        logging_obj: Any,
        model: Optional[
            str
        ] = "imagegeneration",  # vertex ai uses imagegeneration as the default model
        client: Optional[Any] = None,
        optional_params: Optional[dict] = None,
        timeout: Optional[int] = None,
        aimg_generation=False,
    ) -> ImageResponse:
        if aimg_generation is True:
            return self.aimage_generation(  # type: ignore
                prompt=prompt,
                vertex_project=vertex_project,
                vertex_location=vertex_location,
                vertex_credentials=vertex_credentials,
                model=model,
                client=client,
                optional_params=optional_params,
                timeout=timeout,
                logging_obj=logging_obj,
                model_response=model_response,
            )

        if client is None:
            _params = {}
            if timeout is not None:
                if isinstance(timeout, float) or isinstance(timeout, int):
                    _httpx_timeout = httpx.Timeout(timeout)
                    _params["timeout"] = _httpx_timeout
            else:
                _params["timeout"] = httpx.Timeout(timeout=600.0, connect=5.0)

            sync_handler: HTTPHandler = HTTPHandler(**_params)  # type: ignore
        else:
            sync_handler = client  # type: ignore

        url = f"https://{vertex_location}-aiplatform.googleapis.com/v1/projects/{vertex_project}/locations/{vertex_location}/publishers/google/models/{model}:predict"

        auth_header, _ = self._ensure_access_token(
            credentials=vertex_credentials,
            project_id=vertex_project,
            custom_llm_provider="vertex_ai",
        )
        optional_params = optional_params or {
            "sampleCount": 1
        }  # default optional params

        request_data = {
            "instances": [{"prompt": prompt}],
            "parameters": optional_params,
        }

        request_str = f"\n curl -X POST \\\n -H \"Authorization: Bearer {auth_header[:10] + 'XXXXXXXXXX'}\" \\\n -H \"Content-Type: application/json; charset=utf-8\" \\\n -d {request_data} \\\n \"{url}\""
        logging_obj.pre_call(
            input=prompt,
            api_key=None,
            additional_args={
                "complete_input_dict": optional_params,
                "request_str": request_str,
            },
        )

        logging_obj.pre_call(
            input=prompt,
            api_key=None,
            additional_args={
                "complete_input_dict": optional_params,
                "request_str": request_str,
            },
        )

        response = sync_handler.post(
            url=url,
            headers={
                "Content-Type": "application/json; charset=utf-8",
                "Authorization": f"Bearer {auth_header}",
            },
            data=json.dumps(request_data),
        )

        if response.status_code != 200:
            raise Exception(f"Error: {response.status_code} {response.text}")

        json_response = response.json()
        return self.process_image_generation_response(
            json_response, model_response, model
        )

    async def aimage_generation(
        self,
        prompt: str,
        vertex_project: Optional[str],
        vertex_location: Optional[str],
        vertex_credentials: Optional[str],
        model_response: litellm.ImageResponse,
        logging_obj: Any,
        model: Optional[
            str
        ] = "imagegeneration",  # vertex ai uses imagegeneration as the default model
        client: Optional[AsyncHTTPHandler] = None,
        optional_params: Optional[dict] = None,
        timeout: Optional[int] = None,
    ):
        response = None
        if client is None:
            _params = {}
            if timeout is not None:
                if isinstance(timeout, float) or isinstance(timeout, int):
                    _httpx_timeout = httpx.Timeout(timeout)
                    _params["timeout"] = _httpx_timeout
            else:
                _params["timeout"] = httpx.Timeout(timeout=600.0, connect=5.0)

            self.async_handler = get_async_httpx_client(
                llm_provider=litellm.LlmProviders.VERTEX_AI,
                params={"timeout": timeout},
            )
        else:
            self.async_handler = client  # type: ignore

        # make POST request to
        # https://us-central1-aiplatform.googleapis.com/v1/projects/PROJECT_ID/locations/us-central1/publishers/google/models/imagegeneration:predict
        url = f"https://{vertex_location}-aiplatform.googleapis.com/v1/projects/{vertex_project}/locations/{vertex_location}/publishers/google/models/{model}:predict"

        """
        Docs link: https://console.cloud.google.com/vertex-ai/publishers/google/model-garden/imagegeneration?project=adroit-crow-413218
        curl -X POST \
        -H "Authorization: Bearer $(gcloud auth print-access-token)" \
        -H "Content-Type: application/json; charset=utf-8" \
        -d {
            "instances": [
                {
                    "prompt": "a cat"
                }
            ],
            "parameters": {
                "sampleCount": 1
            }
        } \
        "https://us-central1-aiplatform.googleapis.com/v1/projects/PROJECT_ID/locations/us-central1/publishers/google/models/imagegeneration:predict"
        """
        auth_header, _ = self._ensure_access_token(
            credentials=vertex_credentials,
            project_id=vertex_project,
            custom_llm_provider="vertex_ai",
        )
        optional_params = optional_params or {
            "sampleCount": 1
        }  # default optional params

        request_data = {
            "instances": [{"prompt": prompt}],
            "parameters": optional_params,
        }

        request_str = f"\n curl -X POST \\\n -H \"Authorization: Bearer {auth_header[:10] + 'XXXXXXXXXX'}\" \\\n -H \"Content-Type: application/json; charset=utf-8\" \\\n -d {request_data} \\\n \"{url}\""
        logging_obj.pre_call(
            input=prompt,
            api_key=None,
            additional_args={
                "complete_input_dict": optional_params,
                "request_str": request_str,
            },
        )

        response = await self.async_handler.post(
            url=url,
            headers={
                "Content-Type": "application/json; charset=utf-8",
                "Authorization": f"Bearer {auth_header}",
            },
            data=json.dumps(request_data),
        )

        if response.status_code != 200:
            raise Exception(f"Error: {response.status_code} {response.text}")

        json_response = response.json()
        return self.process_image_generation_response(
            json_response, model_response, model
        )

    def is_image_generation_response(self, json_response: Dict[str, Any]) -> bool:
        if "predictions" in json_response:
            if "bytesBase64Encoded" in json_response["predictions"][0]:
                return True
        return False