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 | class HttpxBackend(OpenAIBackendMixin):
SERIALIZER = HttpxSerializer
DEFAULT_SRC = "https://api.reka.ai/v1/chat"
DEFAULT_MODEL = "reka-core-20240501"
def __init__(
self, api_key=None, endpoint: str | None = None, serializer: Serializer | None = None, **kwargs
) -> None:
"""Initializes the CompleteBackend. Defaults to using the API key from the environment and.
Args:
api_key (Optional[str]): The API key for the Complete service.
endpoint (str): The base URL for the Complete API.
serializer (Optional[Serializer]): The serializer to use for serializing messages.
"""
self.base_url = endpoint or self.DEFAULT_SRC
self.api_key = api_key or os.getenv("MBODI_API_KEY")
self.headers = None
if not self.api_key:
self.headers = {"Content-Type": "application/json"}
else:
self.headers = {"X-Api-Key": self.api_key, "Content-Type": "application/json"}
self.serialized = serializer or self.SERIALIZER
self.kwargs = kwargs
self.DEFAULT_MODEL = kwargs.get("model", self.DEFAULT_MODEL)
@overload
def predict(self, messages: List[Message], model: str | None = None, **kwargs) -> str: ...
@overload
def predict(
self, message: Message, context: List[Message] | None = None, model: str | None = None, **kwargs
) -> str: ...
def predict(
self, message_or_messages, context: List[Message] | None = None, model: str | None = None, **kwargs
) -> str:
# Determine if the input is a list of messages or a single message
if isinstance(message_or_messages, list):
messages = message_or_messages
else:
# If a single message is provided, append it to the context
messages = (context or []) + [message_or_messages]
model = model or self.DEFAULT_MODEL
data = {
"messages": [self.serialized(msg).serialize() for msg in messages],
"model": model,
"stream": False,
**kwargs,
}
with httpx.Client(trust_env=True) as client:
response = client.post(
self.base_url, headers=self.headers, json=data, timeout=kwargs.get("timeout", 60), follow_redirects=True
)
# Process response
if response.status_code == 200:
response_data = response.json()
return self.serialized.extract_response(response_data)
response.raise_for_status()
return response.text
@overload
def stream(self, messages: List[Message], model: str | None = None, **kwargs) -> Generator[str, None, None]: ...
@overload
def stream(
self, message: Message, context: List[Message] | None = None, model: str | None = None, **kwargs
) -> Generator[str, None, None]: ...
def stream(
self, message_or_messages, context: List[Message] | None = None, model: str | None = None, **kwargs
) -> Generator[str, None, None]:
if isinstance(message_or_messages, list):
messages = message_or_messages
else:
# If a single message is provided, append it to the context
messages = (context or []) + [message_or_messages]
model = model or self.DEFAULT_MODEL
yield from self._stream_completion(messages, model, **kwargs)
@overload
async def astream(
self, messages: List[Message], model: str | None = None, **kwargs
) -> AsyncGenerator[str, None]: ...
@overload
async def astream(
self, message: Message, context: List[Message] | None = None, model: str | None = None, **kwargs
) -> AsyncGenerator[str, None]: ...
async def astream(
self, message_or_messages, context: List[Message] | None = None, model: str | None = None, **kwargs
) -> AsyncGenerator[str, None]:
if isinstance(message_or_messages, list):
messages = message_or_messages
else:
# If a single message is provided, append it to the context
messages = (context or []) + [message_or_messages]
model = model or self.DEFAULT_MODEL
async for chunk in self._astream_completion(messages, model, **kwargs):
yield chunk
def _stream_completion(
self,
messages: List[Message],
model: str | None = None,
**kwargs,
) -> Generator[str, None, None]:
model = model or self.DEFAULT_MODEL
data = {
"messages": [self.serialized(msg).serialize() for msg in messages],
"model": model,
"stream": True,
**kwargs,
}
data.update(kwargs)
with (
httpx.Client(follow_redirects=True) as client,
client.stream(
"post", self.base_url, headers=self.headers, json=data, timeout=kwargs.get("timeout", 60)
) as stream,
):
for chunk in stream.iter_text():
yield self.serialized.extract_stream(chunk)
async def _acreate_completion(self, messages: List[Message], model: str | None = None, **kwargs) -> str:
model = model or self.DEFAULT_MODEL
data = {
"messages": [self.serialized(msg).serialize() for msg in messages],
"model": model,
"stream": False,
**kwargs,
}
data.update(kwargs)
async with httpx.AsyncClient(timeout=-1) as client:
response = await client.post(
self.base_url, headers=self.headers, json=data, timeout=kwargs.get("timeout", 60)
)
if response.status_code == 200:
response_data = response.json()
return self.serialized.extract_response(response_data)
return response.text
async def _astream_completion(
self, messages: List[Message], model: str | None = None, **kwargs
) -> AsyncGenerator[str, None]:
model = model or self.DEFAULT_MODEL
data = {
"messages": [self.serialized(msg).serialize() for msg in messages],
"model": model,
"stream": True,
**kwargs,
}
data.update(kwargs)
async with httpx.AsyncClient(follow_redirects=True) as client:
stream = client.stream("POST", self.base_url, headers=self.headers, json=data, timeout=60)
async with stream as response:
async for chunk in response.aiter_text():
yield self.serialized.extract_stream(chunk)
|