File size: 10,205 Bytes
447ebeb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import os
import sys
import time
import traceback
import uuid

from dotenv import load_dotenv

load_dotenv()
sys.path.insert(
    0, os.path.abspath("../..")
)  # Adds the parent directory to the system path
import asyncio
import hashlib
import random

import pytest

import litellm
from litellm import aembedding, completion, embedding
from litellm.caching.caching import Cache

from unittest.mock import AsyncMock, patch, MagicMock
from litellm.caching.caching_handler import LLMCachingHandler, CachingHandlerResponse
from litellm.caching.caching import LiteLLMCacheType
from litellm.types.utils import CallTypes
from litellm.types.rerank import RerankResponse
from litellm.types.utils import (
    ModelResponse,
    EmbeddingResponse,
    TextCompletionResponse,
    TranscriptionResponse,
    Embedding,
)
from datetime import timedelta, datetime
from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLogging
from litellm._logging import verbose_logger
import logging


def setup_cache():
    # Set up the cache
    cache = Cache(type=LiteLLMCacheType.LOCAL)
    litellm.cache = cache
    return cache


chat_completion_response = litellm.ModelResponse(
    id=str(uuid.uuid4()),
    choices=[
        litellm.Choices(
            message=litellm.Message(
                role="assistant", content="Hello, how can I help you today?"
            )
        )
    ],
)

text_completion_response = litellm.TextCompletionResponse(
    id=str(uuid.uuid4()),
    choices=[litellm.utils.TextChoices(text="Hello, how can I help you today?")],
)


@pytest.mark.asyncio
@pytest.mark.parametrize(
    "response", [chat_completion_response, text_completion_response]
)
async def test_async_set_get_cache(response):
    litellm.set_verbose = True
    setup_cache()
    verbose_logger.setLevel(logging.DEBUG)
    caching_handler = LLMCachingHandler(
        original_function=completion, request_kwargs={}, start_time=datetime.now()
    )

    messages = [{"role": "user", "content": f"Unique message {datetime.now()}"}]

    logging_obj = LiteLLMLogging(
        litellm_call_id=str(datetime.now()),
        call_type=CallTypes.completion.value,
        model="gpt-3.5-turbo",
        messages=messages,
        function_id=str(uuid.uuid4()),
        stream=False,
        start_time=datetime.now(),
    )

    result = response
    print("result", result)

    original_function = (
        litellm.acompletion
        if isinstance(response, litellm.ModelResponse)
        else litellm.atext_completion
    )
    if isinstance(response, litellm.ModelResponse):
        kwargs = {"messages": messages}
        call_type = CallTypes.acompletion.value
    else:
        kwargs = {"prompt": f"Hello, how can I help you today? {datetime.now()}"}
        call_type = CallTypes.atext_completion.value

    await caching_handler.async_set_cache(
        result=result, original_function=original_function, kwargs=kwargs
    )

    await asyncio.sleep(2)

    # Verify the result was cached
    cached_response = await caching_handler._async_get_cache(
        model="gpt-3.5-turbo",
        original_function=original_function,
        logging_obj=logging_obj,
        start_time=datetime.now(),
        call_type=call_type,
        kwargs=kwargs,
    )

    assert cached_response.cached_result is not None
    assert cached_response.cached_result.id == result.id


@pytest.mark.asyncio
async def test_async_log_cache_hit_on_callbacks():
    """
    Assert logging callbacks are called after a cache hit
    """
    # Setup
    caching_handler = LLMCachingHandler(
        original_function=completion, request_kwargs={}, start_time=datetime.now()
    )

    mock_logging_obj = MagicMock()
    mock_logging_obj.async_success_handler = AsyncMock()
    mock_logging_obj.success_handler = MagicMock()

    cached_result = "Mocked cached result"
    start_time = datetime.now()
    end_time = start_time + timedelta(seconds=1)
    cache_hit = True

    # Call the method
    caching_handler._async_log_cache_hit_on_callbacks(
        logging_obj=mock_logging_obj,
        cached_result=cached_result,
        start_time=start_time,
        end_time=end_time,
        cache_hit=cache_hit,
    )

    # Wait for the async task to complete
    await asyncio.sleep(0.5)

    print("mock logging obj methods called", mock_logging_obj.mock_calls)

    # Assertions
    mock_logging_obj.async_success_handler.assert_called_once_with(
        cached_result, start_time, end_time, cache_hit
    )

    # Wait for the thread to complete
    await asyncio.sleep(0.5)

    mock_logging_obj.success_handler.assert_called_once_with(
        cached_result, start_time, end_time, cache_hit
    )


@pytest.mark.parametrize(
    "call_type, cached_result, expected_type",
    [
        (
            CallTypes.completion.value,
            {
                "id": "test",
                "choices": [{"message": {"role": "assistant", "content": "Hello"}}],
            },
            ModelResponse,
        ),
        (
            CallTypes.text_completion.value,
            {"id": "test", "choices": [{"text": "Hello"}]},
            TextCompletionResponse,
        ),
        (
            CallTypes.embedding.value,
            {"data": [{"embedding": [0.1, 0.2, 0.3]}]},
            EmbeddingResponse,
        ),
        (
            CallTypes.rerank.value,
            {"id": "test", "results": [{"index": 0, "relevance_score": 0.9}]},
            RerankResponse,
        ),
        (
            CallTypes.transcription.value,
            {"text": "Hello, world!"},
            TranscriptionResponse,
        ),
    ],
)
def test_convert_cached_result_to_model_response(
    call_type, cached_result, expected_type
):
    """
    Assert that the cached result is converted to the correct type
    """
    caching_handler = LLMCachingHandler(
        original_function=lambda: None, request_kwargs={}, start_time=datetime.now()
    )
    logging_obj = LiteLLMLogging(
        litellm_call_id=str(datetime.now()),
        call_type=call_type,
        model="gpt-3.5-turbo",
        messages=[{"role": "user", "content": "Hello, how can I help you today?"}],
        function_id=str(uuid.uuid4()),
        stream=False,
        start_time=datetime.now(),
    )

    result = caching_handler._convert_cached_result_to_model_response(
        cached_result=cached_result,
        call_type=call_type,
        kwargs={},
        logging_obj=logging_obj,
        model="test-model",
        args=(),
    )

    assert isinstance(result, expected_type)
    assert result is not None


def test_combine_cached_embedding_response_with_api_result():
    """
    If the cached response has [cache_hit, None, cache_hit]
    result should be [cache_hit, api_result, cache_hit]
    """
    # Setup
    caching_handler = LLMCachingHandler(
        original_function=lambda: None, request_kwargs={}, start_time=datetime.now()
    )

    start_time = datetime.now()
    end_time = start_time + timedelta(seconds=1)

    # Create a CachingHandlerResponse with some cached and some None values
    cached_response = EmbeddingResponse(
        data=[
            Embedding(embedding=[0.1, 0.2, 0.3], index=0, object="embedding"),
            None,
            Embedding(embedding=[0.7, 0.8, 0.9], index=2, object="embedding"),
        ]
    )
    caching_handler_response = CachingHandlerResponse(
        final_embedding_cached_response=cached_response
    )

    # Create an API EmbeddingResponse for the missing value
    api_response = EmbeddingResponse(
        data=[Embedding(embedding=[0.4, 0.5, 0.6], index=1, object="embedding")]
    )

    # Call the method
    result = caching_handler._combine_cached_embedding_response_with_api_result(
        _caching_handler_response=caching_handler_response,
        embedding_response=api_response,
        start_time=start_time,
        end_time=end_time,
    )

    # Assertions
    assert isinstance(result, EmbeddingResponse)
    assert len(result.data) == 3
    assert result.data[0].embedding == [0.1, 0.2, 0.3]
    assert result.data[1].embedding == [0.4, 0.5, 0.6]
    assert result.data[2].embedding == [0.7, 0.8, 0.9]
    assert result._hidden_params["cache_hit"] == True
    assert isinstance(result._response_ms, float)
    assert result._response_ms > 0


def test_combine_cached_embedding_response_multiple_missing_values():
    """
    If the cached response has [cache_hit, None, None, cache_hit, None]
    result should be            [cache_hit, api_result, api_result, cache_hit, api_result]
    """

    # Setup
    caching_handler = LLMCachingHandler(
        original_function=lambda: None, request_kwargs={}, start_time=datetime.now()
    )

    start_time = datetime.now()
    end_time = start_time + timedelta(seconds=1)

    # Create a CachingHandlerResponse with some cached and some None values
    cached_response = EmbeddingResponse(
        data=[
            Embedding(embedding=[0.1, 0.2, 0.3], index=0, object="embedding"),
            None,
            None,
            Embedding(embedding=[0.7, 0.8, 0.9], index=3, object="embedding"),
            None,
        ]
    )

    caching_handler_response = CachingHandlerResponse(
        final_embedding_cached_response=cached_response
    )

    # Create an API EmbeddingResponse for the missing values
    api_response = EmbeddingResponse(
        data=[
            Embedding(embedding=[0.4, 0.5, 0.6], index=1, object="embedding"),
            Embedding(embedding=[0.4, 0.5, 0.6], index=2, object="embedding"),
            Embedding(embedding=[0.4, 0.5, 0.6], index=4, object="embedding"),
        ]
    )

    # Call the method
    result = caching_handler._combine_cached_embedding_response_with_api_result(
        _caching_handler_response=caching_handler_response,
        embedding_response=api_response,
        start_time=start_time,
        end_time=end_time,
    )

    # Assertions
    assert isinstance(result, EmbeddingResponse)
    assert len(result.data) == 5
    assert result.data[0].embedding == [0.1, 0.2, 0.3]
    assert result.data[1].embedding == [0.4, 0.5, 0.6]
    assert result.data[2].embedding == [0.4, 0.5, 0.6]
    assert result.data[3].embedding == [0.7, 0.8, 0.9]