openaivec 0.13.0__py3-none-any.whl → 0.13.1__py3-none-any.whl

openaivec/embeddings.py

@@ -4,7 +4,7 @@ from typing import List
 
 import numpy as np
 from numpy.typing import NDArray
-from openai import AsyncOpenAI, OpenAI, RateLimitError
+from openai import AsyncOpenAI, InternalServerError, OpenAI, RateLimitError
 
 from .log import observe
 from .proxy import AsyncBatchingMapProxy, BatchingMapProxy
@@ -47,7 +47,7 @@ class BatchEmbeddings:
         return cls(client=client, model_name=model_name, cache=BatchingMapProxy(batch_size=batch_size))
 
     @observe(_LOGGER)
-    @backoff(exception=RateLimitError, scale=15, max_retries=8)
+    @backoff(exceptions=[RateLimitError, InternalServerError], scale=1, max_retries=12)
     def _embed_chunk(self, inputs: List[str]) -> List[NDArray[np.float32]]:
         """Embed one minibatch of strings.
 
@@ -155,7 +155,7 @@ class AsyncBatchEmbeddings:
         )
 
     @observe(_LOGGER)
-    @backoff_async(exception=RateLimitError, scale=15, max_retries=8)
+    @backoff_async(exceptions=[RateLimitError, InternalServerError], scale=1, max_retries=12)
     async def _embed_chunk(self, inputs: List[str]) -> List[NDArray[np.float32]]:
         """Embed one minibatch of strings asynchronously.
 

openaivec/responses.py

@@ -2,7 +2,7 @@ from dataclasses import dataclass, field
 from logging import Logger, getLogger
 from typing import Generic, List, Type, cast
 
-from openai import AsyncOpenAI, OpenAI, RateLimitError
+from openai import AsyncOpenAI, InternalServerError, OpenAI, RateLimitError
 from openai.types.responses import ParsedResponse
 from pydantic import BaseModel
 
@@ -206,7 +206,7 @@ class BatchResponses(Generic[ResponseFormat]):
         )
 
     @observe(_LOGGER)
-    @backoff(exception=RateLimitError, scale=15, max_retries=8)
+    @backoff(exceptions=[RateLimitError, InternalServerError], scale=1, max_retries=12)
     def _request_llm(self, user_messages: List[Message[str]]) -> ParsedResponse[Response[ResponseFormat]]:
         """Make a single call to the OpenAI JSON‑mode endpoint.
 
@@ -400,7 +400,7 @@ class AsyncBatchResponses(Generic[ResponseFormat]):
         )
 
     @observe(_LOGGER)
-    @backoff_async(exception=RateLimitError, scale=15, max_retries=8)
+    @backoff_async(exceptions=[RateLimitError, InternalServerError], scale=1, max_retries=12)
     async def _request_llm(self, user_messages: List[Message[str]]) -> ParsedResponse[Response[ResponseFormat]]:
         """Make a single async call to the OpenAI JSON‑mode endpoint.
 

openaivec/util.py

@@ -3,7 +3,7 @@ import functools
 import re
 import time
 from dataclasses import dataclass
-from typing import Awaitable, Callable, List, TypeVar
+from typing import Awaitable, Callable, List, Type, TypeVar
 
 import numpy as np
 import tiktoken
@@ -34,24 +34,28 @@ def get_exponential_with_cutoff(scale: float) -> float:
             return v
 
 
-def backoff(exception: type[Exception], scale: int | None = None, max_retries: int | None = None) -> Callable[..., V]:
+def backoff(
+    exceptions: List[Type[Exception]],
+    scale: int | None = None,
+    max_retries: int | None = None,
+) -> Callable[..., V]:
     """Decorator implementing exponential back‑off retry logic.
 
     Args:
-        exception (type[Exception]): Exception type that triggers a retry.
+        exceptions (List[Type[Exception]]): List of exception types that trigger a retry.
         scale (int | None): Initial scale parameter for the exponential jitter.
             This scale is used as the mean for the first delay's exponential
             distribution and doubles with each subsequent retry. If ``None``,
             an initial scale of 1.0 is used.
-        max_retries (Optional[int]): Maximum number of retries. ``None`` means
+        max_retries (int | None): Maximum number of retries. ``None`` means
             retry indefinitely.
 
     Returns:
         Callable[..., V]: A decorated function that retries on the specified
-            exception with exponential back‑off.
+            exceptions with exponential back‑off.
 
     Raises:
-        exception: Re‑raised when the maximum number of retries is exceeded.
+        Exception: Re‑raised when the maximum number of retries is exceeded.
     """
 
     def decorator(func: Callable[..., V]) -> Callable[..., V]:
@@ -65,7 +69,7 @@ def backoff(exception: type[Exception], scale: int | None = None, max_retries: i
             while True:
                 try:
                     return func(*args, **kwargs)
-                except exception:
+                except tuple(exceptions):
                     attempt += 1
                     if max_retries is not None and attempt >= max_retries:
                         raise
@@ -83,12 +87,14 @@ def backoff(exception: type[Exception], scale: int | None = None, max_retries: i
 
 
 def backoff_async(
-    exception: type[Exception], scale: int | None = None, max_retries: int | None = None
+    exceptions: List[Type[Exception]],
+    scale: int | None = None,
+    max_retries: int | None = None,
 ) -> Callable[..., Awaitable[V]]:
     """Asynchronous version of the backoff decorator.
 
     Args:
-        exception (type[Exception]): Exception type that triggers a retry.
+        exceptions (List[Type[Exception]]): List of exception types that trigger a retry.
         scale (int | None): Initial scale parameter for the exponential jitter.
             This scale is used as the mean for the first delay's exponential
             distribution and doubles with each subsequent retry. If ``None``,
@@ -98,10 +104,10 @@ def backoff_async(
 
     Returns:
         Callable[..., Awaitable[V]]: A decorated asynchronous function that
-            retries on the specified exception with exponential back‑off.
+            retries on the specified exceptions with exponential back‑off.
 
     Raises:
-        exception: Re‑raised when the maximum number of retries is exceeded.
+        Exception: Re‑raised when the maximum number of retries is exceeded.
     """
 
     def decorator(func: Callable[..., Awaitable[V]]) -> Callable[..., Awaitable[V]]:
@@ -115,7 +121,7 @@ def backoff_async(
             while True:
                 try:
                     return await func(*args, **kwargs)
-                except exception:
+                except tuple(exceptions):
                     attempt += 1
                     if max_retries is not None and attempt >= max_retries:
                         raise

{openaivec-0.13.0.dist-info → openaivec-0.13.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: openaivec
-Version: 0.13.0
+Version: 0.13.1
 Summary: Generative mutation for tabular calculation
 Project-URL: Homepage, https://microsoft.github.io/openaivec/
 Project-URL: Repository, https://github.com/microsoft/openaivec

{openaivec-0.13.0.dist-info → openaivec-0.13.1.dist-info}/RECORD

@@ -1,16 +1,16 @@
 openaivec/__init__.py,sha256=CuUAtLtX5RhFUbgF94UmXjurgL2VaerHTFPjMl0rRlE,236
 openaivec/di.py,sha256=RSAiS9PqtoeobX-MZvAN8VIgaQa5EUMysj9aJwGuM9Y,10646
-openaivec/embeddings.py,sha256=zY2_jLXSOXXWzWQlXomC91fwARo8U3L4qAXBO4TaKQc,6881
+openaivec/embeddings.py,sha256=aAAzGAPYg5RybY95-ZHQdnxP12L4XPEJ3tHqqOAkcNY,6950
 openaivec/log.py,sha256=GofgzUpv_xDVuGC-gYmit5Oyu06it1SBXRck6COR5go,1439
 openaivec/model.py,sha256=cXgXMceQpTR46xdn9_9ZQKdLcy6X0i7s1x2lxXJOqyU,2434
 openaivec/pandas_ext.py,sha256=g3fpyCupPMOG4OR60uemySlWfIAsoEO5P15tqxpXkd0,51029
 openaivec/prompt.py,sha256=p_AyCLIhlrxp7JXLoywPuDuBiOVmhuAp0XWlKcRlAGY,17715
 openaivec/provider.py,sha256=2MDI9oyOK-abES5bqa_RS4fXX3rdVu6Wph69KHPFIE0,4077
 openaivec/proxy.py,sha256=36d5Rb69HOhENBpS0xzx0eeJZuIfFBzKH6vEyurK-CA,24337
-openaivec/responses.py,sha256=lPVC0BH2rg8CVvub0QlNNotdwO2QtDq1TfxiCTSE0Tw,17623
+openaivec/responses.py,sha256=335OgA4tFp_4jJf4fOIAx6LbyKIzvQ2pO3IuExheifw,17692
 openaivec/serialize.py,sha256=YbYOAxhw67ELcq48wCCelqp7J4vX42liyIWdl9pjJwU,7325
 openaivec/spark.py,sha256=3cwizW0zjbqA7hb8vmT2POzYlNGNoAp2mnEvcHsmLzg,22855
-openaivec/util.py,sha256=05DXh7ToIg0pt-Z_4nint74ifhMrzGLIdHFjLS-_Bp4,6261
+openaivec/util.py,sha256=cyiATh_IGnkMU0IBPcLpFMUHjjFB6TjLAug5TdC3vS8,6350
 openaivec/task/__init__.py,sha256=26Kd2eRRfUqRJnYH83sRrbuEuRwshypRRSkTWtT50Tw,6177
 openaivec/task/customer_support/__init__.py,sha256=p_OPWdONia_62F5PS-w83-Rngql_4r8Xhn0p2ArzSUk,1067
 openaivec/task/customer_support/customer_sentiment.py,sha256=2DdJHmG5OcLGC-u12F4b2EaFhy5jEjYunCHU2OUrSus,7587
@@ -28,7 +28,7 @@ openaivec/task/nlp/sentiment_analysis.py,sha256=-P56GL_krrthu2g6oQiQ7MeA9QR8-m76
 openaivec/task/nlp/translation.py,sha256=bp_yr4_0_CwrX9Lw3UQMXdmlXBkc6hsAapAEZudCK00,6618
 openaivec/task/table/__init__.py,sha256=qJ-oPcCUZzn2la_NFO1eGRUTNHgTO2jCR4L5xHp6qpM,83
 openaivec/task/table/fillna.py,sha256=ZKYfcweiA6ODGna986pxG9RXYJZthbfp5XXOeT-WUjg,6557
-openaivec-0.13.0.dist-info/METADATA,sha256=6nwmPUNHzW2hGtfernqomBY4zQyfvz-Y5Saow1C3Jns,26015
-openaivec-0.13.0.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-openaivec-0.13.0.dist-info/licenses/LICENSE,sha256=ws_MuBL-SCEBqPBFl9_FqZkaaydIJmxHrJG2parhU4M,1141
-openaivec-0.13.0.dist-info/RECORD,,
+openaivec-0.13.1.dist-info/METADATA,sha256=yr7unxTrWeoKYJX1fgQPfUmbAEDRYMFgQvtvamsJA8Y,26015
+openaivec-0.13.1.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+openaivec-0.13.1.dist-info/licenses/LICENSE,sha256=ws_MuBL-SCEBqPBFl9_FqZkaaydIJmxHrJG2parhU4M,1141
+openaivec-0.13.1.dist-info/RECORD,,