fix(client): add support for streaming binary responses

examples/audio.py

@@ -13,7 +13,10 @@
 def main() -> None:
     # Create text-to-speech audio file
     response = openai.audio.speech.create(
-        model="tts-1", voice="alloy", input="the quick brown fox jumped over the lazy dogs"
+        model="tts-1",
+        voice="alloy",
+        input="the quick brown fox jumped over the lazy dogs",
+        stream=True,
     )
 
     response.stream_to_file(speech_file_path)

src/openai/_base_client.py

@@ -863,7 +863,7 @@ def _request(
         self._prepare_request(request)
 
         try:
-            response = self._client.send(request, auth=self.custom_auth, stream=stream)
+            response = self._client.send(request, auth=self.custom_auth, stream=stream or options.stream or False)
             log.debug(
                 'HTTP Request: %s %s "%i %s"', request.method, request.url, response.status_code, response.reason_phrase
             )
@@ -1304,7 +1304,7 @@ async def _request(
         await self._prepare_request(request)
 
         try:
-            response = await self._client.send(request, auth=self.custom_auth, stream=stream)
+            response = await self._client.send(request, auth=self.custom_auth, stream=stream or options.stream or False)
             log.debug(
                 'HTTP Request: %s %s "%i %s"', request.method, request.url, response.status_code, response.reason_phrase
             )
@@ -1541,6 +1541,7 @@ def make_request_options(
     idempotency_key: str | None = None,
     timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
     post_parser: PostParser | NotGiven = NOT_GIVEN,
+    stream: bool | None = None,
 ) -> RequestOptions:
     """Create a dict of type RequestOptions without keys of NotGiven values."""
     options: RequestOptions = {}
@@ -1562,6 +1563,9 @@ def make_request_options(
     if idempotency_key is not None:
         options["idempotency_key"] = idempotency_key
 
+    if stream is not None:
+        options["stream"] = stream
+
     if is_given(post_parser):
         # internal
         options["post_parser"] = post_parser  # type: ignore

src/openai/_models.py

@@ -403,6 +403,7 @@ class FinalRequestOptionsInput(TypedDict, total=False):
     params: Query
     headers: Headers
     max_retries: int
+    stream: bool | None
     timeout: float | Timeout | None
     files: HttpxRequestFiles | None
     idempotency_key: str
@@ -420,6 +421,7 @@ class FinalRequestOptions(pydantic.BaseModel):
     timeout: Union[float, Timeout, None, NotGiven] = NotGiven()
     files: Union[HttpxRequestFiles, None] = None
     idempotency_key: Union[str, None] = None
+    stream: Union[bool, None] = None
     post_parser: Union[Callable[[Any], Any], NotGiven] = NotGiven()
 
     # It should be noted that we cannot use `json` here as that would override

src/openai/_types.py

@@ -130,7 +130,16 @@ def stream_to_file(
         chunk_size: int | None = None,
     ) -> None:
         """
-        Stream the output to the given file.
+        Stream the output to the given file. NOTE, requires passing `stream=True`
+        to the request for expected behavior, e.g.,
+
+            response = openai.audio.speech.create(
+              model="tts-1",
+              voice="alloy",
+              input="the quick brown fox jumped over the lazy dogs",
+              stream=True,
+            )
+            response.stream_to_file(speech_file_path)
         """
         pass
 
@@ -185,7 +194,16 @@ async def astream_to_file(
         chunk_size: int | None = None,
     ) -> None:
         """
-        Stream the output to the given file.
+        Stream the output to the given file. NOTE, requires passing `stream=True`
+        to the request for expected behavior, e.g.,
+
+            response = await openai.audio.speech.create(
+              model="tts-1",
+              voice="alloy",
+              input="the quick brown fox jumped over the lazy dogs",
+              stream=True,
+            )
+            response.stream_to_file(speech_file_path)
         """
         pass
 
@@ -257,6 +275,7 @@ async def aclose(self) -> None:
 class RequestOptions(TypedDict, total=False):
     headers: Headers
     max_retries: int
+    stream: bool
     timeout: float | Timeout | None
     params: Query
     extra_json: AnyMapping

src/openai/resources/audio/speech.py

@@ -41,6 +41,7 @@ def create(
         extra_query: Query | None = None,
         extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+        stream: bool | None = None,
     ) -> HttpxBinaryResponseContent:
         """
         Generates audio from the input text.
@@ -67,6 +68,9 @@ def create(
           extra_body: Add additional JSON properties to the request
 
           timeout: Override the client-level default timeout for this request, in seconds
+
+          stream: Whether or not the response content should be streamed (i.e. not read to
+              completion immediately), default False
         """
         return self._post(
             "/audio/speech",
@@ -81,7 +85,11 @@ def create(
                 speech_create_params.SpeechCreateParams,
             ),
             options=make_request_options(
-                extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout
+                extra_headers=extra_headers,
+                extra_query=extra_query,
+                extra_body=extra_body,
+                timeout=timeout,
+                stream=stream,
             ),
             cast_to=HttpxBinaryResponseContent,
         )
@@ -108,6 +116,7 @@ async def create(
         extra_query: Query | None = None,
         extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+        stream: bool | None = None,
     ) -> HttpxBinaryResponseContent:
         """
         Generates audio from the input text.
@@ -134,6 +143,9 @@ async def create(
           extra_body: Add additional JSON properties to the request
 
           timeout: Override the client-level default timeout for this request, in seconds
+
+          stream: Whether or not the response content should be streamed (i.e. not read to
+              completion immediately), default False
         """
         return await self._post(
             "/audio/speech",
@@ -148,7 +160,11 @@ async def create(
                 speech_create_params.SpeechCreateParams,
             ),
             options=make_request_options(
-                extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout
+                extra_headers=extra_headers,
+                extra_query=extra_query,
+                extra_body=extra_body,
+                timeout=timeout,
+                stream=stream,
             ),
             cast_to=HttpxBinaryResponseContent,
         )

src/openai/resources/files.py

@@ -212,6 +212,7 @@ def content(
         extra_query: Query | None = None,
         extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+        stream: bool | None = None,
     ) -> HttpxBinaryResponseContent:
         """
         Returns the contents of the specified file.
@@ -224,11 +225,18 @@ def content(
           extra_body: Add additional JSON properties to the request
 
           timeout: Override the client-level default timeout for this request, in seconds
+
+          stream: Whether or not the response content should be streamed (i.e. not read to
+              completion immediately), default False
         """
         return self._get(
             f"/files/{file_id}/content",
             options=make_request_options(
-                extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout
+                extra_headers=extra_headers,
+                extra_query=extra_query,
+                extra_body=extra_body,
+                timeout=timeout,
+                stream=stream,
             ),
             cast_to=HttpxBinaryResponseContent,
         )
@@ -475,6 +483,7 @@ async def content(
         extra_query: Query | None = None,
         extra_body: Body | None = None,
         timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+        stream: bool | None = None,
     ) -> HttpxBinaryResponseContent:
         """
         Returns the contents of the specified file.
@@ -487,11 +496,18 @@ async def content(
           extra_body: Add additional JSON properties to the request
 
           timeout: Override the client-level default timeout for this request, in seconds
+
+          stream: Whether or not the response content should be streamed (i.e. not read to
+              completion immediately), default False
         """
         return await self._get(
             f"/files/{file_id}/content",
             options=make_request_options(
-                extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout
+                extra_headers=extra_headers,
+                extra_query=extra_query,
+                extra_body=extra_body,
+                timeout=timeout,
+                stream=stream,
             ),
             cast_to=HttpxBinaryResponseContent,
         )