Record response when streaming OpenAI Responses API (#1457)

Co-authored-by: Alex Hall <[email protected]>

Author: stevenh

logfire/_internal/integrations/llm_providers/openai.py

@@ -6,6 +6,7 @@
 
 import openai
 from openai._legacy_response import LegacyAPIResponse
+from openai.lib.streaming.responses import ResponseStreamState
 from openai.types.chat.chat_completion import ChatCompletion
 from openai.types.chat.chat_completion_chunk import ChatCompletionChunk
 from openai.types.completion import Completion
@@ -54,15 +55,21 @@ def get_endpoint_config(options: FinalRequestOptions) -> EndpointConfig:
         if is_current_agent_span('Responses API', 'Responses API with {gen_ai.request.model!r}'):
             return EndpointConfig(message_template='', span_data={})
 
+        span_data: dict[str, Any] = {
+            'gen_ai.request.model': json_data['model'],
+        }
+        if json_data.get('stream'):  # type: ignore
+            span_data['request_data'] = json_data
+        else:
+            span_data['events'] = inputs_to_events(
+                json_data['input'],  # type: ignore
+                json_data.get('instructions'),  # type: ignore
+            )
+
         return EndpointConfig(
             message_template='Responses API with {gen_ai.request.model!r}',
-            span_data={
-                'gen_ai.request.model': json_data['model'],
-                'events': inputs_to_events(
-                    json_data['input'],  # type: ignore
-                    json_data.get('instructions'),  # type: ignore
-                ),
-            },
+            span_data=span_data,
+            stream_state_cls=OpenaiResponsesStreamState,
         )
     elif url == '/completions':
         return EndpointConfig(
@@ -81,7 +88,7 @@ def get_endpoint_config(options: FinalRequestOptions) -> EndpointConfig:
             span_data={'request_data': json_data, 'gen_ai.request.model': json_data['model']},
         )
     else:
-        span_data: dict[str, Any] = {'request_data': json_data, 'url': url}
+        span_data = {'request_data': json_data, 'url': url}
         if 'model' in json_data:
             span_data['gen_ai.request.model'] = json_data['model']
         return EndpointConfig(
@@ -119,6 +126,21 @@ def get_response_data(self) -> Any:
         return {'combined_chunk_content': ''.join(self._content), 'chunk_count': len(self._content)}
 
 
+class OpenaiResponsesStreamState(StreamState):
+    def __init__(self):
+        self._state = ResponseStreamState(input_tools=openai.omit, text_format=openai.omit)
+
+    def record_chunk(self, chunk: Any) -> None:
+        self._state.handle_event(chunk)
+
+    def get_response_data(self) -> Any:
+        response = self._state._completed_response  # pyright: ignore[reportPrivateUsage]
+        if not response:  # pragma: no cover
+            raise RuntimeError("Didn't receive a `response.completed` event.")
+
+        return response
+
+
 try:
     # ChatCompletionStreamState only exists in openai>=1.40.0
     from openai.lib.streaming.chat._completions import ChatCompletionStreamState

tests/conftest.py

@@ -186,6 +186,37 @@ def multiple_credentials(tmp_path: Path) -> Path:
     return auth_file
 
 
+SENSITIVE_HEADERS = [
+    # All sensitive headers should be lower case as
+    # that's how they are presented by VCR.
+    'authorization',
+    'cookie',
+    'set-cookie',
+    'x-goog-api-key',
+    'openai-organization',
+    'openai-project',
+    'x-request-id',
+    'cf-ray',
+]
+
+
+def scrub_headers(response: dict[str, Any]) -> dict[str, Any]:
+    """Remove sensitive headers from the response.
+
+    Parameters:
+        response: The response dictionary to scrub.
+
+    Returns:
+        The scrubbed response dictionary with sensitive headers removed.
+    """
+    response['headers'] = {k: v for k, v in response.get('headers', {}).items() if k.lower() not in SENSITIVE_HEADERS}
+
+    return response
+
+
 @pytest.fixture(scope='module')
-def vcr_config():
-    return {'filter_headers': ['authorization', 'cookie', 'Set-Cookie', 'x-goog-api-key']}
+def vcr_config() -> dict[str, Any]:
+    return {
+        'filter_headers': SENSITIVE_HEADERS,
+        'before_record_response': scrub_headers,
+    }

tests/otel_integrations/cassettes/test_openai/test_responses_api.yaml

@@ -1,7 +1,7 @@
 interactions:
 - request:
-    body: "{\"input\":[{\"role\":\"user\",\"content\":\"What is the weather like in
-      Paris today?\"}],\"model\":\"gpt-4.1\",\"instructions\":\"Be nice\",\"tools\":[{\"type\":\"function\",\"name\":\"get_weather\",\"description\":\"Get
+    body: "{\"input\":\"What is the weather like in Paris today?\",\"instructions\":\"Be
+      nice\",\"model\":\"gpt-4.1\",\"tools\":[{\"type\":\"function\",\"name\":\"get_weather\",\"description\":\"Get
       current temperature for a given location.\",\"parameters\":{\"type\":\"object\",\"properties\":{\"location\":{\"type\":\"string\",\"description\":\"City
       and country e.g. Bogot\xE1, Colombia\"}},\"required\":[\"location\"],\"additionalProperties\":false}}]}"
     headers:
@@ -12,13 +12,13 @@ interactions:
       connection:
       - keep-alive
       content-length:
-      - '420'
+      - '392'
       content-type:
       - application/json
       host:
       - api.openai.com
       user-agent:
-      - OpenAI/Python 1.82.0
+      - OpenAI/Python 1.109.1
       x-stainless-arch:
       - arm64
       x-stainless-async:
@@ -28,52 +28,45 @@ interactions:
       x-stainless-os:
       - MacOS
       x-stainless-package-version:
-      - 1.82.0
+      - 1.109.1
       x-stainless-read-timeout:
       - '600'
       x-stainless-retry-count:
       - '0'
       x-stainless-runtime:
       - CPython
       x-stainless-runtime-version:
-      - 3.12.6
+      - 3.12.9
     method: POST
     uri: https://api.openai.com/v1/responses
   response:
     body:
       string: !!binary |
-        H4sIAAAAAAAAA3xUTY/bOAy951cIOmcCO8nk69gCbVH00BbYPXRTGIxEe7QjSy5FTTsY5L8vpCS2
-        k073Yth8ftTjI8WXiRDSaLkTkjB01WqzWJe1wsV9rTbltjysV/Viq7Eo14fFSi+Koi7WsEGlarXY
-        KpDTlMAf/kXFlyTeBTzFFSEw6goSVq6Xm+VyUWxWGQsMHEPiKN92Fhn1iXQA9diQjy6pqsEGzGEk
-        8iR3wkVrc8C4C7HSyGBsuEYDU1RsvMuHvEHhjDrrauFX5SN3kSv2j+iumK3XaBOl6fhuOSvv5sX8
-        /q5Y3pXLc7WZKXfin4kQQrzkZ29jrXoTVb1aJhO3uC22y3qt9Fxv5wd41cScgp87zEmiy9IrBdYO
-        8J88yyBQE1t0nPGXvbReQcqxl7u9/AxkwlS8I3AK9/I40NIJ1Ul6ft3OH8uPmy+PH+DH39/evfff
-        /irh09PX9cBw0GaRDXL1E4EfkGTGjhMhvmeLOiCwFm3F3ttcRFLFFE+97AifjI+huozLSUDfAkII
-        3hnXyN3ZXol17YlHPyU7YtsCPZ+DEyGOp8lCejIKKzaYBkZqrCFaluex84RjLYxthwQcc7icFefo
-        Lx4Orz21MHyPGpX/64s/nX+q+cGncdsJCZG97IHw+9zcNn1wWmNQZLoc3An5HlmoSISOxUi2qD0J
-        EI15QicuXZ/9f796MHWqRUYKo/JGos5XezogHfkOiQ1eM4ToJ+4mPsoWmFJXp9foTZlvDT8LcFoo
-        Hx3Ts8BZMxNvfON5H4sCy6l4661vDwbkKNGxfz+O1BL+iIZQ97bfau2D30cs0NokFOzncbV5G01u
-        DslF5e2XZurmIrDvqm48VxRdb5LUJsDBXhZfDNDgMHTGXe2n1f309/ho870MF1o9oB6IxWQkV96u
-        vXL9GvBa3v5S/ik1ewY7gJt5fyNiwKvV3SKDBoaU/jg5/gcAAP//AwBBmSDhhQYAAA==
+        H4sIAAAAAAAAA3xUy47jNhC8+ysInj2GpPh929lNckqwQA4BEi+EFtnSMEORWrLpXWPgfw9I2Xp4
+        Hzepi12s7i7224IxriQ/Mu7Qd2X2ywF3aym32yrPi0ORZdu9rDHPNsW+2u/zg9gXMi/y9f6wy/L1
+        Ybvmy0hhq/9Q0J3GGo99XDgEQllCxPLd5rDJis1hlzBPQMHHHGHbTiOh7JMqEK+Ns8FEXTVoj31Y
+        aa1Mw4/sbcEYY7yDC7qYL/GM2nbo+IKxazqMztmImaB1Cihzv6WUSKC0n6OeXBCkrEmKnpEZJW5F
+        tPC1tIG6QCXZVzSzzAiStboUoOecrZWoI1nT0dN6lT8VWbF5ytZP+b1piZMf2b+pnr6qYR61+Mk0
+        RFWv0zTkBqv9WkK12cC+PiTiREKXDhNNMKmspG+Ef9T8BIJrQouGEv524toKiBwnfjzxj+CUX7Lf
+        HBiBJ34d0+INZS8+fQal//nr1+Lzu/DHu+LPvwXtig/PX+R2zDDQJpENUvkFgV76ETJ2XTD2KTWp
+        Awdao543mVzoTdE5PCsbfHn3XS9gGELnbNtRKUC8YPmKlynmELw1M0thXVtHk0OxVaFtwd0zB4d5
+        qJEupZJoSNUKZ27z6M5KYEnq7tAagiZ+M751OC2CsO3QAYUUzlfZLfqVRmW1dS2M/5MJp3N9126K
+        z+gq6xVFzbxFqUI7voy+jy822vvIOASyfAD8t258NNI4PYleONWl4JHx35GYCM6hITapiNXWMWCN
+        OqNhdyetfu6BAYzTb5HQ+UnlE1G3vbMckc7FRUAK5xmMDS5+iE/YPLnohuUcfSjzvaILAyOZsMGQ
+        uzBcNSv2bBtLp5BlmC/Ze6ttWyngE6Lr8H2dqHX4OSiHcmj7o9Yh+GmSBVKqiIL+OK02rcrFwyWp
+        qLSao90eHhfZrtS26ZytIkE2BLupD10wQ+e4VB4qfV/VwUODo0mVmS3J7Wb5bXyyft/GzSFeUI6J
+        2czOj7s3330P+B7v8MJ/RE2WQI/gvhieSfDzF90igQSCSH9dXP8HAAD//wMAl3wlJTkHAAA=
     headers:
-      CF-RAY:
-      - 946e7e67de3ddf1a-JNB
       Connection:
       - keep-alive
       Content-Encoding:
       - gzip
       Content-Type:
       - application/json
       Date:
-      - Wed, 28 May 2025 14:38:07 GMT
+      - Fri, 03 Oct 2025 14:43:18 GMT
       Server:
       - cloudflare
-      Set-Cookie:
-      - __cf_bm=K.2f_Ec4Dafi9GQWf3x4es8GXSoEO_bgT7U4CjGi5Ss-1748443087-1.0.1.1-YZRtgk19Vn46L4lNNSLnubSabnBA3_L3.1k8qFpqzRLm1BGESaenng7i2EQdNzy4kbPOrmK5gDBpQ3bgN2gGYiUec5nJdiMoki35p9dG060;
-        path=/; expires=Wed, 28-May-25 15:08:07 GMT; domain=.api.openai.com; HttpOnly;
-        Secure; SameSite=None
-      - _cfuvid=XgEwvrVxz73nhFuqsHDjyuHQwVIWMuU6XgQEDg2dUbA-1748443087163-0.0.1.1-604800000;
-        path=/; domain=.api.openai.com; HttpOnly; Secure; SameSite=None
       Transfer-Encoding:
       - chunked
       X-Content-Type-Options:
@@ -82,34 +75,32 @@ interactions:
       - h3=":443"; ma=86400
       cf-cache-status:
       - DYNAMIC
-      openai-organization:
-      - pydantic-28gund
       openai-processing-ms:
-      - '855'
+      - '816'
       openai-version:
       - '2020-10-01'
       strict-transport-security:
       - max-age=31536000; includeSubDomains; preload
+      x-envoy-upstream-service-time:
+      - '819'
       x-ratelimit-limit-requests:
-      - '10000'
+      - '5000'
       x-ratelimit-limit-tokens:
-      - '2000000'
+      - '450000'
       x-ratelimit-remaining-requests:
-      - '9999'
+      - '4999'
       x-ratelimit-remaining-tokens:
-      - '1999718'
+      - '449718'
       x-ratelimit-reset-requests:
-      - 6ms
+      - 12ms
       x-ratelimit-reset-tokens:
-      - 8ms
-      x-request-id:
-      - req_82a8e571d6bacbdca1f76e0acf26f1bf
+      - 37ms
     status:
       code: 200
       message: OK
 - request:
     body: '{"input":[{"role":"user","content":"What is the weather like in Paris today?"},{"arguments":"{\"location\":\"Paris,
-      France\"}","call_id":"call_92k1J8QkHaqVZFGoZU1aLvR7","name":"get_weather","type":"function_call","id":"fc_68371fcecf6481919e9094f7cd2d92ba00f07a8eccfc39ca","status":"completed"},{"type":"function_call_output","call_id":"call_92k1J8QkHaqVZFGoZU1aLvR7","output":"Rainy"}],"model":"gpt-4.1"}'
+      France\"}","call_id":"call_uilZSE2qAuMA2NWct72DBwd6","name":"get_weather","type":"function_call","id":"fc_039e74dd66b112920068dfe105cbf4819c8d5eb84dab55a8f9","status":"completed"},{"type":"function_call_output","call_id":"call_uilZSE2qAuMA2NWct72DBwd6","output":"Rainy"}],"model":"gpt-4.1"}'
     headers:
       accept:
       - application/json
@@ -118,13 +109,13 @@ interactions:
       connection:
       - keep-alive
       content-length:
-      - '404'
+      - '406'
       content-type:
       - application/json
       host:
       - api.openai.com
       user-agent:
-      - OpenAI/Python 1.82.0
+      - OpenAI/Python 1.109.1
       x-stainless-arch:
       - arm64
       x-stainless-async:
@@ -134,43 +125,41 @@ interactions:
       x-stainless-os:
       - MacOS
       x-stainless-package-version:
-      - 1.82.0
+      - 1.109.1
       x-stainless-read-timeout:
       - '600'
       x-stainless-retry-count:
       - '0'
       x-stainless-runtime:
       - CPython
       x-stainless-runtime-version:
-      - 3.12.6
+      - 3.12.9
     method: POST
     uri: https://api.openai.com/v1/responses
   response:
     body:
       string: !!binary |
-        H4sIAAAAAAAAAwAAAP//dFRNb9pAEL3zK6Y+h8jGBgz/oFIPPfSWVtZ4PYZt1jvW7mwSFPHfK6+N
-        gTS5IHhv5u18vOF9AZDoJtlD4sj31abMt1mr2nVdpGW2y2qVteUG63WTt1tcrdK0TbdYklKtyncK
-        k4dBgOu/pOQiwtbTiCtHKNRUOHDZtiiLIk/LbeS8oAQ/5CjuekNCzZhUo3o+OA52qKpF4ynC5By7
-        ZA82GBMBbS+JVUOC2vh71osLSjTbO7zDt4qD9EEq4Wf6QHJDZijp0MuyeMyWq3S1XqbFMiumRmNm
-        soenBQDAe/ycJ9j5wzxAlY8DLDeb1SbdqfWqzVW92X46wKghp56iCnmPB7oSX00qkoqtkL2WdFvW
-        neylaXqTOTsGoLUseBnU0587MobvIfl1JHgllCM50BZ+otMehBs8gfbgUNvTI3xv4cThd1il2c4R
-        9Aat1fYAwnBg4CAP0LAdeYGW3YFkyHAQutqRMfgNfpBAR/Bs+RV01ANL1EDHjmDaM7ADOdKgQAp9
-        lIqApTeBll6hwZN/TOZGztO3ubfEsYlTQe+1F7QyBg+BMSjp0aExZCphNpVCE+0lLoxu7B29aA6+
-        uhi+ihaYneQIPQ/NJ/tpHQm1LTu5CRpWG7oO3WkCFwDn8TbIvWhFlWgaLJ801GIw494SL+zothah
-        rieHEiKcPaYTGjc3Pd6y6/D6+8YXMW5ufnx/7PnIWo1DCsLJTFxdkgj3VX/7pgtWRSvFqrXH2lzO
-        OkRTzwVpe3eCRf7wP35z13PZCtWRmmtiOpY+ZX+87CL9jPhMd17YV9LCguZKlvk8reDp7o+pI8EG
-        BQf58+L8DwAA//8DAGjwVuhjBQAA
+        H4sIAAAAAAAAA3RUy3LbMAy8+ytYXXKJM5Lf9h/01kNvmY4GIiGZDUVwSNCNJuN/74iyZKtNbhIW
+        WALYJT8WQmRaZSeReQyuzNdH3G+U2u2qolgdV3m+O6gai3x32FebQ3GUVV7J3fpQHKGSCtfb7Lmn
+        oOo3Sh5pyAYc4tIjMKoSeqzYb4/bfLU9HhIWGDiGvkZS6wwyqqGoAvnWeIq276sGE3AIa2O0bbKT
+        +FgIIUTmoEPf1yu8oCGHPlsIcU3J6D31mI3GpIC24ymlQgZtwhwN7KNkTXYWb+G9pMgucsn0hv+D
+        TGRKCWZO15JC03fWOF5uXorlKl9tl/lmWWxu+0qc2Um8plGGgSYp2tB8rcQ+X1eHXgmA/UrJvCgq
+        3Bcg14k5sXDnMPFgCNDgHfhq5QmUZBntvanHxma040LwnafqlADWEsO4xNdfM9BQ4zxVnyCJ6CSy
+        n2cUfxD4jF5oK36A10EwKeiEDsKDtt2L+F6LjuKTR+EMWKttI5hEQ4IiPwtF9olFTb5BFmBFbCuP
+        xsC3bDrwevuaesg8mTQXhKADg+UhuU9MSZkDD8agmavNPg7GdB4vmmIoR++XScbJDc5T67iUIM9Y
+        vmH3iHmEQHZma6xr8vyQ1AsX2xb8WDm5PECN3JVaoWVda5w5PqC/aIkl6/GW1BDNIFkWmDw+DsHY
+        OvTAMYWLl/wWTdLcOqvJt3D/f7BEyhu2duv4gr6ioLkbjKh0bO+3c9jjmbQcFh+Zsgm4OyRjcuWD
+        b/Ip6B579NHK5Lo0pQ5QmfEpicn/0wDazm7yZv38f/zheZjGTNKpe2E+G/XfB2JVfAZ8xjup/xU1
+        E4O5g7vNtMIY5mq3yKCAoae/Lq5/AQAA//8DAInpJWzZBQAA
     headers:
-      CF-RAY:
-      - 946e7e6f1a1fdf1a-JNB
       Connection:
       - keep-alive
       Content-Encoding:
       - gzip
       Content-Type:
       - application/json
       Date:
-      - Wed, 28 May 2025 14:38:08 GMT
+      - Fri, 03 Oct 2025 14:43:19 GMT
       Server:
       - cloudflare
       Transfer-Encoding:
@@ -181,28 +170,26 @@ interactions:
       - h3=":443"; ma=86400
       cf-cache-status:
       - DYNAMIC
-      openai-organization:
-      - pydantic-28gund
       openai-processing-ms:
-      - '1276'
+      - '1023'
       openai-version:
       - '2020-10-01'
       strict-transport-security:
       - max-age=31536000; includeSubDomains; preload
+      x-envoy-upstream-service-time:
+      - '1025'
       x-ratelimit-limit-requests:
-      - '10000'
+      - '5000'
       x-ratelimit-limit-tokens:
-      - '2000000'
+      - '450000'
       x-ratelimit-remaining-requests:
-      - '9999'
+      - '4999'
       x-ratelimit-remaining-tokens:
-      - '1999938'
+      - '449937'
       x-ratelimit-reset-requests:
-      - 6ms
+      - 12ms
       x-ratelimit-reset-tokens:
-      - 1ms
-      x-request-id:
-      - req_4a16ad9f9da83f8def54db828cce54ff
+      - 8ms
     status:
       code: 200
       message: OK