Allow using custom LLM providers in chat APIs

.env.example

@@ -1,3 +1,4 @@
 QSTASH_TOKEN="YOUR_TOKEN"
 QSTASH_CURRENT_SIGNING_KEY="<YOUR_CURRENT_SIGNING_KEY>"
-QSTASH_NEXT_SIGNING_KEY="<YOUR_NEXT_SIGNING_KEY>"
+QSTASH_NEXT_SIGNING_KEY="<YOUR_NEXT_SIGNING_KEY>"
+OPENAI_API_KEY = "<YOUR_OPENAI_API_KEY>"

.github/workflows/tests.yml

@@ -47,4 +47,5 @@ jobs:
           export QSTASH_TOKEN="${{ secrets.QSTASH_TOKEN }}"
           export QSTASH_CURRENT_SIGNING_KEY="${{ secrets.QSTASH_CURRENT_SIGNING_KEY }}"
           export QSTASH_NEXT_SIGNING_KEY="${{ secrets.QSTASH_NEXT_SIGNING_KEY }}"
-          poetry run pytest
+          export OPENAI_API_KEY="${{ secrets.OPENAI_API_KEY }}"
+          poetry run pytest

README.md

@@ -3,7 +3,7 @@
 > [!NOTE]  
 > **This project is in GA Stage.**
 >
-> The Upstash Professional Support fully covers this project. It receives regular updates, and bug fixes. 
+> The Upstash Professional Support fully covers this project. It receives regular updates, and bug fixes.
 > The Upstash team is committed to maintaining and improving its functionality.
 
 **QStash** is an HTTP based messaging and scheduling solution for serverless and edge runtimes.
@@ -73,11 +73,13 @@ receiver.verify(
 
 ```python
 from upstash_qstash import QStash
+from upstash_qstash.chat import upstash
 
 qstash = QStash("<QSTASH_TOKEN>")
 
 res = qstash.chat.create(
     model="meta-llama/Meta-Llama-3-8B-Instruct",
+    provider=upstash(),
     messages=[
         {
             "role": "user",
@@ -89,6 +91,78 @@ res = qstash.chat.create(
 print(res.choices[0].message.content)
 ```
 
+#### Create Chat Completions Using Custom Providers
+
+```python
+from upstash_qstash import QStash
+from upstash_qstash.chat import openai
+
+qstash = QStash("<QSTASH_TOKEN>")
+
+res = qstash.chat.create(
+    model="gpt-3.5-turbo",
+    provider=openai("<OPENAI_API_KEY>"),
+    messages=[
+        {
+            "role": "user",
+            "content": "What is the capital of Turkey?",
+        }
+    ],
+)
+
+print(res.choices[0].message.content)
+```
+
+#### Publish a JSON message to LLM
+
+```python
+from upstash_qstash import QStash
+from upstash_qstash.chat import upstash
+
+qstash = QStash("<QSTASH_TOKEN>")
+
+res = qstash.message.publish_json(
+    api={"name": "llm", "provider": upstash()},
+    body={
+        "model": "meta-llama/Meta-Llama-3-8B-Instruct",
+        "messages": [
+            {
+                "role": "user",
+                "content": "What is the capital of Turkey?",
+            }
+        ],
+    },
+    callback="https://example-cb.com",
+)
+
+print(res.message_id)
+```
+
+#### Publish a JSON message to LLM Using Custom Providers
+
+```python
+from upstash_qstash import QStash
+from upstash_qstash.chat import openai
+
+qstash = QStash("<QSTASH_TOKEN>")
+
+res = qstash.message.publish_json(
+    api={"name": "llm", "provider": openai("<OPENAI_API_KEY>")},
+    body={
+        "model": "gpt-3.5-turbo",
+        "messages": [
+            {
+                "role": "user",
+                "content": "What is the capital of Turkey?",
+            }
+        ],
+    },
+    callback="https://example-cb.com",
+)
+
+print(res.message_id)
+```
+
 #### Additional configuration
 
 ```python

examples/llm.py

@@ -6,6 +6,7 @@
 """
 
 from upstash_qstash import QStash
+from upstash_qstash.chat import upstash
 
 
 def main():
@@ -14,7 +15,7 @@ def main():
     )
 
     qstash.message.publish_json(
-        api="llm",
+        api={"name": "llm", "provider": upstash()},
         body={
             "model": "meta-llama/Meta-Llama-3-8B-Instruct",
             "messages": [

tests/__init__.py

@@ -20,6 +20,11 @@
     dotenv.dotenv_values().get("QSTASH_NEXT_SIGNING_KEY"),
 )
 
+OPENAI_API_KEY = os.environ.get(
+    "OPENAI_API_KEY",
+    dotenv.dotenv_values().get("OPENAI_API_KEY"),
+)
+
 
 def assert_eventually(
     assertion: Callable[[], None],

tests/asyncio/test_chat.py

@@ -1,14 +1,16 @@
 import pytest
+
+from tests import OPENAI_API_KEY
 from upstash_qstash import AsyncQStash
 from upstash_qstash.asyncio.chat import AsyncChatCompletionChunkStream
-from upstash_qstash.chat import ChatCompletion
+from upstash_qstash.chat import ChatCompletion, upstash, openai
 
 
 @pytest.mark.asyncio
 async def test_chat_async(async_qstash: AsyncQStash) -> None:
     res = await async_qstash.chat.create(
         model="meta-llama/Meta-Llama-3-8B-Instruct",
-        messages=[{"role": "user", "content": "hello"}],
+        messages=[{"role": "user", "content": "just say hello"}],
     )
 
     assert isinstance(res, ChatCompletion)
@@ -21,7 +23,7 @@ async def test_chat_async(async_qstash: AsyncQStash) -> None:
 async def test_chat_streaming_async(async_qstash: AsyncQStash) -> None:
     res = await async_qstash.chat.create(
         model="meta-llama/Meta-Llama-3-8B-Instruct",
-        messages=[{"role": "user", "content": "hello"}],
+        messages=[{"role": "user", "content": "just say hello"}],
         stream=True,
     )
 
@@ -41,7 +43,7 @@ async def test_chat_streaming_async(async_qstash: AsyncQStash) -> None:
 async def test_prompt_async(async_qstash: AsyncQStash) -> None:
     res = await async_qstash.chat.prompt(
         model="meta-llama/Meta-Llama-3-8B-Instruct",
-        user="hello",
+        user="just say hello",
     )
 
     assert isinstance(res, ChatCompletion)
@@ -54,7 +56,44 @@ async def test_prompt_async(async_qstash: AsyncQStash) -> None:
 async def test_prompt_streaming_async(async_qstash: AsyncQStash) -> None:
     res = await async_qstash.chat.prompt(
         model="meta-llama/Meta-Llama-3-8B-Instruct",
-        user="hello",
+        user="just say hello",
+        stream=True,
+    )
+
+    assert isinstance(res, AsyncChatCompletionChunkStream)
+
+    i = 0
+    async for r in res:
+        if i == 0:
+            assert r.choices[0].delta.role is not None
+        else:
+            assert r.choices[0].delta.content is not None
+
+        i += 1
+
+
+@pytest.mark.asyncio
+async def test_chat_explicit_upstash_provider_async(async_qstash: AsyncQStash) -> None:
+    res = await async_qstash.chat.create(
+        model="meta-llama/Meta-Llama-3-8B-Instruct",
+        messages=[{"role": "user", "content": "just say hello"}],
+        provider=upstash(),
+    )
+
+    assert isinstance(res, ChatCompletion)
+
+    assert len(res.choices[0].message.content) > 0
+    assert res.choices[0].message.role == "assistant"
+
+
+@pytest.mark.asyncio
+async def test_chat_explicit_upstash_provider_streaming_async(
+    async_qstash: AsyncQStash,
+) -> None:
+    res = await async_qstash.chat.create(
+        model="meta-llama/Meta-Llama-3-8B-Instruct",
+        messages=[{"role": "user", "content": "just say hello"}],
+        provider=upstash(),
         stream=True,
     )
 
@@ -68,3 +107,120 @@ async def test_prompt_streaming_async(async_qstash: AsyncQStash) -> None:
             assert r.choices[0].delta.content is not None
 
         i += 1
+
+
+@pytest.mark.asyncio
+async def test_prompt_explicit_upstash_provider_async(
+    async_qstash: AsyncQStash,
+) -> None:
+    res = await async_qstash.chat.prompt(
+        model="meta-llama/Meta-Llama-3-8B-Instruct",
+        user="just say hello",
+        provider=upstash(),
+    )
+
+    assert isinstance(res, ChatCompletion)
+
+    assert len(res.choices[0].message.content) > 0
+    assert res.choices[0].message.role == "assistant"
+
+
+@pytest.mark.asyncio
+async def test_prompt_explicit_upstash_provider_streaming_async(
+    async_qstash: AsyncQStash,
+) -> None:
+    res = await async_qstash.chat.prompt(
+        model="meta-llama/Meta-Llama-3-8B-Instruct",
+        user="just say hello",
+        provider=upstash(),
+        stream=True,
+    )
+
+    assert isinstance(res, AsyncChatCompletionChunkStream)
+
+    i = 0
+    async for r in res:
+        if i == 0:
+            assert r.choices[0].delta.role is not None
+        else:
+            assert r.choices[0].delta.content is not None
+
+        i += 1
+
+
+@pytest.mark.asyncio
+async def test_chat_custom_provider_async(async_qstash: AsyncQStash) -> None:
+    res = await async_qstash.chat.create(
+        model="gpt-3.5-turbo",
+        messages=[{"role": "user", "content": "just say hello"}],
+        provider=openai(token=OPENAI_API_KEY),  # type:ignore[arg-type]
+    )
+
+    assert isinstance(res, ChatCompletion)
+
+    assert len(res.choices[0].message.content) > 0
+    assert res.choices[0].message.role == "assistant"
+
+
+@pytest.mark.asyncio
+async def test_chat_custom_provider_streaming_async(async_qstash: AsyncQStash) -> None:
+    res = await async_qstash.chat.create(
+        model="gpt-3.5-turbo",
+        messages=[{"role": "user", "content": "just say hello"}],
+        provider=openai(token=OPENAI_API_KEY),  # type:ignore[arg-type]
+        stream=True,
+    )
+
+    assert isinstance(res, AsyncChatCompletionChunkStream)
+
+    i = 0
+    async for r in res:
+        if i == 0:
+            assert r.choices[0].delta.role is not None
+        else:
+            assert (
+                r.choices[0].delta.content is not None
+                or r.choices[0].finish_reason is not None
+            )
+
+        i += 1
+
+
+@pytest.mark.asyncio
+async def test_prompt_custom_provider_async(async_qstash: AsyncQStash) -> None:
+    res = await async_qstash.chat.prompt(
+        model="gpt-3.5-turbo",
+        user="just say hello",
+        provider=openai(token=OPENAI_API_KEY),  # type:ignore[arg-type]
+    )
+
+    assert isinstance(res, ChatCompletion)
+
+    assert len(res.choices[0].message.content) > 0
+    assert res.choices[0].message.role == "assistant"
+
+
+@pytest.mark.asyncio
+async def test_prompt_custom_provider_streaming_async(
+    async_qstash: AsyncQStash,
+) -> None:
+    res = await async_qstash.chat.prompt(
+        model="gpt-3.5-turbo",
+        user="just say hello",
+        provider=openai(token=OPENAI_API_KEY),  # type:ignore[arg-type]
+        stream=True,
+    )
+
+    assert isinstance(res, AsyncChatCompletionChunkStream)
+
+    i = 0
+    async for r in res:
+        if i == 0:
+            assert r.choices[0].delta.role is not None
+        else:
+            assert (
+                r.choices[0].delta.content is not None
+                or r.choices[0].finish_reason is not None
+            )
+
+        i += 1