sunholo 0.96.3__tar.gz → 0.96.5__tar.gz

{sunholo-0.96.3 → sunholo-0.96.5}/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: sunholo
-Version: 0.96.3
+Version: 0.96.5
 Summary: Large Language Model DevOps - a package to help deploy LLMs to the Cloud.
 Home-page: https://github.com/sunholo-data/sunholo-py
-Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.96.3.tar.gz
+Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.96.5.tar.gz
 Author: Holosun ApS
 Author-email: multivac@sunholo.com
 License: Apache License, Version 2.0

{sunholo-0.96.3 → sunholo-0.96.5}/setup.py

@@ -1,6 +1,6 @@
 from setuptools import setup, find_packages
 
-version = '0.96.3'
+version = '0.96.5'
 
 setup(
     name='sunholo',

sunholo-0.96.5/sunholo/invoke/direct_vac_func.py

@@ -0,0 +1,279 @@
+from ..custom_logging import log
+from ..agents import send_to_qa, send_to_qa_async
+from ..qna.parsers import parse_output
+from ..streaming import generate_proxy_stream
+from ..utils import ConfigManager
+from ..utils.api_key import has_multivac_api_key
+import asyncio
+from threading import Thread
+
+def direct_vac(vac_input: dict, vac_name: str, chat_history=[]):
+    """
+    This lets VACs call other VAC Q&A endpoints within their code
+    """
+
+    log.info(f"Invoking VAC Q&A endpoints for {vac_name}")
+
+    if 'user_input' not in vac_input:
+        raise ValueError(f'vac_input must contain at least "user_input" key - got {vac_input}')
+
+    global_config = ConfigManager('global')
+    config = ConfigManager(vac_name)
+
+    agent_name = config.vacConfig('agent')
+    agent_url = config.vacConfig("agent_url")
+
+    if agent_url:
+        log.info("Found agent_url within vacConfig: {agent_url}")
+    # via public cloud endpoints - assumes no gcloud auth
+    override_endpoint = None
+    if has_multivac_api_key():
+        print("Found MULTIVAC_API_KEY")
+        gcp_config = global_config.vacConfig("gcp_config")
+        endpoints_base_url = gcp_config.get("endpoints_base_url")
+        if not endpoints_base_url:
+            raise ValueError("MULTIVAC_API_KEY env var is set but no config.gcp_config.endpoints_base_url can be found")
+
+        override_endpoint = f"{endpoints_base_url}/v1/{agent_name}"
+
+    override_endpoint = agent_url or override_endpoint
+
+    print(f"Using {override_endpoint=}")
+    
+    # Prepare the kwargs for send_to_qa by copying vac_input and adding more values
+    qa_kwargs = vac_input.copy()
+
+    # Add additional arguments
+    qa_kwargs.update({
+        'vector_name': vac_name,
+        'chat_history': chat_history,
+        'image_url': vac_input.get('image_url') or vac_input.get('image_uri'),
+        'override_endpoint': override_endpoint,
+        'message_source': "sunholo.invoke_vac_qa.invoke",
+        'stream': False,
+        'configurable': {
+            "vector_name": vac_name,
+        },
+    })
+
+    log.info(f'Batch invoke_vac_qa {vac_name} with {qa_kwargs=}')
+
+    vac_response = send_to_qa(**qa_kwargs)
+        
+    # ensures {'answer': answer}
+    answer = parse_output(vac_response)
+    chat_history.append({"name": "Human", "content": vac_input})
+    chat_history.append({"name": "AI", "content": answer})
+    answer["chat_history"] = chat_history
+    
+    return answer
+
+async def async_direct_vac(vac_input: dict, vac_name: str, chat_history=[]):
+    """
+    Asynchronous version of direct_vac using send_to_qa_async.
+    Allows VACs to call other VAC Q&A endpoints without blocking the event loop.
+    """
+    log.info(f"Invoking VAC Q&A endpoints for {vac_name}")
+
+    if 'user_input' not in vac_input:
+        raise ValueError(f'vac_input must contain at least "user_input" key - got {vac_input}')
+
+    global_config = ConfigManager('global')
+    config = ConfigManager(vac_name)
+
+    agent_name = config.vacConfig('agent')
+    agent_url = config.vacConfig("agent_url")
+
+    if agent_url:
+        log.info(f"Found agent_url within vacConfig: {agent_url}")
+
+    # Via public cloud endpoints - assumes no gcloud auth
+    override_endpoint = None
+    if has_multivac_api_key():
+        print("Found MULTIVAC_API_KEY")
+        gcp_config = global_config.vacConfig("gcp_config")
+        endpoints_base_url = gcp_config.get("endpoints_base_url")
+        if not endpoints_base_url:
+            raise ValueError("MULTIVAC_API_KEY env var is set but no config.gcp_config.endpoints_base_url can be found")
+
+        override_endpoint = f"{endpoints_base_url}/v1/{agent_name}"
+
+    override_endpoint = agent_url or override_endpoint
+
+    print(f"Using override_endpoint={override_endpoint}")
+
+    # Prepare the kwargs for send_to_qa_async by copying vac_input and adding more values
+    qa_kwargs = vac_input.copy()
+
+    # Add additional arguments
+    qa_kwargs.update({
+        'vector_name': vac_name,
+        'chat_history': chat_history,
+        'image_url': vac_input.get('image_url') or vac_input.get('image_uri'),
+        'override_endpoint': override_endpoint,
+        'message_source': "sunholo.invoke_vac_qa.invoke",
+        'stream': False,
+        'configurable': {
+            "vector_name": vac_name,
+        },
+    })
+
+    log.info(f'Batch invoke_vac_qa {vac_name} with qa_kwargs={qa_kwargs}')
+
+    # Call send_to_qa_async directly
+    vac_response_generator = send_to_qa_async(**qa_kwargs)
+
+    # Since send_to_qa_async returns an async generator, we can get the response
+    vac_response = None
+    async for response in vac_response_generator:
+        vac_response = response  # Since stream=False, we expect only one response
+        break
+
+    # Call parse_output synchronously (since it's non-blocking)
+    answer = parse_output(vac_response)
+
+    chat_history.append({"name": "Human", "content": vac_input})
+    chat_history.append({"name": "AI", "content": answer})
+    answer["chat_history"] = chat_history
+
+    return answer
+
+def direct_vac_stream(vac_input: dict, vac_name: str, chat_history=[]):
+
+    if 'user_input' not in vac_input:
+        raise ValueError('vac_input must contain at least "user_input" key - got {vac_input}')
+
+    user_id = vac_input.get('user_id')
+    session_id = vac_input.get('session_id')
+    image_uri = vac_input.get('image_url') or vac_input.get('image_uri')
+    
+    log.info(f"Streaming invoke_vac_qa with {vac_input=}")
+    def stream_response():
+        generate = generate_proxy_stream(
+                send_to_qa,
+                vac_input["user_input"],
+                vector_name=vac_name,
+                chat_history=chat_history,
+                generate_f_output=lambda x: x,  # Replace with actual processing function
+                stream_wait_time=0.5,
+                stream_timeout=120,
+                message_author=user_id,
+                #TODO: populate these
+                image_url=image_uri,
+                source_filters=None,
+                search_kwargs=None,
+                private_docs=None,
+                whole_document=False,
+                source_filters_and_or=False,
+                # system kwargs
+                configurable={
+                    "vector_name": vac_name,
+                },
+                user_id=user_id,
+                session_id=session_id, 
+                message_source="sunholo.invoke_vac_qa.stream"
+        )
+        for part in generate():
+            yield part
+
+    answer = ""
+
+    for token in stream_response():
+        if isinstance(token, bytes):
+            token = token.decode('utf-8')
+            yield token
+        if isinstance(token, dict):
+            # ?
+            pass
+        elif isinstance(token, str):
+            answer += token
+
+    if answer:
+        chat_history.append({"name": "Human", "content": vac_input})
+        chat_history.append({"name": "AI", "content": answer})
+
+    return chat_history
+
+
+
+async def async_direct_vac_stream(vac_input: dict, vac_name: str, chat_history=[]):
+    """
+    Asynchronous version of direct_vac_stream.
+    Streams responses from VAC Q&A endpoints without blocking the event loop.
+    """
+    if 'user_input' not in vac_input:
+        raise ValueError(f'vac_input must contain at least "user_input" key - got {vac_input}')
+
+    user_id = vac_input.get('user_id')
+    session_id = vac_input.get('session_id')
+    image_uri = vac_input.get('image_url') or vac_input.get('image_uri')
+
+    log.info(f"Streaming invoke_vac_qa with vac_input={vac_input}")
+
+    def sync_stream_response():
+        generate = generate_proxy_stream(
+            send_to_qa,
+            vac_input["user_input"],
+            vector_name=vac_name,
+            chat_history=chat_history,
+            generate_f_output=lambda x: x,  # Replace with actual processing function
+            stream_wait_time=0.5,
+            stream_timeout=120,
+            message_author=user_id,
+            # TODO: populate these
+            image_url=image_uri,
+            source_filters=None,
+            search_kwargs=None,
+            private_docs=None,
+            whole_document=False,
+            source_filters_and_or=False,
+            # system kwargs
+            configurable={
+                "vector_name": vac_name,
+            },
+            user_id=user_id,
+            session_id=session_id,
+            message_source="sunholo.invoke_vac_qa.stream"
+        )
+        for part in generate():
+            yield part
+
+    async def async_stream_response():
+        loop = asyncio.get_event_loop()
+        queue = asyncio.Queue()
+
+        def run_sync_gen():
+            try:
+                for item in sync_stream_response():
+                    loop.call_soon_threadsafe(queue.put_nowait, item)
+            finally:
+                loop.call_soon_threadsafe(queue.put_nowait, None)  # Sentinel
+
+        thread = Thread(target=run_sync_gen)
+        thread.start()
+
+        while True:
+            item = await queue.get()
+            if item is None:
+                break
+            yield item
+
+        thread.join()
+
+    answer = ""
+
+    async for token in async_stream_response():
+        if isinstance(token, bytes):
+            token = token.decode('utf-8')
+            yield token
+        if isinstance(token, dict):
+            # Process dict token if necessary
+            pass
+        elif isinstance(token, str):
+            answer += token
+
+    if answer:
+        chat_history.append({"name": "Human", "content": vac_input})
+        chat_history.append({"name": "AI", "content": answer})
+
+    yield chat_history

{sunholo-0.96.3 → sunholo-0.96.5}/sunholo.egg-info/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: sunholo
-Version: 0.96.3
+Version: 0.96.5
 Summary: Large Language Model DevOps - a package to help deploy LLMs to the Cloud.
 Home-page: https://github.com/sunholo-data/sunholo-py
-Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.96.3.tar.gz
+Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.96.5.tar.gz
 Author: Holosun ApS
 Author-email: multivac@sunholo.com
 License: Apache License, Version 2.0

sunholo-0.96.3/sunholo/invoke/direct_vac_func.py

@@ -1,123 +0,0 @@
-from ..custom_logging import log
-from ..agents import send_to_qa
-from ..qna.parsers import parse_output
-from ..streaming import generate_proxy_stream
-from ..utils import ConfigManager
-from ..utils.api_key import has_multivac_api_key
-
-def direct_vac(vac_input: dict, vac_name: str, chat_history=[]):
-    """
-    This lets VACs call other VAC Q&A endpoints within their code
-    """
-
-    log.info(f"Invoking VAC Q&A endpoints for {vac_name}")
-
-    if 'user_input' not in vac_input:
-        raise ValueError(f'vac_input must contain at least "user_input" key - got {vac_input}')
-
-    global_config = ConfigManager('global')
-    config = ConfigManager(vac_name)
-
-    agent_name = config.vacConfig('agent')
-    agent_url = config.vacConfig("agent_url")
-
-    if agent_url:
-        log.info("Found agent_url within vacConfig: {agent_url}")
-    # via public cloud endpoints - assumes no gcloud auth
-    override_endpoint = None
-    if has_multivac_api_key():
-        print("Found MULTIVAC_API_KEY")
-        gcp_config = global_config.vacConfig("gcp_config")
-        endpoints_base_url = gcp_config.get("endpoints_base_url")
-        if not endpoints_base_url:
-            raise ValueError("MULTIVAC_API_KEY env var is set but no config.gcp_config.endpoints_base_url can be found")
-
-        override_endpoint = f"{endpoints_base_url}/v1/{agent_name}"
-
-    override_endpoint = agent_url or override_endpoint
-
-    print(f"Using {override_endpoint=}")
-    
-    # Prepare the kwargs for send_to_qa by copying vac_input and adding more values
-    qa_kwargs = vac_input.copy()
-
-    # Add additional arguments
-    qa_kwargs.update({
-        'vector_name': vac_name,
-        'chat_history': chat_history,
-        'image_url': vac_input.get('image_url') or vac_input.get('image_uri'),
-        'override_endpoint': override_endpoint,
-        'message_source': "sunholo.invoke_vac_qa.invoke",
-        'stream': False,
-        'configurable': {
-            "vector_name": vac_name,
-        },
-    })
-
-    log.info(f'Batch invoke_vac_qa {vac_name} with {qa_kwargs=}')
-
-    vac_response = send_to_qa(**qa_kwargs)
-        
-    # ensures {'answer': answer}
-    answer = parse_output(vac_response)
-    chat_history.append({"name": "Human", "content": vac_input})
-    chat_history.append({"name": "AI", "content": answer})
-    answer["chat_history"] = chat_history
-    
-    return answer
-
-def direct_vac_stream(vac_input: dict, vac_name: str, chat_history=[]):
-
-    if 'user_input' not in vac_input:
-        raise ValueError('vac_input must contain at least "user_input" key - got {vac_input}')
-
-    user_id = vac_input.get('user_id')
-    session_id = vac_input.get('session_id')
-    image_uri = vac_input.get('image_url') or vac_input.get('image_uri')
-    
-    log.info(f"Streaming invoke_vac_qa with {vac_input=}")
-    def stream_response():
-        generate = generate_proxy_stream(
-                send_to_qa,
-                vac_input["user_input"],
-                vector_name=vac_name,
-                chat_history=chat_history,
-                generate_f_output=lambda x: x,  # Replace with actual processing function
-                stream_wait_time=0.5,
-                stream_timeout=120,
-                message_author=user_id,
-                #TODO: populate these
-                image_url=image_uri,
-                source_filters=None,
-                search_kwargs=None,
-                private_docs=None,
-                whole_document=False,
-                source_filters_and_or=False,
-                # system kwargs
-                configurable={
-                    "vector_name": vac_name,
-                },
-                user_id=user_id,
-                session_id=session_id, 
-                message_source="sunholo.invoke_vac_qa.stream"
-        )
-        for part in generate():
-            yield part
-
-    answer = ""
-
-    for token in stream_response():
-        if isinstance(token, bytes):
-            token = token.decode('utf-8')
-            yield token
-        if isinstance(token, dict):
-            # ?
-            pass
-        elif isinstance(token, str):
-            answer += token
-
-    if answer:
-        chat_history.append({"name": "Human", "content": vac_input})
-        chat_history.append({"name": "AI", "content": answer})
-
-    return chat_history