sunholo 0.95.2__tar.gz → 0.95.4__tar.gz

{sunholo-0.95.2 → sunholo-0.95.4}/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: sunholo
-Version: 0.95.2
+Version: 0.95.4
 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/v0.95.2.tar.gz
+Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.95.4.tar.gz
 Author: Holosun ApS
 Author-email: multivac@sunholo.com
 License: Apache License, Version 2.0

{sunholo-0.95.2 → sunholo-0.95.4}/setup.py

@@ -1,6 +1,6 @@
 from setuptools import setup, find_packages
 
-version = '0.95.2'
+version = '0.95.4'
 
 setup(
     name='sunholo',

{sunholo-0.95.2 → sunholo-0.95.4}/sunholo/agents/route.py

@@ -65,14 +65,14 @@ def route_endpoint(vector_name=None, method = 'post', override_endpoint=None, co
     
     agents_config = config.agentConfig(agent_type)
     
-    log.info(f"agents_config: {agents_config}")
+    log.debug(f"agents_config: {agents_config}")
     if method not in agents_config:
         raise ValueError(f"Invalid method '{method}' for agent configuration.")
 
     # 'post' or 'get'
     endpoints_config = agents_config[method]
 
-    log.info(f"endpoints_config: {endpoints_config}")
+    log.debug(f"endpoints_config: {endpoints_config}")
     # Replace placeholders in the config
     endpoints = {}
     for key, value in endpoints_config.items():

{sunholo-0.95.2 → sunholo-0.95.4}/sunholo/genai/process_funcs_cls.py

@@ -351,7 +351,7 @@ class GenAIFunctionProcessor:
             log.error(f"Error initializing model: {str(err)}")
             return None
 
-    def run_agent_loop(self, chat, content, callback, guardrail_max=10):
+    def run_agent_loop(self, chat, content, callback, guardrail_max=10, loop_return=3):
         """
         Runs the agent loop, sending messages to the orchestrator, processing responses, and executing functions.
 
@@ -360,12 +360,13 @@ class GenAIFunctionProcessor:
             content: The initial content to send to the agent.
             callback: The callback object for handling intermediate responses.
             guardrail_max (int): The maximum number of iterations for the loop.
+            loop_return (int): The number of last loop iterations to return. Default 3 will return last 3 iterations. If loop_return > guardrail_max then all iterations are returned.
 
         Returns:
             tuple: (big_text, usage_metadata) from the loop execution.
         """
         guardrail = 0
-        big_text = ""
+        big_result = []
         usage_metadata = {
             "prompt_token_count": 0,
             "candidates_token_count": 0,
@@ -422,7 +423,6 @@ class GenAIFunctionProcessor:
                     if hasattr(chunk, 'text') and isinstance(chunk.text, str):
                         token = chunk.text
                         token_queue.append(token)
-                        big_text += token
                         this_text += token
                     else:
                         log.info("skipping chunk with no text")
@@ -486,18 +486,17 @@ class GenAIFunctionProcessor:
                         else:
                             token += f"{fn_result}\n--- end ---\n"
                     
-                    big_text += token
                     this_text += token
                     token_queue.append(token)
             else:
                 token = "\nNo function executions were found\n"
                 token_queue.append(token)
-                big_text += token
                 this_text += token
 
             if this_text:
                 content.append(f"Agent: {this_text}")    
                 log.info(f"[{guardrail}] Updated content:\n{this_text}")
+                big_result.append(this_text)
             else:
                 log.warning(f"[{guardrail}] No content created this loop")
                 content.append(f"Agent: No response was found for loop [{guardrail}]")
@@ -523,7 +522,8 @@ class GenAIFunctionProcessor:
             callback.on_llm_new_token(token=token)
         
         usage_metadata["functions_called"] = functions_called
-        #usage_metadata["function_results"] = function_results
+
+        big_text = "\n".join(big_result[loop_return:])
 
         return big_text, usage_metadata
 

{sunholo-0.95.2 → sunholo-0.95.4}/sunholo/invoke/async_class.py

@@ -39,17 +39,13 @@ class AsyncTaskRunner:
     async def _task_wrapper(self, name: str, func: Callable[..., Any], args: Any) -> Any:
         """Wraps the task function to process its output and handle retries."""
         async def run_func():
-            gen = func(*args)
-            if hasattr(gen, '__aiter__'):
-                # If it's an async generator, consume it completely
-                output = ''
-                async for chunk in gen:
-                    output += chunk
-                return output
+            if asyncio.iscoroutinefunction(func):
+                # If the function is async, await it
+                return await func(*args)
             else:
-                # If it's a coroutine, await the result
-                return await gen
-
+                # If the function is sync, run it in a thread to prevent blocking
+                return await asyncio.to_thread(func, *args)
+    
         if self.retry_enabled:
             retry_kwargs = {
                 'wait': wait_random_exponential(multiplier=1, max=60),
@@ -65,12 +61,4 @@ class AsyncTaskRunner:
                 return await run_func()
             except Exception as e:
                 log.error(f"Error in task {name}: {e}\n{traceback.format_exc()}")
-                raise
-
-    def _log_results(self, results: List[Any]):
-        """Logs the results of the task executions."""
-        for result in results:
-            if isinstance(result, Exception):
-                log.error(f"Task resulted in an error: {result}")
-            else:
-                log.info(f"Task completed successfully: {result}")
+                raise

{sunholo-0.95.2 → sunholo-0.95.4}/sunholo/streaming/streaming.py

@@ -175,6 +175,16 @@ async def start_streaming_chat_async(question, vector_name, qna_func_async, chat
     # Run start_chat asynchronously
     chat_task = asyncio.create_task(start_chat())
 
+     # Allow the event loop to process the scheduled tasks
+    await asyncio.sleep(0)
+
+    # Read and yield any initial content from the content buffer
+    content_to_send = content_buffer.read()
+    if content_to_send:
+        log.info(f"Initial content: {content_to_send}")
+        yield content_to_send
+        content_buffer.clear()
+
     start = time.time()
 
     while not chat_callback_handler.stream_finished.is_set() and not stop_event.is_set():

{sunholo-0.95.2 → sunholo-0.95.4}/sunholo/vertex/memory_tools.py

@@ -174,15 +174,11 @@ def print_grounding_response(response):
         markdown_text += (
             f"\n**Web Search Queries:** {grounding_metadata.web_search_queries}\n"
         )
-        if grounding_metadata.search_entry_point:
-            markdown_text += f"\n**Search Entry Point:**\n {grounding_metadata.search_entry_point.rendered_content}\n"
     elif grounding_metadata.retrieval_queries:
         markdown_text += (
             f"\n**Retrieval Queries:** {grounding_metadata.retrieval_queries}\n"
         )
 
-    markdown_text += "### Grounding Chunks\n"
-
     for index, grounding_chunk in enumerate(
         grounding_metadata.grounding_chunks, start=1
     ):
@@ -190,7 +186,9 @@ def print_grounding_response(response):
         if not context:
             print(f"Skipping Grounding Chunk {grounding_chunk}")
             continue
+        markdown_text += "### Grounding Chunks\n"
 
         markdown_text += f"{index}. [{context.title}]({context.uri})\n"
 
-    return markdown_text
+    return {"markdown_text": markdown_text, 
+            "search_entry_point": grounding_metadata.search_entry_point.rendered_content if grounding_metadata.search_entry_point else ""}

{sunholo-0.95.2 → sunholo-0.95.4}/sunholo.egg-info/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: sunholo
-Version: 0.95.2
+Version: 0.95.4
 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/v0.95.2.tar.gz
+Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.95.4.tar.gz
 Author: Holosun ApS
 Author-email: multivac@sunholo.com
 License: Apache License, Version 2.0