sunholo 0.55.14__tar.gz → 0.56.1__tar.gz

{sunholo-0.55.14 → sunholo-0.56.1}/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: sunholo
-Version: 0.55.14
+Version: 0.56.1
 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/v0.55.14.tar.gz
+Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.56.1.tar.gz
 Author: Holosun ApS
 Author-email: multivac@sunholo.com
 License: Apache License, Version 2.0
@@ -30,6 +30,7 @@ Requires-Dist: google-cloud-alloydb-connector[pg8000]; extra == "all"
 Requires-Dist: google-cloud-logging; extra == "all"
 Requires-Dist: google-cloud-storage; extra == "all"
 Requires-Dist: google-cloud-pubsub; extra == "all"
+Requires-Dist: gunicorn; extra == "all"
 Requires-Dist: httpcore; extra == "all"
 Requires-Dist: httpx; extra == "all"
 Requires-Dist: lancedb; extra == "all"
@@ -63,11 +64,12 @@ Requires-Dist: langchain-openai; extra == "openai"
 Provides-Extra: anthropic
 Requires-Dist: langchain-anthropic; extra == "anthropic"
 Provides-Extra: http
-Requires-Dist: langfuse; extra == "http"
 Requires-Dist: fastapi; extra == "http"
 Requires-Dist: flask; extra == "http"
+Requires-Dist: gunicorn; extra == "http"
 Requires-Dist: httpcore; extra == "http"
 Requires-Dist: httpx; extra == "http"
+Requires-Dist: langfuse; extra == "http"
 Requires-Dist: python-socketio; extra == "http"
 
 ## Introduction

{sunholo-0.55.14 → sunholo-0.56.1}/setup.py

@@ -1,7 +1,7 @@
 from setuptools import setup, find_packages
 
 # Define your base version
-version = '0.55.14'
+version = '0.56.1'
 
 setup(
     name='sunholo',
@@ -16,6 +16,11 @@ setup(
     url = 'https://github.com/sunholo-data/sunholo-py',
     download_url=f'https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v{version}.tar.gz',
     keywords=['llms', 'devops','google_cloud_platform'],
+    entry_points={
+        'console_scripts': [
+            'sunholo=sunholo.cli.cli:main', 
+        ],
+    },
     package_data={
         'sunholo.database': ['sql/sb/*.sql'],
         'sunholo.lookup': ['*.yaml']
@@ -38,6 +43,7 @@ setup(
             "google-cloud-logging",
             "google-cloud-storage",
             "google-cloud-pubsub",
+            "gunicorn",
             "httpcore",
             "httpx",
             "lancedb",
@@ -76,11 +82,12 @@ setup(
             "langchain-anthropic",        
         ],
         'http': [
-            "langfuse",
             "fastapi",
             "flask",
+            "gunicorn",
             "httpcore",
             "httpx",
+            "langfuse",
             "python-socketio",
         ]
     },

{sunholo-0.55.14 → sunholo-0.56.1}/sunholo/agents/flask/qna_routes.py

@@ -23,13 +23,6 @@ from ...streaming import start_streaming_chat
 from ...archive import archive_qa
 from ...logging import log
 from ...utils.config import load_config
-
-try:
-    from langfuse import Langfuse
-    langfuse = Langfuse()
-except ImportError as err:
-    print(f"No langfuse installed for agents.flask.register_qna_routes, install via `pip install sunholo[http]` - {str(err)}")
-    langfuse = None
  
 try:
     from flask import request, jsonify, Response
@@ -71,7 +64,7 @@ def register_qna_routes(app, stream_interpreter, vac_interpreter):
                                               wait_time=all_input["stream_wait_time"],
                                               timeout=all_input["stream_timeout"],
                                               #kwargs
-                                              message_author=all_input["message_author"]
+                                              **all_input["kwargs"]
                                               ):
                 if isinstance(chunk, dict) and 'answer' in chunk:
                     # When we encounter the dictionary, we yield it as a JSON string
@@ -98,13 +91,11 @@ def register_qna_routes(app, stream_interpreter, vac_interpreter):
         response = Response(generate_response_content(), content_type='text/plain; charset=utf-8')
         response.headers['Transfer-Encoding'] = 'chunked'  
 
+        log.debug(f"streaming response: {response}")
         if trace:
             generation.end(output=response)
             span.end(output=response)
             trace.update(output=response)
-        
-        if langfuse:
-            langfuse.flush()  
 
         return response
 
@@ -133,7 +124,7 @@ def register_qna_routes(app, stream_interpreter, vac_interpreter):
                 question=all_input["user_input"],
                 vector_name=vector_name,
                 chat_history=all_input["chat_history"],
-                message_author=all_input["message_author"]
+                **all_input["kwargs"]
             )
             if generation:
                 generation.end(output=bot_output)
@@ -152,9 +143,6 @@ def register_qna_routes(app, stream_interpreter, vac_interpreter):
         if trace:
             span.end(output=jsonify(bot_output))
             trace.update(output=jsonify(bot_output)) 
-       
-        if langfuse:
-            langfuse.flush()
 
         return jsonify(bot_output)
 
@@ -200,18 +188,18 @@ def prep_vac(request, vector_name):
     if trace:
         trace.update(input=data, metadata=vac_config)
 
-    user_input = data['user_input'].strip()
-    message_author = data.get('message_author', None)
-    stream_wait_time = data.get('stream_wait_time', 7)
-    stream_timeout = data.get('stream_timeout', 120)
+    user_input = data.pop('user_input').strip()
+    stream_wait_time = data.pop('stream_wait_time', 7)
+    stream_timeout = data.pop('stream_timeout', 120)
+    chat_history = data.pop('chat_history', None)
+    paired_messages = extract_chat_history(chat_history)
 
-    paired_messages = extract_chat_history(data.get('chat_history', None))
     all_input = {'user_input': user_input, 
                  'vector_name': vector_name, 
                  'chat_history': paired_messages, 
-                 'message_author': message_author,
                  'stream_wait_time': stream_wait_time,
-                 'stream_timeout':stream_timeout}
+                 'stream_timeout':stream_timeout,
+                 'kwargs': data}
 
     if trace:
         span = trace.span(

sunholo-0.56.1/sunholo/cli/cli.py

@@ -0,0 +1,12 @@
+# sunholo/cli/cli.py
+import argparse
+
+def main():
+    parser = argparse.ArgumentParser(description="sunholo CLI tool.")
+    parser.add_argument('--echo', help='Echo the string you use here')
+    args = parser.parse_args()
+    if args.echo:
+        print(args.echo)
+
+if __name__ == "__main__":
+    main()

{sunholo-0.55.14 → sunholo-0.56.1}/sunholo/components/llm.py

@@ -19,7 +19,7 @@ import os
 def pick_llm(vector_name):
     log.debug('Picking llm')
     
-    llm_str = load_config_key("llm", vector_name, filename = "config/llm_config.yaml")
+    llm_str = load_config_key("llm", vector_name, kind="vacConfig")
     
     if llm_str == 'openai':
         llm_chat = get_llm_chat(vector_name)
@@ -49,7 +49,7 @@ def pick_llm(vector_name):
 
 def pick_streaming(vector_name):
     
-    llm_str = load_config_key("llm", vector_name, filename = "config/llm_config.yaml")
+    llm_str = load_config_key("llm", vector_name, kind="vacConfig")
     
     if llm_str == 'openai' or llm_str == 'gemini' or llm_str == 'vertex':
         return True
@@ -57,7 +57,7 @@ def pick_streaming(vector_name):
     return False
 
 
-def llm_str_to_llm(llm_str, model=None, vector_name=None, config_file="config/llm_config.yaml"):
+def llm_str_to_llm(llm_str, model=None, vector_name=None):
     if llm_str == 'openai':
         # Setup for OpenAI LLM
         from langchain_openai import ChatOpenAI
@@ -79,7 +79,7 @@ def llm_str_to_llm(llm_str, model=None, vector_name=None, config_file="config/ll
 
     elif llm_str == 'model_garden':
         from ..patches.langchain.vertexai import VertexAIModelGarden
-        model_garden_config = load_config_key("gcp_config", vector_name, filename = config_file)
+        model_garden_config = load_config_key("gcp_config", vector_name, kind="vacConfig")
         if model_garden_config is None:
             raise ValueError("llm='model_garden' requires a gcp_config entry in config yaml file")
         
@@ -97,21 +97,21 @@ def llm_str_to_llm(llm_str, model=None, vector_name=None, config_file="config/ll
     if llm_str is None:
         raise NotImplementedError(f'No llm implemented for {llm_str}') 
 
-def get_llm(vector_name, model=None, config_file="config/llm_config.yaml"):
-    llm_str = load_config_key("llm", vector_name, filename=config_file)
+def get_llm(vector_name, model=None):
+    llm_str = load_config_key("llm", vector_name, kind="vacConfig")
     #model_lookup_filepath = get_module_filepath("lookup/model_lookup.yaml")
     #model_lookup, _ = load_config(model_lookup_filepath)
 
     if not model:
-        model = load_config_key("model", vector_name, filename=config_file)
+        model = load_config_key("model", vector_name, kind="vacConfig")
 
     log.debug(f"Chose LLM: {llm_str}")
-    return llm_str_to_llm(llm_str, model=model, vector_name=vector_name, config_file=config_file)
+    return llm_str_to_llm(llm_str, model=model, vector_name=vector_name, kind="vacConfig")
 
 def get_llm_chat(vector_name, model=None, config_file="config/llm_config.yaml"):
-    llm_str = load_config_key("llm", vector_name, filename=config_file)
+    llm_str = load_config_key("llm", vector_name, kind="vacConfig")
     if not model:
-        model = load_config_key("model", vector_name, filename=config_file)
+        model = load_config_key("model", vector_name, kind="vacConfig")
 
     log.debug(f"Chose LLM: {llm_str}")
     # Configure LLMs based on llm_str
@@ -150,7 +150,7 @@ def get_llm_chat(vector_name, model=None, config_file="config/llm_config.yaml"):
         return ChatAnthropic(model_name = model, temperature=0)
     elif llm_str == 'azure':
         from langchain_openai import AzureChatOpenAI
-        azure_config = load_config_key("azure", vector_name, filename="config/llm_config.yaml")
+        azure_config = load_config_key("azure", vector_name, kind="vacConfig")
         if not azure_config:
             raise ValueError("Need to configure azure.config if llm='azure'")
 
@@ -195,13 +195,13 @@ def get_llm_chat(vector_name, model=None, config_file="config/llm_config.yaml"):
 def get_embeddings(vector_name):
 
     llm_str = None
-    embed_dict = load_config_key("embedder", vector_name, filename="config/llm_config.yaml")
+    embed_dict = load_config_key("embedder", vector_name, kind="vacConfig")
 
     if embed_dict:
         llm_str = embed_dict.get('llm')
 
     if llm_str is None:
-        llm_str = load_config_key("llm", vector_name, filename="config/llm_config.yaml")
+        llm_str = load_config_key("llm", vector_name, kind="vacConfig")
 
     return pick_embedding(llm_str, vector_name=vector_name)
 
@@ -227,7 +227,7 @@ def pick_embedding(llm_str: str, vector_name: str=None):
     elif llm_str == 'azure':
         from langchain_openai import AzureOpenAIEmbeddings
         
-        azure_config = load_config_key("azure", vector_name, filename="config/llm_config.yaml")
+        azure_config = load_config_key("azure", vector_name, kind="vacConfig")
         if not azure_config:
             raise ValueError("Need to configure azure.config if llm='azure'")
 

{sunholo-0.55.14 → sunholo-0.56.1}/sunholo/logging.py

@@ -220,11 +220,16 @@ def setup_logging(logger_name=None, log_level=logging.INFO, project_id=None):
     if logger_name is None:
         logger_name = "sunholo"
 
-    if Client:
+    if Client and os.environ.get('GOOGLE_CLOUD_LOGGING') != "1":
+        print("GOOGLE_CLOUD_LOGGING != 1 but authentication with Google Cloud Logging enabled - missing env var setting?")
+
+    if not Client and os.environ.get('GOOGLE_CLOUD_LOGGING') == "1":
+        print("Found GOOGLE_CLOUD_LOGGING=1 but no GCP Client available, install via `pip install sunholo[gcp]` and/or authenticate")
+
+    if Client and os.environ.get('GOOGLE_CLOUD_LOGGING') == "1":
         # Instantiate the GoogleCloudLogging class
         gc_logger = GoogleCloudLogging(project_id, log_level=log_level, logger_name=logger_name)
     else:
-        print("Could not find GoogleCloudLogging Client, install via `pip install sunholo[gcp]`")
         return logger
     
     # Setup logging and return the logger instance

{sunholo-0.55.14 → sunholo-0.56.1}/sunholo/utils/config.py

@@ -16,6 +16,7 @@ import os
 import json
 import yaml
 from datetime import datetime, timedelta
+from collections import defaultdict
 
 try:
     from google.cloud import storage
@@ -95,6 +96,58 @@ def get_module_filepath(filepath: str):
 
 # Global cache
 config_cache = {}
+def load_all_configs():
+    """
+    Load all configuration files from the specified directory into a dictionary.
+    Files are expected to be either YAML or JSON and must contain a 'kind' key at the root.
+    Caching is used to avoid reloading files within a 5-minute window.
+    """
+    from ..logging import log
+
+    config_folder = os.getenv("_CONFIG_FOLDER", os.getcwd())
+    log.debug(f"Loading all configs from folder: {config_folder}")
+    current_time = datetime.now()
+
+    configs_by_kind = defaultdict(dict)
+    for filename in os.listdir(config_folder):
+        if filename.endswith(('.yaml', '.yml', '.json')):
+            config_file = os.path.join(config_folder, filename)
+            
+            # Check cache first
+            if filename in config_cache:
+                cached_config, cache_time = config_cache[filename]
+                if (current_time - cache_time) < timedelta(minutes=5):
+                    log.debug(f"Returning cached config for {filename}")
+                    config = cached_config
+                else:
+                    config = reload_config_file(config_file, filename)
+            else:
+                config = reload_config_file(config_file, filename)
+
+            kind = config.get('kind')
+            if kind:
+                configs_by_kind[kind] = config
+            else:
+                log.warning(f"No 'kind' found in {filename}")
+
+    return configs_by_kind
+
+def reload_config_file(config_file, filename):
+    """
+    Helper function to load a config file and update the cache.
+    """
+    from ..logging import log
+    with open(config_file, 'r') as file:
+        if filename.endswith('.json'):
+            config = json.load(file)
+        else:
+            config = yaml.safe_load(file)
+    
+    config_cache[filename] = (config, datetime.now())
+    log.debug(f"Loaded and cached {filename}")
+    return config
+
+
 
 def load_config(filename: str=None) -> tuple[dict, str]:
     """
@@ -156,14 +209,15 @@ def load_config(filename: str=None) -> tuple[dict, str]:
     
     return config, filename
 
-def load_config_key(key: str, vector_name: str, filename: str=None):
+def load_config_key(key: str, vector_name: str, filename: str=None, kind: str=None):
     """
     Load a specific key from a configuration file.
 
     Args:
         key (str): The key to fetch from the configuration.
         vector_name (str): The name of the vector in the configuration file.
-        filename (str, optional): The configuration file name. Defaults to the `_CONFIG_FILE` environment variable.
+        filename (str, optional): The configuration file name. Defaults to the `_CONFIG_FILE` environment variable. Deprecated - use 'kind' instead
+        kind: (str, optional): Specify the type of configuration to retrieve e.g. 'vacConfig' which will pick from files within `_CONFIG_FOLDER`
 
     Returns:
         str: The value associated with the specified key.
@@ -179,7 +233,13 @@ def load_config_key(key: str, vector_name: str, filename: str=None):
     assert isinstance(key, str), f"key must be a string got a {type(key)}"
     assert isinstance(vector_name, str), f"vector_name must be a string, got a {type(vector_name)}"
     
-    config, filename = load_config(filename)
+    configs_by_kind = load_all_configs()
+    if kind and configs_by_kind.get(kind):
+        config = configs_by_kind[kind]
+        filename = kind
+    else:
+        config, filename = load_config(filename)
+
     log.info(f"Fetching {key} for {vector_name}")
     apiVersion = config.get('apiVersion')
     kind = config.get('kind')

{sunholo-0.55.14 → sunholo-0.56.1}/sunholo.egg-info/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: sunholo
-Version: 0.55.14
+Version: 0.56.1
 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/v0.55.14.tar.gz
+Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.56.1.tar.gz
 Author: Holosun ApS
 Author-email: multivac@sunholo.com
 License: Apache License, Version 2.0
@@ -30,6 +30,7 @@ Requires-Dist: google-cloud-alloydb-connector[pg8000]; extra == "all"
 Requires-Dist: google-cloud-logging; extra == "all"
 Requires-Dist: google-cloud-storage; extra == "all"
 Requires-Dist: google-cloud-pubsub; extra == "all"
+Requires-Dist: gunicorn; extra == "all"
 Requires-Dist: httpcore; extra == "all"
 Requires-Dist: httpx; extra == "all"
 Requires-Dist: lancedb; extra == "all"
@@ -63,11 +64,12 @@ Requires-Dist: langchain-openai; extra == "openai"
 Provides-Extra: anthropic
 Requires-Dist: langchain-anthropic; extra == "anthropic"
 Provides-Extra: http
-Requires-Dist: langfuse; extra == "http"
 Requires-Dist: fastapi; extra == "http"
 Requires-Dist: flask; extra == "http"
+Requires-Dist: gunicorn; extra == "http"
 Requires-Dist: httpcore; extra == "http"
 Requires-Dist: httpx; extra == "http"
+Requires-Dist: langfuse; extra == "http"
 Requires-Dist: python-socketio; extra == "http"
 
 ## Introduction

{sunholo-0.55.14 → sunholo-0.56.1}/sunholo.egg-info/SOURCES.txt

@@ -8,6 +8,7 @@ sunholo/logging.py
 sunholo.egg-info/PKG-INFO
 sunholo.egg-info/SOURCES.txt
 sunholo.egg-info/dependency_links.txt
+sunholo.egg-info/entry_points.txt
 sunholo.egg-info/requires.txt
 sunholo.egg-info/top_level.txt
 sunholo/agents/__init__.py
@@ -40,6 +41,8 @@ sunholo/chunker/message_data.py
 sunholo/chunker/pdfs.py
 sunholo/chunker/publish.py
 sunholo/chunker/splitter.py
+sunholo/cli/__init__.py
+sunholo/cli/cli.py
 sunholo/components/__init__.py
 sunholo/components/llm.py
 sunholo/components/prompt.py

sunholo-0.56.1/sunholo.egg-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+sunholo = sunholo.cli.cli:main

{sunholo-0.55.14 → sunholo-0.56.1}/sunholo.egg-info/requires.txt

@@ -11,6 +11,7 @@ google-cloud-alloydb-connector[pg8000]
 google-cloud-logging
 google-cloud-storage
 google-cloud-pubsub
+gunicorn
 httpcore
 httpx
 lancedb
@@ -46,11 +47,12 @@ google-api-python-client
 google-cloud-alloydb-connector[pg8000]
 
 [http]
-langfuse
 fastapi
 flask
+gunicorn
 httpcore
 httpx
+langfuse
 python-socketio
 
 [openai]