sunholo 0.126.3__tar.gz → 0.127.0__tar.gz

{sunholo-0.126.3/src/sunholo.egg-info → sunholo-0.127.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sunholo
-Version: 0.126.3
+Version: 0.127.0
 Summary: Large Language Model DevOps - a package to help deploy LLMs to the Cloud.
 Author-email: Holosun ApS <multivac@sunholo.com>
 License: Apache License, Version 2.0
@@ -142,6 +142,7 @@ Requires-Dist: google-cloud-discoveryengine>=0.13.4; extra == "gcp"
 Requires-Dist: google-cloud-texttospeech; extra == "gcp"
 Requires-Dist: google-genai>=0.2.2; extra == "gcp"
 Requires-Dist: google-generativeai>=0.8.3; extra == "gcp"
+Requires-Dist: langchain; extra == "gcp"
 Requires-Dist: langchain-google-genai>=2.0.0; extra == "gcp"
 Requires-Dist: langchain_google_alloydb_pg>=0.2.2; extra == "gcp"
 Requires-Dist: langchain-google-vertexai; extra == "gcp"
@@ -164,6 +165,7 @@ Requires-Dist: flask; extra == "http"
 Requires-Dist: gunicorn; extra == "http"
 Requires-Dist: httpcore; extra == "http"
 Requires-Dist: httpx; extra == "http"
+Requires-Dist: langchain; extra == "http"
 Requires-Dist: langfuse; extra == "http"
 Requires-Dist: python-socketio; extra == "http"
 Requires-Dist: requests; extra == "http"

{sunholo-0.126.3 → sunholo-0.127.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "sunholo"
-version = "0.126.3"
+version = "0.127.0"
 description = "Large Language Model DevOps - a package to help deploy LLMs to the Cloud."
 readme = "README.md"
 requires-python = ">=3.10"
@@ -179,6 +179,7 @@ gcp = [
     "google-cloud-texttospeech",
     "google-genai>=0.2.2",
     "google-generativeai>=0.8.3",
+    "langchain",
     "langchain-google-genai>=2.0.0",
     "langchain_google_alloydb_pg>=0.2.2",
     "langchain-google-vertexai",
@@ -206,6 +207,7 @@ http = [
     "gunicorn",
     "httpcore",
     "httpx",
+    "langchain",
     "langfuse",
     "python-socketio",
     "requests",

{sunholo-0.126.3 → sunholo-0.127.0}/src/sunholo/agents/dispatch_to_qa.py

@@ -21,11 +21,6 @@ import traceback
 from .route import route_endpoint
 import os
 
-try:
-    from langfuse import Langfuse
-    langfuse = Langfuse()
-except ImportError:
-    langfuse = None
 
 def prep_request_payload(user_input, chat_history, vector_name, stream, **kwargs):
     """
@@ -93,7 +88,11 @@ def prep_request_payload(user_input, chat_history, vector_name, stream, **kwargs
     return qna_endpoint, qna_data
 
 def add_langfuse_trace(qna_endpoint):
-    if not langfuse:
+    try:
+        from langfuse import Langfuse
+        langfuse = Langfuse()
+    except Exception as err:
+        log.error(err)
         return None
     
     trace = langfuse.trace(name = f'dispatch/{os.path.basename(qna_endpoint)}')

{sunholo-0.126.3 → sunholo-0.127.0}/src/sunholo/discovery_engine/chunker_handler.py

@@ -36,6 +36,7 @@ def do_discovery_engine(message_data:str, metadata:dict, config:ConfigManager=No
             if vectorstore == "discovery_engine" or vectorstore == "vertex_ai_search":
                 log.info(f"Found vectorstore {vectorstore}")
                 if value.get('read_only'):
+                    log.info(f"{vectorstore} is read only, skipping")
                     continue
 
                 project_id = value.get("project_id")
@@ -50,6 +51,7 @@ def do_discovery_engine(message_data:str, metadata:dict, config:ConfigManager=No
                 if not project_id:
                     raise ValueError("Couldn't retrieve project_id for vertex_ai_search")        
 
+                log.info(f"Using {project_id} and {location} for DiscoveryEngineClient")
                 corpus = DiscoveryEngineClient(
                     data_store_id=config.vector_name, 
                     project_id=project_id,
@@ -67,10 +69,13 @@ def do_discovery_engine(message_data:str, metadata:dict, config:ConfigManager=No
     if message_data.startswith("gs://"):
         log.info(f"DiscoveryEngineClient.import_files for {message_data}")
         if "/pdf_parts/" in message_data:
+            log.info(f"Not processing files with /pdf_parts/ - {message_data}")
             return None
         for corp in corpuses:
             try:
+                
                 metadata = audit_metadata(metadata, chunk_length=500)
+                log.info(f"Importing {message_data} {metadata=} to {corp}")
                 response = corp.import_document_with_metadata(
                     gcs_uri=message_data,
                     metadata=metadata
@@ -160,7 +165,10 @@ def discovery_engine_chunker_check(message_data,
         try:
             log.info(f"Process discovery engine for {metadata}")
             disc_meta = do_discovery_engine(message_data, metadata, config=config)
-            log.info(f"Processed discovery engine: {disc_meta}")
+            if disc_meta is None:
+                log.error(f"No disc_meta found for {metadata}")
+            else:
+                log.info(f"Processed discovery engine: {disc_meta}")
         except Exception as err:
             log.error(f"Error processing discovery engine: {str(err)} {traceback.format_exc()}")
             disc_meta = None

{sunholo-0.126.3 → sunholo-0.127.0}/src/sunholo/discovery_engine/discovery_engine_client.py

@@ -16,6 +16,7 @@ import asyncio
 import json
 import uuid
 from ..utils.mime import guess_mime_type
+import traceback
 
 class DiscoveryEngineClient:
     """
@@ -80,8 +81,10 @@ class DiscoveryEngineClient:
             self.async_search_client = discoveryengine.SearchServiceAsyncClient(client_options=client_options)
         except RuntimeError:
             # No event loop in non-async environment, set async client to None
-            log.info("No event loop detected; skipping async client initialization")
+            log.info("No event loop detected; skipping Discoveryengine async client initialization")
             self.async_search_client = None
+        
+        log.info(f"Discovery Engine client initialized with {self.project_id=}, {self.data_store_id=}, {self.location=}")
 
     @classmethod
     def my_retry(cls):
@@ -490,6 +493,7 @@ class DiscoveryEngineClient:
             return doc_client.import_documents(request=request)
         
         try:
+            log.debug(f"Requesting import of documents: {request=}")
             operation = import_documents_with_retry(self.doc_client, request)
         except ResourceExhausted as e:
             log.error(f"DiscoveryEngine Operation failed after retries due to quota exceeded: {e}")
@@ -610,6 +614,7 @@ class DiscoveryEngineClient:
             reconciliation_mode=discoveryengine.ImportDocumentsRequest.ReconciliationMode.INCREMENTAL,
         )
 
+        log.debug(f"Making import_document_request: {request}")
         return self._import_document_request(request)
 
     def _create_unique_gsuri_docid(self, gcs_uri:str):
@@ -632,6 +637,7 @@ class DiscoveryEngineClient:
             str: The operation name.
         """
         try:
+            log.info(f"Importing doc with metadata: {gcs_uri=}, {metadata=}")
             # 1. Generate a unique document ID
             document_id = self._create_unique_gsuri_docid(gcs_uri)
 
@@ -662,7 +668,7 @@ class DiscoveryEngineClient:
             return self._import_document_request(request)
 
         except Exception as e:
-            log.error(f"Error importing document with metadata: {e}")
+            log.error(f"Error importing document with metadata: {e} {traceback.format_exc()}")
             raise e
     
     def get_mime_type(self, uri:str):

{sunholo-0.126.3 → sunholo-0.127.0}/src/sunholo/embedder/embed_metadata.py

@@ -6,7 +6,7 @@ from ..utils.mime import guess_mime_type
 
 from ..custom_logging import log
 
-def audit_metadata(metadata, chunk_length=None):
+def audit_metadata(metadata:dict, chunk_length:int=None) -> dict:
     
     if 'eventTime' not in metadata:
         metadata['eventTime'] = datetime.datetime.now().isoformat(timespec='microseconds') + "Z"

{sunholo-0.126.3 → sunholo-0.127.0}/src/sunholo/utils/config_class.py

@@ -112,18 +112,6 @@ class ConfigManager:
         """
         Helper function to load a config file and update the cache.
 
-        Args:
-            config_file (str): The path to the configuration file.
-            filename (str): The name of the configuration file.
-            is_local (bool): Indicates if the config file is from the local folder.
-
-        Returns:
-            dict: The loaded configuration.
-        """
-    def _reload_config_file(self, config_file, filename, is_local=False):
-        """
-        Helper function to load a config file and update the cache.
-
         Args:
             config_file (str): The path to the configuration file.
             filename (str): The name of the configuration file.
@@ -244,4 +232,23 @@ class ConfigManager:
             return agents[key]
         else:
             return agents.get("default")
+    
+    def permissionConfig(self, key: str):
+        """
+        Fetch a key from 'permissionConfig' kind configuration.
 
+        Args:
+            key (str): The key to fetch from the configuration.
+
+        Returns:
+            str: The value associated with the specified key.
+        """        
+        self._check_and_reload_configs()
+        config = self.configs_by_kind.get('permissionConfig')
+        if not config:
+            return None
+        agents = config.get('permissions')
+        if key in agents:
+            return agents[key]
+        else:
+            return None

{sunholo-0.126.3 → sunholo-0.127.0/src/sunholo.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sunholo
-Version: 0.126.3
+Version: 0.127.0
 Summary: Large Language Model DevOps - a package to help deploy LLMs to the Cloud.
 Author-email: Holosun ApS <multivac@sunholo.com>
 License: Apache License, Version 2.0
@@ -142,6 +142,7 @@ Requires-Dist: google-cloud-discoveryengine>=0.13.4; extra == "gcp"
 Requires-Dist: google-cloud-texttospeech; extra == "gcp"
 Requires-Dist: google-genai>=0.2.2; extra == "gcp"
 Requires-Dist: google-generativeai>=0.8.3; extra == "gcp"
+Requires-Dist: langchain; extra == "gcp"
 Requires-Dist: langchain-google-genai>=2.0.0; extra == "gcp"
 Requires-Dist: langchain_google_alloydb_pg>=0.2.2; extra == "gcp"
 Requires-Dist: langchain-google-vertexai; extra == "gcp"
@@ -164,6 +165,7 @@ Requires-Dist: flask; extra == "http"
 Requires-Dist: gunicorn; extra == "http"
 Requires-Dist: httpcore; extra == "http"
 Requires-Dist: httpx; extra == "http"
+Requires-Dist: langchain; extra == "http"
 Requires-Dist: langfuse; extra == "http"
 Requires-Dist: python-socketio; extra == "http"
 Requires-Dist: requests; extra == "http"

{sunholo-0.126.3 → sunholo-0.127.0}/src/sunholo.egg-info/requires.txt

@@ -117,6 +117,7 @@ google-cloud-discoveryengine>=0.13.4
 google-cloud-texttospeech
 google-genai>=0.2.2
 google-generativeai>=0.8.3
+langchain
 langchain-google-genai>=2.0.0
 langchain_google_alloydb_pg>=0.2.2
 langchain-google-vertexai
@@ -128,6 +129,7 @@ flask
 gunicorn
 httpcore
 httpx
+langchain
 langfuse
 python-socketio
 requests