sunholo 0.80.4__tar.gz → 0.80.5__tar.gz

{sunholo-0.80.4 → sunholo-0.80.5}/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: sunholo
-Version: 0.80.4
+Version: 0.80.5
 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/v0.80.4.tar.gz
+Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.80.5.tar.gz
 Author: Holosun ApS
 Author-email: multivac@sunholo.com
 License: Apache License, Version 2.0

{sunholo-0.80.4 → sunholo-0.80.5}/setup.py

@@ -1,7 +1,7 @@
 from setuptools import setup, find_packages
 
 # Define your base version
-version = '0.80.4'
+version = '0.80.5'
 
 setup(
     name='sunholo',

{sunholo-0.80.4 → sunholo-0.80.5}/sunholo/components/retriever.py

@@ -60,6 +60,9 @@ def pick_retriever(vector_name:str=None, config:ConfigManager=None, embeddings=N
 
                 if vectorstore == "vertex_ai_search" or vectorstore == "discovery_engine":
                     # use direct retriever
+                    if value.get('chunks'):
+                        log.warning(f"{config.vector_name} will not be using GoogleVertexAISearchRetriever with chunks vertex AI search as not supported yet")
+                        continue
                     from langchain.retrievers import GoogleVertexAISearchRetriever
                     gcp_config = config.vacConfig('gcp_config')
                     try:
@@ -76,6 +79,7 @@ def pick_retriever(vector_name:str=None, config:ConfigManager=None, embeddings=N
                         continue
                     
                     retriever_list.append(gcp_retriever)
+                    continue
 
                 from_metadata_id = value.get('from_metadata_id')
                 if from_metadata_id:

{sunholo-0.80.4 → sunholo-0.80.5}/sunholo/discovery_engine/chunker_handler.py

@@ -1,5 +1,5 @@
 from ..custom_logging import log
-from ..utils.config import load_config_key
+from ..utils import load_config_key, ConfigManager
 from ..utils.gcp_project import get_gcp_project
 from ..components import load_memories
 
@@ -86,8 +86,9 @@ def do_discovery_engine(message_data, metadata, vector_name):
         log.warning("Only gs:// data is supported for Discovery Engine")
 
 
-def check_discovery_engine_in_memory(vector_name):
-    memories = load_config_key("memory", vector_name=vector_name, kind="vacConfig")
+def check_discovery_engine_in_memory(config:ConfigManager):
+    memories = config.vacConfig("memory")
+
     for memory in memories:  # Iterate over the list
         for key, value in memory.items():  # Now iterate over the dictionary
             log.info(f"Found memory {key}")
@@ -99,15 +100,32 @@ def check_discovery_engine_in_memory(vector_name):
     
     return False
 
-def discovery_engine_chunker_check(message_data, metadata, vector_name):
+def check_write_memories(config:ConfigManager):
+    write_mem = []
+    memories = config.vacConfig("memory")
+    for memory in memories:
+        for key, value in memory.items():
+            if value.get('read_only'):
+                continue
+            write_mem.append(memory)
+    
+    return write_mem
+
+def discovery_engine_chunker_check(message_data, metadata, vector_name:str=None, config:ConfigManager=None):
+
+    if config is None:
+        if vector_name is None:
+            raise ValueError("Must provide config or vector_name")
+        config = ConfigManager(vector_name=vector_name)
+
     # discovery engine handles its own chunking/embedding
-    memories = load_config_key("memory", vector_name=vector_name, kind="vacConfig")
+    memories = config.vacConfig("memory")
     if not memories:
         return None
     
-    total_memories = len(memories)
+    total_memories = len(check_write_memories(config))
     llama = None
-    if check_discovery_engine_in_memory(vector_name):
+    if check_discovery_engine_in_memory(config):
         llama = do_discovery_engine(message_data, metadata, vector_name)
         log.info(f"Processed discovery engine: {llama}")
 

{sunholo-0.80.4 → sunholo-0.80.5}/sunholo/gcs/add_file.py

@@ -118,12 +118,13 @@ def add_file_to_gcs(filename: str,
     if os.getenv('EXTENSIONS_BUCKET'):
         bucket_filepath = os.path.basename(filename)
 
-    if vector_name is None:
-            vector_name = "global"
+    if not vector_name:
+        vector_name = "global"
     
     if not bucket_filepath:
         
         bucket_filepath = f"{vector_name}/{year}/{month}/{day}/{hour}/{os.path.basename(filename)}"
+    
     bucket_filepath_prev = f"{vector_name}/{year}/{month}/{day}/{hour_prev}/{os.path.basename(filename)}"
 
     blob = bucket.blob(bucket_filepath)

{sunholo-0.80.4 → sunholo-0.80.5}/sunholo.egg-info/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: sunholo
-Version: 0.80.4
+Version: 0.80.5
 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/v0.80.4.tar.gz
+Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.80.5.tar.gz
 Author: Holosun ApS
 Author-email: multivac@sunholo.com
 License: Apache License, Version 2.0