sunholo 0.119.7__tar.gz → 0.119.9__tar.gz

{sunholo-0.119.7/src/sunholo.egg-info → sunholo-0.119.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: sunholo
-Version: 0.119.7
+Version: 0.119.9
 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

{sunholo-0.119.7 → sunholo-0.119.9}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "sunholo"
-version = "0.119.7"
+version = "0.119.9"
 description = "Large Language Model DevOps - a package to help deploy LLMs to the Cloud."
 readme = "README.md"
 requires-python = ">=3.10"

{sunholo-0.119.7 → sunholo-0.119.9}/src/sunholo/chunker/process_chunker_data.py

@@ -38,7 +38,7 @@ def process_chunker_data(message_data, metadata, vector_name):
         return llamacheck
     
     # if only a discovery engine memory, return early as no other processing needed
-    discovery_check = discovery_engine_chunker_check(message_data, metadata, config=config)
+    discovery_check = discovery_engine_chunker_check(message_data, metadata, config=config, process=False)
     if discovery_check:
         return discovery_check
 

{sunholo-0.119.7 → sunholo-0.119.9}/src/sunholo/discovery_engine/chunker_handler.py

@@ -125,7 +125,11 @@ def check_write_memories(config:ConfigManager):
     
     return write_mem
 
-def discovery_engine_chunker_check(message_data, metadata, vector_name:str=None, config:ConfigManager=None):
+def discovery_engine_chunker_check(message_data, 
+                                   metadata, 
+                                   vector_name:str=None, 
+                                   config:ConfigManager=None,
+                                   process:bool=True):
 
     if config is None:
         if vector_name is None:
@@ -139,7 +143,7 @@ def discovery_engine_chunker_check(message_data, metadata, vector_name:str=None,
     
     total_memories = len(check_write_memories(config))
     llama = None
-    if check_discovery_engine_in_memory(config):
+    if check_discovery_engine_in_memory(config) and process:
         llama = do_discovery_engine(message_data, metadata, config=config)
         log.info(f"Processed discovery engine: {llama}")
 

{sunholo-0.119.7 → sunholo-0.119.9}/src/sunholo/discovery_engine/cli.py

@@ -223,12 +223,11 @@ def setup_discovery_engine_subparser(subparsers):
     # Search subcommand
     search_parser = discovery_engine_subparsers.add_parser('search', help='Search a Discovery Engine datastore')
     search_parser.add_argument('--query', required=True, help='The search query')
-    search_parser.add_argument('--num-previous-chunks', type=int, default=3, help='Number of previous chunks to return for context')
-    search_parser.add_argument('--num-next-chunks', type=int, default=3, help='Number of next chunks to return for context')
+    search_parser.add_argument('--data-store-id', required=True, help='Data store ID to search')
     search_parser.add_argument('--page-size', type=int, default=10, help='The maximum number of results to return per page')
     search_parser.add_argument('--parse-chunks-to-string', action='store_true', help='Combine chunks into a single string')
     search_parser.add_argument('--serving-config', default='default_serving_config', help='The serving configuration to use')
-    search_parser.add_argument('--data-store-ids', nargs='+', help='List of data store IDs to search (optional)')
+
     search_parser.set_defaults(func=discovery_engine_command)
 
     # Search by ID and/or Date subcommand

{sunholo-0.119.7 → sunholo-0.119.9}/src/sunholo/discovery_engine/discovery_engine_client.py

@@ -607,6 +607,12 @@ class DiscoveryEngineClient:
 
         return self._import_document_request(request)
 
+    def _create_unique_gsuri_docid(self, gcs_uri:str):
+        import hashlib
+        # Create SHA-256 hash of the URI
+        hash_object = hashlib.sha256(gcs_uri.encode())
+        # Take first 16 bytes (128 bits) and encode as hex
+        return hash_object.hexdigest()[:32]
     
     def import_document_with_metadata(self, gcs_uri: str, metadata: dict, branch="default_branch"):
         """
@@ -622,7 +628,7 @@ class DiscoveryEngineClient:
         """
         try:
             # 1. Generate a unique document ID
-            document_id = str(uuid.uuid4())
+            document_id = self._create_unique_gsuri_docid(gcs_uri)
 
             # 2. Create a Document object
             parent = self.doc_client.branch_path(

{sunholo-0.119.7 → sunholo-0.119.9/src/sunholo.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: sunholo
-Version: 0.119.7
+Version: 0.119.9
 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