sunholo 0.119.9__tar.gz → 0.119.11__tar.gz

{sunholo-0.119.9/src/sunholo.egg-info → sunholo-0.119.11}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: sunholo
-Version: 0.119.9
+Version: 0.119.11
 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
@@ -95,6 +95,7 @@ Requires-Dist: langchain; extra == "langchain"
 Requires-Dist: langchain_experimental; extra == "langchain"
 Requires-Dist: langchain-community; extra == "langchain"
 Requires-Dist: langsmith; extra == "langchain"
+Requires-Dist: langchain-unstructured; extra == "langchain"
 Provides-Extra: azure
 Requires-Dist: azure-identity; extra == "azure"
 Requires-Dist: azure-storage-blob; extra == "azure"

{sunholo-0.119.9 → sunholo-0.119.11}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "sunholo"
-version = "0.119.9"
+version = "0.119.11"
 description = "Large Language Model DevOps - a package to help deploy LLMs to the Cloud."
 readme = "README.md"
 requires-python = ">=3.10"
@@ -126,7 +126,8 @@ langchain = [
     "langchain",
     "langchain_experimental",
     "langchain-community",
-    "langsmith"
+    "langsmith",
+    "langchain-unstructured",
 ]
 azure = [
     "azure-identity",

{sunholo-0.119.9 → sunholo-0.119.11}/src/sunholo/chunker/loaders.py

@@ -160,9 +160,8 @@ def read_gdrive_to_document(url: str, metadata: dict = None):
     return docs
 
 def read_url_to_document(url: str, metadata: dict = None):
+    from langchain_unstructured import UnstructuredLoader
 
-    if not  UnstructuredLoader:
-        raise ImportError("UnstructuredLoader requires 'langchain_unstructured' to be installed")
     unstructured_kwargs = {"skip_infer_table_types": [],
                             "extract_image_block_types":  ["Image", "Table"]
                             }
@@ -179,7 +178,7 @@ def read_url_to_document(url: str, metadata: dict = None):
     return docs
 
 def read_file_to_documents(gs_file: pathlib.Path, metadata: dict = None):
-    
+    from langchain_unstructured import UnstructuredLoader
     docs = []
     pdf_path = str(pathlib.Path(gs_file))
 
@@ -276,7 +275,7 @@ def read_file_to_documents(gs_file: pathlib.Path, metadata: dict = None):
     return docs
 
 def convert_to_txt_and_extract(gs_file, split=False):
-
+    from langchain_unstructured import UnstructuredLoader
     if not  UnstructuredLoader:
         raise ImportError("UnstructuredLoader requires 'langchain_unstructured' to be installed")
     

{sunholo-0.119.9 → sunholo-0.119.11}/src/sunholo/discovery_engine/get_ai_search_chunks.py

@@ -5,12 +5,13 @@ from .discovery_engine_client import DiscoveryEngineClient
 from ..components import load_memories
 import traceback
 
-def get_all_chunks(question:str, config:ConfigManager):
+def get_all_chunks(question:str, config:ConfigManager, filter_str=None):
     """
     Look through a config memory key and find all Vertex AI Search retrievers, call them and return a joined string of chunks
 
         args: question - question to search similarity for
         config: A ConfigManager object
+        filter_str: A filter that will restrict ai search via its metadata. See https://cloud.google.com/generative-ai-app-builder/docs/filter-search-metadata
 
         returns: a big string of chunks
     """
@@ -39,7 +40,7 @@ def get_all_chunks(question:str, config:ConfigManager):
                 project_id = gcp_config.get('project_id')
                 serving_config = value.get('serving_config')
 
-                chunk = get_chunks(question, vector_name, num_chunks, project_id=project_id, serving_config=serving_config)
+                chunk = get_chunks(question, vector_name, num_chunks, filter_str=filter_str, project_id=project_id, serving_config=serving_config)
                 if chunk:
                     chunks.append(chunk)
     if chunks:
@@ -48,23 +49,30 @@ def get_all_chunks(question:str, config:ConfigManager):
     log.warning(f"No chunks found for {vector_name}")
     return None
 
-def get_chunks(question, vector_name, num_chunks, project_id=None, serving_config=None):
+def get_chunks(question, vector_name, num_chunks, filter_str=None, project_id=None, serving_config=None):
     if serving_config is None:
         serving_config = "default_serving_config"
     de = DiscoveryEngineClient(vector_name, project_id=project_id or get_gcp_project(include_config=True))
     try:
-        return de.get_chunks(question, num_previous_chunks=num_chunks, num_next_chunks=num_chunks, serving_config=serving_config)
+        if filter_str:
+            return de.search_with_filters(query=question, 
+                                         filter_str=filter_str,
+                                         num_previous_chunks=num_chunks, 
+                                         num_next_chunks=num_chunks)
+        else:
+            return de.get_chunks(question, num_previous_chunks=num_chunks, num_next_chunks=num_chunks, serving_config=serving_config)
     except Exception as err:
-        log.error(f"No discovery engine chunks found: {str(err)}")
+        log.error(f"No discovery engine chunks found: {str(err)} {traceback.format_exc()}")
     
 
 
-async def async_get_all_chunks(question:str, config:ConfigManager):
+async def async_get_all_chunks(question:str, config:ConfigManager, filter_str=None):
     """
     Look through a config memory key and find all Vertex AI Search retrievers, call them and return a joined string of chunks
 
         args: question - question to search similarity for
         config: A ConfigManager object
+        filter_str: A filter that will restrict ai search via its metadata. See https://cloud.google.com/generative-ai-app-builder/docs/filter-search-metadata
 
         returns: a big string of chunks
     """
@@ -87,10 +95,16 @@ async def async_get_all_chunks(question:str, config:ConfigManager):
                         continue
                     else:
                         vector_name = new_vector_name
-                
+
+                project_id = value.get('project_id') or get_gcp_project(include_config=True)
                 num_chunks = value.get('num_chunks') or 3
 
-                chunk = await async_get_chunks(question, vector_name, num_chunks)
+                chunk = await async_get_chunks(question, 
+                                               vector_name=vector_name, 
+                                               num_chunks=num_chunks, 
+                                               filter_str=filter_str, 
+                                               project_id=project_id)
+
                 if chunk:
                     chunks.append(chunk)
     if chunks:
@@ -99,9 +113,12 @@ async def async_get_all_chunks(question:str, config:ConfigManager):
     log.warning(f"No chunks found for {vector_name}")
     return None
 
-async def async_get_chunks(question, vector_name, num_chunks):
-    de = DiscoveryEngineClient(vector_name, project_id=get_gcp_project(include_config=True))
+async def async_get_chunks(question, vector_name, num_chunks, filter_str, project_id=None):
+    de = DiscoveryEngineClient(vector_name, project_id=project_id)
     try:
-        return await de.async_get_chunks(question, num_previous_chunks=num_chunks, num_next_chunks=num_chunks)
+        return await de.search_with_filters(query=question, 
+                                            filter_str=filter_str,
+                                            num_previous_chunks=num_chunks, 
+                                            num_next_chunks=num_chunks)
     except Exception as err:
         log.error(f"No discovery engine chunks found: {str(err)} {traceback.format_exc()}")

{sunholo-0.119.9 → sunholo-0.119.11/src/sunholo.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: sunholo
-Version: 0.119.9
+Version: 0.119.11
 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
@@ -95,6 +95,7 @@ Requires-Dist: langchain; extra == "langchain"
 Requires-Dist: langchain_experimental; extra == "langchain"
 Requires-Dist: langchain-community; extra == "langchain"
 Requires-Dist: langsmith; extra == "langchain"
+Requires-Dist: langchain-unstructured; extra == "langchain"
 Provides-Extra: azure
 Requires-Dist: azure-identity; extra == "azure"
 Requires-Dist: azure-storage-blob; extra == "azure"

{sunholo-0.119.9 → sunholo-0.119.11}/src/sunholo.egg-info/requires.txt

@@ -137,6 +137,7 @@ langchain
 langchain_experimental
 langchain-community
 langsmith
+langchain-unstructured
 
 [openai]
 langchain-openai>=0.3.2