sunholo 0.119.9__py3-none-any.whl → 0.119.11__py3-none-any.whl

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/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.dist-info → sunholo-0.119.11.dist-info}/METADATA

@@ -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.dist-info → sunholo-0.119.11.dist-info}/RECORD

@@ -35,7 +35,7 @@ sunholo/chunker/azure.py,sha256=MVF9_-QdKUoJqlpEJ49pv2sdjMDxEiMNxzmO7w5nWDQ,3270
 sunholo/chunker/doc_handling.py,sha256=t_lDazHfJbs4Q2Ruq2MvBBeJRfsjjQkzMxKuX8qQKBI,9087
 sunholo/chunker/encode_metadata.py,sha256=hxxd9KU35Xi0Z_EL8kt_oD66pKfBLhEjBImC16ew-Eo,1919
 sunholo/chunker/images.py,sha256=id2PBu6XyGEOtgafq2v0c9_O6kxaC_pYFMnbsIitkSg,1868
-sunholo/chunker/loaders.py,sha256=5NXrMxV-WdbFpxeLhFzccw0_zhf1UQ7yKFFeaMkc9Bc,11105
+sunholo/chunker/loaders.py,sha256=KEFPHBr32DPJnRUNjOecLiZKsFD-Gk3BFIKyXSVHgbY,11143
 sunholo/chunker/message_data.py,sha256=bpb8QWQttqazm5lr7fTFJ5JDwf-P0SQ5cOIf6NikNyI,10836
 sunholo/chunker/pdfs.py,sha256=xwbuMJrbypcyPXfZ8tiUidWeMr80C2NhfTC1mwa8SHY,2477
 sunholo/chunker/process_chunker_data.py,sha256=xjOAf1FvHDwQaBm7kgDzLQUnwm6AW8qf4fTrwDnwmtc,3613
@@ -76,7 +76,7 @@ sunholo/discovery_engine/chunker_handler.py,sha256=CYsyZLz_VZ0S_TcVlYmPxYz3JiAiP
 sunholo/discovery_engine/cli.py,sha256=KGVle5rkLL49oF9TQhrGI--8017IvvLOEoYur545Qb0,12790
 sunholo/discovery_engine/create_new.py,sha256=WUi4_xh_dFaGX3xA9jkNKZhaR6LCELjMPeRb0hyj4FU,1226
 sunholo/discovery_engine/discovery_engine_client.py,sha256=0h3h_Vy8vCcvfSBhGl8qSgVzDakOmsaPPFqHIGh3mv8,33053
-sunholo/discovery_engine/get_ai_search_chunks.py,sha256=hsFGOQugSeTMPEaQ16XTs_D45F8NABBm2IsAEdTk7kQ,4316
+sunholo/discovery_engine/get_ai_search_chunks.py,sha256=IDb9cm4X5XqQbUytZVQrBsm_oiSQxe3KjBAxaSZ3JyQ,5548
 sunholo/embedder/__init__.py,sha256=sI4N_CqgEVcrMDxXgxKp1FsfsB4FpjoXgPGkl4N_u4I,44
 sunholo/embedder/embed_chunk.py,sha256=did2pKkWM2o0KkRcb0H9l2x_WjCq6OyuHDxGbITFKPM,6530
 sunholo/embedder/embed_metadata.py,sha256=2ziUIdVwnbCUU8gOwQWEvkrRcyp-7IeyZfSsWNkMquA,866
@@ -166,9 +166,9 @@ sunholo/vertex/init.py,sha256=1OQwcPBKZYBTDPdyU7IM4X4OmiXLdsNV30C-fee2scQ,2875
 sunholo/vertex/memory_tools.py,sha256=tBZxqVZ4InTmdBvLlOYwoSEWu4-kGquc-gxDwZCC4FA,7667
 sunholo/vertex/safety.py,sha256=S9PgQT1O_BQAkcqauWncRJaydiP8Q_Jzmu9gxYfy1VA,2482
 sunholo/vertex/type_dict_to_json.py,sha256=uTzL4o9tJRao4u-gJOFcACgWGkBOtqACmb6ihvCErL8,4694
-sunholo-0.119.9.dist-info/LICENSE.txt,sha256=SdE3QjnD3GEmqqg9EX3TM9f7WmtOzqS1KJve8rhbYmU,11345
-sunholo-0.119.9.dist-info/METADATA,sha256=XDQ0g73OJrI5qnlk8SuhbYxaUPF7yuAZQyMUUqkZ0_4,9654
-sunholo-0.119.9.dist-info/WHEEL,sha256=nn6H5-ilmfVryoAQl3ZQ2l8SH5imPWFpm1A5FgEuFV4,91
-sunholo-0.119.9.dist-info/entry_points.txt,sha256=bZuN5AIHingMPt4Ro1b_T-FnQvZ3teBes-3OyO0asl4,49
-sunholo-0.119.9.dist-info/top_level.txt,sha256=wt5tadn5--5JrZsjJz2LceoUvcrIvxjHJe-RxuudxAk,8
-sunholo-0.119.9.dist-info/RECORD,,
+sunholo-0.119.11.dist-info/LICENSE.txt,sha256=SdE3QjnD3GEmqqg9EX3TM9f7WmtOzqS1KJve8rhbYmU,11345
+sunholo-0.119.11.dist-info/METADATA,sha256=gj6We8d7T27etGlhriYOvyJmNLF-bX0hbHui6sc_zmo,9715
+sunholo-0.119.11.dist-info/WHEEL,sha256=nn6H5-ilmfVryoAQl3ZQ2l8SH5imPWFpm1A5FgEuFV4,91
+sunholo-0.119.11.dist-info/entry_points.txt,sha256=bZuN5AIHingMPt4Ro1b_T-FnQvZ3teBes-3OyO0asl4,49
+sunholo-0.119.11.dist-info/top_level.txt,sha256=wt5tadn5--5JrZsjJz2LceoUvcrIvxjHJe-RxuudxAk,8
+sunholo-0.119.11.dist-info/RECORD,,