sunholo 0.69.9__py3-none-any.whl → 0.69.12__py3-none-any.whl

sunholo/discovery_engine/chunker_handler.py

@@ -51,6 +51,8 @@ def do_discovery_engine(message_data, metadata, vector_name):
 
     if message_data.startswith("gs://"):
         log.info(f"DiscoveryEngineClient.import_files for {message_data}")
+        if "/pdf_parts/" in message_data:
+            return None
         for corp in corpuses:
             try:
                 response = corp.import_documents(

sunholo/discovery_engine/discovery_engine_client.py

@@ -155,6 +155,7 @@ class DiscoveryEngineClient:
         num_previous_chunks: int = 3,
         num_next_chunks: int = 3,
         page_size: int = 10,
+        parse_chunks_to_string: bool = True,
         doc_or_chunks: str = "CHUNKS",  # or DOCUMENTS
         serving_config: str = "default_serving_config",
     ):
@@ -166,6 +167,7 @@ class DiscoveryEngineClient:
             num_previous_chunks (int, optional): Number of previous chunks to return for context (default is 3).
             num_next_chunks (int, optional): Number of next chunks to return for context (default is 3).
             page_size (int, optional): The maximum number of results to return per page (default is 10).
+            parse_chunks_to_string: If True will put chunks in one big string, False will return object
 
         Returns:
             discoveryengine.SearchResponse: The search response object containing the search results.
@@ -199,10 +201,69 @@ class DiscoveryEngineClient:
             ),
         )
 
+        log.info(f"Discovery engine request: {search_request=}")
         search_response = self.search_client.search(search_request)
+        
+
+        if parse_chunks_to_string:
+
+            big_string = self.process_chunks(search_response)
+            log.info(f"Discovery engine chunks string sample: {big_string[:100]}")
 
+            return big_string
+        
+        log.info("Discovery engine response object")
         return search_response
 
+    def process_chunks(self, response):
+        all_chunks = []
+
+        if 'results' not in response:
+            raise ValueError(f'No results found in response: {response=}')
+        
+        for result in response['results']:
+            chunk = result['chunk']
+            chunk_metadata = chunk['chunkMetadata']
+
+            if 'previousChunks' in chunk_metadata:
+                # Process previous chunks
+                for prev_chunk in chunk['chunkMetadata']['previousChunks']:
+                    prev_chunk_string = (
+                        f"# {prev_chunk['id']}\n"
+                        f"{prev_chunk['content']}\n"
+                        f"## metadata\n"
+                        f"Document URI: {prev_chunk['documentMetadata']['uri']}\n"
+                        f"Document Title: {prev_chunk['documentMetadata']['title']}\n"
+                    )
+                    all_chunks.append(prev_chunk_string)
+
+            # Process fetched chunk
+            fetched_chunk_string = (
+                f"# {chunk['id']}\n"
+                f"{chunk['content']}\n"
+                f"## metadata\n"
+                f"Document URI: {chunk['documentMetadata']['uri']}\n"
+                f"Document Title: {chunk['documentMetadata']['title']}\n"
+            )
+            all_chunks.append(fetched_chunk_string)
+
+            # Process next chunks
+            if 'nextChunks' in chunk_metadata:
+                for next_chunk in chunk_metadata['nextChunks']:
+                    next_chunk_string = (
+                        f"# {next_chunk['id']}\n"
+                        f"{next_chunk['content']}\n"
+                        f"## metadata\n"
+                        f"Document URI: {next_chunk['documentMetadata']['uri']}\n"
+                        f"Document Title: {next_chunk['documentMetadata']['title']}\n"
+                    )
+                    all_chunks.append(next_chunk_string)
+
+        # Combine all chunks into one long string
+        result_string = "\n".join(all_chunks)
+
+        return result_string
+
     def import_documents(self,
         gcs_uri: Optional[str] = None,
         data_schema="content",
@@ -256,9 +317,9 @@ class DiscoveryEngineClient:
         try:
             operation = import_documents_with_retry(self.doc_client, request)
         except ResourceExhausted as e:
-            print(f"Operation failed after retries due to quota exceeded: {e}")
+            log.error(f"Operation failed after retries due to quota exceeded: {e}")
         except Exception as e:
-            print(f"An unexpected error occurred: {e}")
+            log.error(f"An unexpected error occurred: {e}")
 
         return operation.operation.name
 

sunholo/llamaindex/import_files.py

@@ -61,7 +61,6 @@ def do_llamaindex(message_data, metadata, vector_name):
     corpuses = []
     for memory in memories:
         for key, value in memory.items():  # Now iterate over the dictionary
-            log.info(f"Found memory {key}")
             vectorstore = value.get('vectorstore')
             if vectorstore == "llamaindex":
                 log.info(f"Found vectorstore {vectorstore}")

sunholo/logging.py

@@ -119,8 +119,14 @@ class GoogleCloudLogging:
         if log_text:
             if isinstance(log_struct, dict):
                 logger.log_struct(log_struct, severity=severity, source_location=caller_info)
-            else:
+            elif isinstance(log_struct, str):
                 logger.log_text(log_text, severity=severity, source_location=caller_info)
+            else:
+                try:
+                    turn_to_text = str(log_text)
+                    logger.log_text(turn_to_text, severity=severity, source_location=caller_info)
+                except Exception as err:
+                    print(f"Could not log this: {log_text=} - {str(err)}")
 
         elif log_struct:
             if not isinstance(log_struct, dict):

{sunholo-0.69.9.dist-info → sunholo-0.69.12.dist-info}/METADATA

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: sunholo
-Version: 0.69.9
+Version: 0.69.12
 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/v0.69.9.tar.gz
+Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.69.12.tar.gz
 Author: Holosun ApS
 Author-email: multivac@sunholo.com
 License: Apache License, Version 2.0
@@ -61,6 +61,7 @@ Requires-Dist: pypdf ; extra == 'all'
 Requires-Dist: python-socketio ; extra == 'all'
 Requires-Dist: rich ; extra == 'all'
 Requires-Dist: supabase ; extra == 'all'
+Requires-Dist: tantivy ; extra == 'all'
 Requires-Dist: tiktoken ; extra == 'all'
 Provides-Extra: anthropic
 Requires-Dist: langchain-anthropic >=0.1.13 ; extra == 'anthropic'
@@ -75,6 +76,7 @@ Requires-Dist: pg8000 ; extra == 'database'
 Requires-Dist: pgvector ; extra == 'database'
 Requires-Dist: psycopg2-binary ; extra == 'database'
 Requires-Dist: lancedb ; extra == 'database'
+Requires-Dist: tantivy ; extra == 'database'
 Provides-Extra: gcp
 Requires-Dist: google-auth-httplib2 ; extra == 'gcp'
 Requires-Dist: google-auth-oauthlib ; extra == 'gcp'

{sunholo-0.69.9.dist-info → sunholo-0.69.12.dist-info}/RECORD

@@ -1,5 +1,5 @@
 sunholo/__init__.py,sha256=0CdpufyRKWyZe7J7UKigL6j_qOorM-p0OjHIAuf9M38,864
-sunholo/logging.py,sha256=00VGGArfWHbJuHHSJ4kXhHTggWnRfbVYMcZNOYIsqnA,11787
+sunholo/logging.py,sha256=UUBl0_oBrW21O5cNAT5lYZ2OmAnVJ92PnAwtA_2Sz_g,12117
 sunholo/agents/__init__.py,sha256=Hb4NXy2rN-83Z0-UDRwX-LXv2R29lcbSFPf8G6q4fZg,380
 sunholo/agents/chat_history.py,sha256=8iX1bgvRW6fdp6r_DQR_caPHYrZ_9QJJgPxCiSDf3q8,5380
 sunholo/agents/dispatch_to_qa.py,sha256=nFNdxhkr7rVYuUwVoBCBNYBI2Dke6-_z_ZApBEWb_cU,8291
@@ -61,9 +61,9 @@ sunholo/database/sql/sb/delete_source_row.sql,sha256=r6fEuUKdbiLHCDGKSbKINDCpJjs
 sunholo/database/sql/sb/return_sources.sql,sha256=89KAnxfK8n_qGK9jy1OQT8f9n4uYUtYL5cCxbC2mj_c,255
 sunholo/database/sql/sb/setup.sql,sha256=CvoFvZQev2uWjmFa3aj3m3iuPFzAAJZ0S7Qi3L3-zZI,89
 sunholo/discovery_engine/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-sunholo/discovery_engine/chunker_handler.py,sha256=9zv72CxgKlKpOcSAt-XFtq0Ra-Z6tBTwMuXQhRPE5mY,4425
+sunholo/discovery_engine/chunker_handler.py,sha256=puNnV6vKnjNqk28FLnsIPMNsC-1Y6s20PK7ioi8qwzc,4491
 sunholo/discovery_engine/create_new.py,sha256=7oZG78T6lW0EspRzlo7-qRyXFSuFxDn2dfSAVEaqlqY,978
-sunholo/discovery_engine/discovery_engine_client.py,sha256=UYMhXtGWXuyyg4fA1piVjqfrWXkSP_MFimh6NzaUlZ8,12413
+sunholo/discovery_engine/discovery_engine_client.py,sha256=WFi0h-JfZd0eKtW6uzm6zfffkGMP2iKBHZU1Tw1uiCw,14944
 sunholo/embedder/__init__.py,sha256=sI4N_CqgEVcrMDxXgxKp1FsfsB4FpjoXgPGkl4N_u4I,44
 sunholo/embedder/embed_chunk.py,sha256=P744zUQJgqrjILunzaqtTerB9AwoXFU6tXBtz4rjWgQ,6673
 sunholo/gcs/__init__.py,sha256=DtVw_AZwQn-IguR5BJuIi2XJeF_FQXizhJikzRNrXiE,50
@@ -76,7 +76,7 @@ sunholo/langfuse/prompts.py,sha256=HO4Zy9usn5tKooBPCKksuw4Lff3c03Ny5wqn4ce_xZM,1
 sunholo/llamaindex/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sunholo/llamaindex/generate.py,sha256=l1Picr-hVwkmAUD7XmTCa63qY9ERliFHQXwyX3BqB2Q,686
 sunholo/llamaindex/get_files.py,sha256=6rhXCDqQ_lrIapISQ_OYQDjiSATXvS_9m3qq53-oIl0,781
-sunholo/llamaindex/import_files.py,sha256=WPaVfYKS08GN0HXodPH_5WCs6d_fqj-N4MQIzvlz5Xw,5610
+sunholo/llamaindex/import_files.py,sha256=LIFBXxgXgRYMdVpq3FkycIoMMhLSXkKdKzK7qDuqOmQ,5566
 sunholo/lookup/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sunholo/lookup/model_lookup.yaml,sha256=O7o-jP53MLA06C8pI-ILwERShO-xf6z_258wtpZBv6A,739
 sunholo/patches/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -111,9 +111,9 @@ sunholo/vertex/__init__.py,sha256=JvHcGFuv6R_nAhY2AdoqqhMpJ5ugeWPZ_svGhWrObBk,13
 sunholo/vertex/init.py,sha256=JDMUaBRdednzbKF-5p33qqLit2LMsvgvWW-NRz0AqO0,1801
 sunholo/vertex/memory_tools.py,sha256=8F1iTWnqEK9mX4W5RzCVKIjydIcNp6OFxjn_dtQ3GXo,5379
 sunholo/vertex/safety.py,sha256=3meAX0HyGZYrH7rXPUAHxtI_3w_zoy_RX7Shtkoa660,1275
-sunholo-0.69.9.dist-info/LICENSE.txt,sha256=SdE3QjnD3GEmqqg9EX3TM9f7WmtOzqS1KJve8rhbYmU,11345
-sunholo-0.69.9.dist-info/METADATA,sha256=53jro7ekONplbxTNVWi4DTYXHawjUEnEuLNPi704EwI,6155
-sunholo-0.69.9.dist-info/WHEEL,sha256=mguMlWGMX-VHnMpKOjjQidIo1ssRlCFu4a4mBpz1s2M,91
-sunholo-0.69.9.dist-info/entry_points.txt,sha256=bZuN5AIHingMPt4Ro1b_T-FnQvZ3teBes-3OyO0asl4,49
-sunholo-0.69.9.dist-info/top_level.txt,sha256=wt5tadn5--5JrZsjJz2LceoUvcrIvxjHJe-RxuudxAk,8
-sunholo-0.69.9.dist-info/RECORD,,
+sunholo-0.69.12.dist-info/LICENSE.txt,sha256=SdE3QjnD3GEmqqg9EX3TM9f7WmtOzqS1KJve8rhbYmU,11345
+sunholo-0.69.12.dist-info/METADATA,sha256=w1SqSvez7MqPL3PlMn4PIrhlFDru7p8XLU_DPiE8_AI,6242
+sunholo-0.69.12.dist-info/WHEEL,sha256=mguMlWGMX-VHnMpKOjjQidIo1ssRlCFu4a4mBpz1s2M,91
+sunholo-0.69.12.dist-info/entry_points.txt,sha256=bZuN5AIHingMPt4Ro1b_T-FnQvZ3teBes-3OyO0asl4,49
+sunholo-0.69.12.dist-info/top_level.txt,sha256=wt5tadn5--5JrZsjJz2LceoUvcrIvxjHJe-RxuudxAk,8
+sunholo-0.69.12.dist-info/RECORD,,