sunholo 0.62.0__tar.gz → 0.62.2__tar.gz

{sunholo-0.62.0 → sunholo-0.62.2}/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: sunholo
-Version: 0.62.0
+Version: 0.62.2
 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/v0.62.0.tar.gz
+Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.62.2.tar.gz
 Author: Holosun ApS
 Author-email: multivac@sunholo.com
 License: Apache License, Version 2.0

{sunholo-0.62.0 → sunholo-0.62.2}/setup.py

@@ -1,7 +1,7 @@
 from setuptools import setup, find_packages
 
 # Define your base version
-version = '0.62.0'
+version = '0.62.2'
 
 setup(
     name='sunholo',

{sunholo-0.62.0 → sunholo-0.62.2}/sunholo/cli/embedder.py

@@ -9,6 +9,7 @@ from .sun_rich import console
 from rich.progress import Progress
 
 from .chat_vac import resolve_service_url, invoke_vac
+from .run_proxy import stop_proxy
 
 def create_metadata(vac, metadata):
     now_utc = datetime.now(timezone.utc)
@@ -103,8 +104,8 @@ def embed_command(args):
             json_data = encode_data(args.vac_name, args.data, args.metadata, args.local_chunks)
             chunk_res = invoke_vac(f"{chunk_url}/pubsub_to_store", json_data)
         
+        stop_proxy("chunker")
         if args.only_chunk:
-
             return chunk_res
 
         if not args.local_chunks:
@@ -124,13 +125,13 @@ def embed_command(args):
     
     if not chunk_res:
         console.print(f"[bold red]ERROR: Did not get any chunks from {chunk_url} for {args.data}")
-
+        stop_proxy("embedder")
         return
 
     chunks = chunk_res.get('chunks')
     if not chunks:
         console.print(f"[bold red]ERROR: No chunks found within json data: {str(chunk_res)} [/bold red]")
-
+        stop_proxy("embedder")
         return
     
     embeds = []
@@ -156,6 +157,7 @@ def embed_command(args):
             embeds.append(embed_res)
             progress.advance(task)
 
+    stop_proxy("embedder")
     console.rule("Embedding pipeline finished")
     
     return embed_res

sunholo-0.62.2/sunholo/llamaindex/get_files.py

@@ -0,0 +1,19 @@
+try:
+    from vertexai.preview import rag
+except ImportError:
+    rag = None
+
+# Create a RAG Corpus, Import Files
+def fetch_corpus(project_id, location, rag_id):
+    corpus_name = f"projects/{project_id}/locations/{location}/ragCorpora/{rag_id}"  
+
+    try:
+        return rag.get_corpus(name=corpus_name)
+    except Exception as err:
+        #log.warning(f"Failed to fetch corpus - creating new corpus {str(err)}")
+        # it does not create a unique corpus, display_name can be in multiple rag_ids
+        #try:
+        #    corpus = rag.create_corpus(display_name=vector_name, description=description)
+        #except Exception as err:
+        #    log.error(f"Failed to get or create corpus {str(err)}")
+        raise ValueError(f"Failed to get or create corpus: {str(err)}")

{sunholo-0.62.0 → sunholo-0.62.2}/sunholo/llamaindex/import_files.py

@@ -6,89 +6,7 @@ except ImportError:
 from ..logging import log
 from ..utils.config import load_config_key
 from ..vertex import init_vertex
-from ..components import load_memories
-
-# Create a RAG Corpus, Import Files
-
-def get_corpus(vector_name):
-    """
-    Retrieves a LlamaIndex corpus from Vertex AI based on the provided Google Cloud configuration.
-    
-    This function constructs a corpus name using project details from the configuration and attempts
-    to fetch the corresponding corpus. If the corpus cannot be retrieved, it raises an error.
-    
-    Parameters:
-    - vector_name: The name of the of VAC
-    
-    Returns:
-    - List of corpus objects fetched from Vertex AI.
-    
-    Raises:
-    - ValueError: If any of the required configurations (project_id, location, or rag_id) are missing,
-      or if the corpus cannot be retrieved.
-
-    Example:
-    ```python
-
-    # Fetch the corpus
-    try:
-        corpus = get_corpus("edmonbrain")
-        print("Corpus fetched successfully:", corpus)
-    except ValueError as e:
-        print("Error fetching corpus:", str(e))
-    ```
-    """
-    gcp_config = load_config_key("gcp_config", vector_name=vector_name, type="vacConfig")
-
-    if not rag:
-        raise ValueError("Need to install vertexai module via `pip install sunholo[gcp]`")
-    
-    global_project_id = gcp_config.get('project_id')
-    global_location = gcp_config.get('location')
-    global_rag_id = gcp_config.get('rag_id')
-
-    memories = load_memories(vector_name)
-    corpii = []
-    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}")
-                rag_id = value.get('rag_id')
-                project_id = gcp_config.get('project_id')
-                location = gcp_config.get('location')
-                corpus = fetch_corpus(
-                    project_id=project_id or global_project_id,
-                    location=location or global_location,
-                    rag_id=rag_id or global_rag_id
-                )
-                corpii.append(corpus)
-                
-    if not project_id or not global_project_id:
-        raise ValueError("Need config.gcp.project_id or config.memory.llamaindex.project_id to configure llamaindex on VertexAI")
-    if not rag_id or not global_project_id:
-        raise ValueError("Need config.gcp.rag_id or config.memory.llamaindex.rag_id to configure llamaindex on VertexAI.  Create via `rag.create_corpus(display_name=vector_name, description=description)`")
-    
-    if not corpii:
-        raise ValueError("No llamaindex Vertex corpus configurations could be found")
-    
-    return corpii
-
-
-def fetch_corpus(project_id, location, rag_id):
-    corpus_name = f"projects/{project_id}/locations/{location}/ragCorpora/{rag_id}"  
-
-    try:
-        return rag.get_corpus(name=corpus_name)
-    except Exception as err:
-        #log.warning(f"Failed to fetch corpus - creating new corpus {str(err)}")
-        # it does not create a unique corpus, display_name can be in multiple rag_ids
-        #try:
-        #    corpus = rag.create_corpus(display_name=vector_name, description=description)
-        #except Exception as err:
-        #    log.error(f"Failed to get or create corpus {str(err)}")
-        raise ValueError(f"Failed to get or create corpus: {str(err)}")
+from .get_files import fetch_corpus
     
 
 def do_llamaindex(message_data, metadata, vector_name):
@@ -127,7 +45,7 @@ def do_llamaindex(message_data, metadata, vector_name):
         raise ValueError(f"Need config.{vector_name}.gcp_config to configure llamaindex on VertexAI")
 
     init_vertex(gcp_config)
-    corpus = get_corpus(gcp_config)
+    corpus = fetch_corpus(gcp_config)
     #display_name = load_config_key("display_name", vector_name=vector_name, filename="config/llm_config.yaml")
     #description = load_config_key("description", vector_name=vector_name, filename="config/llm_config.yaml")
 

sunholo-0.62.2/sunholo/vertex/__init__.py

@@ -0,0 +1,2 @@
+from .init import init_vertex
+from .memory_tools import get_vertex_memories

sunholo-0.62.2/sunholo/vertex/memory_tools.py

@@ -0,0 +1,91 @@
+try:
+    from vertexai.preview import rag
+    from vertexai.preview.generative_models import Tool, grounding
+except ImportError:
+    rag = None
+
+from ..logging import log
+from ..utils.config import load_config_key
+from ..components import load_memories
+from ..llamaindex.get_files import fetch_corpus
+
+def get_vertex_memories(vector_name):
+    """
+    Retrieves a LlamaIndex corpus from Vertex AI based on the provided Google Cloud configuration.
+    
+    This function constructs a corpus name using project details from the configuration and attempts
+    to fetch the corresponding corpus. If the corpus cannot be retrieved, it raises an error.
+    
+    Parameters:
+    - vector_name: The name of the of VAC
+    
+    Returns:
+    - List of corpus objects fetched from Vertex AI.
+    
+    Raises:
+    - ValueError: If any of the required configurations (project_id, location, or rag_id) are missing,
+      or if the corpus cannot be retrieved.
+
+    Example:
+    ```python
+
+    # Fetch the corpus
+    try:
+        corpus = get_corpus("edmonbrain")
+        print("Corpus fetched successfully:", corpus)
+    except ValueError as e:
+        print("Error fetching corpus:", str(e))
+    ```
+    """
+    gcp_config = load_config_key("gcp_config", vector_name=vector_name, type="vacConfig")
+
+    if not rag:
+        raise ValueError("Need to install vertexai module via `pip install sunholo[gcp]`")
+    
+    global_project_id = gcp_config.get('project_id')
+    global_location = gcp_config.get('location')
+    global_rag_id = gcp_config.get('rag_id')
+    global_data_store_id = gcp_config.get('data_store_id')
+
+    memories = load_memories(vector_name)
+    tools = []
+    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}")
+                rag_id = value.get('rag_id')
+                project_id = gcp_config.get('project_id')
+                location = gcp_config.get('location')
+                corpus = fetch_corpus(
+                    project_id=project_id or global_project_id,
+                    location=location or global_location,
+                    rag_id=rag_id or global_rag_id
+                )
+                corpus_tool = Tool.from_retrieval(
+                    retrieval=rag.Retrieval(
+                        source=rag.VertexRagStore(
+                            rag_corpora=[corpus.name],  # Currently only 1 corpus is allowed.
+                            similarity_top_k=10,  # Optional
+                        ),
+                    )
+                )
+                tools.append(corpus_tool)
+            elif vectorstore == "vertexai_agent_builder":
+                log.info(f"Found vectorstore {vectorstore}")
+                data_store_id = value.get('data_store_id') or global_data_store_id
+                project_id = gcp_config.get('project_id') or global_project_id
+                location = gcp_config.get('location') or global_location
+                data_store_path=f"projects/{project_id}/locations/{location}/collections/default_collection/dataStores/{data_store_id}"
+
+                corpus_tool = Tool.from_retrieval(
+                    grounding.Retrieval(grounding.VertexAISearch(datastore=data_store_path))
+                )
+                tools.append(corpus_tool)
+                
+
+    if not tools:
+        log.warning("No llamaindex Vertex corpus configurations could be found")
+    
+    return tools

{sunholo-0.62.0 → sunholo-0.62.2}/sunholo.egg-info/PKG-INFO

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: sunholo
-Version: 0.62.0
+Version: 0.62.2
 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/v0.62.0.tar.gz
+Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.62.2.tar.gz
 Author: Holosun ApS
 Author-email: multivac@sunholo.com
 License: Apache License, Version 2.0

{sunholo-0.62.0 → sunholo-0.62.2}/sunholo.egg-info/SOURCES.txt

@@ -79,6 +79,7 @@ sunholo/langfuse/callback.py
 sunholo/langfuse/prompts.py
 sunholo/llamaindex/__init__.py
 sunholo/llamaindex/generate.py
+sunholo/llamaindex/get_files.py
 sunholo/llamaindex/import_files.py
 sunholo/lookup/__init__.py
 sunholo/lookup/model_lookup.yaml
@@ -106,6 +107,7 @@ sunholo/utils/gcp.py
 sunholo/utils/parsers.py
 sunholo/utils/user_ids.py
 sunholo/vertex/__init__.py
-sunholo/vertex/init_vertex.py
+sunholo/vertex/init.py
+sunholo/vertex/memory_tools.py
 tests/test_chat_history.py
 tests/test_config.py