sunholo 0.62.17__py3-none-any.whl → 0.63.0__py3-none-any.whl

sunholo/auth/run.py

@@ -53,26 +53,23 @@ def get_header(vector_name) -> Optional[dict]:
     else:
         run_url = "http://127.0.0.1:8080"
 
-    if "http://" in run_url:
-        return None
-    else:
-        # Append ID Token to make authenticated requests to Cloud Run services
-        frame = inspect.currentframe()
-        caller_frame = frame.f_back if frame is not None else None  # One level up in the stack
+    # Append ID Token to make authenticated requests to Cloud Run services
+    frame = inspect.currentframe()
+    caller_frame = frame.f_back if frame is not None else None  # One level up in the stack
+    deets = {
+        'message': 'Authenticating for run_url',
+        'run_url': run_url
+    }
+    if caller_frame:
         deets = {
-            'message': 'Authenticating for run_url',
-            'run_url': run_url
-        }
-        if caller_frame:
-            deets = {
-                    'message': 'Authenticating for run_url',
-                    'file': caller_frame.f_code.co_filename,
-                    'line': str(caller_frame.f_lineno),  
-                    'function': caller_frame.f_code.co_name,
-                    'run_url': run_url
-                }
-        log.info(f"Authenticating for run_url {run_url} from {caller_frame.f_code.co_name}")
-        id_token = get_id_token(run_url)
-        headers = {"Authorization": f"Bearer {id_token}"}
-        #log.info(f"id_token {id_token}")
-        return headers
+                'message': 'Authenticating for run_url',
+                'file': caller_frame.f_code.co_filename,
+                'line': str(caller_frame.f_lineno),  
+                'function': caller_frame.f_code.co_name,
+                'run_url': run_url
+            }
+    log.info(f"Authenticating for run_url {run_url} from {caller_frame.f_code.co_name}")
+    id_token = get_id_token(run_url)
+    headers = {"Authorization": f"Bearer {id_token}"}
+    #log.info(f"id_token {id_token}")
+    return headers

sunholo/cli/chat_vac.py

@@ -149,46 +149,72 @@ def headless_mode(service_url, service_name, user_input, chat_history=None, stre
     user_id = generate_user_id()
     session_id = str(uuid.uuid4())
 
-    def stream_response():
-        generate = generate_proxy_stream(
-                send_to_qa,
-                user_input,
-                vector_name=service_name,
-                chat_history=chat_history,
-                generate_f_output=lambda x: x,  # Replace with actual processing function
-                stream_wait_time=0.5,
-                stream_timeout=120,
-                message_author=user_id,
-                #TODO: populate these
-                image_url=None,
-                source_filters=None,
-                search_kwargs=None,
-                private_docs=None,
-                whole_document=False,
-                source_filters_and_or=False,
-                # system kwargs
-                configurable={
-                    "vector_name": service_name,
-                },
-                user_id=user_id,
-                session_id=session_id, 
-                message_source="cli",
-                override_endpoint=service_url
-        )
-        for part in generate():
-            yield part
+    if not stream:
+        vac_response = send_to_qa(user_input,
+            vector_name=service_name,
+            chat_history=chat_history,
+            message_author=user_id,
+            #TODO: populate these
+            image_url=None,
+            source_filters=None,
+            search_kwargs=None,
+            private_docs=None,
+            whole_document=False,
+            source_filters_and_or=False,
+            # system kwargs
+            configurable={
+                "vector_name": service_name,
+            },
+            user_id=user_id,
+            session_id=session_id, 
+            message_source="cli",
+            override_endpoint=service_url)
+        
+        # ensures {'answer': answer}
+        answer = parse_output(vac_response)
+        
+        console.print(answer.get('answer'))
+    else:
+        def stream_response():
+            generate = generate_proxy_stream(
+                    send_to_qa,
+                    user_input,
+                    vector_name=service_name,
+                    chat_history=chat_history,
+                    generate_f_output=lambda x: x,  # Replace with actual processing function
+                    stream_wait_time=0.5,
+                    stream_timeout=120,
+                    message_author=user_id,
+                    #TODO: populate these
+                    image_url=None,
+                    source_filters=None,
+                    search_kwargs=None,
+                    private_docs=None,
+                    whole_document=False,
+                    source_filters_and_or=False,
+                    # system kwargs
+                    configurable={
+                        "vector_name": service_name,
+                    },
+                    user_id=user_id,
+                    session_id=session_id, 
+                    message_source="cli",
+                    override_endpoint=service_url
+            )
+            for part in generate():
+                yield part
 
-    vac_response = ""
+        answer = ""
 
-    for token in stream_response():
-        if isinstance(token, bytes):
-            token = token.decode('utf-8')
-        print(token, end='', flush=True)
-        vac_response += token
+        for token in stream_response():
+            if isinstance(token, bytes):
+                token = token.decode('utf-8')
+            print(token, end='', flush=True)
+            answer += token
 
-    if vac_response:
+    if answer:
         chat_history.append({"name": "Human", "content": user_input})
-        chat_history.append({"name": "AI", "content": vac_response})
+        chat_history.append({"name": "AI", "content": answer})
     print()  # For new line after streaming ends
 
     return chat_history
@@ -273,6 +299,7 @@ def vac_command(args):
 def invoke_vac(service_url, data, vector_name=None, metadata=None, is_file=False):
     try:
         if is_file:
+            console.print("Uploading file to chunker...")
             # Handle file upload
             if not isinstance(data, Path) or not data.is_file():
                 raise ValueError("For file uploads, 'data' must be a Path object pointing to a valid file.")
@@ -287,12 +314,20 @@ def invoke_vac(service_url, data, vector_name=None, metadata=None, is_file=False
 
             response = requests.post(service_url, files=files, data=form_data)
         else:
+            console.print("Uploading JSON to chunker...")
             try:
-                json_data = json.loads(data)
+                if isinstance(data, dict):
+                    json_data = data
+                else:
+                    json_data = json.loads(data)
             except json.JSONDecodeError as err:
                 console.print(f"[bold red]ERROR: invalid JSON: {str(err)} [/bold red]")
                 sys.exit(1)
+            except Exception as err:
+                console.print(f"[bold red]ERROR: could not parse JSON: {str(err)} [/bold red]")
+                sys.exit(1)
 
+            log.debug(f"Sending data: {data} or json_data: {json.dumps(json_data)}")
             # Handle JSON data
             headers = {"Content-Type": "application/json"}
             response = requests.post(service_url, headers=headers, data=json.dumps(json_data))

sunholo/cli/embedder.py

@@ -18,12 +18,15 @@ def create_metadata(vac, metadata):
     # Default metadata if none provided
     default_metadata = {"vector_name": vac, "source": "sunholo-cli", "eventTime": formatted_time}
 
-    # Merge default metadata with provided metadata
-    if metadata:
-        if not isinstance(metadata, dict):
-            metadata = json.loads(metadata)
-    else:
-        metadata = {}    
+    try:
+        # Merge default metadata with provided metadata
+        if metadata:
+            if not isinstance(metadata, dict):
+                metadata = json.loads(metadata)
+        else:
+            metadata = {}    
+    except Exception as err:
+        console.print(f"[bold red]ERROR: metadata not parsed: {err} for {metadata}")
 
     # Update metadata with default values if not present
     metadata.update(default_metadata)
@@ -87,7 +90,7 @@ def embed_command(args):
     chunk_args = Namespace(**chunk_args)
     chunk_url = resolve_service_url(chunk_args, no_config=True)
 
-    with console.status(f"[bold orange]Sending {args.data} to chunk via {chunk_url}[/bold orange]", spinner="star"):
+    with console.status(f"[bold orange]Sending '{args.data}' to chunk via {chunk_url}[/bold orange]", spinner="star"):
         if args.is_file:
 
             metadata = create_metadata(args.vac_name, args.metadata)  
@@ -102,6 +105,7 @@ def embed_command(args):
         
         else:
             json_data = encode_data(args.vac_name, args.data, args.metadata, args.local_chunks)
+            console.print(f"Chunk JSON data: {json_data}")
             chunk_res = invoke_vac(f"{chunk_url}/pubsub_to_store", json_data)
         
         stop_proxy("chunker")
@@ -113,6 +117,11 @@ def embed_command(args):
 
             return
     
+    if not chunk_res:
+        console.rule(f"[bold orange]No chunks were found for processing of {args.data}[/bold orange]")
+
+        return
+    
     console.rule("Processing chunks locally")
 
     if args.embed_override:

{sunholo-0.62.17.dist-info → sunholo-0.63.0.dist-info}/METADATA

@@ -1,9 +1,9 @@
 Metadata-Version: 2.1
 Name: sunholo
-Version: 0.62.17
+Version: 0.63.0
 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/v0.62.17.tar.gz
+Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.63.0.tar.gz
 Author: Holosun ApS
 Author-email: multivac@sunholo.com
 License: Apache License, Version 2.0
@@ -111,54 +111,21 @@ This is the Sunholo Python project, a comprehensive toolkit for working with lan
 
 Please refer to the website for full documentation at https://dev.sunholo.com/
 
-# sunholo-py
-
-(draft release https://pypi.org/project/sunholo/ )
-## Table of Contents
-- [Agents](#agents)
-- [Archive](#archive)
-- [Bots](#bots)
-- [Chunker](#chunker)
-- [Components](#components)
-- [Database](#database)
-- [Embedder](#embedder)
-- [PubSub](#pubsub)
-- [QnA](#qna)
-- [Streaming](#streaming)
-- [Summarise](#summarise)
-- [Utils](#utils)
+## Demos
 
+Using https://github.com/charmbracelet/vhs
 
 ```sh
-pip install sunholo
+vhs record > cassette.tape
 ```
 
-A python library to enable LLMOps within cloud environments
-
-`sunholo` provides utilities to help manage LLM operations on Google Cloud Platform at first, but it is hoped that making it open source will help it support other clouds in the future.  A lot of the functionality is not Google Cloud Platform specific, so still may be helpful.
-
-It is derived from the Edmonbrain project, the original blog post you can read here: https://code.markedmondson.me/running-llms-on-gcp/ and owes a lot to Langchain ( https://github.com/langchain-ai/langchain )
-
-The package includes:
+Then make gif:
 
-* `agents/` - functions for working with agents, including easy flask apps, parsing chat history and dispatching requests to different agent endpoints
-* `archive/` - functions to record all Q&A activity to BigQuery via PubSub
-* `bots/` - functions for special cases regarding frontend bots such as GChat, Web Apps, Discord and Slack
-* `chunker/` - functions to slice up documents for sending into vector stores
-* `components/` - functions to help configure which LLM, prompt, vectorstore or document retriever you will use based on a yaml config file
-* `database/` - database setup functions and SQL to run on those sources such as Supabase
-* `embedder/` - functions to send chunks into embedding vector stores
-* `pubsub/` - use of PubSub for a message queue between components
-* `qna/` - utilities for running agents such as retry strats and parsing of output/input
-* `streaming/` - creation of streaming responses from LLM bots
-* `summarise/` - creation of summaries of large documents
-* `utils/` - reading configuration files, Google Cloud Platform metadata
-
-## Configuration
+```sh
+vhs docs/tapes/config-list.tape
+```
 
-The library uses the config specifications that some examples are given in the `config/` folder.
 
-When using the functions, make sure to have the `config/` folder in the root of where your application is running (usually `$HOME/config`)
 
 ```
    Copyright [2024] [Holosun ApS]

{sunholo-0.62.17.dist-info → sunholo-0.63.0.dist-info}/RECORD

@@ -16,7 +16,7 @@ sunholo/agents/flask/qna_routes.py,sha256=Twx6bitUAQ8Urtuwu9PtdoxYv-3aBDbF4cOUl9
 sunholo/archive/__init__.py,sha256=qNHWm5rGPVOlxZBZCpA1wTYPbalizRT7f8X4rs2t290,31
 sunholo/archive/archive.py,sha256=C-UhG5x-XtZ8VheQp92IYJqgD0V3NFQjniqlit94t18,1197
 sunholo/auth/__init__.py,sha256=4owDjSaWYkbTlPK47UHTOC0gCWbZsqn4ZIEw5NWZTlg,28
-sunholo/auth/run.py,sha256=oJ8JcimcDcdIogf7fvXLAi9EFIYhTR2_mjpR2-b97HE,2829
+sunholo/auth/run.py,sha256=SoXkyRSSO3ln4Uq71Zj2J2P9_TRp0Pg_wU1D9kI24Io,2690
 sunholo/bots/__init__.py,sha256=EMFd7e2z68l6pzYOnkzHbLd2xJRvxTKFRNCTuhZ8hIw,130
 sunholo/bots/discord.py,sha256=cCFae5K1BCa6JVkWGLh_iZ9qFO1JpXb6K4eJrlDfEro,2442
 sunholo/bots/github_webhook.py,sha256=5pQPRLM_wxxcILVaIzUDV8Kt7Arcm2dL1r1kMMHA524,9629
@@ -31,12 +31,12 @@ sunholo/chunker/pdfs.py,sha256=daCZ1xjn1YvxlifIyxskWNpLJLe-Q9D_Jq12MWx3tZo,2473
 sunholo/chunker/publish.py,sha256=GNXV6IPdKM2GZUcjGXIERu49D0ITYtizsLIktKVtMjM,2768
 sunholo/chunker/splitter.py,sha256=FLkDhkePkg_zGQpFBK13Cznw575D-Rf9pcaCpc1HUxY,6726
 sunholo/cli/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-sunholo/cli/chat_vac.py,sha256=Ht8EsGTeubV_-4KEUAA1P5M545Lt_wMAZU8esR6rHyI,16419
+sunholo/cli/chat_vac.py,sha256=RbhKICZtWZEPwL9wFmTNzhCWKHHRmNSqQHIfA-wi04U,17820
 sunholo/cli/cli.py,sha256=1-grjOE5sB5k0wrccsqCWSXB1vfrYNgdHBFH5F94pa0,3632
 sunholo/cli/cli_init.py,sha256=JMZ9AX2cPDZ-_mv3adiv2ToFVNyRPtjk9Biszl1kiR0,2358
 sunholo/cli/configs.py,sha256=QUM9DvKOdZmEQRM5uI3Nh887T0YDiSMr7O240zTLqws,4546
 sunholo/cli/deploy.py,sha256=zxdwUsRTRMC8U5vyRv0JiKBLFn84Ug_Tc88-_h9hJSs,1609
-sunholo/cli/embedder.py,sha256=5m_FaAqF6bnLJQj93HGySC-OOCXo2NCUVG_ZBjE1oUM,6937
+sunholo/cli/embedder.py,sha256=w7LT1CANSoQbOz8xAP3Zt4C_hP4lzGJOf8XD2jY5jBQ,7291
 sunholo/cli/merge_texts.py,sha256=U9vdMwKmcPoc6iPOWX5MKSxn49dNGbNzVLw8ui5PhEU,1823
 sunholo/cli/run_proxy.py,sha256=8mrZj0GPYO1q5e6cxej-PHdns5pGguWoDrX7RFg0FWo,11581
 sunholo/cli/sun_rich.py,sha256=UpMqeJ0C8i0pkue1AHnnyyX0bFJ9zZeJ7HBR6yhuA8A,54
@@ -101,9 +101,9 @@ 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-0.62.17.dist-info/LICENSE.txt,sha256=SdE3QjnD3GEmqqg9EX3TM9f7WmtOzqS1KJve8rhbYmU,11345
-sunholo-0.62.17.dist-info/METADATA,sha256=6SXkLQJWIi1y927EmV2uKqHTmgMQfG6c8xelbGYbu1U,8059
-sunholo-0.62.17.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
-sunholo-0.62.17.dist-info/entry_points.txt,sha256=bZuN5AIHingMPt4Ro1b_T-FnQvZ3teBes-3OyO0asl4,49
-sunholo-0.62.17.dist-info/top_level.txt,sha256=wt5tadn5--5JrZsjJz2LceoUvcrIvxjHJe-RxuudxAk,8
-sunholo-0.62.17.dist-info/RECORD,,
+sunholo-0.63.0.dist-info/LICENSE.txt,sha256=SdE3QjnD3GEmqqg9EX3TM9f7WmtOzqS1KJve8rhbYmU,11345
+sunholo-0.63.0.dist-info/METADATA,sha256=0jHRQDDREVfY0MVPf6CZe6wZ_kiSePMGUg5li128Yf8,5939
+sunholo-0.63.0.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
+sunholo-0.63.0.dist-info/entry_points.txt,sha256=bZuN5AIHingMPt4Ro1b_T-FnQvZ3teBes-3OyO0asl4,49
+sunholo-0.63.0.dist-info/top_level.txt,sha256=wt5tadn5--5JrZsjJz2LceoUvcrIvxjHJe-RxuudxAk,8
+sunholo-0.63.0.dist-info/RECORD,,