Feature/chat sessions (#12)

* added cli option to set ollama ps polling interval
* added cli option to purge all chats
* fixed dupe toast messages
* various bug fixes
* updated deps
* added pre-commit run to makefile
* changed /save to /export
* renamed /clear to /new and added /delete
* updated help with table of contents and workflow sections
* add ability to select, load and delete existing sessions
* clear button renamed to new on chat screen

.pre-commit-config.yaml

@@ -20,7 +20,7 @@ repos:
 #-   repo: https://github.com/asottile/pyupgrade
 #    rev: v3.16.0
 #    hooks:
-#    -   id: pyupgrade
+#    -   message_id: pyupgrade
 
 -   repo: https://github.com/psf/black
     rev: 22.10.0
@@ -38,7 +38,7 @@ repos:
 #-   repo: https://github.com/pre-commit/mirrors-mypy
 #    rev: v1.10.1
 #    hooks:
-#    -   id: mypy
+#    -   message_id: mypy
 #        additional_dependencies: [textual, rich, pydantic, ollama, docker, types-beautifulsoup4, types-requests, types-pytz, types-simplejson, typing-extensions, asyncio, humanize, argparse, python-dotenv]
 #        exclude: tests(/\w*)*/functional/|tests/input|tests(/.*)+/conftest.py|doc/data/messages|tests(/\w*)*data/
 

README.md

@@ -1,5 +1,30 @@
 # PAR LLAMA
 
+## Table of Contents
+
+- [About](#about)
+- [Screenshots](#screenshots)
+- [Prerequisites](#prerequisites-for-running)
+  - [For Running](#prerequisites-for-running)
+  - [For Development](#prerequisites-for-dev)
+  - [For Model Quantization](#prerequisites-for-model-quantization)
+- [Installation](#installing-from-mypi-using-pipx)
+  - [Using pipx](#installing-from-mypi-using-pipx)
+  - [Using pip](#installing-from-mypi-using-pip)
+  - [For Development](#installing-for-dev-mode)
+- [Command Line Arguments](#command-line-arguments)
+- [Environment Variables](#environment-variables)
+- [Running PAR_LLAMA](#running-par_llama)
+  - [With pipx installation](#with-pipx-installation)
+  - [With pip installation](#with-pip-installation)
+  - [Under Windows WSL](#running-under-windows-wsl)
+  - [In Development Mode](#dev-mode)
+- [Example Workflow](#example-workflow)
+- [Themes](#themes)
+- [Contributing](#contributing)
+- [Roadmap](#roadmap)
+- [What's New](#whats-new)
+
 ## About
 PAR LLAMA is a TUI application designed for easy management and use of Ollama based LLMs.
 The application was built with [Textual](https://textual.textualize.io/) and [Rich](https://github.com/Textualize/rich?tab=readme-ov-file)
@@ -20,6 +45,7 @@ Supports Dark and Light mode as well as custom themes.
 ## Prerequisites for running
 * Install and run [Ollama](https://ollama.com/download)
 * Install Python 3.11 or newer
+  * On Windows the [Scoop](https://scoop.sh/) tool makes it easy to install and manage things like python.
 
 ## Prerequisites for dev
 * Install pipenv
@@ -66,8 +92,8 @@ make first-setup
 
 ## Command line arguments
 ```
-usage: parllama [-h] [-v] [-d DATA_DIR] [-u OLLAMA_URL] [-t THEME_NAME] [-m {dark,light}] [-s {local,site,tools,create,logs}]
-                [--restore-defaults] [--clear-cache] [--no-save]
+usage: parllama [-h] [-v] [-d DATA_DIR] [-u OLLAMA_URL] [-t THEME_NAME] [-m {dark,light}] [-s {local,site,tools,create,chat,logs}] [-p PS_POLL]
+                [--restore-defaults] [--clear-cache] [--purge-chats] [--no-save]
 
 PAR LLAMA -- Ollama TUI.
 
@@ -82,10 +108,13 @@ options:
                         Theme name. Defaults to par
   -m {dark,light}, --theme-mode {dark,light}
                         Dark / Light mode. Defaults to dark
-  -s {local,site,tools,create,logs}, --starting-screen {local,site,tools,create,logs}
+  -s {local,site,tools,create,chat,logs}, --starting-screen {local,site,tools,create,chat,logs}
                         Starting screen. Defaults to local
+  -p PS_POLL, --ps-poll PS_POLL
+                        Interval in seconds to poll ollama ps command. 0 = disable. Defaults to 3
   --restore-defaults    Restore default settings and theme
   --clear-cache         Clear cached data
+  --purge-chats         Purge all chat history
   --no-save             Prevent saving settings for this session.
 ```
 
@@ -141,6 +170,21 @@ From repo root:
 make dev
 ```
 
+## Example workflow
+* Start parllama.
+* Click the "Site" tab.
+* Use ^R to fetch the latest models from Ollama.com.
+* User the "Filter Site models" text box and type "llama3".
+* Find the entry with title of "llama3".
+* Click the blue tag "8B" to update the search box to read "llama3:8b".
+* Press ^P to pull the model from Ollama to your local machine. Depending on the size of the model and your internet connection this can take a few min.
+* Click the "Local" tab to see models that have been locally downloaded
+* Select the "llama3:8b" entry and press ^C to jump to the "Chat" tab and auto select the model
+* Type a message to the model such as "Why is the sky blue?". It will take a few seconds for Ollama to load the model. After which the LLMs answer will stream in.
+* Towards the very top of the app you will see what model is loaded and what percent of it is loaded into the GPU / CPU. If a model cant be loaded 100% on the GPU it will run slower.
+* To export your conversation as a Markdown file type "/export" in the message input box. This will open a export dialog.
+* Type "/help" to see what other slash commands are available.
+
 ## Themes
 Themes are json files stored in the themes folder in the data directory which defaults to **~/.parllama/themes**  
 
@@ -206,16 +250,19 @@ if anything remains to be fixed before the commit is allowed.
 **Where we are**  
 * Initial release - Find, maintain and create new models
 * Basic chat with LLM
+* Chat history / conversation management
 
 **Where we're going**
-* Chat history / conversation management
 * Chat with multiple models at same time to compare outputs
 * LLM tool use
 
 
 ## What's new
 
 ### v0.2.6
+* Added chat history panel and management to chat page
+
+### v0.2.51
 * Fix missing dependency in package
 
 ### v0.2.5

parllama/__init__.py

@@ -6,7 +6,7 @@
 __credits__ = ["Paul Robello"]
 __maintainer__ = "Paul Robello"
 __email__ = "probello@gmail.com"
-__version__ = "0.2.51"
+__version__ = "0.2.6"
 __licence__ = "MIT"
 __application_title__ = "PAR LLAMA"
 __application_binary__ = "parllama"

parllama/app.py

@@ -34,6 +34,7 @@
 from parllama.messages.main import AppRequest
 from parllama.messages.main import ChangeTab
 from parllama.messages.main import CreateModelFromExistingRequested
+from parllama.messages.main import DeleteSession
 from parllama.messages.main import LocalModelCopied
 from parllama.messages.main import LocalModelCopyRequested
 from parllama.messages.main import LocalModelDelete
@@ -51,6 +52,8 @@
 from parllama.messages.main import NotifyInfoMessage
 from parllama.messages.main import PsMessage
 from parllama.messages.main import SendToClipboard
+from parllama.messages.main import SessionListChanged
+from parllama.messages.main import SessionSelected
 from parllama.messages.main import SetModelNameLoading
 from parllama.messages.main import SiteModelsLoaded
 from parllama.messages.main import SiteModelsRefreshRequested
@@ -105,6 +108,7 @@ def __init__(self) -> None:
         """Initialize the application."""
         super().__init__()
         self.notify_subs = {"*": set[Widget]()}
+        chat_manager.set_app(self)
 
         self.title = __application_title__
         self.dark = settings.theme_mode != "light"
@@ -114,7 +118,6 @@ def __init__(self) -> None:
         self.is_refreshing = False
         self.last_status = ""
         self.main_screen = MainScreen()
-        chat_manager.set_app(self)
 
     def _watch_dark(self, value: bool) -> None:
         """Watch the dark property."""
@@ -155,6 +158,12 @@ async def on_mount(self) -> None:
         self.main_screen.post_message(
             StatusMessage(f"Using Ollama server url: {settings.ollama_host}")
         )
+        self.main_screen.post_message(
+            StatusMessage(
+                f"Polling Ollama ps every: {settings.ollama_ps_poll_interval} seconds"
+            )
+        )
+
         self.main_screen.post_message(
             StatusMessage(
                 f"""Theme: "{settings.theme_name}" in {settings.theme_mode} mode"""
@@ -168,7 +177,8 @@ async def on_mount(self) -> None:
         )
 
         self.set_timer(1, self.do_jobs)
-        self.set_timer(1, self.update_ps)
+        if settings.ollama_ps_poll_interval > 0:
+            self.set_timer(1, self.update_ps)
 
     def action_noop(self) -> None:
         """Do nothing"""
@@ -220,7 +230,7 @@ def send_to_clipboard(self, msg: SendToClipboard) -> None:
         # works for local sessions
         pyperclip.copy(msg.message)
         if msg.notify:
-            self.post_message(NotifyInfoMessage("Value copied to clipboard"))
+            self.post_message(NotifyInfoMessage("Copied to clipboard"))
 
     @on(ModelPushRequested)
     def on_model_push_requested(self, msg: ModelPushRequested) -> None:
@@ -501,6 +511,9 @@ def on_app_request(self, msg: AppRequest) -> None:
         """Add any widget that requests an action to notify_subs"""
         if msg.widget:
             self.notify_subs["*"].add(msg.widget)
+            if msg.__class__.__name__ not in self.notify_subs:
+                self.notify_subs[msg.__class__.__name__] = set()
+            self.notify_subs[msg.__class__.__name__].add(msg.widget)
 
     @on(LocalModelListRefreshRequested)
     def on_model_list_refresh_requested(self) -> None:
@@ -520,7 +533,9 @@ async def refresh_models(self):
             )
             dm.refresh_models()
             self.main_screen.post_message(StatusMessage("Local model list refreshed"))
-            self.post_message_all(LocalModelListLoaded())
+            # self.post_message_all(LocalModelListLoaded())
+            self.main_screen.local_view.post_message(LocalModelListLoaded())
+            self.main_screen.chat_view.post_message(LocalModelListLoaded())
         finally:
             self.is_refreshing = False
 
@@ -554,7 +569,7 @@ async def refresh_site_models(self, msg: SiteModelsRefreshRequested):
                 )
             )
             dm.refresh_site_models(msg.ollama_namespace, None, msg.force)
-            self.post_message_all(
+            self.main_screen.site_view.post_message(
                 SiteModelsLoaded(ollama_namespace=msg.ollama_namespace)
             )
             self.main_screen.post_message(
@@ -571,7 +586,10 @@ async def update_ps(self) -> None:
         """Update ps status bar msg"""
         was_blank = False
         while self.is_running:
-            await asyncio.sleep(2)
+            if settings.ollama_ps_poll_interval < 1:
+                self.main_screen.post_message(PsMessage(msg=""))
+                break
+            await asyncio.sleep(settings.ollama_ps_poll_interval)
             ret = dm.model_ps()
             if not ret:
                 if not was_blank:
@@ -600,14 +618,14 @@ def status_notify(self, msg: str, severity: SeverityLevel = "information") -> No
         self.notify(msg, severity=severity)
         self.main_screen.post_message(StatusMessage(msg))
 
-    def post_message_all(self, msg: Message) -> None:
+    def post_message_all(self, msg: Message, sub_name: str = "*") -> None:
         """Post a message to all screens"""
         if isinstance(msg, StatusMessage):
             self.log(msg.msg)
             self.last_status = msg.msg
-
-        for w in list(self.notify_subs["*"]):
-            w.post_message(msg)
+        if sub_name in self.notify_subs:
+            for w in list(self.notify_subs[sub_name]):
+                w.post_message(msg)
         if self.main_screen:
             self.main_screen.post_message(msg)
 
@@ -622,8 +640,6 @@ def on_create_model_from_existing_requested(
         self, msg: CreateModelFromExistingRequested
     ) -> None:
         """Create model from existing event"""
-        msg.stop()
-
         self.main_screen.create_view.name_input.value = f"my-{msg.model_name}:latest"
         self.main_screen.create_view.text_area.text = msg.model_code
         self.main_screen.create_view.quantize_input.value = msg.quantization_level or ""
@@ -633,7 +649,25 @@ def on_create_model_from_existing_requested(
     @on(ModelInteractRequested)
     def on_model_interact_requested(self, msg: ModelInteractRequested) -> None:
         """Model interact requested event"""
-        msg.stop()
         self.main_screen.change_tab("Chat")
         self.main_screen.chat_view.model_select.value = msg.model_name
         self.main_screen.chat_view.user_input.focus()
+
+    @on(SessionListChanged)
+    def on_session_list_changed(self) -> None:
+        """Session list changed event"""
+        self.main_screen.chat_view.session_list.post_message(SessionListChanged())
+
+    @on(SessionSelected)
+    def on_session_selected(self, msg: SessionSelected) -> None:
+        """Session selected event"""
+        self.main_screen.chat_view.session_list.post_message(
+            SessionSelected(session_id=msg.session_id)
+        )
+
+    @on(DeleteSession)
+    def on_delete_session(self, msg: DeleteSession) -> None:
+        """Delete session event"""
+        self.main_screen.chat_view.post_message(
+            DeleteSession(session_id=msg.session_id)
+        )