lollms-client 0.23.0__py3-none-any.whl → 0.24.1__py3-none-any.whl

examples/console_discussion/console_app.py

@@ -0,0 +1,266 @@
+import os
+import shutil
+import subprocess
+import sys
+import json
+from pathlib import Path
+from typing import Optional, List, Dict, Any
+
+# Correctly import all necessary classes from the lollms_client package
+from lollms_client import LollmsClient, LollmsDataManager, LollmsDiscussion, LollmsPersonality
+from ascii_colors import ASCIIColors, trace_exception
+
+# --- Configuration ---
+MAX_CONTEXT_SIZE_FOR_TEST = 2048 # Increased for agentic turns
+
+# Database and workspace configuration
+WORKSPACE_DIR = Path("./test_workspace_agentic")
+DATABASE_PATH = f"sqlite:///{WORKSPACE_DIR / 'test_discussion_agentic.db'}"
+DISCUSSION_ID = "console-agentic-test-1" # Use a fixed ID for easy resumption
+
+# --- MOCK KNOWLEDGE BASE for RAG ---
+MOCK_KNOWLEDGE_BASE = {
+    "python_basics.md": [
+        {"chunk_id": 1, "text": "Python is a high-level, interpreted programming language known for its readability. It was created by Guido van Rossum and released in 1991."},
+        {"chunk_id": 2, "text": "Key features of Python include dynamic typing, garbage collection, and a large standard library. It supports procedural, object-oriented, and functional programming."},
+    ],
+    "javascript_info.js": [
+        {"chunk_id": 1, "text": "JavaScript is a scripting language for front-end web development. It is also used in back-end development (Node.js)."},
+        {"chunk_id": 2, "text": "Popular JavaScript frameworks include React, Angular, and Vue.js."},
+    ],
+    "ai_concepts.txt": [
+        {"chunk_id": 1, "text": "Retrieval Augmented Generation (RAG) is an AI framework for improving LLM responses by grounding the model on external knowledge sources."},
+    ]
+}
+
+# --- Dummy MCP Server Scripts ---
+TIME_SERVER_PY = """
+import asyncio
+from datetime import datetime
+from mcp.server.fastmcp import FastMCP
+mcp_server = FastMCP("TimeMCP", description="A server that provides the current time.", host="localhost",
+            port=9624,
+            log_level="DEBUG")
+@mcp_server.tool()
+def get_current_time(user_id: str = "unknown"):
+    return {"time": datetime.now().isoformat(), "user_id": user_id}
+if __name__ == "__main__": mcp_server.run(transport="streamable-http")
+"""
+CALCULATOR_SERVER_PY = """
+import asyncio
+from typing import List, Union
+from mcp.server.fastmcp import FastMCP
+mcp_server = FastMCP("TimeMCP", description="A server that provides the current time.", host="localhost",
+            port=9625,
+            log_level="DEBUG")
+@mcp_server.tool()
+def add_numbers(numbers: List[Union[int, float]]):
+    if not isinstance(numbers, list): return {"error": "Input must be a list"}
+    return {"sum": sum(numbers)}
+if __name__ == "__main__": mcp_server.run(transport="streamable-http")
+"""
+
+# --- RAG Mock Function ---
+def mock_rag_query_function(query_text: str, top_k: int = 3, **kwargs) -> List[Dict[str, Any]]:
+    ASCIIColors.magenta(f"\n  [MOCK RAG] Querying knowledge base for: '{query_text}'")
+    results = []
+    query_lower = query_text.lower()
+    for file_path, chunks in MOCK_KNOWLEDGE_BASE.items():
+        for chunk in chunks:
+            if any(word in chunk["text"].lower() for word in query_lower.split() if len(word) > 2):
+                results.append({"file_path": file_path, "chunk_text": chunk["text"]})
+    ASCIIColors.magenta(f"  [MOCK RAG] Found {len(results[:top_k])} relevant chunks.")
+    return results[:top_k]
+
+def start_mcp_servers():
+    """Starts the dummy MCP servers in the background."""
+    ASCIIColors.yellow("--- Starting background MCP servers ---")
+    server_dir = WORKSPACE_DIR / "mcp_servers"
+    server_dir.mkdir(exist_ok=True, parents=True)
+    
+    (server_dir / "time_server.py").write_text(TIME_SERVER_PY)
+    (server_dir / "calculator_server.py").write_text(CALCULATOR_SERVER_PY)
+
+    procs = []
+    procs.append(subprocess.Popen([sys.executable, str(server_dir / "time_server.py")], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL))
+    procs.append(subprocess.Popen([sys.executable, str(server_dir / "calculator_server.py")], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL))
+    ASCIIColors.yellow("--- MCP servers launched ---")
+    return procs
+
+def setup_client_and_discussion() -> LollmsDiscussion:
+    """Sets up the LollmsClient with MCP, the DB manager, and the discussion."""
+    print("--- Setting up Lollms Environment ---")
+    WORKSPACE_DIR.mkdir(exist_ok=True)
+
+    mcp_config = {
+        "servers_infos": {
+            "time_machine": {"server_url": "http://localhost:9624/mcp"},
+            "calc_unit": {"server_url": "http://localhost:9625/mcp"},
+        }
+    }
+
+    try:
+        client = LollmsClient(
+            "ollama",
+            model_name="mistral-nemo:latest",
+            mcp_binding_name="remote_mcp",
+            mcp_binding_config=mcp_config
+        )
+    except Exception as e:
+        trace_exception(e)
+        print("\n---FATAL ERROR---")
+        print("Could not initialize LollmsClient. Ensure Ollama is running and mcp is installed.")
+        exit()
+
+    print(f"-> Using model: {client.binding.model_name}")
+    print(f"-> Using MCP binding: {client.mcp.binding_name}")
+
+    db_manager = LollmsDataManager(db_path=DATABASE_PATH)
+    discussion = db_manager.get_discussion(client, DISCUSSION_ID)
+    
+    if discussion:
+        print(f"-> Resuming discussion (ID: {DISCUSSION_ID})")
+        discussion.max_context_size = MAX_CONTEXT_SIZE_FOR_TEST
+    else:
+        print(f"-> Creating new discussion (ID: {DISCUSSION_ID})")
+        discussion = LollmsDiscussion.create_new(
+            lollms_client=client,
+            db_manager=db_manager,
+            id=DISCUSSION_ID,
+            title="Console Agentic Test",
+            max_context_size=MAX_CONTEXT_SIZE_FOR_TEST
+        )
+
+    print("--- Setup Complete. Ready to chat! ---\n")
+    return discussion
+
+def print_help():
+    print("\n--- Commands ---")
+    print("!agent <prompt>  - Run a prompt using all available tools (MCP).")
+    print("!rag <prompt>    - Run a prompt using the mock knowledge base (RAG).")
+    print("!both <prompt>   - Run a prompt using both MCP tools and RAG.")
+    print("!status          - Show current discussion state (pruning, message count).")
+    print("!regen           - Regenerate the last AI response.")
+    print("!exit            - Exit the application.")
+    print("----------------\n")
+
+def print_agentic_results(response_dict):
+    """Renders a beautiful report of the agent's turn."""
+    ai_message = response_dict.get('ai_message')
+    if not ai_message:
+        return
+
+    ASCIIColors.cyan("\n" + "="*22 + " Agentic Turn Report " + "="*22)
+
+    # --- Final Answer ---
+    ASCIIColors.blue("\nFinal Answer:")
+    ASCIIColors.green(f"  {ai_message.content}")
+
+    # --- Agent's Internal Monologue (The Scratchpad) ---
+    if ai_message.scratchpad:
+        ASCIIColors.blue("\nAgent's Reasoning Log (Scratchpad):")
+        # Print scratchpad line by line for better color coding
+        for line in ai_message.scratchpad.split('\n'):
+            if line.startswith("### Step"):
+                ASCIIColors.yellow(line)
+            elif line.startswith("- **Action**:") or line.startswith("- **Result**:") or line.startswith("- **Error**:") :
+                ASCIIColors.magenta(line)
+            else:
+                print(line)
+
+    # --- Sources Used (from metadata) ---
+    if ai_message.metadata and "sources" in ai_message.metadata:
+        sources = ai_message.metadata.get("sources", [])
+        if sources:
+            ASCIIColors.blue("\nSources Consulted (RAG):")
+            for i, source in enumerate(sources):
+                print(f"  [{i+1}] Path: {source.get('file_path', 'N/A')}")
+                # Indent the content for readability
+                content = source.get('chunk_text', 'N/A').replace('\n', '\n      ')
+                print(f"      Content: \"{content}\"")
+    
+    ASCIIColors.cyan("\n" + "="*61 + "\n")
+
+def run_chat_console(discussion: LollmsDiscussion):
+    print_help()
+    while True:
+        user_input = input("You: ")
+        if not user_input:
+            continue
+
+        use_mcps_flag = False
+        use_data_store_flag = False
+        prompt = user_input
+
+        # --- Command Handling ---
+        if user_input.lower().startswith("!exit"):
+            break
+        elif user_input.lower().startswith("!help"):
+            print_help()
+            continue
+        elif user_input.lower().startswith("!status"):
+            # Assuming a print_status function exists
+            # print_status(discussion)
+            continue
+        elif user_input.lower().startswith("!regen"):
+            # Assuming a regenerate_branch method exists
+            # discussion.regenerate_branch(...)
+            continue
+        elif user_input.lower().startswith("!agent "):
+            use_mcps_flag = True
+            prompt = user_input[7:].strip()
+            ASCIIColors.yellow(f"Agentic MCP turn initiated for: '{prompt}'")
+        elif user_input.lower().startswith("!rag "):
+            use_data_store_flag = True
+            prompt = user_input[5:].strip()
+            ASCIIColors.yellow(f"Agentic RAG turn initiated for: '{prompt}'")
+        elif user_input.lower().startswith("!both "):
+            use_mcps_flag = True
+            use_data_store_flag = True
+            prompt = user_input[6:].strip()
+            ASCIIColors.yellow(f"Agentic MCP+RAG turn initiated for: '{prompt}'")
+        
+        # --- Streaming Callback ---
+        def stream_callback(chunk, msg_type, metadata={}, **kwargs):
+            # Render steps and thoughts in real-time
+            if msg_type == 12: # MSG_TYPE.MSG_TYPE_STEP_START
+                 ASCIIColors.cyan(f"\n> Starting: {chunk}")
+            elif msg_type == 13: # MSG_TYPE.MSG_TYPE_STEP_END
+                 ASCIIColors.cyan(f"> Finished: {chunk}")
+            elif msg_type == 2: # MSG_TYPE.MSG_TYPE_INFO (for thoughts)
+                 ASCIIColors.yellow(f"\n  (Thought): {chunk}")
+            else: # Final answer chunks are printed by the main loop
+                pass # The final answer is printed after the full report
+            return True
+
+        # --- Main Chat Logic ---
+        try:
+            #print("\nAI: ", end="", flush=True)
+            
+            response_dict = discussion.chat(
+                user_message=prompt,
+                use_mcps=use_mcps_flag,
+                use_data_store={"coding_store": mock_rag_query_function} if use_data_store_flag else None,
+                streaming_callback=stream_callback
+            )
+            if use_mcps_flag or use_data_store_flag:
+                print_agentic_results(response_dict)
+            print("\nAI: ", end="")
+            ASCIIColors.green(response_dict['ai_message'].content)            
+
+        except Exception as e:
+            trace_exception(e)
+            print(f"\nAn error occurred during generation.")
+
+if __name__ == "__main__":
+    mcp_procs = start_mcp_servers()
+    try:
+        discussion_session = setup_client_and_discussion()
+        run_chat_console(discussion_session)
+    finally:
+        ASCIIColors.red("\n--- Shutting down MCP servers ---")
+        for proc in mcp_procs:
+            proc.terminate()
+            proc.wait()
+        shutil.rmtree(WORKSPACE_DIR, ignore_errors=True)
+        print("Cleanup complete. Goodbye!")

lollms_client/__init__.py

@@ -8,7 +8,7 @@ from lollms_client.lollms_utilities import PromptReshaper # Keep general utiliti
 from lollms_client.lollms_mcp_binding import LollmsMCPBinding, LollmsMCPBindingManager
 
 
-__version__ = "0.23.0" # Updated version
+__version__ = "0.24.1" # Updated version
 
 # Optionally, you could define __all__ if you want to be explicit about exports
 __all__ = [