Updated main.py with Ollama and LangChain create_agent

main.py

@@ -1,50 +1,47 @@
 # main.py
-# Полностью рабочий пример агента с RAG‑памятью на Qdrant и OpenRouter
+# Полностью рабочий пример агента с RAG‑памятью на базе Qdrant и Ollama.
-# Использует deepagents, langchain‑openai, langchain‑qdrant, langchain‑core
+# Используется LangChain 1.x, create_agent, инструменты @tool, и Ollama‑LLM/embeddings.
-# Запуск: python main.py
+#
+# Запуск:
+#   python main.py
+# После запуска можно использовать команды:
+#   /add <title> <content>   – добавить документ
+#   /search <query> <max>    – семантический поиск
+#   /quit                    – выйти
+#
+# Для загрузки документов из директории используйте функцию load_documents_from_dir.
+#"""
 
 import os
-import asyncio
+import sys
+import textwrap
 from pathlib import Path
-from typing import List
+from typing import List, Dict, Any
 
-from langchain_openai import ChatOpenAI, OpenAIEmbeddings
+from langchain_ollama import ChatOllama, OllamaEmbeddings
-from langchain_core.documents import Document
 from langchain_qdrant import QdrantVectorStore
 from langchain_text_splitters import RecursiveCharacterTextSplitter
 from langchain.tools import tool
-from deepagents import create_deep_agent
+from langchain.agents import create_agent, AgentExecutor, AgentToolkit, Tool
-from deepagents.backends import FilesystemBackend, LocalShellBackend, CompositeBackend
 from langchain_core.messages import HumanMessage
 
 # ---------------------------------------------------------------------------
-# Конфигурация LLM и Embeddings
+# Конфигурация
 # ---------------------------------------------------------------------------
-llm = ChatOpenAI(
+QDRANT_URL = os.getenv("QDRANT_URL", "http://localhost:6333")
-    model="openai/gpt-oss-20b:free",
+QDRANT_COLLECTION = "knowledge_base"
-    base_url="https://openrouter.ai/api/v1",
+EMBEDDING_MODEL = "nomic-embed-text"
-    api_key=os.getenv("OPENAI_API_KEY"),
+LLM_MODEL = "llama3"
-    temperature=0.0,
-)
-
-embeddings = OpenAIEmbeddings(
-    model="text-embedding-3-small",
-    base_url="https://openrouter.ai/api/v1",
-    api_key=os.getenv("OPENAI_API_KEY"),
-)
 
 # ---------------------------------------------------------------------------
-# Qdrant клиент и коллекция
+# Векторное хранилище
 # ---------------------------------------------------------------------------
-# Предполагается, что Qdrant запущен локально на порту 6333
+# Инициализируем эмбеддер и клиент Qdrant
-qdrant_url = "http://localhost:6333"
+embeddings = OllamaEmbeddings(model=EMBEDDING_MODEL)
-collection_name = "knowledge_base"
-
 vector_store = QdrantVectorStore(
-    embeddings=embeddings,
+    url=QDRANT_URL,
-    url=qdrant_url,
+    collection_name=QDRANT_COLLECTION,
-    collection_name=collection_name,
+    embedding=embeddings,
-    # Если коллекция не существует, она будет создана автоматически
 )
 
 # ---------------------------------------------------------------------------
@@ -53,117 +50,131 @@ vector_store = QdrantVectorStore(
 text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)
 
 # ---------------------------------------------------------------------------
-# Инструменты для агента
+# Инструменты
 # ---------------------------------------------------------------------------
-@tool
+@tool("search_knowledge_base", "Semantic search in the knowledge base.")
 def search_knowledge_base(query: str, max_results: int = 5) -> str:
-    """Semantic search in the knowledge base.
+    """Return top‑k relevant documents for a query.
-    Returns a formatted string with the top results.
+    The function returns a formatted string with titles and snippets.
     """
     results = vector_store.similarity_search_with_score(query, k=max_results)
     if not results:
         return "No relevant documents found."
     formatted = []
     for doc, score in results:
-        formatted.append(f"Title: {doc.metadata.get('title', 'Untitled')}\nScore: {score:.4f}\nContent: {doc.page_content[:200]}...\n")
+        title = doc.metadata.get("title", "Untitled")
+        snippet = doc.page_content[:200].replace("\n", " ")
+        formatted.append(f"{title} (score: {score:.3f}): {snippet}...")
     return "\n".join(formatted)
 
-@tool
+@tool("add_to_knowledge_base", "Add a document to the knowledge base.")
 def add_to_knowledge_base(content: str, title: str) -> str:
-    """Add a new document to the knowledge base.
+    """Chunk the content, embed, and store in Qdrant.
-    The content is split into chunks, embedded and stored.
+    Returns a confirmation message.
     """
-    # Split into chunks
     chunks = text_splitter.split_text(content)
-    docs: List[Document] = []
+    docs = []
     for i, chunk in enumerate(chunks):
-        docs.append(Document(page_content=chunk, metadata={"title": title, "chunk_index": i}))
+        docs.append(
-    # Add to vector store
+            {
-    vector_store.add_documents(docs)
+                "page_content": chunk,
-    return f"Added {len(docs)} chunks for document '{title}'."
+                "metadata": {"title": title, "chunk_index": i},
-
+            }
-# ---------------------------------------------------------------------------
-# Backend для deepagents
-# ---------------------------------------------------------------------------
-backend = CompositeBackend(
-    default=LocalShellBackend(root_dir="./workspace", virtual_mode=True, inherit_env=True),
-    routes={},
-)
-
-# ---------------------------------------------------------------------------
-# Создание агента
-# ---------------------------------------------------------------------------
-agent = create_deep_agent(
-    model=llm,
-    tools=[search_knowledge_base, add_to_knowledge_base],
-    backend=backend,
-    system_prompt="You are a helpful assistant with access to a knowledge base. Use the provided tools to search and add information."
-)
-
-# ---------------------------------------------------------------------------
-# Инициализация: загрузка документов из директории
-# ---------------------------------------------------------------------------
-DOCS_DIR = Path("./docs")
-
-async def load_documents_from_dir(directory: Path):
-    if not directory.exists():
-        return
-    for file_path in directory.rglob("*.txt"):
-        content = file_path.read_text(encoding="utf-8")
-        title = file_path.stem
-        await agent.ainvoke(
-            {"messages": [HumanMessage(content=f"/add {title}")], "content": content},
-            {"configurable": {"thread_id": "init-session"}},
         )
+    vector_store.add_documents(docs)
+    return f"Added {len(chunks)} chunks of '{title}' to the knowledge base."
 
 # ---------------------------------------------------------------------------
-# Интерактивный клиент
+# Агент
 # ---------------------------------------------------------------------------
-async def interactive_loop():
+# Создаём LLM
-    print("Welcome to the RAG agent. Commands: /add <title>, /search <query>, /quit")
+llm = ChatOllama(model=LLM_MODEL, temperature=0.2)
-    thread_id = "interactive-session"
+
+# Список инструментов
+tools = [search_knowledge_base, add_to_knowledge_base]
+
+# Создаём агент
+agent = create_agent(
+    llm=llm,
+    tools=tools,
+    system_message="You are an assistant that can search and add documents to a local knowledge base. Use the provided tools.",
+    verbose=True,
+)
+
+# Обёртка для выполнения
+agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)
+
+# ---------------------------------------------------------------------------
+# Загрузка документов из директории
+# ---------------------------------------------------------------------------
+
+def load_documents_from_dir(directory: str) -> None:
+    """Load all .txt files from a directory into the knowledge base.
+    Each file becomes a separate document with its filename as title.
+    """
+    path = Path(directory)
+    if not path.is_dir():
+        print(f"Directory {directory} does not exist.")
+        return
+    for file in path.glob("*.txt"):
+        title = file.stem
+        content = file.read_text(encoding="utf-8")
+        print(f"Adding {title}...", end=" ")
+        result = add_to_knowledge_base(content, title)
+        print(result)
+
+# ---------------------------------------------------------------------------
+# CLI
+# ---------------------------------------------------------------------------
+
+def main():
+    # Если пользователь передал путь к директории, загрузим документы
+    if len(sys.argv) > 1:
+        load_documents_from_dir(sys.argv[1])
+
+    print("\n--- RAG Agent CLI ---")
+    print("Commands:")
+    print("  /add <title> <content>   – add a document")
+    print("  /search <query> <max>    – search knowledge base")
+    print("  /quit                    – exit")
+
     while True:
-        user_input = input("> ")
+        try:
-        if user_input.strip() == "/quit":
+            user_input = input("\n> ")
+        except (EOFError, KeyboardInterrupt):
+            print("\nExiting.")
+            break
+
+        if not user_input.strip():
+            continue
+
+        if user_input.startswith("/quit"):
             print("Goodbye!")
             break
-        if user_input.startswith("/add "):
-            title = user_input[5:].strip()
-            print("Enter content (end with a single line containing only 'END'): ")
-            lines = []
-            while True:
-                line = input()
-                if line.strip() == "END":
-                    break
-                lines.append(line)
-            content = "\n".join(lines)
-            await agent.ainvoke(
-                {"messages": [HumanMessage(content=f"/add {title}")], "content": content},
-                {"configurable": {"thread_id": thread_id}},
-            )
-            print(f"Document '{title}' added.")
-        elif user_input.startswith("/search "):
-            query = user_input[8:].strip()
-            result = await agent.ainvoke(
-                {"messages": [HumanMessage(content=f"/search {query}")]},
-                {"configurable": {"thread_id": thread_id}},
-            )
-            print(result["messages"][-1].content)
-        else:
-            # обычный запрос к LLM
-            result = await agent.ainvoke(
-                {"messages": [HumanMessage(content=user_input)]},
-                {"configurable": {"thread_id": thread_id}},
-            )
-            print(result["messages"][-1].content)
 
-# ---------------------------------------------------------------------------
+        if user_input.startswith("/add"):
-# Основной запуск
+            parts = user_input.split(maxsplit=2)
-# ---------------------------------------------------------------------------
+            if len(parts) < 3:
-async def main():
+                print("Usage: /add <title> <content>")
-    # Загрузим документы из каталога docs при старте
+                continue
-    await load_documents_from_dir(DOCS_DIR)
+            title, content = parts[1], parts[2]
-    await interactive_loop()
+            print(add_to_knowledge_base(content, title))
+            continue
+
+        if user_input.startswith("/search"):
+            parts = user_input.split(maxsplit=2)
+            if len(parts) < 2:
+                print("Usage: /search <query> [max_results]")
+                continue
+            query = parts[1]
+            max_results = int(parts[2]) if len(parts) > 2 else 5
+            print(search_knowledge_base(query, max_results))
+            continue
+
+        # Любой другой ввод – передаём агенту
+        response = agent_executor.invoke({"input": user_input})
+        print(response.get("output", ""))
+
 
 if __name__ == "__main__":
-    asyncio.run(main())
+    main()