feat: solution for 6a02e23da6fe2e4ac16acf65

solutions/6a02e23da6fe2e4ac16acf65/solution.py

@@ -1,23 +1,21 @@
-from langchain_openai import ChatOpenAI
-from pydantic import SecretStr
-from langchain.tools import tool
+from pathlib import Path
+import sys
+
+from langchain_ollama import Ollama, OllamaEmbeddings
 from langchain_qdrant import QdrantVectorStore
 from qdrant_client import QdrantClient
 from qdrant_client.http.models import Distance, VectorParams
-from langchain_ollama import OllamaEmbeddings
-from langchain_core.documents import Document
 from langchain_text_splitters import RecursiveCharacterTextSplitter
+from langchain.tools import tool
 from langchain.agents import create_agent
+from langchain_core.documents import Document
 
-# ---------- LLM ----------
-llm = ChatOpenAI(
-    model="openai/gpt-oss-20b",
-    base_url='https://platform.brojs.ru/jrnl-bh/api/inference/v1',
-    api_key=SecretStr("jrnl_30283ab953615cbb6846ff9940a1eedce0b76d7b2f59a2394f29e74643e6a90d"),
+# ---------- LLM and embeddings ----------
+llm = Ollama(
+    model="llama3",          # local Ollama model
     temperature=0.7,
 )
 
-# ---------- Embeddings ----------
 embeddings = OllamaEmbeddings(model="nomic-embed-text")
 
 # ---------- Qdrant client ----------
@@ -26,81 +24,97 @@ client.create_collection(
     collection_name="knowledge_base",
     vectors_config=VectorParams(size=embeddings.embedding_size, distance=Distance.COSINE),
 )
-vector_store = QdrantVectorStore(
-    client=client,
-    collection_name="knowledge_base",
-    embedding=embeddings,
-)
+vector_store = QdrantVectorStore(client=client, collection_name="knowledge_base", embedding=embeddings)
 
 # ---------- Text splitter ----------
-splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=100)
+splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)
 
 # ---------- Tools ----------
 @tool
 def search_knowledge_base(query: str, max_results: int = 5) -> str:
     """Search the knowledge base for relevant documents."""
-    docs_with_score = vector_store.similarity_search_with_score(query, k=max_results)
-    if not docs_with_score:
-        return "No results found."
-    result_lines = [
-        f"{i+1}. {doc.page_content[:200]}..." for i, (doc, _) in enumerate(docs_with_score)
-    ]
-    return "\n".join(result_lines)
+    results = vector_store.similarity_search_with_score(query, k=max_results)
+    if not results:
+        return "No relevant information found."
+    out_lines = []
+    for doc, score in results:
+        title = doc.metadata.get("title", "Untitled")
+        content_preview = doc.page_content[:200] + ("..." if len(doc.page_content) > 200 else "")
+        out_lines.append(f"Score: {score:.3f}\nTitle: {title}\nContent: {content_preview}")
+    return "\n\n".join(out_lines)
 
 @tool
-def add_to_knowledge_base(content: str, title: str) -> str:
+def add_to_knowledge_base(content: str, title: str = "Untitled") -> str:
     """Add a new document to the knowledge base."""
     chunks = splitter.split_text(content)
-    documents = [Document(page_content=chunk, metadata={"title": title}) for chunk in chunks]
+    documents = [
+        Document(page_content=chunk, metadata={"title": f"{title} (part {i+1})"})
+        for i, chunk in enumerate(chunks)
+    ]
     vector_store.add_documents(documents)
-    return f"Added {len(documents)} chunks from '{title}'."
+    return f"Added {len(chunks)} chunks to the knowledge base under title '{title}'."
 
 # ---------- Agent ----------
+system_prompt = """
+You are an assistant that can search and add information to a local knowledge base.
+Use the tools `search_knowledge_base` and `add_to_knowledge_base` as needed.
+"""
+
 agent = create_agent(
     model=llm,
     tools=[search_knowledge_base, add_to_knowledge_base],
-    system_prompt="You are an assistant that can search and add documents to the knowledge base.",
+    system_prompt=system_prompt,
 )
 
 # ---------- CLI ----------
+def load_documents_from_dir(directory: Path):
+    for file_path in directory.rglob("*"):
+        if file_path.is_file() and file_path.suffix.lower() in {".txt", ".md"}:
+            content = file_path.read_text(encoding="utf-8")
+            title = file_path.stem
+            add_to_knowledge_base(content=content, title=title)
+
 def main():
-    print("RAG Agent CLI. Commands: /add <title> <content>, /search <query>, /quit")
+    # Load initial docs if provided as first arg
+    if len(sys.argv) > 1:
+        load_documents_from_dir(Path(sys.argv[1]))
+
+    print("Agent ready. Commands: /add <title> <file>, /search <query>, /quit")
     while True:
-        try:
-            inp = input("> ").strip()
-        except EOFError:
-            break
-        if not inp:
+        user_input = input("> ").strip()
+        if not user_input:
             continue
-        if inp.lower() in ("quit", "exit"):
-            print("Bye!")
+        if user_input.lower() in {"quit", "exit"} or user_input == "/quit":
+            print("Goodbye!")
             break
 
-        # Add document
-        if inp.startswith("/add "):
-            parts = inp[5:].split(None, 1)
-            if len(parts) != 2:
-                print("Usage: /add <title> <content>")
+        if user_input.startswith("/add"):
+            parts = user_input.split(maxsplit=2)
+            if len(parts) < 3:
+                print("Usage: /add <title> <file_path>")
                 continue
-            title, content = parts
-            res = agent.invoke({"messages": [{"role": "human", "content": f"Add document '{title}'"}]})
-            for msg in res["messages"]:
-                if hasattr(msg, "tool_calls"):
-                    print(msg.content)
+            title, file_path = parts[1], Path(parts[2])
+            if not file_path.is_file():
+                print(f"File {file_path} does not exist.")
                 continue
+            content = file_path.read_text(encoding="utf-8")
+            result = add_to_knowledge_base(content=content, title=title)
+            print(result)
 
-        # Search query
-        if inp.startswith("/search "):
-            query = inp[8:].strip()
-            res = agent.invoke({"messages": [{"role": "human", "content": f"Search: {query}"}]})
-            for msg in res["messages"]:
-                if hasattr(msg, "tool_calls"):
-                    print(msg.content)
+        elif user_input.startswith("/search"):
+            query = user_input[len("/search"):].strip()
+            if not query:
+                print("Usage: /search <query>")
                 continue
+            response = agent.invoke({"messages": [{"role": "human", "content": query}]})
+            for msg in response["messages"]:
+                if hasattr(msg, "content"):
+                    print(msg.content)
 
-        # General chat
-        res = agent.invoke({"messages": [{"role": "human", "content": inp}]})
-        for msg in res["messages"]:
+        else:
+            # Regular chat with agent
+            response = agent.invoke({"messages": [{"role": "human", "content": user_input}]})
+            for msg in response["messages"]:
                 if hasattr(msg, "content"):
                     print(msg.content)