06_langgraph_intro.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Wieso LangChain, LCEL und LangGraph\n",
    "\n",
    "Wieso gibt es diese Abstraktionen LCEL und LangGraph.\n",
    "\n",
    "Das hat vornehmlich historische Gründe.\n",
    "\n",
    "- LangChain ganz zu Anfang war der Versuch, einheitliche Interfaces für LLMs, VectorStores etc. bereit zu stellen und daraus vorgefertigte Chains zu bauen.\n",
    "- Damals gab es noch keine LCEL - Diese lustige Operatorschreibweise mit den Pipes |\n",
    "- Dann kam die LCEL, weil man etwas Hübscheres wollte. Aus Chains wurden Runnables und ein paar andere Sachen haben sich geändert.\n",
    "- Die LCEL hatte nicht vorgesehen, zyklische Datenflüsse abzubilden. Aber die Anforderungen gingen immer mehr in Richtung Agent.\n",
    "- Daher die Idee, speziell für agentische, zyklische Anwendungen neue Abstraktionen zu bauen -> LangGraph.\n",
    "\n",
    "Nun koexistieren die LCEL und LangGraph. Man möchte meist beides in größeren Anwendungen benutzen. Für simple Chains die LCEL, für komplexe Graphen LangGraph.\n",
    "\n",
    "Die beiden Baukästen haben auch viele gemeinsame Schnittpunkte:\n",
    "\n",
    "- Das Callback-System\n",
    "- Das Config-Objekt\n",
    "- Der Graph erbt auch von der Runnable Klasse.\n",
    "- Man kann Chains in den Graph stecken und Graphen chainen...\n",
    "- Langsmith-Tracing\n",
    "- ...\n",
    "\n",
    "Legacy-Chains im LangChain-Repo, die nicht mit LCEL oder LangGraph erstellt wurden, werden je nach Fleiß der Commnity schrittweise umgebaut. Die alten Chains werden dann nicht mehr gepflegt.\n",
    "\n",
    "Mit den schnell besser werdenden LLMs ändern sich die Möglichkeiten und die Anforderungen an Frameworks.\n",
    "\n",
    "Vielleicht sieht LangChain in zwei Jahren schon wieder ganz anders aus oder ist überhaupt nicht mehr das Framework der Wahl.\n",
    "\n",
    "Multimodale Inputs und Outputs werden die LLM-Welt auch noch einmal eine Runde weiter drehen und vielleicht neue Frameworks auf den Markt werfen.\n",
    "\n",
    "Damit muss man leben können. Aktuell findet LangChain allerdings sehr großen Anklang. Machen wir uns also die Mühe und schauen etwas unter die Motorhaube.\n",
    "\n",
    "### Die LCEL kennen wir schon:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from helpers import llm\n",
    "from langchain.schema import StrOutputParser\n",
    "from langchain.prompts import ChatPromptTemplate\n",
    "\n",
    "prompt = ChatPromptTemplate.from_template(\n",
    "    \"hallo Freund! Erzähle mir einen Witz über {topic}\"\n",
    ")\n",
    "model = llm()\n",
    "\n",
    "lcel_chain = prompt | model | StrOutputParser()\n",
    "\n",
    "answer_str = lcel_chain.invoke({\"topic\": \"star trek\"})\n",
    "print(answer_str)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Nun wollen wir uns LangGraph anschauen und uns einen Agenten selbst bauen ohne Import von Prebuilt\n",
    "\n",
    "Das Erstellen von eigenen Graphen ist die eigentliche Super-Power von LangGraph.\n",
    "\n",
    "Unser einfachster Graph hat nicht einmal ein Tool. Er hat nur eine Node, das LLM. Das tut natürlich nicht viel, aber wir fangen simpel an.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from langgraph.graph import MessageGraph\n",
    "\n",
    "# MessageGraph ist eine kleine Abstraktion. Das ist nichts anderes als StateGraph(List[MessageLikeObject])\n",
    "graph_builder = MessageGraph()\n",
    "# Wir definieren unser LLM als einzigen Knoten\n",
    "graph_builder.add_node(\"chatbot_node\", model)\n",
    "# Dieser Knoten wird der Entrypoint\n",
    "graph_builder.set_entry_point(\"chatbot_node\")\n",
    "# Von diesem Knoten geht es direkt zu \"END\", dem vordefinierten Terminalknoten\n",
    "graph_builder.set_finish_point(\"chatbot_node\")\n",
    "\n",
    "# Jetzt kompilieren wir\n",
    "simple_graph = graph_builder.compile()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Den Graph darstellen\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from IPython.display import Image\n",
    "\n",
    "display(Image(simple_graph.get_graph().draw_mermaid_png()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from langchain_core.messages import HumanMessage\n",
    "\n",
    "inputs = HumanMessage(content=\"Hello, my name is Bob\")\n",
    "\n",
    "\n",
    "message_history = simple_graph.invoke(inputs)\n",
    "for message in message_history:\n",
    "    message.pretty_print()"
   ]
  },
  {
   "attachments": {
    "3a047eb5-c3ea-439f-93db-132c12ed322a.png": {
     "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXMAAAHrCAYAAAA9j328AAAAAXNSR0IArs4c6QAAJyh0RVh0bXhm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\nRK5CYII=\n"
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Nun bauen wir zur Übung den Agenten aus dem bekannten Diagramm selbst.\n",
    "\n",
    "![simple_agent.png](attachment:3a047eb5-c3ea-439f-93db-132c12ed322a.png)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Wir definieren ein Tool\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from langchain.tools.arxiv.tool import ArxivQueryRun\n",
    "\n",
    "tool = ArxivQueryRun()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Wir definieren zur Übung manuell unseren AgentState\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from typing import TypedDict, Annotated, Sequence\n",
    "import operator\n",
    "from langchain_core.messages import BaseMessage\n",
    "\n",
    "\n",
    "class AgentState(TypedDict):\n",
    "    messages: Annotated[Sequence[BaseMessage], operator.add]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Wir binden unser tool ans LLM\n",
    "\n",
    "Hä? Tools ans LLM binden?\n",
    "\n",
    "- Wir hatten oben gesehen, dass unser LLM genau wissen muss, welche Tools es benutzen darf.\n",
    "- Bei OpenAI und andere großen LLM Providern kann man einfach die Tools an das LLM binden. Damit werden die erlaubten Tools automatisch von OpenAI ans Prompt gehängt.\n",
    "- Was bringt das? Die Toolbeschreibungen landen in dem von OpenAI gewünschten Format im Prompt. Das macht Toolcalling zuverlässiger. Außerdem erspart es Arbeit.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_with_tools = model.bind_tools([tool])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Wir betten das Tool und das LLM nun in Funktionen ein, die die folgende Signatur haben\n",
    "\n",
    "- Input: Der State\n",
    "- Output: Modifikation des State\n",
    "\n",
    "Diese Funktionen werden die Knoten.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from langchain_core.messages import ToolMessage\n",
    "from langchain_core.runnables import chain\n",
    "from langgraph.prebuilt import ToolNode\n",
    "\n",
    "tools = [tool]\n",
    "\n",
    "\n",
    "@chain\n",
    "def tool_executor(tool_call):\n",
    "    tool = {tool.name: tool for tool in tools}[tool_call[\"name\"]]\n",
    "    return ToolMessage(tool.invoke(tool_call[\"args\"]), tool_call_id=tool_call[\"id\"])\n",
    "\n",
    "\n",
    "def agent(state):\n",
    "    messages = state[\"messages\"]\n",
    "    response = model_with_tools.invoke(messages)\n",
    "    return {\"messages\": [response]}\n",
    "\n",
    "\n",
    "tool_node = ToolNode(tools)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Die Kanten bekommen als Input den State, aber modifizieren ihn nicht\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from typing import Literal\n",
    "\n",
    "from langgraph.graph import END\n",
    "\n",
    "\n",
    "def should_continue(state) -> Literal[\"call_tools\", END]:  # type: ignore\n",
    "    return \"call_tools\" if state[\"messages\"][-1].tool_calls else END"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Kompilieren. Kennt man schon\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from langgraph.graph import StateGraph\n",
    "\n",
    "graph_builder = StateGraph(AgentState)\n",
    "\n",
    "graph_builder.add_node(\"agent\", agent)\n",
    "graph_builder.add_node(\"call_tools\", tool_node)\n",
    "\n",
    "graph_builder.set_entry_point(\"agent\")\n",
    "\n",
    "graph_builder.add_conditional_edges(\"agent\", should_continue)\n",
    "\n",
    "graph_builder.add_edge(\"call_tools\", \"agent\")\n",
    "\n",
    "graph_with_tools = graph_builder.compile()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "display(Image(graph_with_tools.get_graph().draw_mermaid_png()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "inputs = {\n",
    "    \"messages\": [\n",
    "        HumanMessage(content=\"Write a very short summary about supernovas. Use Arxiv.\")\n",
    "    ]\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for chunk in graph_with_tools.stream(inputs):\n",
    "    for state in chunk.values():\n",
    "        for message in state[\"messages\"]:\n",
    "            message.pretty_print()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## ✅ Aufgabe\n",
    "\n",
    "Der Surfer im Netz hat sicherlich nicht genug Zeit, unseren gesamten tollen Artikel über Supernovas zu lesen. Wir wollen deshalb ganz sicher sein, dass die Zusammenfassung nicht mehr als 120 Zeichen beinhaltet.\n",
    "\n",
    "Wir können in unserer conditional_edge einfach prüfen, wie lange der String ist. Wenn er zu lange ist, fragen wir das LLM (in einer Node, die wir neu definieren), ob es das bitte kürzer fassen kann.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def should_continue(state) -> Literal[\"call_tools\", \"rewrite\", END]:  # type: ignore\n",
    "    if state[\"messages\"][-1].tool_calls:\n",
    "        return \"call_tools\"\n",
    "    elif len(state[\"messages\"][-1].content) <= 120:\n",
    "        return END\n",
    "    # Hier die Conditional Edge vervollständigen"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from langchain.schema import SystemMessage\n",
    "\n",
    "\n",
    "def rewrite(state):\n",
    "    too_long_summary = state[\"messages\"][-1].content\n",
    "    rewriter_input = [\n",
    "        SystemMessage(\n",
    "            content=\"Please shorten this text. It must be less than 120 characters.\"\n",
    "        ),\n",
    "        HumanMessage(content=too_long_summary),\n",
    "    ]\n",
    "    # Den Output des Rewriters zurückgeben\n",
    "    return {\"messages\": [response]}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Den Graph vervollständigen. Eine zusätzliche Node und eine zusätzliche Conditional Edge sind ausreichend.\n",
    "from langgraph.graph import StateGraph\n",
    "\n",
    "graph_builder = StateGraph(AgentState)\n",
    "\n",
    "graph_builder.add_node(\"agent\", agent)\n",
    "graph_builder.add_node(\"call_tools\", tool_node)\n",
    "# graph_builder.add_node...\n",
    "\n",
    "graph_builder.set_entry_point(\"agent\")\n",
    "graph_builder.add_conditional_edges(\"agent\", should_continue)\n",
    "# graph_builder.add_conditional_edges...\n",
    "graph_builder.add_edge(\"call_tools\", \"agent\")\n",
    "\n",
    "graph_with_many_loops = graph_builder.compile()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "display(Image(graph_with_many_loops.get_graph().draw_mermaid_png()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for chunk in graph_with_many_loops.stream(inputs):\n",
    "    for state in chunk.values():\n",
    "        for message in state[\"messages\"]:\n",
    "            message.pretty_print()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": ".virtualenv",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}