{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "\n",
    "with open('../data/task1.json', 'r') as f:\n",
    "    test = json.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "draft_ids = []\n",
    "authors = []\n",
    "coauthors = []\n",
    "choices_2 = []\n",
    "choices_3 = []\n",
    "choices_4 = []\n",
    "choices_5 = []\n",
    "\n",
    "for instance in test:\n",
    "    draft_ids.append(instance['folder_id'])\n",
    "    authors.append(instance['author'])\n",
    "    coauthors.append(instance['coauthor'])\n",
    "    choices_2.append(instance['choices_2'])\n",
    "    choices_3.append(instance['choices_3'])\n",
    "    choices_4.append(instance['choices_4'])\n",
    "    choices_5.append(instance['choices_5'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "path = '../data/task1'\n",
    "drafts = []\n",
    "for i in draft_ids:\n",
    "    folder_path = os.path.join(path, str(i))\n",
    "    files = os.listdir(folder_path)\n",
    "    json_file = [file for file in files if file.endswith('EN.json')][0]\n",
    "    with open(os.path.join(folder_path, json_file)) as f:\n",
    "        draft = json.load(f)\n",
    "    drafts.append(draft['Content'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# If use Together API\n",
    "from together import Together\n",
    "\n",
    "your_model_name = 'xxxxxxxxxxxxxxxxxxxxxxxx'\n",
    "your_api_key = 'xxxxxxxxxxxxxxxxxxxxxxxx'\n",
    "client = Together(api_key=your_api_key)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "from tqdm import tqdm\n",
    "\n",
    "def run_task3_k_choices(client, drafts, authors, choices_k):\n",
    "    results = []\n",
    "    invalid_responses = []\n",
    "    for i, (draft, author, choice_k) in tqdm(enumerate(zip(drafts, authors, choices_k))):\n",
    "        system_prompt = f\"\"\"You are representing {author}, a country drafting a resolution for submission to the United Nations Security Council. \n",
    "        Your task is to review the draft resolution and select a coauthor from the following list: {', '.join(choice_k)}. \n",
    "        Respond only with the name of the chosen coauthor and provide no additional explanation.\"\"\"\n",
    "        user_prompt = f\"\"\"{author} is drafting a resolution to submit to the United Nations Security Council and is seeking a coauthor. \n",
    "\n",
    "        Review the following draft resolution and choose a coauthor from the list below. \n",
    "        Respond only with the name of the chosen country and provide no additional explanation.\n",
    "\n",
    "        Available coauthors: {', '.join(choice_k)}\n",
    "\n",
    "        Draft Resolution:\n",
    "        {draft}\n",
    "\n",
    "        Answer:\n",
    "        \"\"\"\n",
    "        \n",
    "        response = client.chat.completions.create(\n",
    "            model=your_model_name,\n",
    "            messages=[\n",
    "                {\"role\": \"system\", \"content\": system_prompt},\n",
    "                {\"role\": \"user\", \"content\": user_prompt}\n",
    "            ],\n",
    "            max_tokens=20, \n",
    "            temperature=0.0\n",
    "        )\n",
    "        result = response.choices[0].message.content.strip()\n",
    "        valid_response = choice_k\n",
    "        if result not in valid_response:\n",
    "            print(f'Invalid response: {result}')\n",
    "            result = random.choice(valid_response)\n",
    "            invalid_responses.append(i)\n",
    "        results.append(result)\n",
    "    return results, invalid_responses"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_2, invalid_responses_2 = run_task3_k_choices(client, drafts, authors, choices_2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_3, invalid_responses_3 = run_task3_k_choices(client, drafts, authors, choices_3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_4, invalid_responses_4 = run_task3_k_choices(client, drafts, authors, choices_4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_5, invalid_responses_5 = run_task3_k_choices(client, drafts, authors, choices_5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def calculate_accuracy(results, ground_truth):\n",
    "    correct = 0\n",
    "    for i in range(len(results)):\n",
    "        if results[i] == ground_truth[i]:\n",
    "            correct += 1\n",
    "    return correct/len(results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "accuracy_2 = calculate_accuracy(results_2, coauthors)\n",
    "accuracy_3 = calculate_accuracy(results_3, coauthors)\n",
    "accuracy_4 = calculate_accuracy(results_4, coauthors)\n",
    "accuracy_5 = calculate_accuracy(results_5, coauthors)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(f'Accuracy for 2 choices: {accuracy_2}')\n",
    "print(f'Accuracy for 3 choices: {accuracy_3}')\n",
    "print(f'Accuracy for 4 choices: {accuracy_4}')\n",
    "print(f'Accuracy for 5 choices: {accuracy_5}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "llm",
   "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": "3.9.19"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}