{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "dc4f5934-9bb9-4bdc-975a-86bed379f9b5",
   "metadata": {},
   "source": [
    "好的，我来修改脚本以满足您的要求：\n",
    "\n",
    "```python\n",
    "import pandas as pd\n",
    "import os\n",
    "\n",
    "# 读取parquet文件\n",
    "df = pd.read_parquet(\"exp2/table.parquet\")\n",
    "\n",
    "# 创建输出目录（如果不存在）\n",
    "output_dir = \"output_txt\"\n",
    "if not os.path.exists(output_dir):\n",
    "    os.makedirs(output_dir)\n",
    "\n",
    "# 遍历每一行数据\n",
    "for _, row in df.iterrows():\n",
    "    # 处理每个字段，将空格分隔的标签转换为逗号加空格分隔\n",
    "    artist = row['tag_string_artist'].replace(' ', ', ')\n",
    "    character = row['tag_string_character'].replace(' ', ', ')\n",
    "    general = row['tag_string_general'].replace(' ', ', ')\n",
    "    meta = row['tag_string_meta'].replace(' ', ', ')\n",
    "    \n",
    "    # 组合所有标签，使用逗号和空格分隔\n",
    "    content = f\"{artist}, {character}, {general}, {meta}\"\n",
    "    \n",
    "    # 使用id作为文件名\n",
    "    filename = os.path.join(output_dir, f\"{row['id']}.txt\")\n",
    "    \n",
    "    # 写入文件\n",
    "    with open(filename, 'w', encoding='utf-8') as f:\n",
    "        f.write(content)\n",
    "\n",
    "print(\"处理完成！文件已保存到output_txt目录下\")\n",
    "print(f\"共处理了 {len(df)} 个文件\")\n",
    "```\n",
    "\n",
    "这样生成的txt文件内容会像这样：\n",
    "```\n",
    "nznzn, jinhsi_(wuthering_waves), 1girl, solo, absurdres, highres\n",
    "```\n",
    "\n",
    "所有标签都会按照 artist, character, general, meta 的顺序排列，并用逗号加空格分隔。需要我运行这个脚本吗？"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fe7391e4-f254-4003-80ef-3b633ae7c092",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import os\n",
    "from datetime import datetime\n",
    "\n",
    "# 读取parquet文件\n",
    "df = pd.read_parquet(\"./table.parquet\")\n",
    "\n",
    "# 创建输出目录（如果不存在）\n",
    "output_dir = \"output_txt\"\n",
    "if not os.path.exists(output_dir):\n",
    "    os.makedirs(output_dir)\n",
    "\n",
    "# 遍历每一行数据\n",
    "for _, row in df.iterrows():\n",
    "    # 处理每个字段,将空格分隔的标签转换为逗号加空格分隔\n",
    "    artist = str(row['tag_string_artist']).replace(' ', ', ') if row['tag_string_artist'] else ''\n",
    "    character = str(row['tag_string_character']).replace(' ', ', ') if row['tag_string_character'] else ''\n",
    "    general = str(row['tag_string_general']).replace(' ', ', ') if row['tag_string_general'] else ''\n",
    "    meta = str(row['tag_string_meta']).replace(' ', ', ') if row['tag_string_meta'] else ''\n",
    "    copyright = str(row['tag_string_copyright']).replace(' ', ', ') if row['tag_string_copyright'] else ''\n",
    "    \n",
    "    # 根据created_at添加年份标签\n",
    "    created_at = datetime.fromisoformat(row['created_at'])\n",
    "    if created_at.year == 2024:\n",
    "        year_tag = \"year_2024\"\n",
    "    elif created_at.year == 2023:\n",
    "        year_tag = \"year_2023\"\n",
    "    elif created_at.year == 2022:\n",
    "        year_tag = \"year_2022\"\n",
    "    elif created_at.year == 2021:\n",
    "        year_tag = \"year_2021\"\n",
    "    else:\n",
    "        year_tag = f\"year_{created_at.year}\"\n",
    "    \n",
    "    # 组合所有标签,使用逗号和空格分隔\n",
    "    # 可以修改顺序\n",
    "    content = f\"{artist}, {character}, {year_tag}, {general}, {meta}, {copyright}\"\n",
    "    \n",
    "    # 使用id作为文件名\n",
    "    filename = os.path.join(output_dir, f\"{row['id']}.txt\")\n",
    "    \n",
    "    # 写入文件\n",
    "    with open(filename, 'w', encoding='utf-8') as f:\n",
    "        f.write(content)\n",
    "\n",
    "print(\"处理完成!文件已保存到output_txt目录下\")\n",
    "print(f\"共处理了 {len(df)} 个文件\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.10.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}