{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "题号: 013235 , 字段: usages 中已有该数据: 20230331\t2023届高三10班\t1.000\n",
      "题号: 013218 , 字段: usages 中已有该数据: 20230331\t2023届高三10班\t1.000\n",
      "题号: 013237 , 字段: usages 中已有该数据: 20230331\t2023届高三10班\t0.000\n",
      "题号: 014640 , 字段: usages 中已有该数据: 20230331\t2023届高三10班\t1.000\n",
      "题号: 014642 , 字段: usages 中已有该数据: 20230331\t2023届高三10班\t1.000\n",
      "题号: 014643 , 字段: usages 中已有该数据: 20230331\t2023届高三10班\t1.000\t1.000\n",
      "题号: 014636 , 字段: usages 中已有该数据: 20230331\t2023届高三10班\t1.000\n",
      "题号: 014088 , 字段: usages 中已有该数据: 20230331\t2023届高三10班\t1.000\n",
      "题号: 014085 , 字段: usages 中已有该数据: 20230331\t2023届高三10班\t1.000\t1.000\t1.000\n",
      "题号: 014644 , 字段: usages 中已有该数据: 20230331\t2023届高三10班\t1.000\t1.000\t0.000\n",
      "题号: 014645 , 字段: usages 中已有该数据: 20230331\t2023届高三10班\t1.000\t0.000\t1.000\n",
      "题号: 014646 , 字段: usages 中已有该数据: 20230331\t2023届高三10班\t1.000\t1.000\t0.000\n",
      "题号: 013235 , 字段: usages 中已有该数据: 20230331\t2023届高三11班\t0.800\n",
      "题号: 013218 , 字段: usages 中已有该数据: 20230331\t2023届高三11班\t1.000\n",
      "题号: 013237 , 字段: usages 中已有该数据: 20230331\t2023届高三11班\t0.350\n",
      "题号: 014640 , 字段: usages 中已有该数据: 20230331\t2023届高三11班\t0.850\n",
      "题号: 014642 , 字段: usages 中已有该数据: 20230331\t2023届高三11班\t0.750\n",
      "题号: 014643 , 字段: usages 中已有该数据: 20230331\t2023届高三11班\t0.950\t0.900\n",
      "题号: 014636 , 字段: usages 中已有该数据: 20230331\t2023届高三11班\t1.000\n",
      "题号: 014088 , 字段: usages 中已有该数据: 20230331\t2023届高三11班\t1.000\n",
      "题号: 014085 , 字段: usages 中已有该数据: 20230331\t2023届高三11班\t0.950\t0.750\t0.850\n",
      "题号: 014644 , 字段: usages 中已有该数据: 20230331\t2023届高三11班\t0.900\t0.650\t0.350\n",
      "题号: 014645 , 字段: usages 中已有该数据: 20230331\t2023届高三11班\t0.650\t0.600\t0.600\n",
      "题号: 014646 , 字段: usages 中已有该数据: 20230331\t2023届高三11班\t0.900\t0.600\t0.350\n",
      "题号: 013235 , 字段: usages 中已有该数据: 20230331\t2023届高三12班\t0.842\n",
      "题号: 013218 , 字段: usages 中已有该数据: 20230331\t2023届高三12班\t1.000\n",
      "题号: 013237 , 字段: usages 中已有该数据: 20230331\t2023届高三12班\t0.105\n",
      "题号: 014640 , 字段: usages 中已有该数据: 20230331\t2023届高三12班\t0.842\n",
      "题号: 014642 , 字段: usages 中已有该数据: 20230331\t2023届高三12班\t0.421\n",
      "题号: 014643 , 字段: usages 中已有该数据: 20230331\t2023届高三12班\t0.947\t0.895\n",
      "题号: 014636 , 字段: usages 中已有该数据: 20230331\t2023届高三12班\t0.947\n",
      "题号: 014088 , 字段: usages 中已有该数据: 20230331\t2023届高三12班\t1.000\n",
      "题号: 014085 , 字段: usages 中已有该数据: 20230331\t2023届高三12班\t1.000\t0.789\t0.895\n",
      "题号: 014644 , 字段: usages 中已有该数据: 20230331\t2023届高三12班\t0.947\t0.947\t0.000\n",
      "题号: 014645 , 字段: usages 中已有该数据: 20230331\t2023届高三12班\t0.895\t0.842\t0.895\n",
      "题号: 014646 , 字段: usages 中已有该数据: 20230331\t2023届高三12班\t0.947\t0.789\t0.842\n",
      "题号: 013235 , 字段: usages 中已有该数据: 20230331\t2023届高三02班\t0.893\n",
      "题号: 013218 , 字段: usages 中已有该数据: 20230331\t2023届高三02班\t0.964\n",
      "题号: 013237 , 字段: usages 中已有该数据: 20230331\t2023届高三02班\t0.071\n",
      "题号: 014640 , 字段: usages 中已有该数据: 20230331\t2023届高三02班\t0.893\n",
      "题号: 014642 , 字段: usages 中已有该数据: 20230331\t2023届高三02班\t0.714\n",
      "题号: 014643 , 字段: usages 中已有该数据: 20230331\t2023届高三02班\t0.964\t0.786\n",
      "题号: 014636 , 字段: usages 中已有该数据: 20230331\t2023届高三02班\t0.929\n",
      "题号: 014088 , 字段: usages 中已有该数据: 20230331\t2023届高三02班\t0.964\n",
      "题号: 014085 , 字段: usages 中已有该数据: 20230331\t2023届高三02班\t1.000\t0.929\t0.821\n",
      "题号: 014644 , 字段: usages 中已有该数据: 20230331\t2023届高三02班\t0.929\t0.929\t0.000\n",
      "题号: 014645 , 字段: usages 中已有该数据: 20230331\t2023届高三02班\t0.964\t0.893\t0.821\n",
      "题号: 014646 , 字段: usages 中已有该数据: 20230331\t2023届高三02班\t0.929\t0.750\t0.821\n",
      "题号: 013235 , 字段: usages 中已有该数据: 20230331\t2023届高三03班\t1.000\n",
      "题号: 013218 , 字段: usages 中已有该数据: 20230331\t2023届高三03班\t0.913\n",
      "题号: 013237 , 字段: usages 中已有该数据: 20230331\t2023届高三03班\t0.391\n",
      "题号: 014640 , 字段: usages 中已有该数据: 20230331\t2023届高三03班\t0.870\n",
      "题号: 014642 , 字段: usages 中已有该数据: 20230331\t2023届高三03班\t0.652\n",
      "题号: 014643 , 字段: usages 中已有该数据: 20230331\t2023届高三03班\t1.000\t0.870\n",
      "题号: 014636 , 字段: usages 中已有该数据: 20230331\t2023届高三03班\t1.000\n",
      "题号: 014088 , 字段: usages 中已有该数据: 20230331\t2023届高三03班\t1.000\n",
      "题号: 014085 , 字段: usages 中已有该数据: 20230331\t2023届高三03班\t0.913\t1.000\t0.957\n",
      "题号: 014644 , 字段: usages 中已有该数据: 20230331\t2023届高三03班\t0.913\t0.913\t0.000\n",
      "题号: 014645 , 字段: usages 中已有该数据: 20230331\t2023届高三03班\t0.696\t0.783\t0.739\n",
      "题号: 014646 , 字段: usages 中已有该数据: 20230331\t2023届高三03班\t0.913\t0.826\t0.696\n",
      "题号: 013235 , 字段: usages 中已有该数据: 20230331\t2023届高三04班\t0.938\n",
      "题号: 013218 , 字段: usages 中已有该数据: 20230331\t2023届高三04班\t0.875\n",
      "题号: 013237 , 字段: usages 中已有该数据: 20230331\t2023届高三04班\t0.000\n",
      "题号: 014640 , 字段: usages 中已有该数据: 20230331\t2023届高三04班\t0.875\n",
      "题号: 014642 , 字段: usages 中已有该数据: 20230331\t2023届高三04班\t0.562\n",
      "题号: 014643 , 字段: usages 中已有该数据: 20230331\t2023届高三04班\t1.000\t0.750\n",
      "题号: 014636 , 字段: usages 中已有该数据: 20230331\t2023届高三04班\t1.000\n",
      "题号: 014088 , 字段: usages 中已有该数据: 20230331\t2023届高三04班\t1.000\n",
      "题号: 014085 , 字段: usages 中已有该数据: 20230331\t2023届高三04班\t1.000\t0.812\t0.812\n",
      "题号: 014644 , 字段: usages 中已有该数据: 20230331\t2023届高三04班\t1.000\t0.938\t0.062\n",
      "题号: 014645 , 字段: usages 中已有该数据: 20230331\t2023届高三04班\t0.812\t0.688\t0.750\n",
      "题号: 014646 , 字段: usages 中已有该数据: 20230331\t2023届高三04班\t0.938\t0.812\t0.438\n",
      "题号: 013235 , 字段: usages 中已有该数据: 20230331\t2023届高三05班\t0.865\n",
      "题号: 013218 , 字段: usages 中已有该数据: 20230331\t2023届高三05班\t0.973\n",
      "题号: 013237 , 字段: usages 中已有该数据: 20230331\t2023届高三05班\t0.135\n",
      "题号: 014640 , 字段: usages 中已有该数据: 20230331\t2023届高三05班\t0.946\n",
      "题号: 014642 , 字段: usages 中已有该数据: 20230331\t2023届高三05班\t0.730\n",
      "题号: 014643 , 字段: usages 中已有该数据: 20230331\t2023届高三05班\t1.000\t1.000\n",
      "题号: 014636 , 字段: usages 中已有该数据: 20230331\t2023届高三05班\t0.973\n",
      "题号: 014088 , 字段: usages 中已有该数据: 20230331\t2023届高三05班\t1.000\n",
      "题号: 014085 , 字段: usages 中已有该数据: 20230331\t2023届高三05班\t0.946\t0.838\t0.946\n",
      "题号: 014644 , 字段: usages 中已有该数据: 20230331\t2023届高三05班\t0.351\t0.919\t0.378\n",
      "题号: 014645 , 字段: usages 中已有该数据: 20230331\t2023届高三05班\t0.378\t0.513\t0.622\n",
      "题号: 014646 , 字段: usages 中已有该数据: 20230331\t2023届高三05班\t0.892\t0.622\t0.513\n",
      "题号: 013235 , 字段: usages 中已有该数据: 20230331\t2023届高三06班\t0.974\n",
      "题号: 013218 , 字段: usages 中已有该数据: 20230331\t2023届高三06班\t0.846\n",
      "题号: 013237 , 字段: usages 中已有该数据: 20230331\t2023届高三06班\t0.179\n",
      "题号: 014640 , 字段: usages 中已有该数据: 20230331\t2023届高三06班\t0.974\n",
      "题号: 014642 , 字段: usages 中已有该数据: 20230331\t2023届高三06班\t0.769\n",
      "题号: 014643 , 字段: usages 中已有该数据: 20230331\t2023届高三06班\t0.974\t0.974\n",
      "题号: 014636 , 字段: usages 中已有该数据: 20230331\t2023届高三06班\t0.974\n",
      "题号: 014088 , 字段: usages 中已有该数据: 20230331\t2023届高三06班\t1.000\n",
      "题号: 014085 , 字段: usages 中已有该数据: 20230331\t2023届高三06班\t1.000\t0.872\t0.974\n",
      "题号: 014644 , 字段: usages 中已有该数据: 20230331\t2023届高三06班\t1.000\t1.000\t0.051\n",
      "题号: 014645 , 字段: usages 中已有该数据: 20230331\t2023届高三06班\t0.692\t0.718\t0.795\n",
      "题号: 014646 , 字段: usages 中已有该数据: 20230331\t2023届高三06班\t0.667\t0.872\t0.590\n",
      "题号: 013235 , 字段: usages 中已有该数据: 20230331\t2023届高三07班\t0.889\n",
      "题号: 013218 , 字段: usages 中已有该数据: 20230331\t2023届高三07班\t0.926\n",
      "题号: 013237 , 字段: usages 中已有该数据: 20230331\t2023届高三07班\t0.037\n",
      "题号: 014640 , 字段: usages 中已有该数据: 20230331\t2023届高三07班\t0.889\n",
      "题号: 014642 , 字段: usages 中已有该数据: 20230331\t2023届高三07班\t0.667\n",
      "题号: 014643 , 字段: usages 中已有该数据: 20230331\t2023届高三07班\t0.963\t0.926\n",
      "题号: 014636 , 字段: usages 中已有该数据: 20230331\t2023届高三07班\t0.926\n",
      "题号: 014088 , 字段: usages 中已有该数据: 20230331\t2023届高三07班\t1.000\n",
      "题号: 014085 , 字段: usages 中已有该数据: 20230331\t2023届高三07班\t0.926\t0.852\t0.704\n",
      "题号: 014644 , 字段: usages 中已有该数据: 20230331\t2023届高三07班\t0.963\t0.963\t0.037\n",
      "题号: 014645 , 字段: usages 中已有该数据: 20230331\t2023届高三07班\t0.852\t0.778\t0.778\n",
      "题号: 014646 , 字段: usages 中已有该数据: 20230331\t2023届高三07班\t1.000\t0.556\t0.518\n",
      "题号: 013235 , 字段: usages 中已有该数据: 20230331\t2023届高三09班\t0.842\n",
      "题号: 013218 , 字段: usages 中已有该数据: 20230331\t2023届高三09班\t1.000\n",
      "题号: 013237 , 字段: usages 中已有该数据: 20230331\t2023届高三09班\t0.053\n",
      "题号: 014640 , 字段: usages 中已有该数据: 20230331\t2023届高三09班\t0.895\n",
      "题号: 014642 , 字段: usages 中已有该数据: 20230331\t2023届高三09班\t0.632\n",
      "题号: 014643 , 字段: usages 中已有该数据: 20230331\t2023届高三09班\t1.000\t0.895\n",
      "题号: 014636 , 字段: usages 中已有该数据: 20230331\t2023届高三09班\t0.947\n",
      "题号: 014088 , 字段: usages 中已有该数据: 20230331\t2023届高三09班\t1.000\n",
      "题号: 014085 , 字段: usages 中已有该数据: 20230331\t2023届高三09班\t1.000\t0.842\t0.947\n",
      "题号: 014644 , 字段: usages 中已有该数据: 20230331\t2023届高三09班\t0.895\t0.895\t0.105\n",
      "题号: 014645 , 字段: usages 中已有该数据: 20230331\t2023届高三09班\t0.789\t0.632\t0.526\n",
      "题号: 014646 , 字段: usages 中已有该数据: 20230331\t2023届高三09班\t0.842\t0.421\t0.105\n"
     ]
    }
   ],
   "source": [
    "import os,re,json\n",
    "\n",
    "\"\"\"---明确数据文件位置---\"\"\"\n",
    "datafile = \"文本文件/metadata.txt\"\n",
    "# 双回车分隔，记录内单回车分隔列表，首行为字段名\n",
    "\"\"\"---文件位置结束---\"\"\"\n",
    "\n",
    "def trim(string):\n",
    "    string = re.sub(r\"^[ \\t\\n]*\",\"\",string)\n",
    "    string = re.sub(r\"[ \\t\\n]*$\",\"\",string)\n",
    "    return string\n",
    "def FloatToInt(string):\n",
    "    f = float(string)\n",
    "    if abs(f-round(f))<0.01:\n",
    "        f = round(f)\n",
    "    return f\n",
    "\n",
    "with open(datafile,\"r\",encoding=\"utf8\") as f:\n",
    "    data = f.read().strip()\n",
    "pos = data.index(\"\\n\")\n",
    "field = data[:pos].strip()\n",
    "appending_data = data[pos:]\n",
    "\n",
    "with open(r\"../题库0.3/Problems.json\",\"r\",encoding = \"utf8\") as f:\n",
    "    database = f.read()\n",
    "pro_dict = json.loads(database)\n",
    "with open(r\"../题库0.3/LessonObj.json\",\"r\",encoding = \"utf8\") as f:\n",
    "    database = f.read()\n",
    "obj_dict = json.loads(database)\n",
    "\n",
    "#该字段列表可能需要更新\n",
    "fields = [\"content\",\"objs\",\"tags\",\"genre\",\"ans\",\"solution\",\"duration\",\"usages\",\"origin\",\"edit\",\"same\",\"related\",\"remark\",\"space\"]\n",
    "\n",
    "if field in fields:\n",
    "    field_type = type(pro_dict[\"000001\"][field])\n",
    "    datalist = [record.strip() for record in appending_data.split(\"\\n\\n\") if len(trim(record)) > 0]\n",
    "    for record in datalist:\n",
    "        id = re.findall(r\"^[\\d]{1,}\",record)[0]\n",
    "        data = record[len(id):].strip()\n",
    "        id = id.zfill(6)\n",
    "        if not id in pro_dict:\n",
    "            print(\"题号:\",id,\"不在数据库中.\")\n",
    "            break\n",
    "        \n",
    "        #字符串类型字段添加数据\n",
    "        elif field_type == str and data in pro_dict[id][field]:\n",
    "            print(\"题号:\",id,\", 字段:\",field,\"中已有该数据:\",data)\n",
    "        elif field_type == str and not data in pro_dict[id][field] and not field == \"ans\" and not field == \"space\":\n",
    "            origin_data = pro_dict[id][field]\n",
    "            new_data = trim(origin_data + \"\\n\" + data)\n",
    "            pro_dict[id][field] = new_data\n",
    "            print(\"题号:\",id,\", 字段:\",field,\"中已添加数据:\",data)\n",
    "        elif field_type == str and not data in pro_dict[id][field] and field == \"ans\" or field == \"space\":\n",
    "            pro_dict[id][field] = data\n",
    "            print(\"题号:\",id,\", 字段:\",field,\"中已修改数据:\",data)\n",
    "        \n",
    "        #数值类型字段添加数据\n",
    "        elif (field_type == int or field_type == float) and abs(float(data) - pro_dict[id][field])<0.01:\n",
    "            print(\"题号:\",id,\", 字段:\",field,\"中已有该数据:\",FloatToInt(data))\n",
    "        elif (field_type == int or field_type == float) and abs(float(data) - pro_dict[id][field])>=0.01:\n",
    "            pro_dict[id][field] = FloatToInt(data)\n",
    "            print(\"题号:\",id,\", 字段:\",field,\"中已修改数据:\",FloatToInt(data))\n",
    "        \n",
    "        #列表类型字段添加数据\n",
    "        elif field_type == list:\n",
    "            cell_data_list = [d.strip() for d in data.split(\"\\n\")]\n",
    "            for cell_data in cell_data_list:\n",
    "                if cell_data in pro_dict[id][field]:\n",
    "                    print(\"题号:\",id,\", 字段:\",field,\"中已有该数据:\",cell_data)\n",
    "                elif not field == \"objs\":\n",
    "                    pro_dict[id][field].append(cell_data)\n",
    "                    print(\"题号:\",id,\", 字段:\",field,\"中已添加数据:\",cell_data)\n",
    "                else:\n",
    "                    if not cell_data in obj_dict and not cell_data.upper() == \"KNONE\":\n",
    "                        print(\"题号:\",id,\", 字段:\",field,\"目标编号有误:\",cell_data)\n",
    "                    else:\n",
    "                        pro_dict[id][field].append(cell_data.upper())\n",
    "                        print(\"题号:\",id,\", 字段:\",field,\"中已添加数据:\",cell_data.upper())\n",
    "else:\n",
    "    print(\"字段名有误\")\n",
    "\n",
    "with open(r\"../题库0.3/Problems.json\",\"w\",encoding = \"utf8\") as f:\n",
    "    f.write(json.dumps(pro_dict,indent=4,ensure_ascii=False))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "mathdept",
   "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": "3.9.15"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "ff3c292c316ba85de6f1ad75f19c731e79d694e741b6f515ec18f14996fe48dc"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}