20220805

高考数据导入.ipynb

@@ -286,26 +286,11 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 112,
+   "execution_count": null,
    "metadata": {
-    "execution": {
-     "iopub.execute_input": "2022-06-15T09:08:30.396155Z",
-     "iopub.status.busy": "2022-06-15T09:08:30.395636Z",
-     "iopub.status.idle": "2022-06-15T09:08:39.230104Z",
-     "shell.execute_reply": "2022-06-15T09:08:39.228838Z",
-     "shell.execute_reply.started": "2022-06-15T09:08:30.396108Z"
-    },
     "tags": []
    },
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "ok\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
     "import pymysql\n",
     "import pymongo\n",
@@ -751,7 +736,9 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {},
+   "metadata": {
+    "tags": []
+   },
    "source": [
     "## 导入2021年录取数据"
    ]
@@ -1005,6 +992,390 @@
     "db.close() "
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## 导入2022年录取数据"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 导出2022年录取数据并生成分数"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "import openpyxl\n",
+    "import json\n",
+    "\n",
+    "filename = 'data/17-22年一分一段表.json'\n",
+    "with open(filename,'r') as fl:\n",
+    "    m_xx = json.load(fl)\n",
+    "data = m_xx['2022']['z']\n",
+    "yfyd = {}\n",
+    "for k, v in data.items():\n",
+    "    list1 = []\n",
+    "    list1 = (v['min_rank'],v['max_rank'])\n",
+    "    yfyd[k] = list1\n",
+    "dict1 = {}\n",
+    "filename = 'data/山东省2022年普通类常规批第一次志愿投档情况表.xlsx'\n",
+    "wb = openpyxl.load_workbook(filename)\n",
+    "sheet = wb.active\n",
+    "data1 =list(sheet.values)\n",
+    "del data1[0:2]\n",
+    "m_num = 1\n",
+    "i = 1\n",
+    "for items in data1:\n",
+    "    m_rank = items[3]    \n",
+    "    list1 = []\n",
+    "    for k, v in yfyd.items():\n",
+    "        if m_rank >=v[1] and m_rank <=v[0]:\n",
+    "            m_num = int(k)\n",
+    "            break\n",
+    "    list1 = (items[0],items[1],items[2],items[3],m_num)\n",
+    "    dict1[i] = list1\n",
+    "    i += 1 \n",
+    "with open('data/2022.json','w') as fl2:\n",
+    "    json.dump(dict1,fl2) \n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 2022年录取数据导入mysql数据库"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "import pymysql\n",
+    "import json\n",
+    "\n",
+    "l_code = []\n",
+    "l_new = []\n",
+    "l_data = []\n",
+    "db = pymysql.connect(host = \"localhost\",user = \"root\",password = \"songyi\",database = \"gaokao\" )\n",
+    "cursor = db.cursor()\n",
+    "\n",
+    "sql = 'select code from college'\n",
+    "cursor.execute(sql)\n",
+    "results = cursor.fetchall()\n",
+    "for result in results:\n",
+    "    l_code.append(result[0])\n",
+    "\n",
+    "with open('data/2022.json','r') as fl:\n",
+    "    m_xx = json.load(fl)\n",
+    "for k, v in m_xx.items():\n",
+    "    #code = v[1][:4]\n",
+    "    l_data.append([v[1][:4],v[0][:2],v[2],v[4],v[3]])\n",
+    "\n",
+    "sql = \"insert into admission_2022 (college,speciality,plan,num_min,rank_min,nian) values(%s,%s,%s,%s,%s,'2022')\"\n",
+    "try:\n",
+    "    cursor.executemany(sql,l_data)\n",
+    "    db.commit()\n",
+    "    print(\"已添加！\")\n",
+    "except:\n",
+    "    # 如果发生错误则回滚\n",
+    "    print(\"error!\")\n",
+    "    db.rollback()   \n",
+    "db.close() "
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 整理2022年招生学校"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pymysql\n",
+    "import json\n",
+    "\n",
+    "l_code = []\n",
+    "l_new = []\n",
+    "l_data = []\n",
+    "db = pymysql.connect(host = \"localhost\",user = \"root\",password = \"songyi\",database = \"gaokao\" )\n",
+    "cursor = db.cursor()\n",
+    "\n",
+    "sql = 'select code from college'\n",
+    "cursor.execute(sql)\n",
+    "results = cursor.fetchall()\n",
+    "for result in results:\n",
+    "    l_code.append(result[0])\n",
+    "\n",
+    "with open('data/2022.json','r') as fl:\n",
+    "    m_xx = json.load(fl)\n",
+    "for k, v in m_xx.items():\n",
+    "    code = v[1][:4]\n",
+    "    if code not in l_code:\n",
+    "        l_code.append(code)\n",
+    "        l_new.append([v[1][0:4],v[1][4:]])\n",
+    "    #l_data.append([v[1][:4],v[0][:2],v[2],v[4],v[3]])\n",
+    "sql = \"insert into college (code,name) values(%s,%s)\"\n",
+    "try:\n",
+    "    cursor.executemany(sql,l_new)\n",
+    "    db.commit()\n",
+    "    print(\"已添加！\")\n",
+    "except:\n",
+    "    # 如果发生错误则回滚\n",
+    "    print(\"error!\")\n",
+    "    db.rollback()   \n",
+    "db.close() "
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 整理2022年新增招生学校"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "import pymongo\n",
+    "import decimal\n",
+    "import json\n",
+    "\n",
+    "\n",
+    "myclient = pymongo.MongoClient('mongodb://localhost:27017/')\n",
+    "mydb = myclient[\"gaokao\"]\n",
+    "mycol = mydb[\"college_2021\"]\n",
+    "mycol1 = mydb[\"admission_2021\"]\n",
+    "\n",
+    "m_col = {}\n",
+    "m_xx = {}\n",
+    "dict1 = {}\n",
+    "list1 = []\n",
+    "for x in mycol.find({\"code\":{'$exists': 'true'}},{\"_id\": 0, \"code\": 1, \"name\": 1}):\n",
+    "    m_col[x['code']] = x['name']\n",
+    "\n",
+    "with open('data/2022.json','r') as fl:\n",
+    "    m_xx = json.load(fl)\n",
+    "for k, v in m_xx.items():\n",
+    "    #code = v[1][:4]\n",
+    "    dict1[v[1][:4]]= v[1][4:]\n",
+    "for code in dict1.keys():\n",
+    "    if code not in m_col.keys():\n",
+    "        #print(dict1[code])\n",
+    "        \n",
+    "        if dict1[code] not in m_col.values():\n",
+    "            list1.append(code)\n",
+    "print(list1)\n",
+    "for x in mycol.find({\"id_code\":{'$exists': 'true'}},{\"_id\": 0, \"id_code\": 1, \"name\": 1}):\n",
+    "    m_col[x['id_code']] = x['name']\n",
+    "for code in list1:\n",
+    "    if dict1[code] in m_col.values():\n",
+    "        myquery = {'name':dict1[code]}\n",
+    "        m_new = {\"$set\":{'code':code}}\n",
+    "        mycol.update_one(myquery,m_new)\n",
+    "        print(dict1[code])"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 汇总2022年学校录取情况"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "import pymysql\n",
+    "import pymongo\n",
+    "\n",
+    "myclient = pymongo.MongoClient('mongodb://localhost:27017/')\n",
+    "mydb = myclient[\"gaokao\"]\n",
+    "mycol = mydb[\"college_2021\"]\n",
+    "m_xx = {}\n",
+    "m_data = {}\n",
+    "m_mongo = {}\n",
+    "db = pymysql.connect(host = \"localhost\",user = \"root\",password = \"songyi\",database = \"gaokao\" )\n",
+    "cursor = db.cursor()\n",
+    "sql = 'SELECT college,SUM(plan),min(num_min),max(RANK_min) FROM admission_2022 GROUP BY college'\n",
+    "cursor.execute(sql)\n",
+    "results = cursor.fetchall()\n",
+    "i = 1\n",
+    "for result in results:\n",
+    "    #m_xx.clear()\n",
+    "    m_code = result[0]\n",
+    "    m_nian ='2022'\n",
+    "    m_lb = 'z'\n",
+    "    m_xx.setdefault(m_code,{})\n",
+    "    m_xx[m_code].setdefault(m_nian,{})\n",
+    "    m_xx[m_code][m_nian].setdefault(m_lb,{})\n",
+    "    m_xx[m_code][m_nian][m_lb] ['name']= '综合'\n",
+    "    m_xx[m_code][m_nian][m_lb] ['dispense']= int(result[1])\n",
+    "    m_xx[m_code][m_nian][m_lb] ['num_min']= result[2]\n",
+    "    m_xx[m_code][m_nian][m_lb] ['rank_min']= result[3]    \n",
+    "    myquery = {'code':m_code}\n",
+    "    colleges = mycol.find(myquery,{ \"_id\": 0, \"admission\": 1 })\n",
+    "    for x in colleges:\n",
+    "        for k, v in x.items():\n",
+    "            for k1, v1 in v.items():\n",
+    "                m_xx[m_code][k1] = v1\n",
+    "for k,v in m_xx.items():\n",
+    "    #m_item ='admission.'+m_nian\n",
+    "    m_mongo.clear()    \n",
+    "    myquery = {'code':k}\n",
+    "    m_new = {\"$set\":{'admission':v}}\n",
+    "    mycol.update_one(myquery,m_new)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 整理2022年新增专业"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pymysql\n",
+    "import json\n",
+    "\n",
+    "l_data = []\n",
+    "db = pymysql.connect(host = \"localhost\",user = \"root\",password = \"songyi\",database = \"gaokao\" )\n",
+    "cursor = db.cursor()\n",
+    "\n",
+    "with open('data/2022.json','r') as fl:\n",
+    "    m_xx = json.load(fl)\n",
+    "for k, v in m_xx.items():\n",
+    "    l_data.append([v[0][:2],v[0][2:],v[1][0:4],'2022'])\n",
+    "sql = \"insert into speciality (code,name,college,nian) values(%s,%s,%s,%s)\"\n",
+    "try:\n",
+    "    cursor.executemany(sql,l_data)\n",
+    "    db.commit()\n",
+    "    print(\"已添加！\")\n",
+    "except:\n",
+    "    # 如果发生错误则回滚\n",
+    "    print(\"error!\")\n",
+    "    db.rollback()   \n",
+    "db.close() "
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 2022年专业录取分数及位次"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 33,
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2022-08-04T03:53:00.735786Z",
+     "iopub.status.busy": "2022-08-04T03:53:00.735257Z",
+     "iopub.status.idle": "2022-08-04T03:53:10.194351Z",
+     "shell.execute_reply": "2022-08-04T03:53:10.192776Z",
+     "shell.execute_reply.started": "2022-08-04T03:53:00.735737Z"
+    },
+    "tags": []
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "ok\n"
+     ]
+    }
+   ],
+   "source": [
+    "import pymysql\n",
+    "import pymongo\n",
+    "import decimal\n",
+    "\n",
+    "myclient = pymongo.MongoClient('mongodb://localhost:27017/')\n",
+    "mydb = myclient[\"gaokao\"]\n",
+    "mycol = mydb[\"college_2021\"]\n",
+    "mycol1 = mydb[\"admission_2022\"]\n",
+    "\n",
+    "m_col = {}\n",
+    "m_spe = {}\n",
+    "m_xx = {}\n",
+    "for x in mycol.find({\"code\":{'$exists': 'true'}},{\"_id\": 0, \"code\": 1, \"name\": 1}):\n",
+    "    m_col[x['code']] = x['name']\n",
+    "\n",
+    "\n",
+    "db = pymysql.connect(host = \"localhost\",user = \"songyi\",password = \"yylzs\",database = \"gaokao\" )\n",
+    "cursor = db.cursor()\n",
+    "sql = 'SELECT a.code,a.college,a.name FROM speciality AS a WHERE a.nian=\"2022\"'\n",
+    "cursor.execute(sql)\n",
+    "results = cursor.fetchall()\n",
+    "for result in results:\n",
+    "    m_spe.setdefault(result[1],{})    \n",
+    "    m_spe[result[1]][result[0]] = result[2]\n",
+    "#print(m_spe)\n",
+    "sql = 'SELECT * FROM admission_2022 AS a where college not in (\"D628\",\"D440\",\"Y010\",\"D245\",\"D597\",\"D641\",\"D501\",\"D659\") ORDER BY a.rank_min'\n",
+    "#sql = 'SELECT * FROM admission_2022 AS a ORDER BY a.rank_min'\n",
+    "cursor.execute(sql)\n",
+    "results = cursor.fetchall()\n",
+    "i = 0\n",
+    "m_min = 0\n",
+    "ii = 0\n",
+    "for result in results:\n",
+    "    m_xx.clear()\n",
+    "    if result[4] == m_min:\n",
+    "        ii = ii\n",
+    "        i = i+1\n",
+    "    else:\n",
+    "        i = i+1\n",
+    "        ii = i\n",
+    "        m_min = result[6]\n",
+    "    \n",
+    "    m_xx['pos'] = ii\n",
+    "    m_xx['col_code'] = result[1]\n",
+    "    m_xx['col_name'] = m_col[result[1]]\n",
+    "    m_xx['spe_code'] = result[2]\n",
+    "    m_xx['spe_name'] = m_spe[result[1]][result[2]]\n",
+    "    m_xx['plan'] = result[3]\n",
+    "    m_xx['num_min'] = result[4]    \n",
+    "    m_xx['rank_min'] = result[5]\n",
+    "    m_xx['nian'] = '2022'    \n",
+    "    mycol1.insert_one(m_xx)\n",
+    "    #print(m_xx)\n",
+    "print('ok')"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {
@@ -1161,6 +1532,51 @@
     "db.close()"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 导入2022年一分一段表"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "import openpyxl\n",
+    "import pymysql\n",
+    "import json\n",
+    "\n",
+    "m_xx = []\n",
+    "db = pymysql.connect(host = \"localhost\",user = \"root\",password = \"songyi\",database = \"gaokao\" )\n",
+    "cursor = db.cursor()\n",
+    "wb = openpyxl.load_workbook('./data/2022一分一段表.xlsx')\n",
+    "sheet = wb.active\n",
+    "i = 1\n",
+    "for n in range(4,sheet.max_row):\n",
+    "        m_score = sheet.cell(n,1).value\n",
+    "        m_num = sheet.cell(n,2).value\n",
+    "        m_sum = sheet.cell(n,3).value\n",
+    "        m_max = m_sum - m_num + 1\n",
+    "        m_xx.append((m_score,m_num,m_max,m_sum,'z','2022'))\n",
+    "#print(m_xx)\n",
+    "sql = 'insert into fenduan(score,per_num,max_rank,min_rank,category,nian) values (%s,%s,%s,%s,%s,%s)'\n",
+    "try:\n",
+    "    cursor.executemany(sql,m_xx)\n",
+    "    db.commit()\n",
+    "    print(\"ok!\")\n",
+    "except:\n",
+    "    # 如果发生错误则回滚\n",
+    "    print(\"error!\")\n",
+    "    db.rollback()   \n",
+    "\n",
+    "db.close()"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},