20020622

数据处理.ipynb

@@ -786,7 +786,6 @@
   {
    "cell_type": "markdown",
    "metadata": {
-    "jp-MarkdownHeadingCollapsed": true,
     "tags": []
    },
    "source": [
@@ -845,15 +844,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 20,
+   "execution_count": null,
    "metadata": {
-    "execution": {
-     "iopub.execute_input": "2021-12-17T01:17:00.335807Z",
-     "iopub.status.busy": "2021-12-17T01:17:00.334807Z",
-     "iopub.status.idle": "2021-12-17T01:17:00.469071Z",
-     "shell.execute_reply": "2021-12-17T01:17:00.469071Z",
-     "shell.execute_reply.started": "2021-12-17T01:17:00.335807Z"
-    },
     "tags": []
    },
    "outputs": [],
@@ -891,15 +883,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 22,
+   "execution_count": null,
    "metadata": {
-    "execution": {
-     "iopub.execute_input": "2021-12-17T01:40:10.503538Z",
-     "iopub.status.busy": "2021-12-17T01:40:10.503538Z",
-     "iopub.status.idle": "2021-12-17T01:40:10.673044Z",
-     "shell.execute_reply": "2021-12-17T01:40:10.672043Z",
-     "shell.execute_reply.started": "2021-12-17T01:40:10.503538Z"
-    },
     "tags": []
    },
    "outputs": [],
@@ -934,15 +919,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 25,
+   "execution_count": null,
    "metadata": {
-    "execution": {
-     "iopub.execute_input": "2021-12-17T02:50:20.560344Z",
-     "iopub.status.busy": "2021-12-17T02:50:20.560344Z",
-     "iopub.status.idle": "2021-12-17T02:50:21.226478Z",
-     "shell.execute_reply": "2021-12-17T02:50:21.226478Z",
-     "shell.execute_reply.started": "2021-12-17T02:50:20.560344Z"
-    },
     "tags": []
    },
    "outputs": [],
@@ -996,15 +974,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 27,
+   "execution_count": null,
    "metadata": {
-    "execution": {
-     "iopub.execute_input": "2021-12-17T03:08:55.050632Z",
-     "iopub.status.busy": "2021-12-17T03:08:55.050632Z",
-     "iopub.status.idle": "2021-12-17T03:08:55.843287Z",
-     "shell.execute_reply": "2021-12-17T03:08:55.843287Z",
-     "shell.execute_reply.started": "2021-12-17T03:08:55.050632Z"
-    },
     "tags": []
    },
    "outputs": [],
@@ -1050,15 +1021,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 29,
+   "execution_count": null,
    "metadata": {
-    "execution": {
-     "iopub.execute_input": "2021-12-17T03:18:41.399609Z",
-     "iopub.status.busy": "2021-12-17T03:18:41.398609Z",
-     "iopub.status.idle": "2021-12-17T03:18:41.467625Z",
-     "shell.execute_reply": "2021-12-17T03:18:41.466624Z",
-     "shell.execute_reply.started": "2021-12-17T03:18:41.399609Z"
-    },
     "tags": []
    },
    "outputs": [],
@@ -1166,6 +1130,35 @@
     "\n"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## excel数据读取"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "import openpyxl\n",
+    "import json\n",
+    "\n",
+    "filename = 'data/长岭全成绩.xlsx'\n",
+    "wb = openpyxl.load_workbook(filename)\n",
+    "sheet = wb.active\n",
+    "data1 =list(sheet.values)\n",
+    "del data1[0]\n",
+    "print(data1)\n",
+    "#for data in data1:\n",
+    "#    print(data)\n",
+    "\n"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
@@ -1186,6 +1179,13 @@
     "print(collist)"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
   {
    "cell_type": "code",
    "execution_count": null,

文件管理1.ipynb

@@ -64,6 +64,42 @@
     "#print(dict1)"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "bbf0b425-c0ce-4608-ae52-85a78ddcd6c0",
+   "metadata": {},
+   "source": [
+    "## 批量更换扩展名"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "2c2bd1d1-4d84-4639-ba20-c97c218ae275",
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2022-05-06T14:01:25.901198Z",
+     "iopub.status.busy": "2022-05-06T14:01:25.900687Z",
+     "iopub.status.idle": "2022-05-06T14:01:25.911729Z",
+     "shell.execute_reply": "2022-05-06T14:01:25.910606Z",
+     "shell.execute_reply.started": "2022-05-06T14:01:25.901153Z"
+    },
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "import glob\n",
+    "import os,shutil\n",
+    "\n",
+    "filepath = 'file/yimo/'\n",
+    "old = 'SGF'\n",
+    "new = 'sgf'\n",
+    "files = glob.glob(f'{filepath}*.{old}')\n",
+    "for f in files:\n",
+    "    new_file =os.path.splitext(f)[0]+'.'+new\n",
+    "    os.rename(f,new_file)"
+   ]
+  },
   {
    "cell_type": "markdown",
    "id": "69712868-6786-430b-8591-f6828d781588",

高考志愿查询.ipynb

@@ -18,59 +18,10 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 80,
+   "execution_count": null,
    "id": "be2371be-89f5-4879-82a9-4a1d9e6376d3",
    "metadata": {},
-   "outputs": [
+   "outputs": [],
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "1 南京中医药大学 中医学（本硕连读5+3一体化） 11019\n",
-      "2 天津医科大学 预防医学 11108\n",
-      "3 哈尔滨医科大学 临床医学 11227\n",
-      "4 空军军医大学 基础医学 11232\n",
-      "5 南京医科大学 预防医学 11275\n",
-      "6 温州医科大学 眼视光医学（5+3一体化） 11564\n",
-      "7 上海中医药大学 中医学（5+3一体化针灸推拿英语方向） 11620\n",
-      "8 兰州大学 临床医学类 11662\n",
-      "9 天津中医药大学 中医学（5+3一体化） 11825\n",
-      "10 哈尔滨医科大学 临床医学（5+3一体化，儿科学硕士） 11837\n",
-      "11 温州医科大学 临床医学（5+3一体化） 11869\n",
-      "12 东北大学 智能医学工程 11916\n",
-      "13 海军军医大学 中医学（中医临床医师） 11955\n",
-      "14 苏州大学 预防医学 11991\n",
-      "15 暨南大学 临床医学 12045\n",
-      "16 中国医科大学 医学影像学 12059\n",
-      "17 南昌大学 临床医学 12252\n",
-      "18 天津医科大学 医学影像技术 12272\n",
-      "19 吉林大学 预防医学 12320\n",
-      "20 南京航空航天大学 生物医学工程 12504\n",
-      "21 江南大学 临床医学 12526\n",
-      "22 广州中医药大学 中医学（5+3一体化） 12715\n",
-      "23 中国医科大学 基础医学 12755\n",
-      "24 重庆医科大学 医学影像学 12784\n",
-      "25 大连医科大学 临床医学（5+3一体化） 13012\n",
-      "26 天津医科大学 智能医学工程 13103\n",
-      "27 南京中医药大学 中医学 13301\n",
-      "28 温州医科大学 眼视光医学 13370\n",
-      "29 天津医科大学 医学检验技术 13371\n",
-      "30 天津医科大学 生物医学工程 13396\n",
-      "31 温州医科大学 临床医学 13421\n",
-      "32 中国医科大学 预防医学 13769\n",
-      "33 汕头大学 临床医学（5+3一体化） 13808\n",
-      "34 郑州大学 临床医学类 13981\n",
-      "35 郑州大学 口腔医学 14147\n",
-      "36 南方医科大学 基础医学 14160\n",
-      "37 哈尔滨医科大学 基础医学 14331\n",
-      "38 南方医科大学 预防医学 14366\n",
-      "39 大连医科大学 临床医学 14423\n",
-      "40 西北大学 临床医学 14492\n",
-      "41 天津中医药大学 中医学（5+3一体化中医儿科学） 14708\n",
-      "42 南京医科大学 智能医学工程 14721\n"
-     ]
-    }
-   ],
    "source": [
     "import pymongo\n",
     "import re\n",

高考志愿管理.ipynb

@@ -3,8 +3,7 @@
   {
    "cell_type": "markdown",
    "metadata": {
-    "tags": [],
+    "tags": []
-    "toc-hr-collapsed": true
    },
    "source": [
     "# 高考志愿管理"
@@ -80,6 +79,161 @@
     "\n"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## 2021年全国高等学校信息更新"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 2021年全国高等学校信息导出至json文件"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2022-06-15T07:00:27.890983Z",
+     "iopub.status.busy": "2022-06-15T07:00:27.890444Z",
+     "iopub.status.idle": "2022-06-15T07:00:28.430955Z",
+     "shell.execute_reply": "2022-06-15T07:00:28.429949Z",
+     "shell.execute_reply.started": "2022-06-15T07:00:27.890936Z"
+    },
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "import openpyxl\n",
+    "import json\n",
+    "\n",
+    "filename = 'data/2021年全国普通高等学校名单.xlsx'\n",
+    "wb = openpyxl.load_workbook(filename)\n",
+    "sheet = wb.active\n",
+    "data1 =list(sheet.values)\n",
+    "del data1[0:3]\n",
+    "i = 1\n",
+    "dict1 = {}\n",
+    "for items in data1:\n",
+    "    list1 = (items[1],str(items[2]),items[3],items[4],items[5],items[6])\n",
+    "    dict1[i] = list1\n",
+    "    i += 1 \n",
+    "with open('data/2021年全国普通高等学校名单.json','w') as fl2:\n",
+    "    json.dump(dict1,fl2) \n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 整理2021年全国高等学校信息"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 25,
+   "metadata": {
+    "execution": {
+     "iopub.execute_input": "2022-06-15T08:26:50.903511Z",
+     "iopub.status.busy": "2022-06-15T08:26:50.902972Z",
+     "iopub.status.idle": "2022-06-15T08:26:50.972239Z",
+     "shell.execute_reply": "2022-06-15T08:26:50.971224Z",
+     "shell.execute_reply.started": "2022-06-15T08:26:50.903463Z"
+    },
+    "tags": []
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "河北工业职业技术大学导入成功！\n",
+      "河北科技工程职业技术大学导入成功！\n",
+      "河北石油职业技术大学导入成功！\n",
+      "邢台应用技术职业学院导入成功！\n",
+      "山西工程科技职业大学导入成功！\n",
+      "吉林通用航空职业技术学院导入成功！\n",
+      "通化医药健康职业学院导入成功！\n",
+      "上海南湖职业技术学院导入成功！\n",
+      "浙江药科职业大学导入成功！\n",
+      "浙江金华科贸职业技术学院导入成功！\n",
+      "宿州航空职业学院导入成功！\n",
+      "和君职业学院导入成功！\n",
+      "滨州科技职业学院导入成功！\n",
+      "洛阳文化旅游职业学院导入成功！\n",
+      "周口文理职业学院导入成功！\n",
+      "信阳艺术职业学院导入成功！\n",
+      "郑州城建职业学院导入成功！\n",
+      "郑州医药健康职业学院导入成功！\n",
+      "湖北孝感美珈职业学院导入成功！\n",
+      "广州幼儿师范高等专科学校导入成功！\n",
+      "广东汕头幼儿师范高等专科学校导入成功！\n",
+      "广东梅州职业技术学院导入成功！\n",
+      "广东潮州卫生健康职业学院导入成功！\n",
+      "广东云浮中医药职业学院导入成功！\n",
+      "广东肇庆航空职业学院导入成功！\n",
+      "广西农业职业技术大学导入成功！\n",
+      "防城港职业技术学院导入成功！\n",
+      "广西信息职业技术学院导入成功！\n",
+      "广西农业工程职业技术学院导入成功！\n",
+      "北海康养职业学院导入成功！\n",
+      "重庆工信职业学院导入成功！\n",
+      "甘孜职业学院导入成功！\n",
+      "自贡职业技术学院导入成功！\n",
+      "贵阳康养职业大学导入成功！\n",
+      "贵州文化旅游职业学院导入成功！\n",
+      "宝鸡中北职业学院导入成功！\n",
+      "兰州石化职业技术大学导入成功！\n",
+      "兰州资源环境职业技术大学导入成功！\n",
+      "兰州航空职业技术学院导入成功！\n",
+      "白银希望职业技术学院导入成功！\n"
+     ]
+    }
+   ],
+   "source": [
+    "import openpyxl\n",
+    "import json\n",
+    "import pymongo\n",
+    "\n",
+    "\n",
+    "filename = 'data/2021年全国普通高等学校名单.json'\n",
+    "with open(filename,'r') as fl:\n",
+    "    m_xx = json.load(fl)\n",
+    "\n",
+    "myclient = pymongo.MongoClient('mongodb://localhost:27017/')\n",
+    "mydb = myclient[\"gaokao\"]\n",
+    "mycol = mydb[\"college_2021\"]\n",
+    "\n",
+    "m_col = {}\n",
+    "\n",
+    "\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",
+    "#print(m_col.keys())\n",
+    "for k, v in m_xx.items():\n",
+    "    m_code = v[1]\n",
+    "    #print(m_code)\n",
+    "    if m_code not in m_col.keys():\n",
+    "        dict1 = {}\n",
+    "        dict1['id_code'] = v[1]\n",
+    "        dict1['name'] = v[0]\n",
+    "        dict1['charge'] = v[2]\n",
+    "        dict1['city'] = v[3]\n",
+    "        dict1['grade'] = v[4]\n",
+    "        dict1['note'] = v[5]\n",
+    "        mycol.insert_one(dict1) \n",
+    "        print(dict1['name'] + '导入成功！')\n",
+    "        #print(v[0],v[1])\n",
+    "        \n",
+    "\n",
+    "\n",
+    "        "
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -602,7 +756,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 9,
+   "execution_count": null,
    "metadata": {},
    "outputs": [],
    "source": [