From 4d957f3ce93f48ac41f93fbe3b5a398d445e9cf9 Mon Sep 17 00:00:00 2001 From: 512song <512song@sina.com> Date: Fri, 17 Dec 2021 11:29:19 +0800 Subject: [PATCH] 20211217 --- 数据处理.ipynb | 437 +++++++++++++++++++++++++++++++++++++++++++------ 1 file changed, 383 insertions(+), 54 deletions(-) diff --git a/数据处理.ipynb b/数据处理.ipynb index e955e04..02caba7 100644 --- a/数据处理.ipynb +++ b/数据处理.ipynb @@ -46,9 +46,33 @@ "#print(list1)" ] }, + { + "cell_type": "code", + "execution_count": null, + "metadata": { + "tags": [] + }, + "outputs": [], + "source": [ + "import re\n", + "\n", + "s = '消化不良、胃下垂、急性胃炎、慢性胃炎、萎缩性胃炎、神经性呕吐、胆囊炎、胆石症、胆道蛔虫症、胸胁痛等'\n", + "ss = '健中和胃，消食止呕，理气疏郁，清热利胆。'\n", + "mo = '等$'\n", + "mo2 = r'[，、；。]'\n", + "s1 = re.sub(mo, '', s) \n", + "list1 = re.split(mo2,s1)\n", + "if '' in list1:\n", + " list1.remove('')\n", + "print(list1)" + ] + }, { "cell_type": "markdown", - "metadata": {}, + "metadata": { + "jp-MarkdownHeadingCollapsed": true, + "tags": [] + }, "source": [ "## 药膳归经明细文件生成" ] @@ -91,7 +115,10 @@ }, { "cell_type": "markdown", - "metadata": {}, + "metadata": { + "jp-MarkdownHeadingCollapsed": true, + "tags": [] + }, "source": [ "## 药膳归经权重生成" ] @@ -169,14 +196,19 @@ }, { "cell_type": "markdown", - "metadata": {}, + "metadata": { + "tags": [], + "toc-hr-collapsed": true + }, "source": [ "## 穴位数据导入" ] }, { "cell_type": "markdown", - "metadata": {}, + "metadata": { + "tags": [] + }, "source": [ "### 简单导出简介、内容" ] @@ -212,7 +244,10 @@ }, { "cell_type": "markdown", - "metadata": {}, + "metadata": { + "jp-MarkdownHeadingCollapsed": true, + "tags": [] + }, "source": [ "### 数据导入文件中" ] @@ -253,6 +288,7 @@ { "cell_type": "markdown", "metadata": { + "jp-MarkdownHeadingCollapsed": true, "tags": [] }, "source": [ @@ -271,7 +307,7 @@ "import re\n", "import pymongo\n", "\n", - "myclient = pymongo.MongoClient('mongodb://192.168.3.100:27017/')\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", "#dblist = myclient.list_database_names()\n", "mydb = myclient['dayi']\n", "mycol = mydb[\"xuewei\"]\n", @@ -322,7 +358,7 @@ "import re\n", "import pymongo\n", "\n", - "myclient = pymongo.MongoClient('mongodb://192.168.3.100:27017/')\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", "mydb = myclient['dayi']\n", "mycol = mydb[\"zhongyizhengzhuang\"]\n", "db_list = []\n", @@ -371,7 +407,7 @@ "import re\n", "import pymongo\n", "\n", - "myclient = pymongo.MongoClient('mongodb://192.168.3.100:27017/')\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", "mydb = myclient['dayi']\n", "mycol = mydb[\"jibing\"]\n", "db_list = []\n", @@ -418,7 +454,7 @@ "import re\n", "import pymongo\n", "\n", - "myclient = pymongo.MongoClient('mongodb://192.168.3.100:27017/')\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", "#dblist = myclient.list_database_names()\n", "mydb = myclient['dayi']\n", "mycol = mydb[\"shuyu\"]\n", @@ -476,7 +512,7 @@ "import re\n", "import pymongo\n", "\n", - "myclient = pymongo.MongoClient('mongodb://192.168.3.100:27017/')\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", "mydb = myclient['dayi']\n", "mycol = mydb[\"xiyizhengzhuang\"]\n", "db_list = []\n", @@ -523,7 +559,7 @@ "import re\n", "import pymongo\n", "\n", - "myclient = pymongo.MongoClient('mongodb://192.168.3.100:27017/')\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", "#dblist = myclient.list_database_names()\n", "mydb = myclient['dayi']\n", "mycol = mydb[\"yaoji\"]\n", @@ -575,7 +611,7 @@ "import re\n", "import pymongo\n", "\n", - "myclient = pymongo.MongoClient('mongodb://192.168.3.100:27017/')\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", "#dblist = myclient.list_database_names()\n", "mydb = myclient['dayi']\n", "mycol = mydb[\"yaoshan\"]\n", @@ -624,7 +660,7 @@ "import re\n", "import pymongo\n", "\n", - "myclient = pymongo.MongoClient('mongodb://192.168.3.100:27017/')\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", "mydb = myclient['dayi']\n", "mycol = mydb[\"zhongcaoyao\"]\n", "db_list = []\n", @@ -653,59 +689,27 @@ }, { "cell_type": "markdown", - "metadata": {}, + "metadata": { + "jp-MarkdownHeadingCollapsed": true, + "tags": [] + }, "source": [ "## 穴位隶属整理" ] }, { "cell_type": "code", - "execution_count": 21, + "execution_count": null, "metadata": { - "execution": { - "iopub.execute_input": "2021-12-16T00:57:08.586442Z", - "iopub.status.busy": "2021-12-16T00:57:08.585443Z", - "iopub.status.idle": "2021-12-16T00:57:08.662953Z", - "shell.execute_reply": "2021-12-16T00:57:08.662953Z", - "shell.execute_reply.started": "2021-12-16T00:57:08.586442Z" - }, "tags": [] }, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "耳穴\n", - "手针穴\n", - "经外奇穴\n", - "任脉\n", - "足太阳膀胱经\n", - "足少阳胆经\n", - "手少阳三焦经\n", - "手太阳小肠经\n", - "足阳明胃经\n", - "手太阴肺经\n", - "手阳明大肠经\n", - "足太阴脾经\n", - "手少阴心经\n", - "足少阴肾经\n", - "手厥阴心包经\n", - "督脉\n", - "足厥阴肝经\n", - "推拿穴位\n", - "头针刺激区\n", - "组合穴\n", - "手穴\n" - ] - } - ], + "outputs": [], "source": [ "import json\n", "import re\n", "import pymongo\n", "\n", - "myclient = pymongo.MongoClient('mongodb://192.168.3.100:27017/')\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", "mydb = myclient['dayi']\n", "mycol = mydb[\"xuewei\"]\n", "dict1 = {}\n", @@ -724,6 +728,61 @@ " print(item)" ] }, + { + "cell_type": "markdown", + "metadata": { + "tags": [] + }, + "source": [ + "## 穴位功能、主治统计" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": { + "tags": [] + }, + "outputs": [], + "source": [ + "import json\n", + "import re\n", + "import pymongo\n", + "\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", + "mydb = myclient['dayi']\n", + "mycol = mydb[\"xuewei\"]\n", + "\n", + "dict1 = {}\n", + "mo = '等$'\n", + "mo2 = r'[，、；。]'\n", + "\n", + "for x in mycol.find({},{ \"_id\": 0,\"name\":1,\"about\":1 }):\n", + " dict1.setdefault(x['name'],{})\n", + " if '主治' in x['about'].keys():\n", + " #dict1[x['name']].setdefault('主治',[])\n", + " s = x['about']['主治']\n", + " s1 = re.sub(mo, '', s)\n", + " list1 = re.split(mo2,s1)\n", + " if '' in list1:\n", + " list1.remove('')\n", + " dict1[x['name']]['主治'] = list1\n", + " if '功能' in x['about'].keys():\n", + " #dict1[x['name']].setdefault('主治',[])\n", + " s = x['about']['功能']\n", + " s1 = re.sub(mo, '', s)\n", + " list1 = re.split(mo2,s1)\n", + " if '' in list1:\n", + " list1.remove('')\n", + " dict1[x['name']]['功能'] = list1\n", + " \n", + "\n", + "filename = './file/穴位主治功能统计.json'\n", + "with open(filename,'w') as fl:\n", + " json.dump(dict1, fl) \n", + "\n" + ] + }, { "cell_type": "markdown", "metadata": { @@ -774,9 +833,279 @@ "\n" ] }, + { + "cell_type": "markdown", + "metadata": { + "jp-MarkdownHeadingCollapsed": true, + "tags": [] + }, + "source": [ + "## 术语数据处理" + ] + }, + { + "cell_type": "code", + "execution_count": 20, + "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": [], + "source": [ + "import json\n", + "import re\n", + "import pymongo\n", + "\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", + "mydb = myclient['dayi']\n", + "mycol = mydb[\"shuyu\"]\n", + "dict1 = {}\n", + "list1 = []\n", + "for x in mycol.find({},{ \"_id\": 0,\"name\":1,\"about\":1 }):\n", + " if '类别' in x['about'].keys():\n", + " m_lb = x['about']['类别'].replace(' ','')\n", + " else:\n", + " m_lb = '无类别'\n", + " dict1.setdefault(re.sub('\\xa0+','',m_lb),[])\n", + " dict1[m_lb].append(x['name'])\n", + "filename = './file/术语类别.json'\n", + "with open(filename,'w') as fl:\n", + " json.dump(dict1, fl) \n" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "jp-MarkdownHeadingCollapsed": true, + "tags": [] + }, + "source": [ + "## 疾病数据处理" + ] + }, + { + "cell_type": "code", + "execution_count": 22, + "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": [], + "source": [ + "import json\n", + "import re\n", + "import pymongo\n", + "\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", + "mydb = myclient['dayi']\n", + "mycol = mydb[\"jibing\"]\n", + "dict1 = {}\n", + "list1 = []\n", + "for x in mycol.find({},{ \"_id\": 0,\"name\":1,\"about\":1 }):\n", + " if '疾病分类' in x['about'].keys():\n", + " m_lb = x['about']['疾病分类'].replace(' ','')\n", + " else:\n", + " m_lb = '无类别'\n", + " dict1.setdefault(re.sub('\\xa0+','',m_lb),[])\n", + " dict1[m_lb].append(x['name'])\n", + "filename = './file/疾病类别.json'\n", + "with open(filename,'w') as fl:\n", + " json.dump(dict1, fl) " + ] + }, { "cell_type": "markdown", "metadata": {}, + "source": [ + "## 中草药数据处理" + ] + }, + { + "cell_type": "code", + "execution_count": 25, + "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": [], + "source": [ + "import json\n", + "import re\n", + "import pymongo\n", + "\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", + "mydb = myclient['dayi']\n", + "mycol = mydb[\"zhongcaoyao\"]\n", + "\n", + "dict1 = {}\n", + "mo = '等$'\n", + "mo2 = r'[，、；。]'\n", + "\n", + "for x in mycol.find({},{ \"_id\": 0,\"name\":1,\"about\":1 }):\n", + " dict1.setdefault(x['name'],{})\n", + " if '别名' in x['about'].keys():\n", + " #dict1[x['name']].setdefault('主治',[])\n", + " s = x['about']['别名']\n", + " s1 = re.sub(mo, '', s)\n", + " list1 = re.split(mo2,s1)\n", + " if '' in list1:\n", + " list1.remove('')\n", + " dict1[x['name']]['别名'] = list1\n", + " if '功能' in x['about'].keys():\n", + " #dict1[x['name']].setdefault('主治',[])\n", + " s = x['about']['功能']\n", + " s1 = re.sub(mo, '', s)\n", + " list1 = re.split(mo2,s1)\n", + " if '' in list1:\n", + " list1.remove('')\n", + " dict1[x['name']]['功能'] = list1\n", + " if '主治' in x['about'].keys():\n", + " #dict1[x['name']].setdefault('主治',[])\n", + " dict1[x['name']]['主治'] = s = x['about']['主治']\n", + " \n", + "\n", + "filename = './file/中草药主治功能统计.json'\n", + "with open(filename,'w') as fl:\n", + " json.dump(dict1, fl) " + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## 药剂数据处理" + ] + }, + { + "cell_type": "code", + "execution_count": 27, + "metadata": { + "execution": { + "iopub.execute_input": "2021-12-17T03:08:55.050632Z", + "iopub.status.busy": "2021-12-17T03:08:55.050632Z", + "iopub.status.idle": "2021-12-17T03:08:55.843287Z", + "shell.execute_reply": "2021-12-17T03:08:55.843287Z", + "shell.execute_reply.started": "2021-12-17T03:08:55.050632Z" + }, + "tags": [] + }, + "outputs": [], + "source": [ + "import json\n", + "import re\n", + "import pymongo\n", + "\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", + "mydb = myclient['dayi']\n", + "mycol = mydb[\"yaoji\"]\n", + "\n", + "dict1 = {}\n", + "mo = '等$'\n", + "mo2 = r'[，、；。]'\n", + "\n", + "for x in mycol.find({},{ \"_id\": 0,\"name\":1,\"about\":1 }):\n", + " dict1.setdefault(x['name'],{})\n", + " if '功用' in x['about'].keys():\n", + " #dict1[x['name']].setdefault('主治',[])\n", + " s = x['about']['功用']\n", + " s1 = re.sub(mo, '', s)\n", + " list1 = re.split(mo2,s1)\n", + " if '' in list1:\n", + " list1.remove('')\n", + " dict1[x['name']]['功用'] = list1\n", + " if '主治' in x['about'].keys():\n", + " #dict1[x['name']].setdefault('主治',[])\n", + " dict1[x['name']]['主治'] = s = x['about']['主治']\n", + " \n", + "\n", + "filename = './file/药剂主治功用统计.json'\n", + "with open(filename,'w') as fl:\n", + " json.dump(dict1, fl) " + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## 药膳数据处理" + ] + }, + { + "cell_type": "code", + "execution_count": 29, + "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": [], + "source": [ + "import json\n", + "import re\n", + "import pymongo\n", + "\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", + "mydb = myclient['dayi']\n", + "mycol = mydb[\"yaoshan\"]\n", + "\n", + "dict1 = {}\n", + "mo = '等$'\n", + "mo2 = r'[，、；。]'\n", + "\n", + "for x in mycol.find({},{ \"_id\": 0,\"name\":1,\"about\":1 }):\n", + " dict1.setdefault(x['name'],{})\n", + " if '功效' in x['about'].keys():\n", + " #dict1[x['name']].setdefault('主治',[])\n", + " s = x['about']['功效']\n", + " s1 = re.sub(mo, '', s)\n", + " list1 = re.split(mo2,s1)\n", + " if '' in list1:\n", + " list1.remove('')\n", + " dict1[x['name']]['功效'] = list1\n", + " if '相关疾病' in x['about'].keys():\n", + " #dict1[x['name']].setdefault('主治',[])\n", + " s = x['about']['相关疾病']\n", + " s1 = re.sub(mo, '', s)\n", + " list1 = re.split(mo2,s1)\n", + " if '' in list1:\n", + " list1.remove('')\n", + " dict1[x['name']]['相关疾病'] = list1 \n", + " \n", + "\n", + "filename = './file/药膳功能统计.json'\n", + "with open(filename,'w') as fl:\n", + " json.dump(dict1, fl) " + ] + }, + { + "cell_type": "markdown", + "metadata": { + "jp-MarkdownHeadingCollapsed": true, + "tags": [] + }, "source": [ "## 北海炼化体检数据提取" ] @@ -869,9 +1198,9 @@ "import json\n", "import pymongo\n", "\n", - "myclient = pymongo.MongoClient('mongodb://192.168.3.100:27017/')\n", + "myclient = pymongo.MongoClient('mongodb://10.147.17.6:27017/')\n", "#dblist = myclient.list_database_names()\n", - "mydb = myclient['tice']\n", + "mydb = myclient['dayi']\n", " \n", "collist = mydb. list_collection_names()\n", "print(collist)"