3.31 上传 dm rewrite

booway_kg_api/Untitled.ipynb

@@ -0,0 +1,142 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "e8f39ebb-71ab-4389-8bc3-29577470f948",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from WikijsTool import WikijsTool"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "id": "0a6bde6e-5507-48d9-8e64-d804f6085723",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "info = WikijsTool.get_all_documents()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 27,
+   "id": "009c3e8d-6ff6-4e0b-83b8-740ed195b5c5",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "html_text = WikijsTool.query_doc_info(8663)['content']"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 29,
+   "id": "621bb76a-aa5c-4f57-8574-c852f24b64e3",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import re\n",
+    "\n",
+    "cleaned_img_text = re.sub(r'<img\\s+[^>]*>', '', html_text)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 31,
+   "id": "941e1e3b-7b8e-47f1-96ff-fa89898a1dd7",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "'<h1>使用场景</h1>\\n<p>组合件或组合件下的消耗量想要修改所属项目划分</p>\\n<h1>功能入口</h1>\\n<p>【组合件】界面-——“组合件列表”页签-——选择已录入消耗量的组合件或组合件下的消耗量，鼠标右键-——选择”设置所属项目“。</p>\\n<figure class=\"image\">\\n  <figcaption>设置所属项目划分</figcaption>\\n</figure>\\n<h1>操作步骤</h1>\\n<p>1.设置所属项目划分</p>\\n<p>方法一：【组合件】界面——“组合件列表”页签-——选择已录入消耗量的组合件或组合件下的消耗量，鼠标右键——选择”设置所属项目划分“，在弹窗中选择项目划分，点击确定；</p>\\n<figure class=\"image\">\\n  <figcaption>批量设置</figcaption>\\n</figure>\\n<p>方法二：对于组合件下单条工程量，在此工程量的所属项目划分列双击，在弹窗中选择项目划分，点击确定；</p>\\n<figure class=\"image\">\\n  <figcaption>单条设置</figcaption>\\n</figure>\\n<h1>内部补充</h1>\\n<p>1.如工程量下方还有子级消耗量，需选择到父级消耗量进行操作，子级工程量的“设置所属项目”为灰色不可选。</p>\\n<figure class=\"image\">\\n  <figcaption>子级不可设置</figcaption>\\n</figure>\\n<p>&nbsp;</p>\\n<p>&nbsp;</p>\\n'"
+      ]
+     },
+     "execution_count": 31,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "cleaned_img_text"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 35,
+   "id": "13973a4a-1f19-4b4d-b3c8-441d69e0091b",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "使用场景\n",
+      "组合件或组合件下的消耗量想要修改所属项目划分\n",
+      "功能入口\n",
+      "【组合件】界面-——“组合件列表”页签-——选择已录入消耗量的组合件或组合件下的消耗量，鼠标右键-——选择”设置所属项目“。\n",
+      "\n",
+      "设置所属项目划分\n",
+      "\n",
+      "操作步骤\n",
+      "1.设置所属项目划分\n",
+      "方法一：【组合件】界面——“组合件列表”页签-——选择已录入消耗量的组合件或组合件下的消耗量，鼠标右键——选择”设置所属项目划分“，在弹窗中选择项目划分，点击确定；\n",
+      "\n",
+      "批量设置\n",
+      "\n",
+      "方法二：对于组合件下单条工程量，在此工程量的所属项目划分列双击，在弹窗中选择项目划分，点击确定；\n",
+      "\n",
+      "单条设置\n",
+      "\n",
+      "内部补充\n",
+      "1.如工程量下方还有子级消耗量，需选择到父级消耗量进行操作，子级工程量的“设置所属项目”为灰色不可选。\n",
+      "\n",
+      "子级不可设置\n",
+      "\n",
+      " \n",
+      " \n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "from bs4 import BeautifulSoup\n",
+    "\n",
+    "soup = BeautifulSoup(cleaned_img_text, \"html.parser\")\n",
+    "plain_text = soup.get_text()\n",
+    "print(plain_text)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "2802c4bd-34a7-4c61-bb15-4dd9739ecc1d",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "dify_lab",
+   "language": "python",
+   "name": "dify_lab"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.9"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}