因为Chrome 频繁的对 TT Rss的标记，甚烦！加之国庆无聊，准备在国庆假期用GPT 来手搓一个极简的自用工具。肝了一晚上，终于提前上线 简阅RSS。
总体上就是，参照TT Rss 做的，因为GPT对CSS的回答和预期差太多，所以花费时间也比预期的长，只能暂时这样，等后面有时间，再搞它。现在只能在PC的Web端可用，对我来已经够用。
前端
因为初衷就是本着，极简自用，所以就是单个页面,虽然GPT在CSS方面回答上，不容易对焦。但只是需要多试几次而已，我只是把TT Rss 的界面截图发给它，然后它就给我生成这个页面。然后在这个页面慢慢问，不停的问就行了。
<template>
<div class="app-layout">
<div class="main-content">
<!-- Sidebar (left) -->
<div class="resizable-sidebar" :style="{ width: sidebarWidth + 'px' }">
<h6 class="p-2">订阅</h6>
<ul class="list-group">
<li v-for="group in treeGroups" :key="group.id" class="list-group-item">
<strong @click="toggleGroup(group)" class="group-header">
<span>{{ group.isOpen ? '[-]' : '[+]' }}</span>
{{ group.name }}
</strong>
<ul v-show="group.isOpen">
<li v-for="node in getNodesForGroup(group.id)" :key="node.id" class="tree-node"
:class="{ active: selectedNode === node }"
@click="selectFeedNode(node)" >
<img :src="'data:image/png;base64,' + node.siteIcon" alt="icon" />
{{ node.siteName }}
<span class="node-count">{{ node.totalFeed }}</span>
</li>
</ul>
</li>
</ul>
</div>
<!-- Resizable divider (vertical) -->
<div class="divider-vertical" @mousedown="startResizingVertical"></div>
<div class="content-area" ref="contentArea">
<!-- Article List (顶部部分) -->
<div class="article-list p-3" :style="{ height: articleListHeight + 'px' }">
<h6>{{selectedNode?selectedNode.siteName :''}}-文章列表</h6>
<ul class="list-group">
<li
v-for="article in selectedArticles"
:key="article.id"
class="list-article-item"
@click="selectArticle(article)"
>
<input type="checkbox" class="checkbox" :checked="article.isSelected" @change="toggleSelectArticle(article)" />
<img src="@/assets/images/star.svg" class="icon bookmark-icon" alt="bookmark icon" />
<img src="@/assets/images/rss.svg" class="icon rss-icon" alt="rss icon" />
<span class="article-title">{{ article.title }}</span>
<span class="pub-date">{{ article.pubDate }}</span>
</li>
</ul>
</div>
<!-- Divider for resizing -->
<div class="divider-horizontal" @mousedown="startResizingHorizontal"></div>
<!-- Article Summary (底部部分) -->
<div class="article-summary p-3" :style="{ height: articleSummaryHeight + 'px', overflowY: 'auto' }">
<a v-if="itm" :href="itm.link || '#'" target="_blank">{{ itm?.title || "" }}</a>
<div v-html="itm?.excerpt || ' '"></div>
</div>
</div>
</div>
</div>
</template>
<script src="./AppLayout.js"></script>
<style src="@/assets/styles/AppLayout.css" scoped></style>
后端
后端虽然代码多，Feed多就得啰嗦，有的给Feed ，直接用 SyndicationFeed 就能解析，有的web直接一个request 就能拿到html，也是扔给GPT，就能出代码。稍麻烦的需要认证的就得跑 playwright把cookie装上，下次请求时自动更新它。 但GPT给力，基本上一问就是标准答案，所以花的时间反而最少的。
PW.cs
using Microsoft.Playwright;
namespace RssAdapter.Common
{
public class PW
{
private static IPlaywright _playwright;
private static IBrowser _browser;
private static IBrowserContext _context;
private static readonly object _lock = new object();
public static async Task<IBrowserContext> GetBrowserContextAsync()
{
if (_context == null)
{
lock (_lock)
{
if (_context == null)
{
_playwright = Playwright.CreateAsync().Result;
_browser = _playwright.Chromium.LaunchAsync(new BrowserTypeLaunchOptions { Headless = true }).Result;
_context = _browser.NewContextAsync( new BrowserNewContextOptions
{
UserAgent = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/127.0.0.0 Safari/537.36",
ViewportSize = new ViewportSize { Width = 1920, Height = 1080 },
//Locale = "en-US", // 模拟你本地的语言环境
//TimezoneId = "America/New_York", // 模拟时区
BypassCSP = true,
JavaScriptEnabled = true
}).Result;
}
}
}
return _context;
}
}
}
linuxdo.cs
public async Task<IActionResult> Getlatest()
{
try
{
string rssFeedUrl = "https://linux.do/latest.rss";
var uri = new Uri(rssFeedUrl);
string CookieFilePath = Path.Combine(AppContext.BaseDirectory, "cookies", uri.Host);
var context = await PW.GetBrowserContextAsync();
if (System.IO.File.Exists(CookieFilePath))
{
// 从本地加载 cookies 并添加到 context
var cookies = LoadCookiesFromFile(CookieFilePath);
if (cookies != null)
{
await context.AddCookiesAsync(cookies);
Console.WriteLine("Cookies loaded from file.");
}
}
var page = await context.NewPageAsync();
await page.GotoAsync(rssFeedUrl, new PageGotoOptions { Timeout = 60000 });
await page.WaitForLoadStateAsync(LoadState.NetworkIdle);
string content = await page.ContentAsync();
var updatedCookies = await context.CookiesAsync();
SaveCookiesToFile(updatedCookies, CookieFilePath);
string xmlContent = ExtractXmlFromHtml(content);
string schemeAndHost = $"{uri.Scheme}://{uri.Host}";
using (XmlReader reader = XmlReader.Create(new System.IO.StringReader(xmlContent)))
{
SyndicationFeed feed = SyndicationFeed.Load(reader);
feed.Links.First().Uri = new Uri(rssFeedUrl);
RSS2RSS rSS = new RSS2RSS(_configuration);
rSS.SaveFeedToDatabase(feed);
var responseData = rSS.ConvertFeedToString(feed);
await page.CloseAsync();
return Ok(responseData);
}
}
catch (Exception ex)
{
return BadRequest($"Failed to retrieve content. Status code: {ex.Message}");
}
}
后续
这是一个开始，RSS 聚合本身就是Feed内容一个载体，需要不断更新和积累，它就是为信息和数据的连接而产生的，“数据”是客观量化的结果，而”信息“ 有传递变化之意。
感谢TT Rss ，它给使用Rss 提供了方便。
感谢 ssh-news.top 的创意，它极大触动了我，让我有自建Web to Rss的动力，把它运行在IDE的终端里，没人分清楚你在上网还是调试代码。
最后，天佑合肥！