ICDYCT / I Can Do, You Can Too

In the previous episode, we implemented the basic framework of an anti-spam bot What if we want to implement a rich way of generating question-answer pairs?

上一期我们实现了防广告机器人的基本框架 如果我们想实现丰富的 问题-答案 生成方式, 怎么做呢? 看一下目前的代码. 是生成了 问题-答案 后, 保存起来, 然后新成员再回答问题, 并检查答案.

Having used Telegram for so long, the anti-spam bots I've seen all use button selection to answer questions, which in theory could be vulnerable to brute-forcing the entire question bank. If your group is attractive enough and there are enough spam bots wanting to join, they can always find the answer to the join question. And if the join question is too extreme ( https://github.com/XTLS/Xray-core/discussions/3487 ) it instead ends up rejecting some less capable genuine humans (they have no intention of spamming; they just want to join to ask questions or chat) So I came up with an idea: what if the bot-check question requires a text answer?

用了这么久的telegram, 看到的防广告机器人, 都是用选按钮的方式回答问题, 理论上有被穷举题库的可能性. 如果你的群足够有吸引力, 想加群的广告机器人足够多, 总是可以找到入群问题的答案的. 而入群问题如果太变态 ( https://github.com/XTLS/Xray-core/discussions/3487 ) 反而拒绝了一些能力不强的正常人类 (TA们本身并不是想发广告, 只是想入群提问或交流) 所以我有了一个思路, 如果这个人机检查的问题需要回答文本呢?

Continuing from the previous post Today I'll continue talking about some thoughts on commercializing the Telegram keyword monitor bot. That is, how you can use the project to make money. Github https://github.com/crazypeace/tg-keyword-monitor-bot Let's review the architecture of this project again

书接上文 今天接着说商业化电报关键词提醒机器人 telegram keyword monitor bot 的一些思考. 也就是, 你可以怎样用项目去卖钱. Github https://github.com/crazypeace/tg-keyword-monitor-bot 再复习一下本项目的架构

Follow the official Telegram project guide https://github.com/telegramdesktop/tdesktop/blob/dev/docs/building-win-x64.md Apply for an api_id, api_hash https://my.telegram.org/apps

参考 telegram 官方项目的指南 https://github.com/telegramdesktop/tdesktop/blob/dev/docs/building-win-x64.md 申请 api_id, api_hash https://my.telegram.org/apps

Recently, I developed a Telegram keyword alert bot uploaded to Github https://github.com/crazypeace/tg-keyword-monitor-bot This article describes the setup process. First, you can take a look at the readme of the Github project, which is a step-by-step illustrated tutorial in operating order from start to finish. If you find that tutorial a bit confusing, then below, I'll describe it from a different angle. System Architecture Information Source -> Collector -> This Program -> Publisher -> Consumer

前几天 开发了一个电报关键词提醒机器人 已上传Github https://github.com/crazypeace/tg-keyword-monitor-bot 本文描述搭建过程. 首先, 你可以看看Github项目的readme, 那是一个从头到尾按操作顺序手把手的图文教程. 如果你觉得那篇教程有点云里雾里, 那么下面, 我换一个角度来描述. 系统架构 信息源 -> 采集者 -> 本程序 -> 发布者 -> 消费者

Preface I previously used a Telegram keyword alert bot project . After using it, I felt the functionality was a bit heavy. The original project was designed for multi-user use, where different users follow different keywords and different groups. Therefore, the original project required a database. I plan to develop a lightweight Telegram keyword alert bot suitable for personal use. I'll write the code with the help of the powerful GPT. The first version of the code was generated using the free Claude. Subsequent code debugging was done using the free M$ Copilot. Designed Features The bot only accepts control commands from users or groups with specified IDs,  After the bot detects a keyword, it sends a notification message to the user, group, or channel with the specified ID, The list of followed keywords supports regular expressions The list of excluded keywords supports regular expressions Both the list of followed keywords and the list of excluded keywords are updated and saved ...

前言 之前使用了一个 电报关键词提醒机器人项目 . 使用下来觉得功能有点重. 原项目考虑到多人使用, 不同的用户关注不同的关键字和不同的群组. 所以原项目需要用到数据库. 我准备自己开发一个轻量的适合个人使用的电报关键词提醒机器人. 借助现在强大的gpt的力量写代码. 第一版代码生成是用的免费的 claude 后续代码调试是用的免费的 M$ copilot

In the previous post I mentioned discovering that the code blocks in my blog posts may contain NBSP issues. Today I happened to discover that clicking the copy button on a code block yields a result that contains NBSP. For example, the following. Selecting with the mouse and copying from the right-click menu gives the result on the left. Clicking the copy button on the code block gives the result on the right. (Contains some NBSP characters) I took my code and the problem description to ask Claude. Solution 1: Change innerText to textContent Implementation https://crazypeace.github.io/blogger-replace-blockquote-to-pre/blockquote-add-copy-button1.js The test result was wrong. Not only did it still contain NBSP, but the line breaks were also gone. I took the above problem and continued asking Claude

上一篇讲到我发现了我的博文的代码段有可能含有NBSP的问题. 今天偶然发现, 复制代码段的按钮, 得到的结果会含有NBSP. 比如, 下面的例子. 用鼠标选择, 右键菜单复制, 得到的结果是左边的. 点击代码段的复制按钮, 得到的结果是右边的. (包含一些NBSP字符) 我拿我的代码和问题描述去问 claude. 方案1 将innerText改为textContent 具体实施 https://crazypeace.github.io/blogger-replace-blockquote-to-pre/blockquote-add-copy-button1.js 测试结果不对. 不仅继续含有NBSP, 而且换行也没有了. 拿上面的问题继续问 claude