refactor(handlers): 拆分RSS工作流为独立的生成与存储阶段
引入两阶段RSS处理架构以支持订阅源摘要预览功能： 第一阶段 - 内容生成(/generateRssContent): * 读取daily目录原始markdown * 调用AI生成精简摘要 * 持久化至GitHub rss目录 第二阶段 - 数据同步(/writeRssData): * 消费rss目录预生成内容 * 转换为HTML并同步至KV存储 其他调整: * 创建官网引流模板(appUrl.js) * 实现智能断行的文本裁剪工具 * 优化日报页面广告展示顺序 * 修正getDailyReportContent异常日志输出
This commit is contained in:
luofeng
27
src/appUrl.js
Normal file
27
src/appUrl.js
Normal file
@@ -0,0 +1,27 @@
|
||||
export function getAppUrl() {
|
||||
return `
|
||||
|
||||
---
|
||||
|
||||
**📢 关于 AI日报 的一次小调整**
|
||||
>
|
||||
坦白说，想要长久地把**AI日报**做下去，单靠“为爱发电”确实面临现实压力。为了更有热情的**投入精力**，我在网站接入了少量 Google 广告。
|
||||
>
|
||||
由于 RSS 无法展示广告带来收入，即日起 RSS 将**试运行“摘要模式”一段时间**。
|
||||
>
|
||||
**💡 您的每一次点击，都是对我最大的支持**
|
||||
诚挚邀请您移步官网阅读全文。那里不仅有更舒适的**排版**和清晰的**代码高亮**，还能在评论区与大家交流。
|
||||
>
|
||||
感谢您的理解与陪伴，让我们一起走得更远！
|
||||
>
|
||||
👇 **点击下方链接，阅读今日完整资讯**
|
||||
### [🚀 前往官网查看完整版 (ai.hubtoday.app)](https://ai.hubtoday.app/)
|
||||
>
|
||||
<small>如有建议，欢迎随时邮件沟通：[justlikemaki@foxmail.com](mailto:justlikemaki@foxmail.com)</small>
|
||||
<br/>
|
||||
<small>或直接扫码进群提供建议:</small>
|
||||
<br/>
|
||||
|
||||
|
||||
`;
|
||||
}
|
||||
2
src/github.js
Normal file → Executable file
2
src/github.js
Normal file → Executable file
@@ -106,7 +106,7 @@ export async function getDailyReportContent(env, filePath) {
|
||||
const data = await callGitHubApi(env, `/contents/${filePath}?ref=${GITHUB_BRANCH}`);
|
||||
return b64DecodeUnicode(data.content);
|
||||
} catch (error) {
|
||||
console.error(`Error fetching daily report content from ${rawUrl}:`, error);
|
||||
console.error(`Error fetching daily report content from ${filePath}:`, error);
|
||||
throw error;
|
||||
}
|
||||
}
|
||||
|
||||
34
src/handlers/genAIContent.js
Normal file → Executable file
34
src/handlers/genAIContent.js
Normal file → Executable file
@@ -77,7 +77,7 @@ export async function handleGenAIPodcastScript(request, env) {
|
||||
let promptsMarkdownContent = `# Prompts for ${dateStr}\n\n`;
|
||||
promptsMarkdownContent += `## Call 3: Podcast Formatting\n\n`;
|
||||
if (fullPromptForCall3_System) promptsMarkdownContent += `### System One Instruction\n\`\`\`\n${fullPromptForCall3_System}\n\`\`\`\n\n`;
|
||||
|
||||
|
||||
|
||||
let fullPromptForCall4_System = getSystemPromptShortPodcastFormatting(env);
|
||||
console.log("Call 4 to Chat (Podcast Formatting): User prompt length:", userPromptPodcastFormattingData.length);
|
||||
@@ -105,13 +105,13 @@ export async function handleGenAIPodcastScript(request, env) {
|
||||
let podcastScriptMarkdownContent = `# ${env.PODCAST_TITLE} ${formatDateToChinese(dateStr)}\n\n${removeMarkdownCodeBlock(finalAiResponseOut)}`;
|
||||
|
||||
const successHtml = generateGenAiPageHtml(
|
||||
env,
|
||||
env,
|
||||
'AI播客脚本',
|
||||
escapeHtml(finalAiResponseOut),
|
||||
escapeHtml(finalAiResponseOut),
|
||||
dateStr, false, selectedItemsParams,
|
||||
null, null, // No Call 1 prompts for this page
|
||||
fullPromptForCallSystem, fullPromptForCall3_User,
|
||||
convertEnglishQuotesToChinese(removeMarkdownCodeBlock(promptsMarkdownContent)),
|
||||
convertEnglishQuotesToChinese(removeMarkdownCodeBlock(promptsMarkdownContent)),
|
||||
outputOfCall1, // No daily summary for this page
|
||||
convertEnglishQuotesToChinese(podcastScriptMarkdownContent)
|
||||
);
|
||||
@@ -119,7 +119,7 @@ export async function handleGenAIPodcastScript(request, env) {
|
||||
|
||||
} catch (error) {
|
||||
console.error("Error in /genAIPodcastScript (outer try-catch):", error);
|
||||
const pageDateForError = dateStr || getISODate();
|
||||
const pageDateForError = dateStr || getISODate();
|
||||
const itemsForActionOnError = Array.isArray(selectedItemsParams) ? selectedItemsParams : [];
|
||||
const errorHtml = generateGenAiPageHtml(env, '生成AI播客脚本出错', `<p><strong>Unexpected error:</strong> ${escapeHtml(error.message)}</p>${error.stack ? `<pre>${escapeHtml(error.stack)}</pre>` : ''}`, pageDateForError, true, itemsForActionOnError, null, null, fullPromptForCall3_System, fullPromptForCall3_User);
|
||||
return new Response(errorHtml, { status: 500, headers: { 'Content-Type': 'text/html; charset=utf-8' } });
|
||||
@@ -146,7 +146,7 @@ export async function handleGenAIContent(request, env) {
|
||||
const errorHtml = generateGenAiPageHtml(env, '生成AI日报出错，未选生成条目', '<p><strong>No items were selected.</strong> Please go back and select at least one item.</p>', dateStr, true, null);
|
||||
return new Response(errorHtml, { status: 400, headers: { 'Content-Type': 'text/html; charset=utf-8' } });
|
||||
}
|
||||
|
||||
|
||||
console.log(`Generating AI content for ${selectedItemsParams.length} selected item references from date ${dateStr}`);
|
||||
|
||||
const allFetchedData = {};
|
||||
@@ -195,7 +195,7 @@ export async function handleGenAIContent(request, env) {
|
||||
if (item.details && item.details.content_html) itemText += `\nContent: ${stripHtml(item.details.content_html)}`;
|
||||
break;
|
||||
}
|
||||
|
||||
|
||||
if (itemText) {
|
||||
selectedContentItems.push(itemText);
|
||||
validItemsProcessedCount++;
|
||||
@@ -209,7 +209,7 @@ export async function handleGenAIContent(request, env) {
|
||||
const errorHtml = generateGenAiPageHtml(env, '生成AI日报出错，可生成条目为空', '<p><strong>Selected items could not be retrieved or resulted in no content.</strong> Please check the data or try different selections.</p>', dateStr, true, selectedItemsParams);
|
||||
return new Response(errorHtml, { status: 404, headers: { 'Content-Type': 'text/html; charset=utf-8' } });
|
||||
}
|
||||
|
||||
|
||||
//提示词内不能有英文引号，否则会存储数据缺失。
|
||||
// fullPromptForCall1_System = getSystemPromptSummarizationStepOne();
|
||||
// fullPromptForCall1_User = '\n\n------\n\n'+selectedContentItems.join('\n\n------\n\n')+'\n\n------\n\n'; // Keep this for logging/error reporting if needed
|
||||
@@ -218,11 +218,11 @@ export async function handleGenAIContent(request, env) {
|
||||
// try {
|
||||
// const chunkSize = 3;
|
||||
// const summaryPromises = [];
|
||||
|
||||
|
||||
// for (let i = 0; i < selectedContentItems.length; i += chunkSize) {
|
||||
// const chunk = selectedContentItems.slice(i, i + chunkSize);
|
||||
// const chunkPrompt = chunk.join('\n\n---\n\n'); // Join selected items with the separator
|
||||
|
||||
|
||||
// summaryPromises.push((async () => {
|
||||
// let summarizedChunks = [];
|
||||
// for await (const streamChunk of callChatAPIStream(env, chunkPrompt, fullPromptForCall1_System)) {
|
||||
@@ -264,7 +264,7 @@ export async function handleGenAIContent(request, env) {
|
||||
const errorHtml = generateGenAiPageHtml(env, '生成AI日报出错(格式化)', `<p><strong>Failed during processing of summarized content:</strong> ${escapeHtml(error.message)}</p>${error.stack ? `<pre>${escapeHtml(error.stack)}</pre>` : ''}`, dateStr, true, selectedItemsParams, fullPromptForCall2_System, fullPromptForCall2_User);
|
||||
return new Response(errorHtml, { status: 500, headers: { 'Content-Type': 'text/html; charset=utf-8' } });
|
||||
}
|
||||
|
||||
|
||||
let promptsMarkdownContent = `# Prompts for ${dateStr}\n\n`;
|
||||
// promptsMarkdownContent += `## Call 1: Content Summarization\n\n`;
|
||||
// if (fullPromptForCall1_System) promptsMarkdownContent += `### System Instruction\n\`\`\`\n${fullPromptForCall1_System}\n\`\`\`\n\n`;
|
||||
@@ -295,27 +295,27 @@ export async function handleGenAIContent(request, env) {
|
||||
return new Response(errorHtml, { status: 500, headers: { 'Content-Type': 'text/html; charset=utf-8' } });
|
||||
}
|
||||
dailySummaryMarkdownContent += '\n\n### **今日摘要**\n\n```\n' + outputOfCall3 + '\n```\n\n';
|
||||
if (env.INSERT_AD=='true') dailySummaryMarkdownContent += insertAd() +`\n`;
|
||||
|
||||
dailySummaryMarkdownContent += `\n\n${removeMarkdownCodeBlock(outputOfCall2)}`;
|
||||
if (env.INSERT_AD=='true') dailySummaryMarkdownContent += insertAd() +`\n`;
|
||||
if (env.INSERT_FOOT=='true') dailySummaryMarkdownContent += insertFoot() +`\n\n`;
|
||||
|
||||
const successHtml = generateGenAiPageHtml(
|
||||
env,
|
||||
env,
|
||||
'AI日报', // Title for Call 1 page
|
||||
escapeHtml(dailySummaryMarkdownContent),
|
||||
escapeHtml(dailySummaryMarkdownContent),
|
||||
dateStr, false, selectedItemsParams,
|
||||
fullPromptForCall2_System, fullPromptForCall2_User,
|
||||
null, null, // Pass Call 2 prompts
|
||||
convertEnglishQuotesToChinese(removeMarkdownCodeBlock(promptsMarkdownContent)),
|
||||
convertEnglishQuotesToChinese(dailySummaryMarkdownContent),
|
||||
convertEnglishQuotesToChinese(removeMarkdownCodeBlock(promptsMarkdownContent)),
|
||||
convertEnglishQuotesToChinese(dailySummaryMarkdownContent),
|
||||
null, // No podcast script for this page
|
||||
);
|
||||
return new Response(successHtml, { headers: { 'Content-Type': 'text/html; charset=utf-8' } });
|
||||
|
||||
} catch (error) {
|
||||
console.error("Error in /genAIContent (outer try-catch):", error);
|
||||
const pageDateForError = dateStr || getISODate();
|
||||
const pageDateForError = dateStr || getISODate();
|
||||
const itemsForActionOnError = Array.isArray(selectedItemsParams) ? selectedItemsParams : [];
|
||||
const errorHtml = generateGenAiPageHtml(env, '生成AI日报出错', `<p><strong>Unexpected error:</strong> ${escapeHtml(error.message)}</p>${error.stack ? `<pre>${escapeHtml(error.stack)}</pre>` : ''}`, pageDateForError, true, itemsForActionOnError, fullPromptForCall2_System, fullPromptForCall2_User);
|
||||
return new Response(errorHtml, { status: 500, headers: { 'Content-Type': 'text/html; charset=utf-8' } });
|
||||
|
||||
135
src/handlers/writeRssData.js
Normal file → Executable file
135
src/handlers/writeRssData.js
Normal file → Executable file
@@ -1,10 +1,80 @@
|
||||
import { replaceImageProxy, formatMarkdownText, formatDateToGMT8WithTime, removeMarkdownCodeBlock } from '../helpers.js';
|
||||
import { getDailyReportContent } from '../github.js';
|
||||
import { getDailyReportContent, getGitHubFileSha, createOrUpdateGitHubFile } from '../github.js';
|
||||
import { storeInKV } from '../kv.js';
|
||||
import { marked } from '../marked.esm.js';
|
||||
import { callChatAPI } from '../chatapi.js'; // 导入 callChatAPI
|
||||
import { getSummarizationSimplifyPrompt } from "../prompt/summarizationSimplifyPrompt";
|
||||
import { getAppUrl } from '../appUrl.js';
|
||||
|
||||
/**
|
||||
* 处理生成RSS内容的请求（从daily目录读取，生成AI内容，写入rss目录）
|
||||
* @param {Request} request - 请求对象
|
||||
* @param {object} env - 环境对象
|
||||
* @returns {Promise<Response>} 包含生成内容的响应
|
||||
*/
|
||||
export async function handleGenerateRssContent(request, env) {
|
||||
const url = new URL(request.url);
|
||||
const dateStr = url.searchParams.get('date');
|
||||
console.log(`[generateRssContent] Received request for date: ${dateStr}`);
|
||||
|
||||
if (!dateStr) {
|
||||
console.error('[generateRssContent] Missing date parameter');
|
||||
return new Response('Missing date parameter', { status: 400 });
|
||||
}
|
||||
|
||||
try {
|
||||
// 从daily目录读取原始内容
|
||||
const dailyPath = `daily/${dateStr}.md`;
|
||||
console.log(`[generateRssContent] Attempting to get content from GitHub path: ${dailyPath}`);
|
||||
let content = await getDailyReportContent(env, dailyPath);
|
||||
|
||||
if (!content) {
|
||||
console.warn(`[generateRssContent] No content found for ${dailyPath}. Returning 404.`);
|
||||
return new Response(`No content found for ${dailyPath}`, { status: 404 });
|
||||
}
|
||||
console.log(`[generateRssContent] Successfully retrieved content for ${dailyPath}. Content length: ${content.length}`);
|
||||
|
||||
content = extractContentFromSecondHash(content);
|
||||
|
||||
// 生成AI内容（内部已包含截断逻辑）
|
||||
const aiContent = await generateAIContent(env, content);
|
||||
|
||||
// 写入到rss目录
|
||||
const rssPath = `rss/${dateStr}.md`;
|
||||
const existingSha = await getGitHubFileSha(env, rssPath);
|
||||
const commitMessage = `${existingSha ? 'Update' : 'Create'} RSS content for ${dateStr}`;
|
||||
await createOrUpdateGitHubFile(env, rssPath, aiContent, commitMessage, existingSha);
|
||||
console.log(`[generateRssContent] Successfully wrote AI content to GitHub: ${rssPath}`);
|
||||
|
||||
// 从 "YYYY-MM-DD" 格式的 dateStr 中提取 "YYYY-MM"
|
||||
const yearMonth = dateStr.substring(0, 7);
|
||||
const result = {
|
||||
report_date: dateStr,
|
||||
title: dateStr + '日刊',
|
||||
link: '/' + yearMonth + '/' + dateStr + '/',
|
||||
content_markdown: aiContent,
|
||||
github_path: rssPath,
|
||||
published_date: formatDateToGMT8WithTime(new Date())
|
||||
};
|
||||
|
||||
console.log(`[generateRssContent] Successfully generated and saved content for ${dateStr}. Content length: ${aiContent.length}`);
|
||||
|
||||
return new Response(JSON.stringify(result), {
|
||||
headers: { 'Content-Type': 'application/json' },
|
||||
status: 200
|
||||
});
|
||||
} catch (error) {
|
||||
console.error('[generateRssContent] Error generating content:', error.message, error.stack);
|
||||
return new Response(`Error generating content: ${error.message}`, { status: 500 });
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* 处理写入RSS数据的请求（从rss目录读取已生成的内容，写入KV）
|
||||
* @param {Request} request - 请求对象
|
||||
* @param {object} env - 环境对象
|
||||
* @returns {Promise<Response>} 包含写入结果的响应
|
||||
*/
|
||||
export async function handleWriteRssData(request, env) {
|
||||
const url = new URL(request.url);
|
||||
const dateStr = url.searchParams.get('date');
|
||||
@@ -16,31 +86,28 @@ export async function handleWriteRssData(request, env) {
|
||||
}
|
||||
|
||||
try {
|
||||
const path = `daily/${dateStr}.md`;
|
||||
console.log(`[writeRssData] Attempting to get content from GitHub path: ${path}`);
|
||||
let content = await getDailyReportContent(env, path);
|
||||
|
||||
if (!content) {
|
||||
console.warn(`[writeRssData] No content found for ${path}. Returning 404.`);
|
||||
return new Response(`No content found for ${path}`, { status: 404 });
|
||||
}
|
||||
console.log(`[writeRssData] Successfully retrieved content for ${path}. Content length: ${content.length}`);
|
||||
// 从rss目录读取已生成的AI内容
|
||||
const rssPath = `rss/${dateStr}.md`;
|
||||
console.log(`[writeRssData] Attempting to get content from GitHub path: ${rssPath}`);
|
||||
let content = await getDailyReportContent(env, rssPath);
|
||||
|
||||
if (!content) {
|
||||
console.warn(`[writeRssData] No content found for ${rssPath}. Returning 404.`);
|
||||
return new Response(`No content found for ${rssPath}. Please run /generateRssContent first.`, { status: 404 });
|
||||
}
|
||||
console.log(`[writeRssData] Successfully retrieved content for ${rssPath}. Content length: ${content.length}`);
|
||||
|
||||
// content = extractContentFromSecondHash(content);
|
||||
// 从 "YYYY-MM-DD" 格式的 dateStr 中提取 "YYYY-MM"
|
||||
const yearMonth = dateStr.substring(0, 7);
|
||||
const report = {
|
||||
report_date: dateStr,
|
||||
title: dateStr+'日刊',
|
||||
link: '/'+yearMonth+'/'+dateStr+'/',
|
||||
content_html: null,
|
||||
title: dateStr + '日刊',
|
||||
link: '/' + yearMonth + '/' + dateStr + '/',
|
||||
content_html: marked.parse(formatMarkdownText(content)),
|
||||
// 可以添加其他相關欄位，例如作者、來源等
|
||||
published_date: formatDateToGMT8WithTime(new Date()) // 記錄保存時間
|
||||
}
|
||||
report.content_html = marked.parse(formatMarkdownText(replaceImageProxy(env, content)));
|
||||
//report.content_html = marked.parse(formatMarkdownText(await generateAIContent(env, content)));
|
||||
};
|
||||
|
||||
|
||||
const kvKey = `${dateStr}-report`;
|
||||
console.log(`[writeRssData] Preparing to store report in KV. Key: ${kvKey}, Report object:`, JSON.stringify(report).substring(0, 200) + '...'); // Log first 200 chars
|
||||
await storeInKV(env.DATA_KV, kvKey, report);
|
||||
@@ -76,6 +143,34 @@ export function extractContentFromSecondHash(content) {
|
||||
return content; // 如果没有找到 ### 或不符合上述条件，则返回原始内容
|
||||
}
|
||||
|
||||
/**
|
||||
* 截断内容到指定字数，并添加省略样式
|
||||
* @param {string} content - 原始内容
|
||||
* @param {number} maxLength - 最大字数，默认150
|
||||
* @returns {string} 截断后的内容
|
||||
*/
|
||||
export function truncateContent(content, maxLength = 150) {
|
||||
if (!content || content.length <= maxLength) {
|
||||
return content;
|
||||
}
|
||||
|
||||
// 截断到指定长度
|
||||
let truncated = content.substring(0, maxLength);
|
||||
|
||||
// 尝试在最后一个换行符处截断
|
||||
const lastNewlineEnd = truncated.lastIndexOf('\n');
|
||||
|
||||
// 如果找到换行符且位置合理（至少保留一半内容），则在换行符处截断
|
||||
if (lastNewlineEnd > maxLength / 2) {
|
||||
truncated = content.substring(0, lastNewlineEnd);
|
||||
}
|
||||
|
||||
// 添加省略样式
|
||||
truncated += '\n\n......\n\n*[剩余内容已省略]*';
|
||||
|
||||
return truncated;
|
||||
}
|
||||
|
||||
/**
|
||||
* 调用 Gemini 或 OpenAI 模型生成指定提示词的内容。
|
||||
* 此方法可供外部调用。
|
||||
@@ -91,7 +186,9 @@ export async function generateAIContent(env, promptText) {
|
||||
let result = await callChatAPI(env, promptText, getSummarizationSimplifyPrompt());
|
||||
console.log(`[generateAIContent] AI model returned content. Length: ${result.length}`);
|
||||
result = removeMarkdownCodeBlock(result);
|
||||
result += "\n\n</br>"+env.INSERT_APP_URL;
|
||||
// 截断内容到360字并添加省略样式
|
||||
result = truncateContent(result, 360);
|
||||
result += "\n\n</br>" + getAppUrl();
|
||||
return result;
|
||||
} catch (error) {
|
||||
console.error('[generateAIContent] Error calling AI model:', error.message, error.stack);
|
||||
|
||||
6
src/index.js
Normal file → Executable file
6
src/index.js
Normal file → Executable file
@@ -6,7 +6,7 @@ import { handleGenAIContent, handleGenAIPodcastScript, handleGenAIDailyAnalysis
|
||||
import { handleGenAIDailyPage } from './handlers/genAIDailyPage.js'; // Import handleGenAIDailyPage
|
||||
import { handleCommitToGitHub } from './handlers/commitToGitHub.js';
|
||||
import { handleRss } from './handlers/getRss.js';
|
||||
import { handleWriteRssData } from './handlers/writeRssData.js';
|
||||
import { handleWriteRssData, handleGenerateRssContent } from './handlers/writeRssData.js';
|
||||
import { dataSources } from './dataFetchers.js';
|
||||
import { handleLogin, isAuthenticated, handleLogout } from './auth.js';
|
||||
|
||||
@@ -32,7 +32,7 @@ export default {
|
||||
<p>Please contact the administrator.</p></body></html>`;
|
||||
return new Response(errorPage, { status: 503, headers: { 'Content-Type': 'text/html; charset=utf-8' } });
|
||||
}
|
||||
|
||||
|
||||
const url = new URL(request.url);
|
||||
const path = url.pathname;
|
||||
console.log(`Request received: ${request.method} ${path}`);
|
||||
@@ -48,6 +48,8 @@ export default {
|
||||
return await handleRss(request, env);
|
||||
} else if (path === '/writeRssData' && request.method === 'GET') {
|
||||
return await handleWriteRssData(request, env);
|
||||
} else if (path === '/generateRssContent' && request.method === 'GET') {
|
||||
return await handleGenerateRssContent(request, env);
|
||||
}
|
||||
|
||||
// Authentication check for all other paths
|
||||
|
||||
Reference in New Issue
Block a user