+ + + 标题 + 来源 · 日期 + + 导语：80～140 字。从一个具体细节、问题或场景切入，不要以「近日/据报道」起笔。 + 可选：用一句话预告下文，不要用「本文看点」作标题 + + 具体的小节标题（写清本节讲什么） + 正文段落 + 图说 + + … 重复 {{MIN_SECTIONS}}～7 个 article-section，段落数可 1～5 段不等 … + + 与题材相关的具体问题（禁止用「怎么看」「总结」） + 2～3 段编辑判断，可出现「我们认为」「值得留意的是」。 + +
， + 加粗只用 ，链接只用 。不要用表格、不要用行内 style 之外的复杂布局。 +5. 事实必须来自给定来源，不要编造数字、人名和时间。至少引用 1 处来源中的具体数字、时间或原话，用「据…报道」标注。 +6. 全文保持一位编辑在说话的感觉，可用「我们认为」「值得留意的是」，避免百科全书口吻。`; + +const ANTI_AI_RULES = `【去 AI 味｜与结构同等重要】 +1. 禁止出现：本文看点、一文读懂、深度解读、全面解析、综上所述、值得注意的是、不可否认、 + 随着…不断发展、在当今社会、引发广泛关注、不容忽视、从某种意义上说。 +2. 禁止以「近日」「近期」「据报道」作为文章第一句。 +3. article-toc 可以省略；若保留，nav 内不要出现「本文看点」字样，只用自然语言预告。 +4. article-verdict 的 h2 必须是与新闻相关的具体问题（如「油价还会涨多久？」），不要用「怎么看」。 +5. 标题要信息量具体，避免「深度解读 XXX」「一文看懂 XXX」。 +6. 写完后默读一遍：如果像通稿汇总而非编辑评论，就重写导语和结尾。`; + +function fill(template, style) { + const spec = NEWS_ITEM_STYLES[style] ?? NEWS_ITEM_STYLES['wechat-feature']; + return template + .replace(/\{\{STYLE\}\}/g, spec.id) + .replace(/\{\{MIN_IMAGES\}\}/g, String(spec.minImages)) + .replace(/\{\{MIN_SECTIONS\}\}/g, String(spec.minSections)) + .replace(/\{\{MIN_CHARS\}\}/g, String(spec.minChars)); +} + +/** + * 生成给 Agent 的任务描述。 + */ +export function buildNewsItemGenerationTaskSpec(newsItem = {}, env = process.env) { + const style = resolveNewsItemStyle(newsItem, env); + const spec = NEWS_ITEM_STYLES[style]; + const title = String(newsItem.canonicalTitle ?? newsItem.title ?? '').trim(); + const sourceLines = [ + title ? `标题：${title}` : '', + newsItem.category ? `栏目：${newsItem.category}` : '', + newsItem.snippet ? `摘要：${newsItem.snippet}` : '', + newsItem.url ? `原文链接：${newsItem.url}` : '', + newsItem.sourceDomain ? `来源站点：${newsItem.sourceDomain}` : '', + newsItem.image ? `新闻原图：${newsItem.image}` : '', + ].filter(Boolean); + + const voiceLine = String(newsItem.accountVoice ?? '').trim(); + const voiceRules = voiceLine + ? [ + '【本号口吻】', + voiceLine, + '全文语气、节奏、用词须符合上述定位；仍须遵守下方去 AI 味规则，不得写成通稿汇总。', + '', + ] + : []; + + const taskSpec = [ + `围绕下面这条新闻，写一篇可以直接发公众号的${spec.label}并生成静态页面。`, + `版式：${spec.description}`, + newsItem.accountLabel ? `发布账号：${newsItem.accountLabel}` : '', + '', + '【新闻素材】', + ...sourceLines, + '', + ...voiceRules, + fill(STRUCTURE_CONTRACT, style), + '', + fill(IMAGE_RULES, style), + '', + fill(WRITING_RULES, style), + '', + fill(ANTI_AI_RULES, style), + '', + `【交付校验】写完后自查：article-section ≥ ${spec.minSections} 个、 ≥ ${spec.minImages} 张、`, + `正文纯文字 ≥ ${spec.minChars} 字。不达标就继续补写补图，不要提前交付。`, + `模板版本：${NEWS_ITEM_VERSION}`, + ].join('\n'); + + return { style, spec, taskSpec }; +} + +// --------------------------------------------------------------------------- +// 生成结果校验 +// --------------------------------------------------------------------------- + +export function countNewsItemImages(html) { + const matches = String(html ?? '').match(/]*>/gi) ?? []; + return matches.filter((tag) => { + const src = /\bsrc=["']([^"']+)["']/i.exec(tag)?.[1] ?? ''; + if (!src) return false; + if (/^data:/i.test(src)) return false; + return !/\.svg(\?|$)/i.test(src); + }).length; +} + +export function countNewsItemSections(html) { + return (String(html ?? '').match(/class=["'][^"']*\barticle-section\b[^"']*["']/gi) ?? []).length; +} + +export function countNewsItemTextChars(html) { + const scope = /]*>([\s\S]*?)<\s*\/\s*article\s*>/i.exec(String(html ?? ''))?.[1] + ?? String(html ?? ''); + return scope + .replace(/<(script|style)\b[^>]*>[\s\S]*?<\s*\/\s*\1\s*>/gi, '') + .replace(/<[^>]*>/g, '') + .replace(/&[a-z#0-9]+;/gi, '') + .replace(/\s+/g, '') + .length; +} + +/** + * 生成阶段的质量闸门：Agent 交付的落地页是否达标。 + * 注意这是「生成是否合格」，推送前还有一道基于编辑后块结构的闸门。 + */ +export function validateNewsItemHtml(html, { style = 'wechat-feature' } = {}) { + const spec = NEWS_ITEM_STYLES[style] ?? NEWS_ITEM_STYLES['wechat-feature']; + const source = String(html ?? ''); + const imageCount = countNewsItemImages(source); + const sectionCount = countNewsItemSections(source); + const charCount = countNewsItemTextChars(source); + const issues = []; + + if (!source.trim()) issues.push('页面为空'); + if (!/ 标题'); + if (imageCount < spec.minImages) issues.push(`配图不足（${imageCount}/${spec.minImages}）`); + if (sectionCount < spec.minSections) issues.push(`正文小节不足（${sectionCount}/${spec.minSections}）`); + if (charCount < spec.minChars) issues.push(`正文字数不足（${charCount}/${spec.minChars}）`); + if (/]*src=["'][^"']*\.svg/i.test(source)) issues.push('存在 SVG 占位图'); + + return { + ok: issues.length === 0, + issues, + style: spec.id, + imageCount, + sectionCount, + charCount, + version: NEWS_ITEM_VERSION, + }; +} diff --git a/scheduled-task-executor.mjs b/scheduled-task-executor.mjs index 72a9b47..60f400c 100644 --- a/scheduled-task-executor.mjs +++ b/scheduled-task-executor.mjs @@ -322,6 +322,131 @@ export async function finalizeScheduledTaskPageDelivery({ return readyPaths; } +function scheduledTaskSleep(delayMs) { + return new Promise((resolve) => { + setTimeout(resolve, delayMs); + }); +} + +/** + * 等 Agent 把预期 HTML 写入磁盘并稳定（大小连续两次不变）。 + * 用于在 Goose 迟迟不发 Finish 时提前结束生成，避免空等会话收尾。 + */ +export async function waitForExpectedPageFile({ + publishDir, + relativePath, + timeoutMs = 15 * 60 * 1000, + pollIntervalMs = 2000, + stablePolls = 2, + minBytes = 100, + isPageReady = null, + sleepFn = scheduledTaskSleep, +} = {}) { + const normalized = String(relativePath ?? '').trim().replace(/^\/+/, ''); + if (!normalized || !publishDir) return null; + const filePath = path.join(publishDir, normalized); + const deadline = Date.now() + Math.max(1, Number(timeoutMs) || 1); + let lastSize = -1; + let stable = 0; + + while (Date.now() < deadline) { + try { + if (fs.existsSync(filePath)) { + const size = fs.statSync(filePath).size; + if (size >= minBytes && size === lastSize) { + stable += 1; + if (stable >= Math.max(1, Number(stablePolls) || 1)) { + if (typeof isPageReady === 'function') { + const html = fs.readFileSync(filePath, 'utf8'); + if (!isPageReady(html)) { + stable = 0; + await sleepFn(Math.max(250, Number(pollIntervalMs) || 2000)); + continue; + } + } + return normalized; + } + } else { + stable = 0; + lastSize = size; + } + } else { + stable = 0; + lastSize = -1; + } + } catch { + stable = 0; + lastSize = -1; + } + await sleepFn(Math.max(250, Number(pollIntervalMs) || 2000)); + } + return null; +} + +async function awaitGooseSessionOrExpectedPage({ + task, + publishDir, + tkmindProxy, + sessionId, + requestId, + userMessage, + timeoutMs, + logger, +}) { + const sessionWait = tkmindProxy.submitSessionReplyAndAwaitFinishForUser( + task.userId, + sessionId, + requestId, + userMessage, + { timeoutMs }, + ); + const expectedPath = String(task.expectedRelativePath ?? '').trim(); + if (!expectedPath || !publishDir) { + await sessionWait; + return { earlyReadyPath: null }; + } + + let earlyReadyPath = null; + const pageWait = waitForExpectedPageFile({ + publishDir, + relativePath: expectedPath, + timeoutMs, + isPageReady: typeof task.isExpectedPageReady === 'function' + ? task.isExpectedPageReady + : null, + }).then((found) => (found + ? 'page' + : new Promise(() => {}))); + + try { + const winner = await Promise.race([ + sessionWait.then(() => 'session'), + pageWait, + ]); + if (winner === 'page') { + earlyReadyPath = expectedPath; + sessionWait.catch((error) => { + logger.warn?.('[ScheduledTask] session still open after page materialized; continuing', { + taskId: task.id, + sessionId, + message: error instanceof Error ? error.message : String(error), + }); + }); + logger.info?.('[ScheduledTask] proceeding without session Finish; page materialized', { + taskId: task.id, + sessionId, + relativePath: expectedPath, + }); + } + } catch (error) { + pageWait.catch(() => {}); + sessionWait.catch(() => {}); + throw error; + } + + return { earlyReadyPath }; +} + export async function awaitScheduledTaskPageDelivery({ pool, userId, @@ -689,13 +814,16 @@ export async function executeScheduledTask(task, { requestId, }); - await tkmindProxy.submitSessionReplyAndAwaitFinishForUser( - task.userId, + const { earlyReadyPath } = await awaitGooseSessionOrExpectedPage({ + task, + publishDir, + tkmindProxy, sessionId, requestId, userMessage, - { timeoutMs }, - ); + timeoutMs, + logger, + }); messages = await refreshScheduledTaskMessages({ userId: task.userId, @@ -703,6 +831,22 @@ export async function executeScheduledTask(task, { tkmindProxy, sessionSnapshotService, }); + + if (earlyReadyPath) { + let deliveryText = extractScheduledTaskDeliveryText(messages, task); + if (task.userId) { + const links = [`${resolveScheduledTaskPublicBaseUrl()}/MindSpace/${task.userId}/${earlyReadyPath}`]; + deliveryText = `${deliveryText}\n\n页面链接：\n${links.join('\n')}`.trim(); + } + return { + sessionId, + requestId, + deliveryText, + messages, + readyPaths: [earlyReadyPath], + executor: 'goose', + }; + } } let deliveryText = extractScheduledTaskDeliveryText(messages, task); diff --git a/scheduled-task-executor.test.mjs b/scheduled-task-executor.test.mjs index 50d0c14..2d54f3f 100644 --- a/scheduled-task-executor.test.mjs +++ b/scheduled-task-executor.test.mjs @@ -18,6 +18,7 @@ import { resendScheduledTaskWechatForReadyPage, resolveScheduledTaskDeliveryPollIntervalMs, resolveScheduledTaskDeliveryRetryDelaysMs, + waitForExpectedPageFile, } from './scheduled-task-executor.mjs'; test('buildScheduledTaskExecutionPrompt includes task spec and automation marker', () => { @@ -296,3 +297,59 @@ test('resendScheduledTaskWechatForReadyPage skips suspended users', async () => assert.equal(sent, false); assert.equal(sendCount, 0); }); + +test('waitForExpectedPageFile resolves once html file size stabilizes', async () => { + const root = await fs.mkdtemp(path.join(os.tmpdir(), 'scheduled-page-')); + const publishDir = path.join(root, 'MindSpace', 'user-1'); + const relativePath = 'public/news-item-demo.html'; + const filePath = path.join(publishDir, relativePath); + await fs.mkdir(path.dirname(filePath), { recursive: true }); + + let polls = 0; + const pending = waitForExpectedPageFile({ + publishDir, + relativePath, + pollIntervalMs: 20, + stablePolls: 2, + minBytes: 10, + sleepFn: (delayMs) => new Promise((resolve) => { + polls += 1; + setTimeout(async () => { + if (polls === 2) { + await fs.writeFile(filePath, 'draft', 'utf8'); + } + resolve(delayMs); + }, 0); + }), + }); + + assert.equal(await pending, relativePath); +}); + +test('waitForExpectedPageFile waits until isPageReady passes', async () => { + const root = await fs.mkdtemp(path.join(os.tmpdir(), 'scheduled-page-ready-')); + const publishDir = path.join(root, 'MindSpace', 'user-2'); + const relativePath = 'public/news-item-ready.html'; + const filePath = path.join(publishDir, relativePath); + await fs.mkdir(path.dirname(filePath), { recursive: true }); + + let polls = 0; + const pending = waitForExpectedPageFile({ + publishDir, + relativePath, + pollIntervalMs: 20, + stablePolls: 2, + minBytes: 10, + isPageReady: (html) => html.includes('DONE'), + sleepFn: (delayMs) => new Promise((resolve) => { + polls += 1; + setTimeout(async () => { + const body = polls >= 4 ? 'DONE' : 'draft'; + await fs.writeFile(filePath, body, 'utf8'); + resolve(delayMs); + }, 0); + }), + }); + + assert.equal(await pending, relativePath); +}); diff --git a/scripts/news-item-generate-page.mjs b/scripts/news-item-generate-page.mjs new file mode 100644 index 0000000..4252c69 --- /dev/null +++ b/scripts/news-item-generate-page.mjs @@ -0,0 +1,128 @@ +#!/usr/bin/env node +/** + * 阶段一：Agent 深度分析单条新闻并生成落地页。只生成，不推送。 + * + * stdin { userId, newsItem, runId? } + * stdout { ok, pageSlug, pageUrl, relativePath, publishDir, html, check, executor } + * + * 落地页生成后即可点链接查看；后续的预览、编辑、推送都不再需要跑 Agent。 + */ +import fs from 'node:fs'; +import path from 'node:path'; +import process from 'node:process'; +import { executeScheduledTask } from '../scheduled-task-executor.mjs'; +import { + buildPublicUrl, + PUBLISH_ROOT_DIR, + resolvePublicBaseUrl, +} from '../user-publish.mjs'; +import { + buildNewsItemGenerationTaskSpec, + validateNewsItemHtml, +} from '../news-item-templates.mjs'; +import { + bootstrapAgentContext, + loadPushEnvironment, + readStdinPayload, + runScript, +} from './news-item-wechat-context.mjs'; + +function slugify(value) { + return String(value ?? '') + .trim() + .toLowerCase() + .replace(/[^\p{L}\p{N}]+/gu, '-') + .replace(/^-+|-+$/g, '') + .slice(0, 48) || 'news-item'; +} + +function buildGenerationTask(userId, newsItem, env, now = Date.now()) { + const date = new Date(now); + const mmdd = `${String(date.getMonth() + 1).padStart(2, '0')}${String(date.getDate()).padStart(2, '0')}`; + const slug = `news-item-${slugify(newsItem.canonicalTitle || newsItem.title)}-${mmdd}`; + const { style, taskSpec } = buildNewsItemGenerationTaskSpec(newsItem, env); + return { + id: `news-item-${slug}`, + userId, + title: `新闻深度稿：${newsItem.canonicalTitle || newsItem.title}`, + taskSpec: [ + taskSpec, + '', + `输出文件名必须是 public/${slug}.html（可覆盖同名旧文件）。`, + `data-style 必须设为 ${style}。`, + ].join('\n'), + recurrence: 'once', + timezone: 'Asia/Shanghai', + notifyChannel: 'web', + expectedSlug: slug, + expectedRelativePath: `public/${slug}.html`, + layoutStyle: style, + isExpectedPageReady: (html) => validateNewsItemHtml(html, { style }).ok, + }; +} + +function resolveTimeoutMs(env) { + return Math.max( + 5 * 60_000, + Number(env.MEMIND_NEWS_ITEM_GENERATE_TIMEOUT_MS ?? 20 * 60_000) || 20 * 60_000, + ); +} + +async function main() { + const env = loadPushEnvironment(process.env); + const payload = readStdinPayload(); + const userId = String(payload.userId ?? '').trim(); + const newsItem = payload.newsItem ?? {}; + if (!userId) throw new Error('缺少 userId'); + + const services = await bootstrapAgentContext(env); + try { + const task = buildGenerationTask(userId, newsItem, env); + const execution = await executeScheduledTask(task, { + userAuth: services.userAuth, + tkmindProxy: services.tkmindProxy, + agentRunGateway: services.agentRunGateway, + cursorExecutorPolicyService: services.cursorExecutorPolicyService, + sessionSnapshotService: services.sessionSnapshotService, + pool: services.pool, + h5Root: services.h5Root, + timeoutMs: resolveTimeoutMs(env), + logger: console, + }); + + const publishDir = path.join(services.h5Root, PUBLISH_ROOT_DIR, userId); + const relativePath = execution.readyPaths?.[0] + ?? (fs.existsSync(path.join(publishDir, task.expectedRelativePath)) + ? task.expectedRelativePath + : null); + if (!relativePath) { + throw new Error('Agent 未生成 public/*.html 页面'); + } + + const localPath = path.join(publishDir, relativePath); + if (!fs.existsSync(localPath)) throw new Error(`生成页面不存在：${relativePath}`); + const html = fs.readFileSync(localPath, 'utf8'); + const check = validateNewsItemHtml(html, { style: task.layoutStyle }); + if (!check.ok) { + throw Object.assign( + new Error(`生成页面不达标：${check.issues.join('；')}`), + { code: 'news_item_page_substandard' }, + ); + } + + return { + pageSlug: path.basename(relativePath, '.html'), + pageUrl: buildPublicUrl(resolvePublicBaseUrl(env), userId, relativePath), + relativePath, + publishDir, + html, + check, + layoutStyle: task.layoutStyle, + executor: execution.executor ?? null, + }; + } finally { + await services.close(); + } +} + +runScript(main); diff --git a/scripts/news-item-wechat-commit.mjs b/scripts/news-item-wechat-commit.mjs new file mode 100644 index 0000000..c86fcbe --- /dev/null +++ b/scripts/news-item-wechat-commit.mjs @@ -0,0 +1,79 @@ +#!/usr/bin/env node +/** + * 阶段三：把用户确认过的正文原样提交到公众号草稿箱。 + * + * 这里刻意不做任何 HTML 转换——content 就是预览页显示的那一份，由 news-engine 的 + * 块渲染器产出。任何在这里的二次加工都会破坏「预览即草稿」。 + * + * stdin { userId, title, author?, digest, content, contentSourceUrl, thumb: { url?, localPath? } } + * stdout { ok, draftMediaId } + */ +import fs from 'node:fs'; +import process from 'node:process'; +import { + addWechatDraftArticle, + uploadWechatPermanentThumb, +} from '../wechat-news-morning-draft.mjs'; +import { + createLightContext, + loadPushEnvironment, + readStdinPayload, + runScript, +} from './news-item-wechat-context.mjs'; + +const PLACEHOLDER_SVG = '' + + '' + + 'TKMind' + + ''; + +async function loadThumbBuffer(thumb) { + const sharp = (await import('sharp')).default; + let raw = null; + const localPath = String(thumb?.localPath ?? '').trim(); + const url = String(thumb?.url ?? '').trim(); + if (localPath && fs.existsSync(localPath)) { + raw = fs.readFileSync(localPath); + } else if (/^https?:\/\//i.test(url)) { + const response = await fetch(url, { signal: AbortSignal.timeout(20_000) }); + if (response.ok) raw = Buffer.from(await response.arrayBuffer()); + } + if (!raw) return sharp(Buffer.from(PLACEHOLDER_SVG)).png().toBuffer(); + return sharp(raw).resize(900, 900, { fit: 'cover' }).png().toBuffer(); +} + +async function main() { + const env = loadPushEnvironment(process.env); + const payload = readStdinPayload(); + const userId = String(payload.userId ?? '').trim(); + const title = String(payload.title ?? '').trim(); + const content = String(payload.content ?? '').trim(); + if (!userId) throw new Error('缺少 userId'); + if (!title) throw new Error('缺少草稿标题'); + if (!content) throw new Error('缺少草稿正文'); + + const context = createLightContext(env); + try { + const { credentials, accessToken } = await context.resolveWechat(userId); + const { wechatFetch } = context; + const thumbMediaId = await uploadWechatPermanentThumb( + accessToken, + await loadThumbBuffer(payload.thumb), + { wechatFetch }, + ); + const draft = await addWechatDraftArticle(accessToken, { + title: title.slice(0, 64), + author: String(payload.author ?? '').trim() || credentials.author || 'TKMind', + digest: String(payload.digest ?? '').slice(0, 120), + content, + content_source_url: String(payload.contentSourceUrl ?? '').trim() || undefined, + thumb_media_id: thumbMediaId, + need_open_comment: 0, + only_fans_can_comment: 0, + }, { wechatFetch }); + return { draftMediaId: draft.draftMediaId, thumbMediaId }; + } finally { + await context.close(); + } +} + +runScript(main); diff --git a/scripts/news-item-wechat-context.mjs b/scripts/news-item-wechat-context.mjs new file mode 100644 index 0000000..7270638 --- /dev/null +++ b/scripts/news-item-wechat-context.mjs @@ -0,0 +1,220 @@ +/** + * 单条新闻 → 公众号草稿流程的共享上下文。 + * + * 流程被拆成四个独立入口（生成 / 传图 / 生成配图 / 提交草稿），只有「生成」需要 + * 完整的 Portal + Agent bootstrap；其余三个只要数据库连接和公众号凭证，因此这里 + * 提供两级上下文，避免预览和推送也背上几十秒的 bootstrap 成本。 + */ +import fs from 'node:fs'; +import path from 'node:path'; +import process from 'node:process'; +import { fileURLToPath } from 'node:url'; +import { createDbPool, isDatabaseConfigured } from '../db.mjs'; +import { createMindSpaceWechatMpConfigService, fetchWechatMpAccessToken } from '../mindspace-wechat-mp-config.mjs'; +import { resolveMindSpaceAnalyticsConfig } from '../mindspace-analytics.mjs'; +import { + resolveMindSpaceServerRuntimeOptions, + resolvePortalH5Root, +} from '../mindspace-runtime-config.mjs'; +import { startWorkspaceAssetSyncWatcher } from '../mindspace-workspace-sync.mjs'; +import { startWorkspaceThumbnailWatcher } from '../mindspace-workspace-thumbnails.mjs'; +import { bootstrapPortalAgentServices } from '../server/portal-agent-services-bootstrap.mjs'; +import { bootstrapPortalAuthServices } from '../server/portal-auth-services-bootstrap.mjs'; +import { bootstrapPortalDomainServices } from '../server/portal-domain-services-bootstrap.mjs'; +import { bootstrapPortalGatewayServices } from '../server/portal-gateway-services-bootstrap.mjs'; +import { bootstrapPortalMemorySessionServices } from '../server/portal-memory-session-services-bootstrap.mjs'; +import { resolveWechatFetch } from '../wechat-egress-fetch.mjs'; +import { loadH5Environment } from './load-env.mjs'; +import { loadMemindEnvFiles } from './memind-runtime-profile.mjs'; + +const scriptDir = path.dirname(fileURLToPath(import.meta.url)); + +export const CODE_ROOT = path.join(scriptDir, '..'); + +export function resolvePortalRoot(env = process.env) { + return path.resolve(String(env.MEMIND_PORTAL_H5_ROOT ?? '/Users/john/Project/Memind').trim()); +} + +/** 读取 stdin 上的 JSON 载荷。四个脚本统一用这种调用约定。 */ +export function readStdinPayload() { + const text = fs.readFileSync(0, 'utf8').trim(); + if (!text) throw new Error('缺少 stdin JSON 输入'); + return JSON.parse(text); +} + +/** + * 统一的环境准备：news-engine 用 MEMIND_DATABASE_URL，Memind 侧用 DATABASE_URL。 + */ +export function loadPushEnvironment(env = process.env) { + loadH5Environment(scriptDir); + loadMemindEnvFiles(resolvePortalRoot(env)); + if (!String(env.DATABASE_URL ?? '').trim() && String(env.MEMIND_DATABASE_URL ?? '').trim()) { + env.DATABASE_URL = String(env.MEMIND_DATABASE_URL).trim(); + } + env.MEMIND_PORTAL_H5_ROOT = resolvePortalRoot(env); + env.MEMIND_WORKSPACE_MAINTENANCE = env.MEMIND_WORKSPACE_MAINTENANCE ?? '0'; + env.MEMIND_PUSH_SKIP_SCHEMA_INIT = env.MEMIND_PUSH_SKIP_SCHEMA_INIT ?? '1'; + if (!isDatabaseConfigured()) { + throw new Error('DATABASE_URL / MEMIND_DATABASE_URL 未配置'); + } + return env; +} + +/** + * 轻量上下文：只要数据库和公众号凭证，用于传图 / 生成配图 / 提交草稿。 + */ +export function createLightContext(env = process.env) { + const pool = createDbPool(); + const h5Root = resolvePortalH5Root(CODE_ROOT, env); + const wechatFetch = resolveWechatFetch(env); + const wechatMpConfigService = createMindSpaceWechatMpConfigService(pool, { env, wechatFetch }); + return { + pool, + h5Root, + wechatFetch, + wechatMpConfigService, + async resolveWechat(userId) { + const credentials = await wechatMpConfigService.getCredentials(userId); + const token = await fetchWechatMpAccessToken(credentials, { wechatFetch }); + return { credentials, accessToken: token.accessToken }; + }, + async close() { + await pool.end().catch(() => {}); + }, + }; +} + +function resolveInitSchemaFn(env = process.env) { + const skip = String(env.MEMIND_PUSH_SKIP_SCHEMA_INIT ?? '1').trim().toLowerCase(); + if (skip === '1' || skip === 'true' || skip === 'yes') return async () => {}; + if (fs.existsSync(path.join(CODE_ROOT, 'schema.sql'))) return undefined; + console.warn('[news-item] schema.sql 不在 memind-lib，跳过 initSchema（假定生产库已由 Portal 初始化）'); + return async () => {}; +} + +/** + * 完整 Agent 上下文：只有页面生成阶段需要。 + */ +export async function bootstrapAgentContext(env = process.env) { + const h5Root = resolvePortalH5Root(CODE_ROOT, env); + const usersRoot = env.H5_USERS_ROOT?.trim() || path.join(h5Root, 'users'); + const runtime = resolveMindSpaceServerRuntimeOptions(h5Root, env); + const apiTarget = String(env.TKMIND_API_TARGET ?? 'https://127.0.0.1:18006').trim(); + const apiSecret = env.TKMIND_SERVER__SECRET_KEY ?? 'local-dev-secret'; + const analyticsConfig = resolveMindSpaceAnalyticsConfig(); + const pool = createDbPool(); + let userAuth = null; + let sessionSnapshotService = null; + + const domainServices = await bootstrapPortalDomainServices({ + pool, + h5Root, + env, + runtime, + analyticsConfig, + getUserAuth: () => userAuth, + getSessionSnapshotService: () => sessionSnapshotService, + workspaceMaintenanceEnabled: env.MEMIND_WORKSPACE_MAINTENANCE !== '0', + logger: console, + initSchemaFn: resolveInitSchemaFn(env), + }); + + const authServices = await bootstrapPortalAuthServices({ + pool, + h5Root, + usersRoot, + mindSearchConfigService: domainServices.mindSearchConfigService, + env, + logger: console, + }); + userAuth = authServices.userAuth; + + const agentServices = await bootstrapPortalAgentServices({ + pool, + h5Root, + env, + runtime, + apiTarget, + apiTargets: [apiTarget], + apiSecret, + userAuth, + sessionAccess: authServices.sessionAccess, + subscriptionService: authServices.subscriptionService, + mindSpaceRuntimeAdapter: domainServices.mindSpaceRuntimeAdapter, + mindSpaceAssets: domainServices.mindSpaceAssets, + resolveUserIdByDirKey: domainServices.resolveUserIdByDirKey, + workspaceMaintenanceEnabled: env.MEMIND_WORKSPACE_MAINTENANCE !== '0', + startWorkspaceThumbnailWatcher, + startWorkspaceAssetSyncWatcher, + logger: console, + }); + + const memorySessionServices = await bootstrapPortalMemorySessionServices({ + pool, + h5Root, + env, + llmProviderService: agentServices.llmProviderService, + userAuth, + sessionAccess: authServices.sessionAccess, + logger: console, + }); + sessionSnapshotService = memorySessionServices.sessionSnapshotService; + + const gatewayServices = bootstrapPortalGatewayServices({ + pool, + h5Root, + env, + apiTarget, + apiTargets: [apiTarget], + apiSecret, + userAuth, + sessionAccess: authServices.sessionAccess, + sessionStreamStore: memorySessionServices.sessionStreamStore, + llmProviderService: agentServices.llmProviderService, + subscriptionService: authServices.subscriptionService, + billingConfigService: authServices.billingConfigService, + sessionSnapshotService, + conversationMemoryService: memorySessionServices.conversationMemoryService, + memoryV2: memorySessionServices.memoryV2, + systemDisclosurePolicyService: memorySessionServices.systemDisclosurePolicyService, + wechatCursorExecutorPolicyService: memorySessionServices.wechatCursorExecutorPolicyService, + mindSpaceAssets: domainServices.mindSpaceAssets, + directChatService: memorySessionServices.directChatService, + chatIntentRouter: memorySessionServices.chatIntentRouter, + syncUserGeneratedPages: domainServices.mindSpacePageSync + ? (userId, options) => domainServices.mindSpacePageSync.syncUserGeneratedPages(userId, options) + : async () => {}, + isSessionPageDeliveryActive: () => false, + experienceService: agentServices.experienceService, + }); + + return { + pool, + h5Root, + userAuth, + tkmindProxy: gatewayServices.tkmindProxy, + agentRunGateway: gatewayServices.agentRunGateway, + cursorExecutorPolicyService: memorySessionServices.wechatCursorExecutorPolicyService, + sessionSnapshotService, + wechatMpConfigService: createMindSpaceWechatMpConfigService(pool, { env }), + async close() { + await pool.end().catch(() => {}); + }, + }; +} + +/** 四个脚本统一的 stdout 协议：最后一行是 JSON。 */ +export function runScript(main) { + main() + .then((result) => { + console.log(JSON.stringify({ ok: true, ...result })); + }) + .catch((error) => { + console.log(JSON.stringify({ + ok: false, + error: error instanceof Error ? error.message : String(error), + code: error?.code ?? null, + })); + process.exit(1); + }); +} diff --git a/scripts/news-item-wechat-images.mjs b/scripts/news-item-wechat-images.mjs new file mode 100644 index 0000000..d6a67ba --- /dev/null +++ b/scripts/news-item-wechat-images.mjs @@ -0,0 +1,184 @@ +#!/usr/bin/env node +/** + * 配图工具：把图片变成微信素材库 URL。预览和草稿共用这批 URL，所以预览里看到的 + * 图就是草稿里的图。 + * + * stdin 三种模式： + * { mode: 'upload-refs', userId, publishDir, htmlRelativePath, refs: [string] } + * { mode: 'upload-buffer', userId, publishDir, filename, dataBase64 } + * { mode: 'generate', userId, publishDir, prompt, preset: 'hero'|'inline_image' } + * + * stdout { ok, imageMap: { [ref]: { url, localPath } } } + */ +import fs from 'node:fs'; +import path from 'node:path'; +import process from 'node:process'; +import { randomUUID } from 'node:crypto'; +import { createImageMakeClientFromEnv } from '../image-make-client.mjs'; +import { uploadWechatArticleContentImage } from '../wechat-news-morning-draft.mjs'; +import { + createLightContext, + loadPushEnvironment, + readStdinPayload, + runScript, +} from './news-item-wechat-context.mjs'; + +const PRESETS = Object.freeze({ + hero: { presetId: 'memind_dark_hero', width: 1024, height: 576 }, + inline_image: { presetId: 'memind_square_illustration', width: 1024, height: 1024 }, +}); + +function normalizeWechatImageUrl(value) { + return String(value ?? '').trim().replace(/^http:\/\//i, 'https://'); +} + +/** 与 mindspace-wechat-page-draft.mjs 的 resolveImageFilePath 保持一致。 */ +function resolveImageFilePath(ref, publishDir, htmlRelativePath = '') { + const value = String(ref ?? '').trim(); + if (!value || /^data:/i.test(value)) return null; + if (/^https?:\/\//i.test(value)) return value; + const clean = value.replace(/[?#].*$/, ''); + const htmlDir = htmlRelativePath + ? path.dirname(path.join(publishDir, htmlRelativePath)) + : publishDir; + const candidates = [ + path.resolve(htmlDir, clean.replace(/^\.\//, '')), + path.resolve(path.join(publishDir, 'public'), clean.replace(/^\.\//, '')), + path.resolve(publishDir, clean.replace(/^\.\//, '')), + path.resolve(publishDir, clean.replace(/^\/+/, '')), + ]; + return candidates.find((candidate) => fs.existsSync(candidate)) ?? null; +} + +async function toPngBuffer(buffer, hint = '') { + if (!/\.(webp|svg|avif)$/i.test(hint)) return buffer; + const sharp = (await import('sharp')).default; + return sharp(buffer).png().toBuffer(); +} + +async function readSource(source) { + if (/^https?:\/\//i.test(source)) { + const response = await fetch(source, { signal: AbortSignal.timeout(20_000) }); + if (!response.ok) throw new Error(`下载配图失败（${response.status}）：${source}`); + return { buffer: Buffer.from(await response.arrayBuffer()), localPath: null }; + } + return { buffer: fs.readFileSync(source), localPath: source }; +} + +async function uploadOne(accessToken, source, { filename, wechatFetch }) { + const { buffer, localPath } = await readSource(source); + const png = await toPngBuffer(buffer, source); + const url = normalizeWechatImageUrl( + await uploadWechatArticleContentImage(accessToken, png, { + filename: `${path.basename(filename, path.extname(filename)) || 'image'}.png`, + wechatFetch, + }), + ); + if (!url) throw new Error(`微信素材上传未返回地址：${filename}`); + return { url, localPath }; +} + +function assetsDir(publishDir) { + const dir = path.join(publishDir, 'public', 'assets', 'news-item'); + fs.mkdirSync(dir, { recursive: true }); + return dir; +} + +async function handleUploadRefs({ accessToken, payload, wechatFetch }) { + const publishDir = String(payload.publishDir ?? '').trim(); + const htmlRelativePath = String(payload.htmlRelativePath ?? '').trim(); + const refs = [...new Set((payload.refs ?? []).map((ref) => String(ref ?? '').trim()).filter(Boolean))]; + const imageMap = {}; + const failures = []; + for (const ref of refs) { + const source = resolveImageFilePath(ref, publishDir, htmlRelativePath); + if (!source) { + failures.push(`找不到配图文件：${ref}`); + continue; + } + try { + imageMap[ref] = await uploadOne(accessToken, source, { filename: path.basename(ref), wechatFetch }); + } catch (error) { + failures.push(`${ref}：${error instanceof Error ? error.message : String(error)}`); + } + } + return { imageMap, failures }; +} + +async function handleUploadBuffer({ accessToken, payload, wechatFetch }) { + const publishDir = String(payload.publishDir ?? '').trim(); + const filename = String(payload.filename ?? 'upload.png').trim(); + const data = Buffer.from(String(payload.dataBase64 ?? ''), 'base64'); + if (!data.length) throw new Error('上传内容为空'); + const ext = path.extname(filename).toLowerCase() || '.png'; + const localName = `upload-${randomUUID().slice(0, 8)}${ext}`; + const localPath = publishDir ? path.join(assetsDir(publishDir), localName) : null; + if (localPath) fs.writeFileSync(localPath, data); + + const png = await toPngBuffer(data, filename); + const url = normalizeWechatImageUrl( + await uploadWechatArticleContentImage(accessToken, png, { + filename: `${path.basename(localName, ext)}.png`, + wechatFetch, + }), + ); + if (!url) throw new Error('微信素材上传未返回地址'); + return { imageMap: { [localName]: { url, localPath } }, failures: [] }; +} + +async function handleGenerate({ accessToken, payload, env, wechatFetch }) { + const client = createImageMakeClientFromEnv(env); + if (!client) throw new Error('未配置 IMAGE_MAKE_BASE_URL / IMAGE_MAKE_TOKEN，无法重新生成配图'); + const prompt = String(payload.prompt ?? '').trim(); + if (!prompt) throw new Error('缺少配图描述'); + const preset = PRESETS[String(payload.preset ?? 'inline_image')] ?? PRESETS.inline_image; + const publishDir = String(payload.publishDir ?? '').trim(); + + const generated = await client.generateImage({ + prompt, + presetId: preset.presetId, + width: preset.width, + height: preset.height, + consumerRef: `news-item:${payload.userId ?? ''}`, + }); + + const localName = `gen-${randomUUID().slice(0, 8)}.webp`; + const localPath = publishDir ? path.join(assetsDir(publishDir), localName) : null; + if (localPath) fs.writeFileSync(localPath, generated.buffer); + + const png = await toPngBuffer(generated.buffer, '.webp'); + const url = normalizeWechatImageUrl( + await uploadWechatArticleContentImage(accessToken, png, { + filename: `${path.basename(localName, '.webp')}.png`, + wechatFetch, + }), + ); + if (!url) throw new Error('微信素材上传未返回地址'); + await client.acknowledge(generated.jobId, localName).catch(() => {}); + return { imageMap: { [localName]: { url, localPath } }, failures: [] }; +} + +const HANDLERS = { + 'upload-refs': handleUploadRefs, + 'upload-buffer': handleUploadBuffer, + generate: handleGenerate, +}; + +async function main() { + const env = loadPushEnvironment(process.env); + const payload = readStdinPayload(); + const userId = String(payload.userId ?? '').trim(); + if (!userId) throw new Error('缺少 userId'); + const handler = HANDLERS[String(payload.mode ?? 'upload-refs')]; + if (!handler) throw new Error(`未知模式：${payload.mode}`); + + const context = createLightContext(env); + try { + const { accessToken } = await context.resolveWechat(userId); + return await handler({ accessToken, payload, env, wechatFetch: context.wechatFetch }); + } finally { + await context.close(); + } +} + +runScript(main);
+ 
+
+