${esc(diag.title)}

[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8,"_9":10},"actionData","errors","root",{"_167":168,"_169":170},"routes/_app.x.$handle.$name",{"_96":95,"_32":44,"_105":86,"_27":35,"_13":14,"_106":-5,"_11":107,"_108":109,"_110":111,"_112":113,"_114":115,"_20":-5,"_116":-5,"_117":-5,"_118":25,"_119":120,"_121":58,"_122":-5,"_123":-5,"_124":125},"routes/_app.x.$handle.$name.branch.$branch.version.$version.diff",{"_11":12,"_13":14,"_15":16,"_17":-5,"_18":-5},"project",{"_42":43,"_32":44,"_37":45,"_46":-5,"_47":25,"_48":49,"_50":49,"_51":52,"_53":-5,"_18":54,"_55":56,"_57":58,"_59":-5,"_60":-5,"_61":62,"_63":64,"_65":-5,"_66":-5,"_67":-5,"_68":-5,"_69":70,"_71":70,"_72":73,"_74":73,"_75":73,"_76":70,"_77":70,"_78":79,"_80":81,"_82":83,"_84":25},"version",15,"changes",[19],"aiChatInfo","userId",{"_20":21,"_22":23,"_24":25,"_26":25,"_27":28,"_29":28},"file",{"_30":31,"_32":28,"_33":20,"_34":36,"_37":38,"_39":14,"_40":41,"_24":25},"previous",{"_30":31,"_32":28,"_33":20,"_34":35},"deleted",false,"deletedHere","path","README.md","latestPath","fileId","5b5faa18-a80f-11f1-8197-1607ee4eb77e","name","fileType","content","","# us-topnews-merge\n\nVal Town: https://www.val.town/x/anmeimei/us-topnews-merge Base:\nhttps://anmeimei--1349d12aa7df11f1abc71607ee4eb77e.web.val.run\n\n| 路由 | 内容 |\n| --------- | -------------------------------------------- |\n| / | CNBC Top + CBS Top 合并去重（重复保留 CNBC） |\n| /politics | CNBC 政治与政策，扣掉已在 Top News 出现过的 |\n| /economy | CNBC 经济，同上 |\n| /tech | CNBC 科技，同上 |\n\n源：\n\n- CNBC Top https://www.cnbc.com/id/100003114/device/rss/rss.html\n- CBS Top https://www.cbsnews.com/latest/rss/main\n- Politics https://www.cnbc.com/id/10000113/device/rss/rss.html\n- Economy https://www.cnbc.com/id/20910258/device/rss/rss.html\n- Tech https://www.cnbc.com/id/19854910/device/rss/rss.html\n\n参数（任何路由通用，多个参数用 & 连接，只有第一个用 ?）： ?sim=0.45 阈值 ?tag=0\n去来源前缀 ?diag=0 去诊断条目 ?debug=1 诊断网页\n\nfeed 第一条是「合并诊断」，置顶。guid 见文末 2026-09-04 那节（已改成不带日期）。\n\n## 实测 2026-09-03\n\n/ CNBC30 + CBS30 → 保留 56，丢弃 4 /politics CNBC30 → 保留 22，丢弃 8 /economy\nCNBC30 → 保留 28，丢弃 2 /tech CNBC30 → 保留 21，丢弃 9（9\n条全部是「同一篇」精确命中）\n\n板块 vs Top News 基本都是同一篇文章（链接相同），走精确匹配，不靠相似度猜。跨\nCBS/CNBC 才需要近似匹配。sim=0.36 会误杀 CBS 理财软文标题，故默认 0.45。\n\n## 2026-09-04 修复：标题完全相同但日期相隔很远的重复\n\nCBS 的 latest feed 会把 60 Minutes 旧片段按**原始播出日期**重发\n（实测出现过同一条 \"When Social Security mistakenly overpays benefits\" 的\n2025-07-27 / 2024-06-23 / 2023-12-20 三份）。原来的近似判重有 WINDOW_HOURS=36\n的时间窗，相隔几年的两条永远不会被拿来比较，所以三条全留下了。\n\n修法：在近似比较之前加一道「标题归一化后完全相同 → 同一篇」，这一步不看日期。\n归一化 = 转小写 + 非字母数字全替换成空格 + trim。零误判风险。\n本地用那三条真实标题跑过，2 条被丢、1 条保留，走的是 same-title 分支。\n\n注意 CBS 这类旧稿是它自己有意重推的内容，不是脏数据，所以只去重、不按年龄过滤。\n\n## 2026-09-04 诊断条目修正（已上线）\n\n两处改动，均在 `diagEntry()`：\n\n1. **日期按美东算**，不再用 UTC。\n `new Date().toLocaleDateString(\"en-CA\", { timeZone: \"America/New_York\" })`\n 之前用 `toISOString().slice(0,10)`，美东晚 8 点后 UTC 已经跳到第二天，\n 所以标题会显示 9/4 而实际是 9/3。\n\n2. **诊断条目的 guid 去掉日期**：`diag-${base.pathname}`。每个 feed\n 永远只有唯一一条「合并诊断」，内容随刷新就地更新，不再每天新增、不会积攒。\n\n验证（2026-09-03 ET）：\n\n| 路由 | guid | 诊断条数 | item 总数 |\n| ----------- | ---------------- | -------- | --------- |\n| `/` | `diag-/` | 1 | 57 |\n| `/politics` | `diag-/politics` | 1 | 24 |\n| `/economy` | `diag-/economy` | 1 | 30 |\n| `/tech` | `diag-/tech` | 1 | 21 |\n\n副作用：改 guid 会在 NetNewsWire 里各留下一条旧的孤儿诊断（四个 feed 各一条），\nNNW 会自行清掉。这是最后一次——以后 guid 固定，不会再产生孤儿。\n\n## 2026-09-04 查证：三个板块之间要不要互相去重 → 不需要\n\n问题：Tech / Politics / Economy 之间是否也有重复，要不要做成级联去重（Tech 只扣\nTop；Politics 扣 Top+Tech；Economy 扣 Top+Tech+Politics）。\n\n实测（当天四个源各 30 条，按文章 URL 比对）：\n\n| 比对 | 重叠 |\n| ------------------- | ---- |\n| Tech ∩ Politics | 1 |\n| Tech ∩ Economy | 0 |\n| Politics ∩ Economy | 0 |\n| Tech ∩ Top News | 10 |\n| Politics ∩ Top News | 7 |\n| Economy ∩ Top News | 1 |\n\n那唯一 1 条（Musk 中期选举花钱）本身就在 Top News 里，两个板块已经各自扣掉了，\n所以实际可见的跨板块重复是 0 条。又把模糊阈值放宽到 0.40 并去掉 36\n小时时间窗重跑，三个板块两两之间零个近似标题。\n\n原因是结构性的：CNBC 每篇文章基本只归一个板块，Top News\n是跨板块挑出来的一层覆盖，所以「板块 ↔ Top」重叠大，「板块 ↔ 板块」几乎不重叠。\n\n结论：不做级联。理由不只是没必要，还有代价—— 跨板块用 sim=0.45 会误杀（Economy\n一篇关税报道 vs Politics 另一篇关税报道\n是两篇不同文章，不是重复），且引入顺序依赖、/economy 每次多抓两个源。以每天 0~1\n条真重复换这些，不划算。\n\n如果以后真的反复看到同一篇出现两次，再加**只按 URL 精确匹配、不做模糊**的级联，\n那个零误杀风险。注意：这个结论基于一天的切片（90 条）， Musk\n那条证明「一篇挂两个板块」是可能的，只是罕见。","createdAt",["D",1788492098625],"branchVersion","branchId","23607ccd-58ae-46e5-8871-5f1a571d8515","id","449fec92-6e09-4aec-b796-e8acf2bfb859","us-topnews-merge",["D",1788471196775],"deletedAt","legacyEnvVars","privacy","public","httpPrivacy","description","CNBC + CBS Top News merged and de-duplicated into one RSS feed","imageUrl","4d930760-aac6-4857-b21b-f01a00f81422","pinnedFileId","1349d12a-a7df-11f1-abc7-1607ee4eb77e","pinPreview",true,"pinMigrated","screenshotVersion","likeCount",0,"orgId","01a06281-3add-7588-9875-998aa2e1101c","tursoName","tursoHostname","tursoMeasuredAt","pinned","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm","user",{"_96":95,"_104":100},"org",{"_96":95,"_97":98,"_99":100,"_18":54,"_101":102,"_103":25},"branch",{"_13":85,"_32":86,"_42":41,"_87":43,"_88":-5,"_89":-5,"_90":-5,"_91":-5,"_92":-5,"_93":44,"_94":95},"isOwned",29,"main","projectId","forkedBranchId","forkedBranchVersion","forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","projectName","userHandle","anmeimei","handle","displayName","Meimei An","avatar","https://lh3.googleusercontent.com/a/ACg8ocIGi6q-4ldhq-uZuUQBjSymuORY8Y80AuFVk-RYPSZTveoWJw=s96-c","tier","free","canRestrictVals","profileImageUrl","branchName","adoptOrgId",{"_42":43,"_32":44,"_37":163,"_46":-5,"_47":25,"_48":49,"_50":49,"_51":52,"_53":-5,"_18":54,"_55":56,"_57":58,"_59":-5,"_60":-5,"_61":62,"_63":64,"_65":-5,"_66":-5,"_67":-5,"_68":-5,"_69":70,"_71":70,"_72":73,"_74":73,"_75":73,"_76":70,"_77":70,"_78":164,"_80":165,"_82":166,"_84":25},"tree",[154,155],"files",[154,155],"counts",{"_143":62,"_144":145,"_146":62,"_147":148,"_149":150},"contents",[126,127],"upstreamProject","upstreamBranch","emptyEnvVars","flags",[],"hasHttpFiles","grants","orgs","customDomains",{},{"_42":139,"_32":28,"_30":31,"_130":-5,"_40":41,"_39":14,"_132":25,"_37":140,"_13":141,"_33":20,"_34":36,"_24":25,"_137":-5,"_27":142},{"_42":128,"_32":129,"_30":56,"_130":-5,"_40":41,"_39":131,"_132":25,"_37":133,"_13":134,"_33":135,"_34":136,"_24":25,"_137":-5,"_27":138},"0e5171ec-3c99-4401-9134-a34c4010e259","main.ts","parentId",11,"latest",["D",1788489864086],10,"http","/**\n * CNBC / CBS 新闻聚合去重\n *\n * 路由：\n * / Top News：CNBC Top + CBS Top 合并去重（重复时保留 CNBC 的版本）\n * /politics CNBC 政治与政策，扣掉已经在 Top News 里出现过的\n * /economy CNBC 经济，同上\n * /tech CNBC 科技，同上\n *\n * 板块 feed 的意义：CNBC 的板块源和它自己的 Top News 大量重叠。\n * 先把 Top News（含 CBS 那一半）当作「已经看过的」，板块里只留没见过的，\n * 这样四个 feed 加起来每条新闻只出现一次。\n *\n * 去重是「近似匹配」：两家写同一件事用词不同，只能靠词汇重合度判断，有阈值。\n * CNBC 板块与 CNBC Top 之间通常是同一篇文章（链接相同），那一步是精确匹配。\n *\n * 参数（任何路由都能用）：\n * ?sim=0.45 相似度阈值，越低去重越狠（默认 0.45）\n * ?tag=0 去掉标题前的来源前缀（只在合并多个源时才有前缀）\n * ?diag=0 不要在 feed 里插那条「合并诊断」条目\n * ?debug=1 改为输出人类可读的诊断网页\n *\n * URL 写法：多个参数之间用 & 连接，只有第一个用 ?\n * 对： /?sim=0.35&debug=1 错： /?sim=0.35/?debug=1\n */\n\ntype Item = {\n tag: string;\n pri: number;\n title: string;\n link: string;\n guid: string;\n date: Date;\n desc: string;\n};\n\ntype Src = { tag: string; pri: number; url: string };\n\n// pri 小的优先保留 → 重复时保留 CNBC 的版本。\nconst TOP: Src[] = [\n {\n tag: \"CNBC\",\n pri: 0,\n url: \"https://www.cnbc.com/id/100003114/device/rss/rss.html\",\n },\n { tag: \"CBS\", pri: 1, url: \"https://www.cbsnews.com/latest/rss/main\" },\n];\n\nconst SECTIONS: Record = {\n politics: {\n title: \"CNBC 政治与政策（已扣除 Top News）\",\n src: {\n tag: \"CNBC\",\n pri: 0,\n url: \"https://www.cnbc.com/id/10000113/device/rss/rss.html\",\n },\n },\n economy: {\n title: \"CNBC 经济（已扣除 Top News）\",\n src: {\n tag: \"CNBC\",\n pri: 0,\n url: \"https://www.cnbc.com/id/20910258/device/rss/rss.html\",\n },\n },\n tech: {\n title: \"CNBC 科技（已扣除 Top News）\",\n src: {\n tag: \"CNBC\",\n pri: 0,\n url: \"https://www.cnbc.com/id/19854910/device/rss/rss.html\",\n },\n },\n};\n\nconst DEFAULT_SIM = 0.45;\nconst WINDOW_HOURS = 36; // 只在这个时间窗内比较，避免跨天误判\n\nconst UA =\n \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 \" +\n \"(KHTML, like Gecko) Chrome/126.0 Safari/537.36\";\n\nasync function fetchText(url: string): Promise {\n const res = await fetch(url, { headers: { \"User-Agent\": UA } });\n if (!res.ok) throw new Error(`${res.status} ${url}`);\n return new TextDecoder(\"utf-8\").decode(await res.arrayBuffer());\n}\n\nconst decodeEntities = (s: string) =>\n s.replace(//g, \"$1\")\n .replace(/ /g, \" \").replace(/’|’/g, \"'\")\n .replace(/‘|‘/g, \"'\").replace(/“|“/g, '\"')\n .replace(/”|”/g, '\"').replace(/—|—/g, \"—\")\n .replace(/&/g, \"&\").replace(/</g, \"<\").replace(/>/g, \">\")\n .replace(/"/g, '\"').replace(/'|'/g, \"'\");\n\nconst tagText = (xml: string, name: string) => {\n const m = xml.match(\n new RegExp(`<${name}[^>]*>([\\\\s\\\\S]*?)<\\\\/${name}>`, \"i\"),\n );\n return m ? decodeEntities(m[1]).replace(/<[^>]*>/g, \"\").trim() : \"\";\n};\n\nfunction parseFeed(xml: string, src: Src): Item[] {\n const out: Item[] = [];\n for (const m of xml.matchAll(/][\\s\\S]*?<\\/item>/gi)) {\n const raw = m[0];\n const title = tagText(raw, \"title\");\n if (!title) continue;\n const link = tagText(raw, \"link\");\n const guid = tagText(raw, \"guid\") || link;\n const d = new Date(tagText(raw, \"pubDate\"));\n out.push({\n tag: src.tag,\n pri: src.pri,\n title,\n link,\n guid,\n date: isNaN(d.getTime()) ? new Date() : d,\n desc: tagText(raw, \"description\").slice(0, 400),\n });\n }\n return out;\n}\n\nasync function collect(srcs: Src[]) {\n const counts: Record = {};\n const errors: string[] = [];\n const items: Item[] = [];\n await Promise.all(srcs.map(async (s) => {\n try {\n const got = parseFeed(await fetchText(s.url), s);\n counts[s.tag] = (counts[s.tag] ?? 0) + got.length;\n items.push(...got);\n } catch (e) {\n counts[s.tag] = counts[s.tag] ?? 0;\n errors.push(`${s.tag}: ${e}`);\n }\n }));\n return { items, counts, errors };\n}\n\n// ---- 近似去重 ----\n\nconst STOP = new Set(\n (\"a an the and or but of to in on at for from by with as is are was were be \" +\n \"been being it its this that these those he she they them his her their \" +\n \"will would can could may might says said say new after over into about \" +\n \"amid up down out off more most than then there here what which who whom \" +\n \"you your we our us i me my not no nor so if because while during has have\")\n .split(/\\s+/),\n);\n\nfunction tokens(title: string): string[] {\n return [\n ...new Set(\n title.toLowerCase()\n .replace(/[^a-z0-9$%\\s]/g, \" \")\n .split(/\\s+/)\n .filter(Boolean)\n .filter((w) => !STOP.has(w))\n .filter((w) => w.length >= 3 || /\\d/.test(w)),\n ),\n ];\n}\n\n/**\n * 相似度 = IDF 加权的 containment。\n * 加权：共享的专有名词（Clancy、Nvidia）权重高，\"says / new / after\" 权重低。\n * containment 而非 Jaccard：两家标题长短经常差很多，Jaccard 会被长的稀释。\n * 天花板：改写到没有共同词的重复抓不到，例如\n * \"Fed holds rates steady\" vs \"Federal Reserve leaves rates unchanged\"（0.29）。\n */\nfunction makeSim(all: string[][]) {\n const n = all.length;\n const df = new Map();\n for (const doc of all) for (const w of doc) df.set(w, (df.get(w) ?? 0) + 1);\n const wt = (w: string) => Math.log((n + 1) / ((df.get(w) ?? 0) + 0.5));\n return (a: string[], b: string[]) => {\n const B = new Set(b);\n let inter = 0, sa = 0, sb = 0;\n for (const w of a) {\n sa += wt(w);\n if (B.has(w)) inter += wt(w);\n }\n for (const w of b) sb += wt(w);\n if (sa <= 0 || sb <= 0) return 0;\n return inter / Math.min(sa, sb);\n };\n}\n\ntype Dropped = { kept: Item; dropped: Item; score: number; exact: boolean };\n\nconst key = (it: Item) => it.link || it.guid;\n\n/**\n * items 内部互相去重；再扣掉与 reference 重复的。\n * reference 为空就是单纯的合并去重。\n */\nfunction reduce(items: Item[], reference: Item[], simThreshold: number) {\n const ordered = [...items].sort(\n (x, y) => x.pri - y.pri || y.date.getTime() - x.date.getTime(),\n );\n const tk = new Map();\n for (const it of [...ordered, ...reference]) tk.set(it, tokens(it.title));\n const sim = makeSim([...tk.values()]);\n\n const norm = (t: string) =>\n t.toLowerCase().replace(/[^a-z0-9]+/g, \" \").trim();\n const titleSeen = new Map();\n for (const r of reference) titleSeen.set(norm(r.title), r);\n const refKeys = new Set(reference.map(key));\n const kept: Item[] = [];\n const dropped: Dropped[] = [];\n const seen = new Set();\n\n for (const it of ordered) {\n const k = key(it);\n if (seen.has(k)) continue;\n\n // 1) 精确命中 reference（CNBC 板块 vs CNBC Top 通常就是同一篇）\n if (refKeys.has(k)) {\n const ref = reference.find((r) => key(r) === k)!;\n dropped.push({ kept: ref, dropped: it, score: 1, exact: true });\n continue;\n }\n\n // 2) 近似命中 reference 或已保留的条目\n /* identical title = same story; ignore the time window (CBS republishes old 60 Minutes segments under their original dates, years apart) */ const nt =\n norm(it.title);\n const twin = titleSeen.get(nt);\n if (twin) {\n dropped.push({ kept: twin, dropped: it, score: 1, exact: true });\n continue;\n }\n let dup: { it: Item; score: number } | null = null;\n for (const other of [...reference, ...kept]) {\n const dh = Math.abs(other.date.getTime() - it.date.getTime()) / 36e5;\n if (dh > WINDOW_HOURS) continue;\n const s = sim(tk.get(it)!, tk.get(other)!);\n if (s >= simThreshold && (!dup || s > dup.score)) {\n dup = { it: other, score: s };\n }\n }\n if (dup) {\n dropped.push({\n kept: dup.it,\n dropped: it,\n score: dup.score,\n exact: false,\n });\n continue;\n }\n\n seen.add(k);\n kept.push(it);\n titleSeen.set(nt, it);\n }\n\n kept.sort((x, y) => y.date.getTime() - x.date.getTime());\n return { kept, dropped };\n}\n\nconst esc = (s: string) =>\n s.replace(/&/g, \"&\").replace(//g, \">\")\n .replace(/\"/g, \""\").replace(/'/g, \"'\");\n\n// ---- 输出 ----\n\nfunction dropTable(dropped: Dropped[]) {\n if (!dropped.length) {\n return `这一轮没有判出重复。`;\n }\n return [...dropped].sort((a, b) => b.score - a.score).map((d) =>\n `${d.exact ? \"同一篇\" : d.score.toFixed(2)}` +\n `丢弃 ${esc(d.dropped.tag)} ${esc(d.dropped.title)}
` +\n `因为已有 ${esc(d.kept.tag)} ${esc(d.kept.title)}`\n ).join(\"\");\n}\n\nfunction summaryLine(\n counts: Record,\n keptN: number,\n droppedN: number,\n sim: number,\n) {\n return `阈值 sim=${sim}　抓到 ${\n Object.entries(counts).map(([k, v]) => `${k} ${v} 条`).join(\"、\")\n }　→ 保留 ${keptN} 条，丢弃 ${droppedN} 条。`;\n}\n\nfunction diagEntry(\n counts: Record,\n kept: Item[],\n dropped: Dropped[],\n sim: number,\n base: URL,\n) {\n const day = (() => {\n const iso = () => new Date().toISOString().slice(0, 10);\n try {\n const s = new Date().toLocaleDateString(\"en-CA\", {\n timeZone: \"America/New_York\",\n });\n return /^\\d{4}-\\d{2}-\\d{2}$/.test(s) ? s : iso();\n } catch {\n return iso();\n }\n })(); /* label the day in US Eastern, not UTC: after 8pm ET the UTC date has already rolled over */\n const html =\n `

${summaryLine(counts, kept.length, dropped.length, sim)}

` +\n `` +\n `${\n dropTable(dropped)\n }

相似度	丢弃了什么 / 为什么

` +\n `

换阈值：订阅地址改成 ${base.origin}${base.pathname}?sim=0.35（越低去重越狠）。` +\n `不想看到这条：地址后面加 &diag=0。

`;\n return {\n title:\n `合并诊断 · ${day}（丢弃 ${dropped.length} 条 / 保留 ${kept.length} 条）`,\n html,\n guid:\n `diag-${base.pathname}`, /* no date: one diagnostic per feed forever, updated in place, never accumulates */\n link: `${base.origin}${base.pathname}?sim=${sim}&debug=1`,\n };\n}\n\nfunction buildRss(\n chanTitle: string,\n chanDesc: string,\n items: Item[],\n showTag: boolean,\n diag: ReturnType | null,\n) {\n const parts: string[] = [];\n if (diag) {\n // 排在最新一条之后一分钟 → 置顶。guid 按天固定，一天只新增一条，不会反复未读。\n const newest = items.length ? items[0].date.getTime() : Date.now();\n parts.push(\n `${esc(diag.title)}${esc(diag.link)}` +\n `${esc(diag.guid)}` +\n `${new Date(newest + 60_000).toUTCString()}` +\n `${esc(diag.html)}`,\n );\n }\n for (const it of items) {\n const title = showTag ? `${it.tag} · ${it.title}` : it.title;\n parts.push(\n `${esc(title)}${esc(it.link)}` +\n `${esc(it.guid)}` +\n `${it.date.toUTCString()}` +\n `${esc(it.desc)}`,\n );\n }\n return `\n\n${esc(chanTitle)}\nhttps://www.cnbc.com/\n${esc(chanDesc)}\nen-US\n${new Date().toUTCString()}\n${parts.join(\"\")}\n`;\n}\n\nfunction debugPage(\n chanTitle: string,\n counts: Record,\n kept: Item[],\n dropped: Dropped[],\n sim: number,\n) {\n return `${\n esc(chanTitle)\n } · 诊断\n\n

${esc(chanTitle)} · 合并诊断

\n\n

${summaryLine(counts, kept.length, dropped.length, sim)}

阈值调低去重更狠（可能误杀），调高更保守（可能漏重复）。多个参数用 & 连接，\n只有第一个用 ?：/tech?sim=0.35&debug=1

\n${\n dropTable(dropped)\n }

相似度	丢弃了什么 / 为什么

`;\n}\n\nexport default async function (req: Request): Promise {\n const url = new URL(req.url);\n const sim = Math.min(\n 0.95,\n Math.max(0.2, Number(url.searchParams.get(\"sim\")) || DEFAULT_SIM),\n );\n const wantTag = url.searchParams.get(\"tag\") !== \"0\";\n const wantDiag = url.searchParams.get(\"diag\") !== \"0\";\n const path = url.pathname.replace(/^\\/|\\/$/g, \"\");\n\n const plain = (body: string, status = 200) =>\n new Response(body, {\n status,\n headers: { \"Content-Type\": \"text/plain; charset=utf-8\" },\n });\n\n if (path !== \"\" && !SECTIONS[path]) {\n return plain(\n \"可用地址：\\n / Top News（CNBC + CBS 去重）\\n\" +\n Object.entries(SECTIONS).map(([k, v]) =>\n ` /${k}${\" \".repeat(11 - k.length)}${v.title}`\n )\n .join(\"\\n\") +\n \"\\n\\n参数：?sim= ?tag=0 ?diag=0 ?debug=1\",\n 404,\n );\n }\n\n let chanTitle: string, chanDesc: string, counts: Record;\n let kept: Item[], dropped: Dropped[], showTag: boolean, errors: string[];\n\n if (path === \"\") {\n const got = await collect(TOP);\n ({ counts, errors } = got);\n const r = reduce(got.items, [], sim);\n kept = r.kept;\n dropped = r.dropped;\n chanTitle = \"US Top News（CNBC + CBS 去重）\";\n chanDesc =\n \"CNBC Top News 与 CBS Top News 合并，同一条新闻只保留一份（优先 CNBC 的版本）。\";\n showTag = wantTag; // 两个来源，标来源有意义\n } else {\n const sec = SECTIONS[path];\n const [got, top] = await Promise.all([collect([sec.src]), collect(TOP)]);\n counts = { ...got.counts, \"Top News（参照）\": top.items.length };\n errors = [...got.errors, ...top.errors];\n const r = reduce(got.items, top.items, sim);\n kept = r.kept;\n dropped = r.dropped;\n chanTitle = sec.title;\n chanDesc = \"只保留没有在 Top News 那个 feed 里出现过的条目。\";\n showTag = false; // 单一来源，前缀是废话\n }\n\n if (!kept.length && dropped.length === 0) {\n return plain(\n \"一条都没抓到。把这句话告诉 Claude。\\n\" + errors.join(\"\\n\"),\n 500,\n );\n }\n\n if (url.searchParams.get(\"debug\") === \"1\") {\n return new Response(debugPage(chanTitle, counts, kept, dropped, sim), {\n headers: { \"Content-Type\": \"text/html; charset=utf-8\" },\n });\n }\n\n const diag = wantDiag ? diagEntry(counts, kept, dropped, sim, url) : null;\n return new Response(buildRss(chanTitle, chanDesc, kept, showTag, diag), {\n headers: {\n \"Content-Type\": \"application/rss+xml; charset=utf-8\",\n \"Cache-Control\": \"public, max-age=300\",\n },\n });\n}","label",[129],"964e5ebe-dea7-437f-96ef-cf0554d57122",["D",1788492098625],3,[28],"pulls","branches",1,"forks","envVars",{"_151":62,"_153":62},"connections",{"_151":62,"_152":62},"count","unfilledCount","emptyCount",{"_42":139,"_32":28,"_30":31,"_130":-5,"_40":41,"_39":14,"_132":25,"_37":160,"_13":141,"_33":20,"_34":36,"_24":25,"_137":-5,"_157":161,"_27":162},{"_42":128,"_32":129,"_30":56,"_130":-5,"_40":41,"_39":131,"_132":25,"_37":156,"_13":134,"_33":135,"_34":136,"_24":25,"_137":-5,"_157":158,"_27":159},["D",1788489864086],"children",[],[129],["D",1788492098625],[],[28],["D",1788471196775],{"_96":95,"_104":100},{"_96":95,"_97":98,"_99":100,"_18":54,"_101":102,"_103":25},{"_13":85,"_32":86,"_42":41,"_87":43,"_88":-5,"_89":-5,"_90":-5,"_91":-5,"_92":-5,"_93":44,"_94":95},"ctx",{"_78":-5,"_123":200,"_201":202,"_203":204,"_205":25,"_206":25,"_207":25,"_208":58,"_209":-5,"_210":-7,"_211":-7,"_212":62,"_213":25,"_214":25,"_215":-5,"_216":25,"_217":25,"_218":25,"_219":-5},"ENV",{"_171":172,"_173":174,"_175":86,"_176":177,"_178":179,"_180":181,"_182":183,"_184":185,"_186":187,"_188":189,"_190":191,"_192":193,"_194":195,"_196":197,"_198":199},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","5015b8df46402d8bba4f485ecc37aec202b0e06b","RENDER_GIT_BRANCH","POSTHOG_API_TOKEN","phc_nmoyuAkStEZuUp69rYlg1DZgrIt19Gx0Y4Hwf8tGgJP","POSTHOG_API_HOST","https://wart.val.town","WEB_API_TEMPLATE","https://{val}.web.val.run","APP_URL","https://www.val.town","MODULE_URL","https://esm.town","VANITY_WEB_API_TEMPLATE","https://{label}.val.run/","DENOLS_URL","https://lsp.val.town","NODE_ENV","production","CLOUDFLARE_TURNSTILE_SITE_KEY","0x4AAAAAACtaIZJH7s9rBMZ6","GOOGLE_CLIENT_ID","875989696899-jtsqkhlbdv5jrlsk8631p7ipjg3c6o62.apps.googleusercontent.com","STRIPE_PUBLISHABLE_KEY","pk_live_51N2zvWEpZTTLipIkVEwXrUG9AT0yeOOEYxszarZT1jpsZV58Ly6wlg3SltZpHN26CUalOBmUSY7adQRQiSkCZz2500qSXNkf8Y","API_URL","https://api.val.town",[],"preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","preferencesCloseBrackets","preferencesChatPanelState","preferencesDprint","preferencesDenofmt","notificationCount","flagDenols","isBusinessOrgMember","businessOrgHandle","isBusinessOrgAdmin","isHigherTownieTier","navCollapsed","currentOrgId"]

Diff · v15