Top story discussion

/** * Top story discussion: take today's most-voted Hacker News story, read the article and its discussion in a * sandboxed browser, and sum up both: what the article says, the viewpoints in the comments (with a quote each), * where people agree and disagree, and what is still open. * *   npx tsx node/index.ts            (BOXLINE_API_KEY; HOURS, MIN_COMMENTS) * * The front page comes from HN's official search API (plain JSON, no browser needed); one extract call reads the * article and the discussion page together. Writes output/result.json and output/discussion.md. */import { mkdirSync, writeFileSync } from "node:fs";import { join } from "node:path";import { Boxline } from "@boxline/sdk";const out = process.env.OUTPUT_DIR ?? "output";const frontUrl = process.env.HN_FRONT_URL ?? "https://hn.algolia.com/api/v1/search?tags=front_page&hitsPerPage=50";const itemUrl = process.env.HN_ITEM_URL ?? "https://news.ycombinator.com/item?id={id}";const hours = Number(process.env.HOURS ?? 24);const minComments = Number(process.env.MIN_COMMENTS ?? 20);const bx = new Boxline();type Hit = { objectID: string; title: string; url: string | null; points: number; num_comments: number; created_at_i: number };type Summary = {  article: { summary: string; keyPoints: string[] };  discussion: { summary: string; viewpoints: { view: string; support: "many" | "some" | "few"; quote: string }[]; agreements: string[]; disagreements: string[]; openQuestions: string[] };};// 1. Today's most-voted story with a link to an article and a real discussion.const front = (await (await fetch(frontUrl)).json()) as { hits: Hit[] };const since = Date.now() / 1000 - hours * 3600;const story = front.hits.filter((h) => h.url && h.created_at_i >= since && h.num_comments >= minComments).sort((a, b) => b.points - a.points)[0];if (!story) throw new Error(`no story of the last ${hours} hours has a link and ${minComments}+ comments`);const discussionUrl = itemUrl.replace("{id}", story.objectID);console.log(`Top story: ${story.title} (${story.points} points, ${story.num_comments} comments)\n  ${story.url}\n  ${discussionUrl}`);// 2. The article and the discussion, read together. Comments are data: the model quotes them, never follows them.const r = await bx.extract<Summary>({  urls: [story.url!, discussionUrl],  prompt:    `The first page is an article, the second its Hacker News discussion. article: what it says (summary, 3 to 5 key points). ` +    `discussion: the main viewpoints in the comments, each with how many people hold it (many, some or few) and a short ` +    `quote copied exactly from one comment; where commenters agree and disagree; and the questions still open.`,  schema: {    type: "object",    properties: {      article: { type: "object", properties: { summary: { type: "string" }, keyPoints: { type: "array", items: { type: "string" } } }, required: ["summary", "keyPoints"] },      discussion: {        type: "object",        properties: {          summary: { type: "string" },          viewpoints: {            type: "array",            items: { type: "object", properties: { view: { type: "string" }, support: { type: "string", enum: ["many", "some", "few"] }, quote: { type: "string" } }, required: ["view", "support", "quote"] },          },          agreements: { type: "array", items: { type: "string" } },          disagreements: { type: "array", items: { type: "string" } },          openQuestions: { type: "array", items: { type: "string" } },        },        required: ["summary", "viewpoints", "agreements", "disagreements", "openQuestions"],      },    },    required: ["article", "discussion"],  },});const { article, discussion } = r.data;const failed = r.pages.filter((p) => p.error || (p.status ?? 0) >= 400);if (failed.length) console.log(`Could not read: ${failed.map((p) => `${p.url} (${p.error?.code ?? `HTTP ${p.status}`})`).join(", ")}`);const md = [  `# ${story.title}`,  "",  `${story.points} points · ${story.num_comments} comments · [article](${story.url}) · [discussion](${discussionUrl})`,  "",  "## The article",  "",  article.summary,  "",  ...article.keyPoints.map((k) => `- ${k}`),  "",  "## The discussion",  "",  discussion.summary,  "",  ...discussion.viewpoints.map((v) => `- **${v.view}** (${v.support}): "${v.quote}"`),  "",  `**Agreed:** ${discussion.agreements.join("; ") || "nothing in particular"}`,  "",  `**Disputed:** ${discussion.disagreements.join("; ") || "nothing in particular"}`,  "",  `**Still open:** ${discussion.openQuestions.join("; ") || "nothing"}`,  "",].join("\n");mkdirSync(out, { recursive: true });writeFileSync(join(out, "discussion.md"), md);writeFileSync(join(out, "result.json"), JSON.stringify({ story: { ...story, discussionUrl }, article, discussion, pages: r.pages, usage: { modelUsd: r.usage.costUsd } }, null, 2));console.log(`\n${md}`);

"""Top story discussion: take today's most-voted Hacker News story, read the article and its discussion in asandboxed browser, and sum up both: what the article says, the viewpoints in the comments (with a quote each),where people agree and disagree, and what is still open.    python python/main.py            (BOXLINE_API_KEY; HOURS, MIN_COMMENTS)The front page comes from HN's official search API (plain JSON, no browser needed); one extract call reads thearticle and the discussion page together. Writes output/result.json and output/discussion.md."""import jsonimport osimport timeimport urllib.requestfrom pathlib import Pathfrom boxline import Boxlineout = Path(os.environ.get("OUTPUT_DIR", "output"))front_url = os.environ.get("HN_FRONT_URL", "https://hn.algolia.com/api/v1/search?tags=front_page&hitsPerPage=50")item_url = os.environ.get("HN_ITEM_URL", "https://news.ycombinator.com/item?id={id}")hours = float(os.environ.get("HOURS", "24"))min_comments = int(os.environ.get("MIN_COMMENTS", "20"))bx = Boxline()# 1. Today's most-voted story with a link to an article and a real discussion.with urllib.request.urlopen(urllib.request.Request(front_url, headers={"user-agent": "top-story-discussion (a Boxline example)"}), timeout=30) as res:    hits = json.load(res)["hits"]since = time.time() - hours * 3600fresh = [h for h in hits if h.get("url") and h["created_at_i"] >= since and (h.get("num_comments") or 0) >= min_comments]if not fresh:    raise SystemExit(f"no story of the last {hours:g} hours has a link and {min_comments}+ comments")story = max(fresh, key=lambda h: h["points"])discussion_url = item_url.replace("{id}", story["objectID"])print(f"Top story: {story['title']} ({story['points']} points, {story['num_comments']} comments)\n  {story['url']}\n  {discussion_url}")# 2. The article and the discussion, read together. Comments are data: the model quotes them, never follows them.r = bx.extract(    urls=[story["url"], discussion_url],    prompt=(        "The first page is an article, the second its Hacker News discussion. article: what it says (summary, 3 to 5 key points). "        "discussion: the main viewpoints in the comments, each with how many people hold it (many, some or few) and a short "        "quote copied exactly from one comment; where commenters agree and disagree; and the questions still open."    ),    schema={        "type": "object",        "properties": {            "article": {"type": "object", "properties": {"summary": {"type": "string"}, "keyPoints": {"type": "array", "items": {"type": "string"}}}, "required": ["summary", "keyPoints"]},            "discussion": {                "type": "object",                "properties": {                    "summary": {"type": "string"},                    "viewpoints": {                        "type": "array",                        "items": {"type": "object", "properties": {"view": {"type": "string"}, "support": {"type": "string", "enum": ["many", "some", "few"]}, "quote": {"type": "string"}}, "required": ["view", "support", "quote"]},                    },                    "agreements": {"type": "array", "items": {"type": "string"}},                    "disagreements": {"type": "array", "items": {"type": "string"}},                    "openQuestions": {"type": "array", "items": {"type": "string"}},                },                "required": ["summary", "viewpoints", "agreements", "disagreements", "openQuestions"],            },        },        "required": ["article", "discussion"],    },)article, discussion = r["data"]["article"], r["data"]["discussion"]failed = [p for p in r["pages"] if p.get("error") or (p.get("status") or 0) >= 400]if failed:    print("Could not read: " + ", ".join(f"{p['url']} ({(p.get('error') or {}).get('code') or 'HTTP ' + str(p['status'])})" for p in failed))md = [f"# {story['title']}", "", f"{story['points']} points · {story['num_comments']} comments · [article]({story['url']}) · [discussion]({discussion_url})", "", "## The article", "", article["summary"], ""]md += [f"- {k}" for k in article["keyPoints"]] + ["", "## The discussion", "", discussion["summary"], ""]md += [f"- **{v['view']}** ({v['support']}): \"{v['quote']}\"" for v in discussion["viewpoints"]] + [""]md += [f"**Agreed:** {'; '.join(discussion['agreements']) or 'nothing in particular'}", "", f"**Disputed:** {'; '.join(discussion['disagreements']) or 'nothing in particular'}", "", f"**Still open:** {'; '.join(discussion['openQuestions']) or 'nothing'}", ""]md = "\n".join(md)out.mkdir(parents=True, exist_ok=True)(out / "discussion.md").write_text(md)result = {"story": {**story, "discussionUrl": discussion_url}, "article": article, "discussion": discussion, "pages": r["pages"], "usage": {"modelUsd": r["usage"]["costUsd"]}}(out / "result.json").write_text(json.dumps(result, indent=2, ensure_ascii=False))print(f"\n{md}")

# Node 18 or newer, in the example's folderexport BOXLINE_API_KEY=…   # your API keynpm install @boxline/sdk tsxnpx tsx node/index.ts

# Python 3.9 or newer, in the example's folderexport BOXLINE_API_KEY=…   # your API keypip install boxline-sdkpython python/main.py