Product alternatives

/** * Product alternatives: read a product page, find comparable products (from a web search, or the candidate pages you * give), rank them by how close they are with a reason each, and show the price difference. Prices are remembered, * so the next run says which went up or down. * *   npx tsx node/index.ts            (BOXLINE_API_KEY; PRODUCT_URL; CANDIDATES comma-separated, else a search) * * Writes output/result.json and output/alternatives.md. */import { createHash } from "node:crypto";import { existsSync, mkdirSync, readFileSync, writeFileSync } from "node:fs";import { join } from "node:path";import { Boxline } from "@boxline/sdk";const out = process.env.OUTPUT_DIR ?? "output";const stateDir = process.env.STATE_DIR ?? join(out, "state");const productUrl = process.env.PRODUCT_URL ?? "https://books.toscrape.com/catalogue/a-light-in-the-attic_1000/index.html";const given = (process.env.CANDIDATES ?? "").split(",").map((u) => u.trim()).filter(Boolean);const bx = new Boxline();type Product = { url: string; name: string; price: number | null; currency: string | null; category: string | null; features: string[] };const PRODUCT = {  type: "object",  properties: {    url: { type: "string" },    name: { type: "string" },    price: { type: ["number", "null"], description: "a number without the currency sign" },    currency: { type: ["string", "null"], description: "ISO 4217" },    category: { type: ["string", "null"] },    features: { type: "array", items: { type: "string" }, description: "what it is and does, in a few words each" },  },  required: ["url", "name", "price", "currency", "category", "features"],};// 1. The product itself.const base = await bx.extract<Product>({ url: productUrl, prompt: "The product on this page. url: the page's address as given.", schema: PRODUCT });let modelUsd = base.usage.costUsd;const product = { ...base.data, url: productUrl };console.log(`${product.name}: ${product.price ?? "?"} ${product.currency ?? ""} (${product.category ?? "no category"})`);// 2. Candidates: the pages given, or product pages a search finds (not this one).let searches = 0;let candidates = given;if (!candidates.length) {  const found = await bx.search({ query: `${product.name} alternatives similar ${product.category ?? ""}`.trim(), limit: 10 });  searches = found.cached ? 0 : 1;  candidates = found.results.map((r) => r.url).filter((u) => u !== productUrl).slice(0, 6);}const r = await bx.extract<{ items: (Product & { similarity: number; why: string })[] }>({  urls: candidates.slice(0, 10),  prompt:    `Each page may sell a product. For each page, the product on it, and how close it is to this one: "${product.name}" (${product.category ?? "?"}; ${product.features.join(", ")}). ` +    `similarity: 0 (nothing alike) to 1 (the same kind of thing for the same buyer); why: one sentence. Pages that sell no product get similarity 0. url: the page's address as given.`,  schema: { type: "object", properties: { items: { type: "array", items: { ...PRODUCT, properties: { ...PRODUCT.properties, similarity: { type: "number" }, why: { type: "string" } }, required: [...PRODUCT.required, "similarity", "why"] } } }, required: ["items"] },});modelUsd += r.usage.costUsd;// 3. Ranked, with the price difference, and the change since the last run.mkdirSync(stateDir, { recursive: true });const stateFile = join(stateDir, `${createHash("sha256").update(productUrl).digest("hex").slice(0, 16)}.json`);const before: Record<string, number | null> = existsSync(stateFile) ? JSON.parse(readFileSync(stateFile, "utf8")) : {};const ranked = r.data.items  .map((it, i) => ({ ...it, url: candidates.find((c) => c === it.url) ?? candidates[i] ?? it.url }))  .filter((it) => it.similarity > 0)  .sort((a, b) => b.similarity - a.similarity)  .map((it) => ({    ...it,    priceDiff: it.price !== null && product.price !== null && it.currency === product.currency ? Math.round((it.price - product.price) * 100) / 100 : null,    priceChange: before[it.url] !== undefined && before[it.url] !== it.price ? { from: before[it.url], to: it.price } : null,  }));writeFileSync(stateFile, JSON.stringify(Object.fromEntries([[productUrl, product.price], ...ranked.map((x) => [x.url, x.price])])));for (const x of ranked) {  const diff = x.priceDiff === null ? "" : x.priceDiff === 0 ? ", same price" : `, ${x.priceDiff > 0 ? "+" : ""}${x.priceDiff} ${x.currency}`;  console.log(`  ${x.similarity.toFixed(2)}  ${x.name}: ${x.price ?? "?"} ${x.currency ?? ""}${diff}${x.priceChange ? ` (was ${x.priceChange.from})` : ""}\n        ${x.why}`);}const md = [  `# Alternatives to ${product.name}`,  "",  `[${product.name}](${productUrl}): ${product.price ?? "?"} ${product.currency ?? ""}`,  "",  "| | Product | Price | Difference | Why |",  "|---|---|---|---|---|",  ...ranked.map((x) => `| ${x.similarity.toFixed(2)} | [${x.name}](${x.url}) | ${x.price ?? "?"} ${x.currency ?? ""} | ${x.priceDiff ?? ""} | ${x.why} |`),  "",].join("\n");mkdirSync(out, { recursive: true });writeFileSync(join(out, "alternatives.md"), md);writeFileSync(join(out, "result.json"), JSON.stringify({ product, candidates, alternatives: ranked, usage: { modelUsd, searches } }, null, 2));

"""Product alternatives: read a product page, find comparable products (from a web search, or the candidate pages yougive), rank them by how close they are with a reason each, and show the price difference. Prices are remembered,so the next run says which went up or down.    python python/main.py            (BOXLINE_API_KEY; PRODUCT_URL; CANDIDATES comma-separated, else a search)Writes output/result.json and output/alternatives.md."""import hashlibimport jsonimport osfrom pathlib import Pathfrom boxline import Boxlineout = Path(os.environ.get("OUTPUT_DIR", "output"))state_dir = Path(os.environ.get("STATE_DIR", str(out / "state")))product_url = os.environ.get("PRODUCT_URL", "https://books.toscrape.com/catalogue/a-light-in-the-attic_1000/index.html")given = [u.strip() for u in os.environ.get("CANDIDATES", "").split(",") if u.strip()]bx = Boxline()PRODUCT = {    "type": "object",    "properties": {        "url": {"type": "string"},        "name": {"type": "string"},        "price": {"type": ["number", "null"], "description": "a number without the currency sign"},        "currency": {"type": ["string", "null"], "description": "ISO 4217"},        "category": {"type": ["string", "null"]},        "features": {"type": "array", "items": {"type": "string"}, "description": "what it is and does, in a few words each"},    },    "required": ["url", "name", "price", "currency", "category", "features"],}# 1. The product itself.base = bx.extract(url=product_url, prompt="The product on this page. url: the page's address as given.", schema=PRODUCT)model_usd = base["usage"]["costUsd"]product = {**base["data"], "url": product_url}print(f"{product['name']}: {product['price']} {product['currency'] or ''} ({product['category'] or 'no category'})")# 2. Candidates: the pages given, or product pages a search finds (not this one).searches, candidates = 0, givenif not candidates:    found = bx.search(f"{product['name']} alternatives similar {product['category'] or ''}".strip(), limit=10)    searches = 0 if found["cached"] else 1    candidates = [x["url"] for x in found["results"] if x["url"] != product_url][:6]item = {**PRODUCT, "properties": {**PRODUCT["properties"], "similarity": {"type": "number"}, "why": {"type": "string"}}, "required": [*PRODUCT["required"], "similarity", "why"]}r = bx.extract(    urls=candidates[:10],    prompt=(        f'Each page may sell a product. For each page, the product on it, and how close it is to this one: "{product["name"]}" ({product["category"] or "?"}; {", ".join(product["features"])}). '        "similarity: 0 (nothing alike) to 1 (the same kind of thing for the same buyer); why: one sentence. Pages that sell no product get similarity 0. url: the page's address as given."    ),    schema={"type": "object", "properties": {"items": {"type": "array", "items": item}}, "required": ["items"]},)model_usd += r["usage"]["costUsd"]# 3. Ranked, with the price difference, and the change since the last run.state_dir.mkdir(parents=True, exist_ok=True)state_file = state_dir / f"{hashlib.sha256(product_url.encode()).hexdigest()[:16]}.json"before = json.loads(state_file.read_text()) if state_file.exists() else {}ranked = []for i, it in enumerate(r["data"]["items"]):    url = it["url"] if it["url"] in candidates else (candidates[i] if i < len(candidates) else it["url"])    if it["similarity"] <= 0:        continue    diff = round(it["price"] - product["price"], 2) if it["price"] is not None and product["price"] is not None and it["currency"] == product["currency"] else None    change = {"from": before[url], "to": it["price"]} if url in before and before[url] != it["price"] else None    ranked.append({**it, "url": url, "priceDiff": diff, "priceChange": change})ranked.sort(key=lambda x: -x["similarity"])state_file.write_text(json.dumps({product_url: product["price"], **{x["url"]: x["price"] for x in ranked}}))for x in ranked:    diff = "" if x["priceDiff"] is None else ", same price" if x["priceDiff"] == 0 else f", {'+' if x['priceDiff'] > 0 else ''}{x['priceDiff']} {x['currency']}"    was = f" (was {x['priceChange']['from']})" if x["priceChange"] else ""    print(f"  {x['similarity']:.2f}  {x['name']}: {x['price']} {x['currency'] or ''}{diff}{was}\n        {x['why']}")md = [f"# Alternatives to {product['name']}", "", f"[{product['name']}]({product_url}): {product['price']} {product['currency'] or ''}", "", "| | Product | Price | Difference | Why |", "|---|---|---|---|---|"]md += [f"| {x['similarity']:.2f} | [{x['name']}]({x['url']}) | {x['price']} {x['currency'] or ''} | {'' if x['priceDiff'] is None else x['priceDiff']} | {x['why']} |" for x in ranked] + [""]out.mkdir(parents=True, exist_ok=True)(out / "alternatives.md").write_text("\n".join(md))(out / "result.json").write_text(json.dumps({"product": product, "candidates": candidates, "alternatives": ranked, "usage": {"modelUsd": model_usd, "searches": searches}}, indent=2, ensure_ascii=False))

# 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