@press2ai/engine 0.6.0 → 1.0.0

package/package.json

@@ -1,40 +1,15 @@
 {
   "name": "@press2ai/engine",
-  "version": "0.6.0",
-  "description": "Multi-tenant runtime + template contracts dla otwarty-* verticali. SSR na Cloudflare Workers, isomorphic renderer (SSG/browser w roadmapie).",
+  "version": "1.0.0",
+  "description": "CEIDG data library for otwarty-* verticals. Loads leads, transforms to profiles, builds sitemap/catalog/llms.txt. Zero opinions about rendering.",
   "type": "module",
   "license": "MIT",
-  "files": [
-    "src",
-    "README.md"
-  ],
-  "publishConfig": {
-    "access": "public"
-  },
-  "exports": {
-    ".": "./src/index.ts",
-    "./template-catalog": "./src/template-catalog.ts",
-    "./template-blog": "./src/template-blog.ts",
-    "./ceidg-vertical": "./src/ceidg-vertical.ts",
-    "./types": "./src/types.ts"
-  },
-  "scripts": {
-    "check": "tsc --noEmit"
-  },
-  "dependencies": {
-    "zod": "^3.23.8"
-  },
-  "peerDependencies": {
-    "hono": "^4.0.0"
-  },
-  "peerDependenciesMeta": {
-    "hono": {
-      "optional": true
-    }
-  },
+  "files": ["src", "README.md"],
+  "publishConfig": { "access": "public" },
+  "exports": { ".": "./src/index.ts" },
+  "scripts": { "check": "tsc --noEmit" },
   "devDependencies": {
     "@cloudflare/workers-types": "^4.20260101.0",
-    "hono": "^4.6.0",
     "typescript": "^5.5.0"
   }
 }

package/src/index.ts

@@ -1,32 +1,137 @@
 /**
- * @press2ai/engine — public exports.
+ * @press2ai/engine — CEIDG data library for otwarty-* verticals.
+ *
+ * Engine is data, not opinions. It loads leads from D1, transforms them
+ * to profiles, and provides helpers for sitemap/catalog/llms.txt.
+ * How your page looks is YOUR decision — compose HTML however you want.
  */
 
-// Types — kontrakty templatów
-export type {
-  OtwartyTemplate,
-  PresenceTemplate,
-  CommerceTemplate,
-  RouteDescriptor,
-  RenderRequest,
-  RenderContext,
-  RenderResult,
-  EditorSchema,
-  EditorField,
-  EditorSection,
-  DataSource,
-  DataSourceInput,
-  SitemapEntry,
-  RssFeed,
-  RssItem,
-  CommerceProviders,
-  PaymentProviderId,
-  ShippingProviderId,
-  InvoiceProviderId,
-  ProviderSlot,
-  BuildFile,
-  TemplateMeta,
-} from './types';
-
-// Runtime — Faza 0 SSR multi-tenant
-export { createSSRWorker, type SSRWorkerOpts, type OtwartyWorker } from './runtime/ssr';
+/* ─────────────── Types ─────────────── */
+
+export interface CatalogProfile {
+  slug: string;
+  firstName: string;
+  lastName: string;
+  jobTitle: string;
+  city?: string;
+  bio?: string;
+  photo?: string;
+  specialties: string[];
+  languages: string[];
+  business: { name?: string; taxId?: string; classification: string[] };
+  social: Record<string, string>;
+}
+
+export interface CeidgLead {
+  nip: string;
+  first_name: string;
+  last_name: string;
+  city: string | null;
+  company_name: string;
+  pkd: string | null;
+  slug: string;
+  claimed: number;
+  external_site_url: string | null;
+  fetched_at: number;
+  opted_out_at: number | null;
+}
+
+export type CeidgBindings = { DB: D1Database; CACHE: KVNamespace };
+
+/* ─────────────── Data (D1) ─────────────── */
+
+const LEADS_SQL = `
+  SELECT l.nip, l.first_name, l.last_name, c.name as city,
+         l.company_name, l.slug, l.claimed, l.external_site_url, l.fetched_at,
+         l.opted_out_at,
+         (SELECT pkd FROM lead_pkd WHERE nip = l.nip LIMIT 1) as pkd
+  FROM leads l
+  LEFT JOIN cities c ON l.city_id = c.id
+  WHERE l.opted_out_at IS NULL
+    AND EXISTS (SELECT 1 FROM lead_categories WHERE lead_nip = l.nip AND category = ?)`;
+
+const CITIES_SQL = `
+  SELECT c.name, COUNT(*) as count
+  FROM leads l
+  JOIN cities c ON l.city_id = c.id
+  INNER JOIN lead_categories lc ON l.nip = lc.lead_nip AND lc.category = ?
+  WHERE l.opted_out_at IS NULL
+  GROUP BY c.name ORDER BY count DESC LIMIT 8`;
+
+export async function loadLeads(db: D1Database, category: string): Promise<CeidgLead[]> {
+  const { results } = await db.prepare(LEADS_SQL + ' ORDER BY c.name, l.last_name').bind(category).all<CeidgLead>();
+  return results ?? [];
+}
+
+export async function loadLeadBySlug(db: D1Database, category: string, slug: string): Promise<CeidgLead | null> {
+  return db.prepare(LEADS_SQL + ' AND l.slug = ?').bind(category, slug).first<CeidgLead>() ?? null;
+}
+
+export async function loadCities(db: D1Database, category: string): Promise<{ name: string; count: number }[]> {
+  const { results } = await db.prepare(CITIES_SQL).bind(category).all<{ name: string; count: number }>();
+  return results ?? [];
+}
+
+/* ─────────────── Transform ─────────────── */
+
+export function leadToProfile(l: CeidgLead, pkdToJobTitle: (pkd: string | null | undefined) => string): CatalogProfile {
+  return {
+    slug: l.slug, firstName: l.first_name, lastName: l.last_name,
+    jobTitle: pkdToJobTitle(l.pkd), city: l.city ?? undefined,
+    specialties: [], languages: [],
+    business: { name: l.company_name, taxId: l.nip, classification: l.pkd ? [l.pkd] : [] },
+    social: {},
+  };
+}
+
+/* ─────────────── Helpers ─────────────── */
+
+export async function kvCached<T>(kv: KVNamespace, key: string, fn: () => Promise<T>, ttl = 86400): Promise<T> {
+  const hit = await kv.get(key);
+  if (hit) return JSON.parse(hit) as T;
+  const data = await fn();
+  await kv.put(key, JSON.stringify(data), { expirationTtl: ttl });
+  return data;
+}
+
+export function filterProfiles(profiles: CatalogProfile[], query: string): CatalogProfile[] {
+  if (!query) return profiles;
+  const ql = query.toLowerCase();
+  return profiles.filter(p =>
+    `${p.firstName} ${p.lastName}`.toLowerCase().includes(ql) ||
+    (p.city ?? '').toLowerCase().includes(ql) ||
+    (p.business?.name ?? '').toLowerCase().includes(ql),
+  );
+}
+
+export function paginate<T>(items: T[], page: number, pageSize = 12): { slice: T[]; total: number; totalPages: number } {
+  const total = items.length;
+  const totalPages = Math.max(1, Math.ceil(total / pageSize));
+  const p = Math.max(1, Math.min(page, totalPages));
+  return { slice: items.slice((p - 1) * pageSize, p * pageSize), total, totalPages };
+}
+
+export function buildSitemapXml(urls: string[]): string {
+  const items = urls.map(u => `  <url><loc>${escXml(u)}</loc></url>`).join('\n');
+  return `<?xml version="1.0" encoding="UTF-8"?>\n<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">\n${items}\n</urlset>`;
+}
+
+export function buildCatalogJson(leads: CeidgLead[], category: string) {
+  return {
+    version: 1, source: 'ceidg', category, count: leads.length,
+    items: leads.map(l => ({
+      slug: l.slug, firstName: l.first_name, lastName: l.last_name,
+      city: l.city, companyName: l.company_name, claimed: !!l.claimed, url: '/' + l.slug,
+    })),
+  };
+}
+
+export function buildLlmsTxt(leads: CeidgLead[], title: string, intro: string): string {
+  return [`# ${title}`, intro, '', '## Wpisy',
+    ...leads.map(l => `- ${l.first_name} ${l.last_name} (${l.city ?? '—'}) — /${l.slug}`),
+  ].join('\n');
+}
+
+export function escXml(s: string): string {
+  return s.replace(/[<>&"']/g, c => ({ '<': '&lt;', '>': '&gt;', '&': '&amp;', '"': '&quot;', "'": '&apos;' })[c]!);
+}

package/src/ceidg-vertical.ts

@@ -1,325 +0,0 @@
-/**
- * createVerticalApp — opinionated helper dla rodziny otwarty-* (CEIDG verticals).
- *
- * Każdy wertykal (`otwarty-trener`, `otwarty-terapeuta`, `otwarty-lekarz`, ...)
- * dzielił dotąd ~230 LOC boilerplate'u: te same Hono routy, ten sam content
- * loader, te same KV-cache helpery, te same opt-out handlery — różnica tylko
- * w stałych. Helper wciąga to całe wnętrze do engine'u; wertykal staje się
- * cienkim configiem (brand, copy, footer, kategoria SQL, pages, theme bundle).
- *
- * Co tu siedzi i dlaczego:
- *   - Engine v0.1 jest GET-only (czysty SSR). Mutacje (`POST /opt-out`)
- *     żyją tutaj, w warstwie `vertical`, bo to są handlery konkretnej rodziny
- *     wertykali, nie generycznego runtime'u. Engine core (`createSSRWorker`)
- *     pozostaje agnostyczny.
- *   - Schema CEIDG (`leads`, `lead_pkd`, `lead_categories`, `cities`) jest
- *     hard-coded w queries — to jest *cel* tego helpera. „Generic vertical
- *     framework" nie istnieje, dopóki nie ma trzeciej rodziny niż otwarty-*.
- *   - Theme bundle (`CatalogTheme`) jest **input**, nie tworzony tutaj. Engine
- *     nie zna theme-specialist-glossy. Wertykal podaje gotowy bundle (~10 LOC
- *     adaptera w landingu), helper tylko go wywołuje.
- *   - `loadContent` filtruje po `EXISTS lead_categories WHERE category=?`
- *     i — fix względem starego trener-landinga — `cities` query też filtruje
- *     po kategorii (wcześniej trener pokazywał miasta z całej tabeli leads,
- *     mieszając wszystkie wertykale).
- *   - Opt-out jest **globalny** — UPDATE leads bez filtra category. RODO
- *     sprzeciw to usunięcie danych osobowych, nie selektywne ukrycie w jednej
- *     kategorii. Soft flag + filtr w queries — hard delete bez proof-of-ownership
- *     jest podatny na sabotage (patrz NEXT-STEPS manifest-1/manifest-2 notatka).
- *
- * Hono jest peer dependency (każdy wertykal i tak go ma). Engine core nie
- * importuje Hono — tylko ten subpath go używa.
- */
-
-import { Hono } from 'hono';
-import { createSSRWorker } from './runtime/ssr';
-import {
-  catalogTemplate,
-  type CatalogContent,
-  type CatalogProfile,
-  type CatalogTheme,
-} from './template-catalog';
-
-/* ─────────────── Public types ─────────────── */
-
-/** Kształt wiersza z `leads` JOIN `cities` JOIN (subselect) `lead_pkd`. */
-export type CeidgLead = {
-  nip: string;
-  first_name: string;
-  last_name: string;
-  city: string | null;
-  company_name: string;
-  pkd: string | null;
-  slug: string;
-  claimed: number;
-  external_site_url: string | null;
-  fetched_at: number;
-  opted_out_at: number | null;
-};
-
-export type CeidgBindings = { DB: D1Database; CACHE: KVNamespace };
-
-/** Statyczna strona prawna/info — pełne body HTML, opcjonalny meta description. */
-export interface StaticPage {
-  title: string;
-  body: string;
-  description?: string;
-}
-
-export interface VerticalConfig {
-  /** Wartość `lead_categories.category`. Używana w SQL i w prefiksach KV cache. */
-  category: string;
-  /** Wersja landinga — wyświetlana w `/health` (debug) i w stopce. */
-  version: string;
-  brand: { siteName: string; description: string };
-  copy: {
-    heroBadge?: string;
-    heroTitle: string;
-    heroSubtitle?: string;
-    searchPlaceholder?: string;
-    /** Plural label for count in statBar (e.g. 'trenerów', 'terapeutów'). Default 'wpisów'. */
-    itemsLabel?: string;
-    /** Singular fallback for 404 (e.g. 'trenera', 'terapeuty'). Default 'wpisu'. */
-    itemSingular?: string;
-  };
-  /** PKD → human-readable jobTitle. Per-wertykal mapping. */
-  pkdToJobTitle(pkd: string | null | undefined): string;
-  /** Statyczne strony — RODO/regulamin/opt-out. Body HTML wkleja się w layout. */
-  pages: {
-    zasady: StaticPage;
-    optOutForm: StaticPage;
-    optOutDone: StaticPage;
-    /** Optional — RODO art. 14 dla linków społecznościowych. */
-    linkiInfo?: StaticPage;
-  };
-  /** llms.txt — header + intro paragraph. Lista profili dopisywana automatycznie. */
-  llms: { title: string; intro: string };
-  /** Static HTML sections injected into the index page between engine-rendered blocks. */
-  sections?: {
-    afterHero?: string;
-    afterGrid?: string;
-  };
-  /** Theme bundle. Wertykal buduje adapter (Profile ↔ CatalogProfile) i podaje gotowy. */
-  theme: CatalogTheme;
-}
-
-/* ─────────────── Implementation ─────────────── */
-
-const KV_TTL = 86400;
-const OPTOUT_LIMIT = 5;
-const OPTOUT_WINDOW_SEC = 3600;
-
-export function createVerticalApp(config: VerticalConfig): Hono<{ Bindings: CeidgBindings }> {
-  const app = new Hono<{ Bindings: CeidgBindings }>();
-  app.onError((err, c) => c.json({ error: err.message, stack: err.stack }, 500));
-
-  const { category, version, brand, copy, pkdToJobTitle, pages, llms, theme } = config;
-
-  // ─────────── SQL ───────────
-  // EXISTS subselect zamiast JOIN: lead z N PKDs nie multiplikuje wierszy.
-  // (SELECT pkd ... LIMIT 1) wybiera jeden PKD per lead — wystarczy do
-  // pkdToJobTitle. To jest świadomy uproszczenie: lead z mieszanymi PKDs
-  // (np. fizjoterapia + psychologia) dostanie tylko jeden label.
-  const leadsQuery = `
-    SELECT l.nip, l.first_name, l.last_name, c.name as city,
-           l.company_name, l.slug, l.claimed, l.external_site_url, l.fetched_at,
-           l.opted_out_at,
-           (SELECT pkd FROM lead_pkd WHERE nip = l.nip LIMIT 1) as pkd
-    FROM leads l
-    LEFT JOIN cities c ON l.city_id = c.id
-    WHERE l.opted_out_at IS NULL
-      AND EXISTS (SELECT 1 FROM lead_categories WHERE lead_nip = l.nip AND category = ?)`;
-
-  // Cities query filtruje po kategorii — fix względem starego trener landinga
-  // który pokazywał miasta z całej tabeli leads (mieszane wertykale).
-  const citiesQuery = `
-    SELECT c.name, COUNT(*) as count
-    FROM leads l
-    JOIN cities c ON l.city_id = c.id
-    INNER JOIN lead_categories lc ON l.nip = lc.lead_nip AND lc.category = ?
-    WHERE l.opted_out_at IS NULL
-    GROUP BY c.name ORDER BY count DESC LIMIT 8`;
-
-  // ─────────── Content loader ───────────
-  function leadToProfile(l: CeidgLead): CatalogProfile {
-    return {
-      slug: l.slug,
-      firstName: l.first_name,
-      lastName: l.last_name,
-      jobTitle: pkdToJobTitle(l.pkd),
-      city: l.city ?? undefined,
-      specialties: [],
-      languages: [],
-      business: { name: l.company_name, taxId: l.nip, classification: l.pkd ? [l.pkd] : [] },
-      social: {},
-    };
-  }
-
-  async function loadContent(_host: string, env: unknown): Promise<CatalogContent> {
-    const e = env as CeidgBindings;
-    const [leadsRes, citiesRes] = await Promise.all([
-      e.DB.prepare(leadsQuery + ' ORDER BY c.name, l.last_name').bind(category).all<CeidgLead>(),
-      e.DB.prepare(citiesQuery).bind(category).all<{ name: string; count: number }>(),
-    ]);
-    return {
-      brand: { siteName: brand.siteName, description: brand.description },
-      copy: {
-        heroBadge: copy.heroBadge,
-        heroTitle: copy.heroTitle,
-        heroSubtitle: copy.heroSubtitle,
-        searchPlaceholder: copy.searchPlaceholder ?? 'Szukaj...',
-        itemsLabel: copy.itemsLabel ?? 'wpisów',
-        itemSingular: copy.itemSingular ?? 'wpisu',
-      },
-      profiles: (leadsRes.results ?? []).map(leadToProfile),
-      cities: citiesRes.results ?? [],
-      sections: {
-        afterHero: config.sections?.afterHero ?? '',
-        afterGrid: config.sections?.afterGrid ?? '',
-      },
-    };
-  }
-
-  const engine = createSSRWorker({ template: catalogTemplate, theme, loadContent });
-
-  // ─────────── KV cache helper ───────────
-  async function cached<T>(kv: KVNamespace, key: string, fn: () => Promise<T>): Promise<T> {
-    const hit = await kv.get(key);
-    if (hit) return JSON.parse(hit) as T;
-    const data = await fn();
-    await kv.put(key, JSON.stringify(data), { expirationTtl: KV_TTL });
-    return data;
-  }
-
-  const ck = (suffix: string) => `vertical:${category}:${suffix}`;
-
-  // ─────────── Static page renderer ───────────
-  // Wszystkie strony statyczne idą przez ten helper — theme.layout dostarcza
-  // siteName/footer/CSS, wertykal podaje tylko {title, body, description}.
-  function renderStatic(page: StaticPage): string {
-    return theme.layout({ title: page.title, description: page.description }, page.body);
-  }
-
-  // ─────────── Routes: legacy / cached / mutations ───────────
-
-  app.get('/health', async (c) => {
-    const { results } = await c.env.DB
-      .prepare('SELECT COUNT(*) as cnt FROM lead_categories WHERE category = ?')
-      .bind(category)
-      .all();
-    return c.json({ ok: true, db: results, version, category });
-  });
-
-  app.get('/catalog.json', async (c) => {
-    const data = await cached(c.env.CACHE, ck('catalog.json'), async () => {
-      const { results } = await c.env.DB
-        .prepare(leadsQuery + ' ORDER BY c.name, l.last_name')
-        .bind(category)
-        .all<CeidgLead>();
-      const leads = results ?? [];
-      return {
-        version: 1,
-        source: 'ceidg',
-        category,
-        count: leads.length,
-        items: leads.map((l) => ({
-          slug: l.slug,
-          firstName: l.first_name,
-          lastName: l.last_name,
-          city: l.city,
-          companyName: l.company_name,
-          claimed: !!l.claimed,
-          url: '/' + l.slug,
-        })),
-      };
-    });
-    return c.json(data);
-  });
-
-  app.get('/llms.txt', async (c) => {
-    const text = await cached(c.env.CACHE, ck('llms.txt'), async () => {
-      const { results } = await c.env.DB
-        .prepare(leadsQuery + ' ORDER BY c.name, l.last_name')
-        .bind(category)
-        .all<CeidgLead>();
-      const leads = results ?? [];
-      return [
-        `# ${llms.title}`,
-        llms.intro,
-        '',
-        '## Wpisy',
-        ...leads.map((l) => `- ${l.first_name} ${l.last_name} (${l.city ?? '—'}) — /${l.slug}`),
-      ].join('\n');
-    });
-    return c.text(text);
-  });
-
-  app.get('/sitemap.xml', async (c) => {
-    const host = new URL(c.req.url).origin;
-    const xml = await cached(c.env.CACHE, ck(`sitemap:${host}`), async () => {
-      const { results } = await c.env.DB
-        .prepare(leadsQuery)
-        .bind(category)
-        .all<CeidgLead>();
-      const urls = [host + '/', ...(results ?? []).map((l) => host + '/' + l.slug)];
-      return (
-        '<?xml version="1.0" encoding="UTF-8"?>\n<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">\n' +
-        urls.map((u) => `  <url><loc>${u}</loc></url>`).join('\n') +
-        '\n</urlset>'
-      );
-    });
-    return c.body(xml, 200, { 'content-type': 'application/xml' });
-  });
-
-  app.get('/zasady', (c) => c.html(renderStatic(pages.zasady)));
-  if (pages.linkiInfo) {
-    app.get('/linki-info', (c) => c.html(renderStatic(pages.linkiInfo!)));
-  }
-  app.get('/opt-out', (c) => c.html(renderStatic(pages.optOutForm)));
-
-  app.post('/opt-out', async (c) => {
-    const form = await c.req.formData();
-    const nip = String(form.get('nip') ?? '').replace(/\D/g, '');
-    if (nip.length !== 10) return c.text('Nieprawidłowy NIP', 400);
-
-    const ip = c.req.header('cf-connecting-ip') ?? '0.0.0.0';
-    const rlKey = `vertical:${category}:rl:optout:${ip}`;
-    const rlVal = parseInt((await c.env.CACHE.get(rlKey)) ?? '0', 10);
-    if (rlVal >= OPTOUT_LIMIT) return c.text('Zbyt wiele prób. Spróbuj ponownie za godzinę.', 429);
-    await c.env.CACHE.put(rlKey, String(rlVal + 1), { expirationTtl: OPTOUT_WINDOW_SEC });
-
-    // Opt-out jest GLOBALNY — usuwa też z innych wertykali. RODO sprzeciw to
-    // usunięcie danych osobowych, nie selektywne ukrycie w jednej kategorii.
-    // Soft flag (nie DELETE) — defense in depth przeciw masowej sabotage:
-    // catalog.json emituje slug+imię+miasto bez NIP, więc adversary musi
-    // zmapować slug→NIP przez CEIDG pod rate limit (6000 req/h); dodatkowo
-    // soft flag jest natychmiast odwracalny jednym UPDATE gdyby coś poszło nie tak.
-    const res = await c.env.DB
-      .prepare(
-        'UPDATE leads SET opted_out_at = ?, opted_out_ip = ?, opted_out_ua = ? WHERE nip = ? AND opted_out_at IS NULL',
-      )
-      .bind(Math.floor(Date.now() / 1000), ip, c.req.header('user-agent') ?? '', nip)
-      .run();
-
-    if (!res.meta.changes) return c.text('Nie znaleziono wpisu', 404);
-    return c.html(renderStatic(pages.optOutDone));
-  });
-
-  // ─────────── /:slug — claim redirect → engine fallback ───────────
-  app.get('/:slug', async (c) => {
-    const slug = c.req.param('slug');
-    const claim = await c.env.DB
-      .prepare('SELECT claimed, external_site_url FROM leads WHERE slug = ?')
-      .bind(slug)
-      .first<{ claimed: number; external_site_url: string | null }>();
-    if (claim?.claimed && claim.external_site_url) {
-      return c.redirect(claim.external_site_url);
-    }
-    return engine.fetch(c.req.raw, c.env);
-  });
-
-  // ─────────── / — engine ───────────
-  app.get('/', (c) => engine.fetch(c.req.raw, c.env));
-
-  return app;
-}