kochwas/src/lib/server/parsers/json-ld-recipe.ts

import { parseHTML } from 'linkedom';
import { parseIso8601Duration } from './iso8601-duration';
import { parseIngredient } from './ingredient';
import type { Recipe, Step } from '$lib/types';

type JsonLdNode = Record<string, unknown>;

function unwrapGraph(node: unknown): JsonLdNode[] {
  if (Array.isArray(node)) return node.flatMap(unwrapGraph);
  if (node && typeof node === 'object') {
    const obj = node as JsonLdNode;
    if (obj['@graph']) return unwrapGraph(obj['@graph']);
    return [obj];
  }
  return [];
}

function isRecipeType(t: unknown): boolean {
  if (typeof t === 'string') return t === 'Recipe' || t.endsWith('/Recipe');
  if (Array.isArray(t)) return t.some(isRecipeType);
  return false;
}

function toText(v: unknown): string | null {
  if (typeof v === 'string') return v.trim() || null;
  if (Array.isArray(v) && v.length > 0) return toText(v[0]);
  if (v && typeof v === 'object') {
    const o = v as JsonLdNode;
    if (typeof o.name === 'string') return o.name.trim();
    if (typeof o.text === 'string') return o.text.trim();
  }
  return null;
}

function toImageUrl(v: unknown): string | null {
  if (typeof v === 'string') return v;
  if (Array.isArray(v) && v.length > 0) return toImageUrl(v[0]);
  if (v && typeof v === 'object') {
    const o = v as JsonLdNode;
    if (typeof o.url === 'string') return o.url;
  }
  return null;
}

function toStringArray(v: unknown): string[] {
  if (Array.isArray(v)) return v.map((x) => toText(x)).filter((x): x is string => x !== null);
  if (typeof v === 'string') return v.split(',').map((s) => s.trim()).filter(Boolean);
  return [];
}

function toSteps(v: unknown): Step[] {
  const out: Step[] = [];
  const walk = (x: unknown): void => {
    if (Array.isArray(x)) {
      for (const item of x) walk(item);
      return;
    }
    if (typeof x === 'string') {
      if (x.trim()) out.push({ position: out.length + 1, text: x.trim() });
      return;
    }
    if (x && typeof x === 'object') {
      const obj = x as JsonLdNode;
      if (obj['@type'] === 'HowToSection' && obj.itemListElement) {
        walk(obj.itemListElement);
        return;
      }
      if (obj['@type'] === 'HowToStep' && typeof obj.text === 'string') {
        if (obj.text.trim()) out.push({ position: out.length + 1, text: obj.text.trim() });
        return;
      }
      if (typeof obj.text === 'string' && obj.text.trim()) {
        out.push({ position: out.length + 1, text: obj.text.trim() });
      }
    }
  };
  walk(v);
  return out;
}

function toServings(v: unknown): number | null {
  if (typeof v === 'number' && Number.isFinite(v)) return Math.trunc(v);
  if (typeof v === 'string') {
    const m = /(\d+)/.exec(v);
    if (m) return parseInt(m[1], 10);
  }
  if (Array.isArray(v) && v.length > 0) return toServings(v[0]);
  return null;
}

function findRecipeNode(html: string): JsonLdNode | null {
  const { document } = parseHTML(html);
  const scripts = document.querySelectorAll('script[type="application/ld+json"]');
  for (const script of scripts) {
    const raw = script.textContent;
    if (!raw) continue;
    try {
      const parsed = JSON.parse(raw);
      for (const node of unwrapGraph(parsed)) {
        if (isRecipeType(node['@type'])) return node;
      }
    } catch {
      // malformed JSON-LD, keep scanning
    }
  }
  return null;
}

// Microdata-Alternative zum JSON-LD: viele SSR-Sites (inkl. rezeptwelt.de)
// nutzen <div itemtype="https://schema.org/Recipe"> statt application/ld+json.
// Ein einfacher Regex reicht — wir brauchen nur das Flag, nicht die Daten.
const MICRODATA_RECIPE = /itemtype\s*=\s*["']https?:\/\/schema\.org\/Recipe["']/i;

export function hasRecipeMarkup(html: string): boolean {
  if (MICRODATA_RECIPE.test(html)) return true;
  try {
    return findRecipeNode(html) !== null;
  } catch {
    return false;
  }
}

// @deprecated use hasRecipeMarkup
export function hasRecipeJsonLd(html: string): boolean {
  return hasRecipeMarkup(html);
}

function microdataValueOf(el: Element): string {
  if (el.hasAttribute('content')) return (el.getAttribute('content') ?? '').trim();
  const tag = el.tagName.toLowerCase();
  if (tag === 'meta') return (el.getAttribute('content') ?? '').trim();
  if (tag === 'a' || tag === 'link' || tag === 'area')
    return (el.getAttribute('href') ?? '').trim();
  if (
    tag === 'img' ||
    tag === 'source' ||
    tag === 'video' ||
    tag === 'audio' ||
    tag === 'embed' ||
    tag === 'iframe' ||
    tag === 'track'
  )
    return (el.getAttribute('src') ?? '').trim();
  if (tag === 'object') return (el.getAttribute('data') ?? '').trim();
  if (tag === 'data' || tag === 'meter')
    return (el.getAttribute('value') ?? '').trim();
  if (tag === 'time')
    return (el.getAttribute('datetime') ?? el.textContent ?? '').trim();
  return (el.textContent ?? '').trim();
}

type MicroProps = Map<string, Element[]>;

function gatherMicrodataProps(scope: Element): MicroProps {
  // Alle itemprop-Descendants sammeln, dabei aber nicht in verschachtelte
  // itemscopes einsteigen (sonst landen z.B. HowToStep.text im Haupt-Scope).
  const map: MicroProps = new Map();
  function walk(el: Element) {
    for (const child of Array.from(el.children) as Element[]) {
      const hasProp = child.hasAttribute('itemprop');
      const hasScope = child.hasAttribute('itemscope');
      if (hasProp) {
        const names = (child.getAttribute('itemprop') ?? '')
          .split(/\s+/)
          .filter(Boolean);
        for (const name of names) {
          const arr = map.get(name) ?? [];
          arr.push(child);
          map.set(name, arr);
        }
      }
      if (!hasScope) walk(child);
    }
  }
  walk(scope);
  return map;
}

function microText(map: MicroProps, name: string): string | null {
  const els = map.get(name);
  if (!els || els.length === 0) return null;
  const v = microdataValueOf(els[0]);
  return v || null;
}

function microAllTexts(map: MicroProps, name: string): string[] {
  const els = map.get(name) ?? [];
  return els.map(microdataValueOf).filter((v) => v !== '');
}

// Rausholen von Text mit erhaltenen Zeilenumbrüchen — <br> → \n, Block-
// Elemente (<p>, <li> …) bekommen ebenfalls Newline-Grenzen. <img>, <script>,
// <style> werden komplett übersprungen, damit alt-Attribute und andere
// Nicht-Text-Content nicht in den Rezepttext bluten.
function textWithLineBreaks(el: Element): string {
  const BLOCK = new Set(['p', 'div', 'li', 'br', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'tr']);
  const SKIP = new Set(['script', 'style', 'img', 'noscript']);
  let out = '';
  const walk = (node: Node): void => {
    if (node.nodeType === 3) {
      out += node.nodeValue ?? '';
      return;
    }
    if (node.nodeType !== 1) return;
    const e = node as Element;
    const tag = e.tagName.toLowerCase();
    if (SKIP.has(tag)) return;
    const block = BLOCK.has(tag);
    if (tag === 'br') {
      out += '\n';
      return;
    }
    if (block && out && !out.endsWith('\n')) out += '\n';
    for (const child of Array.from(node.childNodes)) walk(child);
    if (block && out && !out.endsWith('\n')) out += '\n';
  };
  walk(el);
  return out;
}

// Teilt extrahierten Rezepttext in einzelne Schritte auf. Rezeptwelt und
// andere SSR-Sites liefern oft einen einzigen HowToStep-Block, der intern
// mit "1. …<br>2. …<br>3. …" mehrere Schritte vereint.
function splitStepText(raw: string): string[] {
  const numbered = /^(\d+)[.)]\s+(.+)$/;
  const lines = raw
    .split(/\n+/)
    .map((l) => l.replace(/\s+/g, ' ').trim())
    .filter(Boolean);
  if (lines.length === 0) return [];
  const numberedCount = lines.filter((l) => numbered.test(l)).length;
  if (numberedCount >= 2) {
    // Mehrere nummerierte Zeilen → jede ist ein eigener Schritt. Nicht-
    // nummerierte Folgezeilen gehören zum vorherigen Schritt.
    const out: string[] = [];
    let current = '';
    for (const l of lines) {
      const m = l.match(numbered);
      if (m) {
        if (current) out.push(current);
        current = m[2];
      } else {
        current += current ? ' ' + l : l;
      }
    }
    if (current) out.push(current);
    return out;
  }
  return [lines.join(' ')];
}

function stepsFromElement(el: Element): string[] {
  const textEl = el.querySelector('[itemprop="text"]') ?? el;
  const raw = textWithLineBreaks(textEl);
  return splitStepText(raw);
}

function microSteps(scope: Element): Step[] {
  const out: Step[] = [];
  let pos = 1;
  const containers = Array.from(scope.querySelectorAll('[itemprop="recipeInstructions"]'));
  for (const el of containers) {
    const itemtype = (el.getAttribute('itemtype') ?? '').toLowerCase();
    if (itemtype.includes('howtosection')) {
      // HowToSection enthält HowToStep-Kinder als itemListElement.
      const steps = Array.from(
        el.querySelectorAll(
          '[itemprop="itemListElement"]'
        )
      );
      for (const step of steps) {
        for (const t of stepsFromElement(step)) out.push({ position: pos++, text: t });
      }
    } else if (itemtype.includes('howtostep')) {
      for (const t of stepsFromElement(el)) out.push({ position: pos++, text: t });
    } else if (el.hasAttribute('itemscope')) {
      // Anderer unbekannter Scope — trotzdem Text versuchen.
      for (const t of stepsFromElement(el)) out.push({ position: pos++, text: t });
    } else {
      const lis = Array.from(el.querySelectorAll('li'));
      if (lis.length > 0) {
        for (const li of lis) {
          for (const t of splitStepText(textWithLineBreaks(li))) {
            out.push({ position: pos++, text: t });
          }
        }
      } else {
        for (const t of splitStepText(textWithLineBreaks(el))) {
          out.push({ position: pos++, text: t });
        }
      }
    }
  }
  return out;
}

export function extractRecipeFromMicrodata(html: string): Recipe | null {
  let document: Document;
  try {
    ({ document } = parseHTML(html));
  } catch {
    return null;
  }
  const scope = document.querySelector(
    '[itemtype*="schema.org/Recipe" i]'
  );
  if (!scope) return null;
  const props = gatherMicrodataProps(scope);

  const title = microText(props, 'name');
  if (!title) return null;

  const ingredients = microAllTexts(props, 'recipeIngredient')
    .map((raw, i) => parseIngredient(raw, i + 1))
    .filter((x): x is NonNullable<typeof x> => x !== null);

  const steps = microSteps(scope);
  const prep = parseIso8601Duration(microText(props, 'prepTime') ?? undefined);
  const cook = parseIso8601Duration(microText(props, 'cookTime') ?? undefined);
  const total = parseIso8601Duration(microText(props, 'totalTime') ?? undefined);

  const tags = new Set<string>([
    ...microAllTexts(props, 'recipeCategory'),
    ...microAllTexts(props, 'recipeCuisine'),
    ...microAllTexts(props, 'keywords')
  ]);

  return {
    id: null,
    title,
    description: microText(props, 'description'),
    source_url: microText(props, 'url'),
    source_domain: null,
    image_path: microText(props, 'image'),
    servings_default: toServings(microText(props, 'recipeYield')),
    servings_unit: null,
    prep_time_min: prep,
    cook_time_min: cook,
    total_time_min: total,
    cuisine: microText(props, 'recipeCuisine'),
    category: microText(props, 'recipeCategory'),
    ingredients,
    steps,
    tags: [...tags]
  };
}

export function extractRecipeFromHtml(html: string): Recipe | null {
  const node = findRecipeNode(html);
  if (!node) {
    // Fallback auf Microdata — rezeptwelt.de & andere SSR-Sites nutzen das
    // anstatt application/ld+json.
    return extractRecipeFromMicrodata(html);
  }

  const title = toText(node.name) ?? '';
  if (!title) return null;

  const ingredients = Array.isArray(node.recipeIngredient)
    ? (node.recipeIngredient as unknown[])
        .map((x, i) => (typeof x === 'string' ? parseIngredient(x, i + 1) : null))
        .filter((x): x is NonNullable<typeof x> => x !== null)
    : [];

  const steps = toSteps(node.recipeInstructions);
  const imageUrl = toImageUrl(node.image);

  const prep = parseIso8601Duration(
    typeof node.prepTime === 'string' ? node.prepTime : undefined
  );
  const cook = parseIso8601Duration(
    typeof node.cookTime === 'string' ? node.cookTime : undefined
  );
  const total = parseIso8601Duration(
    typeof node.totalTime === 'string' ? node.totalTime : undefined
  );

  const tags = new Set<string>([
    ...toStringArray(node.recipeCategory),
    ...toStringArray(node.recipeCuisine),
    ...toStringArray(node.keywords)
  ]);

  return {
    id: null,
    title,
    description: toText(node.description),
    source_url: typeof node.url === 'string' ? node.url : null,
    source_domain: null,
    image_path: imageUrl,
    servings_default: toServings(node.recipeYield),
    servings_unit: null,
    prep_time_min: prep,
    cook_time_min: cook,
    total_time_min: total,
    cuisine: toText(node.recipeCuisine),
    category: toText(node.recipeCategory),
    ingredients,
    steps,
    tags: [...tags]
  };
}