nutrition: extract shared ref resolution, fix HTML in ingredient names

- Move parseAnchorRecipeRef and resolveReferencedNutrition from the items endpoint into nutritionMatcher.ts for reuse - JSON-LD endpoint now includes nutrition from referenced recipes (base recipe refs and anchor-tag ingredient refs) - Strip HTML tags in normalizeIngredientName/De before matching to prevent regex crash on ingredients containing anchor tags - Escape regex special chars in substringMatchScore word-boundary check
nutrition: include NutritionInformation in recipe JSON-LD
2026-04-03 11:15:55 +02:00 · 2026-04-03 11:15:49 +02:00
4 changed files with 202 additions and 74 deletions
--- a/src/lib/js/recipeJsonLd.ts
+++ b/src/lib/js/recipeJsonLd.ts
@@ -26,7 +26,7 @@ function parseTimeToISO8601(timeString: string | undefined): string | undefined
  return undefined;
 }

-import type { RecipeModelType } from '$types/types';
+import type { RecipeModelType, NutritionMapping } from '$types/types';

 interface HowToStep {
  "@type": "HowToStep";
@@ -55,7 +55,14 @@ interface RecipeJsonLd {
  [key: string]: unknown;
 }

-export function generateRecipeJsonLd(data: RecipeModelType) {
+type ReferencedNutrition = {
+  shortName: string;
+  name: string;
+  nutrition: Record<string, number>;
+  baseMultiplier: number;
+};
+
+export function generateRecipeJsonLd(data: RecipeModelType, referencedNutrition?: ReferencedNutrition[]) {
  const jsonLd: RecipeJsonLd = {
    "@context": "https://schema.org",
    "@type": "Recipe",
@@ -144,6 +151,12 @@ export function generateRecipeJsonLd(data: RecipeModelType) {
    }
  }

+  // Add nutrition information from stored mappings
+  const nutritionInfo = computeNutritionInfo(data.ingredients || [], data.nutritionMappings, data.portions, referencedNutrition);
+  if (nutritionInfo) {
+    jsonLd.nutrition = nutritionInfo;
+  }
+
  // Clean up undefined values
  Object.keys(jsonLd).forEach(key => {
    if (jsonLd[key] === undefined) {
@@ -153,3 +166,110 @@ export function generateRecipeJsonLd(data: RecipeModelType) {

  return jsonLd;
 }
+
+function parseAmount(amount: string): number {
+  if (!amount?.trim()) return 0;
+  const s = amount.trim().replace(',', '.');
+  const rangeMatch = s.match(/^(\d+(?:\.\d+)?)\s*[-–]\s*(\d+(?:\.\d+)?)$/);
+  if (rangeMatch) return (parseFloat(rangeMatch[1]) + parseFloat(rangeMatch[2])) / 2;
+  const fractionMatch = s.match(/^(\d+)\s*\/\s*(\d+)$/);
+  if (fractionMatch) return parseInt(fractionMatch[1]) / parseInt(fractionMatch[2]);
+  const mixedMatch = s.match(/^(\d+)\s+(\d+)\s*\/\s*(\d+)$/);
+  if (mixedMatch) return parseInt(mixedMatch[1]) + parseInt(mixedMatch[2]) / parseInt(mixedMatch[3]);
+  const parsed = parseFloat(s);
+  return isNaN(parsed) ? 0 : parsed;
+}
+
+function stripHtml(html: string): string {
+  return html.replace(/<[^>]*>/g, '');
+}
+
+function computeNutritionInfo(
+  ingredients: any[],
+  mappings: NutritionMapping[] | undefined,
+  portions: string | undefined,
+  referencedNutrition?: ReferencedNutrition[],
+): Record<string, string> | null {
+  if ((!mappings || mappings.length === 0) && (!referencedNutrition || referencedNutrition.length === 0)) return null;
+
+  const index = new Map(
+    mappings.map(m => [`${m.sectionIndex}-${m.ingredientIndex}`, m])
+  );
+
+  const totals = { calories: 0, protein: 0, fat: 0, saturatedFat: 0, carbs: 0, fiber: 0, sugars: 0, sodium: 0, cholesterol: 0 };
+
+  // Collect section names for dedup
+  const sectionNames = new Set<string>();
+  for (const section of ingredients) {
+    if (section.name) sectionNames.add(stripHtml(section.name).toLowerCase().trim());
+  }
+
+  for (let si = 0; si < ingredients.length; si++) {
+    const section = ingredients[si];
+    if (section.type === 'reference' || !section.list) continue;
+    const currentSectionName = section.name ? stripHtml(section.name).toLowerCase().trim() : '';
+
+    for (let ii = 0; ii < section.list.length; ii++) {
+      const item = section.list[ii];
+      const rawName = item.name || '';
+      const itemName = stripHtml(rawName).toLowerCase().trim();
+      if (/<a\s/i.test(rawName)) continue;
+      if (itemName && sectionNames.has(itemName) && itemName !== currentSectionName) continue;
+
+      const m = index.get(`${si}-${ii}`);
+      if (!m || m.matchMethod === 'none' || m.excluded || !m.per100g) continue;
+
+      const amount = parseAmount(item.amount || '') || (m.defaultAmountUsed ? 1 : 0);
+      const grams = amount * (m.gramsPerUnit || 0);
+      const factor = grams / 100;
+
+      totals.calories += factor * m.per100g.calories;
+      totals.protein += factor * m.per100g.protein;
+      totals.fat += factor * m.per100g.fat;
+      totals.saturatedFat += factor * m.per100g.saturatedFat;
+      totals.carbs += factor * m.per100g.carbs;
+      totals.fiber += factor * m.per100g.fiber;
+      totals.sugars += factor * m.per100g.sugars;
+      totals.sodium += factor * m.per100g.sodium;
+      totals.cholesterol += factor * m.per100g.cholesterol;
+    }
+  }
+
+  // Add nutrition from referenced recipes (base refs + anchor-tag refs)
+  if (referencedNutrition) {
+    for (const ref of referencedNutrition) {
+      const scale = ref.baseMultiplier;
+      for (const key of Object.keys(totals) as (keyof typeof totals)[]) {
+        totals[key] += (ref.nutrition[key] || 0) * scale;
+      }
+    }
+  }
+
+  if (totals.calories === 0) return null;
+
+  // Parse portion count for per-serving values
+  const portionMatch = portions?.match(/^(\d+(?:[.,]\d+)?)/);
+  const portionCount = portionMatch ? parseFloat(portionMatch[1].replace(',', '.')) : 0;
+  const div = portionCount > 0 ? portionCount : 1;
+
+  const fmt = (val: number, unit: string) => `${Math.round(val / div)} ${unit}`;
+
+  const info: Record<string, string> = {
+    '@type': 'NutritionInformation',
+    calories: `${Math.round(totals.calories / div)} calories`,
+    proteinContent: fmt(totals.protein, 'g'),
+    fatContent: fmt(totals.fat, 'g'),
+    saturatedFatContent: fmt(totals.saturatedFat, 'g'),
+    carbohydrateContent: fmt(totals.carbs, 'g'),
+    fiberContent: fmt(totals.fiber, 'g'),
+    sugarContent: fmt(totals.sugars, 'g'),
+    sodiumContent: fmt(totals.sodium, 'mg'),
+    cholesterolContent: fmt(totals.cholesterol, 'mg'),
+  };
+
+  if (portionCount > 0) {
+    info.servingSize = `1 portion (${portionCount} total)`;
+  }
+
+  return info;
+}
--- a/src/lib/server/nutritionMatcher.ts
+++ b/src/lib/server/nutritionMatcher.ts
@@ -132,7 +132,7 @@ async function getBlsEmbeddingIndex() {

 /** Normalize an ingredient name for matching (English) */
 export function normalizeIngredientName(name: string): string {
-	let normalized = name.toLowerCase().trim();
+	let normalized = name.replace(/<[^>]*>/g, '').toLowerCase().trim();
 	normalized = normalized.replace(/\(.*?\)/g, '').trim();
 	for (const mod of STRIP_MODIFIERS) {
 		normalized = normalized.replace(new RegExp(`\\b${mod}\\b,?\\s*`, 'gi'), '').trim();
@@ -143,7 +143,7 @@ export function normalizeIngredientName(name: string): string {

 /** Normalize a German ingredient name for matching */
 export function normalizeIngredientNameDe(name: string): string {
-	let normalized = name.toLowerCase().trim();
+	let normalized = name.replace(/<[^>]*>/g, '').toLowerCase().trim();
 	normalized = normalized.replace(/\(.*?\)/g, '').trim();
 	for (const mod of STRIP_MODIFIERS_DE) {
 		normalized = normalized.replace(new RegExp(`\\b${mod}\\b,?\\s*`, 'gi'), '').trim();
@@ -280,7 +280,8 @@ function substringMatchScore(
 		const pos = nameLower.indexOf(form);
 		if (pos >= 0 && pos < 15) hasEarlyMatch = true;
 		// Word-boundary match
-		const wordBoundary = new RegExp(`(^|[\\s,/])${form}([\\s,/]|$)`);
+		const escaped = form.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
+		const wordBoundary = new RegExp(`(^|[\\s,/])${escaped}([\\s,/]|$)`);
 		if (wordBoundary.test(nameLower)) hasWordBoundaryMatch = true;
 	}

@@ -813,3 +814,72 @@ export function computeRecipeNutritionTotals(
 function stripHtml(html: string): string {
 	return html.replace(/<[^>]*>/g, '');
 }
+
+/** Parse anchor href from ingredient name, return recipe short_name or null */
+export function parseAnchorRecipeRef(ingredientName: string): string | null {
+	const match = ingredientName.match(/<a\s+href=["']?([^"' >]+)["']?[^>]*>/i);
+	if (!match) return null;
+	let href = match[1].trim();
+	href = href.split('?')[0];
+	if (href.startsWith('http') || href.includes('://')) return null;
+	href = href.replace(/^(\.?\/?rezepte\/|\.\/|\/)/, '');
+	if (href.includes('.')) return null;
+	return href || null;
+}
+
+export type ReferencedNutritionResult = {
+	shortName: string;
+	name: string;
+	nutrition: Record<string, number>;
+	baseMultiplier: number;
+};
+
+/**
+ * Build nutrition totals for referenced recipes:
+ * 1. Base recipe references (type='reference' with populated baseRecipeRef)
+ * 2. Anchor-tag references in ingredient names (<a href=...>)
+ */
+export async function resolveReferencedNutrition(
+	ingredients: any[],
+): Promise<ReferencedNutritionResult[]> {
+	const { Recipe } = await import('$models/Recipe');
+	const results: ReferencedNutritionResult[] = [];
+	const processedSlugs = new Set<string>();
+
+	for (const section of ingredients) {
+		// Type 1: Base recipe references
+		if (section.type === 'reference' && section.baseRecipeRef) {
+			const ref = section.baseRecipeRef;
+			const slug = ref.short_name;
+			if (processedSlugs.has(slug)) continue;
+			processedSlugs.add(slug);
+
+			if (ref.nutritionMappings?.length > 0) {
+				const mult = section.baseMultiplier || 1;
+				const nutrition = computeRecipeNutritionTotals(ref.ingredients || [], ref.nutritionMappings, 1);
+				results.push({ shortName: slug, name: ref.name, nutrition, baseMultiplier: mult });
+			}
+		}
+
+		// Type 2: Anchor-tag references in ingredient names
+		if (section.list) {
+			for (const item of section.list) {
+				const refSlug = parseAnchorRecipeRef(item.name || '');
+				if (!refSlug || processedSlugs.has(refSlug)) continue;
+				processedSlugs.add(refSlug);
+
+				const refRecipe = await Recipe.findOne({ short_name: refSlug })
+					.select('short_name name ingredients nutritionMappings portions')
+					.lean();
+				if (!refRecipe?.nutritionMappings?.length) continue;
+
+				const nutrition = computeRecipeNutritionTotals(
+					refRecipe.ingredients || [], refRecipe.nutritionMappings, 1
+				);
+				results.push({ shortName: refSlug, name: refRecipe.name, nutrition, baseMultiplier: 1 });
+			}
+		}
+	}
+
+	return results;
+}
--- a/src/routes/api/[recipeLang=recipeLang]/items/[name]/+server.ts
+++ b/src/routes/api/[recipeLang=recipeLang]/items/[name]/+server.ts
@@ -4,7 +4,7 @@ import { dbConnect } from '$utils/db';
 import { error } from '@sveltejs/kit';
 import type { RecipeModelType, IngredientItem, InstructionItem } from '$types/types';
 import { isEnglish } from '$lib/server/recipeHelpers';
-import { getNutritionEntryByFdcId, getBlsEntryByCode, computeRecipeNutritionTotals } from '$lib/server/nutritionMatcher';
+import { getNutritionEntryByFdcId, getBlsEntryByCode, resolveReferencedNutrition } from '$lib/server/nutritionMatcher';

 /** Recursively map populated baseRecipeRef to resolvedRecipe field */
 function mapBaseRecipeRefs(items: any[]): any[] {
@@ -42,72 +42,6 @@ function resolveNutritionData(mappings: any[]): any[] {
  });
 }

-/** Parse anchor href from ingredient name, return short_name or null */
-function parseAnchorRecipeRef(ingredientName: string): string | null {
-  const match = ingredientName.match(/<a\s+href=["']?([^"' >]+)["']?[^>]*>/i);
-  if (!match) return null;
-  let href = match[1].trim();
-  // Strip query params (e.g., ?multiplier={{multiplier}})
-  href = href.split('?')[0];
-  // Skip external links
-  if (href.startsWith('http') || href.includes('://')) return null;
-  // Strip leading path components like /rezepte/ or ./
-  href = href.replace(/^(\.?\/?rezepte\/|\.\/|\/)/, '');
-  // Skip if contains a dot (file extensions, external domains)
-  if (href.includes('.')) return null;
-  return href || null;
-}
-
-/**
- * Build nutrition totals for referenced recipes:
- * 1. Base recipe references (type='reference' with populated baseRecipeRef)
- * 2. Anchor-tag references in ingredient names (<a href=...>)
- */
-async function resolveReferencedNutrition(
-  ingredients: any[],
-): Promise<{ shortName: string; name: string; nutrition: Record<string, number>; baseMultiplier: number }[]> {
-  const results: { shortName: string; name: string; nutrition: Record<string, number>; baseMultiplier: number }[] = [];
-  const processedSlugs = new Set<string>();
-
-  for (const section of ingredients) {
-    // Type 1: Base recipe references
-    if (section.type === 'reference' && section.baseRecipeRef) {
-      const ref = section.baseRecipeRef;
-      const slug = ref.short_name;
-      if (processedSlugs.has(slug)) continue;
-      processedSlugs.add(slug);
-
-      if (ref.nutritionMappings?.length > 0) {
-        const mult = section.baseMultiplier || 1;
-        const nutrition = computeRecipeNutritionTotals(ref.ingredients || [], ref.nutritionMappings, 1);
-        results.push({ shortName: slug, name: ref.name, nutrition, baseMultiplier: mult });
-      }
-    }
-
-    // Type 2: Anchor-tag references in ingredient names
-    if (section.list) {
-      for (const item of section.list) {
-        const refSlug = parseAnchorRecipeRef(item.name || '');
-        if (!refSlug || processedSlugs.has(refSlug)) continue;
-        processedSlugs.add(refSlug);
-
-        // Look up the referenced recipe
-        const refRecipe = await Recipe.findOne({ short_name: refSlug })
-          .select('short_name name ingredients nutritionMappings portions')
-          .lean();
-        if (!refRecipe?.nutritionMappings?.length) continue;
-
-        const nutrition = computeRecipeNutritionTotals(
-          refRecipe.ingredients || [], refRecipe.nutritionMappings, 1
-        );
-        results.push({ shortName: refSlug, name: refRecipe.name, nutrition, baseMultiplier: 1 });
-      }
-    }
-  }
-
-  return results;
-}
-
 export const GET: RequestHandler = async ({ params }) => {
  await dbConnect();
  const en = isEnglish(params.recipeLang!);
--- a/src/routes/api/[recipeLang=recipeLang]/json-ld/[name]/+server.ts
+++ b/src/routes/api/[recipeLang=recipeLang]/json-ld/[name]/+server.ts
@@ -2,19 +2,23 @@ import { json, type RequestHandler } from '@sveltejs/kit';
 import { Recipe } from '$models/Recipe';
 import { dbConnect } from '$utils/db';
 import { generateRecipeJsonLd } from '$lib/js/recipeJsonLd';
+import { resolveReferencedNutrition } from '$lib/server/nutritionMatcher';
 import type { RecipeModelType } from '$types/types';
 import { error } from '@sveltejs/kit';

 export const GET: RequestHandler = async ({ params, setHeaders }) => {
  await dbConnect();
-  let recipe = (await Recipe.findOne({ short_name: params.name }).lean()) as unknown as RecipeModelType;
+  let recipe = (await Recipe.findOne({ short_name: params.name })
+    .populate({ path: 'ingredients.baseRecipeRef', select: 'short_name name ingredients nutritionMappings' })
+    .lean()) as unknown as RecipeModelType;

  recipe = JSON.parse(JSON.stringify(recipe));
  if (recipe == null) {
    throw error(404, "Recipe not found");
  }

-  const jsonLd = generateRecipeJsonLd(recipe);
+  const referencedNutrition = await resolveReferencedNutrition(recipe.ingredients || []);
+  const jsonLd = generateRecipeJsonLd(recipe, referencedNutrition);

  // Set appropriate headers for JSON-LD
  setHeaders({