soustack 0.2.3 → 0.3.0

package/dist/scrape.d.mts

@@ -1,10 +1,24 @@
 /**
- * Soustack Recipe Schema v0.2.1
+ * Soustack Recipe Schema v0.3.0
  * A portable, scalable, interoperable recipe format.
  */
 interface SoustackRecipe {
+    /** Document marker for Soustack recipes */
+    '@type'?: 'Recipe';
     /** Optional $schema pointer for profile-aware validation */
     $schema?: string;
+    /** Optional declared validation profile */
+    profile?: string;
+    /** Enabled module identifiers (e.g., "nutrition@1") */
+    modules?: string[];
+    /** Attribution module payload */
+    attribution?: AttributionModule;
+    /** Taxonomy module payload */
+    taxonomy?: TaxonomyModule;
+    /** Media module payload */
+    media?: MediaModule;
+    /** Times module payload */
+    times?: TimesModule;
     /** Unique identifier (slug or UUID) */
     id?: string;
     /** Optional display title */
@@ -190,15 +204,27 @@ interface Alternative {
     dietary?: string[];
 }
 interface NutritionFacts {
-    calories?: string;
-    fatContent?: string;
-    carbohydrateContent?: string;
-    proteinContent?: string;
-    fiberContent?: string;
-    sugarContent?: string;
-    sodiumContent?: string;
-    servingSize?: string;
-    [key: string]: string | number | null | string[] | undefined;
+    calories?: number;
+    protein_g?: number;
+}
+interface AttributionModule {
+    url?: string;
+    author?: string;
+    datePublished?: string;
+}
+interface TaxonomyModule {
+    keywords?: string[];
+    category?: string;
+    cuisine?: string;
+}
+interface MediaModule {
+    images?: string[];
+    videos?: string[];
+}
+interface TimesModule {
+    prepMinutes?: number;
+    cookMinutes?: number;
+    totalMinutes?: number;
 }
 
 interface HowToStep {

package/dist/scrape.d.ts

@@ -1,10 +1,24 @@
 /**
- * Soustack Recipe Schema v0.2.1
+ * Soustack Recipe Schema v0.3.0
  * A portable, scalable, interoperable recipe format.
  */
 interface SoustackRecipe {
+    /** Document marker for Soustack recipes */
+    '@type'?: 'Recipe';
     /** Optional $schema pointer for profile-aware validation */
     $schema?: string;
+    /** Optional declared validation profile */
+    profile?: string;
+    /** Enabled module identifiers (e.g., "nutrition@1") */
+    modules?: string[];
+    /** Attribution module payload */
+    attribution?: AttributionModule;
+    /** Taxonomy module payload */
+    taxonomy?: TaxonomyModule;
+    /** Media module payload */
+    media?: MediaModule;
+    /** Times module payload */
+    times?: TimesModule;
     /** Unique identifier (slug or UUID) */
     id?: string;
     /** Optional display title */
@@ -190,15 +204,27 @@ interface Alternative {
     dietary?: string[];
 }
 interface NutritionFacts {
-    calories?: string;
-    fatContent?: string;
-    carbohydrateContent?: string;
-    proteinContent?: string;
-    fiberContent?: string;
-    sugarContent?: string;
-    sodiumContent?: string;
-    servingSize?: string;
-    [key: string]: string | number | null | string[] | undefined;
+    calories?: number;
+    protein_g?: number;
+}
+interface AttributionModule {
+    url?: string;
+    author?: string;
+    datePublished?: string;
+}
+interface TaxonomyModule {
+    keywords?: string[];
+    category?: string;
+    cuisine?: string;
+}
+interface MediaModule {
+    images?: string[];
+    videos?: string[];
+}
+interface TimesModule {
+    prepMinutes?: number;
+    cookMinutes?: number;
+    totalMinutes?: number;
 }
 
 interface HowToStep {

package/dist/scrape.js

@@ -143,8 +143,22 @@ function fromSchemaOrg(input) {
   const tags = collectTags(recipeNode.recipeCuisine, recipeNode.keywords);
   const category = extractFirst(recipeNode.recipeCategory);
   const source = convertSource(recipeNode);
-  const nutrition = recipeNode.nutrition && typeof recipeNode.nutrition === "object" ? recipeNode.nutrition : void 0;
+  const dateModified = recipeNode.dateModified || void 0;
+  const nutrition = convertNutrition(recipeNode.nutrition);
+  const attribution = convertAttribution(recipeNode);
+  const taxonomy = convertTaxonomy(tags, category, extractFirst(recipeNode.recipeCuisine));
+  const media = convertMedia(recipeNode.image, recipeNode.video);
+  const times = convertTimes(time);
+  const modules = [];
+  if (attribution) modules.push("attribution@1");
+  if (taxonomy) modules.push("taxonomy@1");
+  if (media) modules.push("media@1");
+  if (nutrition) modules.push("nutrition@1");
+  if (times) modules.push("times@1");
   return {
+    "@type": "Recipe",
+    profile: "minimal",
+    modules: modules.sort(),
     name: recipeNode.name.trim(),
     description: recipeNode.description?.trim() || void 0,
     image: normalizeImage(recipeNode.image),
@@ -152,12 +166,16 @@ function fromSchemaOrg(input) {
     tags: tags.length ? tags : void 0,
     source,
     dateAdded: recipeNode.datePublished || void 0,
-    dateModified: recipeNode.dateModified || void 0,
     yield: recipeYield,
     time,
     ingredients,
     instructions,
-    nutrition
+    ...dateModified ? { dateModified } : {},
+    ...nutrition ? { nutrition } : {},
+    ...attribution ? { attribution } : {},
+    ...taxonomy ? { taxonomy } : {},
+    ...media ? { media } : {},
+    ...times ? { times } : {}
   };
 }
 function extractRecipeNode(input) {
@@ -370,6 +388,90 @@ function extractEntityName(value) {
   }
   return void 0;
 }
+function convertAttribution(recipe) {
+  const attribution = {};
+  const url = (recipe.url || recipe.mainEntityOfPage)?.trim();
+  const author = extractEntityName(recipe.author);
+  const datePublished = recipe.datePublished?.trim();
+  if (url) attribution.url = url;
+  if (author) attribution.author = author;
+  if (datePublished) attribution.datePublished = datePublished;
+  return Object.keys(attribution).length ? attribution : void 0;
+}
+function convertTaxonomy(keywords, category, cuisine) {
+  const taxonomy = {};
+  if (keywords.length) taxonomy.keywords = keywords;
+  if (category) taxonomy.category = category;
+  if (cuisine) taxonomy.cuisine = cuisine;
+  return Object.keys(taxonomy).length ? taxonomy : void 0;
+}
+function normalizeMediaList(value) {
+  if (!value) return [];
+  if (typeof value === "string") return [value.trim()].filter(Boolean);
+  if (Array.isArray(value)) {
+    return value.map((item) => typeof item === "string" ? item.trim() : extractMediaUrl(item)).filter((entry) => Boolean(entry?.length));
+  }
+  const url = extractMediaUrl(value);
+  return url ? [url] : [];
+}
+function extractMediaUrl(value) {
+  if (value && typeof value === "object" && "url" in value && typeof value.url === "string") {
+    const trimmed = value.url.trim();
+    return trimmed || void 0;
+  }
+  return void 0;
+}
+function convertMedia(image, video) {
+  const normalizedImage = normalizeImage(image);
+  const images = normalizedImage ? Array.isArray(normalizedImage) ? normalizedImage : [normalizedImage] : [];
+  const videos = normalizeMediaList(video);
+  const media = {};
+  if (images.length) media.images = images;
+  if (videos.length) media.videos = videos;
+  return Object.keys(media).length ? media : void 0;
+}
+function convertTimes(time) {
+  if (!time) return void 0;
+  const times = {};
+  if (typeof time.prep === "number") times.prepMinutes = time.prep;
+  if (typeof time.active === "number") times.cookMinutes = time.active;
+  if (typeof time.total === "number") times.totalMinutes = time.total;
+  return Object.keys(times).length ? times : void 0;
+}
+function convertNutrition(nutrition) {
+  if (!nutrition || typeof nutrition !== "object") {
+    return void 0;
+  }
+  const result = {};
+  let hasData = false;
+  if ("calories" in nutrition) {
+    const calories = nutrition.calories;
+    if (typeof calories === "number") {
+      result.calories = calories;
+      hasData = true;
+    } else if (typeof calories === "string") {
+      const parsed = parseFloat(calories.replace(/[^\d.-]/g, ""));
+      if (!isNaN(parsed)) {
+        result.calories = parsed;
+        hasData = true;
+      }
+    }
+  }
+  if ("proteinContent" in nutrition || "protein_g" in nutrition) {
+    const protein = nutrition.proteinContent || nutrition.protein_g;
+    if (typeof protein === "number") {
+      result.protein_g = protein;
+      hasData = true;
+    } else if (typeof protein === "string") {
+      const parsed = parseFloat(protein.replace(/[^\d.-]/g, ""));
+      if (!isNaN(parsed)) {
+        result.protein_g = parsed;
+        hasData = true;
+      }
+    }
+  }
+  return hasData ? result : void 0;
+}
 
 // src/scraper/fetch.ts
 var DEFAULT_USER_AGENTS = [