soustack 0.1.2 → 0.1.3

package/dist/index.d.mts

@@ -245,7 +245,7 @@ declare function validateRecipe(data: any): data is Recipe;
 
 declare function fromSchemaOrg(input: unknown): Recipe | null;
 
-interface SchemaOrgRecipe {
+interface SchemaOrgRecipe$1 {
     '@context'?: string;
     '@type'?: string | string[];
     name: string;
@@ -270,7 +270,7 @@ interface SchemaOrgRecipe {
     '@graph'?: unknown;
 }
 type SchemaOrgIngredientList = string | string[];
-type SchemaOrgInstructionList = string | HowToStep | HowToSection | Array<string | HowToStep | HowToSection>;
+type SchemaOrgInstructionList = string | HowToStep$1 | HowToSection | Array<string | HowToStep$1 | HowToSection>;
 interface SchemaOrgImageObject {
     '@type'?: string;
     url?: string;
@@ -282,7 +282,7 @@ interface SchemaOrgYield {
     unit?: string;
     description?: string;
 }
-interface HowToStep {
+interface HowToStep$1 {
     '@type': 'HowToStep';
     text?: string;
     name?: string;
@@ -292,7 +292,7 @@ interface HowToStep {
 interface HowToSection {
     '@type': 'HowToSection';
     name: string;
-    itemListElement: Array<string | HowToStep | HowToSection>;
+    itemListElement: Array<string | HowToStep$1 | HowToSection>;
 }
 interface SchemaOrgPersonOrOrganization {
     '@type'?: 'Person' | 'Organization';
@@ -303,8 +303,30 @@ interface NutritionInformation {
     [key: string]: string | number | null | undefined;
 }
 
-declare function toSchemaOrg(recipe: Recipe): SchemaOrgRecipe;
+declare function toSchemaOrg(recipe: Recipe): SchemaOrgRecipe$1;
 
+interface HowToStep {
+    '@type'?: 'HowToStep' | 'HowToSection' | string;
+    name?: string;
+    text?: string;
+    itemListElement?: Array<string | HowToStep>;
+}
+interface SchemaOrgRecipe {
+    '@type': string | string[];
+    name?: string;
+    description?: string;
+    image?: string | string[];
+    recipeIngredient?: string[];
+    recipeInstructions?: Array<string | HowToStep>;
+    recipeYield?: string | number;
+    prepTime?: string;
+    cookTime?: string;
+    totalTime?: string;
+    author?: unknown;
+    datePublished?: string;
+    aggregateRating?: unknown;
+    [key: string]: unknown;
+}
 interface FetchRequestInit {
     headers?: Record<string, string>;
     signal?: AbortSignal;
@@ -357,6 +379,33 @@ declare function scrapeRecipe(url: string, options?: ScrapeRecipeOptions): Promi
  * @throws Error if no recipe is found
  */
 declare function extractRecipeFromHTML(html: string): Recipe;
+/**
+ * Extract Schema.org recipe data from HTML string (browser-compatible).
+ *
+ * Returns the raw Schema.org recipe object, which can then be converted
+ * to Soustack format using fromSchemaOrg(). This gives you access to the
+ * original Schema.org data for inspection, debugging, or custom transformations.
+ *
+ * @param html - HTML string containing Schema.org recipe data
+ * @returns Schema.org recipe object, or null if not found
+ *
+ * @example
+ * ```ts
+ * // In browser:
+ * const response = await fetch('https://example.com/recipe');
+ * const html = await response.text();
+ * const schemaOrgRecipe = extractSchemaOrgRecipeFromHTML(html);
+ *
+ * if (schemaOrgRecipe) {
+ *   // Inspect or modify Schema.org data before converting
+ *   console.log('Found recipe:', schemaOrgRecipe.name);
+ *
+ *   // Convert to Soustack format
+ *   const soustackRecipe = fromSchemaOrg(schemaOrgRecipe);
+ * }
+ * ```
+ */
+declare function extractSchemaOrgRecipeFromHTML(html: string): SchemaOrgRecipe | null;
 
 declare function normalizeIngredientInput(input: string): string;
 declare function parseIngredient(text: string): ParsedIngredient;
@@ -376,4 +425,4 @@ declare function normalizeYield(text: string): string;
 declare function parseYield(text: string): ParsedYield | null;
 declare function formatYield(value: ParsedYield): string;
 
-export { type Alternative, type ComputedIngredient, type ComputedInstruction, type ComputedRecipe, type Equipment, type FrozenStorageMethod, type Ingredient, type IngredientItem, type IngredientSubsection, type Instruction, type InstructionItem, type InstructionSubsection, type MakeAheadComponent, type NutritionFacts, type ParsedIngredient, type ParsedYield, type Quantity, type Recipe, type Scaling, type ScalingBakersPercentage, type ScalingBase, type ScalingDiscrete, type ScalingFixed, type ScalingLinear, type ScalingProportional, type SimpleTime, type Source, type StepTiming, type Storage, type StorageMethod, type StructuredTime, type Substitution, type Time, type Yield, extractRecipeFromHTML, formatDuration, formatYield, fromSchemaOrg, normalizeIngredientInput, normalizeYield, parseDuration, parseHumanDuration, parseIngredient, parseIngredientLine, parseIngredients, parseYield, scaleRecipe, scrapeRecipe, smartParseDuration, toSchemaOrg, validateRecipe };
+export { type Alternative, type ComputedIngredient, type ComputedInstruction, type ComputedRecipe, type Equipment, type FrozenStorageMethod, type Ingredient, type IngredientItem, type IngredientSubsection, type Instruction, type InstructionItem, type InstructionSubsection, type MakeAheadComponent, type NutritionFacts, type ParsedIngredient, type ParsedYield, type Quantity, type Recipe, type Scaling, type ScalingBakersPercentage, type ScalingBase, type ScalingDiscrete, type ScalingFixed, type ScalingLinear, type ScalingProportional, type SchemaOrgRecipe, type SimpleTime, type Source, type StepTiming, type Storage, type StorageMethod, type StructuredTime, type Substitution, type Time, type Yield, extractRecipeFromHTML, extractSchemaOrgRecipeFromHTML, formatDuration, formatYield, fromSchemaOrg, normalizeIngredientInput, normalizeYield, parseDuration, parseHumanDuration, parseIngredient, parseIngredientLine, parseIngredients, parseYield, scaleRecipe, scrapeRecipe, smartParseDuration, toSchemaOrg, validateRecipe };

package/dist/index.d.ts

@@ -245,7 +245,7 @@ declare function validateRecipe(data: any): data is Recipe;
 
 declare function fromSchemaOrg(input: unknown): Recipe | null;
 
-interface SchemaOrgRecipe {
+interface SchemaOrgRecipe$1 {
     '@context'?: string;
     '@type'?: string | string[];
     name: string;
@@ -270,7 +270,7 @@ interface SchemaOrgRecipe {
     '@graph'?: unknown;
 }
 type SchemaOrgIngredientList = string | string[];
-type SchemaOrgInstructionList = string | HowToStep | HowToSection | Array<string | HowToStep | HowToSection>;
+type SchemaOrgInstructionList = string | HowToStep$1 | HowToSection | Array<string | HowToStep$1 | HowToSection>;
 interface SchemaOrgImageObject {
     '@type'?: string;
     url?: string;
@@ -282,7 +282,7 @@ interface SchemaOrgYield {
     unit?: string;
     description?: string;
 }
-interface HowToStep {
+interface HowToStep$1 {
     '@type': 'HowToStep';
     text?: string;
     name?: string;
@@ -292,7 +292,7 @@ interface HowToStep {
 interface HowToSection {
     '@type': 'HowToSection';
     name: string;
-    itemListElement: Array<string | HowToStep | HowToSection>;
+    itemListElement: Array<string | HowToStep$1 | HowToSection>;
 }
 interface SchemaOrgPersonOrOrganization {
     '@type'?: 'Person' | 'Organization';
@@ -303,8 +303,30 @@ interface NutritionInformation {
     [key: string]: string | number | null | undefined;
 }
 
-declare function toSchemaOrg(recipe: Recipe): SchemaOrgRecipe;
+declare function toSchemaOrg(recipe: Recipe): SchemaOrgRecipe$1;
 
+interface HowToStep {
+    '@type'?: 'HowToStep' | 'HowToSection' | string;
+    name?: string;
+    text?: string;
+    itemListElement?: Array<string | HowToStep>;
+}
+interface SchemaOrgRecipe {
+    '@type': string | string[];
+    name?: string;
+    description?: string;
+    image?: string | string[];
+    recipeIngredient?: string[];
+    recipeInstructions?: Array<string | HowToStep>;
+    recipeYield?: string | number;
+    prepTime?: string;
+    cookTime?: string;
+    totalTime?: string;
+    author?: unknown;
+    datePublished?: string;
+    aggregateRating?: unknown;
+    [key: string]: unknown;
+}
 interface FetchRequestInit {
     headers?: Record<string, string>;
     signal?: AbortSignal;
@@ -357,6 +379,33 @@ declare function scrapeRecipe(url: string, options?: ScrapeRecipeOptions): Promi
  * @throws Error if no recipe is found
  */
 declare function extractRecipeFromHTML(html: string): Recipe;
+/**
+ * Extract Schema.org recipe data from HTML string (browser-compatible).
+ *
+ * Returns the raw Schema.org recipe object, which can then be converted
+ * to Soustack format using fromSchemaOrg(). This gives you access to the
+ * original Schema.org data for inspection, debugging, or custom transformations.
+ *
+ * @param html - HTML string containing Schema.org recipe data
+ * @returns Schema.org recipe object, or null if not found
+ *
+ * @example
+ * ```ts
+ * // In browser:
+ * const response = await fetch('https://example.com/recipe');
+ * const html = await response.text();
+ * const schemaOrgRecipe = extractSchemaOrgRecipeFromHTML(html);
+ *
+ * if (schemaOrgRecipe) {
+ *   // Inspect or modify Schema.org data before converting
+ *   console.log('Found recipe:', schemaOrgRecipe.name);
+ *
+ *   // Convert to Soustack format
+ *   const soustackRecipe = fromSchemaOrg(schemaOrgRecipe);
+ * }
+ * ```
+ */
+declare function extractSchemaOrgRecipeFromHTML(html: string): SchemaOrgRecipe | null;
 
 declare function normalizeIngredientInput(input: string): string;
 declare function parseIngredient(text: string): ParsedIngredient;
@@ -376,4 +425,4 @@ declare function normalizeYield(text: string): string;
 declare function parseYield(text: string): ParsedYield | null;
 declare function formatYield(value: ParsedYield): string;
 
-export { type Alternative, type ComputedIngredient, type ComputedInstruction, type ComputedRecipe, type Equipment, type FrozenStorageMethod, type Ingredient, type IngredientItem, type IngredientSubsection, type Instruction, type InstructionItem, type InstructionSubsection, type MakeAheadComponent, type NutritionFacts, type ParsedIngredient, type ParsedYield, type Quantity, type Recipe, type Scaling, type ScalingBakersPercentage, type ScalingBase, type ScalingDiscrete, type ScalingFixed, type ScalingLinear, type ScalingProportional, type SimpleTime, type Source, type StepTiming, type Storage, type StorageMethod, type StructuredTime, type Substitution, type Time, type Yield, extractRecipeFromHTML, formatDuration, formatYield, fromSchemaOrg, normalizeIngredientInput, normalizeYield, parseDuration, parseHumanDuration, parseIngredient, parseIngredientLine, parseIngredients, parseYield, scaleRecipe, scrapeRecipe, smartParseDuration, toSchemaOrg, validateRecipe };
+export { type Alternative, type ComputedIngredient, type ComputedInstruction, type ComputedRecipe, type Equipment, type FrozenStorageMethod, type Ingredient, type IngredientItem, type IngredientSubsection, type Instruction, type InstructionItem, type InstructionSubsection, type MakeAheadComponent, type NutritionFacts, type ParsedIngredient, type ParsedYield, type Quantity, type Recipe, type Scaling, type ScalingBakersPercentage, type ScalingBase, type ScalingDiscrete, type ScalingFixed, type ScalingLinear, type ScalingProportional, type SchemaOrgRecipe, type SimpleTime, type Source, type StepTiming, type Storage, type StorageMethod, type StructuredTime, type Substitution, type Time, type Yield, extractRecipeFromHTML, extractSchemaOrgRecipeFromHTML, formatDuration, formatYield, fromSchemaOrg, normalizeIngredientInput, normalizeYield, parseDuration, parseHumanDuration, parseIngredient, parseIngredientLine, parseIngredients, parseYield, scaleRecipe, scrapeRecipe, smartParseDuration, toSchemaOrg, validateRecipe };

package/dist/index.js

@@ -1990,14 +1990,30 @@ function extractRecipe(html) {
     return extractRecipeBrowser(html);
   }
   const jsonLdRecipe = extractJsonLd(html);
-  fetch("http://127.0.0.1:7243/ingest/7225c3b5-9ac2-4c94-b561-807ca9003b66", { method: "POST", headers: { "Content-Type": "application/json" }, body: JSON.stringify({ location: "scraper/extractors/index.ts:6", message: "JSON-LD extraction result", data: { hasJsonLd: !!jsonLdRecipe }, timestamp: Date.now(), sessionId: "debug-session", runId: "run1", hypothesisId: "C,D" }) }).catch(() => {
-  });
+  if (typeof process === "undefined" || process.env.NODE_ENV !== "test") {
+    try {
+      const globalFetch = typeof globalThis !== "undefined" && typeof globalThis.fetch !== "undefined" ? globalThis.fetch : null;
+      if (globalFetch) {
+        globalFetch("http://127.0.0.1:7243/ingest/7225c3b5-9ac2-4c94-b561-807ca9003b66", { method: "POST", headers: { "Content-Type": "application/json" }, body: JSON.stringify({ location: "scraper/extractors/index.ts:6", message: "JSON-LD extraction result", data: { hasJsonLd: !!jsonLdRecipe }, timestamp: Date.now(), sessionId: "debug-session", runId: "run1", hypothesisId: "C,D" }) }).catch(() => {
+        });
+      }
+    } catch {
+    }
+  }
   if (jsonLdRecipe) {
     return { recipe: jsonLdRecipe, source: "jsonld" };
   }
   const microdataRecipe = extractMicrodata(html);
-  fetch("http://127.0.0.1:7243/ingest/7225c3b5-9ac2-4c94-b561-807ca9003b66", { method: "POST", headers: { "Content-Type": "application/json" }, body: JSON.stringify({ location: "scraper/extractors/index.ts:12", message: "Microdata extraction result", data: { hasMicrodata: !!microdataRecipe }, timestamp: Date.now(), sessionId: "debug-session", runId: "run1", hypothesisId: "D" }) }).catch(() => {
-  });
+  if (typeof process === "undefined" || process.env.NODE_ENV !== "test") {
+    try {
+      const globalFetch = typeof globalThis !== "undefined" && typeof globalThis.fetch !== "undefined" ? globalThis.fetch : null;
+      if (globalFetch) {
+        globalFetch("http://127.0.0.1:7243/ingest/7225c3b5-9ac2-4c94-b561-807ca9003b66", { method: "POST", headers: { "Content-Type": "application/json" }, body: JSON.stringify({ location: "scraper/extractors/index.ts:12", message: "Microdata extraction result", data: { hasMicrodata: !!microdataRecipe }, timestamp: Date.now(), sessionId: "debug-session", runId: "run1", hypothesisId: "D" }) }).catch(() => {
+        });
+      }
+    } catch {
+    }
+  }
   if (microdataRecipe) {
     return { recipe: microdataRecipe, source: "microdata" };
   }
@@ -2006,20 +2022,52 @@ function extractRecipe(html) {
 
 // src/scraper/index.ts
 async function scrapeRecipe(url, options = {}) {
-  fetch("http://127.0.0.1:7243/ingest/7225c3b5-9ac2-4c94-b561-807ca9003b66", { method: "POST", headers: { "Content-Type": "application/json" }, body: JSON.stringify({ location: "scraper/index.ts:7", message: "scrapeRecipe entry", data: { url, hasOptions: !!options }, timestamp: Date.now(), sessionId: "debug-session", runId: "run1", hypothesisId: "A,B,C,D,E" }) }).catch(() => {
-  });
+  if (typeof process === "undefined" || process.env.NODE_ENV !== "test") {
+    try {
+      const globalFetch = typeof globalThis !== "undefined" && typeof globalThis.fetch !== "undefined" ? globalThis.fetch : null;
+      if (globalFetch) {
+        globalFetch("http://127.0.0.1:7243/ingest/7225c3b5-9ac2-4c94-b561-807ca9003b66", { method: "POST", headers: { "Content-Type": "application/json" }, body: JSON.stringify({ location: "scraper/index.ts:7", message: "scrapeRecipe entry", data: { url, hasOptions: !!options }, timestamp: Date.now(), sessionId: "debug-session", runId: "run1", hypothesisId: "A,B,C,D,E" }) }).catch(() => {
+        });
+      }
+    } catch {
+    }
+  }
   const html = await fetchPage(url, options);
-  fetch("http://127.0.0.1:7243/ingest/7225c3b5-9ac2-4c94-b561-807ca9003b66", { method: "POST", headers: { "Content-Type": "application/json" }, body: JSON.stringify({ location: "scraper/index.ts:9", message: "HTML fetched", data: { htmlLength: html?.length, htmlPreview: html?.substring(0, 200) }, timestamp: Date.now(), sessionId: "debug-session", runId: "run1", hypothesisId: "B" }) }).catch(() => {
-  });
+  if (typeof process === "undefined" || process.env.NODE_ENV !== "test") {
+    try {
+      const globalFetch = typeof globalThis !== "undefined" && typeof globalThis.fetch !== "undefined" ? globalThis.fetch : null;
+      if (globalFetch) {
+        globalFetch("http://127.0.0.1:7243/ingest/7225c3b5-9ac2-4c94-b561-807ca9003b66", { method: "POST", headers: { "Content-Type": "application/json" }, body: JSON.stringify({ location: "scraper/index.ts:9", message: "HTML fetched", data: { htmlLength: html?.length, htmlPreview: html?.substring(0, 200) }, timestamp: Date.now(), sessionId: "debug-session", runId: "run1", hypothesisId: "B" }) }).catch(() => {
+        });
+      }
+    } catch {
+    }
+  }
   const { recipe } = extractRecipe(html);
-  fetch("http://127.0.0.1:7243/ingest/7225c3b5-9ac2-4c94-b561-807ca9003b66", { method: "POST", headers: { "Content-Type": "application/json" }, body: JSON.stringify({ location: "scraper/index.ts:11", message: "extractRecipe result", data: { hasRecipe: !!recipe, recipeType: recipe?.["@type"], recipeName: recipe?.name }, timestamp: Date.now(), sessionId: "debug-session", runId: "run1", hypothesisId: "A,C,D" }) }).catch(() => {
-  });
+  if (typeof process === "undefined" || process.env.NODE_ENV !== "test") {
+    try {
+      const globalFetch = typeof globalThis !== "undefined" && typeof globalThis.fetch !== "undefined" ? globalThis.fetch : null;
+      if (globalFetch) {
+        globalFetch("http://127.0.0.1:7243/ingest/7225c3b5-9ac2-4c94-b561-807ca9003b66", { method: "POST", headers: { "Content-Type": "application/json" }, body: JSON.stringify({ location: "scraper/index.ts:11", message: "extractRecipe result", data: { hasRecipe: !!recipe, recipeType: recipe?.["@type"], recipeName: recipe?.name }, timestamp: Date.now(), sessionId: "debug-session", runId: "run1", hypothesisId: "A,C,D" }) }).catch(() => {
+        });
+      }
+    } catch {
+    }
+  }
   if (!recipe) {
     throw new Error("No Schema.org recipe data found in page");
   }
   const soustackRecipe = fromSchemaOrg(recipe);
-  fetch("http://127.0.0.1:7243/ingest/7225c3b5-9ac2-4c94-b561-807ca9003b66", { method: "POST", headers: { "Content-Type": "application/json" }, body: JSON.stringify({ location: "scraper/index.ts:17", message: "fromSchemaOrg result", data: { hasSoustackRecipe: !!soustackRecipe, soustackRecipeName: soustackRecipe?.name }, timestamp: Date.now(), sessionId: "debug-session", runId: "run1", hypothesisId: "A" }) }).catch(() => {
-  });
+  if (typeof process === "undefined" || process.env.NODE_ENV !== "test") {
+    try {
+      const globalFetch = typeof globalThis !== "undefined" && typeof globalThis.fetch !== "undefined" ? globalThis.fetch : null;
+      if (globalFetch) {
+        globalFetch("http://127.0.0.1:7243/ingest/7225c3b5-9ac2-4c94-b561-807ca9003b66", { method: "POST", headers: { "Content-Type": "application/json" }, body: JSON.stringify({ location: "scraper/index.ts:17", message: "fromSchemaOrg result", data: { hasSoustackRecipe: !!soustackRecipe, soustackRecipeName: soustackRecipe?.name }, timestamp: Date.now(), sessionId: "debug-session", runId: "run1", hypothesisId: "A" }) }).catch(() => {
+        });
+      }
+    } catch {
+    }
+  }
   if (!soustackRecipe) {
     throw new Error("Schema.org data did not include a valid recipe");
   }
@@ -2036,6 +2084,10 @@ function extractRecipeFromHTML(html) {
   }
   return soustackRecipe;
 }
+function extractSchemaOrgRecipeFromHTML(html) {
+  const { recipe } = extractRecipe(html);
+  return recipe;
+}
 
 // src/parsers/yield.ts
 var RANGE_PATTERN = /^(\d+)(?:\s*(?:[-–—]|to)\s*)(\d+)\s+(.+)$/i;
@@ -2280,6 +2332,7 @@ function wordToNumber(word) {
 }
 
 exports.extractRecipeFromHTML = extractRecipeFromHTML;
+exports.extractSchemaOrgRecipeFromHTML = extractSchemaOrgRecipeFromHTML;
 exports.formatDuration = formatDuration;
 exports.formatYield = formatYield2;
 exports.fromSchemaOrg = fromSchemaOrg;