soustack 0.1.2 → 0.2.1

package/dist/index.js

@@ -140,8 +140,8 @@ function flattenInstructions(items) {
 // src/schema.json
 var schema_default = {
   $schema: "http://json-schema.org/draft-07/schema#",
-  $id: "http://soustack.org/schema/v0.1",
-  title: "Soustack Recipe Schema v0.1",
+  $id: "http://soustack.org/schema/v0.2",
+  title: "Soustack Recipe Schema v0.2",
   description: "A portable, scalable, interoperable recipe format.",
   type: "object",
   required: ["name", "ingredients", "instructions"],
@@ -171,8 +171,21 @@ var schema_default = {
       items: { type: "string" }
     },
     image: {
-      type: "string",
-      format: "uri"
+      description: "Recipe-level hero image(s)",
+      anyOf: [
+        {
+          type: "string",
+          format: "uri"
+        },
+        {
+          type: "array",
+          minItems: 1,
+          items: {
+            type: "string",
+            format: "uri"
+          }
+        }
+      ]
     },
     dateAdded: {
       type: "string",
@@ -337,6 +350,11 @@ var schema_default = {
       properties: {
         id: { type: "string" },
         text: { type: "string" },
+        image: {
+          type: "string",
+          format: "uri",
+          description: "Optional image that illustrates this instruction"
+        },
         destination: { type: "string" },
         dependsOn: {
           type: "array",
@@ -1234,6 +1252,40 @@ function smartParseDuration(input) {
   return parseHumanDuration(input);
 }
 
+// src/utils/image.ts
+function normalizeImage(image) {
+  if (!image) {
+    return void 0;
+  }
+  if (typeof image === "string") {
+    const trimmed = image.trim();
+    return trimmed || void 0;
+  }
+  if (Array.isArray(image)) {
+    const urls = image.map((entry) => typeof entry === "string" ? entry.trim() : extractUrl(entry)).filter((url) => typeof url === "string" && Boolean(url));
+    if (urls.length === 0) {
+      return void 0;
+    }
+    if (urls.length === 1) {
+      return urls[0];
+    }
+    return urls;
+  }
+  return extractUrl(image);
+}
+function extractUrl(value) {
+  if (!value || typeof value !== "object") {
+    return void 0;
+  }
+  const record = value;
+  const candidate = typeof record.url === "string" ? record.url : typeof record.contentUrl === "string" ? record.contentUrl : void 0;
+  if (!candidate) {
+    return void 0;
+  }
+  const trimmed = candidate.trim();
+  return trimmed || void 0;
+}
+
 // src/fromSchemaOrg.ts
 function fromSchemaOrg(input) {
   const recipeNode = extractRecipeNode(input);
@@ -1246,13 +1298,12 @@ function fromSchemaOrg(input) {
   const recipeYield = parseYield(recipeNode.recipeYield);
   const tags = collectTags(recipeNode.recipeCuisine, recipeNode.keywords);
   const category = extractFirst(recipeNode.recipeCategory);
-  const image = convertImage(recipeNode.image);
   const source = convertSource(recipeNode);
   const nutrition = recipeNode.nutrition && typeof recipeNode.nutrition === "object" ? recipeNode.nutrition : void 0;
   return {
     name: recipeNode.name.trim(),
     description: recipeNode.description?.trim() || void 0,
-    image,
+    image: normalizeImage(recipeNode.image),
     category,
     tags: tags.length ? tags : void 0,
     source,
@@ -1335,9 +1386,9 @@ function convertInstructions(value) {
       continue;
     }
     if (isHowToStep(entry)) {
-      const text = extractInstructionText(entry);
-      if (text) {
-        result.push(text);
+      const parsed = convertHowToStep(entry);
+      if (parsed) {
+        result.push(parsed);
       }
     }
   }
@@ -1355,9 +1406,9 @@ function extractSectionItems(items = []) {
       continue;
     }
     if (isHowToStep(item)) {
-      const text = extractInstructionText(item);
-      if (text) {
-        result.push(text);
+      const parsed = convertHowToStep(item);
+      if (parsed) {
+        result.push(parsed);
       }
       continue;
     }
@@ -1371,6 +1422,17 @@ function extractInstructionText(value) {
   const text = typeof value.text === "string" ? value.text : value.name;
   return typeof text === "string" ? text.trim() || void 0 : void 0;
 }
+function convertHowToStep(step) {
+  const text = extractInstructionText(step);
+  if (!text) {
+    return void 0;
+  }
+  const normalizedImage = normalizeImage(step.image);
+  if (typeof normalizedImage === "string") {
+    return { text, image: normalizedImage };
+  }
+  return text;
+}
 function isHowToStep(value) {
   return Boolean(value) && typeof value === "object" && value["@type"] === "HowToStep";
 }
@@ -1412,26 +1474,6 @@ function extractFirst(value) {
   const arr = flattenStrings(value);
   return arr.length ? arr[0] : void 0;
 }
-function convertImage(value) {
-  if (!value) return void 0;
-  if (typeof value === "string") {
-    return value;
-  }
-  if (Array.isArray(value)) {
-    for (const item of value) {
-      const url = typeof item === "string" ? item : extractImageUrl(item);
-      if (url) return url;
-    }
-    return void 0;
-  }
-  return extractImageUrl(value);
-}
-function extractImageUrl(value) {
-  if (!value || typeof value !== "object") return void 0;
-  const record = value;
-  const candidate = typeof record.url === "string" ? record.url : typeof record.contentUrl === "string" ? record.contentUrl : void 0;
-  return candidate?.trim() || void 0;
-}
 function convertSource(recipe) {
   const author = extractEntityName(recipe.author);
   const publisher = extractEntityName(recipe.publisher);
@@ -1527,7 +1569,7 @@ function convertInstruction(entry) {
     return createHowToStep(entry);
   }
   if ("subsection" in entry) {
-    const steps = entry.items.map((item) => typeof item === "string" ? createHowToStep(item) : createHowToStep(item.text)).filter((step) => Boolean(step));
+    const steps = entry.items.map((item) => createHowToStep(item)).filter((step) => Boolean(step));
     if (!steps.length) {
       return null;
     }
@@ -1538,18 +1580,34 @@ function convertInstruction(entry) {
     };
   }
   if ("text" in entry) {
-    return createHowToStep(entry.text);
+    return createHowToStep(entry);
   }
   return createHowToStep(String(entry));
 }
-function createHowToStep(text) {
-  if (!text) return null;
-  const trimmed = text.trim();
-  if (!trimmed) return null;
-  return {
+function createHowToStep(entry) {
+  if (!entry) return null;
+  if (typeof entry === "string") {
+    const trimmed2 = entry.trim();
+    if (!trimmed2) {
+      return null;
+    }
+    return {
+      "@type": "HowToStep",
+      text: trimmed2
+    };
+  }
+  const trimmed = entry.text?.trim();
+  if (!trimmed) {
+    return null;
+  }
+  const step = {
     "@type": "HowToStep",
     text: trimmed
   };
+  if (entry.image) {
+    step.image = entry.image;
+  }
+  return step;
 }
 function convertTime2(time) {
   if (!time) {
@@ -1712,7 +1770,7 @@ async function fetchPage(url, options = {}) {
       };
       const response = await resolvedFetch(url, requestInit);
       clearTimeout(timeoutId);
-      if (response && (typeof process === "undefined" || process.env.NODE_ENV !== "test")) {
+      if (response && typeof process !== "undefined" && process.env.NODE_ENV !== "test") {
         try {
           const globalFetch = typeof globalThis !== "undefined" && typeof globalThis.fetch !== "undefined" ? globalThis.fetch : null;
           if (globalFetch) {
@@ -1730,7 +1788,7 @@ async function fetchPage(url, options = {}) {
         throw error;
       }
       const html = await response.text();
-      if (typeof process === "undefined" || process.env.NODE_ENV !== "test") {
+      if (typeof process !== "undefined" && process.env.NODE_ENV !== "test") {
         try {
           const globalFetch = typeof globalThis !== "undefined" && typeof globalThis.fetch !== "undefined" ? globalThis.fetch : null;
           if (globalFetch) {
@@ -1990,14 +2048,30 @@ 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 +2080,52 @@ 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 +2142,10 @@ 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;
@@ -2280,9 +2390,11 @@ function wordToNumber(word) {
 }
 
 exports.extractRecipeFromHTML = extractRecipeFromHTML;
+exports.extractSchemaOrgRecipeFromHTML = extractSchemaOrgRecipeFromHTML;
 exports.formatDuration = formatDuration;
 exports.formatYield = formatYield2;
 exports.fromSchemaOrg = fromSchemaOrg;
+exports.normalizeImage = normalizeImage;
 exports.normalizeIngredientInput = normalizeIngredientInput;
 exports.normalizeYield = normalizeYield;
 exports.parseDuration = parseDuration;