recipe-scrapers-js 0.1.0-alpha.6 → 0.1.0

package/dist/index.js

@@ -25,8 +25,7 @@ function isString(value) {
 */
 function getHostName(value) {
 	try {
-		const url = new URL(value.replace("www.", ""));
-		return url.host;
+		return new URL(value.replace("www.", "")).host;
 	} catch {
 		throw new Error(`Invalid URL: ${value}`);
 	}
@@ -139,8 +138,7 @@ function splitToList(value, separator) {
 * @TODO Implement [Temporal.Duration](https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Temporal/Duration) once it lands.
 */
 function parseMinutes(value) {
-	const duration = parse(value);
-	const totalSeconds = toSeconds(duration);
+	const totalSeconds = toSeconds(parse(value));
 	return Math.round(totalSeconds / 60);
 }
 
@@ -181,8 +179,7 @@ function scoreSentenceSimilarity(first, second) {
 	const bigrams = (s) => new Set(Array.from({ length: s.length - 1 }, (_, i) => s.slice(i, i + 2)));
 	const firstBigrams = bigrams(first);
 	const secondBigrams = bigrams(second);
-	const intersectionSize = [...firstBigrams].filter((b) => secondBigrams.has(b)).length;
-	return 2 * intersectionSize / (firstBigrams.size + secondBigrams.size);
+	return 2 * [...firstBigrams].filter((b) => secondBigrams.has(b)).length / (firstBigrams.size + secondBigrams.size);
 }
 function bestMatch(testString, targetStrings) {
 	if (targetStrings.length === 0) throw new Error("targetStrings cannot be empty");
@@ -227,8 +224,7 @@ function groupIngredients($, ingredientsList, headingSelector, itemSelector) {
 	for (const el of elements) {
 		const $el = $(el);
 		if ($el.is(groupNameSelector)) {
-			const headingText = normalizeString($el.text());
-			currentHeading = headingText || DEFAULT_INGREDIENTS_GROUP_NAME;
+			currentHeading = normalizeString($el.text()) || DEFAULT_INGREDIENTS_GROUP_NAME;
 			if (!groupings.has(currentHeading)) groupings.set(currentHeading, /* @__PURE__ */ new Set());
 		} else if ($el.is(ingredientSelector)) {
 			const text = normalizeString($el.text());
@@ -383,8 +379,7 @@ const extractValueFromElement = (element) => {
 	return element.text().trim();
 };
 const extractSchemaType = (itemType) => {
-	const typeMatch = itemType.match(/schema\.org\/(\w+)/);
-	return typeMatch?.[1];
+	return itemType.match(/schema\.org\/(\w+)/)?.[1];
 };
 /**
 * Extracts microdata from HTML elements using itemtype and itemprop attributes
@@ -395,8 +390,7 @@ const extractSchemaType = (itemType) => {
 */
 function extractMicrodata($, selector) {
 	const results = [];
-	const elements = $(selector);
-	elements.each((_, el) => {
+	$(selector).each((_, el) => {
 		const $element = $(el);
 		const itemType = $element.attr("itemtype");
 		const rootObject = {};
@@ -406,13 +400,11 @@ function extractMicrodata($, selector) {
 		}
 		const allProps = $element.find("[itemprop]").addBack("[itemprop]");
 		const nestedItemTypes = $element.find("[itemtype]");
-		const rootLevelProps = allProps.filter((_$1, propEl) => {
+		allProps.filter((_$1, propEl) => {
 			const $prop = $(propEl);
 			if ($prop.attr("itemtype")) return true;
-			const isInsideNestedType = nestedItemTypes.toArray().some((nestedEl) => $(nestedEl).find($prop).length > 0);
-			return !isInsideNestedType;
-		});
-		rootLevelProps.each((_$1, propEl) => {
+			return !nestedItemTypes.toArray().some((nestedEl) => $(nestedEl).find($prop).length > 0);
+		}).each((_$1, propEl) => {
 			const $prop = $(propEl);
 			const propName = $prop.attr("itemprop");
 			if (!propName) return;
@@ -497,8 +489,7 @@ function parseYields(element) {
 		const splitMatch = serveText.match(SERVE_REGEX_TO);
 		if (splitMatch && splitMatch.index !== void 0) serveText = serveText.slice(splitMatch.index + splitMatch[0].length).trim();
 	}
-	const match = serveText.match(SERVE_REGEX_NUMBER);
-	const matched = match?.groups?.items || "0";
+	const matched = serveText.match(SERVE_REGEX_NUMBER)?.groups?.items || "0";
 	const serveTextLower = serveText.toLowerCase();
 	let bestMatch$1 = null;
 	let bestMatchLength = 0;
@@ -535,8 +526,7 @@ function isSchemaOrgData(obj) {
 }
 function isThingType(obj, type) {
 	if (!isBaseType(obj)) return false;
-	const thingType = Array.isArray(obj["@type"]) ? obj["@type"][0] : obj["@type"];
-	return thingType === type;
+	return (Array.isArray(obj["@type"]) ? obj["@type"][0] : obj["@type"]) === type;
 }
 function isAggregateRating(obj) {
 	return isThingType(obj, "AggregateRating");
@@ -645,7 +635,6 @@ var SchemaOrgPlugin = class SchemaOrgPlugin extends ExtractorPlugin {
 	pickFromObject(obj, props) {
 		if (!isPlainObject(obj)) return void 0;
 		for (const prop of props) if (isString(obj[prop])) return obj[prop];
-		return void 0;
 	}
 	getSchemaTextValue(value, props = [
 		"textValue",
@@ -714,10 +703,7 @@ var SchemaOrgPlugin = class SchemaOrgPlugin extends ExtractorPlugin {
 			return value;
 		}
 		if (isString(value)) return parseMinutes(value);
-		if (isBaseType(value) && "maxValue" in value) {
-			const maxValue = this.getSchemaTextValue(value.maxValue);
-			return parseMinutes(maxValue);
-		}
+		if (isBaseType(value) && "maxValue" in value) return parseMinutes(this.getSchemaTextValue(value.maxValue));
 		return null;
 	}
 	parseInstructions(value) {
@@ -921,8 +907,7 @@ var RecipeExtractor = class RecipeExtractor {
 		this.logger.debug(`Extracting field: ${field}`);
 		for (const plugin of this.plugins) {
 			const pluginLogger = new Logger(this.getContext(plugin.name), this.options.logLevel);
-			const isSupported = plugin.supports(field);
-			if (isSupported && !isDefined(result)) try {
+			if (plugin.supports(field) && !isDefined(result)) try {
 				result = await plugin.extract(field);
 			} catch (err) {
 				if (err instanceof ExtractionFailedException) pluginLogger.verbose(err.message);
@@ -964,9 +949,7 @@ var AbstractScraper = class {
 		const { extraExtractors = [], extraPostProcessors = [], logLevel = LogLevel.WARN } = options;
 		this.logger = new Logger(this.constructor.name, logLevel);
 		this.$ = cheerio.load(html);
-		const baseExtractors = [new OpenGraphPlugin(this.$), new SchemaOrgPlugin(this.$, logLevel)];
-		const basePostProcessors = [new HtmlStripperPlugin()];
-		this.pluginManager = new PluginManager(baseExtractors, basePostProcessors, extraExtractors, extraPostProcessors);
+		this.pluginManager = new PluginManager([new OpenGraphPlugin(this.$), new SchemaOrgPlugin(this.$, logLevel)], [new HtmlStripperPlugin()], extraExtractors, extraPostProcessors);
 		this.recipeExtractor = new RecipeExtractor(this.pluginManager.getExtractors(), this.constructor.name, { logLevel });
 	}
 	/**
@@ -1104,7 +1087,8 @@ const recipeDataSchema = z.object({
 	recipeTimeNote: z.string().optional(),
 	ingredientGroups: z.array(recipeIngredientGroupSchema),
 	headnote: z.string().optional(),
-	instructions: z.array(recipeInstructionSchema)
+	instructions: z.array(recipeInstructionSchema),
+	metaData: z.object({ fields: z.object({ photo: z.object({ url: z.url() }) }) })
 });
 const pagePropsDataSchema = z.object({ props: z.object({ pageProps: z.object({ data: recipeDataSchema }) }) });
 var AmericasTestKitchen = class extends AbstractScraper {
@@ -1113,6 +1097,7 @@ var AmericasTestKitchen = class extends AbstractScraper {
 		return "americastestkitchen.com";
 	}
 	extractors = {
+		image: this.image.bind(this),
 		ingredients: this.ingredients.bind(this),
 		instructions: this.instructions.bind(this),
 		siteName: this.siteName.bind(this)
@@ -1120,6 +1105,14 @@ var AmericasTestKitchen = class extends AbstractScraper {
 	siteName() {
 		return "America's Test Kitchen";
 	}
+	image(prevValue) {
+		const data = this.getRecipeData();
+		if (!data) {
+			if (prevValue) return prevValue;
+			throw new Error("Failed to extract image");
+		}
+		return data.metaData.fields.photo.url;
+	}
 	ingredients(prevValue) {
 		let ingredients = this.parseIngredients();
 		if (!ingredients) ingredients = this.parseHtmlIngredients(prevValue);
@@ -1142,23 +1135,18 @@ var AmericasTestKitchen = class extends AbstractScraper {
 	parseHtmlIngredients(prevValue) {
 		const headingSelector = "[class*=\"RecipeIngredientGroups_group\"] > span";
 		const ingredientSelector = "[class*=\"RecipeIngredient\"] label";
-		if (isList(prevValue) && prevValue.size > 0) {
-			const result = groupIngredients(this.$, prevValue, headingSelector, ingredientSelector);
-			return result;
-		}
+		if (isList(prevValue) && prevValue.size > 0) return groupIngredients(this.$, prevValue, headingSelector, ingredientSelector);
 		return null;
 	}
 	getRecipeData() {
 		if (this.data === null) {
-			const jsonElement = this.$("script[type=\"application/json\"]");
-			const jsonString = jsonElement.html();
+			const jsonString = this.$("script[type=\"application/json\"]").html();
 			if (!jsonString) {
 				this.logger.warn("Could not find JSON data script tag");
 				return null;
 			}
 			try {
-				const parsed = pagePropsDataSchema.parse(JSON.parse(jsonString));
-				this.data = parsed.props.pageProps.data;
+				this.data = pagePropsDataSchema.parse(JSON.parse(jsonString)).props.pageProps.data;
 			} catch (error) {
 				this.logger.error("Failed to parse JSON data:", error);
 				return null;
@@ -1208,10 +1196,7 @@ var BBCGoodFood = class extends AbstractScraper {
 	ingredients(prevValue) {
 		const headingSelector = ".recipe__ingredients h3";
 		const ingredientSelector = ".recipe__ingredients li";
-		if (isList(prevValue) && prevValue.size > 0) {
-			const result = groupIngredients(this.$, prevValue, headingSelector, ingredientSelector);
-			return result;
-		}
+		if (isList(prevValue) && prevValue.size > 0) return groupIngredients(this.$, prevValue, headingSelector, ingredientSelector);
 		throw new Error("No ingredients found to group");
 	}
 };
@@ -1233,8 +1218,7 @@ var Epicurious = class extends AbstractScraper {
 	}
 	extractors = { author: this.author.bind(this) };
 	author() {
-		const author = this.$("a[itemprop=\"author\"]").text().trim();
-		return author;
+		return this.$("a[itemprop=\"author\"]").text().trim();
 	}
 };
 
@@ -1248,10 +1232,7 @@ var NYTimes = class extends AbstractScraper {
 	ingredients(prevValue) {
 		const headingSelector = "h3[class*=\"ingredientgroup_name\"]";
 		const ingredientSelector = "li[class*=\"ingredient\"]";
-		if (isList(prevValue) && prevValue.size > 0) {
-			const result = groupIngredients(this.$, prevValue, headingSelector, ingredientSelector);
-			return result;
-		}
+		if (isList(prevValue) && prevValue.size > 0) return groupIngredients(this.$, prevValue, headingSelector, ingredientSelector);
 		throw new Error("No ingredients found to group");
 	}
 };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "recipe-scrapers-js",
-  "version": "0.1.0-alpha.6",
+  "version": "0.1.0",
   "license": "MIT",
   "description": "A recipe scrapers library",
   "author": {
@@ -39,18 +39,18 @@
     "prepublishOnly": "bun run lint && bun run build"
   },
   "peerDependencies": {
-    "cheerio": "^1.1.0",
-    "zod": "^3.25.74"
+    "cheerio": "^1.1.2",
+    "zod": "^4.1.12"
   },
   "dependencies": {
-    "iso8601-duration": "^2.1.2",
+    "iso8601-duration": "^2.1.3",
     "schema-dts": "^1.1.5"
   },
   "devDependencies": {
-    "@biomejs/biome": "^2.0.6",
-    "@types/bun": "^1.2.17",
-    "cheerio": "^1.1.0",
-    "tsdown": "^0.12.9",
-    "typescript": "^5.8.3"
+    "@biomejs/biome": "^2.2.6",
+    "@types/bun": "^1.3.0",
+    "cheerio": "^1.1.2",
+    "tsdown": "^0.15.7",
+    "typescript": "^5.9.3"
   }
 }