feedcanon 1.0.0 → 1.1.0

package/README.md

@@ -36,11 +36,14 @@ The 9 URLs below all work and return identical content. None redirect to each ot
 
 ### How It Works
 
+This is a simplified flow. For complete details, see [How It Works](https://feedcanon.dev/how-it-works) in the docs.
+
 1. Fetch the input URL and parse the feed to establish reference content.
-2. Extract the feed's declared self URL and validate it serves identical content.
-3. Generate URL variants ordered from cleanest to least clean.
-4. Test variants in order—the first one serving identical content wins.
-5. Upgrade HTTP to HTTPS if both serve identical content.
+2. Extract the feed's declared self URL (if present).
+3. Validate the self URL by fetching and comparing content.
+4. Generate URL variants ordered from cleanest to least clean.
+5. Test variants in order—the first one serving identical content wins.
+6. Upgrade HTTP to HTTPS if both serve identical content.
 
 ### Customization
 

package/dist/defaults.cjs

@@ -185,6 +185,13 @@ const defaultFetch = async (url, options) => {
 		status: response.status
 	};
 };
+const findSelfLink = (parsed) => {
+	switch (parsed.format) {
+		case "atom": return parsed.feed.links?.find((link) => link.rel === "self");
+		case "rss":
+		case "rdf": return parsed.feed.atom?.links?.find((link) => link.rel === "self");
+	}
+};
 const defaultParser = {
 	parse: (body) => {
 		try {
@@ -192,15 +199,36 @@ const defaultParser = {
 		} catch {}
 	},
 	getSelfUrl: (parsed) => {
-		switch (parsed.format) {
-			case "atom": return parsed.feed.links?.find((link) => link.rel === "self")?.href;
-			case "rss":
-			case "rdf": return parsed.feed.atom?.links?.find((link) => link.rel === "self")?.href;
-			case "json": return parsed.feed.feed_url;
-		}
+		return parsed.format === "json" ? parsed.feed.feed_url : findSelfLink(parsed)?.href;
 	},
 	getSignature: (parsed) => {
-		return parsed.feed;
+		if (parsed.format === "json") {
+			const originalSelfUrl = parsed.feed.feed_url;
+			parsed.feed.feed_url = void 0;
+			const signature$1 = JSON.stringify(parsed.feed);
+			parsed.feed.feed_url = originalSelfUrl;
+			return signature$1;
+		}
+		let signature;
+		let originalBuildDate;
+		if (parsed.format === "rss") {
+			originalBuildDate = parsed.feed.lastBuildDate;
+			parsed.feed.lastBuildDate = void 0;
+		} else if (parsed.format === "atom") {
+			originalBuildDate = parsed.feed.updated;
+			parsed.feed.updated = void 0;
+		}
+		const link = findSelfLink(parsed);
+		if (!link) signature = JSON.stringify(parsed.feed);
+		else {
+			const originalSelfUrl = link.href;
+			link.href = void 0;
+			signature = JSON.stringify(parsed.feed);
+			link.href = originalSelfUrl;
+		}
+		if (parsed.format === "rss") parsed.feed.lastBuildDate = originalBuildDate;
+		else if (parsed.format === "atom") parsed.feed.updated = originalBuildDate;
+		return signature;
 	}
 };
 const defaultTiers = [

package/dist/defaults.d.cts

@@ -1,11 +1,11 @@
-import { FeedsmithFeed, FetchFn, NormalizeOptions, ParserAdapter, PlatformHandler, Tier } from "./types.cjs";
+import { DefaultParserResult, FetchFn, NormalizeOptions, ParserAdapter, PlatformHandler, Tier } from "./types.cjs";
 
 //#region src/defaults.d.ts
 declare const defaultPlatforms: Array<PlatformHandler>;
 declare const defaultStrippedParams: string[];
 declare const defaultNormalizeOptions: NormalizeOptions;
 declare const defaultFetch: FetchFn;
-declare const defaultParser: ParserAdapter<FeedsmithFeed>;
+declare const defaultParser: ParserAdapter<DefaultParserResult>;
 declare const defaultTiers: Array<Tier>;
 //#endregion
 export { defaultFetch, defaultNormalizeOptions, defaultParser, defaultPlatforms, defaultStrippedParams, defaultTiers };

package/dist/defaults.d.ts

@@ -1,11 +1,11 @@
-import { FeedsmithFeed, FetchFn, NormalizeOptions, ParserAdapter, PlatformHandler, Tier } from "./types.js";
+import { DefaultParserResult, FetchFn, NormalizeOptions, ParserAdapter, PlatformHandler, Tier } from "./types.js";
 
 //#region src/defaults.d.ts
 declare const defaultPlatforms: Array<PlatformHandler>;
 declare const defaultStrippedParams: string[];
 declare const defaultNormalizeOptions: NormalizeOptions;
 declare const defaultFetch: FetchFn;
-declare const defaultParser: ParserAdapter<FeedsmithFeed>;
+declare const defaultParser: ParserAdapter<DefaultParserResult>;
 declare const defaultTiers: Array<Tier>;
 //#endregion
 export { defaultFetch, defaultNormalizeOptions, defaultParser, defaultPlatforms, defaultStrippedParams, defaultTiers };

package/dist/defaults.js

@@ -185,6 +185,13 @@ const defaultFetch = async (url, options) => {
 		status: response.status
 	};
 };
+const findSelfLink = (parsed) => {
+	switch (parsed.format) {
+		case "atom": return parsed.feed.links?.find((link) => link.rel === "self");
+		case "rss":
+		case "rdf": return parsed.feed.atom?.links?.find((link) => link.rel === "self");
+	}
+};
 const defaultParser = {
 	parse: (body) => {
 		try {
@@ -192,15 +199,36 @@ const defaultParser = {
 		} catch {}
 	},
 	getSelfUrl: (parsed) => {
-		switch (parsed.format) {
-			case "atom": return parsed.feed.links?.find((link) => link.rel === "self")?.href;
-			case "rss":
-			case "rdf": return parsed.feed.atom?.links?.find((link) => link.rel === "self")?.href;
-			case "json": return parsed.feed.feed_url;
-		}
+		return parsed.format === "json" ? parsed.feed.feed_url : findSelfLink(parsed)?.href;
 	},
 	getSignature: (parsed) => {
-		return parsed.feed;
+		if (parsed.format === "json") {
+			const originalSelfUrl = parsed.feed.feed_url;
+			parsed.feed.feed_url = void 0;
+			const signature$1 = JSON.stringify(parsed.feed);
+			parsed.feed.feed_url = originalSelfUrl;
+			return signature$1;
+		}
+		let signature;
+		let originalBuildDate;
+		if (parsed.format === "rss") {
+			originalBuildDate = parsed.feed.lastBuildDate;
+			parsed.feed.lastBuildDate = void 0;
+		} else if (parsed.format === "atom") {
+			originalBuildDate = parsed.feed.updated;
+			parsed.feed.updated = void 0;
+		}
+		const link = findSelfLink(parsed);
+		if (!link) signature = JSON.stringify(parsed.feed);
+		else {
+			const originalSelfUrl = link.href;
+			link.href = void 0;
+			signature = JSON.stringify(parsed.feed);
+			link.href = originalSelfUrl;
+		}
+		if (parsed.format === "rss") parsed.feed.lastBuildDate = originalBuildDate;
+		else if (parsed.format === "atom") parsed.feed.updated = originalBuildDate;
+		return signature;
 	}
 };
 const defaultTiers = [

package/dist/exports.cjs

@@ -11,6 +11,7 @@ exports.defaultStrippedParams = require_defaults.defaultStrippedParams;
 exports.defaultTiers = require_defaults.defaultTiers;
 exports.feedburnerHandler = require_feedburner.feedburnerHandler;
 exports.findCanonical = require_index.findCanonical;
+exports.fixMalformedProtocol = require_utils.fixMalformedProtocol;
 exports.normalizeUrl = require_utils.normalizeUrl;
 exports.resolveFeedProtocol = require_utils.resolveFeedProtocol;
 exports.resolveUrl = require_utils.resolveUrl;

package/dist/exports.d.cts

@@ -1,6 +1,6 @@
-import { ExistsFn, FeedsmithFeed, FetchFn, FetchFnOptions, FetchFnResponse, FindCanonicalOptions, NormalizeOptions, OnExistsFn, OnFetchFn, OnMatchFn, ParserAdapter, PlatformHandler } from "./types.cjs";
+import { DefaultParserResult, ExistsFn, FetchFn, FetchFnOptions, FetchFnResponse, FindCanonicalOptions, NormalizeOptions, OnExistsFn, OnFetchFn, OnMatchFn, ParserAdapter, PlatformHandler } from "./types.cjs";
 import { defaultFetch, defaultParser, defaultPlatforms, defaultStrippedParams, defaultTiers } from "./defaults.cjs";
 import { findCanonical } from "./index.cjs";
 import { feedburnerHandler } from "./platforms/feedburner.cjs";
-import { addMissingProtocol, normalizeUrl, resolveFeedProtocol, resolveUrl } from "./utils.cjs";
-export { type ExistsFn, type FeedsmithFeed, type FetchFn, type FetchFnOptions, type FetchFnResponse, type FindCanonicalOptions, type NormalizeOptions, type OnExistsFn, type OnFetchFn, type OnMatchFn, type ParserAdapter, type PlatformHandler, addMissingProtocol, defaultFetch, defaultParser, defaultPlatforms, defaultStrippedParams, defaultTiers, feedburnerHandler, findCanonical, normalizeUrl, resolveFeedProtocol, resolveUrl };
+import { addMissingProtocol, fixMalformedProtocol, normalizeUrl, resolveFeedProtocol, resolveUrl } from "./utils.cjs";
+export { type DefaultParserResult, type ExistsFn, type FetchFn, type FetchFnOptions, type FetchFnResponse, type FindCanonicalOptions, type NormalizeOptions, type OnExistsFn, type OnFetchFn, type OnMatchFn, type ParserAdapter, type PlatformHandler, addMissingProtocol, defaultFetch, defaultParser, defaultPlatforms, defaultStrippedParams, defaultTiers, feedburnerHandler, findCanonical, fixMalformedProtocol, normalizeUrl, resolveFeedProtocol, resolveUrl };

package/dist/exports.d.ts

@@ -1,6 +1,6 @@
-import { ExistsFn, FeedsmithFeed, FetchFn, FetchFnOptions, FetchFnResponse, FindCanonicalOptions, NormalizeOptions, OnExistsFn, OnFetchFn, OnMatchFn, ParserAdapter, PlatformHandler } from "./types.js";
+import { DefaultParserResult, ExistsFn, FetchFn, FetchFnOptions, FetchFnResponse, FindCanonicalOptions, NormalizeOptions, OnExistsFn, OnFetchFn, OnMatchFn, ParserAdapter, PlatformHandler } from "./types.js";
 import { defaultFetch, defaultParser, defaultPlatforms, defaultStrippedParams, defaultTiers } from "./defaults.js";
 import { findCanonical } from "./index.js";
 import { feedburnerHandler } from "./platforms/feedburner.js";
-import { addMissingProtocol, normalizeUrl, resolveFeedProtocol, resolveUrl } from "./utils.js";
-export { type ExistsFn, type FeedsmithFeed, type FetchFn, type FetchFnOptions, type FetchFnResponse, type FindCanonicalOptions, type NormalizeOptions, type OnExistsFn, type OnFetchFn, type OnMatchFn, type ParserAdapter, type PlatformHandler, addMissingProtocol, defaultFetch, defaultParser, defaultPlatforms, defaultStrippedParams, defaultTiers, feedburnerHandler, findCanonical, normalizeUrl, resolveFeedProtocol, resolveUrl };
+import { addMissingProtocol, fixMalformedProtocol, normalizeUrl, resolveFeedProtocol, resolveUrl } from "./utils.js";
+export { type DefaultParserResult, type ExistsFn, type FetchFn, type FetchFnOptions, type FetchFnResponse, type FindCanonicalOptions, type NormalizeOptions, type OnExistsFn, type OnFetchFn, type OnMatchFn, type ParserAdapter, type PlatformHandler, addMissingProtocol, defaultFetch, defaultParser, defaultPlatforms, defaultStrippedParams, defaultTiers, feedburnerHandler, findCanonical, fixMalformedProtocol, normalizeUrl, resolveFeedProtocol, resolveUrl };

package/dist/exports.js

@@ -1,6 +1,6 @@
 import { feedburnerHandler } from "./platforms/feedburner.js";
 import { defaultFetch, defaultParser, defaultPlatforms, defaultStrippedParams, defaultTiers } from "./defaults.js";
-import { addMissingProtocol, normalizeUrl, resolveFeedProtocol, resolveUrl } from "./utils.js";
+import { addMissingProtocol, fixMalformedProtocol, normalizeUrl, resolveFeedProtocol, resolveUrl } from "./utils.js";
 import { findCanonical } from "./index.js";
 
-export { addMissingProtocol, defaultFetch, defaultParser, defaultPlatforms, defaultStrippedParams, defaultTiers, feedburnerHandler, findCanonical, normalizeUrl, resolveFeedProtocol, resolveUrl };
+export { addMissingProtocol, defaultFetch, defaultParser, defaultPlatforms, defaultStrippedParams, defaultTiers, feedburnerHandler, findCanonical, fixMalformedProtocol, normalizeUrl, resolveFeedProtocol, resolveUrl };

package/dist/index.cjs

@@ -52,8 +52,8 @@ async function findCanonical(inputUrl, options) {
 		if (initialResponseBody === comparedResponseBody) return true;
 		const comparedResponseFeed = await parser.parse(comparedResponseBody);
 		if (comparedResponseFeed) {
-			initialResponseSignature ||= JSON.stringify(parser.getSignature(initialResponseFeed));
-			const comparedResponseSignature = JSON.stringify(parser.getSignature(comparedResponseFeed));
+			initialResponseSignature ||= parser.getSignature(initialResponseFeed);
+			const comparedResponseSignature = parser.getSignature(comparedResponseFeed);
 			return initialResponseSignature === comparedResponseSignature;
 		}
 		return false;

package/dist/index.d.cts

@@ -1,7 +1,7 @@
-import { FeedsmithFeed, FetchFnResponse, FindCanonicalOptions, ParserAdapter } from "./types.cjs";
+import { DefaultParserResult, FetchFnResponse, FindCanonicalOptions, ParserAdapter } from "./types.cjs";
 
 //#region src/index.d.ts
-declare function findCanonical<TResponse extends FetchFnResponse = FetchFnResponse, TExisting = unknown>(inputUrl: string, options?: Omit<FindCanonicalOptions<FeedsmithFeed, TResponse, TExisting>, 'parser'>): Promise<string | undefined>;
+declare function findCanonical<TResponse extends FetchFnResponse = FetchFnResponse, TExisting = unknown>(inputUrl: string, options?: Omit<FindCanonicalOptions<DefaultParserResult, TResponse, TExisting>, 'parser'>): Promise<string | undefined>;
 declare function findCanonical<TFeed, TResponse extends FetchFnResponse = FetchFnResponse, TExisting = unknown>(inputUrl: string, options: FindCanonicalOptions<TFeed, TResponse, TExisting> & {
   parser: ParserAdapter<TFeed>;
 }): Promise<string | undefined>;

package/dist/index.d.ts

@@ -1,7 +1,7 @@
-import { FeedsmithFeed, FetchFnResponse, FindCanonicalOptions, ParserAdapter } from "./types.js";
+import { DefaultParserResult, FetchFnResponse, FindCanonicalOptions, ParserAdapter } from "./types.js";
 
 //#region src/index.d.ts
-declare function findCanonical<TResponse extends FetchFnResponse = FetchFnResponse, TExisting = unknown>(inputUrl: string, options?: Omit<FindCanonicalOptions<FeedsmithFeed, TResponse, TExisting>, 'parser'>): Promise<string | undefined>;
+declare function findCanonical<TResponse extends FetchFnResponse = FetchFnResponse, TExisting = unknown>(inputUrl: string, options?: Omit<FindCanonicalOptions<DefaultParserResult, TResponse, TExisting>, 'parser'>): Promise<string | undefined>;
 declare function findCanonical<TFeed, TResponse extends FetchFnResponse = FetchFnResponse, TExisting = unknown>(inputUrl: string, options: FindCanonicalOptions<TFeed, TResponse, TExisting> & {
   parser: ParserAdapter<TFeed>;
 }): Promise<string | undefined>;

package/dist/index.js

@@ -52,8 +52,8 @@ async function findCanonical(inputUrl, options) {
 		if (initialResponseBody === comparedResponseBody) return true;
 		const comparedResponseFeed = await parser.parse(comparedResponseBody);
 		if (comparedResponseFeed) {
-			initialResponseSignature ||= JSON.stringify(parser.getSignature(initialResponseFeed));
-			const comparedResponseSignature = JSON.stringify(parser.getSignature(comparedResponseFeed));
+			initialResponseSignature ||= parser.getSignature(initialResponseFeed);
+			const comparedResponseSignature = parser.getSignature(comparedResponseFeed);
 			return initialResponseSignature === comparedResponseSignature;
 		}
 		return false;

package/dist/types.d.cts

@@ -1,11 +1,11 @@
 import * as feedsmith0 from "feedsmith";
 
 //#region src/types.d.ts
-type FeedsmithFeed = ReturnType<typeof feedsmith0.parseFeed>;
+type DefaultParserResult = ReturnType<typeof feedsmith0.parseFeed>;
 type ParserAdapter<T> = {
   parse: (body: string) => Promise<T | undefined> | T | undefined;
   getSelfUrl: (parsed: T) => string | undefined;
-  getSignature: (parsed: T) => object;
+  getSignature: (parsed: T) => string;
 };
 type PlatformHandler = {
   match: (url: URL) => boolean;
@@ -41,7 +41,7 @@ type OnExistsFn<T> = (data: {
   url: string;
   data: T;
 }) => void;
-type FindCanonicalOptions<TFeed = FeedsmithFeed, TResponse extends FetchFnResponse = FetchFnResponse, TExisting = unknown> = {
+type FindCanonicalOptions<TFeed = DefaultParserResult, TResponse extends FetchFnResponse = FetchFnResponse, TExisting = unknown> = {
   parser?: ParserAdapter<TFeed>;
   fetchFn?: FetchFn<TResponse>;
   existsFn?: ExistsFn<TExisting>;
@@ -65,4 +65,4 @@ type FetchFnResponse = {
 };
 type FetchFn<TResponse extends FetchFnResponse = FetchFnResponse> = (url: string, options?: FetchFnOptions) => Promise<TResponse>;
 //#endregion
-export { ExistsFn, FeedsmithFeed, FetchFn, FetchFnOptions, FetchFnResponse, FindCanonicalOptions, NormalizeOptions, OnExistsFn, OnFetchFn, OnMatchFn, ParserAdapter, PlatformHandler, Tier };
+export { DefaultParserResult, ExistsFn, FetchFn, FetchFnOptions, FetchFnResponse, FindCanonicalOptions, NormalizeOptions, OnExistsFn, OnFetchFn, OnMatchFn, ParserAdapter, PlatformHandler, Tier };

package/dist/types.d.ts

@@ -1,11 +1,11 @@
 import * as feedsmith0 from "feedsmith";
 
 //#region src/types.d.ts
-type FeedsmithFeed = ReturnType<typeof feedsmith0.parseFeed>;
+type DefaultParserResult = ReturnType<typeof feedsmith0.parseFeed>;
 type ParserAdapter<T> = {
   parse: (body: string) => Promise<T | undefined> | T | undefined;
   getSelfUrl: (parsed: T) => string | undefined;
-  getSignature: (parsed: T) => object;
+  getSignature: (parsed: T) => string;
 };
 type PlatformHandler = {
   match: (url: URL) => boolean;
@@ -41,7 +41,7 @@ type OnExistsFn<T> = (data: {
   url: string;
   data: T;
 }) => void;
-type FindCanonicalOptions<TFeed = FeedsmithFeed, TResponse extends FetchFnResponse = FetchFnResponse, TExisting = unknown> = {
+type FindCanonicalOptions<TFeed = DefaultParserResult, TResponse extends FetchFnResponse = FetchFnResponse, TExisting = unknown> = {
   parser?: ParserAdapter<TFeed>;
   fetchFn?: FetchFn<TResponse>;
   existsFn?: ExistsFn<TExisting>;
@@ -65,4 +65,4 @@ type FetchFnResponse = {
 };
 type FetchFn<TResponse extends FetchFnResponse = FetchFnResponse> = (url: string, options?: FetchFnOptions) => Promise<TResponse>;
 //#endregion
-export { ExistsFn, FeedsmithFeed, FetchFn, FetchFnOptions, FetchFnResponse, FindCanonicalOptions, NormalizeOptions, OnExistsFn, OnFetchFn, OnMatchFn, ParserAdapter, PlatformHandler, Tier };
+export { DefaultParserResult, ExistsFn, FetchFn, FetchFnOptions, FetchFnResponse, FindCanonicalOptions, NormalizeOptions, OnExistsFn, OnFetchFn, OnMatchFn, ParserAdapter, PlatformHandler, Tier };

package/dist/utils.cjs

@@ -15,6 +15,27 @@ const getStrippedParamsSet = (params) => {
 const ipv4Pattern = /^\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}$/;
 const ipv6Pattern = /^([0-9a-f]{0,4}:){2,7}[0-9a-f]{0,4}$/i;
 const safePathChars = /[a-zA-Z0-9._~!$&'()*+,;=:@-]/;
+const validUrlPattern = /^https?:\/\/(?:www\.|[a-vx-z0-9])/i;
+const doubledProtocolPattern = /^\/?[htps]{2,7}[:\s=.\\/]+([htps]{2,7})[:\s=.\\/]+[.,:/]*(www[./]+)?/i;
+const singleMalformedPattern = /^\/?(?:h[htps():]{1,10}|t{1,2}ps?)[:\s=.\\/]+[.,:/]*(www[./]+)?/i;
+const fixMalformedProtocol = (url) => {
+	if (validUrlPattern.test(url) && !doubledProtocolPattern.test(url)) return url;
+	const doubledMatch = doubledProtocolPattern.exec(url);
+	if (doubledMatch) {
+		const inner = doubledMatch[1];
+		const www = doubledMatch[2];
+		const rest = url.slice(doubledMatch[0].length);
+		return (/s/i.test(inner) ? "https://" : "http://") + (www ? "www." : "") + rest;
+	}
+	const singleMatch = singleMalformedPattern.exec(url);
+	if (singleMatch) {
+		const fullMatch = singleMatch[0];
+		const www = singleMatch[1];
+		const rest = url.slice(fullMatch.length);
+		return (/s/i.test(fullMatch) ? "https://" : "http://") + (www ? "www." : "") + rest;
+	}
+	return url;
+};
 const feedProtocols = [
 	"feed:",
 	"rss:",
@@ -58,6 +79,7 @@ const resolveUrl = (url, base) => {
 	let resolvedUrl;
 	resolvedUrl = url.includes("&") ? (0, entities.decodeHTML)(url) : url;
 	resolvedUrl = resolveFeedProtocol(resolvedUrl);
+	resolvedUrl = fixMalformedProtocol(resolvedUrl);
 	if (base) try {
 		resolvedUrl = new URL(resolvedUrl, base).href;
 	} catch {
@@ -114,6 +136,7 @@ const normalizeUrl = (url, options = require_defaults.defaultNormalizeOptions) =
 		if (options.sortQueryParams) parsed.searchParams.sort();
 		if (options.stripEmptyQuery && parsed.href.endsWith("?")) parsed.search = "";
 		let result = parsed.href;
+		if (options.stripRootSlash && result === `${parsed.origin}/`) result = parsed.origin;
 		if (options.stripProtocol) result = result.replace(/^https?:\/\//, "");
 		return result;
 	} catch {
@@ -136,6 +159,7 @@ const applyPlatformHandlers = (url, platforms) => {
 //#endregion
 exports.addMissingProtocol = addMissingProtocol;
 exports.applyPlatformHandlers = applyPlatformHandlers;
+exports.fixMalformedProtocol = fixMalformedProtocol;
 exports.normalizeUrl = normalizeUrl;
 exports.resolveFeedProtocol = resolveFeedProtocol;
 exports.resolveUrl = resolveUrl;

package/dist/utils.d.cts

@@ -1,9 +1,10 @@
 import { NormalizeOptions } from "./types.cjs";
 
 //#region src/utils.d.ts
+declare const fixMalformedProtocol: (url: string) => string;
 declare const resolveFeedProtocol: (url: string, protocol?: "http" | "https") => string;
 declare const addMissingProtocol: (url: string, protocol?: "http" | "https") => string;
 declare const resolveUrl: (url: string, base?: string) => string | undefined;
 declare const normalizeUrl: (url: string, options?: NormalizeOptions) => string;
 //#endregion
-export { addMissingProtocol, normalizeUrl, resolveFeedProtocol, resolveUrl };
+export { addMissingProtocol, fixMalformedProtocol, normalizeUrl, resolveFeedProtocol, resolveUrl };

package/dist/utils.d.ts

@@ -1,9 +1,10 @@
 import { NormalizeOptions } from "./types.js";
 
 //#region src/utils.d.ts
+declare const fixMalformedProtocol: (url: string) => string;
 declare const resolveFeedProtocol: (url: string, protocol?: "http" | "https") => string;
 declare const addMissingProtocol: (url: string, protocol?: "http" | "https") => string;
 declare const resolveUrl: (url: string, base?: string) => string | undefined;
 declare const normalizeUrl: (url: string, options?: NormalizeOptions) => string;
 //#endregion
-export { addMissingProtocol, normalizeUrl, resolveFeedProtocol, resolveUrl };
+export { addMissingProtocol, fixMalformedProtocol, normalizeUrl, resolveFeedProtocol, resolveUrl };

package/dist/utils.js

@@ -15,6 +15,27 @@ const getStrippedParamsSet = (params) => {
 const ipv4Pattern = /^\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}$/;
 const ipv6Pattern = /^([0-9a-f]{0,4}:){2,7}[0-9a-f]{0,4}$/i;
 const safePathChars = /[a-zA-Z0-9._~!$&'()*+,;=:@-]/;
+const validUrlPattern = /^https?:\/\/(?:www\.|[a-vx-z0-9])/i;
+const doubledProtocolPattern = /^\/?[htps]{2,7}[:\s=.\\/]+([htps]{2,7})[:\s=.\\/]+[.,:/]*(www[./]+)?/i;
+const singleMalformedPattern = /^\/?(?:h[htps():]{1,10}|t{1,2}ps?)[:\s=.\\/]+[.,:/]*(www[./]+)?/i;
+const fixMalformedProtocol = (url) => {
+	if (validUrlPattern.test(url) && !doubledProtocolPattern.test(url)) return url;
+	const doubledMatch = doubledProtocolPattern.exec(url);
+	if (doubledMatch) {
+		const inner = doubledMatch[1];
+		const www = doubledMatch[2];
+		const rest = url.slice(doubledMatch[0].length);
+		return (/s/i.test(inner) ? "https://" : "http://") + (www ? "www." : "") + rest;
+	}
+	const singleMatch = singleMalformedPattern.exec(url);
+	if (singleMatch) {
+		const fullMatch = singleMatch[0];
+		const www = singleMatch[1];
+		const rest = url.slice(fullMatch.length);
+		return (/s/i.test(fullMatch) ? "https://" : "http://") + (www ? "www." : "") + rest;
+	}
+	return url;
+};
 const feedProtocols = [
 	"feed:",
 	"rss:",
@@ -58,6 +79,7 @@ const resolveUrl = (url, base) => {
 	let resolvedUrl;
 	resolvedUrl = url.includes("&") ? decodeHTML(url) : url;
 	resolvedUrl = resolveFeedProtocol(resolvedUrl);
+	resolvedUrl = fixMalformedProtocol(resolvedUrl);
 	if (base) try {
 		resolvedUrl = new URL(resolvedUrl, base).href;
 	} catch {
@@ -114,6 +136,7 @@ const normalizeUrl = (url, options = defaultNormalizeOptions) => {
 		if (options.sortQueryParams) parsed.searchParams.sort();
 		if (options.stripEmptyQuery && parsed.href.endsWith("?")) parsed.search = "";
 		let result = parsed.href;
+		if (options.stripRootSlash && result === `${parsed.origin}/`) result = parsed.origin;
 		if (options.stripProtocol) result = result.replace(/^https?:\/\//, "");
 		return result;
 	} catch {
@@ -134,4 +157,4 @@ const applyPlatformHandlers = (url, platforms) => {
 };
 
 //#endregion
-export { addMissingProtocol, applyPlatformHandlers, normalizeUrl, resolveFeedProtocol, resolveUrl };
+export { addMissingProtocol, applyPlatformHandlers, fixMalformedProtocol, normalizeUrl, resolveFeedProtocol, resolveUrl };

package/package.json

@@ -60,8 +60,8 @@
   "devDependencies": {
     "@types/bun": "^1.3.5",
     "kvalita": "1.9.0",
-    "tsdown": "^0.18.3",
+    "tsdown": "^0.18.4",
     "vitepress": "^1.6.4"
   },
-  "version": "1.0.0"
+  "version": "1.1.0"
 }