urlpurify 1.4.2 → 1.6.0

package/dist/unwraps/yandexTurbo.mjs

@@ -1,9 +1,28 @@
+import { createParamExtractor } from "../utils.mjs";
 //#region src/unwraps/yandexTurbo.ts
 const turbopagesHostRegex = /\.turbopages\.org$/;
 const turbopagesPathRegex = /^\/[^/]+\/s\/(.+)$/;
+const hostInPathRegex = /^\/turbo\/([^/]+)\/s\/(.+)$/;
+const yandexSlashSFirstRegex = /^\/turbo\/s\/([^/]+)\/(.*)$/;
+const turbopagesSlashSFirstRegex = /^\/s\/([^/]+)\/(.*)$/;
 const dashRegex = /-/g;
+const yandexHostRegex = /^(?:www\.)?yandex\.(?:com\.tr|[a-z]{2,3})$/;
+const extractTurboTarget = createParamExtractor({
+	hosts: yandexHostRegex,
+	path: "/turbo",
+	params: ["text"]
+});
 const unwrapYandexTurbo = (url) => {
-	if (!turbopagesHostRegex.test(url.hostname)) return;
+	const turboTarget = extractTurboTarget(url);
+	if (turboTarget) return turboTarget;
+	const isTurbopagesHost = turbopagesHostRegex.test(url.hostname);
+	if (!isTurbopagesHost && !yandexHostRegex.test(url.hostname)) return;
+	const slashSFirstRegex = isTurbopagesHost ? turbopagesSlashSFirstRegex : yandexSlashSFirstRegex;
+	const slashSFirstMatch = url.pathname.match(slashSFirstRegex);
+	if (slashSFirstMatch) return `https://${slashSFirstMatch[1]}/${slashSFirstMatch[2]}`;
+	const hostInPathMatch = url.pathname.match(hostInPathRegex);
+	if (hostInPathMatch) return `https://${hostInPathMatch[1]}/${hostInPathMatch[2]}`;
+	if (!isTurbopagesHost) return;
 	const match = url.pathname.match(turbopagesPathRegex);
 	if (!match) return;
 	return `https://${url.hostname.replace(turbopagesHostRegex, "").replace(dashRegex, ".")}/${match[1]}`;

package/dist/unwraps/yelp.d.mts

@@ -0,0 +1,4 @@
+import { UrlUnwrapper } from "../types.mjs";
+//#region src/unwraps/yelp.d.ts
+export declare const unwrapYelp: UrlUnwrapper;
+//#endregion

package/dist/unwraps/yelp.mjs

@@ -0,0 +1,13 @@
+import { createParamExtractor } from "../utils.mjs";
+//#region src/unwraps/yelp.ts
+const redirectPaths = ["/biz_redir", "/redir"];
+const extractTarget = createParamExtractor({
+	hosts: "www.yelp.com",
+	params: ["url"]
+});
+const unwrapYelp = (url) => {
+	if (!redirectPaths.includes(url.pathname)) return;
+	return extractTarget(url);
+};
+//#endregion
+export { unwrapYelp };

package/dist/unwraps/yesware.d.mts

@@ -0,0 +1,4 @@
+import { UrlUnwrapper } from "../types.mjs";
+//#region src/unwraps/yesware.d.ts
+export declare const unwrapYesware: UrlUnwrapper;
+//#endregion

package/dist/unwraps/yesware.mjs

@@ -0,0 +1,19 @@
+import { isHostOf, isHttpUrl } from "../node_modules/trousse/dist/urls.mjs";
+import { createParamExtractor } from "../utils.mjs";
+//#region src/unwraps/yesware.ts
+const trackPathRegex = /^\/tt\/[\da-f]{40}\/[\da-f]{32}\/[\da-f]{32}\/(.+)$/;
+const linkPathRegex = /^\/tl\/[\da-f]{40}\/[\da-f]{32}\/[\da-f]{32}$/;
+const extractLinkTarget = createParamExtractor({
+	hosts: "t.yesware.com",
+	params: ["ytl"]
+});
+const unwrapYesware = (url) => {
+	if (linkPathRegex.test(url.pathname)) return extractLinkTarget(url);
+	if (!isHostOf(url, "t.yesware.com")) return;
+	const match = trackPathRegex.exec(url.pathname);
+	if (!match?.[1]) return;
+	const target = `https://${match[1]}${url.search}${url.hash}`;
+	if (isHttpUrl(target)) return target;
+};
+//#endregion
+export { unwrapYesware };

package/dist/unwraps/yourMembership.d.mts

@@ -0,0 +1,4 @@
+import { UrlUnwrapper } from "../types.mjs";
+//#region src/unwraps/yourMembership.d.ts
+export declare const unwrapYourMembership: UrlUnwrapper;
+//#endregion

package/dist/unwraps/yourMembership.mjs

@@ -0,0 +1,13 @@
+import { isHttpUrl } from "../node_modules/trousse/dist/urls.mjs";
+import { getParamValues, percentDecode } from "../utils.mjs";
+//#region src/unwraps/yourMembership.ts
+const ymlinkRegex = /^\d+$/;
+const encodedSchemeRegex = /^https?%3A/i;
+const unwrapYourMembership = (url) => {
+	if (url.pathname !== "/link.asp" || !ymlinkRegex.test(url.searchParams.get("ymlink") ?? "")) return;
+	let target = getParamValues(url, "finalurl").at(0);
+	if (target && encodedSchemeRegex.test(target)) target = percentDecode(target);
+	if (target && isHttpUrl(target)) return target;
+};
+//#endregion
+export { unwrapYourMembership };

package/dist/unwraps/youtube.mjs

@@ -1,7 +1,7 @@
 import { createParamExtractor } from "../utils.mjs";
 //#region src/unwraps/youtube.ts
 const unwrapYouTube = createParamExtractor({
-	hosts: "www.youtube.com",
+	hosts: ["www.youtube.com", "youtube.com"],
 	path: "/redirect",
 	params: ["q"]
 });

package/dist/unwraps/zanox.d.mts

@@ -0,0 +1,4 @@
+import { UrlUnwrapper } from "../types.mjs";
+//#region src/unwraps/zanox.d.ts
+export declare const unwrapZanox: UrlUnwrapper;
+//#endregion

package/dist/unwraps/zanox.mjs

@@ -0,0 +1,17 @@
+import { createParamExtractor } from "../utils.mjs";
+//#region src/unwraps/zanox.ts
+const bracketRegex = /[?&]ulp=(?:\[\[|%5B%5B)(.+?)(?:\]\]|%5D%5D)/i;
+const extractUlp = createParamExtractor({
+	hosts: "ad.zanox.com",
+	path: "/ppc/",
+	params: ["ULP", "ulp"]
+});
+const unwrapZanox = (url) => {
+	const target = extractUlp(url);
+	if (!target) return;
+	const bracketed = url.search.match(bracketRegex);
+	if (bracketed) return bracketed[1];
+	return target;
+};
+//#endregion
+export { unwrapZanox };

package/dist/unwraps/zayads.d.mts

@@ -0,0 +1,4 @@
+import { UrlUnwrapper } from "../types.mjs";
+//#region src/unwraps/zayads.d.ts
+export declare const unwrapZayads: UrlUnwrapper;
+//#endregion

package/dist/unwraps/zayads.mjs

@@ -0,0 +1,10 @@
+import { isHostOf } from "../node_modules/trousse/dist/urls.mjs";
+import { getPathTarget } from "../pathTarget.mjs";
+//#region src/unwraps/zayads.ts
+const campaignPrefixRegex = /^\/[\w-]{22}\//;
+const unwrapZayads = (url) => {
+	if (!isHostOf(url, "track.zayads.ru")) return;
+	return getPathTarget(url, campaignPrefixRegex);
+};
+//#endregion
+export { unwrapZayads };

package/dist/unwraps/zemanta.d.mts

@@ -0,0 +1,4 @@
+import { UrlUnwrapper } from "../types.mjs";
+//#region src/unwraps/zemanta.d.ts
+export declare const unwrapZemanta: UrlUnwrapper;
+//#endregion

package/dist/unwraps/zemanta.mjs

@@ -0,0 +1,9 @@
+import { createParamExtractor } from "../utils.mjs";
+//#region src/unwraps/zemanta.ts
+const unwrapZemanta = createParamExtractor({
+	hosts: "r.zemanta.com",
+	path: "/",
+	params: ["u"]
+});
+//#endregion
+export { unwrapZemanta };

package/dist/unwraps/zencastr.d.mts

@@ -0,0 +1,4 @@
+import { UrlUnwrapper } from "../types.mjs";
+//#region src/unwraps/zencastr.d.ts
+export declare const unwrapZencastr: UrlUnwrapper;
+//#endregion

package/dist/unwraps/zencastr.mjs

@@ -0,0 +1,10 @@
+import { isHostOf } from "../node_modules/trousse/dist/urls.mjs";
+import { getPathTarget } from "../pathTarget.mjs";
+//#region src/unwraps/zencastr.ts
+const redirectPrefixRegex = /^\/r\//;
+const unwrapZencastr = (url) => {
+	if (!isHostOf(url, ["r.zencastr.com", "r.zen.ai"])) return;
+	return getPathTarget(url, redirectPrefixRegex);
+};
+//#endregion
+export { unwrapZencastr };

package/dist/unwraps/zhihu.mjs

@@ -2,6 +2,7 @@ import { createParamExtractor } from "../utils.mjs";
 //#region src/unwraps/zhihu.ts
 const unwrapZhihu = createParamExtractor({
 	hosts: "link.zhihu.com",
+	path: "/",
 	params: ["target"]
 });
 //#endregion

package/dist/unwraps/zix.d.mts

@@ -0,0 +1,4 @@
+import { UrlUnwrapper } from "../types.mjs";
+//#region src/unwraps/zix.d.ts
+export declare const unwrapZix: UrlUnwrapper;
+//#endregion

package/dist/unwraps/zix.mjs

@@ -0,0 +1,13 @@
+import { createParamExtractor } from "../utils.mjs";
+//#region src/unwraps/zix.ts
+const linkPathRegex = /^\/u\/[0-9a-f]{8}\/[A-Za-z0-9_-]+$/;
+const extractTarget = createParamExtractor({
+	hosts: "link.zixcentral.com",
+	params: ["u"]
+});
+const unwrapZix = (url) => {
+	if (!linkPathRegex.test(url.pathname)) return;
+	return extractTarget(url);
+};
+//#endregion
+export { unwrapZix };

package/dist/unwraps/zscalerIsolation.d.mts

@@ -0,0 +1,4 @@
+import { UrlUnwrapper } from "../types.mjs";
+//#region src/unwraps/zscalerIsolation.d.ts
+export declare const unwrapZscalerIsolation: UrlUnwrapper;
+//#endregion

package/dist/unwraps/zscalerIsolation.mjs

@@ -0,0 +1,13 @@
+import { createParamExtractor } from "../utils.mjs";
+//#region src/unwraps/zscalerIsolation.ts
+const sessionPathRegex = /^\/profile\/[0-9a-f-]{36}\/zia-session\/$/;
+const extractTarget = createParamExtractor({
+	hosts: /^[0-9a-f]{8}\.isolation\.zscaler\.com$/,
+	params: ["original_url"]
+});
+const unwrapZscalerIsolation = (url) => {
+	if (!sessionPathRegex.test(url.pathname)) return;
+	return extractTarget(url);
+};
+//#endregion
+export { unwrapZscalerIsolation };

package/dist/utils.d.mts

@@ -1,7 +1,12 @@
 import { UrlUnwrapper } from "./types.mjs";
 //#region src/utils.d.ts
-export type ParamExtractorConfig = {
+export type ParamExtractorConfig = ({
+  domains: string | Array<string>;
+  hosts?: never;
+} | {
   hosts: string | Array<string> | RegExp;
+  domains?: never;
+}) & {
   path?: string;
   params: Array<string>;
 };

package/dist/utils.mjs

@@ -1,16 +1,47 @@
-import { isHostOf } from "./node_modules/trousse/dist/urls.mjs";
+import { isAnyOf } from "./node_modules/trousse/dist/matching.mjs";
+import { isHostOf, isHostOrSubdomainOf } from "./node_modules/trousse/dist/urls.mjs";
 //#region src/utils.ts
-const matchesHost = (url, hosts) => {
-	if (hosts instanceof RegExp) return hosts.test(url.hostname);
-	return isHostOf(url, hosts);
+const percentDecode = (value) => {
+	const escaped = value.replaceAll("+", "%2B").replaceAll("&", "%26");
+	return new URLSearchParams(`value=${escaped}`).get("value") ?? "";
+};
+const encodedSchemeRegex = /^https?%3A/i;
+const decodeEncodedScheme = (value) => {
+	if (!encodedSchemeRegex.test(value)) return value;
+	return percentDecode(value);
+};
+const unencodedTargetRegex = /^https?:\/\//i;
+const getParamValues = (url, name) => {
+	const values = [];
+	for (const pair of url.search.slice(1).split("&")) {
+		const entry = new URLSearchParams(`&${pair}`).entries().next().value;
+		if (!entry || entry[0] !== name) continue;
+		const raw = pair.slice(pair.indexOf("=") + 1);
+		if (!unencodedTargetRegex.test(raw)) {
+			values.push(entry[1]);
+			continue;
+		}
+		values.push(percentDecode(raw));
+	}
+	return values;
+};
+const getParamTarget = (url, name, index = 0) => {
+	const value = getParamValues(url, name).at(index);
+	if (!value) return;
+	return decodeEncodedScheme(value);
 };
 const createParamExtractor = (config) => {
+	const isHostMatch = (url) => {
+		if (config.domains !== void 0) return isHostOrSubdomainOf(url, config.domains);
+		if (config.hosts instanceof RegExp) return isAnyOf(url.hostname, config.hosts);
+		return isHostOf(url, config.hosts);
+	};
 	return (url) => {
-		if (!matchesHost(url, config.hosts)) return;
+		if (!isHostMatch(url)) return;
 		if (config.path && url.pathname !== config.path) return;
 		for (const param of config.params) {
-			const value = url.searchParams.get(param);
-			if (value) return value;
+			const value = getParamValues(url, param).at(0);
+			if (value) return decodeEncodedScheme(value);
 		}
 	};
 };
@@ -27,11 +58,18 @@ const decodeBase64 = (value) => {
 	for (let index = 0; index < binary.length; index++) bytes[index] = binary.charCodeAt(index);
 	return utf8Decoder.decode(bytes);
 };
+const toBase64 = (base64Url) => {
+	return base64Url.replace(/-/g, "+").replace(/_/g, "/");
+};
 const decodeBase64Url = (value) => {
-	return decodeBase64(value.replace(/-/g, "+").replace(/_/g, "/"));
+	return decodeBase64(toBase64(value));
+};
+const decodeBase64UrlBinary = (value) => {
+	return decodeBase64Binary(toBase64(value));
 };
+const utf8Encoder = new TextEncoder();
 const getUtf8ByteLength = (value) => {
-	return new TextEncoder().encode(value).length;
+	return utf8Encoder.encode(value).length;
 };
 //#endregion
-export { createParamExtractor, decodeBase64, decodeBase64Binary, decodeBase64Url, getUtf8ByteLength };
+export { createParamExtractor, decodeBase64, decodeBase64Binary, decodeBase64Url, decodeBase64UrlBinary, getParamTarget, getParamValues, getUtf8ByteLength, percentDecode };

package/package.json

@@ -41,9 +41,9 @@
   },
   "devDependencies": {
     "@types/bun": "^1.4.2",
-    "kvalita": "^1.24.0",
-    "trousse": "^2.6.0",
+    "kvalita": "^1.24.3",
+    "trousse": "^2.6.1",
     "tsdown": "^0.23.0"
   },
-  "version": "1.4.2"
+  "version": "1.6.0"
 }

package/dist/unwraps/idealoPartner.d.mts

@@ -1,4 +0,0 @@
-import { UrlUnwrapper } from "../types.mjs";
-//#region src/unwraps/idealoPartner.d.ts
-export declare const unwrapIdealoPartner: UrlUnwrapper;
-//#endregion

package/dist/unwraps/idealoPartner.mjs

@@ -1,8 +0,0 @@
-import { createParamExtractor } from "../utils.mjs";
-//#region src/unwraps/idealoPartner.ts
-const unwrapIdealoPartner = createParamExtractor({
-	hosts: /\.idealo-partner\.com$/,
-	params: ["trg"]
-});
-//#endregion
-export { unwrapIdealoPartner };

package/dist/unwraps/pxf.d.mts

@@ -1,4 +0,0 @@
-import { UrlUnwrapper } from "../types.mjs";
-//#region src/unwraps/pxf.d.ts
-export declare const unwrapPxf: UrlUnwrapper;
-//#endregion

package/dist/unwraps/pxf.mjs

@@ -1,8 +0,0 @@
-import { createParamExtractor } from "../utils.mjs";
-//#region src/unwraps/pxf.ts
-const unwrapPxf = createParamExtractor({
-	hosts: /\.pxf\.io$/,
-	params: ["u"]
-});
-//#endregion
-export { unwrapPxf };

package/dist/unwraps/sjv.d.mts

@@ -1,4 +0,0 @@
-import { UrlUnwrapper } from "../types.mjs";
-//#region src/unwraps/sjv.d.ts
-export declare const unwrapSjv: UrlUnwrapper;
-//#endregion

package/dist/unwraps/sjv.mjs

@@ -1,8 +0,0 @@
-import { createParamExtractor } from "../utils.mjs";
-//#region src/unwraps/sjv.ts
-const unwrapSjv = createParamExtractor({
-	hosts: /\.sjv\.io$/,
-	params: ["u"]
-});
-//#endregion
-export { unwrapSjv };