Safari compat fix (capture group preserving version)

src/index.ts

@@ -25,6 +25,15 @@ const list = function (input: string, skipHTML:boolean=true) {
 
 	while ((result = finalRegex.exec(input)) !== null) {
 		const start = result.index;
+
+		// To support environments without lookbehind, we use a normal capture
+		// group to check for presence of "_", then truncate it if it exists as
+		// it's not part of the URL.
+		if (result[1]) {
+			result.index -= result[1].length;
+			result[0] = result[0].slice(result[1].length);
+		}
+
 		let end = start + result[0].length;
 		let string = result[0];
 
@@ -152,12 +161,7 @@ const list = function (input: string, skipHTML:boolean=true) {
 				reason: "email",
 			});
 		} else {
-			found.push({
-				start,
-				end,
-				string,
-				reason: "unknown",
-			});
+			throw new Error("Unreachable");
 		}
 	}
 	return found;

src/regex.ts

@@ -15,16 +15,11 @@ export const email = `\\b(mailto:)?${emailAddress}@(${domain}|${ipv4})`;
 export const url = `(${fqdn})${path}?`;
 export const file = `(file:\\/\\/\\/)(?:[a-z]+:(?:\\/|\\\\)+)?([\\w.]+(?:[\\/\\\\]?)+)+`;
 
-// since safari doesn't like lookbehind, we're trying an alternative
-export const final1 = `(?<=\\b|_)((${email})|(${file})|(${url}))(\\b)?`;
-export const final2 = `((\\b)(${email})|(\\b)(${file})|(\\b)(${url}))(\\b)?`;
-
-export let finalRegex = new RegExp(final2, "gi");
-try {
-	finalRegex = new RegExp(final1, "gi");
-} catch (e) {
-	finalRegex = new RegExp(final2, "gi");
-}
+// Since Safari doesn't like lookbehind, we're trying an alternative.
+// Upon matching, we truncate whatever is in the very first capture group
+// ("" or "_") as it's not part of the URL.
+export const final = `(\\b|_)((${email})|(${file})|(${url}))(\\b)?`;
+export const finalRegex = new RegExp(final, "gi");
 
 // for validation purposes
 export const ipRegex = new RegExp(`^(${ipv4}|${ipv6})$`, "i");
@@ -35,7 +30,7 @@ export const urlRegex = new RegExp(`^(${url})$`, "i");
 // identifying parts of the link
 // the initial value of this object is precomputed.
 // https://github.com/alexcorvi/anchorme.js/blob/098843bc0d042601cff592c4f8c9f6d0424c09cd/src/regex.ts
-const iidxes = {"isFile":8,"file":{"fileName":10,"protocol":9},"isEmail":2,"email":{"protocol":3,"local":4,"host":5},"isURL":11,"url":{"TLD":[18,6],"protocol":[15,22],"host":[17],"ipv4":19,"byProtocol":13,"port":21,"protocolWithDomain":12,"path":24}};
+const iidxes = {"isFile":9,"file":{"fileName":11,"protocol":10},"isEmail":3,"email":{"protocol":4,"local":5,"host":6},"isURL":12,"url":{"TLD":[19,7],"protocol":[16,23],"host":[18],"ipv4":20,"byProtocol":14,"port":22,"protocolWithDomain":13,"path":25}};
 
 
 /***
@@ -107,4 +102,4 @@ for (let i = 0; i < testers.length; i++) {
 console.log(JSON.stringify(iidxes));
 */
 
-export { iidxes };
+export { iidxes };

src/types.ts

@@ -1,40 +1,47 @@
 export interface BaseTokenProps {
 	start: number;
 	end: number;
-	reason: string;
 	string: string;
 }
 
 export interface Email extends BaseTokenProps {
+	reason: "email";
 	isEmail: true;
-	protocol: string;
+	isURL?: false;
+	isFile?: false;
+	protocol?: string;
 	local: string;
 	host: string;
 }
 
 export interface URL extends BaseTokenProps {
+	reason: "url";
 	isURL: true;
-	protocol: string;
-	host: string;
-	port: string;
-	ipv4: string;
-	ipv6: string;
+	isEmail?: false;
+	isFile?: false;
+	protocol?: string;
+	host?: string;
+	port?: string;
+	ipv4?: string;
+	ipv6?: string;
 	confirmedByProtocol: boolean;
-	path: string;
-	query: string;
-	fragment: string;
+	path?: string;
+	query?: string;
+	fragment?: string;
 }
 
 export interface File extends BaseTokenProps {
+	reason: "file";
 	isFile: true;
-	filename: string;
+	isURL?: false;
+	isEmail?: false;
+	protocol: string;
+	filename?: string;
 	filePath: string;
-	fileDirectory: string;
+	fileDirectory?: string;
 }
 
-type TokenProps = Email & File & URL;
-
-export type ListingProps = Partial<TokenProps> & BaseTokenProps;
+export type ListingProps = Email | File | URL;
 
 export type DesiredValues =
 	| { [key: string]: string | undefined | true }

test/_global.test.ts

@@ -0,0 +1,15 @@
+const OriginalRegExp = globalThis.RegExp;
+
+// Mock RegExp constructor used in tests to ensure regexes contain no lookbehind syntax
+class SafariRegExp extends OriginalRegExp {
+	constructor(pattern: string | RegExp, flags?: string) {
+		super(pattern, flags);
+		pattern = String(pattern)
+		if (pattern.includes("(?<=") || pattern.includes("(?<!")) {
+			throw new SyntaxError("Lookbehind syntax (?<=...) and (?<!...) is disallowed to support legacy Safari");
+		}
+	}
+}
+
+// @ts-ignore
+globalThis.RegExp = SafariRegExp;

test/a_unit/groups.test.ts

@@ -0,0 +1,145 @@
+/// <reference types="mocha" />
+import anchorme from "../../src/index";
+import * as expect from "expect";
+
+describe("UNIT: groups", () => {
+	it("groups", () => {
+		const matches = anchorme.list(
+			"https://example.xyz example.com [email protected] file:///filename.txt 192.168.1.1",
+			false,
+		);
+
+		const expected = [
+			{
+				start: 0,
+				end: 19,
+				string: "https://example.xyz",
+				isURL: true,
+				protocol: "https://",
+				host: "example.xyz",
+				confirmedByProtocol: true,
+				reason: "url",
+			},
+			{
+				start: 20,
+				end: 31,
+				string: "example.com",
+				isURL: true,
+				host: "example.com",
+				confirmedByProtocol: false,
+				reason: "url",
+			},
+			{
+				start: 32,
+				end: 46,
+				string: "[email protected]",
+				isEmail: true,
+				local: "user",
+				host: "email.com",
+				reason: "email",
+			},
+			{
+				start: 47,
+				end: 67,
+				string: "file:///filename.txt",
+				isFile: true,
+				protocol: "file:///",
+				filename: "filename.txt",
+				filePath: "filename.txt",
+				fileDirectory: "",
+				reason: "file",
+			},
+			{
+				start: 68,
+				end: 79,
+				string: "192.168.1.1",
+				isURL: true,
+				ipv4: "192.168.1.1",
+				host: "192.168.1.1",
+				confirmedByProtocol: false,
+				reason: "url",
+			},
+		];
+
+		expect(matches).toEqual(expected);
+
+		for (const match of matches) {
+			switch (match.reason) {
+				case "email": {
+					expect(match.isEmail).toBe(true);
+					expect(match.isFile).toBeFalsy();
+					expect(match.isURL).toBeFalsy();
+
+					// ts
+					match.protocol;
+					match.local;
+					match.host;
+					expect(match).toHaveProperty("protocol");
+					expect(match).toHaveProperty("local");
+					expect(match).toHaveProperty("host");
+
+					// @ts-expect-error
+					match.filename;
+					expect(match).not.toHaveProperty("filename");
+
+					break;
+				}
+				case "file": {
+					expect(match.isFile).toBe(true);
+					expect(match.isEmail).toBeFalsy();
+					expect(match.isURL).toBeFalsy();
+
+					// ts
+					match.protocol;
+					match.filename;
+					match.filePath;
+					match.fileDirectory;
+					expect(match).toHaveProperty("protocol");
+					expect(match).toHaveProperty("filename");
+					expect(match).toHaveProperty("filePath");
+					expect(match).toHaveProperty("fileDirectory");
+
+					// @ts-expect-error
+					match.fragment;
+					expect(match).not.toHaveProperty("fragment");
+
+					break;
+				}
+				case "url": {
+					expect(match.isURL).toBe(true);
+					expect(match.isEmail).toBeFalsy();
+					expect(match.isFile).toBeFalsy();
+
+					// ts
+					match.protocol;
+					match.host;
+					match.port;
+					match.ipv4;
+					match.ipv6;
+					match.confirmedByProtocol;
+					match.path;
+					match.query;
+					match.fragment;
+					expect(match).toHaveProperty("protocol");
+					expect(match).toHaveProperty("host");
+					expect(match).toHaveProperty("port");
+					expect(match).toHaveProperty("ipv4");
+					expect(match).toHaveProperty("ipv6");
+					expect(match).toHaveProperty("confirmedByProtocol");
+					expect(match).toHaveProperty("path");
+					expect(match).toHaveProperty("query");
+					expect(match).toHaveProperty("fragment");
+
+					// @ts-expect-error
+					match.local;
+					expect(match).not.toHaveProperty("local");
+
+					break;
+				}
+				default: {
+					throw new Error("Unreachable");
+				}
+			}
+		}
+	});
+});