Add checks for mixed numerics, sequences of non-spacing marks, and initial blacklist

src/index.spec.ts

@@ -48,9 +48,15 @@ describe('test', () => {
     expect(validate('\u0369')).toEqual(true);
   });
   it('should return false when sequence of combining marks are present', () => {
-    expect(validate('\u0369\u0369\u0369')).toEqual(false);
+    expect(validate('\u0369\u0369')).toEqual(false);
   });
-  it('shoud return false', () => {
-    expect(validate('googlé.com')).toEqual(false);
+  it('should return false when sequence of kana combining marks are present', () => {
+    expect(validate('\u3099\u3099')).toEqual(false);
+  });
+  it('should return true when a single combining mark is present', () => {
+    expect(validate('\u0369')).toEqual(true);
+  });
+  it('should return true when a single kana combining mark is present', () => {
+    expect(validate('\u3099')).toEqual(true);
   });
 });

src/index.ts

@@ -9,6 +9,8 @@ import {
   kanaCharacterException,
   latinGreekCyrillicAscii,
   nonAsciiLatin,
+  nonSpacingMark,
+  numeric,
   pureAscii,
   removed,
   scripts,
@@ -16,6 +18,8 @@ import {
 
 import { cyrillic } from './regexes/scripts';
 
+// const blacklist = require('./domains.json');
+
 class UrlFormatter {
   public readonly url: string;
   public readonly isTldAscii: boolean;
@@ -30,7 +34,31 @@ class UrlFormatter {
   }
 }
 
-export function detectScripts(validationString: string): object {
+function detectScript(character: string): string {
+  let s = '';
+  for (const [scriptName, script] of Object.entries(scripts)) {
+    if (script.test(character)) {
+      s = scriptName;
+      break;
+    }
+  }
+  return s;
+}
+
+function mixedNumerics(validationString: string) {
+  const result: string[] = [];
+  for (const character of validationString) {
+    if (numeric.test(character)) {
+      const script: string = detectScript(character);
+      if (!result.includes(script)) {
+        result.push(script);
+      }
+    }
+  }
+  return result.length > 1;
+}
+
+function detectScripts(validationString: string): object {
   return Object.entries(scripts)
     .filter(script => {
       return script[1].test(validationString);
@@ -93,6 +121,22 @@ function singleScript(detectedScripts: object): boolean {
   );
 }
 
+// //Only do Levenshtein if it's not blacklisted
+// //Levenshtein - @sogoiii
+// var blHolisticStatus = false;
+// if(isBlacklisted === false && arrWhitelistedDomains.indexOf(strCurrentTab) < 0) {
+//     var strCurrentTab = punycode.toUnicode(strCurrentTab);
+//     var source = strCurrentTab.replace(/\./g, '');
+//     var intHolisticMetric = levenshtein(source, 'myetherwallet');
+//     var intHolisticLimit = 5; // How different can the word be?
+//     blHolisticStatus = (intHolisticMetric > 0 && intHolisticMetric < intHolisticLimit) ? true : false;
+//     if(blHolisticStatus === false) {
+//         //Do edit distance against mycrypto
+//         var intHolisticMetric = levenshtein(source, 'mycrypto');
+//         blHolisticStatus = (intHolisticMetric > 0 && intHolisticMetric < 3) ? true : false;
+//     }
+// }
+
 // function lookupMatchInTopDomains(labels: string[]): boolean {
 //       return labels.every(l => {
 //           if(found) return true
@@ -107,9 +151,27 @@ function inAllowedSets(validationString: string) {
   );
 }
 
+function sequenceOfCombiningMarks(validationString: string) {
+  const characters: string[] = Array.from(validationString);
+  for (let i = 0; i < characters.length; i++) {
+    if (
+      nonSpacingMark.test(characters[i]) &&
+      nonSpacingMark.test(characters[i + 1])
+    ) {
+      return true;
+    }
+    continue;
+  }
+}
+
 export function validate(validationString: string): boolean {
   try {
-    if (!inAllowedSets(validationString)) {
+    if (
+      // blacklist.indexOf(validationString) > -1 ||
+      !inAllowedSets(validationString) ||
+      mixedNumerics(validationString) ||
+      sequenceOfCombiningMarks(validationString)
+    ) {
       return false;
     }
     // scripts

src/regexes/index.ts

@@ -36,6 +36,7 @@ export const dangerous = [
 // for(let i = 0; i < dangerous.length; i++) console.log('dangerous' + i, new RegExp(dangerous[i]).test('abc'))
 
 export const nonSpacingMark = X('\\p{Non_Spacing_Mark}');
+export const numeric = X('\\p{Number}');
 export const decimalDigitNumber = X('\\p{Nd}');
 
 export const allowed = X(

src/types.ts

@@ -0,0 +1,4 @@
+declare module '*.json' {
+  const value: string;
+  export default value;
+}

tslint.json

@@ -21,7 +21,7 @@
       "no-reference": true,
       "unified-signatures": true,
       "await-promise": true,
-      "cyclomatic-complexity": [true, 11],
+      "cyclomatic-complexity": [true, 15],
       "prefer-const": true,
       "no-boolean-literal-compare": true,
       "no-unnecessary-callback-wrapper": true,