Introduce 'lenient' mode for hostname validation.

index.js

@@ -9,7 +9,7 @@ var extractHostname = require('./lib/clean-host.js');
 var getDomain = require('./lib/domain.js');
 var getPublicSuffix = require('./lib/public-suffix.js');
 var getSubdomain = require('./lib/subdomain.js');
-var isValidHostname = require('./lib/is-valid.js');
+var isValid = require('./lib/is-valid.js');
 var isIp = require('./lib/is-ip.js');
 var tldExists = require('./lib/tld-exists.js');
 
@@ -33,8 +33,23 @@ var ALL = 5;
 function factory(options) {
   var rules = options.rules || allRules || {};
   var validHosts = options.validHosts || [];
+  var lenientHostnameValidation = options.lenientHostnameValidation === true;
   var _extractHostname = options.extractHostname || extractHostname;
 
+  // @see https://github.com/oncletom/tld.js/pull/122
+  var isValidHostname = isValid;
+  if (lenientHostnameValidation) {
+    // If 'lenient' mode is enabled, then underscores are allowed to appear in
+    // hostnames. By default, only alphanumeric characters as well as '-' is
+    // allowed.
+    isValidHostname = function (hostname) {
+      return isValid(hostname, function (code) {
+        return code === 45 || code === 95;
+        //              ^ - (dash)     ^ _ (underscore)
+      });
+    };
+  }
+
   /**
    * Process a given url and extract all information. This is a higher level API
    * around private functions of `tld.js`. It allows to remove duplication (only

lib/is-valid.js

@@ -37,7 +37,12 @@ function isAlpha(code) {
  * @param {string} hostname
  * @return {boolean}
  */
-module.exports = function isValid(hostname) {
+module.exports = function isValid(hostname, _extraValidCode) {
+  var extraValidCode = function (code) { return code === 45; };
+  if (_extraValidCode !== undefined) {
+    extraValidCode = _extraValidCode;
+  }
+
   if (typeof hostname !== 'string') {
     return false;
   }
@@ -72,13 +77,13 @@ module.exports = function isValid(hostname) {
         // Check that previous character was not already a '.'
         lastCharCode === 46 ||
         // Check that the previous label does not end with a '-'
-        lastCharCode === 45
+        extraValidCode(lastCharCode)
       ) {
         return false;
       }
 
       lastDotIndex = i;
-    } else if (!(isAlpha(code) || isDigit(code) || code === 45)) {
+    } else if (!(isAlpha(code) || isDigit(code) || extraValidCode(code))) {
       // Check if there is a forbidden character in the label: [^a-zA-Z0-9-]
       return false;
     }

test/tld.js

@@ -70,6 +70,9 @@ describe('tld.js', function () {
       expect(tld.isValid('example.' + repeat('a', 63) + '.')).to.be(true);
       expect(tld.isValid('example.' + repeat('a', 63))).to.be(true);
 
+      // Rejects domains with '_'
+      expect(tld.isValid('foo.bar_baz.com')).to.be(false);
+
       //@see https://github.com/oncletom/tld.js/issues/95
       expect(tld.isValid('miam.miam.google.com.')).to.be(true);
 
@@ -92,6 +95,12 @@ describe('tld.js', function () {
       expect(tld.isValid('.google.com')).to.be(false);
       expect(tld.isValid('.com')).to.be(false);
     });
+
+    it('should accept extra code points in domain with lenien mode', function () {
+      // @see https://github.com/oncletom/tld.js/pull/122
+      var customTld = tld.fromUserSettings({ lenientHostnameValidation: true });
+      expect(customTld.isValid('foo.bar_baz.com')).to.be(true);
+    });
   });
 
   describe('isIp method', function () {