Handle BOM. Also fix regexp for hyphenated languages. Fixes #144.

features/with_bom.feature

@@ -0,0 +1,53 @@
+Feature: User logs into the system
+  In order to be able to use eraNET components
+  As a user
+  I want to log in to the system
+
+
+  Scenario: Unlogged user sees welcome page with login
+    Given I have not logged or have logged out before
+    When I visit initial page
+    Then Default app should be loaded
+    And I should see login request
+    And I should not see any username
+
+
+  Scenario: Minimal user sees welcome page with its username and logout
+    Given I have logged as guest named "Guest"
+    When I visit initial page
+    Then Default app should be loaded
+    And I should see logout request
+    And I should see "Guest" as username
+
+
+  Scenario: Unlogged user logs in
+    Given I have not logged or have logged out before
+    And I have visited initial page
+    And I have seen login request
+    When I ask to log in
+    Then I should be taken to login page
+
+
+  Scenario: Logged user logs out
+    Given I have logged as guest named "Guest"
+    And I have visited initial page
+    And I have seen logout request
+    When I ask to log out
+    Then I should be taken to logout page
+
+
+  Scenario: Disconnected user sees welcome page and reconnect option
+    Given I have had a broken connection with api site
+    When I visit initial page
+    Then Default app should be loaded
+    And I should see reconnect request
+    And I should not see any username
+
+
+  Scenario: User sees 'connecting' while connecting
+    Given I have had lagging api site
+    When I visit initial page
+    Then I should see connecting message
+    And I should not see any username
+
+

lib/cucumber/gherkin_lexer.js

@@ -0,0 +1,19 @@
+var gherkin = require('gherkin');
+
+/**
+ * Small wrapper around Gherkin that strips BOM and loads the correct lexer based
+ * on the language header.
+ */
+module.exports = function GherkinLexer(content, gherkinListener) {
+  // Strip BOM
+  content = content.replace(/^\ufeff/g, '');
+
+  var languageMatch = /^\s*#\s*language:\s*([a-zA-Z-]+)\s*$/m.exec(content);
+  var language = languageMatch == null ? 'en' : languageMatch[1].toLowerCase();
+  var Lexer = gherkin.Lexer(language);
+  var lexer = new Lexer(gherkinListener);
+
+  this.scan = function() {
+    lexer.scan(content);
+  }
+};

lib/cucumber/parser.js

@@ -1,5 +1,5 @@
 var Parser = function(featureSources, astFilter) {
-  var Gherkin      = require('gherkin');
+  var GherkinLexer = require('./gherkin_lexer');
   var Cucumber     = require('../cucumber');
 
   var features     = Cucumber.Ast.Features();
@@ -8,22 +8,12 @@ var Parser = function(featureSources, astFilter) {
 
   var self = {
     parse: function parse() {
-      var lexers = {};
-      var lexer = function (lang) {
-        if (!(lang in lexers)) {
-          lexers[lang] = new (Gherkin.Lexer(lang))(self.getEventHandlers());
-        }
-
-        return lexers[lang];
-      };
-
       for (i in featureSources) {
         var currentSourceUri = featureSources[i][Parser.FEATURE_NAME_SOURCE_PAIR_URI_INDEX];
         var featureSource    = featureSources[i][Parser.FEATURE_NAME_SOURCE_PAIR_SOURCE_INDEX];
         self.setCurrentSourceUri(currentSourceUri);
-        var languageMatch = /^\s*#\s*language:\s*([a-z_]*)/.exec(featureSource.toString());
-        var language = languageMatch == null ? 'en' : languageMatch[1];
-        lexer(language).scan(featureSource);
+        var lexer = new GherkinLexer(featureSource.toString(), self.getEventHandlers());
+        lexer.scan();
       }
       return features;
     },

spec/cucumber/parser_spec.js

@@ -78,16 +78,6 @@ describe("Cucumber.Parser", function () {
       expect(parser.setCurrentSourceUri).toHaveBeenCalledWith(featureSources[1][0]);
     });
 
-    it("asks the English lexer to scan the first feature source", function () {
-      parser.parse();
-      expect(gherkinENLexer.scan).toHaveBeenCalledWith(featureSources[0][1]);
-    });
-
-    it("asks the French lexer to scan the second feature source", function () {
-      parser.parse();
-      expect(gherkinFRLexer.scan).toHaveBeenCalledWith(featureSources[1][1]);
-    });
-
     it("returns the features root element", function () {
       expect(parser.parse()).toBe(features);
     });