Rewrite filename guessing logic

 - Closes #261

gulpfile.js

@@ -1,7 +1,8 @@
 var gulp     = require('gulp'),
     jshint   = require('gulp-jshint'),
-    jsonlint = require("gulp-jsonlint");
+    jsonlint = require("gulp-jsonlint"),
     sass     = require('gulp-sass'),
+    karma    = require('karma'),
     zip      = require('gulp-zip');
 
 var sources = {
@@ -21,6 +22,19 @@ var sources = {
   ]
 };
 
+// Default Karma configuration.
+var karmaConfig = {
+  browsers: ['FirefoxHeadless'],
+  customLaunchers: {
+    FirefoxHeadless: {
+      base: 'Firefox',
+      flags: [ '-headless' ],
+    },
+  },
+  frameworks: [ 'jasmine' ],
+  singleRun: true
+};
+
 gulp.task('default', [ 'lint', 'jsonlint', 'sass', 'watch' ]);
 
 gulp.task('watch', [ 'sass' ], function () {
@@ -47,6 +61,17 @@ gulp.task('jsonlint', function () {
     .pipe(jsonlint.reporter());
 });
 
+gulp.task('test', function (done) {
+  // background/helpers.js
+  new karma.Server(Object.assign(karmaConfig, {
+    files: [
+      'test/background/webex-mocks.js',
+      'src/background/helpers.js',
+      'test/background/helpers.spec.js'
+    ]
+  }), done).start();
+});
+
 gulp.task('dist', [ 'sass' ], function () {
   return gulp.src(sources.dist)
     .pipe(zip('foxygestures.xpi', {

package.json

@@ -21,8 +21,12 @@
     "gulp-jsonlint": "^1.2.1",
     "gulp-sass": "^3.1.0",
     "gulp-zip": "^4.0.0",
+    "jasmine-core": "^3.1.0",
     "jshint": "^2.9.4",
-    "jshint-stylish": "^2.2.1"
+    "jshint-stylish": "^2.2.1",
+    "karma": "^2.0.4",
+    "karma-firefox-launcher": "^1.1.0",
+    "karma-jasmine": "^1.1.2"
   },
   "dependencies": {
     "bootstrap-sass": "^3.3.7"

src/background/helpers.js

@@ -26,14 +26,15 @@ modules.helpers = (function (module) {
     // Document
     'text/plain': '.txt',
     'text/html': '.html',
-    'text/css' : '.css',
-    'text/javascript' : '.js',
+    'text/css': '.css',
+    'text/javascript': '.js',
+    'text/json': '.json',
     // Image
-    'image/png' : '.png',
-    'image/jpeg' : '.jpg',
-    'image/gif' : '.gif',
-    'image/bmp' : '.bmp',
-    'image/webp' : '.webp',
+    'image/png': '.png',
+    'image/jpeg': '.jpg',
+    'image/gif': '.gif',
+    'image/bmp': '.bmp',
+    'image/webp': '.webp',
     // Video
     'video/mp4': '.mp4',
     'video/ogg': '.ogg',
@@ -48,6 +49,7 @@ modules.helpers = (function (module) {
     'audio/wave': '.wav',
     'audio/webm': '.webm',
     // Other
+    'application/json': '.json',
     'application/octet-stream': '.bin'
   };
 
@@ -85,9 +87,14 @@ modules.helpers = (function (module) {
     return format;
   };
 
+  // Remove invalid characters in a Windows path.
+  module.cleanPath = (input, replace = '') => {
+    return input.replace(/[\\/:"*?<>|]+/gi, replace);
+  };
+
   // Attempt to determine the filename from a media URL. If the media source does not contain a file extension but the
   // mime type is known, select the extension automatically.
-  module.suggestFilename = (mediaSource, mediaType) => {
+  module.suggestFilename = (mediaSource, mediaType = null) => {
     // Data URIs do not have a file name so try generate a name like 'data.ext' using mime type.
     if (mediaSource.startsWith('data:')) {
       // Extract the mime type if present.
@@ -98,19 +105,30 @@ modules.helpers = (function (module) {
       return (mime === '') ? 'data.txt' : 'data' + (mimeToExtensionMap[mime] || '.bin');
     }
 
-    // Extract the filename from the URL.
-    let match = /\/([^\/?#]+)($|\?|#)/i.exec(decodeURI(mediaSource));
-    if (match && match[1]) {
-      // Extract the extension from the filename.
-      match = /([^.]+)(\.[a-z0-9]+)?/i.exec(match[1]);
-      if (match && match[2]) {
-        // Filename seems to have an extension
-        return match[1] + match[2];
-      } else {
-        // Try to guess the extension from the type.
-        return match[1] + (mimeToExtensionMap[mediaType] || '');
+    try {
+      // Extract the filename from the URL.
+      // Take everything from the final / to the query, fragment, or end of URL.
+      mediaSource = decodeURI(mediaSource);
+      let match = /\/([^\/?#]+)($|\?|#)/i.exec(mediaSource);
+      if (mediaSource && match && match[1]) {
+        let basename = match[1];
+        basename = decodeURIComponent(basename);
+        basename = module.cleanPath(basename, ' ');
+
+        // Split the basename into file and extension.
+        let lastDot = basename.lastIndexOf('.');
+        if (!!~lastDot)  {
+          let filename = basename.substring(0, lastDot);
+          let extension = basename.substring(lastDot);
+
+          // Trim any trailing text from the extension.
+          return filename + /\.\w+/.exec(extension);
+        } else {
+          // Try to guess the extension from the type.
+          return basename + (mimeToExtensionMap[mediaType] || '');
+        }
       }
-    }
+    } catch (error) {}
 
     // Couldn't determine the filename; let the browser guess.
     return null;

test/background/helpers.spec.js

@@ -0,0 +1,75 @@
+'use strict';
+
+describe("background/helper.js", function() {
+
+  var helpers;
+
+  beforeEach(function () {
+    helpers = modules.helpers;
+  });
+
+  it("should suggest filenames for normal URLs", function() {
+    // Return null so the browser can guess the filename.
+    expect(helpers.suggestFilename('https://www.example.com/'))
+      .toBe(null);
+
+    // Simple URLs with normal features.
+    expect(helpers.suggestFilename('https://www.example.com/example.html'))
+      .toBe('example.html');
+    expect(helpers.suggestFilename('https://www.example.com/example.html?q=search'))
+      .toBe('example.html');
+    expect(helpers.suggestFilename('https://www.example.com/example.html?q=search#anchor'))
+      .toBe('example.html');
+    expect(helpers.suggestFilename('https://www.example.com/example.html#anchor'))
+      .toBe('example.html');
+    expect(helpers.suggestFilename('https://www.example.com/example.jpg'))
+      .toBe('example.jpg');
+    expect(helpers.suggestFilename('https://www.example.com/subdir/example.jpg'))
+      .toBe('example.jpg');
+    expect(helpers.suggestFilename('https://www.example.com/subdir/example'))
+      .toBe('example');
+    expect(helpers.suggestFilename('https://www.example.com/subdir/video.mp4'))
+      .toBe('video.mp4');
+
+    // Simple URLs with characters that are not valid for paths.
+    expect(helpers.suggestFilename('https://www.example.com/bad:file.txt'))
+      .toBe('bad file.txt');
+    expect(helpers.suggestFilename('https://www.example.com/bad%3Ffile.txt'))
+      .toBe('bad file.txt');
+
+    // Examples from #251 - trailing text after extension.
+    expect(helpers.suggestFilename('https://pbs.twimg.com/media/Deh3bIAXUAAfyxP.jpg:large'))
+      .toBe('Deh3bIAXUAAfyxP.jpg');
+
+    // Examples from #261 - repeated dots in filename.
+    expect(helpers.suggestFilename('https://www.example.com/string01.string02_string03.jpg'))
+      .toBe('string01.string02_string03.jpg');
+    expect(helpers.suggestFilename('https://www.example.com/string01.string02/string03.string04_string05.jpg'))
+      .toBe('string03.string04_string05.jpg');
+
+    // Video with no extension but mime-type is known.
+    expect(helpers.suggestFilename('https://www.example.com/sample_video', 'video/webm'))
+      .toBe('sample_video.webm');
+  });
+
+  it("should suggest filenames for data: URLs", function() {
+    // No mime-type should default to text.
+    expect(helpers.suggestFilename('data:,Hello%2C%20World!')).toBe('data.txt');
+
+    // Known mime-types should get valid extensions.
+    expect(helpers.suggestFilename('data:text/plain;base64,SGVsbG8sIFdvcmxkIQ%3D%3D')).toBe('data.txt');
+    expect(helpers.suggestFilename('data:text/html,%3Ch1%3EHello%2C%20World!%3C%2Fh1%3E')).toBe('data.html');
+    expect(helpers.suggestFilename('data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7')).toBe('data.gif');
+    expect(helpers.suggestFilename('data:application/octet-stream;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7')).toBe('data.bin');
+
+    // Unknown mime type defaults to .bin
+    expect(helpers.suggestFilename('data:unknown/mime;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7')).toBe('data.bin');
+  });
+
+  it("should suggest filenames and handle invalid URLs", function() {
+    // URL contains an invalid escape sequence.
+    expect(helpers.suggestFilename('https://www.example.com/bad-sequence%E0%A4%A.jpg'))
+      .toBe(null);
+  });
+
+});

test/background/webex-mocks.js

@@ -0,0 +1,5 @@
+var browser = {
+  i18n: jasmine.createSpyObj([
+    'getMessage'
+  ])
+};