chrome/browser/google/google_util_unittest.cc

   1 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include "base/command_line.h"
   6 #include "chrome/browser/google/google_url_tracker.h"
   7 #include "chrome/browser/google/google_util.h"
   8 #include "chrome/common/chrome_switches.h"
   9 #include "testing/gtest/include/gtest/gtest.h"
  10
  11 using google_util::IsGoogleDomainUrl;
  12
  13
  14 // Helpers --------------------------------------------------------------------
  15
  16 namespace {
  17
  18 // These functions merely provide brevity in the callers.
  19
  20 bool IsHomePage(const std::string& url) {
  21   return google_util::IsGoogleHomePageUrl(GURL(url));
  22 }
  23
  24 bool IsSearch(const std::string& url) {
  25   return google_util::IsGoogleSearchUrl(GURL(url));
  26 }
  27
  28 bool StartsWithBaseURL(const std::string& url) {
  29   return google_util::StartsWithCommandLineGoogleBaseURL(GURL(url));
  30 }
  31
  32 }  // namespace
  33
  34
  35 // Actual tests ---------------------------------------------------------------
  36
  37 TEST(GoogleUtilTest, GoodHomePagesNonSecure) {
  38   // Valid home page hosts.
  39   EXPECT_TRUE(IsHomePage(GoogleURLTracker::kDefaultGoogleHomepage));
  40   EXPECT_TRUE(IsHomePage("http://google.com"));
  41   EXPECT_TRUE(IsHomePage("http://www.google.com"));
  42   EXPECT_TRUE(IsHomePage("http://www.google.ca"));
  43   EXPECT_TRUE(IsHomePage("http://www.google.co.uk"));
  44   EXPECT_TRUE(IsHomePage("http://www.google.com:80/"));
  45
  46   // Only the paths /, /webhp, and /ig.* are valid.  Query parameters are
  47   // ignored.
  48   EXPECT_TRUE(IsHomePage("http://www.google.com/"));
  49   EXPECT_TRUE(IsHomePage("http://www.google.com/webhp"));
  50   EXPECT_TRUE(IsHomePage("http://www.google.com/webhp?rlz=TEST"));
  51   EXPECT_TRUE(IsHomePage("http://www.google.com/ig"));
  52   EXPECT_TRUE(IsHomePage("http://www.google.com/ig/foo"));
  53   EXPECT_TRUE(IsHomePage("http://www.google.com/ig?rlz=TEST"));
  54   EXPECT_TRUE(IsHomePage("http://www.google.com/ig/foo?rlz=TEST"));
  55 }
  56
  57 TEST(GoogleUtilTest, GoodHomePagesSecure) {
  58   // Valid home page hosts.
  59   EXPECT_TRUE(IsHomePage("https://google.com"));
  60   EXPECT_TRUE(IsHomePage("https://www.google.com"));
  61   EXPECT_TRUE(IsHomePage("https://www.google.ca"));
  62   EXPECT_TRUE(IsHomePage("https://www.google.co.uk"));
  63   EXPECT_TRUE(IsHomePage("https://www.google.com:443/"));
  64
  65   // Only the paths /, /webhp, and /ig.* are valid.  Query parameters are
  66   // ignored.
  67   EXPECT_TRUE(IsHomePage("https://www.google.com/"));
  68   EXPECT_TRUE(IsHomePage("https://www.google.com/webhp"));
  69   EXPECT_TRUE(IsHomePage("https://www.google.com/webhp?rlz=TEST"));
  70   EXPECT_TRUE(IsHomePage("https://www.google.com/ig"));
  71   EXPECT_TRUE(IsHomePage("https://www.google.com/ig/foo"));
  72   EXPECT_TRUE(IsHomePage("https://www.google.com/ig?rlz=TEST"));
  73   EXPECT_TRUE(IsHomePage("https://www.google.com/ig/foo?rlz=TEST"));
  74 }
  75
  76 TEST(GoogleUtilTest, BadHomePages) {
  77   EXPECT_FALSE(IsHomePage(std::string()));
  78
  79   // If specified, only the "www" subdomain is OK.
  80   EXPECT_FALSE(IsHomePage("http://maps.google.com"));
  81   EXPECT_FALSE(IsHomePage("http://foo.google.com"));
  82
  83   // No non-standard port numbers.
  84   EXPECT_FALSE(IsHomePage("http://www.google.com:1234"));
  85   EXPECT_FALSE(IsHomePage("https://www.google.com:5678"));
  86
  87   // Invalid TLDs.
  88   EXPECT_FALSE(IsHomePage("http://www.google.abc"));
  89   EXPECT_FALSE(IsHomePage("http://www.google.com.abc"));
  90   EXPECT_FALSE(IsHomePage("http://www.google.abc.com"));
  91   EXPECT_FALSE(IsHomePage("http://www.google.ab.cd"));
  92   EXPECT_FALSE(IsHomePage("http://www.google.uk.qq"));
  93
  94   // Must be http or https.
  95   EXPECT_FALSE(IsHomePage("ftp://www.google.com"));
  96   EXPECT_FALSE(IsHomePage("file://does/not/exist"));
  97   EXPECT_FALSE(IsHomePage("bad://www.google.com"));
  98   EXPECT_FALSE(IsHomePage("www.google.com"));
  99
 100   // Only the paths /, /webhp, and /ig.* are valid.
 101   EXPECT_FALSE(IsHomePage("http://www.google.com/abc"));
 102   EXPECT_FALSE(IsHomePage("http://www.google.com/webhpabc"));
 103   EXPECT_FALSE(IsHomePage("http://www.google.com/webhp/abc"));
 104   EXPECT_FALSE(IsHomePage("http://www.google.com/abcig"));
 105   EXPECT_FALSE(IsHomePage("http://www.google.com/webhp/ig"));
 106
 107   // A search URL should not be identified as a home page URL.
 108   EXPECT_FALSE(IsHomePage("http://www.google.com/search?q=something"));
 109
 110   // Path is case sensitive.
 111   EXPECT_FALSE(IsHomePage("https://www.google.com/WEBHP"));
 112 }
 113
 114 TEST(GoogleUtilTest, GoodSearchPagesNonSecure) {
 115   // Queries with path "/search" need to have the query parameter in either
 116   // the url parameter or the hash fragment.
 117   EXPECT_TRUE(IsSearch("http://www.google.com/search?q=something"));
 118   EXPECT_TRUE(IsSearch("http://www.google.com/search#q=something"));
 119   EXPECT_TRUE(IsSearch("http://www.google.com/search?name=bob&q=something"));
 120   EXPECT_TRUE(IsSearch("http://www.google.com/search?name=bob#q=something"));
 121   EXPECT_TRUE(IsSearch("http://www.google.com/search?name=bob#age=24&q=thing"));
 122   EXPECT_TRUE(IsSearch("http://www.google.co.uk/search?q=something"));
 123   // It's actually valid for both to have the query parameter.
 124   EXPECT_TRUE(IsSearch("http://www.google.com/search?q=something#q=other"));
 125
 126   // Queries with path "/webhp", "/" or "" need to have the query parameter in
 127   // the hash fragment.
 128   EXPECT_TRUE(IsSearch("http://www.google.com/webhp#q=something"));
 129   EXPECT_TRUE(IsSearch("http://www.google.com/webhp#name=bob&q=something"));
 130   EXPECT_TRUE(IsSearch("http://www.google.com/webhp?name=bob#q=something"));
 131   EXPECT_TRUE(IsSearch("http://www.google.com/webhp?name=bob#age=24&q=thing"));
 132
 133   EXPECT_TRUE(IsSearch("http://www.google.com/#q=something"));
 134   EXPECT_TRUE(IsSearch("http://www.google.com/#name=bob&q=something"));
 135   EXPECT_TRUE(IsSearch("http://www.google.com/?name=bob#q=something"));
 136   EXPECT_TRUE(IsSearch("http://www.google.com/?name=bob#age=24&q=something"));
 137
 138   EXPECT_TRUE(IsSearch("http://www.google.com#q=something"));
 139   EXPECT_TRUE(IsSearch("http://www.google.com#name=bob&q=something"));
 140   EXPECT_TRUE(IsSearch("http://www.google.com?name=bob#q=something"));
 141   EXPECT_TRUE(IsSearch("http://www.google.com?name=bob#age=24&q=something"));
 142 }
 143
 144 TEST(GoogleUtilTest, GoodSearchPagesSecure) {
 145   // Queries with path "/search" need to have the query parameter in either
 146   // the url parameter or the hash fragment.
 147   EXPECT_TRUE(IsSearch("https://www.google.com/search?q=something"));
 148   EXPECT_TRUE(IsSearch("https://www.google.com/search#q=something"));
 149   EXPECT_TRUE(IsSearch("https://www.google.com/search?name=bob&q=something"));
 150   EXPECT_TRUE(IsSearch("https://www.google.com/search?name=bob#q=something"));
 151   EXPECT_TRUE(IsSearch("https://www.google.com/search?name=bob#age=24&q=q"));
 152   EXPECT_TRUE(IsSearch("https://www.google.co.uk/search?q=something"));
 153   // It's actually valid for both to have the query parameter.
 154   EXPECT_TRUE(IsSearch("https://www.google.com/search?q=something#q=other"));
 155
 156   // Queries with path "/webhp", "/" or "" need to have the query parameter in
 157   // the hash fragment.
 158   EXPECT_TRUE(IsSearch("https://www.google.com/webhp#q=something"));
 159   EXPECT_TRUE(IsSearch("https://www.google.com/webhp#name=bob&q=something"));
 160   EXPECT_TRUE(IsSearch("https://www.google.com/webhp?name=bob#q=something"));
 161   EXPECT_TRUE(IsSearch("https://www.google.com/webhp?name=bob#age=24&q=thing"));
 162
 163   EXPECT_TRUE(IsSearch("https://www.google.com/#q=something"));
 164   EXPECT_TRUE(IsSearch("https://www.google.com/#name=bob&q=something"));
 165   EXPECT_TRUE(IsSearch("https://www.google.com/?name=bob#q=something"));
 166   EXPECT_TRUE(IsSearch("https://www.google.com/?name=bob#age=24&q=something"));
 167
 168   EXPECT_TRUE(IsSearch("https://www.google.com#q=something"));
 169   EXPECT_TRUE(IsSearch("https://www.google.com#name=bob&q=something"));
 170   EXPECT_TRUE(IsSearch("https://www.google.com?name=bob#q=something"));
 171   EXPECT_TRUE(IsSearch("https://www.google.com?name=bob#age=24&q=something"));
 172 }
 173
 174 TEST(GoogleUtilTest, BadSearches) {
 175   // A home page URL should not be identified as a search URL.
 176   EXPECT_FALSE(IsSearch(GoogleURLTracker::kDefaultGoogleHomepage));
 177   EXPECT_FALSE(IsSearch("http://google.com"));
 178   EXPECT_FALSE(IsSearch("http://www.google.com"));
 179   EXPECT_FALSE(IsSearch("http://www.google.com/search"));
 180   EXPECT_FALSE(IsSearch("http://www.google.com/search?"));
 181
 182   // Must be http or https
 183   EXPECT_FALSE(IsSearch("ftp://www.google.com/search?q=something"));
 184   EXPECT_FALSE(IsSearch("file://does/not/exist/search?q=something"));
 185   EXPECT_FALSE(IsSearch("bad://www.google.com/search?q=something"));
 186   EXPECT_FALSE(IsSearch("www.google.com/search?q=something"));
 187
 188   // Can't have an empty query parameter.
 189   EXPECT_FALSE(IsSearch("http://www.google.com/search?q="));
 190   EXPECT_FALSE(IsSearch("http://www.google.com/search?name=bob&q="));
 191   EXPECT_FALSE(IsSearch("http://www.google.com/webhp#q="));
 192   EXPECT_FALSE(IsSearch("http://www.google.com/webhp#name=bob&q="));
 193
 194   // Home page searches without a hash fragment query parameter are invalid.
 195   EXPECT_FALSE(IsSearch("http://www.google.com/webhp?q=something"));
 196   EXPECT_FALSE(IsSearch("http://www.google.com/webhp?q=something#no=good"));
 197   EXPECT_FALSE(IsSearch("http://www.google.com/webhp?name=bob&q=something"));
 198   EXPECT_FALSE(IsSearch("http://www.google.com/?q=something"));
 199   EXPECT_FALSE(IsSearch("http://www.google.com?q=something"));
 200
 201   // Some paths are outright invalid as searches.
 202   EXPECT_FALSE(IsSearch("http://www.google.com/notreal?q=something"));
 203   EXPECT_FALSE(IsSearch("http://www.google.com/chrome?q=something"));
 204   EXPECT_FALSE(IsSearch("http://www.google.com/search/nogood?q=something"));
 205   EXPECT_FALSE(IsSearch("http://www.google.com/webhp/nogood#q=something"));
 206   EXPECT_FALSE(IsSearch(std::string()));
 207
 208   // Case sensitive paths.
 209   EXPECT_FALSE(IsSearch("http://www.google.com/SEARCH?q=something"));
 210   EXPECT_FALSE(IsSearch("http://www.google.com/WEBHP#q=something"));
 211 }
 212
 213 TEST(GoogleUtilTest, GoogleDomains) {
 214   // Test some good Google domains (valid TLDs).
 215   EXPECT_TRUE(IsGoogleDomainUrl(GURL("http://www.google.com"),
 216                                 google_util::ALLOW_SUBDOMAIN,
 217                                 google_util::DISALLOW_NON_STANDARD_PORTS));
 218   EXPECT_TRUE(IsGoogleDomainUrl(GURL("http://google.com"),
 219                                 google_util::ALLOW_SUBDOMAIN,
 220                                 google_util::DISALLOW_NON_STANDARD_PORTS));
 221   EXPECT_TRUE(IsGoogleDomainUrl(GURL("http://www.google.ca"),
 222                                 google_util::ALLOW_SUBDOMAIN,
 223                                 google_util::DISALLOW_NON_STANDARD_PORTS));
 224   EXPECT_TRUE(IsGoogleDomainUrl(GURL("http://www.google.biz.tj"),
 225                                 google_util::ALLOW_SUBDOMAIN,
 226                                 google_util::DISALLOW_NON_STANDARD_PORTS));
 227   EXPECT_TRUE(IsGoogleDomainUrl(GURL("http://www.google.com/search?q=thing"),
 228                                 google_util::ALLOW_SUBDOMAIN,
 229                                 google_util::DISALLOW_NON_STANDARD_PORTS));
 230   EXPECT_TRUE(IsGoogleDomainUrl(GURL("http://www.google.com/webhp"),
 231                                 google_util::ALLOW_SUBDOMAIN,
 232                                 google_util::DISALLOW_NON_STANDARD_PORTS));
 233
 234   // Test some bad Google domains (invalid TLDs).
 235   EXPECT_FALSE(IsGoogleDomainUrl(GURL("http://www.google.notrealtld"),
 236                                  google_util::ALLOW_SUBDOMAIN,
 237                                  google_util::DISALLOW_NON_STANDARD_PORTS));
 238   EXPECT_FALSE(IsGoogleDomainUrl(GURL("http://www.google.faketld/search?q=q"),
 239                                  google_util::ALLOW_SUBDOMAIN,
 240                                  google_util::DISALLOW_NON_STANDARD_PORTS));
 241   EXPECT_FALSE(IsGoogleDomainUrl(GURL("http://www.yahoo.com"),
 242                                  google_util::ALLOW_SUBDOMAIN,
 243                                  google_util::DISALLOW_NON_STANDARD_PORTS));
 244
 245   // Test subdomain checks.
 246   EXPECT_TRUE(IsGoogleDomainUrl(GURL("http://images.google.com"),
 247                                 google_util::ALLOW_SUBDOMAIN,
 248                                 google_util::DISALLOW_NON_STANDARD_PORTS));
 249   EXPECT_FALSE(IsGoogleDomainUrl(GURL("http://images.google.com"),
 250                                  google_util::DISALLOW_SUBDOMAIN,
 251                                  google_util::DISALLOW_NON_STANDARD_PORTS));
 252   EXPECT_TRUE(IsGoogleDomainUrl(GURL("http://google.com"),
 253                                 google_util::DISALLOW_SUBDOMAIN,
 254                                 google_util::DISALLOW_NON_STANDARD_PORTS));
 255   EXPECT_TRUE(IsGoogleDomainUrl(GURL("http://www.google.com"),
 256                                 google_util::DISALLOW_SUBDOMAIN,
 257                                 google_util::DISALLOW_NON_STANDARD_PORTS));
 258
 259   // Port and scheme checks.
 260   EXPECT_TRUE(IsGoogleDomainUrl(GURL("http://www.google.com:80"),
 261                                 google_util::DISALLOW_SUBDOMAIN,
 262                                 google_util::DISALLOW_NON_STANDARD_PORTS));
 263   EXPECT_FALSE(IsGoogleDomainUrl(GURL("http://www.google.com:123"),
 264                                  google_util::DISALLOW_SUBDOMAIN,
 265                                  google_util::DISALLOW_NON_STANDARD_PORTS));
 266   EXPECT_TRUE(IsGoogleDomainUrl(GURL("https://www.google.com:443"),
 267                                 google_util::DISALLOW_SUBDOMAIN,
 268                                 google_util::DISALLOW_NON_STANDARD_PORTS));
 269   EXPECT_FALSE(IsGoogleDomainUrl(GURL("http://www.google.com:123"),
 270                                  google_util::DISALLOW_SUBDOMAIN,
 271                                  google_util::DISALLOW_NON_STANDARD_PORTS));
 272   EXPECT_TRUE(IsGoogleDomainUrl(GURL("http://www.google.com:123"),
 273                                 google_util::DISALLOW_SUBDOMAIN,
 274                                 google_util::ALLOW_NON_STANDARD_PORTS));
 275   EXPECT_TRUE(IsGoogleDomainUrl(GURL("https://www.google.com:123"),
 276                                 google_util::DISALLOW_SUBDOMAIN,
 277                                 google_util::ALLOW_NON_STANDARD_PORTS));
 278   EXPECT_TRUE(IsGoogleDomainUrl(GURL("http://www.google.com:80"),
 279                                 google_util::DISALLOW_SUBDOMAIN,
 280                                 google_util::ALLOW_NON_STANDARD_PORTS));
 281   EXPECT_TRUE(IsGoogleDomainUrl(GURL("https://www.google.com:443"),
 282                                 google_util::DISALLOW_SUBDOMAIN,
 283                                 google_util::ALLOW_NON_STANDARD_PORTS));
 284   EXPECT_FALSE(IsGoogleDomainUrl(GURL("file://www.google.com"),
 285                                  google_util::DISALLOW_SUBDOMAIN,
 286                                  google_util::DISALLOW_NON_STANDARD_PORTS));
 287   EXPECT_FALSE(IsGoogleDomainUrl(GURL("doesnotexist://www.google.com"),
 288                                  google_util::DISALLOW_SUBDOMAIN,
 289                                  google_util::DISALLOW_NON_STANDARD_PORTS));
 290 }
 291
 292 TEST(GoogleUtilTest, GoogleBaseURLNotSpecified) {
 293   // When no command-line flag is specified, no input to
 294   // StartsWithCommandLineGoogleBaseURL() should return true.
 295   EXPECT_FALSE(StartsWithBaseURL(std::string()));
 296   EXPECT_FALSE(StartsWithBaseURL("http://www.foo.com/"));
 297   EXPECT_FALSE(StartsWithBaseURL("http://www.google.com/"));
 298
 299   // By default, none of the IsGoogleXXX functions should return true for a
 300   // "foo.com" URL.
 301   EXPECT_FALSE(IsGoogleHostname("www.foo.com",
 302                                 google_util::DISALLOW_SUBDOMAIN));
 303   EXPECT_FALSE(IsGoogleDomainUrl(GURL("http://www.foo.com/xyz"),
 304                                  google_util::DISALLOW_SUBDOMAIN,
 305                                  google_util::DISALLOW_NON_STANDARD_PORTS));
 306   EXPECT_FALSE(IsGoogleDomainUrl(GURL("https://www.foo.com/"),
 307                                  google_util::DISALLOW_SUBDOMAIN,
 308                                  google_util::DISALLOW_NON_STANDARD_PORTS));
 309   EXPECT_FALSE(IsHomePage("https://www.foo.com/webhp"));
 310   EXPECT_FALSE(IsSearch("http://www.foo.com/search?q=a"));
 311
 312   // Override the Google base URL on the command line.
 313   CommandLine::ForCurrentProcess()->AppendSwitchASCII(switches::kGoogleBaseURL,
 314                                                       "http://www.foo.com/");
 315
 316   // Only URLs which start with exactly the string on the command line should
 317   // cause StartsWithCommandLineGoogleBaseURL() to return true.
 318   EXPECT_FALSE(StartsWithBaseURL(std::string()));
 319   EXPECT_TRUE(StartsWithBaseURL("http://www.foo.com/"));
 320   EXPECT_TRUE(StartsWithBaseURL("http://www.foo.com/abc"));
 321   EXPECT_FALSE(StartsWithBaseURL("https://www.foo.com/"));
 322   EXPECT_FALSE(StartsWithBaseURL("http://www.google.com/"));
 323
 324   // The various IsGoogleXXX functions should respect the command-line flag.
 325   EXPECT_TRUE(IsGoogleHostname("www.foo.com", google_util::DISALLOW_SUBDOMAIN));
 326   EXPECT_FALSE(IsGoogleHostname("foo.com", google_util::ALLOW_SUBDOMAIN));
 327   EXPECT_TRUE(IsGoogleDomainUrl(GURL("http://www.foo.com/xyz"),
 328                                 google_util::DISALLOW_SUBDOMAIN,
 329                                 google_util::DISALLOW_NON_STANDARD_PORTS));
 330   EXPECT_TRUE(IsGoogleDomainUrl(GURL("https://www.foo.com/"),
 331                                 google_util::DISALLOW_SUBDOMAIN,
 332                                 google_util::DISALLOW_NON_STANDARD_PORTS));
 333   EXPECT_TRUE(IsHomePage("https://www.foo.com/webhp"));
 334   EXPECT_FALSE(IsHomePage("http://www.foo.com/xyz"));
 335   EXPECT_TRUE(IsSearch("http://www.foo.com/search?q=a"));
 336 }
 337
 338 TEST(GoogleUtilTest, GoogleBaseURLDisallowQuery) {
 339   CommandLine::ForCurrentProcess()->AppendSwitchASCII(switches::kGoogleBaseURL,
 340                                                       "http://www.foo.com/?q=");
 341   EXPECT_FALSE(google_util::CommandLineGoogleBaseURL().is_valid());
 342 }
 343
 344 TEST(GoogleUtilTest, GoogleBaseURLDisallowRef) {
 345   CommandLine::ForCurrentProcess()->AppendSwitchASCII(switches::kGoogleBaseURL,
 346                                                       "http://www.foo.com/#q=");
 347   EXPECT_FALSE(google_util::CommandLineGoogleBaseURL().is_valid());
 348 }
 349
 350 TEST(GoogleUtilTest, GoogleBaseURLFixup) {
 351   CommandLine::ForCurrentProcess()->AppendSwitchASCII(switches::kGoogleBaseURL,
 352                                                       "www.foo.com");
 353   ASSERT_TRUE(google_util::CommandLineGoogleBaseURL().is_valid());
 354   EXPECT_EQ("http://www.foo.com/",
 355             google_util::CommandLineGoogleBaseURL().spec());
 356 }