components/url_formatter/elide_url_unittest.cc

   1 // Copyright 2014 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 "components/url_formatter/elide_url.h"
   6
   7 #include "base/ios/ios_util.h"
   8 #include "base/strings/utf_string_conversions.h"
   9 #include "testing/gtest/include/gtest/gtest.h"
  10 #include "url/gurl.h"
  11
  12 #if !defined(OS_ANDROID)
  13 #include "ui/gfx/font_list.h"  // nogncheck
  14 #include "ui/gfx/text_elider.h"  // nogncheck
  15 #include "ui/gfx/text_utils.h"  // nogncheck
  16 #endif
  17
  18 namespace {
  19
  20 struct Testcase {
  21   const std::string input;
  22   const std::string output;
  23   enum SupportedPlatforms {
  24     ALL = 0,
  25     NO_IOS9_OR_LATER,
  26     NO_IOS,
  27   } platforms;
  28 };
  29
  30 #if !defined(OS_ANDROID)
  31 void RunUrlTest(Testcase* testcases, size_t num_testcases) {
  32   static const gfx::FontList font_list;
  33   for (size_t i = 0; i < num_testcases; ++i) {
  34     const GURL url(testcases[i].input);
  35     // Should we test with non-empty language list?
  36     // That's kinda redundant with net_util_unittests.
  37     const float available_width =
  38         gfx::GetStringWidthF(base::UTF8ToUTF16(testcases[i].output), font_list);
  39     EXPECT_EQ(base::UTF8ToUTF16(testcases[i].output),
  40               url_formatter::ElideUrl(url, font_list, available_width,
  41                                       std::string()));
  42   }
  43 }
  44
  45 // Test eliding of commonplace URLs.
  46 TEST(TextEliderTest, TestGeneralEliding) {
  47   const std::string kEllipsisStr(gfx::kEllipsis);
  48   Testcase testcases[] = {
  49       {"http://www.google.com/intl/en/ads/", "www.google.com/intl/en/ads/"},
  50       {"http://www.google.com/intl/en/ads/", "www.google.com/intl/en/ads/"},
  51       {"http://www.google.com/intl/en/ads/",
  52        "google.com/intl/" + kEllipsisStr + "/ads/"},
  53       {"http://www.google.com/intl/en/ads/",
  54        "google.com/" + kEllipsisStr + "/ads/"},
  55       {"http://www.google.com/intl/en/ads/", "google.com/" + kEllipsisStr},
  56       {"http://www.google.com/intl/en/ads/", "goog" + kEllipsisStr},
  57       {"https://subdomain.foo.com/bar/filename.html",
  58        "subdomain.foo.com/bar/filename.html"},
  59       {"https://subdomain.foo.com/bar/filename.html",
  60        "subdomain.foo.com/" + kEllipsisStr + "/filename.html"},
  61       {"http://subdomain.foo.com/bar/filename.html",
  62        kEllipsisStr + "foo.com/" + kEllipsisStr + "/filename.html"},
  63       {"http://www.google.com/intl/en/ads/?aLongQueryWhichIsNotRequired",
  64        "www.google.com/intl/en/ads/?aLongQ" + kEllipsisStr},
  65   };
  66
  67   RunUrlTest(testcases, arraysize(testcases));
  68 }
  69
  70 // When there is very little space available, the elision code will shorten
  71 // both path AND file name to an ellipsis - ".../...". To avoid this result,
  72 // there is a hack in place that simply treats them as one string in this
  73 // case.
  74 TEST(TextEliderTest, TestTrailingEllipsisSlashEllipsisHack) {
  75   const std::string kEllipsisStr(gfx::kEllipsis);
  76
  77   // Very little space, would cause double ellipsis.
  78   gfx::FontList font_list;
  79   GURL url("http://battersbox.com/directory/foo/peter_paul_and_mary.html");
  80   float available_width = gfx::GetStringWidthF(
  81       base::UTF8ToUTF16("battersbox.com/" + kEllipsisStr + "/" + kEllipsisStr),
  82       font_list);
  83
  84   // Create the expected string, after elision. Depending on font size, the
  85   // directory might become /dir... or /di... or/d... - it never should be
  86   // shorter than that. (If it is, the font considers d... to be longer
  87   // than .../... -  that should never happen).
  88   ASSERT_GT(
  89       gfx::GetStringWidthF(base::UTF8ToUTF16(kEllipsisStr + "/" + kEllipsisStr),
  90                            font_list),
  91       gfx::GetStringWidthF(base::UTF8ToUTF16("d" + kEllipsisStr), font_list));
  92   GURL long_url("http://battersbox.com/directorynameisreallylongtoforcetrunc");
  93   base::string16 expected = url_formatter::ElideUrl(
  94       long_url, font_list, available_width, std::string());
  95   // Ensure that the expected result still contains part of the directory name.
  96   ASSERT_GT(expected.length(), std::string("battersbox.com/d").length());
  97   EXPECT_EQ(expected, url_formatter::ElideUrl(url, font_list, available_width,
  98                                               std::string()));
  99
 100   // More space available - elide directories, partially elide filename.
 101   Testcase testcases[] = {
 102       {"http://battersbox.com/directory/foo/peter_paul_and_mary.html",
 103        "battersbox.com/" + kEllipsisStr + "/peter" + kEllipsisStr},
 104   };
 105   RunUrlTest(testcases, arraysize(testcases));
 106 }
 107
 108 // Test eliding of empty strings, URLs with ports, passwords, queries, etc.
 109 TEST(TextEliderTest, TestMoreEliding) {
 110   const std::string kEllipsisStr(gfx::kEllipsis);
 111   Testcase testcases[] = {
 112       {"http://www.google.com/foo?bar", "www.google.com/foo?bar"},
 113       {"http://xyz.google.com/foo?bar", "xyz.google.com/foo?" + kEllipsisStr},
 114       {"http://xyz.google.com/foo?bar", "xyz.google.com/foo" + kEllipsisStr},
 115       {"http://xyz.google.com/foo?bar", "xyz.google.com/fo" + kEllipsisStr},
 116       {"http://a.b.com/pathname/c?d", "a.b.com/" + kEllipsisStr + "/c?d"},
 117       {"", ""},
 118       {"http://foo.bar..example.com...hello/test/filename.html",
 119        "foo.bar..example.com...hello/" + kEllipsisStr + "/filename.html"},
 120       {"http://foo.bar../", "foo.bar.."},
 121       {"http://xn--1lq90i.cn/foo", "\xe5\x8c\x97\xe4\xba\xac.cn/foo"},
 122       {"http://me:mypass@secrethost.com:99/foo?bar#baz",
 123        "secrethost.com:99/foo?bar#baz"},
 124       {"http://me:mypass@ss%xxfdsf.com/foo", "ss%25xxfdsf.com/foo"},
 125       {"mailto:elgoato@elgoato.com", "mailto:elgoato@elgoato.com"},
 126       {"javascript:click(0)", "javascript:click(0)"},
 127       {"https://chess.eecs.berkeley.edu:4430/login/arbitfilename",
 128        "chess.eecs.berkeley.edu:4430/login/arbitfilename"},
 129       {"https://chess.eecs.berkeley.edu:4430/login/arbitfilename",
 130        kEllipsisStr + "berkeley.edu:4430/" + kEllipsisStr + "/arbitfilename"},
 131
 132       // Unescaping.
 133       {"http://www/%E4%BD%A0%E5%A5%BD?q=%E4%BD%A0%E5%A5%BD#\xe4\xbd\xa0",
 134        "www/\xe4\xbd\xa0\xe5\xa5\xbd?q=\xe4\xbd\xa0\xe5\xa5\xbd#\xe4\xbd\xa0"},
 135
 136       // Invalid unescaping for path. The ref will always be valid UTF-8. We
 137       // don't
 138       // bother to do too many edge cases, since these are handled by the
 139       // escaper
 140       // unittest.
 141       {"http://www/%E4%A0%E5%A5%BD?q=%E4%BD%A0%E5%A5%BD#\xe4\xbd\xa0",
 142        "www/%E4%A0%E5%A5%BD?q=\xe4\xbd\xa0\xe5\xa5\xbd#\xe4\xbd\xa0"},
 143   };
 144
 145   RunUrlTest(testcases, arraysize(testcases));
 146 }
 147
 148 // Test eliding of file: URLs.
 149 TEST(TextEliderTest, TestFileURLEliding) {
 150   const std::string kEllipsisStr(gfx::kEllipsis);
 151   Testcase testcases[] = {
 152     {"file:///C:/path1/path2/path3/filename",
 153      "file:///C:/path1/path2/path3/filename"},
 154     {"file:///C:/path1/path2/path3/filename", "C:/path1/path2/path3/filename"},
 155 // GURL parses "file:///C:path" differently on windows than it does on posix.
 156 #if defined(OS_WIN)
 157     {"file:///C:path1/path2/path3/filename",
 158      "C:/path1/path2/" + kEllipsisStr + "/filename"},
 159     {"file:///C:path1/path2/path3/filename",
 160      "C:/path1/" + kEllipsisStr + "/filename"},
 161     {"file:///C:path1/path2/path3/filename",
 162      "C:/" + kEllipsisStr + "/filename"},
 163 #endif  // defined(OS_WIN)
 164     {"file://filer/foo/bar/file", "filer/foo/bar/file"},
 165     {"file://filer/foo/bar/file", "filer/foo/" + kEllipsisStr + "/file"},
 166     {"file://filer/foo/bar/file", "filer/" + kEllipsisStr + "/file"},
 167     {"file://filer/foo/", "file://filer/foo/"},
 168     {"file://filer/foo/", "filer/foo/"},
 169     {"file://filer/foo/", "filer" + kEllipsisStr},
 170     // Eliding file URLs with nothing after the ':' shouldn't crash.
 171     {"file:///aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa:", "aaa" + kEllipsisStr},
 172     {"file:///aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa:/", "aaa" + kEllipsisStr},
 173   };
 174
 175   RunUrlTest(testcases, arraysize(testcases));
 176 }
 177
 178 TEST(TextEliderTest, TestHostEliding) {
 179   const std::string kEllipsisStr(gfx::kEllipsis);
 180   Testcase testcases[] = {
 181       {"http://google.com", "google.com"},
 182       // iOS width calculations are off by a letter from other platforms for
 183       // strings with too many kerned letters on the default font set.
 184       // TODO(rohitrao): Fix secure_display::ElideHost for iOS
 185       // (crbug.com/517604).
 186       {"http://subdomain.google.com", kEllipsisStr + ".google.com",
 187        Testcase::NO_IOS9_OR_LATER},
 188       {"http://reallyreallyreallylongdomainname.com",
 189        "reallyreallyreallylongdomainname.com"},
 190       {"http://a.b.c.d.e.f.com", kEllipsisStr + "f.com",
 191        Testcase::NO_IOS9_OR_LATER},
 192       {"http://foo", "foo"},
 193       {"http://foo.bar", "foo.bar"},
 194       {"http://subdomain.foo.bar", kEllipsisStr + "in.foo.bar",
 195        Testcase::NO_IOS9_OR_LATER},
 196       {"http://subdomain.reallylongdomainname.com",
 197        kEllipsisStr + "ain.reallylongdomainname.com", Testcase::NO_IOS},
 198       {"http://a.b.c.d.e.f.com", kEllipsisStr + ".e.f.com", Testcase::NO_IOS},
 199   };
 200
 201   for (size_t i = 0; i < arraysize(testcases); ++i) {
 202 #if defined(OS_IOS)
 203     if (testcases[i].platforms == Testcase::NO_IOS ||
 204         (testcases[i].platforms == Testcase::NO_IOS9_OR_LATER &&
 205          base::ios::IsRunningOnIOS9OrLater())) {
 206       continue;
 207     }
 208 #endif
 209     const float available_width = gfx::GetStringWidthF(
 210         base::UTF8ToUTF16(testcases[i].output), gfx::FontList());
 211     EXPECT_EQ(base::UTF8ToUTF16(testcases[i].output),
 212               url_formatter::ElideHost(GURL(testcases[i].input),
 213                                        gfx::FontList(), available_width));
 214   }
 215
 216   // Trying to elide to a really short length will still keep the full TLD+1
 217   EXPECT_EQ(
 218       base::ASCIIToUTF16("google.com"),
 219       url_formatter::ElideHost(GURL("http://google.com"), gfx::FontList(), 2));
 220   EXPECT_EQ(base::UTF8ToUTF16(kEllipsisStr + ".google.com"),
 221             url_formatter::ElideHost(GURL("http://subdomain.google.com"),
 222                                      gfx::FontList(), 2));
 223   EXPECT_EQ(
 224       base::ASCIIToUTF16("foo.bar"),
 225       url_formatter::ElideHost(GURL("http://foo.bar"), gfx::FontList(), 2));
 226 }
 227
 228 #endif  // !defined(OS_ANDROID)
 229
 230 TEST(TextEliderTest, FormatUrlForSecurityDisplay) {
 231   struct OriginTestData {
 232     const char* const description;
 233     const char* const input;
 234     const wchar_t* const output;
 235     const wchar_t* const output_omit_scheme;
 236   };
 237
 238   const OriginTestData tests[] = {
 239       {"Empty URL", "", L"", L""},
 240       {"HTTP URL", "http://www.google.com/", L"http://www.google.com",
 241        L"www.google.com"},
 242       {"HTTPS URL", "https://www.google.com/", L"https://www.google.com",
 243        L"www.google.com"},
 244       {"Standard HTTP port", "http://www.google.com:80/",
 245        L"http://www.google.com", L"www.google.com"},
 246       {"Standard HTTPS port", "https://www.google.com:443/",
 247        L"https://www.google.com", L"www.google.com"},
 248       {"Standard HTTP port, IDN Chinese",
 249        "http://\xe4\xb8\xad\xe5\x9b\xbd.icom.museum:80",
 250        L"http://xn--fiqs8s.icom.museum", L"xn--fiqs8s.icom.museum"},
 251       {"HTTP URL, IDN Hebrew (RTL)",
 252        "http://"
 253        "\xd7\x90\xd7\x99\xd7\xa7\xd7\x95\xd7\xb4\xd7\x9d."
 254        "\xd7\x99\xd7\xa9\xd7\xa8\xd7\x90\xd7\x9c.museum/",
 255        L"http://xn--4dbklr2c8d.xn--4dbrk0ce.museum",
 256        L"xn--4dbklr2c8d.xn--4dbrk0ce.museum"},
 257       {"HTTP URL with query string, IDN Arabic (RTL)",
 258        "http://\xd9\x85\xd8\xb5\xd8\xb1.icom.museum/foo.html?yes=no",
 259        L"http://xn--wgbh1c.icom.museum", L"xn--wgbh1c.icom.museum"},
 260       {"Non-standard HTTP port", "http://www.google.com:9000/",
 261        L"http://www.google.com:9000", L"www.google.com:9000"},
 262       {"Non-standard HTTPS port", "https://www.google.com:9000/",
 263        L"https://www.google.com:9000", L"www.google.com:9000"},
 264       {"File URI", "file:///usr/example/file.html",
 265        L"file:///usr/example/file.html", L"file:///usr/example/file.html"},
 266       {"File URI with hostname", "file://localhost/usr/example/file.html",
 267        L"file:///usr/example/file.html", L"file:///usr/example/file.html"},
 268       {"UNC File URI 1", "file:///CONTOSO/accounting/money.xls",
 269        L"file:///CONTOSO/accounting/money.xls",
 270        L"file:///CONTOSO/accounting/money.xls"},
 271       {"UNC File URI 2",
 272        "file:///C:/Program%20Files/Music/Web%20Sys/main.html?REQUEST=RADIO",
 273        L"file:///C:/Program%20Files/Music/Web%20Sys/main.html",
 274        L"file:///C:/Program%20Files/Music/Web%20Sys/main.html"},
 275       {"HTTP URL with path", "http://www.google.com/test.html",
 276        L"http://www.google.com", L"www.google.com"},
 277       {"HTTPS URL with path", "https://www.google.com/test.html",
 278        L"https://www.google.com", L"www.google.com"},
 279       {"Unusual secure scheme (wss)", "wss://www.google.com/",
 280        L"wss://www.google.com", L"wss://www.google.com"},
 281       {"Unusual non-secure scheme (gopher)", "gopher://www.google.com/",
 282        L"gopher://www.google.com", L"gopher://www.google.com"},
 283       {"Unlisted scheme (chrome)", "chrome://version", L"chrome://version",
 284        L"chrome://version"},
 285       {"HTTP IP address", "http://173.194.65.103", L"http://173.194.65.103",
 286        L"173.194.65.103"},
 287       {"HTTPS IP address", "https://173.194.65.103", L"https://173.194.65.103",
 288        L"173.194.65.103"},
 289       {"HTTP IPv6 address", "http://[FE80:0000:0000:0000:0202:B3FF:FE1E:8329]/",
 290        L"http://[fe80::202:b3ff:fe1e:8329]", L"[fe80::202:b3ff:fe1e:8329]"},
 291       {"HTTPS IPv6 address with port", "https://[2001:db8:0:1]:443/",
 292        L"https://[2001:db8:0:1]", L"https://[2001:db8:0:1]"},
 293       {"HTTPS IP address, non-default port", "https://173.194.65.103:8443",
 294        L"https://173.194.65.103:8443", L"173.194.65.103:8443"},
 295       {"HTTP filesystem: URL with path",
 296        "filesystem:http://www.google.com/temporary/test.html",
 297        L"filesystem:http://www.google.com",
 298        L"filesystem:http://www.google.com"},
 299       {"File filesystem: URL with path",
 300        "filesystem:file://localhost/temporary/stuff/test.html?z=fun&goat=billy",
 301        L"filesystem:file:///temporary/stuff/test.html",
 302        L"filesystem:file:///temporary/stuff/test.html"},
 303       {"Invalid scheme 1", "twelve://www.cyber.org/wow.php",
 304        L"twelve://www.cyber.org/wow.php", L"twelve://www.cyber.org/wow.php"},
 305       {"Invalid scheme 2", "://www.cyber.org/wow.php",
 306        L"://www.cyber.org/wow.php", L"://www.cyber.org/wow.php"},
 307       {"Invalid host 1", "https://www.cyber../wow.php", L"https://www.cyber..",
 308        L"www.cyber.."},
 309       {"Invalid host 2", "https://www...cyber/wow.php", L"https://www...cyber",
 310        L"www...cyber"},
 311       {"Invalid port 1", "https://173.194.65.103:000",
 312        L"https://173.194.65.103:0", L"173.194.65.103:0"},
 313       {"Invalid port 2", "https://173.194.65.103:gruffle",
 314        L"https://173.194.65.103:gruffle", L"https://173.194.65.103:gruffle"},
 315       {"Invalid port 3", "https://173.194.65.103:/hello.aspx",
 316        L"https://173.194.65.103", L"173.194.65.103"},
 317       {"Trailing dot in DNS name", "https://www.example.com./get/goat",
 318        L"https://www.example.com.", L"www.example.com."},
 319       {"Blob URL",
 320        "blob:http%3A//www.html5rocks.com/4d4ff040-6d61-4446-86d3-13ca07ec9ab9",
 321        L"blob:http%3A//www.html5rocks.com/"
 322        L"4d4ff040-6d61-4446-86d3-13ca07ec9ab9",
 323        L"blob:http%3A//www.html5rocks.com/"
 324        L"4d4ff040-6d61-4446-86d3-13ca07ec9ab9"}};
 325
 326   const char languages[] = "zh-TW,en-US,en,am,ar-EG,ar";
 327   for (size_t i = 0; i < arraysize(tests); ++i) {
 328     base::string16 formatted = url_formatter::FormatUrlForSecurityDisplay(
 329         GURL(tests[i].input), std::string());
 330     EXPECT_EQ(base::WideToUTF16(tests[i].output), formatted)
 331         << tests[i].description;
 332
 333     base::string16 formatted_omit_scheme =
 334         url_formatter::FormatUrlForSecurityDisplayOmitScheme(
 335             GURL(tests[i].input), std::string());
 336     EXPECT_EQ(base::WideToUTF16(tests[i].output_omit_scheme),
 337               formatted_omit_scheme)
 338         << tests[i].description;
 339
 340     base::string16 formatted_with_languages =
 341         url_formatter::FormatUrlForSecurityDisplay(GURL(tests[i].input),
 342                                                    languages);
 343     EXPECT_EQ(base::WideToUTF16(tests[i].output), formatted_with_languages)
 344         << tests[i].description;
 345
 346     base::string16 formatted_with_languages_omit_scheme =
 347         url_formatter::FormatUrlForSecurityDisplayOmitScheme(
 348             GURL(tests[i].input), languages);
 349     EXPECT_EQ(base::WideToUTF16(tests[i].output_omit_scheme),
 350               formatted_with_languages_omit_scheme)
 351         << tests[i].description;
 352   }
 353
 354   base::string16 formatted =
 355       url_formatter::FormatUrlForSecurityDisplay(GURL(), std::string());
 356   EXPECT_EQ(base::string16(), formatted)
 357       << "Explicitly test the 0-argument GURL constructor";
 358
 359   base::string16 formatted_omit_scheme =
 360       url_formatter::FormatUrlForSecurityDisplayOmitScheme(GURL(),
 361                                                            std::string());
 362   EXPECT_EQ(base::string16(), formatted_omit_scheme)
 363       << "Explicitly test the 0-argument GURL constructor";
 364 }
 365
 366 }  // namespace