chrome/utility/importer/bookmark_html_reader_unittest.cc

   1 // Copyright 2013 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 "chrome/utility/importer/bookmark_html_reader.h"
   6
   7 #include "base/bind.h"
   8 #include "base/bind_helpers.h"
   9 #include "base/callback.h"
  10 #include "base/files/file_path.h"
  11 #include "base/path_service.h"
  12 #include "base/strings/string16.h"
  13 #include "base/strings/string_util.h"
  14 #include "base/strings/utf_string_conversions.h"
  15 #include "chrome/common/chrome_paths.h"
  16 #include "chrome/common/importer/imported_bookmark_entry.h"
  17 #include "testing/gtest/include/gtest/gtest.h"
  18
  19 using base::ASCIIToUTF16;
  20 using base::UTF16ToWide;
  21
  22 namespace bookmark_html_reader {
  23
  24 TEST(BookmarkHTMLReaderTest, ParseTests) {
  25   bool result;
  26
  27   // Tests charset.
  28   std::string charset;
  29   result = internal::ParseCharsetFromLine(
  30       "<META HTTP-EQUIV=\"Content-Type\" "
  31       "CONTENT=\"text/html; charset=UTF-8\">",
  32       &charset);
  33   EXPECT_TRUE(result);
  34   EXPECT_EQ("UTF-8", charset);
  35
  36   // Escaped characters in name.
  37   base::string16 folder_name;
  38   bool is_toolbar_folder;
  39   base::Time folder_add_date;
  40   result = internal::ParseFolderNameFromLine(
  41       "<DT><H3 ADD_DATE=\"1207558707\" >&lt; &gt;"
  42       " &amp; &quot; &#39; \\ /</H3>",
  43       charset, &folder_name, &is_toolbar_folder, &folder_add_date);
  44   EXPECT_TRUE(result);
  45   EXPECT_EQ(ASCIIToUTF16("< > & \" ' \\ /"), folder_name);
  46   EXPECT_FALSE(is_toolbar_folder);
  47   EXPECT_TRUE(base::Time::FromTimeT(1207558707) == folder_add_date);
  48
  49   // Empty name and toolbar folder attribute.
  50   result = internal::ParseFolderNameFromLine(
  51       "<DT><H3 PERSONAL_TOOLBAR_FOLDER=\"true\"></H3>",
  52       charset, &folder_name, &is_toolbar_folder, &folder_add_date);
  53   EXPECT_TRUE(result);
  54   EXPECT_EQ(base::string16(), folder_name);
  55   EXPECT_TRUE(is_toolbar_folder);
  56
  57   // Unicode characters in title and shortcut.
  58   base::string16 title;
  59   GURL url, favicon;
  60   base::string16 shortcut;
  61   base::string16 post_data;
  62   base::Time add_date;
  63   result = internal::ParseBookmarkFromLine(
  64       "<DT><A HREF=\"http://chinese.site.cn/path?query=1#ref\" "
  65       "SHORTCUTURL=\"\xE4\xB8\xAD\">\xE4\xB8\xAD\xE6\x96\x87</A>",
  66       charset, &title, &url, &favicon, &shortcut, &add_date, &post_data);
  67   EXPECT_TRUE(result);
  68   EXPECT_EQ(L"\x4E2D\x6587", UTF16ToWide(title));
  69   EXPECT_EQ("http://chinese.site.cn/path?query=1#ref", url.spec());
  70   EXPECT_EQ(L"\x4E2D", UTF16ToWide(shortcut));
  71   EXPECT_EQ(base::string16(), post_data);
  72   EXPECT_TRUE(base::Time() == add_date);
  73
  74   // No shortcut, and url contains %22 ('"' character).
  75   result = internal::ParseBookmarkFromLine(
  76       "<DT><A HREF=\"http://domain.com/?q=%22<>%22\">name</A>",
  77       charset, &title, &url, &favicon, &shortcut, &add_date, &post_data);
  78   EXPECT_TRUE(result);
  79   EXPECT_EQ(ASCIIToUTF16("name"), title);
  80   EXPECT_EQ("http://domain.com/?q=%22%3C%3E%22", url.spec());
  81   EXPECT_EQ(base::string16(), shortcut);
  82   EXPECT_EQ(base::string16(), post_data);
  83   EXPECT_TRUE(base::Time() == add_date);
  84
  85   result = internal::ParseBookmarkFromLine(
  86       "<DT><A HREF=\"http://domain.com/?g=&quot;\"\">name</A>",
  87       charset, &title, &url, &favicon, &shortcut, &add_date, &post_data);
  88   EXPECT_TRUE(result);
  89   EXPECT_EQ(ASCIIToUTF16("name"), title);
  90   EXPECT_EQ("http://domain.com/?g=%22", url.spec());
  91   EXPECT_EQ(base::string16(), shortcut);
  92   EXPECT_EQ(base::string16(), post_data);
  93   EXPECT_TRUE(base::Time() == add_date);
  94
  95   // Creation date.
  96   result = internal::ParseBookmarkFromLine(
  97       "<DT><A HREF=\"http://site/\" ADD_DATE=\"1121301154\">name</A>",
  98       charset, &title, &url, &favicon, &shortcut, &add_date, &post_data);
  99   EXPECT_TRUE(result);
 100   EXPECT_EQ(ASCIIToUTF16("name"), title);
 101   EXPECT_EQ(GURL("http://site/"), url);
 102   EXPECT_EQ(base::string16(), shortcut);
 103   EXPECT_EQ(base::string16(), post_data);
 104   EXPECT_TRUE(base::Time::FromTimeT(1121301154) == add_date);
 105
 106   // Post-data
 107   result = internal::ParseBookmarkFromLine(
 108       "<DT><A HREF=\"http://localhost:8080/test/hello.html\" ADD_DATE=\""
 109       "1212447159\" LAST_VISIT=\"1212447251\" LAST_MODIFIED=\"1212447248\""
 110       "SHORTCUTURL=\"post\" ICON=\"data:\" POST_DATA=\"lname%3D%25s\""
 111       "LAST_CHARSET=\"UTF-8\" ID=\"rdf:#$weKaR3\">Test Post keyword</A>",
 112       charset, &title, &url, &favicon, &shortcut, &add_date, &post_data);
 113   EXPECT_TRUE(result);
 114   EXPECT_EQ(ASCIIToUTF16("Test Post keyword"), title);
 115   EXPECT_EQ("http://localhost:8080/test/hello.html", url.spec());
 116   EXPECT_EQ(ASCIIToUTF16("post"), shortcut);
 117   EXPECT_EQ(ASCIIToUTF16("lname%3D%25s"), post_data);
 118   EXPECT_TRUE(base::Time::FromTimeT(1212447159) == add_date);
 119
 120   // Invalid case.
 121   result = internal::ParseBookmarkFromLine(
 122       "<DT><A HREF=\"http://domain.com/?q=%22",
 123       charset, &title, &url, &favicon, &shortcut, &add_date, &post_data);
 124   EXPECT_FALSE(result);
 125   EXPECT_EQ(base::string16(), title);
 126   EXPECT_EQ("", url.spec());
 127   EXPECT_EQ(base::string16(), shortcut);
 128   EXPECT_EQ(base::string16(), post_data);
 129   EXPECT_TRUE(base::Time() == add_date);
 130
 131   // Epiphany format.
 132   result = internal::ParseMinimumBookmarkFromLine(
 133       "<dt><a href=\"http://www.google.com/\">Google</a></dt>",
 134       charset, &title, &url);
 135   EXPECT_TRUE(result);
 136   EXPECT_EQ(ASCIIToUTF16("Google"), title);
 137   EXPECT_EQ("http://www.google.com/", url.spec());
 138 }
 139
 140 TEST(BookmarkHTMLReaderTest, CanImportURLAsSearchEngineTest) {
 141   struct TestCase {
 142     const std::string url;
 143     const bool can_be_imported_as_search_engine;
 144   } test_cases[] = {
 145     { "http://www.example.%s.com", true },
 146     { "http://www.example.%S.com", true },
 147     { "http://www.example.%x.com", false },
 148     { "http://www.example.com", false },
 149     { "http://%s.example.com", true },
 150     { "http://www.example.%s.test.%s.com", true },
 151     { "http://www.test&test.%s.com", true },
 152     { "http://www.example.com?q=%s&foo=bar", true },
 153     { "http://www.example.com/%s/?q=%s&foo=bar", true },
 154     { "http//google.com", false },
 155     { "path", false },
 156     { "http:/path/%s/", true },
 157     { "path", false },
 158     { "", false },
 159   };
 160
 161   std::string search_engine_url;
 162   for (size_t i = 0; i < arraysize(test_cases); ++i) {
 163     EXPECT_EQ(test_cases[i].can_be_imported_as_search_engine,
 164         CanImportURLAsSearchEngine(GURL(test_cases[i].url),
 165                                    &search_engine_url));
 166   }
 167 }
 168
 169 namespace {
 170
 171 class BookmarkHTMLReaderTestWithData : public testing::Test {
 172  public:
 173   void SetUp() override;
 174
 175  protected:
 176   void ExpectFirstFirefox2Bookmark(const ImportedBookmarkEntry& entry);
 177   void ExpectSecondFirefox2Bookmark(const ImportedBookmarkEntry& entry);
 178   void ExpectThirdFirefox2Bookmark(const ImportedBookmarkEntry& entry);
 179   void ExpectFirstEpiphanyBookmark(const ImportedBookmarkEntry& entry);
 180   void ExpectSecondEpiphanyBookmark(const ImportedBookmarkEntry& entry);
 181   void ExpectFirstFirefox23Bookmark(const ImportedBookmarkEntry& entry);
 182   void ExpectSecondFirefox23Bookmark(const ImportedBookmarkEntry& entry);
 183   void ExpectThirdFirefox23Bookmark(const ImportedBookmarkEntry& entry);
 184   void ExpectFirstFirefoxBookmarkWithKeyword(
 185       const importer::SearchEngineInfo& info);
 186   void ExpectSecondFirefoxBookmarkWithKeyword(
 187       const importer::SearchEngineInfo& info);
 188
 189   base::FilePath test_data_path_;
 190 };
 191
 192 void BookmarkHTMLReaderTestWithData::SetUp() {
 193   ASSERT_TRUE(PathService::Get(chrome::DIR_TEST_DATA, &test_data_path_));
 194   test_data_path_ = test_data_path_.AppendASCII("bookmark_html_reader");
 195 }
 196
 197 void BookmarkHTMLReaderTestWithData::ExpectFirstFirefox2Bookmark(
 198     const ImportedBookmarkEntry& entry) {
 199   EXPECT_EQ(ASCIIToUTF16("Empty"), entry.title);
 200   EXPECT_TRUE(entry.is_folder);
 201   EXPECT_EQ(base::Time::FromTimeT(1295938143), entry.creation_time);
 202   EXPECT_EQ(1U, entry.path.size());
 203   if (entry.path.size() == 1)
 204     EXPECT_EQ(ASCIIToUTF16("Empty's Parent"), entry.path.front());
 205 }
 206
 207 void BookmarkHTMLReaderTestWithData::ExpectSecondFirefox2Bookmark(
 208     const ImportedBookmarkEntry& entry) {
 209   EXPECT_EQ(ASCIIToUTF16("[Tamura Yukari.com]"), entry.title);
 210   EXPECT_FALSE(entry.is_folder);
 211   EXPECT_EQ(base::Time::FromTimeT(1234567890), entry.creation_time);
 212   EXPECT_EQ(1U, entry.path.size());
 213   if (entry.path.size() == 1)
 214     EXPECT_EQ(ASCIIToUTF16("Not Empty"), entry.path.front());
 215   EXPECT_EQ("http://www.tamurayukari.com/", entry.url.spec());
 216 }
 217
 218 void BookmarkHTMLReaderTestWithData::ExpectThirdFirefox2Bookmark(
 219     const ImportedBookmarkEntry& entry) {
 220   EXPECT_EQ(ASCIIToUTF16("Google"), entry.title);
 221   EXPECT_FALSE(entry.is_folder);
 222   EXPECT_EQ(base::Time::FromTimeT(0000000000), entry.creation_time);
 223   EXPECT_EQ(1U, entry.path.size());
 224   if (entry.path.size() == 1)
 225     EXPECT_EQ(ASCIIToUTF16("Not Empty But Default"), entry.path.front());
 226   EXPECT_EQ("http://www.google.com/", entry.url.spec());
 227 }
 228
 229 void BookmarkHTMLReaderTestWithData::ExpectFirstEpiphanyBookmark(
 230     const ImportedBookmarkEntry& entry) {
 231   EXPECT_EQ(ASCIIToUTF16("[Tamura Yukari.com]"), entry.title);
 232   EXPECT_EQ("http://www.tamurayukari.com/", entry.url.spec());
 233   EXPECT_EQ(0U, entry.path.size());
 234 }
 235
 236 void BookmarkHTMLReaderTestWithData::ExpectSecondEpiphanyBookmark(
 237     const ImportedBookmarkEntry& entry) {
 238   EXPECT_EQ(ASCIIToUTF16("Google"), entry.title);
 239   EXPECT_EQ("http://www.google.com/", entry.url.spec());
 240   EXPECT_EQ(0U, entry.path.size());
 241 }
 242
 243 void BookmarkHTMLReaderTestWithData::ExpectFirstFirefox23Bookmark(
 244     const ImportedBookmarkEntry& entry) {
 245   EXPECT_EQ(ASCIIToUTF16("Google"), entry.title);
 246   EXPECT_FALSE(entry.is_folder);
 247   EXPECT_EQ(base::Time::FromTimeT(1376102167), entry.creation_time);
 248   EXPECT_EQ(0U, entry.path.size());
 249   EXPECT_EQ("https://www.google.com/", entry.url.spec());
 250 }
 251
 252 void BookmarkHTMLReaderTestWithData::ExpectSecondFirefox23Bookmark(
 253     const ImportedBookmarkEntry& entry) {
 254   EXPECT_EQ(ASCIIToUTF16("Issues"), entry.title);
 255   EXPECT_FALSE(entry.is_folder);
 256   EXPECT_EQ(base::Time::FromTimeT(1376102304), entry.creation_time);
 257   EXPECT_EQ(1U, entry.path.size());
 258   EXPECT_EQ(ASCIIToUTF16("Chromium"), entry.path.front());
 259   EXPECT_EQ("https://code.google.com/p/chromium/issues/list", entry.url.spec());
 260 }
 261
 262 void BookmarkHTMLReaderTestWithData::ExpectThirdFirefox23Bookmark(
 263     const ImportedBookmarkEntry& entry) {
 264   EXPECT_EQ(ASCIIToUTF16("CodeSearch"), entry.title);
 265   EXPECT_FALSE(entry.is_folder);
 266   EXPECT_EQ(base::Time::FromTimeT(1376102224), entry.creation_time);
 267   EXPECT_EQ(1U, entry.path.size());
 268   EXPECT_EQ(ASCIIToUTF16("Chromium"), entry.path.front());
 269   EXPECT_EQ("http://code.google.com/p/chromium/codesearch", entry.url.spec());
 270 }
 271
 272 void BookmarkHTMLReaderTestWithData::ExpectFirstFirefoxBookmarkWithKeyword(
 273     const importer::SearchEngineInfo& info) {
 274   EXPECT_EQ(ASCIIToUTF16("http://example.{searchTerms}.com/"), info.url);
 275   EXPECT_EQ(ASCIIToUTF16("keyword"), info.keyword);
 276   EXPECT_EQ(ASCIIToUTF16("Bookmark Keyword"), info.display_name);
 277 }
 278
 279 void BookmarkHTMLReaderTestWithData::ExpectSecondFirefoxBookmarkWithKeyword(
 280     const importer::SearchEngineInfo& info) {
 281   EXPECT_EQ(ASCIIToUTF16("http://example.com/?q={searchTerms}"), info.url);
 282   EXPECT_EQ(ASCIIToUTF16("keyword"), info.keyword);
 283   EXPECT_EQ(ASCIIToUTF16("BookmarkName"), info.display_name);
 284 }
 285
 286 }  // namespace
 287
 288 TEST_F(BookmarkHTMLReaderTestWithData, Firefox2BookmarkFileImport) {
 289   base::FilePath path = test_data_path_.AppendASCII("firefox2.html");
 290
 291   std::vector<ImportedBookmarkEntry> bookmarks;
 292   ImportBookmarksFile(base::Callback<bool(void)>(),
 293                       base::Callback<bool(const GURL&)>(),
 294                       path, &bookmarks, NULL, NULL);
 295
 296   ASSERT_EQ(3U, bookmarks.size());
 297   ExpectFirstFirefox2Bookmark(bookmarks[0]);
 298   ExpectSecondFirefox2Bookmark(bookmarks[1]);
 299   ExpectThirdFirefox2Bookmark(bookmarks[2]);
 300 }
 301
 302 TEST_F(BookmarkHTMLReaderTestWithData, BookmarkFileWithHrTagImport) {
 303   base::FilePath path = test_data_path_.AppendASCII("firefox23.html");
 304
 305   std::vector<ImportedBookmarkEntry> bookmarks;
 306   ImportBookmarksFile(base::Callback<bool(void)>(),
 307                       base::Callback<bool(const GURL&)>(),
 308                       path, &bookmarks, NULL, NULL);
 309
 310   ASSERT_EQ(3U, bookmarks.size());
 311   ExpectFirstFirefox23Bookmark(bookmarks[0]);
 312   ExpectSecondFirefox23Bookmark(bookmarks[1]);
 313   ExpectThirdFirefox23Bookmark(bookmarks[2]);
 314 }
 315
 316 TEST_F(BookmarkHTMLReaderTestWithData, EpiphanyBookmarkFileImport) {
 317   base::FilePath path = test_data_path_.AppendASCII("epiphany.html");
 318
 319   std::vector<ImportedBookmarkEntry> bookmarks;
 320   ImportBookmarksFile(base::Callback<bool(void)>(),
 321                       base::Callback<bool(const GURL&)>(),
 322                       path, &bookmarks, NULL, NULL);
 323
 324   ASSERT_EQ(2U, bookmarks.size());
 325   ExpectFirstEpiphanyBookmark(bookmarks[0]);
 326   ExpectSecondEpiphanyBookmark(bookmarks[1]);
 327 }
 328
 329 TEST_F(BookmarkHTMLReaderTestWithData, FirefoxBookmarkFileWithKeywordImport) {
 330   base::FilePath path = test_data_path_.AppendASCII(
 331       "firefox_bookmark_keyword.html");
 332
 333   std::vector<importer::SearchEngineInfo> search_engines;
 334   ImportBookmarksFile(base::Callback<bool(void)>(),
 335                       base::Callback<bool(const GURL&)>(),
 336                       path, NULL, &search_engines, NULL);
 337
 338   ASSERT_EQ(2U, search_engines.size());
 339   ExpectFirstFirefoxBookmarkWithKeyword(search_engines[0]);
 340   ExpectSecondFirefoxBookmarkWithKeyword(search_engines[1]);
 341 }
 342
 343 // Verifies that importing a bookmarks file without a charset specified succeeds
 344 // (by falling back to a default charset). Per [ http://crbug.com/460423 ], this
 345 // sort of bookmarks file is generated by IE.
 346 TEST_F(BookmarkHTMLReaderTestWithData,
 347        InternetExplorerBookmarkFileWithoutCharsetImport) {
 348   base::FilePath path = test_data_path_.AppendASCII("ie_sans_charset.html");
 349
 350   std::vector<ImportedBookmarkEntry> bookmarks;
 351   ImportBookmarksFile(base::Callback<bool(void)>(),
 352                       base::Callback<bool(const GURL&)>(),
 353                       path, &bookmarks, NULL, NULL);
 354
 355   ASSERT_EQ(3U, bookmarks.size());
 356   EXPECT_EQ(ASCIIToUTF16("Google"), bookmarks[0].title);
 357   EXPECT_EQ(ASCIIToUTF16("Outlook"), bookmarks[1].title);
 358   EXPECT_EQ(ASCIIToUTF16("Speed Test"), bookmarks[2].title);
 359 }
 360
 361 namespace {
 362
 363 class CancelAfterFifteenCalls {
 364   int count;
 365  public:
 366   CancelAfterFifteenCalls() : count(0) { }
 367   bool ShouldCancel() {
 368     return ++count > 16;
 369   }
 370 };
 371
 372 }  // namespace
 373
 374 TEST_F(BookmarkHTMLReaderTestWithData, CancellationCallback) {
 375   // Use a file for testing that has multiple bookmarks.
 376   base::FilePath path = test_data_path_.AppendASCII("firefox2.html");
 377
 378   std::vector<ImportedBookmarkEntry> bookmarks;
 379   CancelAfterFifteenCalls cancel_fifteen;
 380   ImportBookmarksFile(base::Bind(&CancelAfterFifteenCalls::ShouldCancel,
 381                                  base::Unretained(&cancel_fifteen)),
 382                       base::Callback<bool(const GURL&)>(),
 383                       path, &bookmarks, NULL, NULL);
 384
 385   // The cancellation callback is checked before each line is read, so fifteen
 386   // lines are imported. The first fifteen lines of firefox2.html include only
 387   // one bookmark.
 388   ASSERT_EQ(1U, bookmarks.size());
 389   ExpectFirstFirefox2Bookmark(bookmarks[0]);
 390 }
 391
 392 namespace {
 393
 394 bool IsURLValid(const GURL& url) {
 395   // No offense to whomever owns this domain...
 396   return !url.DomainIs("tamurayukari.com");
 397 }
 398
 399 }  // namespace
 400
 401 TEST_F(BookmarkHTMLReaderTestWithData, ValidURLCallback) {
 402   // Use a file for testing that has multiple bookmarks.
 403   base::FilePath path = test_data_path_.AppendASCII("firefox2.html");
 404
 405   std::vector<ImportedBookmarkEntry> bookmarks;
 406   ImportBookmarksFile(base::Callback<bool(void)>(),
 407                       base::Bind(&IsURLValid),
 408                       path, &bookmarks, NULL, NULL);
 409
 410   ASSERT_EQ(2U, bookmarks.size());
 411   ExpectFirstFirefox2Bookmark(bookmarks[0]);
 412   ExpectThirdFirefox2Bookmark(bookmarks[1]);
 413 }
 414
 415 }  // namespace bookmark_html_reader