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 "chrome/browser/autocomplete/bookmark_provider.h" 6 7 #include <algorithm> 8 #include <string> 9 #include <vector> 10 11 #include "base/memory/ref_counted.h" 12 #include "base/memory/scoped_ptr.h" 13 #include "base/strings/string16.h" 14 #include "base/strings/string_number_conversions.h" 15 #include "base/strings/string_split.h" 16 #include "base/strings/utf_string_conversions.h" 17 #include "chrome/browser/autocomplete/autocomplete_provider.h" 18 #include "chrome/browser/autocomplete/autocomplete_provider_listener.h" 19 #include "chrome/test/base/testing_profile.h" 20 #include "components/bookmarks/browser/bookmark_match.h" 21 #include "components/bookmarks/browser/bookmark_model.h" 22 #include "components/bookmarks/test/test_bookmark_client.h" 23 #include "components/metrics/proto/omnibox_event.pb.h" 24 #include "testing/gtest/include/gtest/gtest.h" 25 26 using bookmarks::BookmarkMatch; 27 28 // The bookmark corpus against which we will simulate searches. 29 struct BookmarksTestInfo { 30 std::string title; 31 std::string url; 32 } bookmark_provider_test_data[] = { 33 { "abc def", "http://www.catsanddogs.com/a" }, 34 { "abcde", "http://www.catsanddogs.com/b" }, 35 { "abcdef", "http://www.catsanddogs.com/c" }, 36 { "a definition", "http://www.catsanddogs.com/d" }, 37 { "carry carbon carefully", "http://www.catsanddogs.com/e" }, 38 { "ghi jkl", "http://www.catsanddogs.com/f" }, 39 { "jkl ghi", "http://www.catsanddogs.com/g" }, 40 { "frankly frankly frank", "http://www.catsanddogs.com/h" }, 41 { "foobar foobar", "http://www.foobar.com/" }, 42 { "domain", "http://www.domain.com/http/" }, 43 { "repeat", "http://www.repeat.com/1/repeat/2/" }, 44 // For testing inline_autocompletion. 45 { "http://blah.com/", "http://blah.com/" }, 46 { "http://fiddle.com/", "http://fiddle.com/" }, 47 { "http://www.www.com/", "http://www.www.com/" }, 48 { "chrome://version", "chrome://version" }, 49 { "chrome://omnibox", "chrome://omnibox" }, 50 // For testing ranking with different URLs. 51 {"achlorhydric featherheads resuscitates mockingbirds", 52 "http://www.featherheads.com/a" }, 53 {"achlorhydric mockingbirds resuscitates featherhead", 54 "http://www.featherheads.com/b" }, 55 {"featherhead resuscitates achlorhydric mockingbirds", 56 "http://www.featherheads.com/c" }, 57 {"mockingbirds resuscitates featherheads achlorhydric", 58 "http://www.featherheads.com/d" }, 59 // For testing URL boosting. 60 {"burning worms #1", "http://www.burned.com/" }, 61 {"burning worms #2", "http://www.worms.com/" }, 62 {"worming burns #10", "http://www.burned.com/" }, 63 {"worming burns #20", "http://www.worms.com/" }, 64 {"jive music", "http://www.worms.com/" }, 65 }; 66 67 class BookmarkProviderTest : public testing::Test, 68 public AutocompleteProviderListener { 69 public: 70 BookmarkProviderTest(); 71 72 // AutocompleteProviderListener: Not called. 73 virtual void OnProviderUpdate(bool updated_matches) OVERRIDE {} 74 75 protected: 76 virtual void SetUp() OVERRIDE; 77 78 test::TestBookmarkClient client_; 79 scoped_ptr<TestingProfile> profile_; 80 scoped_ptr<BookmarkModel> model_; 81 scoped_refptr<BookmarkProvider> provider_; 82 83 private: 84 DISALLOW_COPY_AND_ASSIGN(BookmarkProviderTest); 85 }; 86 87 BookmarkProviderTest::BookmarkProviderTest() { 88 model_ = client_.CreateModel(false); 89 } 90 91 void BookmarkProviderTest::SetUp() { 92 profile_.reset(new TestingProfile()); 93 DCHECK(profile_.get()); 94 provider_ = new BookmarkProvider(this, profile_.get()); 95 DCHECK(provider_.get()); 96 provider_->set_bookmark_model_for_testing(model_.get()); 97 98 const BookmarkNode* other_node = model_->other_node(); 99 for (size_t i = 0; i < ARRAYSIZE_UNSAFE(bookmark_provider_test_data); ++i) { 100 const BookmarksTestInfo& cur(bookmark_provider_test_data[i]); 101 const GURL url(cur.url); 102 model_->AddURL(other_node, other_node->child_count(), 103 base::ASCIIToUTF16(cur.title), url); 104 } 105 } 106 107 // Structures and functions supporting the BookmarkProviderTest.Positions 108 // unit test. 109 110 struct TestBookmarkPosition { 111 TestBookmarkPosition(size_t begin, size_t end) 112 : begin(begin), end(end) {} 113 114 size_t begin; 115 size_t end; 116 }; 117 typedef std::vector<TestBookmarkPosition> TestBookmarkPositions; 118 119 // Return |positions| as a formatted string for unit test diagnostic output. 120 std::string TestBookmarkPositionsAsString( 121 const TestBookmarkPositions& positions) { 122 std::string position_string("{"); 123 for (TestBookmarkPositions::const_iterator i = positions.begin(); 124 i != positions.end(); ++i) { 125 if (i != positions.begin()) 126 position_string += ", "; 127 position_string += "{" + base::IntToString(i->begin) + ", " + 128 base::IntToString(i->end) + "}"; 129 } 130 position_string += "}\n"; 131 return position_string; 132 } 133 134 // Return the positions in |matches| as a formatted string for unit test 135 // diagnostic output. 136 base::string16 MatchesAsString16(const ACMatches& matches) { 137 base::string16 matches_string; 138 for (ACMatches::const_iterator i = matches.begin(); i != matches.end(); ++i) { 139 matches_string.append(base::ASCIIToUTF16(" '")); 140 matches_string.append(i->description); 141 matches_string.append(base::ASCIIToUTF16("'\n")); 142 } 143 return matches_string; 144 } 145 146 // Comparison function for sorting search terms by descending length. 147 bool TestBookmarkPositionsEqual(const TestBookmarkPosition& pos_a, 148 const TestBookmarkPosition& pos_b) { 149 return pos_a.begin == pos_b.begin && pos_a.end == pos_b.end; 150 } 151 152 // Convience function to make comparing ACMatchClassifications against the 153 // test expectations structure easier. 154 TestBookmarkPositions PositionsFromAutocompleteMatch( 155 const AutocompleteMatch& match) { 156 TestBookmarkPositions positions; 157 bool started = false; 158 size_t start = 0; 159 for (AutocompleteMatch::ACMatchClassifications::const_iterator 160 i = match.description_class.begin(); 161 i != match.description_class.end(); ++i) { 162 if (i->style & AutocompleteMatch::ACMatchClassification::MATCH) { 163 // We have found the start of a match. 164 EXPECT_FALSE(started); 165 started = true; 166 start = i->offset; 167 } else if (started) { 168 // We have found the end of a match. 169 started = false; 170 positions.push_back(TestBookmarkPosition(start, i->offset)); 171 start = 0; 172 } 173 } 174 // Record the final position if the last match goes to the end of the 175 // candidate string. 176 if (started) 177 positions.push_back(TestBookmarkPosition(start, match.description.size())); 178 return positions; 179 } 180 181 // Convience function to make comparing test expectations structure against the 182 // actual ACMatchClassifications easier. 183 TestBookmarkPositions PositionsFromExpectations( 184 const size_t expectations[9][2]) { 185 TestBookmarkPositions positions; 186 size_t i = 0; 187 // The array is zero-terminated in the [1]th element. 188 while (expectations[i][1]) { 189 positions.push_back( 190 TestBookmarkPosition(expectations[i][0], expectations[i][1])); 191 ++i; 192 } 193 return positions; 194 } 195 196 TEST_F(BookmarkProviderTest, Positions) { 197 // Simulate searches. 198 // Description of |positions|: 199 // The first index represents the collection of positions for each expected 200 // match. The count of the actual subarrays in each instance of |query_data| 201 // must equal |match_count|. The second index represents each expected 202 // match position. The third index represents the |start| and |end| of the 203 // expected match's position within the |test_data|. This array must be 204 // terminated by an entry with a value of '0' for |end|. 205 // Example: 206 // Consider the line for 'def' below: 207 // {"def", 2, {{{4, 7}, {XXX, 0}}, {{2, 5}, {11, 14}, {XXX, 0}}}}, 208 // There are two expected matches: 209 // 0. {{4, 7}, {XXX, 0}} 210 // 1. {{2, 5}, {11 ,14}, {XXX, 0}} 211 // For the first match, [0], there is one match within the bookmark's title 212 // expected, {4, 7}, which maps to the 'def' within "abc def". The 'XXX' 213 // value is ignored. The second match, [1], indicates that two matches are 214 // expected within the bookmark title "a definite definition". In each case, 215 // the {XXX, 0} indicates the end of the subarray. Or: 216 // Match #1 Match #2 217 // ------------------ ---------------------------- 218 // Pos1 Term Pos1 Pos2 Term 219 // ------ -------- ------ -------- -------- 220 // {"def", 2, {{{4, 7}, {999, 0}}, {{2, 5}, {11, 14}, {999, 0}}}}, 221 // 222 struct QueryData { 223 const std::string query; 224 const size_t match_count; // This count must match the number of major 225 // elements in the following |positions| array. 226 const size_t positions[99][9][2]; 227 } query_data[] = { 228 // This first set is primarily for position detection validation. 229 {"abc", 3, {{{0, 3}, {0, 0}}, 230 {{0, 3}, {0, 0}}, 231 {{0, 3}, {0, 0}}}}, 232 {"abcde", 2, {{{0, 5}, {0, 0}}, 233 {{0, 5}, {0, 0}}}}, 234 {"foo bar", 0, {{{0, 0}}}}, 235 {"fooey bark", 0, {{{0, 0}}}}, 236 {"def", 2, {{{2, 5}, {0, 0}}, 237 {{4, 7}, {0, 0}}}}, 238 {"ghi jkl", 2, {{{0, 3}, {4, 7}, {0, 0}}, 239 {{0, 3}, {4, 7}, {0, 0}}}}, 240 // NB: GetBookmarksWithTitlesMatching(...) uses exact match for "a". 241 {"a", 1, {{{0, 1}, {0, 0}}}}, 242 {"a d", 0, {{{0, 0}}}}, 243 {"carry carbon", 1, {{{0, 5}, {6, 12}, {0, 0}}}}, 244 // NB: GetBookmarksWithTitlesMatching(...) sorts the match positions. 245 {"carbon carry", 1, {{{0, 5}, {6, 12}, {0, 0}}}}, 246 {"arbon", 0, {{{0, 0}}}}, 247 {"ar", 0, {{{0, 0}}}}, 248 {"arry", 0, {{{0, 0}}}}, 249 // Quoted terms are single terms. 250 {"\"carry carbon\"", 1, {{{0, 12}, {0, 0}}}}, 251 {"\"carry carbon\" care", 1, {{{0, 12}, {13, 17}, {0, 0}}}}, 252 // Quoted terms require complete word matches. 253 {"\"carry carbo\"", 0, {{{0, 0}}}}, 254 // This set uses duplicated and/or overlaps search terms in the title. 255 {"frank", 1, {{{0, 5}, {8, 13}, {16, 21}, {0, 0}}}}, 256 {"frankly", 1, {{{0, 7}, {8, 15}, {0, 0}}}}, 257 {"frankly frankly", 1, {{{0, 7}, {8, 15}, {0, 0}}}}, 258 {"foobar foo", 1, {{{0, 6}, {7, 13}, {0, 0}}}}, 259 {"foo foobar", 1, {{{0, 6}, {7, 13}, {0, 0}}}}, 260 }; 261 262 for (size_t i = 0; i < ARRAYSIZE_UNSAFE(query_data); ++i) { 263 AutocompleteInput input(base::ASCIIToUTF16(query_data[i].query), 264 base::string16::npos, base::string16(), GURL(), 265 metrics::OmniboxEventProto::INVALID_SPEC, false, 266 false, false, true); 267 provider_->Start(input, false); 268 const ACMatches& matches(provider_->matches()); 269 // Validate number of results is as expected. 270 EXPECT_LE(matches.size(), query_data[i].match_count) 271 << "One or more of the following matches were unexpected:\n" 272 << MatchesAsString16(matches) 273 << "For query '" << query_data[i].query << "'."; 274 EXPECT_GE(matches.size(), query_data[i].match_count) 275 << "One or more expected matches are missing. Matches found:\n" 276 << MatchesAsString16(matches) 277 << "for query '" << query_data[i].query << "'."; 278 // Validate positions within each match is as expected. 279 for (size_t j = 0; j < matches.size(); ++j) { 280 // Collect the expected positions as a vector, collect the match's 281 // classifications for match positions as a vector, then compare. 282 TestBookmarkPositions expected_positions( 283 PositionsFromExpectations(query_data[i].positions[j])); 284 TestBookmarkPositions actual_positions( 285 PositionsFromAutocompleteMatch(matches[j])); 286 EXPECT_TRUE(std::equal(expected_positions.begin(), 287 expected_positions.end(), 288 actual_positions.begin(), 289 TestBookmarkPositionsEqual)) 290 << "EXPECTED: " << TestBookmarkPositionsAsString(expected_positions) 291 << "ACTUAL: " << TestBookmarkPositionsAsString(actual_positions) 292 << " for query: '" << query_data[i].query << "'."; 293 } 294 } 295 } 296 297 TEST_F(BookmarkProviderTest, Rankings) { 298 // Simulate searches. 299 struct QueryData { 300 const std::string query; 301 // |match_count| must match the number of elements in the following 302 // |matches| array. 303 const size_t match_count; 304 // |matches| specifies the titles for all bookmarks expected to be matched 305 // by the |query| 306 const std::string matches[3]; 307 } query_data[] = { 308 // Basic ranking test. 309 {"abc", 3, {"abcde", // Most complete match. 310 "abcdef", 311 "abc def"}}, // Least complete match. 312 {"ghi", 2, {"ghi jkl", // Matched earlier. 313 "jkl ghi", // Matched later. 314 ""}}, 315 // Rankings of exact-word matches with different URLs. 316 {"achlorhydric", 317 3, {"achlorhydric mockingbirds resuscitates featherhead", 318 "achlorhydric featherheads resuscitates mockingbirds", 319 "featherhead resuscitates achlorhydric mockingbirds"}}, 320 {"achlorhydric featherheads", 321 2, {"achlorhydric featherheads resuscitates mockingbirds", 322 "mockingbirds resuscitates featherheads achlorhydric", 323 ""}}, 324 {"mockingbirds resuscitates", 325 3, {"mockingbirds resuscitates featherheads achlorhydric", 326 "achlorhydric mockingbirds resuscitates featherhead", 327 "featherhead resuscitates achlorhydric mockingbirds"}}, 328 // Ranking of exact-word matches with URL boost. 329 {"worms", 2, {"burning worms #2", // boosted 330 "burning worms #1", // not boosted 331 ""}}, 332 // Ranking of prefix matches with URL boost. Note that a query of 333 // "worm burn" will have the same results. 334 {"burn worm", 3, {"burning worms #2", // boosted 335 "worming burns #20", // boosted 336 "burning worms #1"}}, // not boosted but shorter 337 }; 338 339 for (size_t i = 0; i < ARRAYSIZE_UNSAFE(query_data); ++i) { 340 AutocompleteInput input(base::ASCIIToUTF16(query_data[i].query), 341 base::string16::npos, base::string16(), GURL(), 342 metrics::OmniboxEventProto::INVALID_SPEC, false, 343 false, false, true); 344 provider_->Start(input, false); 345 const ACMatches& matches(provider_->matches()); 346 // Validate number and content of results is as expected. 347 for (size_t j = 0; j < std::max(query_data[i].match_count, matches.size()); 348 ++j) { 349 EXPECT_LT(j, query_data[i].match_count) << " Unexpected match '" 350 << base::UTF16ToUTF8(matches[j].description) << "' for query: '" 351 << query_data[i].query << "'."; 352 if (j >= query_data[i].match_count) 353 continue; 354 EXPECT_LT(j, matches.size()) << " Missing match '" 355 << query_data[i].matches[j] << "' for query: '" 356 << query_data[i].query << "'."; 357 if (j >= matches.size()) 358 continue; 359 EXPECT_EQ(query_data[i].matches[j], 360 base::UTF16ToUTF8(matches[j].description)) 361 << " Mismatch at [" << base::IntToString(j) << "] for query '" 362 << query_data[i].query << "'."; 363 } 364 } 365 } 366 367 TEST_F(BookmarkProviderTest, InlineAutocompletion) { 368 // Simulate searches. 369 struct QueryData { 370 const std::string query; 371 const std::string url; 372 const bool allowed_to_be_default_match; 373 const std::string inline_autocompletion; 374 } query_data[] = { 375 { "bla", "http://blah.com/", true, "h.com" }, 376 { "blah ", "http://blah.com/", false, ".com" }, 377 { "http://bl", "http://blah.com/", true, "ah.com" }, 378 { "fiddle.c", "http://fiddle.com/", true, "om" }, 379 { "www", "http://www.www.com/", true, ".com" }, 380 { "chro", "chrome://version", true, "me://version" }, 381 { "chrome://ve", "chrome://version", true, "rsion" }, 382 { "chrome ver", "chrome://version", false, "" }, 383 { "versi", "chrome://version", false, "" }, 384 { "abou", "chrome://omnibox", false, "" }, 385 { "about:om", "chrome://omnibox", true, "nibox" } 386 // Note: when adding a new URL to this test, be sure to add it to the list 387 // of bookmarks at the top of the file as well. All items in this list 388 // need to be in the bookmarks list because BookmarkProvider's 389 // TitleMatchToACMatch() has an assertion that verifies the URL is 390 // actually bookmarked. 391 }; 392 393 for (size_t i = 0; i < ARRAYSIZE_UNSAFE(query_data); ++i) { 394 const std::string description = "for query=" + query_data[i].query + 395 " and url=" + query_data[i].url; 396 AutocompleteInput input(base::ASCIIToUTF16(query_data[i].query), 397 base::string16::npos, base::string16(), GURL(), 398 metrics::OmniboxEventProto::INVALID_SPEC, false, 399 false, false, true); 400 const base::string16 fixed_up_input( 401 provider_->FixupUserInput(input).second); 402 BookmarkNode node(GURL(query_data[i].url)); 403 node.SetTitle(base::ASCIIToUTF16(query_data[i].url)); 404 BookmarkMatch bookmark_match; 405 bookmark_match.node = &node; 406 const AutocompleteMatch& ac_match = provider_->BookmarkMatchToACMatch( 407 input, fixed_up_input, bookmark_match); 408 EXPECT_EQ(query_data[i].allowed_to_be_default_match, 409 ac_match.allowed_to_be_default_match) << description; 410 EXPECT_EQ(base::ASCIIToUTF16(query_data[i].inline_autocompletion), 411 ac_match.inline_autocompletion) << description; 412 } 413 } 414 415 TEST_F(BookmarkProviderTest, StripHttpAndAdjustOffsets) { 416 // Simulate searches. 417 struct QueryData { 418 const std::string query; 419 const std::string expected_contents; 420 // |expected_contents_class| is in format offset:style,offset:style,... 421 const std::string expected_contents_class; 422 } query_data[] = { 423 { "foo", "www.foobar.com", "0:1,4:3,7:1" }, 424 { "www foo", "www.foobar.com", "0:3,3:1,4:3,7:1" }, 425 { "foo www", "www.foobar.com", "0:3,3:1,4:3,7:1" }, 426 { "foo http", "http://www.foobar.com", "0:3,4:1,11:3,14:1" }, 427 { "blah", "blah.com", "0:3,4:1" }, 428 { "http blah", "http://blah.com", "0:3,4:1,7:3,11:1" }, 429 { "dom", "www.domain.com/http/", "0:1,4:3,7:1" }, 430 { "dom http", "http://www.domain.com/http/", 431 "0:3,4:1,11:3,14:1,22:3,26:1" }, 432 { "rep", "www.repeat.com/1/repeat/2/", "0:1,4:3,7:1,17:3,20:1" }, 433 { "versi", "chrome://version", "0:1,9:3,14:1" } 434 }; 435 436 // Reload the bookmarks index with |index_urls| == true. 437 model_ = client_.CreateModel(true); 438 SetUp(); 439 440 for (size_t i = 0; i < ARRAYSIZE_UNSAFE(query_data); ++i) { 441 std::string description = "for query=" + query_data[i].query; 442 AutocompleteInput input(base::ASCIIToUTF16(query_data[i].query), 443 base::string16::npos, base::string16(), GURL(), 444 metrics::OmniboxEventProto::INVALID_SPEC, false, 445 false, false, true); 446 provider_->Start(input, false); 447 const ACMatches& matches(provider_->matches()); 448 ASSERT_EQ(1U, matches.size()) << description; 449 const AutocompleteMatch& match = matches[0]; 450 EXPECT_EQ(base::ASCIIToUTF16(query_data[i].expected_contents), 451 match.contents) << description; 452 std::vector<std::string> class_strings; 453 base::SplitString( 454 query_data[i].expected_contents_class, ',', &class_strings); 455 ASSERT_EQ(class_strings.size(), match.contents_class.size()) 456 << description; 457 for (size_t i = 0; i < class_strings.size(); ++i) { 458 std::vector<std::string> chunks; 459 base::SplitString(class_strings[i], ':', &chunks); 460 ASSERT_EQ(2U, chunks.size()) << description; 461 size_t offset; 462 EXPECT_TRUE(base::StringToSizeT(chunks[0], &offset)) << description; 463 EXPECT_EQ(offset, match.contents_class[i].offset) << description; 464 int style; 465 EXPECT_TRUE(base::StringToInt(chunks[1], &style)) << description; 466 EXPECT_EQ(style, match.contents_class[i].style) << description; 467 } 468 } 469 } 470