?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Componentize component_updater: Copy over test data with executable bit.
[chromium-blink-merge.git] / chrome / browser / autocomplete / bookmark_provider_unittest.cc
blob241eea3f4bb60702644b4e8208e9efe389d5d100
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/chrome_autocomplete_scheme_classifier.h"
18 #include "chrome/test/base/testing_profile.h"
19 #include "components/bookmarks/browser/bookmark_match.h"
20 #include "components/bookmarks/browser/bookmark_model.h"
21 #include "components/bookmarks/test/test_bookmark_client.h"
22 #include "components/metrics/proto/omnibox_event.pb.h"
23 #include "components/omnibox/autocomplete_provider.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 { "carry carbon carefully", "http://www.catsanddogs.com/d" },
37 { "a definition", "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.manylongwords.com/1a" },
53 {"achlorhydric mockingbirds resuscitates featherhead",
54 "http://www.manylongwords.com/2b" },
55 {"featherhead resuscitates achlorhydric mockingbirds",
56 "http://www.manylongwords.com/3c" },
57 {"mockingbirds resuscitates featherheads achlorhydric",
58 "http://www.manylongwords.com/4d" },
59 // For testing URL boosting. (URLs referenced multiple times are boosted.)
60 {"burning worms #1", "http://www.burns.com/" },
61 {"burning worms #2", "http://www.worms.com/" },
62 {"worming burns #10", "http://www.burns.com/" },
63 };
64
65 class BookmarkProviderTest : public testing::Test {
66 public:
67 BookmarkProviderTest();
68
69 protected:
70 virtual void SetUp() OVERRIDE;
71
72 test::TestBookmarkClient client_;
73 scoped_ptr<TestingProfile> profile_;
74 scoped_ptr<BookmarkModel> model_;
75 scoped_refptr<BookmarkProvider> provider_;
76
77 private:
78 DISALLOW_COPY_AND_ASSIGN(BookmarkProviderTest);
79 };
80
81 BookmarkProviderTest::BookmarkProviderTest() {
82 model_ = client_.CreateModel();
83 }
84
85 void BookmarkProviderTest::SetUp() {
86 profile_.reset(new TestingProfile());
87 DCHECK(profile_.get());
88 provider_ = new BookmarkProvider(profile_.get());
89 DCHECK(provider_.get());
90 provider_->set_bookmark_model_for_testing(model_.get());
91
92 const BookmarkNode* other_node = model_->other_node();
93 for (size_t i = 0; i < ARRAYSIZE_UNSAFE(bookmark_provider_test_data); ++i) {
94 const BookmarksTestInfo& cur(bookmark_provider_test_data[i]);
95 const GURL url(cur.url);
96 model_->AddURL(other_node, other_node->child_count(),
97 base::ASCIIToUTF16(cur.title), url);
98 }
99 }
100
101 // Structures and functions supporting the BookmarkProviderTest.Positions
102 // unit test.
103
104 struct TestBookmarkPosition {
105 TestBookmarkPosition(size_t begin, size_t end)
106 : begin(begin), end(end) {}
107
108 size_t begin;
109 size_t end;
110 };
111 typedef std::vector<TestBookmarkPosition> TestBookmarkPositions;
112
113 // Return |positions| as a formatted string for unit test diagnostic output.
114 std::string TestBookmarkPositionsAsString(
115 const TestBookmarkPositions& positions) {
116 std::string position_string("{");
117 for (TestBookmarkPositions::const_iterator i = positions.begin();
118 i != positions.end(); ++i) {
119 if (i != positions.begin())
120 position_string += ", ";
121 position_string += "{" + base::IntToString(i->begin) + ", " +
122 base::IntToString(i->end) + "}";
123 }
124 position_string += "}\n";
125 return position_string;
126 }
127
128 // Return the positions in |matches| as a formatted string for unit test
129 // diagnostic output.
130 base::string16 MatchesAsString16(const ACMatches& matches) {
131 base::string16 matches_string;
132 for (ACMatches::const_iterator i = matches.begin(); i != matches.end(); ++i) {
133 matches_string.append(base::ASCIIToUTF16(" '"));
134 matches_string.append(i->description);
135 matches_string.append(base::ASCIIToUTF16("'\n"));
136 }
137 return matches_string;
138 }
139
140 // Comparison function for sorting search terms by descending length.
141 bool TestBookmarkPositionsEqual(const TestBookmarkPosition& pos_a,
142 const TestBookmarkPosition& pos_b) {
143 return pos_a.begin == pos_b.begin && pos_a.end == pos_b.end;
144 }
145
146 // Convience function to make comparing ACMatchClassifications against the
147 // test expectations structure easier.
148 TestBookmarkPositions PositionsFromAutocompleteMatch(
149 const AutocompleteMatch& match) {
150 TestBookmarkPositions positions;
151 bool started = false;
152 size_t start = 0;
153 for (AutocompleteMatch::ACMatchClassifications::const_iterator
154 i = match.description_class.begin();
155 i != match.description_class.end(); ++i) {
156 if (i->style & AutocompleteMatch::ACMatchClassification::MATCH) {
157 // We have found the start of a match.
158 EXPECT_FALSE(started);
159 started = true;
160 start = i->offset;
161 } else if (started) {
162 // We have found the end of a match.
163 started = false;
164 positions.push_back(TestBookmarkPosition(start, i->offset));
165 start = 0;
166 }
167 }
168 // Record the final position if the last match goes to the end of the
169 // candidate string.
170 if (started)
171 positions.push_back(TestBookmarkPosition(start, match.description.size()));
172 return positions;
173 }
174
175 // Convience function to make comparing test expectations structure against the
176 // actual ACMatchClassifications easier.
177 TestBookmarkPositions PositionsFromExpectations(
178 const size_t expectations[9][2]) {
179 TestBookmarkPositions positions;
180 size_t i = 0;
181 // The array is zero-terminated in the [1]th element.
182 while (expectations[i][1]) {
183 positions.push_back(
184 TestBookmarkPosition(expectations[i][0], expectations[i][1]));
185 ++i;
186 }
187 return positions;
188 }
189
190 TEST_F(BookmarkProviderTest, Positions) {
191 // Simulate searches.
192 // Description of |positions|:
193 // The first index represents the collection of positions for each expected
194 // match. The count of the actual subarrays in each instance of |query_data|
195 // must equal |match_count|. The second index represents each expected
196 // match position. The third index represents the |start| and |end| of the
197 // expected match's position within the |test_data|. This array must be
198 // terminated by an entry with a value of '0' for |end|.
199 // Example:
200 // Consider the line for 'def' below:
201 // {"def", 2, {{{4, 7}, {XXX, 0}}, {{2, 5}, {11, 14}, {XXX, 0}}}},
202 // There are two expected matches:
203 // 0. {{4, 7}, {XXX, 0}}
204 // 1. {{2, 5}, {11 ,14}, {XXX, 0}}
205 // For the first match, [0], there is one match within the bookmark's title
206 // expected, {4, 7}, which maps to the 'def' within "abc def". The 'XXX'
207 // value is ignored. The second match, [1], indicates that two matches are
208 // expected within the bookmark title "a definite definition". In each case,
209 // the {XXX, 0} indicates the end of the subarray. Or:
210 // Match #1 Match #2
211 // ------------------ ----------------------------
212 // Pos1 Term Pos1 Pos2 Term
213 // ------ -------- ------ -------- --------
214 // {"def", 2, {{{4, 7}, {999, 0}}, {{2, 5}, {11, 14}, {999, 0}}}},
215 //
216 struct QueryData {
217 const std::string query;
218 const size_t match_count; // This count must match the number of major
219 // elements in the following |positions| array.
220 const size_t positions[99][9][2];
221 } query_data[] = {
222 // This first set is primarily for position detection validation.
223 {"abc", 3, {{{0, 3}, {0, 0}},
224 {{0, 3}, {0, 0}},
225 {{0, 3}, {0, 0}}}},
226 {"abcde", 2, {{{0, 5}, {0, 0}},
227 {{0, 5}, {0, 0}}}},
228 {"foo bar", 0, {{{0, 0}}}},
229 {"fooey bark", 0, {{{0, 0}}}},
230 {"def", 2, {{{2, 5}, {0, 0}},
231 {{4, 7}, {0, 0}}}},
232 {"ghi jkl", 2, {{{0, 3}, {4, 7}, {0, 0}},
233 {{0, 3}, {4, 7}, {0, 0}}}},
234 // NB: GetBookmarksMatching(...) uses exact match for "a" in title or URL.
235 {"a", 2, {{{0, 1}, {0, 0}},
236 {{0, 0}}}},
237 {"a d", 0, {{{0, 0}}}},
238 {"carry carbon", 1, {{{0, 5}, {6, 12}, {0, 0}}}},
239 // NB: GetBookmarksMatching(...) sorts the match positions.
240 {"carbon carry", 1, {{{0, 5}, {6, 12}, {0, 0}}}},
241 {"arbon", 0, {{{0, 0}}}},
242 {"ar", 0, {{{0, 0}}}},
243 {"arry", 0, {{{0, 0}}}},
244 // Quoted terms are single terms.
245 {"\"carry carbon\"", 1, {{{0, 12}, {0, 0}}}},
246 {"\"carry carbon\" care", 1, {{{0, 12}, {13, 17}, {0, 0}}}},
247 // Quoted terms require complete word matches.
248 {"\"carry carbo\"", 0, {{{0, 0}}}},
249 // This set uses duplicated and/or overlaps search terms in the title.
250 {"frank", 1, {{{0, 5}, {8, 13}, {16, 21}, {0, 0}}}},
251 {"frankly", 1, {{{0, 7}, {8, 15}, {0, 0}}}},
252 {"frankly frankly", 1, {{{0, 7}, {8, 15}, {0, 0}}}},
253 {"foobar foo", 1, {{{0, 6}, {7, 13}, {0, 0}}}},
254 {"foo foobar", 1, {{{0, 6}, {7, 13}, {0, 0}}}},
255 };
256
257 for (size_t i = 0; i < ARRAYSIZE_UNSAFE(query_data); ++i) {
258 AutocompleteInput input(base::ASCIIToUTF16(query_data[i].query),
259 base::string16::npos, base::string16(), GURL(),
260 metrics::OmniboxEventProto::INVALID_SPEC, false,
261 false, false, true,
262 ChromeAutocompleteSchemeClassifier(profile_.get()));
263 provider_->Start(input, false);
264 const ACMatches& matches(provider_->matches());
265 // Validate number of results is as expected.
266 EXPECT_LE(matches.size(), query_data[i].match_count)
267 << "One or more of the following matches were unexpected:\n"
268 << MatchesAsString16(matches)
269 << "For query '" << query_data[i].query << "'.";
270 EXPECT_GE(matches.size(), query_data[i].match_count)
271 << "One or more expected matches are missing. Matches found:\n"
272 << MatchesAsString16(matches)
273 << "for query '" << query_data[i].query << "'.";
274 // Validate positions within each match is as expected.
275 for (size_t j = 0; j < matches.size(); ++j) {
276 // Collect the expected positions as a vector, collect the match's
277 // classifications for match positions as a vector, then compare.
278 TestBookmarkPositions expected_positions(
279 PositionsFromExpectations(query_data[i].positions[j]));
280 TestBookmarkPositions actual_positions(
281 PositionsFromAutocompleteMatch(matches[j]));
282 EXPECT_TRUE(std::equal(expected_positions.begin(),
283 expected_positions.end(),
284 actual_positions.begin(),
285 TestBookmarkPositionsEqual))
286 << "EXPECTED: " << TestBookmarkPositionsAsString(expected_positions)
287 << "ACTUAL: " << TestBookmarkPositionsAsString(actual_positions)
288 << " for query: '" << query_data[i].query << "'.";
289 }
290 }
291 }
292
293 TEST_F(BookmarkProviderTest, Rankings) {
294 // Simulate searches.
295 struct QueryData {
296 const std::string query;
297 // |match_count| must match the number of elements in the following
298 // |matches| array.
299 const size_t match_count;
300 // |matches| specifies the titles for all bookmarks expected to be matched
301 // by the |query|
302 const std::string matches[3];
303 } query_data[] = {
304 // Basic ranking test.
305 {"abc", 3, {"abcde", // Most complete match.
306 "abcdef",
307 "abc def"}}, // Least complete match.
308 {"ghi", 2, {"ghi jkl", // Matched earlier.
309 "jkl ghi", // Matched later.
310 ""}},
311 // Rankings of exact-word matches with different URLs.
312 {"achlorhydric",
313 3, {"achlorhydric mockingbirds resuscitates featherhead",
314 "achlorhydric featherheads resuscitates mockingbirds",
315 "featherhead resuscitates achlorhydric mockingbirds"}},
316 {"achlorhydric featherheads",
317 2, {"achlorhydric featherheads resuscitates mockingbirds",
318 "mockingbirds resuscitates featherheads achlorhydric",
319 ""}},
320 {"mockingbirds resuscitates",
321 3, {"mockingbirds resuscitates featherheads achlorhydric",
322 "achlorhydric mockingbirds resuscitates featherhead",
323 "featherhead resuscitates achlorhydric mockingbirds"}},
324 // Ranking of exact-word matches with URL boosts.
325 {"worms", 2, {"burning worms #1", // boosted
326 "burning worms #2", // not boosted
327 ""}},
328 // Ranking of prefix matches with URL boost.
329 {"burn worm", 3, {"burning worms #1", // boosted
330 "worming burns #10", // boosted but longer title
331 "burning worms #2"}}, // not boosted
332 // A query of "worm burn" will have the same results.
333 {"worm burn", 3, {"burning worms #1", // boosted
334 "worming burns #10", // boosted but longer title
335 "burning worms #2"}}, // not boosted
336 };
337
338 for (size_t i = 0; i < ARRAYSIZE_UNSAFE(query_data); ++i) {
339 AutocompleteInput input(base::ASCIIToUTF16(query_data[i].query),
340 base::string16::npos, base::string16(), GURL(),
341 metrics::OmniboxEventProto::INVALID_SPEC, false,
342 false, false, true,
343 ChromeAutocompleteSchemeClassifier(profile_.get()));
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 ChromeAutocompleteSchemeClassifier(profile_.get()));
401 const base::string16 fixed_up_input(
402 provider_->FixupUserInput(input).second);
403 BookmarkNode node(GURL(query_data[i].url));
404 node.SetTitle(base::ASCIIToUTF16(query_data[i].url));
405 BookmarkMatch bookmark_match;
406 bookmark_match.node = &node;
407 const AutocompleteMatch& ac_match = provider_->BookmarkMatchToACMatch(
408 input, fixed_up_input, bookmark_match);
409 EXPECT_EQ(query_data[i].allowed_to_be_default_match,
410 ac_match.allowed_to_be_default_match) << description;
411 EXPECT_EQ(base::ASCIIToUTF16(query_data[i].inline_autocompletion),
412 ac_match.inline_autocompletion) << description;
413 }
414 }
415
416 TEST_F(BookmarkProviderTest, StripHttpAndAdjustOffsets) {
417 // Simulate searches.
418 struct QueryData {
419 const std::string query;
420 const std::string expected_contents;
421 // |expected_contents_class| is in format offset:style,offset:style,...
422 const std::string expected_contents_class;
423 } query_data[] = {
424 { "foo", "www.foobar.com", "0:1,4:3,7:1" },
425 { "www foo", "www.foobar.com", "0:3,3:1,4:3,7:1" },
426 { "foo www", "www.foobar.com", "0:3,3:1,4:3,7:1" },
427 { "foo http", "http://www.foobar.com", "0:3,4:1,11:3,14:1" },
428 { "blah", "blah.com", "0:3,4:1" },
429 { "http blah", "http://blah.com", "0:3,4:1,7:3,11:1" },
430 { "dom", "www.domain.com/http/", "0:1,4:3,7:1" },
431 { "dom http", "http://www.domain.com/http/",
432 "0:3,4:1,11:3,14:1,22:3,26:1" },
433 { "rep", "www.repeat.com/1/repeat/2/", "0:1,4:3,7:1,17:3,20:1" },
434 { "versi", "chrome://version", "0:1,9:3,14:1" }
435 };
436
437 for (size_t i = 0; i < ARRAYSIZE_UNSAFE(query_data); ++i) {
438 std::string description = "for query=" + query_data[i].query;
439 AutocompleteInput input(base::ASCIIToUTF16(query_data[i].query),
440 base::string16::npos, base::string16(), GURL(),
441 metrics::OmniboxEventProto::INVALID_SPEC, false,
442 false, false, true,
443 ChromeAutocompleteSchemeClassifier(profile_.get()));
444 provider_->Start(input, false);
445 const ACMatches& matches(provider_->matches());
446 ASSERT_EQ(1U, matches.size()) << description;
447 const AutocompleteMatch& match = matches[0];
448 EXPECT_EQ(base::ASCIIToUTF16(query_data[i].expected_contents),
449 match.contents) << description;
450 std::vector<std::string> class_strings;
451 base::SplitString(
452 query_data[i].expected_contents_class, ',', &class_strings);
453 ASSERT_EQ(class_strings.size(), match.contents_class.size())
454 << description;
455 for (size_t i = 0; i < class_strings.size(); ++i) {
456 std::vector<std::string> chunks;
457 base::SplitString(class_strings[i], ':', &chunks);
458 ASSERT_EQ(2U, chunks.size()) << description;
459 size_t offset;
460 EXPECT_TRUE(base::StringToSizeT(chunks[0], &offset)) << description;
461 EXPECT_EQ(offset, match.contents_class[i].offset) << description;
462 int style;
463 EXPECT_TRUE(base::StringToInt(chunks[1], &style)) << description;
464 EXPECT_EQ(style, match.contents_class[i].style) << description;
465 }
466 }
467 }