chrome/browser/autocomplete/autocomplete_match.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 "chrome/browser/autocomplete/autocomplete_match.h"
   6
   7 #include "base/i18n/time_formatting.h"
   8 #include "base/logging.h"
   9 #include "base/strings/string16.h"
  10 #include "base/strings/string_number_conversions.h"
  11 #include "base/strings/string_util.h"
  12 #include "base/strings/utf_string_conversions.h"
  13 #include "base/time/time.h"
  14 #include "chrome/browser/autocomplete/autocomplete_provider.h"
  15 #include "chrome/browser/search_engines/template_url.h"
  16 #include "chrome/browser/search_engines/template_url_service.h"
  17 #include "chrome/browser/search_engines/template_url_service_factory.h"
  18 #include "content/public/common/url_constants.h"
  19 #include "grit/theme_resources.h"
  20
  21 namespace {
  22
  23 bool IsTrivialClassification(const ACMatchClassifications& classifications) {
  24   return classifications.empty() ||
  25       ((classifications.size() == 1) &&
  26        (classifications.back().style == ACMatchClassification::NONE));
  27 }
  28
  29 }  // namespace
  30
  31 // AutocompleteMatch ----------------------------------------------------------
  32
  33 // static
  34 const base::char16 AutocompleteMatch::kInvalidChars[] = {
  35   '\n', '\r', '\t',
  36   0x2028,  // Line separator
  37   0x2029,  // Paragraph separator
  38   0
  39 };
  40
  41 AutocompleteMatch::AutocompleteMatch()
  42     : provider(NULL),
  43       relevance(0),
  44       typed_count(-1),
  45       deletable(false),
  46       allowed_to_be_default_match(false),
  47       transition(content::PAGE_TRANSITION_GENERATED),
  48       is_history_what_you_typed_match(false),
  49       type(AutocompleteMatchType::SEARCH_WHAT_YOU_TYPED),
  50       starred(false),
  51       from_previous(false) {
  52 }
  53
  54 AutocompleteMatch::AutocompleteMatch(AutocompleteProvider* provider,
  55                                      int relevance,
  56                                      bool deletable,
  57                                      Type type)
  58     : provider(provider),
  59       relevance(relevance),
  60       typed_count(-1),
  61       deletable(deletable),
  62       allowed_to_be_default_match(false),
  63       transition(content::PAGE_TRANSITION_TYPED),
  64       is_history_what_you_typed_match(false),
  65       type(type),
  66       starred(false),
  67       from_previous(false) {
  68 }
  69
  70 AutocompleteMatch::AutocompleteMatch(const AutocompleteMatch& match)
  71     : provider(match.provider),
  72       relevance(match.relevance),
  73       typed_count(match.typed_count),
  74       deletable(match.deletable),
  75       fill_into_edit(match.fill_into_edit),
  76       inline_autocompletion(match.inline_autocompletion),
  77       allowed_to_be_default_match(match.allowed_to_be_default_match),
  78       destination_url(match.destination_url),
  79       stripped_destination_url(match.stripped_destination_url),
  80       contents(match.contents),
  81       contents_class(match.contents_class),
  82       description(match.description),
  83       description_class(match.description_class),
  84       transition(match.transition),
  85       is_history_what_you_typed_match(match.is_history_what_you_typed_match),
  86       type(match.type),
  87       associated_keyword(match.associated_keyword.get() ?
  88           new AutocompleteMatch(*match.associated_keyword) : NULL),
  89       keyword(match.keyword),
  90       starred(match.starred),
  91       from_previous(match.from_previous),
  92       search_terms_args(match.search_terms_args.get() ?
  93           new TemplateURLRef::SearchTermsArgs(*match.search_terms_args) :
  94           NULL),
  95       additional_info(match.additional_info),
  96       duplicate_matches(match.duplicate_matches) {
  97 }
  98
  99 AutocompleteMatch::~AutocompleteMatch() {
 100 }
 101
 102 AutocompleteMatch& AutocompleteMatch::operator=(
 103     const AutocompleteMatch& match) {
 104   if (this == &match)
 105     return *this;
 106
 107   provider = match.provider;
 108   relevance = match.relevance;
 109   typed_count = match.typed_count;
 110   deletable = match.deletable;
 111   fill_into_edit = match.fill_into_edit;
 112   inline_autocompletion = match.inline_autocompletion;
 113   allowed_to_be_default_match = match.allowed_to_be_default_match;
 114   destination_url = match.destination_url;
 115   stripped_destination_url = match.stripped_destination_url;
 116   contents = match.contents;
 117   contents_class = match.contents_class;
 118   description = match.description;
 119   description_class = match.description_class;
 120   transition = match.transition;
 121   is_history_what_you_typed_match = match.is_history_what_you_typed_match;
 122   type = match.type;
 123   associated_keyword.reset(match.associated_keyword.get() ?
 124       new AutocompleteMatch(*match.associated_keyword) : NULL);
 125   keyword = match.keyword;
 126   starred = match.starred;
 127   from_previous = match.from_previous;
 128   search_terms_args.reset(match.search_terms_args.get() ?
 129       new TemplateURLRef::SearchTermsArgs(*match.search_terms_args) : NULL);
 130   additional_info = match.additional_info;
 131   duplicate_matches = match.duplicate_matches;
 132   return *this;
 133 }
 134
 135 // static
 136 int AutocompleteMatch::TypeToIcon(Type type) {
 137   int icons[] = {
 138     IDR_OMNIBOX_HTTP,
 139     IDR_OMNIBOX_HTTP,
 140     IDR_OMNIBOX_HTTP,
 141     IDR_OMNIBOX_HTTP,
 142     IDR_OMNIBOX_HTTP,
 143     IDR_OMNIBOX_HTTP,
 144     IDR_OMNIBOX_SEARCH,
 145     IDR_OMNIBOX_SEARCH,
 146     IDR_OMNIBOX_SEARCH,
 147     IDR_OMNIBOX_SEARCH,
 148     IDR_OMNIBOX_SEARCH,
 149     IDR_OMNIBOX_SEARCH,
 150     IDR_OMNIBOX_SEARCH,
 151     IDR_OMNIBOX_SEARCH,
 152     IDR_OMNIBOX_EXTENSION_APP,
 153     IDR_OMNIBOX_SEARCH,
 154     IDR_OMNIBOX_HTTP,
 155   };
 156   COMPILE_ASSERT(arraysize(icons) == AutocompleteMatchType::NUM_TYPES,
 157                  icons_array_must_match_type_enum);
 158   return icons[type];
 159 }
 160
 161 // static
 162 int AutocompleteMatch::TypeToLocationBarIcon(Type type) {
 163   int id = TypeToIcon(type);
 164   if (id == IDR_OMNIBOX_HTTP)
 165     return IDR_LOCATION_BAR_HTTP;
 166   return id;
 167 }
 168
 169 // static
 170 bool AutocompleteMatch::MoreRelevant(const AutocompleteMatch& elem1,
 171                                      const AutocompleteMatch& elem2) {
 172   // For equal-relevance matches, we sort alphabetically, so that providers
 173   // who return multiple elements at the same priority get a "stable" sort
 174   // across multiple updates.
 175   return (elem1.relevance == elem2.relevance) ?
 176       (elem1.contents < elem2.contents) : (elem1.relevance > elem2.relevance);
 177 }
 178
 179 // static
 180 bool AutocompleteMatch::DestinationsEqual(const AutocompleteMatch& elem1,
 181                                           const AutocompleteMatch& elem2) {
 182   if (elem1.stripped_destination_url.is_empty() &&
 183       elem2.stripped_destination_url.is_empty())
 184     return false;
 185   return elem1.stripped_destination_url == elem2.stripped_destination_url;
 186 }
 187
 188 // static
 189 void AutocompleteMatch::ClassifyMatchInString(
 190     const base::string16& find_text,
 191     const base::string16& text,
 192     int style,
 193     ACMatchClassifications* classification) {
 194   ClassifyLocationInString(text.find(find_text), find_text.length(),
 195                            text.length(), style, classification);
 196 }
 197
 198 // static
 199 void AutocompleteMatch::ClassifyLocationInString(
 200     size_t match_location,
 201     size_t match_length,
 202     size_t overall_length,
 203     int style,
 204     ACMatchClassifications* classification) {
 205   classification->clear();
 206
 207   // Don't classify anything about an empty string
 208   // (AutocompleteMatch::Validate() checks this).
 209   if (overall_length == 0)
 210     return;
 211
 212   // Mark pre-match portion of string (if any).
 213   if (match_location != 0) {
 214     classification->push_back(ACMatchClassification(0, style));
 215   }
 216
 217   // Mark matching portion of string.
 218   if (match_location == base::string16::npos) {
 219     // No match, above classification will suffice for whole string.
 220     return;
 221   }
 222   // Classifying an empty match makes no sense and will lead to validation
 223   // errors later.
 224   DCHECK_GT(match_length, 0U);
 225   classification->push_back(ACMatchClassification(match_location,
 226       (style | ACMatchClassification::MATCH) & ~ACMatchClassification::DIM));
 227
 228   // Mark post-match portion of string (if any).
 229   const size_t after_match(match_location + match_length);
 230   if (after_match < overall_length) {
 231     classification->push_back(ACMatchClassification(after_match, style));
 232   }
 233 }
 234
 235 // static
 236 AutocompleteMatch::ACMatchClassifications
 237     AutocompleteMatch::MergeClassifications(
 238     const ACMatchClassifications& classifications1,
 239     const ACMatchClassifications& classifications2) {
 240   // We must return the empty vector only if both inputs are truly empty.
 241   // The result of merging an empty vector with a single (0, NONE)
 242   // classification is the latter one-entry vector.
 243   if (IsTrivialClassification(classifications1))
 244     return classifications2.empty() ? classifications1 : classifications2;
 245   if (IsTrivialClassification(classifications2))
 246     return classifications1;
 247
 248   ACMatchClassifications output;
 249   for (ACMatchClassifications::const_iterator i = classifications1.begin(),
 250        j = classifications2.begin(); i != classifications1.end();) {
 251     AutocompleteMatch::AddLastClassificationIfNecessary(&output,
 252         std::max(i->offset, j->offset), i->style | j->style);
 253     const size_t next_i_offset = (i + 1) == classifications1.end() ?
 254         static_cast<size_t>(-1) : (i + 1)->offset;
 255     const size_t next_j_offset = (j + 1) == classifications2.end() ?
 256         static_cast<size_t>(-1) : (j + 1)->offset;
 257     if (next_i_offset >= next_j_offset)
 258       ++j;
 259     if (next_j_offset >= next_i_offset)
 260       ++i;
 261   }
 262
 263   return output;
 264 }
 265
 266 // static
 267 std::string AutocompleteMatch::ClassificationsToString(
 268     const ACMatchClassifications& classifications) {
 269   std::string serialized_classifications;
 270   for (size_t i = 0; i < classifications.size(); ++i) {
 271     if (i)
 272       serialized_classifications += ',';
 273     serialized_classifications += base::IntToString(classifications[i].offset) +
 274         ',' + base::IntToString(classifications[i].style);
 275   }
 276   return serialized_classifications;
 277 }
 278
 279 // static
 280 ACMatchClassifications AutocompleteMatch::ClassificationsFromString(
 281     const std::string& serialized_classifications) {
 282   ACMatchClassifications classifications;
 283   std::vector<std::string> tokens;
 284   Tokenize(serialized_classifications, ",", &tokens);
 285   DCHECK(!(tokens.size() & 1));  // The number of tokens should be even.
 286   for (size_t i = 0; i < tokens.size(); i += 2) {
 287     int classification_offset = 0;
 288     int classification_style = ACMatchClassification::NONE;
 289     if (!base::StringToInt(tokens[i], &classification_offset) ||
 290         !base::StringToInt(tokens[i + 1], &classification_style)) {
 291       NOTREACHED();
 292       return classifications;
 293     }
 294     classifications.push_back(ACMatchClassification(classification_offset,
 295                                                     classification_style));
 296   }
 297   return classifications;
 298 }
 299
 300 // static
 301 void AutocompleteMatch::AddLastClassificationIfNecessary(
 302     ACMatchClassifications* classifications,
 303     size_t offset,
 304     int style) {
 305   DCHECK(classifications);
 306   if (classifications->empty() || classifications->back().style != style) {
 307     DCHECK(classifications->empty() ||
 308            (offset > classifications->back().offset));
 309     classifications->push_back(ACMatchClassification(offset, style));
 310   }
 311 }
 312
 313 // static
 314 base::string16 AutocompleteMatch::SanitizeString(const base::string16& text) {
 315   // NOTE: This logic is mirrored by |sanitizeString()| in
 316   // omnibox_custom_bindings.js.
 317   base::string16 result;
 318   base::TrimWhitespace(text, base::TRIM_LEADING, &result);
 319   base::RemoveChars(result, kInvalidChars, &result);
 320   return result;
 321 }
 322
 323 // static
 324 bool AutocompleteMatch::IsSearchType(Type type) {
 325   return type == AutocompleteMatchType::SEARCH_WHAT_YOU_TYPED ||
 326          type == AutocompleteMatchType::SEARCH_HISTORY ||
 327          type == AutocompleteMatchType::SEARCH_SUGGEST ||
 328          type == AutocompleteMatchType::SEARCH_OTHER_ENGINE ||
 329          IsSpecializedSearchType(type);
 330 }
 331
 332 // static
 333 bool AutocompleteMatch::IsSpecializedSearchType(Type type) {
 334   return type == AutocompleteMatchType::SEARCH_SUGGEST_ENTITY ||
 335          type == AutocompleteMatchType::SEARCH_SUGGEST_INFINITE ||
 336          type == AutocompleteMatchType::SEARCH_SUGGEST_PERSONALIZED ||
 337          type == AutocompleteMatchType::SEARCH_SUGGEST_PROFILE;
 338 }
 339
 340 void AutocompleteMatch::ComputeStrippedDestinationURL(Profile* profile) {
 341   stripped_destination_url = destination_url;
 342   if (!stripped_destination_url.is_valid())
 343     return;
 344
 345   // If the destination URL looks like it was generated from a TemplateURL,
 346   // remove all substitutions other than the search terms.  This allows us
 347   // to eliminate cases like past search URLs from history that differ only
 348   // by some obscure query param from each other or from the search/keyword
 349   // provider matches.
 350   TemplateURL* template_url = GetTemplateURL(profile, true);
 351   if (template_url != NULL && template_url->SupportsReplacement()) {
 352     base::string16 search_terms;
 353     if (template_url->ExtractSearchTermsFromURL(stripped_destination_url,
 354                                                 &search_terms)) {
 355       stripped_destination_url =
 356           GURL(template_url->url_ref().ReplaceSearchTerms(
 357               TemplateURLRef::SearchTermsArgs(search_terms)));
 358     }
 359   }
 360
 361   // |replacements| keeps all the substitions we're going to make to
 362   // from {destination_url} to {stripped_destination_url}.  |need_replacement|
 363   // is a helper variable that helps us keep track of whether we need
 364   // to apply the replacement.
 365   bool needs_replacement = false;
 366   GURL::Replacements replacements;
 367
 368   // Remove the www. prefix from the host.
 369   static const char prefix[] = "www.";
 370   static const size_t prefix_len = arraysize(prefix) - 1;
 371   std::string host = stripped_destination_url.host();
 372   if (host.compare(0, prefix_len, prefix) == 0) {
 373     host = host.substr(prefix_len);
 374     replacements.SetHostStr(host);
 375     needs_replacement = true;
 376   }
 377
 378   // Replace https protocol with http protocol.
 379   if (stripped_destination_url.SchemeIs(content::kHttpsScheme)) {
 380     replacements.SetScheme(content::kHttpScheme,
 381                            url::Component(0, strlen(content::kHttpScheme)));
 382     needs_replacement = true;
 383   }
 384
 385   if (needs_replacement)
 386     stripped_destination_url = stripped_destination_url.ReplaceComponents(
 387         replacements);
 388 }
 389
 390 void AutocompleteMatch::GetKeywordUIState(Profile* profile,
 391                                           base::string16* keyword,
 392                                           bool* is_keyword_hint) const {
 393   *is_keyword_hint = associated_keyword.get() != NULL;
 394   keyword->assign(*is_keyword_hint ? associated_keyword->keyword :
 395       GetSubstitutingExplicitlyInvokedKeyword(profile));
 396 }
 397
 398 base::string16 AutocompleteMatch::GetSubstitutingExplicitlyInvokedKeyword(
 399     Profile* profile) const {
 400   if (transition != content::PAGE_TRANSITION_KEYWORD)
 401     return base::string16();
 402   const TemplateURL* t_url = GetTemplateURL(profile, false);
 403   return (t_url && t_url->SupportsReplacement()) ? keyword : base::string16();
 404 }
 405
 406 TemplateURL* AutocompleteMatch::GetTemplateURL(
 407     Profile* profile, bool allow_fallback_to_destination_host) const {
 408   DCHECK(profile);
 409   TemplateURLService* template_url_service =
 410       TemplateURLServiceFactory::GetForProfile(profile);
 411   if (template_url_service == NULL)
 412     return NULL;
 413   TemplateURL* template_url = keyword.empty() ? NULL :
 414       template_url_service->GetTemplateURLForKeyword(keyword);
 415   if (template_url == NULL && allow_fallback_to_destination_host) {
 416     template_url = template_url_service->GetTemplateURLForHost(
 417         destination_url.host());
 418   }
 419   return template_url;
 420 }
 421
 422 void AutocompleteMatch::RecordAdditionalInfo(const std::string& property,
 423                                              const std::string& value) {
 424   DCHECK(!property.empty());
 425   DCHECK(!value.empty());
 426   additional_info[property] = value;
 427 }
 428
 429 void AutocompleteMatch::RecordAdditionalInfo(const std::string& property,
 430                                              int value) {
 431   RecordAdditionalInfo(property, base::IntToString(value));
 432 }
 433
 434 void AutocompleteMatch::RecordAdditionalInfo(const std::string& property,
 435                                              const base::Time& value) {
 436   RecordAdditionalInfo(property,
 437                        base::UTF16ToUTF8(
 438                            base::TimeFormatShortDateAndTime(value)));
 439 }
 440
 441 std::string AutocompleteMatch::GetAdditionalInfo(
 442     const std::string& property) const {
 443   AdditionalInfo::const_iterator i(additional_info.find(property));
 444   return (i == additional_info.end()) ? std::string() : i->second;
 445 }
 446
 447 bool AutocompleteMatch::IsVerbatimType() const {
 448   const bool is_keyword_verbatim_match =
 449       (type == AutocompleteMatchType::SEARCH_OTHER_ENGINE &&
 450        provider != NULL &&
 451        provider->type() == AutocompleteProvider::TYPE_SEARCH);
 452   return type == AutocompleteMatchType::SEARCH_WHAT_YOU_TYPED ||
 453       type == AutocompleteMatchType::URL_WHAT_YOU_TYPED ||
 454       is_keyword_verbatim_match;
 455 }
 456
 457 bool AutocompleteMatch::SupportsDeletion() const {
 458   if (deletable)
 459     return true;
 460
 461   for (ACMatches::const_iterator it(duplicate_matches.begin());
 462        it != duplicate_matches.end(); ++it) {
 463     if (it->deletable)
 464       return true;
 465   }
 466   return false;
 467 }
 468
 469 #ifndef NDEBUG
 470 void AutocompleteMatch::Validate() const {
 471   ValidateClassifications(contents, contents_class);
 472   ValidateClassifications(description, description_class);
 473 }
 474
 475 void AutocompleteMatch::ValidateClassifications(
 476     const base::string16& text,
 477     const ACMatchClassifications& classifications) const {
 478   if (text.empty()) {
 479     DCHECK(classifications.empty());
 480     return;
 481   }
 482
 483   // The classifications should always cover the whole string.
 484   DCHECK(!classifications.empty()) << "No classification for \"" << text << '"';
 485   DCHECK_EQ(0U, classifications[0].offset)
 486       << "Classification misses beginning for \"" << text << '"';
 487   if (classifications.size() == 1)
 488     return;
 489
 490   // The classifications should always be sorted.
 491   size_t last_offset = classifications[0].offset;
 492   for (ACMatchClassifications::const_iterator i(classifications.begin() + 1);
 493        i != classifications.end(); ++i) {
 494     const char* provider_name = provider ? provider->GetName() : "None";
 495     DCHECK_GT(i->offset, last_offset)
 496         << " Classification for \"" << text << "\" with offset of " << i->offset
 497         << " is unsorted in relation to last offset of " << last_offset
 498         << ". Provider: " << provider_name << ".";
 499     DCHECK_LT(i->offset, text.length())
 500         << " Classification of [" << i->offset << "," << text.length()
 501         << "] is out of bounds for \"" << text << "\". Provider: "
 502         << provider_name << ".";
 503     last_offset = i->offset;
 504   }
 505 }
 506 #endif