components/omnibox/keyword_provider.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/omnibox/keyword_provider.h"
   6
   7 #include <algorithm>
   8 #include <vector>
   9
  10 #include "base/strings/string16.h"
  11 #include "base/strings/string_util.h"
  12 #include "base/strings/utf_string_conversions.h"
  13 #include "components/metrics/proto/omnibox_input_type.pb.h"
  14 #include "components/omnibox/autocomplete_match.h"
  15 #include "components/omnibox/autocomplete_provider_client.h"
  16 #include "components/omnibox/autocomplete_provider_listener.h"
  17 #include "components/omnibox/keyword_extensions_delegate.h"
  18 #include "components/search_engines/template_url.h"
  19 #include "components/search_engines/template_url_service.h"
  20 #include "grit/components_strings.h"
  21 #include "net/base/escape.h"
  22 #include "net/base/net_util.h"
  23 #include "ui/base/l10n/l10n_util.h"
  24
  25 namespace {
  26
  27 // Helper functor for Start(), for sorting keyword matches by quality.
  28 class CompareQuality {
  29  public:
  30   // A keyword is of higher quality when a greater fraction of it has been
  31   // typed, that is, when it is shorter.
  32   //
  33   // TODO(pkasting): Most recent and most frequent keywords are probably
  34   // better rankings than the fraction of the keyword typed.  We should
  35   // always put any exact matches first no matter what, since the code in
  36   // Start() assumes this (and it makes sense).
  37   bool operator()(const TemplateURL* t_url1, const TemplateURL* t_url2) const {
  38     return t_url1->keyword().length() < t_url2->keyword().length();
  39   }
  40 };
  41
  42 // Helper for KeywordProvider::Start(), for ending keyword mode unless
  43 // explicitly told otherwise.
  44 class ScopedEndExtensionKeywordMode {
  45  public:
  46   explicit ScopedEndExtensionKeywordMode(KeywordExtensionsDelegate* delegate);
  47   ~ScopedEndExtensionKeywordMode();
  48
  49   void StayInKeywordMode();
  50
  51  private:
  52   KeywordExtensionsDelegate* delegate_;
  53
  54   DISALLOW_COPY_AND_ASSIGN(ScopedEndExtensionKeywordMode);
  55 };
  56
  57 ScopedEndExtensionKeywordMode::ScopedEndExtensionKeywordMode(
  58     KeywordExtensionsDelegate* delegate)
  59     : delegate_(delegate) {
  60 }
  61
  62 ScopedEndExtensionKeywordMode::~ScopedEndExtensionKeywordMode() {
  63   if (delegate_)
  64     delegate_->MaybeEndExtensionKeywordMode();
  65 }
  66
  67 void ScopedEndExtensionKeywordMode::StayInKeywordMode() {
  68   delegate_ = NULL;
  69 }
  70
  71 }  // namespace
  72
  73 KeywordProvider::KeywordProvider(AutocompleteProviderClient* client,
  74                                  AutocompleteProviderListener* listener)
  75     : AutocompleteProvider(AutocompleteProvider::TYPE_KEYWORD),
  76       listener_(listener),
  77       model_(client->GetTemplateURLService()),
  78       extensions_delegate_(client->GetKeywordExtensionsDelegate(this)) {
  79 }
  80
  81 // static
  82 base::string16 KeywordProvider::SplitKeywordFromInput(
  83     const base::string16& input,
  84     bool trim_leading_whitespace,
  85     base::string16* remaining_input) {
  86   // Find end of first token.  The AutocompleteController has trimmed leading
  87   // whitespace, so we need not skip over that.
  88   const size_t first_white(input.find_first_of(base::kWhitespaceUTF16));
  89   DCHECK_NE(0U, first_white);
  90   if (first_white == base::string16::npos)
  91     return input;  // Only one token provided.
  92
  93   // Set |remaining_input| to everything after the first token.
  94   DCHECK(remaining_input != NULL);
  95   const size_t remaining_start = trim_leading_whitespace ?
  96       input.find_first_not_of(base::kWhitespaceUTF16, first_white) :
  97       first_white + 1;
  98
  99   if (remaining_start < input.length())
 100     remaining_input->assign(input.begin() + remaining_start, input.end());
 101
 102   // Return first token as keyword.
 103   return input.substr(0, first_white);
 104 }
 105
 106 // static
 107 base::string16 KeywordProvider::SplitReplacementStringFromInput(
 108     const base::string16& input,
 109     bool trim_leading_whitespace) {
 110   // The input may contain leading whitespace, strip it.
 111   base::string16 trimmed_input;
 112   base::TrimWhitespace(input, base::TRIM_LEADING, &trimmed_input);
 113
 114   // And extract the replacement string.
 115   base::string16 remaining_input;
 116   SplitKeywordFromInput(trimmed_input, trim_leading_whitespace,
 117       &remaining_input);
 118   return remaining_input;
 119 }
 120
 121 // static
 122 const TemplateURL* KeywordProvider::GetSubstitutingTemplateURLForInput(
 123     TemplateURLService* model,
 124     AutocompleteInput* input) {
 125   if (!input->allow_exact_keyword_match())
 126     return NULL;
 127
 128   base::string16 keyword, remaining_input;
 129   if (!ExtractKeywordFromInput(*input, &keyword, &remaining_input))
 130     return NULL;
 131
 132   DCHECK(model);
 133   const TemplateURL* template_url = model->GetTemplateURLForKeyword(keyword);
 134   if (template_url &&
 135       template_url->SupportsReplacement(model->search_terms_data())) {
 136     // Adjust cursor position iff it was set before, otherwise leave it as is.
 137     size_t cursor_position = base::string16::npos;
 138     // The adjustment assumes that the keyword was stripped from the beginning
 139     // of the original input.
 140     if (input->cursor_position() != base::string16::npos &&
 141         !remaining_input.empty() &&
 142         base::EndsWith(input->text(), remaining_input, true)) {
 143       int offset = input->text().length() - input->cursor_position();
 144       // The cursor should never be past the last character or before the
 145       // first character.
 146       DCHECK_GE(offset, 0);
 147       DCHECK_LE(offset, static_cast<int>(input->text().length()));
 148       if (offset <= 0) {
 149         // Normalize the cursor to be exactly after the last character.
 150         cursor_position = remaining_input.length();
 151       } else {
 152         // If somehow the cursor was before the remaining text, set it to 0,
 153         // otherwise adjust it relative to the remaining text.
 154         cursor_position = offset > static_cast<int>(remaining_input.length()) ?
 155             0u : remaining_input.length() - offset;
 156       }
 157     }
 158     input->UpdateText(remaining_input, cursor_position, input->parts());
 159     return template_url;
 160   }
 161
 162   return NULL;
 163 }
 164
 165 base::string16 KeywordProvider::GetKeywordForText(
 166     const base::string16& text) const {
 167   const base::string16 keyword(TemplateURLService::CleanUserInputKeyword(text));
 168
 169   if (keyword.empty())
 170     return keyword;
 171
 172   TemplateURLService* url_service = GetTemplateURLService();
 173   if (!url_service)
 174     return base::string16();
 175
 176   // Don't provide a keyword if it doesn't support replacement.
 177   const TemplateURL* const template_url =
 178       url_service->GetTemplateURLForKeyword(keyword);
 179   if (!template_url ||
 180       !template_url->SupportsReplacement(url_service->search_terms_data()))
 181     return base::string16();
 182
 183   // Don't provide a keyword for inactive/disabled extension keywords.
 184   if ((template_url->GetType() == TemplateURL::OMNIBOX_API_EXTENSION) &&
 185       extensions_delegate_ &&
 186       !extensions_delegate_->IsEnabledExtension(template_url->GetExtensionId()))
 187     return base::string16();
 188
 189   return keyword;
 190 }
 191
 192 AutocompleteMatch KeywordProvider::CreateVerbatimMatch(
 193     const base::string16& text,
 194     const base::string16& keyword,
 195     const AutocompleteInput& input) {
 196   // A verbatim match is allowed to be the default match.
 197   return CreateAutocompleteMatch(
 198       GetTemplateURLService()->GetTemplateURLForKeyword(keyword), input,
 199       keyword.length(), SplitReplacementStringFromInput(text, true), true, 0);
 200 }
 201
 202 void KeywordProvider::Start(const AutocompleteInput& input,
 203                             bool minimal_changes) {
 204   // This object ensures we end keyword mode if we exit the function without
 205   // toggling keyword mode to on.
 206   ScopedEndExtensionKeywordMode keyword_mode_toggle(extensions_delegate_.get());
 207
 208   matches_.clear();
 209
 210   if (!minimal_changes) {
 211     done_ = true;
 212
 213     // Input has changed. Increment the input ID so that we can discard any
 214     // stale extension suggestions that may be incoming.
 215     if (extensions_delegate_)
 216       extensions_delegate_->IncrementInputId();
 217   }
 218
 219   if (input.from_omnibox_focus())
 220     return;
 221
 222   // Split user input into a keyword and some query input.
 223   //
 224   // We want to suggest keywords even when users have started typing URLs, on
 225   // the assumption that they might not realize they no longer need to go to a
 226   // site to be able to search it.  So we call CleanUserInputKeyword() to strip
 227   // any initial scheme and/or "www.".  NOTE: Any heuristics or UI used to
 228   // automatically/manually create keywords will need to be in sync with
 229   // whatever we do here!
 230   //
 231   // TODO(pkasting): http://crbug/347744 If someday we remember usage frequency
 232   // for keywords, we might suggest keywords that haven't even been partially
 233   // typed, if the user uses them enough and isn't obviously typing something
 234   // else.  In this case we'd consider all input here to be query input.
 235   base::string16 keyword, remaining_input;
 236   if (!ExtractKeywordFromInput(input, &keyword, &remaining_input))
 237     return;
 238
 239   // Get the best matches for this keyword.
 240   //
 241   // NOTE: We could cache the previous keywords and reuse them here in the
 242   // |minimal_changes| case, but since we'd still have to recalculate their
 243   // relevances and we can just recreate the results synchronously anyway, we
 244   // don't bother.
 245   TemplateURLService::TemplateURLVector matches;
 246   GetTemplateURLService()->FindMatchingKeywords(
 247       keyword, !remaining_input.empty(), &matches);
 248
 249   for (TemplateURLService::TemplateURLVector::iterator i(matches.begin());
 250        i != matches.end(); ) {
 251     const TemplateURL* template_url = *i;
 252
 253     // Prune any extension keywords that are disallowed in incognito mode (if
 254     // we're incognito), or disabled.
 255     if (template_url->GetType() == TemplateURL::OMNIBOX_API_EXTENSION &&
 256         extensions_delegate_ &&
 257         !extensions_delegate_->IsEnabledExtension(
 258             template_url->GetExtensionId())) {
 259       i = matches.erase(i);
 260       continue;
 261     }
 262
 263     // Prune any substituting keywords if there is no substitution.
 264     if (template_url->SupportsReplacement(
 265             GetTemplateURLService()->search_terms_data()) &&
 266         remaining_input.empty() &&
 267         !input.allow_exact_keyword_match()) {
 268       i = matches.erase(i);
 269       continue;
 270     }
 271
 272     ++i;
 273   }
 274   if (matches.empty())
 275     return;
 276   std::sort(matches.begin(), matches.end(), CompareQuality());
 277
 278   // Limit to one exact or three inexact matches, and mark them up for display
 279   // in the autocomplete popup.
 280   // Any exact match is going to be the highest quality match, and thus at the
 281   // front of our vector.
 282   if (matches.front()->keyword() == keyword) {
 283     const TemplateURL* template_url = matches.front();
 284     const bool is_extension_keyword =
 285         template_url->GetType() == TemplateURL::OMNIBOX_API_EXTENSION;
 286
 287     // Only create an exact match if |remaining_input| is empty or if
 288     // this is an extension keyword.  If |remaining_input| is a
 289     // non-empty non-extension keyword (i.e., a regular keyword that
 290     // supports replacement and that has extra text following it),
 291     // then SearchProvider creates the exact (a.k.a. verbatim) match.
 292     if (!remaining_input.empty() && !is_extension_keyword)
 293       return;
 294
 295     // TODO(pkasting): We should probably check that if the user explicitly
 296     // typed a scheme, that scheme matches the one in |template_url|.
 297
 298     // When creating an exact match (either for the keyword itself, no
 299     // remaining query or an extension keyword, possibly with remaining
 300     // input), allow the match to be the default match.
 301     matches_.push_back(CreateAutocompleteMatch(
 302         template_url, input, keyword.length(), remaining_input, true, -1));
 303
 304     if (is_extension_keyword && extensions_delegate_) {
 305       if (extensions_delegate_->Start(input, minimal_changes, template_url,
 306                                       remaining_input))
 307         keyword_mode_toggle.StayInKeywordMode();
 308     }
 309   } else {
 310     if (matches.size() > kMaxMatches)
 311       matches.erase(matches.begin() + kMaxMatches, matches.end());
 312     for (TemplateURLService::TemplateURLVector::const_iterator i(
 313          matches.begin()); i != matches.end(); ++i) {
 314       matches_.push_back(CreateAutocompleteMatch(
 315           *i, input, keyword.length(), remaining_input, false, -1));
 316     }
 317   }
 318 }
 319
 320 void KeywordProvider::Stop(bool clear_cached_results,
 321                            bool due_to_user_inactivity) {
 322   done_ = true;
 323   // Only end an extension's request if the user did something to explicitly
 324   // cancel it; mere inactivity shouldn't terminate long-running extension
 325   // operations since the user likely explicitly requested them.
 326   if (extensions_delegate_ && !due_to_user_inactivity)
 327     extensions_delegate_->MaybeEndExtensionKeywordMode();
 328 }
 329
 330 KeywordProvider::~KeywordProvider() {}
 331
 332 // static
 333 bool KeywordProvider::ExtractKeywordFromInput(const AutocompleteInput& input,
 334                                               base::string16* keyword,
 335                                               base::string16* remaining_input) {
 336   if ((input.type() == metrics::OmniboxInputType::INVALID) ||
 337       (input.type() == metrics::OmniboxInputType::FORCED_QUERY))
 338     return false;
 339
 340   *keyword = TemplateURLService::CleanUserInputKeyword(
 341       SplitKeywordFromInput(input.text(), true, remaining_input));
 342   return !keyword->empty();
 343 }
 344
 345 // static
 346 int KeywordProvider::CalculateRelevance(metrics::OmniboxInputType::Type type,
 347                                         bool complete,
 348                                         bool supports_replacement,
 349                                         bool prefer_keyword,
 350                                         bool allow_exact_keyword_match) {
 351   // This function is responsible for scoring suggestions of keywords
 352   // themselves and the suggestion of the verbatim query on an
 353   // extension keyword.  SearchProvider::CalculateRelevanceForKeywordVerbatim()
 354   // scores verbatim query suggestions for non-extension keywords.
 355   // These two functions are currently in sync, but there's no reason
 356   // we couldn't decide in the future to score verbatim matches
 357   // differently for extension and non-extension keywords.  If you
 358   // make such a change, however, you should update this comment to
 359   // describe it, so it's clear why the functions diverge.
 360   if (!complete)
 361     return (type == metrics::OmniboxInputType::URL) ? 700 : 450;
 362   if (!supports_replacement || (allow_exact_keyword_match && prefer_keyword))
 363     return 1500;
 364   return (allow_exact_keyword_match &&
 365           (type == metrics::OmniboxInputType::QUERY)) ?
 366       1450 : 1100;
 367 }
 368
 369 AutocompleteMatch KeywordProvider::CreateAutocompleteMatch(
 370     const TemplateURL* template_url,
 371     const AutocompleteInput& input,
 372     size_t prefix_length,
 373     const base::string16& remaining_input,
 374     bool allowed_to_be_default_match,
 375     int relevance) {
 376   DCHECK(template_url);
 377   const bool supports_replacement =
 378       template_url->url_ref().SupportsReplacement(
 379           GetTemplateURLService()->search_terms_data());
 380
 381   // Create an edit entry of "[keyword] [remaining input]".  This is helpful
 382   // even when [remaining input] is empty, as the user can select the popup
 383   // choice and immediately begin typing in query input.
 384   const base::string16& keyword = template_url->keyword();
 385   const bool keyword_complete = (prefix_length == keyword.length());
 386   if (relevance < 0) {
 387     relevance =
 388         CalculateRelevance(input.type(), keyword_complete,
 389                            // When the user wants keyword matches to take
 390                            // preference, score them highly regardless of
 391                            // whether the input provides query text.
 392                            supports_replacement, input.prefer_keyword(),
 393                            input.allow_exact_keyword_match());
 394   }
 395   AutocompleteMatch match(this, relevance, false,
 396       supports_replacement ? AutocompleteMatchType::SEARCH_OTHER_ENGINE :
 397                              AutocompleteMatchType::HISTORY_KEYWORD);
 398   match.allowed_to_be_default_match = allowed_to_be_default_match;
 399   match.fill_into_edit = keyword;
 400   if (!remaining_input.empty() || supports_replacement)
 401     match.fill_into_edit.push_back(L' ');
 402   match.fill_into_edit.append(remaining_input);
 403   // If we wanted to set |result.inline_autocompletion| correctly, we'd need
 404   // CleanUserInputKeyword() to return the amount of adjustment it's made to
 405   // the user's input.  Because right now inexact keyword matches can't score
 406   // more highly than a "what you typed" match from one of the other providers,
 407   // we just don't bother to do this, and leave inline autocompletion off.
 408
 409   // Create destination URL and popup entry content by substituting user input
 410   // into keyword templates.
 411   FillInURLAndContents(remaining_input, template_url, &match);
 412
 413   match.keyword = keyword;
 414   match.transition = ui::PAGE_TRANSITION_KEYWORD;
 415
 416   return match;
 417 }
 418
 419 void KeywordProvider::FillInURLAndContents(
 420     const base::string16& remaining_input,
 421     const TemplateURL* element,
 422     AutocompleteMatch* match) const {
 423   DCHECK(!element->short_name().empty());
 424   const TemplateURLRef& element_ref = element->url_ref();
 425   DCHECK(element_ref.IsValid(GetTemplateURLService()->search_terms_data()));
 426   int message_id = (element->GetType() == TemplateURL::OMNIBOX_API_EXTENSION) ?
 427       IDS_EXTENSION_KEYWORD_COMMAND : IDS_KEYWORD_SEARCH;
 428   if (remaining_input.empty()) {
 429     // Allow extension keyword providers to accept empty string input. This is
 430     // useful to allow extensions to do something in the case where no input is
 431     // entered.
 432     if (element_ref.SupportsReplacement(
 433             GetTemplateURLService()->search_terms_data()) &&
 434         (element->GetType() != TemplateURL::OMNIBOX_API_EXTENSION)) {
 435       // No query input; return a generic, no-destination placeholder.
 436       match->contents.assign(
 437           l10n_util::GetStringFUTF16(message_id,
 438               element->AdjustedShortNameForLocaleDirection(),
 439               l10n_util::GetStringUTF16(IDS_EMPTY_KEYWORD_VALUE)));
 440       match->contents_class.push_back(
 441           ACMatchClassification(0, ACMatchClassification::DIM));
 442     } else {
 443       // Keyword that has no replacement text (aka a shorthand for a URL).
 444       match->destination_url = GURL(element->url());
 445       match->contents.assign(element->short_name());
 446       AutocompleteMatch::ClassifyLocationInString(0, match->contents.length(),
 447           match->contents.length(), ACMatchClassification::NONE,
 448           &match->contents_class);
 449     }
 450   } else {
 451     // Create destination URL by escaping user input and substituting into
 452     // keyword template URL.  The escaping here handles whitespace in user
 453     // input, but we rely on later canonicalization functions to do more
 454     // fixup to make the URL valid if necessary.
 455     DCHECK(element_ref.SupportsReplacement(
 456         GetTemplateURLService()->search_terms_data()));
 457     TemplateURLRef::SearchTermsArgs search_terms_args(remaining_input);
 458     search_terms_args.append_extra_query_params =
 459         element == GetTemplateURLService()->GetDefaultSearchProvider();
 460     match->destination_url = GURL(element_ref.ReplaceSearchTerms(
 461         search_terms_args, GetTemplateURLService()->search_terms_data()));
 462     std::vector<size_t> content_param_offsets;
 463     match->contents.assign(l10n_util::GetStringFUTF16(message_id,
 464                                                       element->short_name(),
 465                                                       remaining_input,
 466                                                       &content_param_offsets));
 467     DCHECK_EQ(2U, content_param_offsets.size());
 468     AutocompleteMatch::ClassifyLocationInString(content_param_offsets[1],
 469         remaining_input.length(), match->contents.length(),
 470         ACMatchClassification::NONE, &match->contents_class);
 471   }
 472 }
 473
 474 TemplateURLService* KeywordProvider::GetTemplateURLService() const {
 475   // Make sure the model is loaded. This is cheap and quickly bails out if
 476   // the model is already loaded.
 477   model_->Load();
 478   return model_;
 479 }