components/omnibox/search_suggestion_parser.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/search_suggestion_parser.h"
   6
   7 #include "base/i18n/icu_string_conversions.h"
   8 #include "base/json/json_string_value_serializer.h"
   9 #include "base/json/json_writer.h"
  10 #include "base/logging.h"
  11 #include "base/metrics/histogram.h"
  12 #include "base/strings/string_number_conversions.h"
  13 #include "base/strings/string_util.h"
  14 #include "base/strings/utf_string_conversions.h"
  15 #include "base/values.h"
  16 #include "components/omnibox/autocomplete_input.h"
  17 #include "components/omnibox/url_prefix.h"
  18 #include "components/url_fixer/url_fixer.h"
  19 #include "net/base/net_util.h"
  20 #include "net/http/http_response_headers.h"
  21 #include "net/url_request/url_fetcher.h"
  22 #include "url/url_constants.h"
  23
  24 namespace {
  25
  26 AutocompleteMatchType::Type GetAutocompleteMatchType(const std::string& type) {
  27   if (type == "ENTITY")
  28     return AutocompleteMatchType::SEARCH_SUGGEST_ENTITY;
  29   if (type == "TAIL")
  30     return AutocompleteMatchType::SEARCH_SUGGEST_TAIL;
  31   if (type == "PERSONALIZED_QUERY")
  32     return AutocompleteMatchType::SEARCH_SUGGEST_PERSONALIZED;
  33   if (type == "PROFILE")
  34     return AutocompleteMatchType::SEARCH_SUGGEST_PROFILE;
  35   if (type == "NAVIGATION")
  36     return AutocompleteMatchType::NAVSUGGEST;
  37   if (type == "PERSONALIZED_NAVIGATION")
  38     return AutocompleteMatchType::NAVSUGGEST_PERSONALIZED;
  39   return AutocompleteMatchType::SEARCH_SUGGEST;
  40 }
  41
  42 }  // namespace
  43
  44 // SearchSuggestionParser::Result ----------------------------------------------
  45
  46 SearchSuggestionParser::Result::Result(bool from_keyword_provider,
  47                                        int relevance,
  48                                        bool relevance_from_server,
  49                                        AutocompleteMatchType::Type type,
  50                                        const std::string& deletion_url)
  51     : from_keyword_provider_(from_keyword_provider),
  52       type_(type),
  53       relevance_(relevance),
  54       relevance_from_server_(relevance_from_server),
  55       received_after_last_keystroke_(true),
  56       deletion_url_(deletion_url) {}
  57
  58 SearchSuggestionParser::Result::~Result() {}
  59
  60 // SearchSuggestionParser::SuggestResult ---------------------------------------
  61
  62 SearchSuggestionParser::SuggestResult::SuggestResult(
  63     const base::string16& suggestion,
  64     AutocompleteMatchType::Type type,
  65     const base::string16& match_contents,
  66     const base::string16& match_contents_prefix,
  67     const base::string16& annotation,
  68     const base::string16& answer_contents,
  69     const base::string16& answer_type,
  70     scoped_ptr<SuggestionAnswer> answer,
  71     const std::string& suggest_query_params,
  72     const std::string& deletion_url,
  73     bool from_keyword_provider,
  74     int relevance,
  75     bool relevance_from_server,
  76     bool should_prefetch,
  77     const base::string16& input_text)
  78     : Result(from_keyword_provider,
  79              relevance,
  80              relevance_from_server,
  81              type,
  82              deletion_url),
  83       suggestion_(suggestion),
  84       match_contents_prefix_(match_contents_prefix),
  85       annotation_(annotation),
  86       suggest_query_params_(suggest_query_params),
  87       answer_contents_(answer_contents),
  88       answer_type_(answer_type),
  89       answer_(answer.Pass()),
  90       should_prefetch_(should_prefetch) {
  91   match_contents_ = match_contents;
  92   DCHECK(!match_contents_.empty());
  93   ClassifyMatchContents(true, input_text);
  94 }
  95
  96 SearchSuggestionParser::SuggestResult::SuggestResult(
  97     const SuggestResult& result)
  98     : Result(result),
  99       suggestion_(result.suggestion_),
 100       match_contents_prefix_(result.match_contents_prefix_),
 101       annotation_(result.annotation_),
 102       suggest_query_params_(result.suggest_query_params_),
 103       answer_contents_(result.answer_contents_),
 104       answer_type_(result.answer_type_),
 105       answer_(SuggestionAnswer::copy(result.answer_.get())),
 106       should_prefetch_(result.should_prefetch_) {
 107 }
 108
 109 SearchSuggestionParser::SuggestResult::~SuggestResult() {}
 110
 111 SearchSuggestionParser::SuggestResult&
 112     SearchSuggestionParser::SuggestResult::operator=(const SuggestResult& rhs) {
 113   if (this == &rhs)
 114     return *this;
 115
 116   // Assign via parent class first.
 117   Result::operator=(rhs);
 118
 119   suggestion_ = rhs.suggestion_;
 120   match_contents_prefix_ = rhs.match_contents_prefix_;
 121   annotation_ = rhs.annotation_;
 122   suggest_query_params_ = rhs.suggest_query_params_;
 123   answer_contents_ = rhs.answer_contents_;
 124   answer_type_ = rhs.answer_type_;
 125   answer_ = SuggestionAnswer::copy(rhs.answer_.get());
 126   should_prefetch_ = rhs.should_prefetch_;
 127
 128   return *this;
 129 }
 130
 131 void SearchSuggestionParser::SuggestResult::ClassifyMatchContents(
 132     const bool allow_bolding_all,
 133     const base::string16& input_text) {
 134   if (input_text.empty()) {
 135     // In case of zero-suggest results, do not highlight matches.
 136     match_contents_class_.push_back(
 137         ACMatchClassification(0, ACMatchClassification::NONE));
 138     return;
 139   }
 140
 141   base::string16 lookup_text = input_text;
 142   if (type_ == AutocompleteMatchType::SEARCH_SUGGEST_TAIL) {
 143     const size_t contents_index =
 144         suggestion_.length() - match_contents_.length();
 145     // Ensure the query starts with the input text, and ends with the match
 146     // contents, and the input text has an overlap with contents.
 147     if (StartsWith(suggestion_, input_text, true) &&
 148         EndsWith(suggestion_, match_contents_, true) &&
 149         (input_text.length() > contents_index)) {
 150       lookup_text = input_text.substr(contents_index);
 151     }
 152   }
 153   size_t lookup_position = match_contents_.find(lookup_text);
 154   if (!allow_bolding_all && (lookup_position == base::string16::npos)) {
 155     // Bail if the code below to update the bolding would bold the whole
 156     // string.  Note that the string may already be entirely bolded; if
 157     // so, leave it as is.
 158     return;
 159   }
 160   match_contents_class_.clear();
 161   // We do intra-string highlighting for suggestions - the suggested segment
 162   // will be highlighted, e.g. for input_text = "you" the suggestion may be
 163   // "youtube", so we'll bold the "tube" section: you*tube*.
 164   if (input_text != match_contents_) {
 165     if (lookup_position == base::string16::npos) {
 166       // The input text is not a substring of the query string, e.g. input
 167       // text is "slasdot" and the query string is "slashdot", so we bold the
 168       // whole thing.
 169       match_contents_class_.push_back(
 170           ACMatchClassification(0, ACMatchClassification::MATCH));
 171     } else {
 172       // We don't iterate over the string here annotating all matches because
 173       // it looks odd to have every occurrence of a substring that may be as
 174       // short as a single character highlighted in a query suggestion result,
 175       // e.g. for input text "s" and query string "southwest airlines", it
 176       // looks odd if both the first and last s are highlighted.
 177       if (lookup_position != 0) {
 178         match_contents_class_.push_back(
 179             ACMatchClassification(0, ACMatchClassification::MATCH));
 180       }
 181       match_contents_class_.push_back(
 182           ACMatchClassification(lookup_position, ACMatchClassification::NONE));
 183       size_t next_fragment_position = lookup_position + lookup_text.length();
 184       if (next_fragment_position < match_contents_.length()) {
 185         match_contents_class_.push_back(ACMatchClassification(
 186             next_fragment_position, ACMatchClassification::MATCH));
 187       }
 188     }
 189   } else {
 190     // Otherwise, match_contents_ is a verbatim (what-you-typed) match, either
 191     // for the default provider or a keyword search provider.
 192     match_contents_class_.push_back(
 193         ACMatchClassification(0, ACMatchClassification::NONE));
 194   }
 195 }
 196
 197 int SearchSuggestionParser::SuggestResult::CalculateRelevance(
 198     const AutocompleteInput& input,
 199     bool keyword_provider_requested) const {
 200   if (!from_keyword_provider_ && keyword_provider_requested)
 201     return 100;
 202   return ((input.type() == metrics::OmniboxInputType::URL) ? 300 : 600);
 203 }
 204
 205 // SearchSuggestionParser::NavigationResult ------------------------------------
 206
 207 SearchSuggestionParser::NavigationResult::NavigationResult(
 208     const AutocompleteSchemeClassifier& scheme_classifier,
 209     const GURL& url,
 210     AutocompleteMatchType::Type type,
 211     const base::string16& description,
 212     const std::string& deletion_url,
 213     bool from_keyword_provider,
 214     int relevance,
 215     bool relevance_from_server,
 216     const base::string16& input_text,
 217     const std::string& languages)
 218     : Result(from_keyword_provider, relevance, relevance_from_server, type,
 219              deletion_url),
 220       url_(url),
 221       formatted_url_(AutocompleteInput::FormattedStringWithEquivalentMeaning(
 222           url, net::FormatUrl(url, languages,
 223                               net::kFormatUrlOmitAll & ~net::kFormatUrlOmitHTTP,
 224                               net::UnescapeRule::SPACES, NULL, NULL, NULL),
 225           scheme_classifier)),
 226       description_(description) {
 227   DCHECK(url_.is_valid());
 228   CalculateAndClassifyMatchContents(true, input_text, languages);
 229 }
 230
 231 SearchSuggestionParser::NavigationResult::~NavigationResult() {}
 232
 233 void
 234 SearchSuggestionParser::NavigationResult::CalculateAndClassifyMatchContents(
 235     const bool allow_bolding_nothing,
 236     const base::string16& input_text,
 237     const std::string& languages) {
 238   if (input_text.empty()) {
 239     // In case of zero-suggest results, do not highlight matches.
 240     match_contents_class_.push_back(
 241         ACMatchClassification(0, ACMatchClassification::NONE));
 242     return;
 243   }
 244
 245   // First look for the user's input inside the formatted url as it would be
 246   // without trimming the scheme, so we can find matches at the beginning of the
 247   // scheme.
 248   const URLPrefix* prefix =
 249       URLPrefix::BestURLPrefix(formatted_url_, input_text);
 250   size_t match_start = (prefix == NULL) ?
 251       formatted_url_.find(input_text) : prefix->prefix.length();
 252   bool trim_http = !AutocompleteInput::HasHTTPScheme(input_text) &&
 253                    (!prefix || (match_start != 0));
 254   const net::FormatUrlTypes format_types =
 255       net::kFormatUrlOmitAll & ~(trim_http ? 0 : net::kFormatUrlOmitHTTP);
 256
 257   base::string16 match_contents = net::FormatUrl(url_, languages, format_types,
 258       net::UnescapeRule::SPACES, NULL, NULL, &match_start);
 259   // If the first match in the untrimmed string was inside a scheme that we
 260   // trimmed, look for a subsequent match.
 261   if (match_start == base::string16::npos)
 262     match_start = match_contents.find(input_text);
 263   // Update |match_contents_| and |match_contents_class_| if it's allowed.
 264   if (allow_bolding_nothing || (match_start != base::string16::npos)) {
 265     match_contents_ = match_contents;
 266     // Safe if |match_start| is npos; also safe if the input is longer than the
 267     // remaining contents after |match_start|.
 268     AutocompleteMatch::ClassifyLocationInString(match_start,
 269         input_text.length(), match_contents_.length(),
 270         ACMatchClassification::URL, &match_contents_class_);
 271   }
 272 }
 273
 274 int SearchSuggestionParser::NavigationResult::CalculateRelevance(
 275     const AutocompleteInput& input,
 276     bool keyword_provider_requested) const {
 277   return (from_keyword_provider_ || !keyword_provider_requested) ? 800 : 150;
 278 }
 279
 280 // SearchSuggestionParser::Results ---------------------------------------------
 281
 282 SearchSuggestionParser::Results::Results()
 283     : verbatim_relevance(-1),
 284       field_trial_triggered(false),
 285       relevances_from_server(false) {}
 286
 287 SearchSuggestionParser::Results::~Results() {}
 288
 289 void SearchSuggestionParser::Results::Clear() {
 290   suggest_results.clear();
 291   navigation_results.clear();
 292   verbatim_relevance = -1;
 293   metadata.clear();
 294 }
 295
 296 bool SearchSuggestionParser::Results::HasServerProvidedScores() const {
 297   if (verbatim_relevance >= 0)
 298     return true;
 299
 300   // Right now either all results of one type will be server-scored or they will
 301   // all be locally scored, but in case we change this later, we'll just check
 302   // them all.
 303   for (SuggestResults::const_iterator i(suggest_results.begin());
 304        i != suggest_results.end(); ++i) {
 305     if (i->relevance_from_server())
 306       return true;
 307   }
 308   for (NavigationResults::const_iterator i(navigation_results.begin());
 309        i != navigation_results.end(); ++i) {
 310     if (i->relevance_from_server())
 311       return true;
 312   }
 313
 314   return false;
 315 }
 316
 317 // SearchSuggestionParser ------------------------------------------------------
 318
 319 // static
 320 std::string SearchSuggestionParser::ExtractJsonData(
 321     const net::URLFetcher* source) {
 322   const net::HttpResponseHeaders* const response_headers =
 323       source->GetResponseHeaders();
 324   std::string json_data;
 325   source->GetResponseAsString(&json_data);
 326
 327   // JSON is supposed to be UTF-8, but some suggest service providers send
 328   // JSON files in non-UTF-8 encodings.  The actual encoding is usually
 329   // specified in the Content-Type header field.
 330   if (response_headers) {
 331     std::string charset;
 332     if (response_headers->GetCharset(&charset)) {
 333       base::string16 data_16;
 334       // TODO(jungshik): Switch to CodePageToUTF8 after it's added.
 335       if (base::CodepageToUTF16(json_data, charset.c_str(),
 336                                 base::OnStringConversionError::FAIL,
 337                                 &data_16))
 338         json_data = base::UTF16ToUTF8(data_16);
 339     }
 340   }
 341   return json_data;
 342 }
 343
 344 // static
 345 scoped_ptr<base::Value> SearchSuggestionParser::DeserializeJsonData(
 346     base::StringPiece json_data) {
 347   // The JSON response should be an array.
 348   for (size_t response_start_index = json_data.find("["), i = 0;
 349        response_start_index != base::StringPiece::npos && i < 5;
 350        response_start_index = json_data.find("[", 1), i++) {
 351     // Remove any XSSI guards to allow for JSON parsing.
 352     json_data.remove_prefix(response_start_index);
 353
 354     JSONStringValueSerializer deserializer(json_data);
 355     deserializer.set_allow_trailing_comma(true);
 356     int error_code = 0;
 357     scoped_ptr<base::Value> data(deserializer.Deserialize(&error_code, NULL));
 358     if (error_code == 0)
 359       return data.Pass();
 360   }
 361   return scoped_ptr<base::Value>();
 362 }
 363
 364 // static
 365 bool SearchSuggestionParser::ParseSuggestResults(
 366     const base::Value& root_val,
 367     const AutocompleteInput& input,
 368     const AutocompleteSchemeClassifier& scheme_classifier,
 369     int default_result_relevance,
 370     const std::string& languages,
 371     bool is_keyword_result,
 372     Results* results) {
 373   base::string16 query;
 374   const base::ListValue* root_list = NULL;
 375   const base::ListValue* results_list = NULL;
 376
 377   if (!root_val.GetAsList(&root_list) || !root_list->GetString(0, &query) ||
 378       query != input.text() || !root_list->GetList(1, &results_list))
 379     return false;
 380
 381   // 3rd element: Description list.
 382   const base::ListValue* descriptions = NULL;
 383   root_list->GetList(2, &descriptions);
 384
 385   // 4th element: Disregard the query URL list for now.
 386
 387   // Reset suggested relevance information.
 388   results->verbatim_relevance = -1;
 389
 390   // 5th element: Optional key-value pairs from the Suggest server.
 391   const base::ListValue* types = NULL;
 392   const base::ListValue* relevances = NULL;
 393   const base::ListValue* suggestion_details = NULL;
 394   const base::DictionaryValue* extras = NULL;
 395   int prefetch_index = -1;
 396   if (root_list->GetDictionary(4, &extras)) {
 397     extras->GetList("google:suggesttype", &types);
 398
 399     // Discard this list if its size does not match that of the suggestions.
 400     if (extras->GetList("google:suggestrelevance", &relevances) &&
 401         (relevances->GetSize() != results_list->GetSize()))
 402       relevances = NULL;
 403     extras->GetInteger("google:verbatimrelevance",
 404                        &results->verbatim_relevance);
 405
 406     // Check if the active suggest field trial (if any) has triggered either
 407     // for the default provider or keyword provider.
 408     results->field_trial_triggered = false;
 409     extras->GetBoolean("google:fieldtrialtriggered",
 410                        &results->field_trial_triggered);
 411
 412     const base::DictionaryValue* client_data = NULL;
 413     if (extras->GetDictionary("google:clientdata", &client_data) && client_data)
 414       client_data->GetInteger("phi", &prefetch_index);
 415
 416     if (extras->GetList("google:suggestdetail", &suggestion_details) &&
 417         suggestion_details->GetSize() != results_list->GetSize())
 418       suggestion_details = NULL;
 419
 420     // Store the metadata that came with the response in case we need to pass it
 421     // along with the prefetch query to Instant.
 422     JSONStringValueSerializer json_serializer(&results->metadata);
 423     json_serializer.Serialize(*extras);
 424   }
 425
 426   // Clear the previous results now that new results are available.
 427   results->suggest_results.clear();
 428   results->navigation_results.clear();
 429   results->answers_image_urls.clear();
 430
 431   base::string16 suggestion;
 432   std::string type;
 433   int relevance = default_result_relevance;
 434   // Prohibit navsuggest in FORCED_QUERY mode.  Users wants queries, not URLs.
 435   const bool allow_navsuggest =
 436       input.type() != metrics::OmniboxInputType::FORCED_QUERY;
 437   const base::string16& trimmed_input =
 438       base::CollapseWhitespace(input.text(), false);
 439   for (size_t index = 0; results_list->GetString(index, &suggestion); ++index) {
 440     // Google search may return empty suggestions for weird input characters,
 441     // they make no sense at all and can cause problems in our code.
 442     if (suggestion.empty())
 443       continue;
 444
 445     // Apply valid suggested relevance scores; discard invalid lists.
 446     if (relevances != NULL && !relevances->GetInteger(index, &relevance))
 447       relevances = NULL;
 448     AutocompleteMatchType::Type match_type =
 449         AutocompleteMatchType::SEARCH_SUGGEST;
 450     if (types && types->GetString(index, &type))
 451       match_type = GetAutocompleteMatchType(type);
 452     const base::DictionaryValue* suggestion_detail = NULL;
 453     std::string deletion_url;
 454
 455     if (suggestion_details &&
 456         suggestion_details->GetDictionary(index, &suggestion_detail))
 457       suggestion_detail->GetString("du", &deletion_url);
 458
 459     if ((match_type == AutocompleteMatchType::NAVSUGGEST) ||
 460         (match_type == AutocompleteMatchType::NAVSUGGEST_PERSONALIZED)) {
 461       // Do not blindly trust the URL coming from the server to be valid.
 462       GURL url(
 463           url_fixer::FixupURL(base::UTF16ToUTF8(suggestion), std::string()));
 464       if (url.is_valid() && allow_navsuggest) {
 465         base::string16 title;
 466         if (descriptions != NULL)
 467           descriptions->GetString(index, &title);
 468         results->navigation_results.push_back(NavigationResult(
 469             scheme_classifier, url, match_type, title, deletion_url,
 470             is_keyword_result, relevance, relevances != NULL, input.text(),
 471             languages));
 472       }
 473     } else {
 474       base::string16 match_contents = suggestion;
 475       base::string16 match_contents_prefix;
 476       base::string16 annotation;
 477       base::string16 answer_contents;
 478       base::string16 answer_type_str;
 479       scoped_ptr<SuggestionAnswer> answer;
 480       std::string suggest_query_params;
 481
 482       if (suggestion_details) {
 483         suggestion_details->GetDictionary(index, &suggestion_detail);
 484         if (suggestion_detail) {
 485           suggestion_detail->GetString("t", &match_contents);
 486           suggestion_detail->GetString("mp", &match_contents_prefix);
 487           // Error correction for bad data from server.
 488           if (match_contents.empty())
 489             match_contents = suggestion;
 490           suggestion_detail->GetString("a", &annotation);
 491           suggestion_detail->GetString("q", &suggest_query_params);
 492
 493           // Extract the Answer, if provided.
 494           const base::DictionaryValue* answer_json = NULL;
 495           if (suggestion_detail->GetDictionary("ansa", &answer_json) &&
 496               suggestion_detail->GetString("ansb", &answer_type_str)) {
 497             bool answer_parsed_successfully = false;
 498             answer = SuggestionAnswer::ParseAnswer(answer_json);
 499             int answer_type = 0;
 500             if (answer && base::StringToInt(answer_type_str, &answer_type)) {
 501               answer_parsed_successfully = true;
 502
 503               answer->set_type(answer_type);
 504               answer->AddImageURLsTo(&results->answers_image_urls);
 505
 506               std::string contents;
 507               base::JSONWriter::Write(answer_json, &contents);
 508               answer_contents = base::UTF8ToUTF16(contents);
 509             } else {
 510               answer_type_str = base::string16();
 511             }
 512             UMA_HISTOGRAM_BOOLEAN("Omnibox.AnswerParseSuccess",
 513                                   answer_parsed_successfully);
 514           }
 515         }
 516       }
 517
 518       bool should_prefetch = static_cast<int>(index) == prefetch_index;
 519       // TODO(kochi): Improve calculator suggestion presentation.
 520       results->suggest_results.push_back(SuggestResult(
 521           base::CollapseWhitespace(suggestion, false), match_type,
 522           base::CollapseWhitespace(match_contents, false),
 523           match_contents_prefix, annotation, answer_contents, answer_type_str,
 524           answer.Pass(), suggest_query_params, deletion_url, is_keyword_result,
 525           relevance, relevances != NULL, should_prefetch, trimmed_input));
 526     }
 527   }
 528   results->relevances_from_server = relevances != NULL;
 529   return true;
 530 }