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 <algorithm>
   8
   9 #include "base/i18n/icu_string_conversions.h"
  10 #include "base/json/json_string_value_serializer.h"
  11 #include "base/json/json_writer.h"
  12 #include "base/logging.h"
  13 #include "base/metrics/histogram.h"
  14 #include "base/strings/string_number_conversions.h"
  15 #include "base/strings/string_util.h"
  16 #include "base/strings/utf_string_conversions.h"
  17 #include "base/values.h"
  18 #include "components/omnibox/autocomplete_input.h"
  19 #include "components/omnibox/url_prefix.h"
  20 #include "components/url_fixer/url_fixer.h"
  21 #include "net/base/net_util.h"
  22 #include "net/http/http_response_headers.h"
  23 #include "net/url_request/url_fetcher.h"
  24 #include "url/url_constants.h"
  25
  26 namespace {
  27
  28 AutocompleteMatchType::Type GetAutocompleteMatchType(const std::string& type) {
  29   if (type == "CALCULATOR")
  30     return AutocompleteMatchType::CALCULATOR;
  31   if (type == "ENTITY")
  32     return AutocompleteMatchType::SEARCH_SUGGEST_ENTITY;
  33   if (type == "TAIL")
  34     return AutocompleteMatchType::SEARCH_SUGGEST_TAIL;
  35   if (type == "PERSONALIZED_QUERY")
  36     return AutocompleteMatchType::SEARCH_SUGGEST_PERSONALIZED;
  37   if (type == "PROFILE")
  38     return AutocompleteMatchType::SEARCH_SUGGEST_PROFILE;
  39   if (type == "NAVIGATION")
  40     return AutocompleteMatchType::NAVSUGGEST;
  41   if (type == "PERSONALIZED_NAVIGATION")
  42     return AutocompleteMatchType::NAVSUGGEST_PERSONALIZED;
  43   return AutocompleteMatchType::SEARCH_SUGGEST;
  44 }
  45
  46 }  // namespace
  47
  48 // SearchSuggestionParser::Result ----------------------------------------------
  49
  50 SearchSuggestionParser::Result::Result(bool from_keyword_provider,
  51                                        int relevance,
  52                                        bool relevance_from_server,
  53                                        AutocompleteMatchType::Type type,
  54                                        const std::string& deletion_url)
  55     : from_keyword_provider_(from_keyword_provider),
  56       type_(type),
  57       relevance_(relevance),
  58       relevance_from_server_(relevance_from_server),
  59       received_after_last_keystroke_(true),
  60       deletion_url_(deletion_url) {}
  61
  62 SearchSuggestionParser::Result::~Result() {}
  63
  64 // SearchSuggestionParser::SuggestResult ---------------------------------------
  65
  66 SearchSuggestionParser::SuggestResult::SuggestResult(
  67     const base::string16& suggestion,
  68     AutocompleteMatchType::Type type,
  69     const base::string16& match_contents,
  70     const base::string16& match_contents_prefix,
  71     const base::string16& annotation,
  72     const base::string16& answer_contents,
  73     const base::string16& answer_type,
  74     scoped_ptr<SuggestionAnswer> answer,
  75     const std::string& suggest_query_params,
  76     const std::string& deletion_url,
  77     bool from_keyword_provider,
  78     int relevance,
  79     bool relevance_from_server,
  80     bool should_prefetch,
  81     const base::string16& input_text)
  82     : Result(from_keyword_provider,
  83              relevance,
  84              relevance_from_server,
  85              type,
  86              deletion_url),
  87       suggestion_(suggestion),
  88       match_contents_prefix_(match_contents_prefix),
  89       annotation_(annotation),
  90       suggest_query_params_(suggest_query_params),
  91       answer_contents_(answer_contents),
  92       answer_type_(answer_type),
  93       answer_(answer.Pass()),
  94       should_prefetch_(should_prefetch) {
  95   match_contents_ = match_contents;
  96   DCHECK(!match_contents_.empty());
  97   ClassifyMatchContents(true, input_text);
  98 }
  99
 100 SearchSuggestionParser::SuggestResult::SuggestResult(
 101     const SuggestResult& result)
 102     : Result(result),
 103       suggestion_(result.suggestion_),
 104       match_contents_prefix_(result.match_contents_prefix_),
 105       annotation_(result.annotation_),
 106       suggest_query_params_(result.suggest_query_params_),
 107       answer_contents_(result.answer_contents_),
 108       answer_type_(result.answer_type_),
 109       answer_(SuggestionAnswer::copy(result.answer_.get())),
 110       should_prefetch_(result.should_prefetch_) {
 111 }
 112
 113 SearchSuggestionParser::SuggestResult::~SuggestResult() {}
 114
 115 SearchSuggestionParser::SuggestResult&
 116     SearchSuggestionParser::SuggestResult::operator=(const SuggestResult& rhs) {
 117   if (this == &rhs)
 118     return *this;
 119
 120   // Assign via parent class first.
 121   Result::operator=(rhs);
 122
 123   suggestion_ = rhs.suggestion_;
 124   match_contents_prefix_ = rhs.match_contents_prefix_;
 125   annotation_ = rhs.annotation_;
 126   suggest_query_params_ = rhs.suggest_query_params_;
 127   answer_contents_ = rhs.answer_contents_;
 128   answer_type_ = rhs.answer_type_;
 129   answer_ = SuggestionAnswer::copy(rhs.answer_.get());
 130   should_prefetch_ = rhs.should_prefetch_;
 131
 132   return *this;
 133 }
 134
 135 void SearchSuggestionParser::SuggestResult::ClassifyMatchContents(
 136     const bool allow_bolding_all,
 137     const base::string16& input_text) {
 138   if (input_text.empty()) {
 139     // In case of zero-suggest results, do not highlight matches.
 140     match_contents_class_.push_back(
 141         ACMatchClassification(0, ACMatchClassification::NONE));
 142     return;
 143   }
 144
 145   base::string16 lookup_text = input_text;
 146   if (type_ == AutocompleteMatchType::SEARCH_SUGGEST_TAIL) {
 147     const size_t contents_index =
 148         suggestion_.length() - match_contents_.length();
 149     // Ensure the query starts with the input text, and ends with the match
 150     // contents, and the input text has an overlap with contents.
 151     if (base::StartsWith(suggestion_, input_text, true) &&
 152         base::EndsWith(suggestion_, match_contents_, true) &&
 153         (input_text.length() > contents_index)) {
 154       lookup_text = input_text.substr(contents_index);
 155     }
 156   }
 157   // Do a case-insensitive search for |lookup_text|.
 158   base::string16::const_iterator lookup_position = std::search(
 159       match_contents_.begin(), match_contents_.end(), lookup_text.begin(),
 160       lookup_text.end(), base::CaseInsensitiveCompare<base::char16>());
 161   if (!allow_bolding_all && (lookup_position == match_contents_.end())) {
 162     // Bail if the code below to update the bolding would bold the whole
 163     // string.  Note that the string may already be entirely bolded; if
 164     // so, leave it as is.
 165     return;
 166   }
 167   match_contents_class_.clear();
 168   // We do intra-string highlighting for suggestions - the suggested segment
 169   // will be highlighted, e.g. for input_text = "you" the suggestion may be
 170   // "youtube", so we'll bold the "tube" section: you*tube*.
 171   if (input_text != match_contents_) {
 172     if (lookup_position == match_contents_.end()) {
 173       // The input text is not a substring of the query string, e.g. input
 174       // text is "slasdot" and the query string is "slashdot", so we bold the
 175       // whole thing.
 176       match_contents_class_.push_back(
 177           ACMatchClassification(0, ACMatchClassification::MATCH));
 178     } else {
 179       // We don't iterate over the string here annotating all matches because
 180       // it looks odd to have every occurrence of a substring that may be as
 181       // short as a single character highlighted in a query suggestion result,
 182       // e.g. for input text "s" and query string "southwest airlines", it
 183       // looks odd if both the first and last s are highlighted.
 184       const size_t lookup_index = lookup_position - match_contents_.begin();
 185       if (lookup_index != 0) {
 186         match_contents_class_.push_back(
 187             ACMatchClassification(0, ACMatchClassification::MATCH));
 188       }
 189       match_contents_class_.push_back(
 190           ACMatchClassification(lookup_index, ACMatchClassification::NONE));
 191       size_t next_fragment_position = lookup_index + lookup_text.length();
 192       if (next_fragment_position < match_contents_.length()) {
 193         match_contents_class_.push_back(ACMatchClassification(
 194             next_fragment_position, ACMatchClassification::MATCH));
 195       }
 196     }
 197   } else {
 198     // Otherwise, match_contents_ is a verbatim (what-you-typed) match, either
 199     // for the default provider or a keyword search provider.
 200     match_contents_class_.push_back(
 201         ACMatchClassification(0, ACMatchClassification::NONE));
 202   }
 203 }
 204
 205 int SearchSuggestionParser::SuggestResult::CalculateRelevance(
 206     const AutocompleteInput& input,
 207     bool keyword_provider_requested) const {
 208   if (!from_keyword_provider_ && keyword_provider_requested)
 209     return 100;
 210   return ((input.type() == metrics::OmniboxInputType::URL) ? 300 : 600);
 211 }
 212
 213 // SearchSuggestionParser::NavigationResult ------------------------------------
 214
 215 SearchSuggestionParser::NavigationResult::NavigationResult(
 216     const AutocompleteSchemeClassifier& scheme_classifier,
 217     const GURL& url,
 218     AutocompleteMatchType::Type type,
 219     const base::string16& description,
 220     const std::string& deletion_url,
 221     bool from_keyword_provider,
 222     int relevance,
 223     bool relevance_from_server,
 224     const base::string16& input_text,
 225     const std::string& languages)
 226     : Result(from_keyword_provider, relevance, relevance_from_server, type,
 227              deletion_url),
 228       url_(url),
 229       formatted_url_(AutocompleteInput::FormattedStringWithEquivalentMeaning(
 230           url, net::FormatUrl(url, languages,
 231                               net::kFormatUrlOmitAll & ~net::kFormatUrlOmitHTTP,
 232                               net::UnescapeRule::SPACES, NULL, NULL, NULL),
 233           scheme_classifier)),
 234       description_(description) {
 235   DCHECK(url_.is_valid());
 236   CalculateAndClassifyMatchContents(true, input_text, languages);
 237 }
 238
 239 SearchSuggestionParser::NavigationResult::~NavigationResult() {}
 240
 241 void
 242 SearchSuggestionParser::NavigationResult::CalculateAndClassifyMatchContents(
 243     const bool allow_bolding_nothing,
 244     const base::string16& input_text,
 245     const std::string& languages) {
 246   if (input_text.empty()) {
 247     // In case of zero-suggest results, do not highlight matches.
 248     match_contents_class_.push_back(
 249         ACMatchClassification(0, ACMatchClassification::NONE));
 250     return;
 251   }
 252
 253   // First look for the user's input inside the formatted url as it would be
 254   // without trimming the scheme, so we can find matches at the beginning of the
 255   // scheme.
 256   const URLPrefix* prefix =
 257       URLPrefix::BestURLPrefix(formatted_url_, input_text);
 258   size_t match_start = (prefix == NULL) ?
 259       formatted_url_.find(input_text) : prefix->prefix.length();
 260   bool trim_http = !AutocompleteInput::HasHTTPScheme(input_text) &&
 261                    (!prefix || (match_start != 0));
 262   const net::FormatUrlTypes format_types =
 263       net::kFormatUrlOmitAll & ~(trim_http ? 0 : net::kFormatUrlOmitHTTP);
 264
 265   base::string16 match_contents = net::FormatUrl(url_, languages, format_types,
 266       net::UnescapeRule::SPACES, NULL, NULL, &match_start);
 267   // If the first match in the untrimmed string was inside a scheme that we
 268   // trimmed, look for a subsequent match.
 269   if (match_start == base::string16::npos)
 270     match_start = match_contents.find(input_text);
 271   // Update |match_contents_| and |match_contents_class_| if it's allowed.
 272   if (allow_bolding_nothing || (match_start != base::string16::npos)) {
 273     match_contents_ = match_contents;
 274     // Safe if |match_start| is npos; also safe if the input is longer than the
 275     // remaining contents after |match_start|.
 276     AutocompleteMatch::ClassifyLocationInString(match_start,
 277         input_text.length(), match_contents_.length(),
 278         ACMatchClassification::URL, &match_contents_class_);
 279   }
 280 }
 281
 282 int SearchSuggestionParser::NavigationResult::CalculateRelevance(
 283     const AutocompleteInput& input,
 284     bool keyword_provider_requested) const {
 285   return (from_keyword_provider_ || !keyword_provider_requested) ? 800 : 150;
 286 }
 287
 288 // SearchSuggestionParser::Results ---------------------------------------------
 289
 290 SearchSuggestionParser::Results::Results()
 291     : verbatim_relevance(-1),
 292       field_trial_triggered(false),
 293       relevances_from_server(false) {}
 294
 295 SearchSuggestionParser::Results::~Results() {}
 296
 297 void SearchSuggestionParser::Results::Clear() {
 298   suggest_results.clear();
 299   navigation_results.clear();
 300   verbatim_relevance = -1;
 301   metadata.clear();
 302 }
 303
 304 bool SearchSuggestionParser::Results::HasServerProvidedScores() const {
 305   if (verbatim_relevance >= 0)
 306     return true;
 307
 308   // Right now either all results of one type will be server-scored or they will
 309   // all be locally scored, but in case we change this later, we'll just check
 310   // them all.
 311   for (SuggestResults::const_iterator i(suggest_results.begin());
 312        i != suggest_results.end(); ++i) {
 313     if (i->relevance_from_server())
 314       return true;
 315   }
 316   for (NavigationResults::const_iterator i(navigation_results.begin());
 317        i != navigation_results.end(); ++i) {
 318     if (i->relevance_from_server())
 319       return true;
 320   }
 321
 322   return false;
 323 }
 324
 325 // SearchSuggestionParser ------------------------------------------------------
 326
 327 // static
 328 std::string SearchSuggestionParser::ExtractJsonData(
 329     const net::URLFetcher* source) {
 330   const net::HttpResponseHeaders* const response_headers =
 331       source->GetResponseHeaders();
 332   std::string json_data;
 333   source->GetResponseAsString(&json_data);
 334
 335   // JSON is supposed to be UTF-8, but some suggest service providers send
 336   // JSON files in non-UTF-8 encodings.  The actual encoding is usually
 337   // specified in the Content-Type header field.
 338   if (response_headers) {
 339     std::string charset;
 340     if (response_headers->GetCharset(&charset)) {
 341       base::string16 data_16;
 342       // TODO(jungshik): Switch to CodePageToUTF8 after it's added.
 343       if (base::CodepageToUTF16(json_data, charset.c_str(),
 344                                 base::OnStringConversionError::FAIL,
 345                                 &data_16))
 346         json_data = base::UTF16ToUTF8(data_16);
 347     }
 348   }
 349   return json_data;
 350 }
 351
 352 // static
 353 scoped_ptr<base::Value> SearchSuggestionParser::DeserializeJsonData(
 354     base::StringPiece json_data) {
 355   // The JSON response should be an array.
 356   for (size_t response_start_index = json_data.find("["), i = 0;
 357        response_start_index != base::StringPiece::npos && i < 5;
 358        response_start_index = json_data.find("[", 1), i++) {
 359     // Remove any XSSI guards to allow for JSON parsing.
 360     json_data.remove_prefix(response_start_index);
 361
 362     JSONStringValueDeserializer deserializer(json_data);
 363     deserializer.set_allow_trailing_comma(true);
 364     int error_code = 0;
 365     scoped_ptr<base::Value> data(deserializer.Deserialize(&error_code, NULL));
 366     if (error_code == 0)
 367       return data.Pass();
 368   }
 369   return scoped_ptr<base::Value>();
 370 }
 371
 372 // static
 373 bool SearchSuggestionParser::ParseSuggestResults(
 374     const base::Value& root_val,
 375     const AutocompleteInput& input,
 376     const AutocompleteSchemeClassifier& scheme_classifier,
 377     int default_result_relevance,
 378     const std::string& languages,
 379     bool is_keyword_result,
 380     Results* results) {
 381   base::string16 query;
 382   const base::ListValue* root_list = NULL;
 383   const base::ListValue* results_list = NULL;
 384
 385   if (!root_val.GetAsList(&root_list) || !root_list->GetString(0, &query) ||
 386       query != input.text() || !root_list->GetList(1, &results_list))
 387     return false;
 388
 389   // 3rd element: Description list.
 390   const base::ListValue* descriptions = NULL;
 391   root_list->GetList(2, &descriptions);
 392
 393   // 4th element: Disregard the query URL list for now.
 394
 395   // Reset suggested relevance information.
 396   results->verbatim_relevance = -1;
 397
 398   // 5th element: Optional key-value pairs from the Suggest server.
 399   const base::ListValue* types = NULL;
 400   const base::ListValue* relevances = NULL;
 401   const base::ListValue* suggestion_details = NULL;
 402   const base::DictionaryValue* extras = NULL;
 403   int prefetch_index = -1;
 404   if (root_list->GetDictionary(4, &extras)) {
 405     extras->GetList("google:suggesttype", &types);
 406
 407     // Discard this list if its size does not match that of the suggestions.
 408     if (extras->GetList("google:suggestrelevance", &relevances) &&
 409         (relevances->GetSize() != results_list->GetSize()))
 410       relevances = NULL;
 411     extras->GetInteger("google:verbatimrelevance",
 412                        &results->verbatim_relevance);
 413
 414     // Check if the active suggest field trial (if any) has triggered either
 415     // for the default provider or keyword provider.
 416     results->field_trial_triggered = false;
 417     extras->GetBoolean("google:fieldtrialtriggered",
 418                        &results->field_trial_triggered);
 419
 420     const base::DictionaryValue* client_data = NULL;
 421     if (extras->GetDictionary("google:clientdata", &client_data) && client_data)
 422       client_data->GetInteger("phi", &prefetch_index);
 423
 424     if (extras->GetList("google:suggestdetail", &suggestion_details) &&
 425         suggestion_details->GetSize() != results_list->GetSize())
 426       suggestion_details = NULL;
 427
 428     // Store the metadata that came with the response in case we need to pass it
 429     // along with the prefetch query to Instant.
 430     JSONStringValueSerializer json_serializer(&results->metadata);
 431     json_serializer.Serialize(*extras);
 432   }
 433
 434   // Clear the previous results now that new results are available.
 435   results->suggest_results.clear();
 436   results->navigation_results.clear();
 437   results->answers_image_urls.clear();
 438
 439   base::string16 suggestion;
 440   std::string type;
 441   int relevance = default_result_relevance;
 442   // Prohibit navsuggest in FORCED_QUERY mode.  Users wants queries, not URLs.
 443   const bool allow_navsuggest =
 444       input.type() != metrics::OmniboxInputType::FORCED_QUERY;
 445   const base::string16& trimmed_input =
 446       base::CollapseWhitespace(input.text(), false);
 447   for (size_t index = 0; results_list->GetString(index, &suggestion); ++index) {
 448     // Google search may return empty suggestions for weird input characters,
 449     // they make no sense at all and can cause problems in our code.
 450     if (suggestion.empty())
 451       continue;
 452
 453     // Apply valid suggested relevance scores; discard invalid lists.
 454     if (relevances != NULL && !relevances->GetInteger(index, &relevance))
 455       relevances = NULL;
 456     AutocompleteMatchType::Type match_type =
 457         AutocompleteMatchType::SEARCH_SUGGEST;
 458     if (types && types->GetString(index, &type))
 459       match_type = GetAutocompleteMatchType(type);
 460     const base::DictionaryValue* suggestion_detail = NULL;
 461     std::string deletion_url;
 462
 463     if (suggestion_details &&
 464         suggestion_details->GetDictionary(index, &suggestion_detail))
 465       suggestion_detail->GetString("du", &deletion_url);
 466
 467     if ((match_type == AutocompleteMatchType::NAVSUGGEST) ||
 468         (match_type == AutocompleteMatchType::NAVSUGGEST_PERSONALIZED)) {
 469       // Do not blindly trust the URL coming from the server to be valid.
 470       GURL url(
 471           url_fixer::FixupURL(base::UTF16ToUTF8(suggestion), std::string()));
 472       if (url.is_valid() && allow_navsuggest) {
 473         base::string16 title;
 474         if (descriptions != NULL)
 475           descriptions->GetString(index, &title);
 476         results->navigation_results.push_back(NavigationResult(
 477             scheme_classifier, url, match_type, title, deletion_url,
 478             is_keyword_result, relevance, relevances != NULL, input.text(),
 479             languages));
 480       }
 481     } else {
 482       // TODO(dschuyler) If the "= " is no longer sent from the back-end
 483       // then this may be removed.
 484       if ((match_type == AutocompleteMatchType::CALCULATOR) &&
 485           !suggestion.compare(0, 2, base::UTF8ToUTF16("= ")))
 486         suggestion.erase(0, 2);
 487
 488       base::string16 match_contents = suggestion;
 489       base::string16 match_contents_prefix;
 490       base::string16 annotation;
 491       base::string16 answer_contents;
 492       base::string16 answer_type_str;
 493       scoped_ptr<SuggestionAnswer> answer;
 494       std::string suggest_query_params;
 495
 496       if (suggestion_details) {
 497         suggestion_details->GetDictionary(index, &suggestion_detail);
 498         if (suggestion_detail) {
 499           suggestion_detail->GetString("t", &match_contents);
 500           suggestion_detail->GetString("mp", &match_contents_prefix);
 501           // Error correction for bad data from server.
 502           if (match_contents.empty())
 503             match_contents = suggestion;
 504           suggestion_detail->GetString("a", &annotation);
 505           suggestion_detail->GetString("q", &suggest_query_params);
 506
 507           // Extract the Answer, if provided.
 508           const base::DictionaryValue* answer_json = NULL;
 509           if (suggestion_detail->GetDictionary("ansa", &answer_json) &&
 510               suggestion_detail->GetString("ansb", &answer_type_str)) {
 511             bool answer_parsed_successfully = false;
 512             answer = SuggestionAnswer::ParseAnswer(answer_json);
 513             int answer_type = 0;
 514             if (answer && base::StringToInt(answer_type_str, &answer_type)) {
 515               answer_parsed_successfully = true;
 516
 517               answer->set_type(answer_type);
 518               answer->AddImageURLsTo(&results->answers_image_urls);
 519
 520               std::string contents;
 521               base::JSONWriter::Write(*answer_json, &contents);
 522               answer_contents = base::UTF8ToUTF16(contents);
 523             } else {
 524               answer_type_str = base::string16();
 525             }
 526             UMA_HISTOGRAM_BOOLEAN("Omnibox.AnswerParseSuccess",
 527                                   answer_parsed_successfully);
 528           }
 529         }
 530       }
 531
 532       bool should_prefetch = static_cast<int>(index) == prefetch_index;
 533       results->suggest_results.push_back(SuggestResult(
 534           base::CollapseWhitespace(suggestion, false), match_type,
 535           base::CollapseWhitespace(match_contents, false),
 536           match_contents_prefix, annotation, answer_contents, answer_type_str,
 537           answer.Pass(), suggest_query_params, deletion_url, is_keyword_result,
 538           relevance, relevances != NULL, should_prefetch, trimmed_input));
 539     }
 540   }
 541   results->relevances_from_server = relevances != NULL;
 542   return true;
 543 }