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_macros.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,
 152                          base::CompareCase::SENSITIVE) &&
 153         base::EndsWith(suggestion_, match_contents_, true) &&
 154         (input_text.length() > contents_index)) {
 155       lookup_text = input_text.substr(contents_index);
 156     }
 157   }
 158   // Do a case-insensitive search for |lookup_text|.
 159   base::string16::const_iterator lookup_position = std::search(
 160       match_contents_.begin(), match_contents_.end(), lookup_text.begin(),
 161       lookup_text.end(), base::CaseInsensitiveCompare<base::char16>());
 162   if (!allow_bolding_all && (lookup_position == match_contents_.end())) {
 163     // Bail if the code below to update the bolding would bold the whole
 164     // string.  Note that the string may already be entirely bolded; if
 165     // so, leave it as is.
 166     return;
 167   }
 168   match_contents_class_.clear();
 169   // We do intra-string highlighting for suggestions - the suggested segment
 170   // will be highlighted, e.g. for input_text = "you" the suggestion may be
 171   // "youtube", so we'll bold the "tube" section: you*tube*.
 172   if (input_text != match_contents_) {
 173     if (lookup_position == match_contents_.end()) {
 174       // The input text is not a substring of the query string, e.g. input
 175       // text is "slasdot" and the query string is "slashdot", so we bold the
 176       // whole thing.
 177       match_contents_class_.push_back(
 178           ACMatchClassification(0, ACMatchClassification::MATCH));
 179     } else {
 180       // We don't iterate over the string here annotating all matches because
 181       // it looks odd to have every occurrence of a substring that may be as
 182       // short as a single character highlighted in a query suggestion result,
 183       // e.g. for input text "s" and query string "southwest airlines", it
 184       // looks odd if both the first and last s are highlighted.
 185       const size_t lookup_index = lookup_position - match_contents_.begin();
 186       if (lookup_index != 0) {
 187         match_contents_class_.push_back(
 188             ACMatchClassification(0, ACMatchClassification::MATCH));
 189       }
 190       match_contents_class_.push_back(
 191           ACMatchClassification(lookup_index, ACMatchClassification::NONE));
 192       size_t next_fragment_position = lookup_index + lookup_text.length();
 193       if (next_fragment_position < match_contents_.length()) {
 194         match_contents_class_.push_back(ACMatchClassification(
 195             next_fragment_position, ACMatchClassification::MATCH));
 196       }
 197     }
 198   } else {
 199     // Otherwise, match_contents_ is a verbatim (what-you-typed) match, either
 200     // for the default provider or a keyword search provider.
 201     match_contents_class_.push_back(
 202         ACMatchClassification(0, ACMatchClassification::NONE));
 203   }
 204 }
 205
 206 int SearchSuggestionParser::SuggestResult::CalculateRelevance(
 207     const AutocompleteInput& input,
 208     bool keyword_provider_requested) const {
 209   if (!from_keyword_provider_ && keyword_provider_requested)
 210     return 100;
 211   return ((input.type() == metrics::OmniboxInputType::URL) ? 300 : 600);
 212 }
 213
 214 // SearchSuggestionParser::NavigationResult ------------------------------------
 215
 216 SearchSuggestionParser::NavigationResult::NavigationResult(
 217     const AutocompleteSchemeClassifier& scheme_classifier,
 218     const GURL& url,
 219     AutocompleteMatchType::Type type,
 220     const base::string16& description,
 221     const std::string& deletion_url,
 222     bool from_keyword_provider,
 223     int relevance,
 224     bool relevance_from_server,
 225     const base::string16& input_text,
 226     const std::string& languages)
 227     : Result(from_keyword_provider, relevance, relevance_from_server, type,
 228              deletion_url),
 229       url_(url),
 230       formatted_url_(AutocompleteInput::FormattedStringWithEquivalentMeaning(
 231           url, net::FormatUrl(url, languages,
 232                               net::kFormatUrlOmitAll & ~net::kFormatUrlOmitHTTP,
 233                               net::UnescapeRule::SPACES, NULL, NULL, NULL),
 234           scheme_classifier)),
 235       description_(description) {
 236   DCHECK(url_.is_valid());
 237   CalculateAndClassifyMatchContents(true, input_text, languages);
 238 }
 239
 240 SearchSuggestionParser::NavigationResult::~NavigationResult() {}
 241
 242 void
 243 SearchSuggestionParser::NavigationResult::CalculateAndClassifyMatchContents(
 244     const bool allow_bolding_nothing,
 245     const base::string16& input_text,
 246     const std::string& languages) {
 247   if (input_text.empty()) {
 248     // In case of zero-suggest results, do not highlight matches.
 249     match_contents_class_.push_back(
 250         ACMatchClassification(0, ACMatchClassification::NONE));
 251     return;
 252   }
 253
 254   // First look for the user's input inside the formatted url as it would be
 255   // without trimming the scheme, so we can find matches at the beginning of the
 256   // scheme.
 257   const URLPrefix* prefix =
 258       URLPrefix::BestURLPrefix(formatted_url_, input_text);
 259   size_t match_start = (prefix == NULL) ?
 260       formatted_url_.find(input_text) : prefix->prefix.length();
 261   bool trim_http = !AutocompleteInput::HasHTTPScheme(input_text) &&
 262                    (!prefix || (match_start != 0));
 263   const net::FormatUrlTypes format_types =
 264       net::kFormatUrlOmitAll & ~(trim_http ? 0 : net::kFormatUrlOmitHTTP);
 265
 266   base::string16 match_contents = net::FormatUrl(url_, languages, format_types,
 267       net::UnescapeRule::SPACES, NULL, NULL, &match_start);
 268   // If the first match in the untrimmed string was inside a scheme that we
 269   // trimmed, look for a subsequent match.
 270   if (match_start == base::string16::npos)
 271     match_start = match_contents.find(input_text);
 272   // Update |match_contents_| and |match_contents_class_| if it's allowed.
 273   if (allow_bolding_nothing || (match_start != base::string16::npos)) {
 274     match_contents_ = match_contents;
 275     // Safe if |match_start| is npos; also safe if the input is longer than the
 276     // remaining contents after |match_start|.
 277     AutocompleteMatch::ClassifyLocationInString(match_start,
 278         input_text.length(), match_contents_.length(),
 279         ACMatchClassification::URL, &match_contents_class_);
 280   }
 281 }
 282
 283 int SearchSuggestionParser::NavigationResult::CalculateRelevance(
 284     const AutocompleteInput& input,
 285     bool keyword_provider_requested) const {
 286   return (from_keyword_provider_ || !keyword_provider_requested) ? 800 : 150;
 287 }
 288
 289 // SearchSuggestionParser::Results ---------------------------------------------
 290
 291 SearchSuggestionParser::Results::Results()
 292     : verbatim_relevance(-1),
 293       field_trial_triggered(false),
 294       relevances_from_server(false) {}
 295
 296 SearchSuggestionParser::Results::~Results() {}
 297
 298 void SearchSuggestionParser::Results::Clear() {
 299   suggest_results.clear();
 300   navigation_results.clear();
 301   verbatim_relevance = -1;
 302   metadata.clear();
 303 }
 304
 305 bool SearchSuggestionParser::Results::HasServerProvidedScores() const {
 306   if (verbatim_relevance >= 0)
 307     return true;
 308
 309   // Right now either all results of one type will be server-scored or they will
 310   // all be locally scored, but in case we change this later, we'll just check
 311   // them all.
 312   for (SuggestResults::const_iterator i(suggest_results.begin());
 313        i != suggest_results.end(); ++i) {
 314     if (i->relevance_from_server())
 315       return true;
 316   }
 317   for (NavigationResults::const_iterator i(navigation_results.begin());
 318        i != navigation_results.end(); ++i) {
 319     if (i->relevance_from_server())
 320       return true;
 321   }
 322
 323   return false;
 324 }
 325
 326 // SearchSuggestionParser ------------------------------------------------------
 327
 328 // static
 329 std::string SearchSuggestionParser::ExtractJsonData(
 330     const net::URLFetcher* source) {
 331   const net::HttpResponseHeaders* const response_headers =
 332       source->GetResponseHeaders();
 333   std::string json_data;
 334   source->GetResponseAsString(&json_data);
 335
 336   // JSON is supposed to be UTF-8, but some suggest service providers send
 337   // JSON files in non-UTF-8 encodings.  The actual encoding is usually
 338   // specified in the Content-Type header field.
 339   if (response_headers) {
 340     std::string charset;
 341     if (response_headers->GetCharset(&charset)) {
 342       base::string16 data_16;
 343       // TODO(jungshik): Switch to CodePageToUTF8 after it's added.
 344       if (base::CodepageToUTF16(json_data, charset.c_str(),
 345                                 base::OnStringConversionError::FAIL,
 346                                 &data_16))
 347         json_data = base::UTF16ToUTF8(data_16);
 348     }
 349   }
 350   return json_data;
 351 }
 352
 353 // static
 354 scoped_ptr<base::Value> SearchSuggestionParser::DeserializeJsonData(
 355     base::StringPiece json_data) {
 356   // The JSON response should be an array.
 357   for (size_t response_start_index = json_data.find("["), i = 0;
 358        response_start_index != base::StringPiece::npos && i < 5;
 359        response_start_index = json_data.find("[", 1), i++) {
 360     // Remove any XSSI guards to allow for JSON parsing.
 361     json_data.remove_prefix(response_start_index);
 362
 363     JSONStringValueDeserializer deserializer(json_data);
 364     deserializer.set_allow_trailing_comma(true);
 365     int error_code = 0;
 366     scoped_ptr<base::Value> data(deserializer.Deserialize(&error_code, NULL));
 367     if (error_code == 0)
 368       return data.Pass();
 369   }
 370   return scoped_ptr<base::Value>();
 371 }
 372
 373 // static
 374 bool SearchSuggestionParser::ParseSuggestResults(
 375     const base::Value& root_val,
 376     const AutocompleteInput& input,
 377     const AutocompleteSchemeClassifier& scheme_classifier,
 378     int default_result_relevance,
 379     const std::string& languages,
 380     bool is_keyword_result,
 381     Results* results) {
 382   base::string16 query;
 383   const base::ListValue* root_list = NULL;
 384   const base::ListValue* results_list = NULL;
 385
 386   if (!root_val.GetAsList(&root_list) || !root_list->GetString(0, &query) ||
 387       query != input.text() || !root_list->GetList(1, &results_list))
 388     return false;
 389
 390   // 3rd element: Description list.
 391   const base::ListValue* descriptions = NULL;
 392   root_list->GetList(2, &descriptions);
 393
 394   // 4th element: Disregard the query URL list for now.
 395
 396   // Reset suggested relevance information.
 397   results->verbatim_relevance = -1;
 398
 399   // 5th element: Optional key-value pairs from the Suggest server.
 400   const base::ListValue* types = NULL;
 401   const base::ListValue* relevances = NULL;
 402   const base::ListValue* suggestion_details = NULL;
 403   const base::DictionaryValue* extras = NULL;
 404   int prefetch_index = -1;
 405   if (root_list->GetDictionary(4, &extras)) {
 406     extras->GetList("google:suggesttype", &types);
 407
 408     // Discard this list if its size does not match that of the suggestions.
 409     if (extras->GetList("google:suggestrelevance", &relevances) &&
 410         (relevances->GetSize() != results_list->GetSize()))
 411       relevances = NULL;
 412     extras->GetInteger("google:verbatimrelevance",
 413                        &results->verbatim_relevance);
 414
 415     // Check if the active suggest field trial (if any) has triggered either
 416     // for the default provider or keyword provider.
 417     results->field_trial_triggered = false;
 418     extras->GetBoolean("google:fieldtrialtriggered",
 419                        &results->field_trial_triggered);
 420
 421     const base::DictionaryValue* client_data = NULL;
 422     if (extras->GetDictionary("google:clientdata", &client_data) && client_data)
 423       client_data->GetInteger("phi", &prefetch_index);
 424
 425     if (extras->GetList("google:suggestdetail", &suggestion_details) &&
 426         suggestion_details->GetSize() != results_list->GetSize())
 427       suggestion_details = NULL;
 428
 429     // Store the metadata that came with the response in case we need to pass it
 430     // along with the prefetch query to Instant.
 431     JSONStringValueSerializer json_serializer(&results->metadata);
 432     json_serializer.Serialize(*extras);
 433   }
 434
 435   // Clear the previous results now that new results are available.
 436   results->suggest_results.clear();
 437   results->navigation_results.clear();
 438   results->answers_image_urls.clear();
 439
 440   base::string16 suggestion;
 441   std::string type;
 442   int relevance = default_result_relevance;
 443   // Prohibit navsuggest in FORCED_QUERY mode.  Users wants queries, not URLs.
 444   const bool allow_navsuggest =
 445       input.type() != metrics::OmniboxInputType::FORCED_QUERY;
 446   const base::string16& trimmed_input =
 447       base::CollapseWhitespace(input.text(), false);
 448   for (size_t index = 0; results_list->GetString(index, &suggestion); ++index) {
 449     // Google search may return empty suggestions for weird input characters,
 450     // they make no sense at all and can cause problems in our code.
 451     if (suggestion.empty())
 452       continue;
 453
 454     // Apply valid suggested relevance scores; discard invalid lists.
 455     if (relevances != NULL && !relevances->GetInteger(index, &relevance))
 456       relevances = NULL;
 457     AutocompleteMatchType::Type match_type =
 458         AutocompleteMatchType::SEARCH_SUGGEST;
 459     if (types && types->GetString(index, &type))
 460       match_type = GetAutocompleteMatchType(type);
 461     const base::DictionaryValue* suggestion_detail = NULL;
 462     std::string deletion_url;
 463
 464     if (suggestion_details &&
 465         suggestion_details->GetDictionary(index, &suggestion_detail))
 466       suggestion_detail->GetString("du", &deletion_url);
 467
 468     if ((match_type == AutocompleteMatchType::NAVSUGGEST) ||
 469         (match_type == AutocompleteMatchType::NAVSUGGEST_PERSONALIZED)) {
 470       // Do not blindly trust the URL coming from the server to be valid.
 471       GURL url(
 472           url_fixer::FixupURL(base::UTF16ToUTF8(suggestion), std::string()));
 473       if (url.is_valid() && allow_navsuggest) {
 474         base::string16 title;
 475         if (descriptions != NULL)
 476           descriptions->GetString(index, &title);
 477         results->navigation_results.push_back(NavigationResult(
 478             scheme_classifier, url, match_type, title, deletion_url,
 479             is_keyword_result, relevance, relevances != NULL, input.text(),
 480             languages));
 481       }
 482     } else {
 483       // TODO(dschuyler) If the "= " is no longer sent from the back-end
 484       // then this may be removed.
 485       if ((match_type == AutocompleteMatchType::CALCULATOR) &&
 486           !suggestion.compare(0, 2, base::UTF8ToUTF16("= ")))
 487         suggestion.erase(0, 2);
 488
 489       base::string16 match_contents = suggestion;
 490       base::string16 match_contents_prefix;
 491       base::string16 annotation;
 492       base::string16 answer_contents;
 493       base::string16 answer_type_str;
 494       scoped_ptr<SuggestionAnswer> answer;
 495       std::string suggest_query_params;
 496
 497       if (suggestion_details) {
 498         suggestion_details->GetDictionary(index, &suggestion_detail);
 499         if (suggestion_detail) {
 500           suggestion_detail->GetString("t", &match_contents);
 501           suggestion_detail->GetString("mp", &match_contents_prefix);
 502           // Error correction for bad data from server.
 503           if (match_contents.empty())
 504             match_contents = suggestion;
 505           suggestion_detail->GetString("a", &annotation);
 506           suggestion_detail->GetString("q", &suggest_query_params);
 507
 508           // Extract the Answer, if provided.
 509           const base::DictionaryValue* answer_json = NULL;
 510           if (suggestion_detail->GetDictionary("ansa", &answer_json) &&
 511               suggestion_detail->GetString("ansb", &answer_type_str)) {
 512             bool answer_parsed_successfully = false;
 513             answer = SuggestionAnswer::ParseAnswer(answer_json);
 514             int answer_type = 0;
 515             if (answer && base::StringToInt(answer_type_str, &answer_type)) {
 516               answer_parsed_successfully = true;
 517
 518               answer->set_type(answer_type);
 519               answer->AddImageURLsTo(&results->answers_image_urls);
 520
 521               std::string contents;
 522               base::JSONWriter::Write(*answer_json, &contents);
 523               answer_contents = base::UTF8ToUTF16(contents);
 524             } else {
 525               answer_type_str = base::string16();
 526             }
 527             UMA_HISTOGRAM_BOOLEAN("Omnibox.AnswerParseSuccess",
 528                                   answer_parsed_successfully);
 529           }
 530         }
 531       }
 532
 533       bool should_prefetch = static_cast<int>(index) == prefetch_index;
 534       results->suggest_results.push_back(SuggestResult(
 535           base::CollapseWhitespace(suggestion, false), match_type,
 536           base::CollapseWhitespace(match_contents, false),
 537           match_contents_prefix, annotation, answer_contents, answer_type_str,
 538           answer.Pass(), suggest_query_params, deletion_url, is_keyword_result,
 539           relevance, relevances != NULL, should_prefetch, trimmed_input));
 540     }
 541   }
 542   results->relevances_from_server = relevances != NULL;
 543   return true;
 544 }