components/search_engines/template_url.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/search_engines/template_url.h"
   6
   7 #include <string>
   8 #include <vector>
   9
  10 #include "base/basictypes.h"
  11 #include "base/command_line.h"
  12 #include "base/format_macros.h"
  13 #include "base/i18n/icu_string_conversions.h"
  14 #include "base/i18n/rtl.h"
  15 #include "base/logging.h"
  16 #include "base/metrics/field_trial.h"
  17 #include "base/rand_util.h"
  18 #include "base/strings/string_number_conversions.h"
  19 #include "base/strings/string_split.h"
  20 #include "base/strings/string_util.h"
  21 #include "base/strings/stringprintf.h"
  22 #include "base/strings/utf_string_conversions.h"
  23 #include "components/google/core/browser/google_util.h"
  24 #include "components/metrics/proto/omnibox_input_type.pb.h"
  25 #include "components/search_engines/search_engines_switches.h"
  26 #include "components/search_engines/search_terms_data.h"
  27 #include "google_apis/google_api_keys.h"
  28 #include "net/base/escape.h"
  29 #include "net/base/mime_util.h"
  30 #include "net/base/net_util.h"
  31
  32 namespace {
  33
  34 // The TemplateURLRef has any number of terms that need to be replaced. Each of
  35 // the terms is enclosed in braces. If the character preceeding the final
  36 // brace is a ?, it indicates the term is optional and can be replaced with
  37 // an empty string.
  38 const char kStartParameter = '{';
  39 const char kEndParameter = '}';
  40 const char kOptional = '?';
  41
  42 // Known parameters found in the URL.
  43 const char kSearchTermsParameter[] = "searchTerms";
  44 const char kSearchTermsParameterFull[] = "{searchTerms}";
  45 const char kCountParameter[] = "count";
  46 const char kStartIndexParameter[] = "startIndex";
  47 const char kStartPageParameter[] = "startPage";
  48 const char kLanguageParameter[] = "language";
  49 const char kInputEncodingParameter[] = "inputEncoding";
  50 const char kOutputEncodingParameter[] = "outputEncoding";
  51
  52 const char kGoogleAssistedQueryStatsParameter[] = "google:assistedQueryStats";
  53
  54 // Host/Domain Google searches are relative to.
  55 const char kGoogleBaseURLParameter[] = "google:baseURL";
  56 const char kGoogleBaseURLParameterFull[] = "{google:baseURL}";
  57
  58 // Like google:baseURL, but for the Search Suggest capability.
  59 const char kGoogleBaseSuggestURLParameter[] = "google:baseSuggestURL";
  60 const char kGoogleBaseSuggestURLParameterFull[] = "{google:baseSuggestURL}";
  61 const char kGoogleBookmarkBarPinnedParameter[] = "google:bookmarkBarPinned";
  62 const char kGoogleContextualSearchContextData[] =
  63     "google:contextualSearchContextData";
  64 const char kGoogleContextualSearchVersion[] = "google:contextualSearchVersion";
  65 const char kGoogleCurrentPageUrlParameter[] = "google:currentPageUrl";
  66 const char kGoogleCursorPositionParameter[] = "google:cursorPosition";
  67 const char kGoogleForceInstantResultsParameter[] = "google:forceInstantResults";
  68 const char kGoogleImageSearchSource[] = "google:imageSearchSource";
  69 const char kGoogleImageThumbnailParameter[] = "google:imageThumbnail";
  70 const char kGoogleImageOriginalWidth[] = "google:imageOriginalWidth";
  71 const char kGoogleImageOriginalHeight[] = "google:imageOriginalHeight";
  72 const char kGoogleImageURLParameter[] = "google:imageURL";
  73 const char kGoogleInputTypeParameter[] = "google:inputType";
  74 const char kGoogleInstantExtendedEnabledParameter[] =
  75     "google:instantExtendedEnabledParameter";
  76 const char kGoogleInstantExtendedEnabledKey[] =
  77     "google:instantExtendedEnabledKey";
  78 const char kGoogleInstantExtendedEnabledKeyFull[] =
  79     "{google:instantExtendedEnabledKey}";
  80 const char kGoogleNTPIsThemedParameter[] = "google:ntpIsThemedParameter";
  81 const char kGoogleOmniboxStartMarginParameter[] =
  82     "google:omniboxStartMarginParameter";
  83 const char kGoogleOriginalQueryForSuggestionParameter[] =
  84     "google:originalQueryForSuggestion";
  85 const char kGooglePageClassificationParameter[] = "google:pageClassification";
  86 const char kGooglePrefetchQuery[] = "google:prefetchQuery";
  87 const char kGoogleRLZParameter[] = "google:RLZ";
  88 const char kGoogleSearchClient[] = "google:searchClient";
  89 const char kGoogleSearchFieldtrialParameter[] =
  90     "google:searchFieldtrialParameter";
  91 const char kGoogleSearchVersion[] = "google:searchVersion";
  92 const char kGoogleSessionToken[] = "google:sessionToken";
  93 const char kGoogleSourceIdParameter[] = "google:sourceId";
  94 const char kGoogleSuggestAPIKeyParameter[] = "google:suggestAPIKeyParameter";
  95 const char kGoogleSuggestClient[] = "google:suggestClient";
  96 const char kGoogleSuggestRequestId[] = "google:suggestRid";
  97
  98 // Same as kSearchTermsParameter, with no escaping.
  99 const char kGoogleUnescapedSearchTermsParameter[] =
 100     "google:unescapedSearchTerms";
 101 const char kGoogleUnescapedSearchTermsParameterFull[] =
 102     "{google:unescapedSearchTerms}";
 103
 104 // Display value for kSearchTermsParameter.
 105 const char kDisplaySearchTerms[] = "%s";
 106
 107 // Display value for kGoogleUnescapedSearchTermsParameter.
 108 const char kDisplayUnescapedSearchTerms[] = "%S";
 109
 110 // Used if the count parameter is not optional. Indicates we want 10 search
 111 // results.
 112 const char kDefaultCount[] = "10";
 113
 114 // Used if the parameter kOutputEncodingParameter is required.
 115 const char kOutputEncodingType[] = "UTF-8";
 116
 117 // Attempts to encode |terms| and |original_query| in |encoding| and escape
 118 // them.  |terms| may be escaped as path or query depending on |is_in_query|;
 119 // |original_query| is always escaped as query.  Returns whether the encoding
 120 // process succeeded.
 121 bool TryEncoding(const base::string16& terms,
 122                  const base::string16& original_query,
 123                  const char* encoding,
 124                  bool is_in_query,
 125                  base::string16* escaped_terms,
 126                  base::string16* escaped_original_query) {
 127   DCHECK(escaped_terms);
 128   DCHECK(escaped_original_query);
 129   std::string encoded_terms;
 130   if (!base::UTF16ToCodepage(terms, encoding,
 131       base::OnStringConversionError::SKIP, &encoded_terms))
 132     return false;
 133   *escaped_terms = base::UTF8ToUTF16(is_in_query ?
 134       net::EscapeQueryParamValue(encoded_terms, true) :
 135       net::EscapePath(encoded_terms));
 136   if (original_query.empty())
 137     return true;
 138   std::string encoded_original_query;
 139   if (!base::UTF16ToCodepage(original_query, encoding,
 140       base::OnStringConversionError::SKIP, &encoded_original_query))
 141     return false;
 142   *escaped_original_query = base::UTF8ToUTF16(
 143       net::EscapeQueryParamValue(encoded_original_query, true));
 144   return true;
 145 }
 146
 147 // Extract query key and host given a list of parameters coming from the URL
 148 // query or ref.
 149 std::string FindSearchTermsKey(const std::string& params) {
 150   if (params.empty())
 151     return std::string();
 152   url::Component query, key, value;
 153   query.len = static_cast<int>(params.size());
 154   while (url::ExtractQueryKeyValue(params.c_str(), &query, &key, &value)) {
 155     if (key.is_nonempty() && value.is_nonempty()) {
 156       std::string value_string = params.substr(value.begin, value.len);
 157       if (value_string.find(kSearchTermsParameterFull, 0) !=
 158           std::string::npos ||
 159           value_string.find(kGoogleUnescapedSearchTermsParameterFull, 0) !=
 160           std::string::npos) {
 161         return params.substr(key.begin, key.len);
 162       }
 163     }
 164   }
 165   return std::string();
 166 }
 167
 168 bool IsTemplateParameterString(const std::string& param) {
 169   return (param.length() > 2) && (*(param.begin()) == kStartParameter) &&
 170       (*(param.rbegin()) == kEndParameter);
 171 }
 172
 173 }  // namespace
 174
 175
 176 // TemplateURLRef::SearchTermsArgs --------------------------------------------
 177
 178 TemplateURLRef::SearchTermsArgs::SearchTermsArgs(
 179     const base::string16& search_terms)
 180     : search_terms(search_terms),
 181       input_type(metrics::OmniboxInputType::INVALID),
 182       accepted_suggestion(NO_SUGGESTIONS_AVAILABLE),
 183       cursor_position(base::string16::npos),
 184       omnibox_start_margin(-1),
 185       page_classification(metrics::OmniboxEventProto::INVALID_SPEC),
 186       bookmark_bar_pinned(false),
 187       append_extra_query_params(false),
 188       force_instant_results(false),
 189       from_app_list(false),
 190       contextual_search_params(ContextualSearchParams()) {
 191 }
 192
 193 TemplateURLRef::SearchTermsArgs::~SearchTermsArgs() {
 194 }
 195
 196 TemplateURLRef::SearchTermsArgs::ContextualSearchParams::
 197     ContextualSearchParams()
 198     : version(-1),
 199       start(base::string16::npos),
 200       end(base::string16::npos) {
 201 }
 202
 203 TemplateURLRef::SearchTermsArgs::ContextualSearchParams::
 204     ContextualSearchParams(
 205         const int version,
 206         const size_t start,
 207         const size_t end,
 208         const std::string& selection,
 209         const std::string& content,
 210         const std::string& base_page_url,
 211         const std::string& encoding)
 212     : version(version),
 213       start(start),
 214       end(end),
 215       selection(selection),
 216       content(content),
 217       base_page_url(base_page_url),
 218       encoding(encoding) {
 219 }
 220
 221 TemplateURLRef::SearchTermsArgs::ContextualSearchParams::
 222     ~ContextualSearchParams() {
 223 }
 224
 225 // TemplateURLRef -------------------------------------------------------------
 226
 227 TemplateURLRef::TemplateURLRef(TemplateURL* owner, Type type)
 228     : owner_(owner),
 229       type_(type),
 230       index_in_owner_(-1),
 231       parsed_(false),
 232       valid_(false),
 233       supports_replacements_(false),
 234       search_term_key_location_(url::Parsed::QUERY),
 235       prepopulated_(false) {
 236   DCHECK(owner_);
 237   DCHECK_NE(INDEXED, type_);
 238 }
 239
 240 TemplateURLRef::TemplateURLRef(TemplateURL* owner, size_t index_in_owner)
 241     : owner_(owner),
 242       type_(INDEXED),
 243       index_in_owner_(index_in_owner),
 244       parsed_(false),
 245       valid_(false),
 246       supports_replacements_(false),
 247       search_term_key_location_(url::Parsed::QUERY),
 248       prepopulated_(false) {
 249   DCHECK(owner_);
 250   DCHECK_LT(index_in_owner_, owner_->URLCount());
 251 }
 252
 253 TemplateURLRef::~TemplateURLRef() {
 254 }
 255
 256 std::string TemplateURLRef::GetURL() const {
 257   switch (type_) {
 258     case SEARCH:            return owner_->url();
 259     case SUGGEST:           return owner_->suggestions_url();
 260     case INSTANT:           return owner_->instant_url();
 261     case IMAGE:             return owner_->image_url();
 262     case NEW_TAB:           return owner_->new_tab_url();
 263     case CONTEXTUAL_SEARCH: return owner_->contextual_search_url();
 264     case INDEXED:           return owner_->GetURL(index_in_owner_);
 265     default:       NOTREACHED(); return std::string();  // NOLINT
 266   }
 267 }
 268
 269 std::string TemplateURLRef::GetPostParamsString() const {
 270   switch (type_) {
 271     case INDEXED:
 272     case SEARCH:            return owner_->search_url_post_params();
 273     case SUGGEST:           return owner_->suggestions_url_post_params();
 274     case INSTANT:           return owner_->instant_url_post_params();
 275     case NEW_TAB:           return std::string();
 276     case CONTEXTUAL_SEARCH: return std::string();
 277     case IMAGE:             return owner_->image_url_post_params();
 278     default:      NOTREACHED(); return std::string();  // NOLINT
 279   }
 280 }
 281
 282 bool TemplateURLRef::UsesPOSTMethod(
 283     const SearchTermsData& search_terms_data) const {
 284   ParseIfNecessary(search_terms_data);
 285   return !post_params_.empty();
 286 }
 287
 288 bool TemplateURLRef::EncodeFormData(const PostParams& post_params,
 289                                     PostContent* post_content) const {
 290   if (post_params.empty())
 291     return true;
 292   if (!post_content)
 293     return false;
 294
 295   const char kUploadDataMIMEType[] = "multipart/form-data; boundary=";
 296   const char kMultipartBoundary[] = "----+*+----%016" PRIx64 "----+*+----";
 297   // Each name/value pair is stored in a body part which is preceded by a
 298   // boundary delimiter line. Uses random number generator here to create
 299   // a unique boundary delimiter for form data encoding.
 300   std::string boundary = base::StringPrintf(kMultipartBoundary,
 301                                             base::RandUint64());
 302   // Sets the content MIME type.
 303   post_content->first = kUploadDataMIMEType;
 304   post_content->first += boundary;
 305   // Encodes the post parameters.
 306   std::string* post_data = &post_content->second;
 307   post_data->clear();
 308   for (PostParams::const_iterator param = post_params.begin();
 309        param != post_params.end(); ++param) {
 310     DCHECK(!param->first.empty());
 311     net::AddMultipartValueForUpload(param->first, param->second, boundary,
 312                                     std::string(), post_data);
 313   }
 314   net::AddMultipartFinalDelimiterForUpload(boundary, post_data);
 315   return true;
 316 }
 317
 318 bool TemplateURLRef::SupportsReplacement(
 319     const SearchTermsData& search_terms_data) const {
 320   ParseIfNecessary(search_terms_data);
 321   return valid_ && supports_replacements_;
 322 }
 323
 324 std::string TemplateURLRef::ReplaceSearchTerms(
 325     const SearchTermsArgs& search_terms_args,
 326     const SearchTermsData& search_terms_data,
 327     PostContent* post_content) const {
 328   ParseIfNecessary(search_terms_data);
 329   if (!valid_)
 330     return std::string();
 331
 332   std::string url(HandleReplacements(search_terms_args, search_terms_data,
 333                                      post_content));
 334
 335   GURL gurl(url);
 336   if (!gurl.is_valid())
 337     return url;
 338
 339   std::vector<std::string> query_params;
 340   if (search_terms_args.append_extra_query_params) {
 341     std::string extra_params(
 342         CommandLine::ForCurrentProcess()->GetSwitchValueASCII(
 343             switches::kExtraSearchQueryParams));
 344     if (!extra_params.empty())
 345       query_params.push_back(extra_params);
 346   }
 347   if (!search_terms_args.suggest_query_params.empty())
 348     query_params.push_back(search_terms_args.suggest_query_params);
 349   if (!gurl.query().empty())
 350     query_params.push_back(gurl.query());
 351
 352   if (query_params.empty())
 353     return url;
 354
 355   GURL::Replacements replacements;
 356   std::string query_str = JoinString(query_params, "&");
 357   replacements.SetQueryStr(query_str);
 358   return gurl.ReplaceComponents(replacements).possibly_invalid_spec();
 359 }
 360
 361 bool TemplateURLRef::IsValid(const SearchTermsData& search_terms_data) const {
 362   ParseIfNecessary(search_terms_data);
 363   return valid_;
 364 }
 365
 366 base::string16 TemplateURLRef::DisplayURL(
 367     const SearchTermsData& search_terms_data) const {
 368   ParseIfNecessary(search_terms_data);
 369   base::string16 result(base::UTF8ToUTF16(GetURL()));
 370   if (valid_ && !replacements_.empty()) {
 371     ReplaceSubstringsAfterOffset(&result, 0,
 372                                  base::ASCIIToUTF16(kSearchTermsParameterFull),
 373                                  base::ASCIIToUTF16(kDisplaySearchTerms));
 374     ReplaceSubstringsAfterOffset(&result, 0,
 375         base::ASCIIToUTF16(kGoogleUnescapedSearchTermsParameterFull),
 376         base::ASCIIToUTF16(kDisplayUnescapedSearchTerms));
 377   }
 378   return result;
 379 }
 380
 381 // static
 382 std::string TemplateURLRef::DisplayURLToURLRef(
 383     const base::string16& display_url) {
 384   base::string16 result = display_url;
 385   ReplaceSubstringsAfterOffset(&result, 0,
 386                                base::ASCIIToUTF16(kDisplaySearchTerms),
 387                                base::ASCIIToUTF16(kSearchTermsParameterFull));
 388   ReplaceSubstringsAfterOffset(
 389       &result, 0,
 390       base::ASCIIToUTF16(kDisplayUnescapedSearchTerms),
 391       base::ASCIIToUTF16(kGoogleUnescapedSearchTermsParameterFull));
 392   return base::UTF16ToUTF8(result);
 393 }
 394
 395 const std::string& TemplateURLRef::GetHost(
 396     const SearchTermsData& search_terms_data) const {
 397   ParseIfNecessary(search_terms_data);
 398   return host_;
 399 }
 400
 401 const std::string& TemplateURLRef::GetPath(
 402     const SearchTermsData& search_terms_data) const {
 403   ParseIfNecessary(search_terms_data);
 404   return path_;
 405 }
 406
 407 const std::string& TemplateURLRef::GetSearchTermKey(
 408     const SearchTermsData& search_terms_data) const {
 409   ParseIfNecessary(search_terms_data);
 410   return search_term_key_;
 411 }
 412
 413 base::string16 TemplateURLRef::SearchTermToString16(
 414     const std::string& term) const {
 415   const std::vector<std::string>& encodings = owner_->input_encodings();
 416   base::string16 result;
 417
 418   std::string unescaped = net::UnescapeURLComponent(
 419       term,
 420       net::UnescapeRule::REPLACE_PLUS_WITH_SPACE |
 421       net::UnescapeRule::URL_SPECIAL_CHARS);
 422   for (size_t i = 0; i < encodings.size(); ++i) {
 423     if (base::CodepageToUTF16(unescaped, encodings[i].c_str(),
 424                               base::OnStringConversionError::FAIL, &result))
 425       return result;
 426   }
 427
 428   // Always fall back on UTF-8 if it works.
 429   if (base::CodepageToUTF16(unescaped, base::kCodepageUTF8,
 430                             base::OnStringConversionError::FAIL, &result))
 431     return result;
 432
 433   // When nothing worked, just use the escaped text. We have no idea what the
 434   // encoding is. We need to substitute spaces for pluses ourselves since we're
 435   // not sending it through an unescaper.
 436   result = base::UTF8ToUTF16(term);
 437   std::replace(result.begin(), result.end(), '+', ' ');
 438   return result;
 439 }
 440
 441 bool TemplateURLRef::HasGoogleBaseURLs(
 442     const SearchTermsData& search_terms_data) const {
 443   ParseIfNecessary(search_terms_data);
 444   for (size_t i = 0; i < replacements_.size(); ++i) {
 445     if ((replacements_[i].type == GOOGLE_BASE_URL) ||
 446         (replacements_[i].type == GOOGLE_BASE_SUGGEST_URL))
 447       return true;
 448   }
 449   return false;
 450 }
 451
 452 bool TemplateURLRef::ExtractSearchTermsFromURL(
 453     const GURL& url,
 454     base::string16* search_terms,
 455     const SearchTermsData& search_terms_data,
 456     url::Parsed::ComponentType* search_terms_component,
 457     url::Component* search_terms_position) const {
 458   DCHECK(search_terms);
 459   search_terms->clear();
 460
 461   ParseIfNecessary(search_terms_data);
 462
 463   // We need a search term in the template URL to extract something.
 464   if (search_term_key_.empty())
 465     return false;
 466
 467   // TODO(beaudoin): Support patterns of the form http://foo/{searchTerms}/
 468   // See crbug.com/153798
 469
 470   // Fill-in the replacements. We don't care about search terms in the pattern,
 471   // so we use the empty string.
 472   // Currently we assume the search term only shows in URL, not in post params.
 473   GURL pattern(ReplaceSearchTerms(SearchTermsArgs(base::string16()),
 474                                   search_terms_data, NULL));
 475   // Host, path and port must match.
 476   if (url.port() != pattern.port() ||
 477       url.host() != host_ ||
 478       url.path() != path_) {
 479     return false;
 480   }
 481
 482   // Parameter must be present either in the query or the ref.
 483   const std::string& params(
 484       (search_term_key_location_ == url::Parsed::QUERY) ?
 485           url.query() : url.ref());
 486
 487   url::Component query, key, value;
 488   query.len = static_cast<int>(params.size());
 489   bool key_found = false;
 490   while (url::ExtractQueryKeyValue(params.c_str(), &query, &key, &value)) {
 491     if (key.is_nonempty()) {
 492       if (params.substr(key.begin, key.len) == search_term_key_) {
 493         // Fail if search term key is found twice.
 494         if (key_found) {
 495           search_terms->clear();
 496           return false;
 497         }
 498         key_found = true;
 499         // Extract the search term.
 500         *search_terms = net::UnescapeAndDecodeUTF8URLComponent(
 501             params.substr(value.begin, value.len),
 502             net::UnescapeRule::SPACES |
 503                 net::UnescapeRule::URL_SPECIAL_CHARS |
 504                 net::UnescapeRule::REPLACE_PLUS_WITH_SPACE);
 505         if (search_terms_component)
 506           *search_terms_component = search_term_key_location_;
 507         if (search_terms_position)
 508           *search_terms_position = value;
 509       }
 510     }
 511   }
 512   return key_found;
 513 }
 514
 515 void TemplateURLRef::InvalidateCachedValues() const {
 516   supports_replacements_ = valid_ = parsed_ = false;
 517   host_.clear();
 518   path_.clear();
 519   search_term_key_.clear();
 520   replacements_.clear();
 521   post_params_.clear();
 522 }
 523
 524 bool TemplateURLRef::ParseParameter(size_t start,
 525                                     size_t end,
 526                                     std::string* url,
 527                                     Replacements* replacements) const {
 528   DCHECK(start != std::string::npos &&
 529          end != std::string::npos && end > start);
 530   size_t length = end - start - 1;
 531   bool optional = false;
 532   if ((*url)[end - 1] == kOptional) {
 533     optional = true;
 534     length--;
 535   }
 536   std::string parameter(url->substr(start + 1, length));
 537   std::string full_parameter(url->substr(start, end - start + 1));
 538   // Remove the parameter from the string.  For parameters who replacement is
 539   // constant and already known, just replace them directly.  For other cases,
 540   // like parameters whose values may change over time, use |replacements|.
 541   url->erase(start, end - start + 1);
 542   if (parameter == kSearchTermsParameter) {
 543     replacements->push_back(Replacement(SEARCH_TERMS, start));
 544   } else if (parameter == kCountParameter) {
 545     if (!optional)
 546       url->insert(start, kDefaultCount);
 547   } else if (parameter == kGoogleAssistedQueryStatsParameter) {
 548     replacements->push_back(Replacement(GOOGLE_ASSISTED_QUERY_STATS, start));
 549   } else if (parameter == kGoogleBaseURLParameter) {
 550     replacements->push_back(Replacement(GOOGLE_BASE_URL, start));
 551   } else if (parameter == kGoogleBaseSuggestURLParameter) {
 552     replacements->push_back(Replacement(GOOGLE_BASE_SUGGEST_URL, start));
 553   } else if (parameter == kGoogleBookmarkBarPinnedParameter) {
 554     replacements->push_back(Replacement(GOOGLE_BOOKMARK_BAR_PINNED, start));
 555   } else if (parameter == kGoogleCurrentPageUrlParameter) {
 556     replacements->push_back(Replacement(GOOGLE_CURRENT_PAGE_URL, start));
 557   } else if (parameter == kGoogleCursorPositionParameter) {
 558     replacements->push_back(Replacement(GOOGLE_CURSOR_POSITION, start));
 559   } else if (parameter == kGoogleForceInstantResultsParameter) {
 560     replacements->push_back(Replacement(GOOGLE_FORCE_INSTANT_RESULTS, start));
 561   } else if (parameter == kGoogleImageOriginalHeight) {
 562     replacements->push_back(
 563         Replacement(TemplateURLRef::GOOGLE_IMAGE_ORIGINAL_HEIGHT, start));
 564   } else if (parameter == kGoogleImageOriginalWidth) {
 565     replacements->push_back(
 566         Replacement(TemplateURLRef::GOOGLE_IMAGE_ORIGINAL_WIDTH, start));
 567   } else if (parameter == kGoogleImageSearchSource) {
 568     replacements->push_back(
 569         Replacement(TemplateURLRef::GOOGLE_IMAGE_SEARCH_SOURCE, start));
 570   } else if (parameter == kGoogleImageThumbnailParameter) {
 571     replacements->push_back(
 572         Replacement(TemplateURLRef::GOOGLE_IMAGE_THUMBNAIL, start));
 573   } else if (parameter == kGoogleImageURLParameter) {
 574     replacements->push_back(Replacement(TemplateURLRef::GOOGLE_IMAGE_URL,
 575                                         start));
 576   } else if (parameter == kGoogleInputTypeParameter) {
 577     replacements->push_back(Replacement(TemplateURLRef::GOOGLE_INPUT_TYPE,
 578                                         start));
 579   } else if (parameter == kGoogleInstantExtendedEnabledParameter) {
 580     replacements->push_back(Replacement(GOOGLE_INSTANT_EXTENDED_ENABLED,
 581                                         start));
 582   } else if (parameter == kGoogleInstantExtendedEnabledKey) {
 583     url->insert(start, google_util::kInstantExtendedAPIParam);
 584   } else if (parameter == kGoogleNTPIsThemedParameter) {
 585     replacements->push_back(Replacement(GOOGLE_NTP_IS_THEMED, start));
 586   } else if (parameter == kGoogleOmniboxStartMarginParameter) {
 587     replacements->push_back(Replacement(GOOGLE_OMNIBOX_START_MARGIN, start));
 588   } else if (parameter == kGoogleContextualSearchVersion) {
 589     replacements->push_back(
 590         Replacement(GOOGLE_CONTEXTUAL_SEARCH_VERSION, start));
 591   } else if (parameter == kGoogleContextualSearchContextData) {
 592     replacements->push_back(
 593         Replacement(GOOGLE_CONTEXTUAL_SEARCH_CONTEXT_DATA, start));
 594   } else if (parameter == kGoogleOriginalQueryForSuggestionParameter) {
 595     replacements->push_back(Replacement(GOOGLE_ORIGINAL_QUERY_FOR_SUGGESTION,
 596                                         start));
 597   } else if (parameter == kGooglePageClassificationParameter) {
 598     replacements->push_back(Replacement(GOOGLE_PAGE_CLASSIFICATION, start));
 599   } else if (parameter == kGooglePrefetchQuery) {
 600     replacements->push_back(Replacement(GOOGLE_PREFETCH_QUERY, start));
 601   } else if (parameter == kGoogleRLZParameter) {
 602     replacements->push_back(Replacement(GOOGLE_RLZ, start));
 603   } else if (parameter == kGoogleSearchClient) {
 604     replacements->push_back(Replacement(GOOGLE_SEARCH_CLIENT, start));
 605   } else if (parameter == kGoogleSearchFieldtrialParameter) {
 606     replacements->push_back(Replacement(GOOGLE_SEARCH_FIELDTRIAL_GROUP, start));
 607   } else if (parameter == kGoogleSearchVersion) {
 608     replacements->push_back(Replacement(GOOGLE_SEARCH_VERSION, start));
 609   } else if (parameter == kGoogleSessionToken) {
 610     replacements->push_back(Replacement(GOOGLE_SESSION_TOKEN, start));
 611   } else if (parameter == kGoogleSourceIdParameter) {
 612 #if defined(OS_ANDROID)
 613     url->insert(start, "sourceid=chrome-mobile&");
 614 #else
 615     url->insert(start, "sourceid=chrome&");
 616 #endif
 617   } else if (parameter == kGoogleSuggestAPIKeyParameter) {
 618     url->insert(start,
 619                 net::EscapeQueryParamValue(google_apis::GetAPIKey(), false));
 620   } else if (parameter == kGoogleSuggestClient) {
 621     replacements->push_back(Replacement(GOOGLE_SUGGEST_CLIENT, start));
 622   } else if (parameter == kGoogleSuggestRequestId) {
 623     replacements->push_back(Replacement(GOOGLE_SUGGEST_REQUEST_ID, start));
 624   } else if (parameter == kGoogleUnescapedSearchTermsParameter) {
 625     replacements->push_back(Replacement(GOOGLE_UNESCAPED_SEARCH_TERMS, start));
 626   } else if (parameter == kInputEncodingParameter) {
 627     replacements->push_back(Replacement(ENCODING, start));
 628   } else if (parameter == kLanguageParameter) {
 629     replacements->push_back(Replacement(LANGUAGE, start));
 630   } else if (parameter == kOutputEncodingParameter) {
 631     if (!optional)
 632       url->insert(start, kOutputEncodingType);
 633   } else if ((parameter == kStartIndexParameter) ||
 634              (parameter == kStartPageParameter)) {
 635     // We don't support these.
 636     if (!optional)
 637       url->insert(start, "1");
 638   } else if (!prepopulated_) {
 639     // If it's a prepopulated URL, we know that it's safe to remove unknown
 640     // parameters, so just ignore this and return true below. Otherwise it could
 641     // be some garbage but can also be a javascript block. Put it back.
 642     url->insert(start, full_parameter);
 643     return false;
 644   }
 645   return true;
 646 }
 647
 648 std::string TemplateURLRef::ParseURL(const std::string& url,
 649                                      Replacements* replacements,
 650                                      PostParams* post_params,
 651                                      bool* valid) const {
 652   *valid = false;
 653   std::string parsed_url = url;
 654   for (size_t last = 0; last != std::string::npos; ) {
 655     last = parsed_url.find(kStartParameter, last);
 656     if (last != std::string::npos) {
 657       size_t template_end = parsed_url.find(kEndParameter, last);
 658       if (template_end != std::string::npos) {
 659         // Since we allow Javascript in the URL, {} pairs could be nested. Match
 660         // only leaf pairs with supported parameters.
 661         size_t next_template_start = parsed_url.find(kStartParameter, last + 1);
 662         if (next_template_start == std::string::npos ||
 663             next_template_start > template_end) {
 664           // If successful, ParseParameter erases from the string as such no
 665           // need to update |last|. If failed, move |last| to the end of pair.
 666           if (!ParseParameter(last, template_end, &parsed_url, replacements)) {
 667             // |template_end| + 1 may be beyond the end of the string.
 668             last = template_end;
 669           }
 670         } else {
 671           last = next_template_start;
 672         }
 673       } else {
 674         // Open brace without a closing brace, return.
 675         return std::string();
 676       }
 677     }
 678   }
 679
 680   // Handles the post parameters.
 681   const std::string& post_params_string = GetPostParamsString();
 682   if (!post_params_string.empty()) {
 683     typedef std::vector<std::string> Strings;
 684     Strings param_list;
 685     base::SplitString(post_params_string, ',', &param_list);
 686
 687     for (Strings::const_iterator iterator = param_list.begin();
 688          iterator != param_list.end(); ++iterator) {
 689       Strings parts;
 690       // The '=' delimiter is required and the name must be not empty.
 691       base::SplitString(*iterator, '=', &parts);
 692       if ((parts.size() != 2U) || parts[0].empty())
 693         return std::string();
 694
 695       std::string& value = parts[1];
 696       size_t replacements_size = replacements->size();
 697       if (IsTemplateParameterString(value))
 698         ParseParameter(0, value.length() - 1, &value, replacements);
 699       post_params->push_back(std::make_pair(parts[0], value));
 700       // If there was a replacement added, points its index to last added
 701       // PostParam.
 702       if (replacements->size() > replacements_size) {
 703         DCHECK_EQ(replacements_size + 1, replacements->size());
 704         Replacement* r = &replacements->back();
 705         r->is_post_param = true;
 706         r->index = post_params->size() - 1;
 707       }
 708     }
 709     DCHECK(!post_params->empty());
 710   }
 711
 712   *valid = true;
 713   return parsed_url;
 714 }
 715
 716 void TemplateURLRef::ParseIfNecessary(
 717     const SearchTermsData& search_terms_data) const {
 718   if (!parsed_) {
 719     InvalidateCachedValues();
 720     parsed_ = true;
 721     parsed_url_ = ParseURL(GetURL(), &replacements_, &post_params_, &valid_);
 722     supports_replacements_ = false;
 723     if (valid_) {
 724       bool has_only_one_search_term = false;
 725       for (Replacements::const_iterator i = replacements_.begin();
 726            i != replacements_.end(); ++i) {
 727         if ((i->type == SEARCH_TERMS) ||
 728             (i->type == GOOGLE_UNESCAPED_SEARCH_TERMS)) {
 729           if (has_only_one_search_term) {
 730             has_only_one_search_term = false;
 731             break;
 732           }
 733           has_only_one_search_term = true;
 734           supports_replacements_ = true;
 735         }
 736       }
 737       // Only parse the host/key if there is one search term. Technically there
 738       // could be more than one term, but it's uncommon; so we punt.
 739       if (has_only_one_search_term)
 740         ParseHostAndSearchTermKey(search_terms_data);
 741     }
 742   }
 743 }
 744
 745 void TemplateURLRef::ParseHostAndSearchTermKey(
 746     const SearchTermsData& search_terms_data) const {
 747   std::string url_string(GetURL());
 748   ReplaceSubstringsAfterOffset(&url_string, 0,
 749                                kGoogleBaseURLParameterFull,
 750                                search_terms_data.GoogleBaseURLValue());
 751   ReplaceSubstringsAfterOffset(&url_string, 0,
 752                                kGoogleBaseSuggestURLParameterFull,
 753                                search_terms_data.GoogleBaseSuggestURLValue());
 754
 755   search_term_key_.clear();
 756   host_.clear();
 757   path_.clear();
 758   search_term_key_location_ = url::Parsed::REF;
 759
 760   GURL url(url_string);
 761   if (!url.is_valid())
 762     return;
 763
 764   std::string query_key = FindSearchTermsKey(url.query());
 765   std::string ref_key = FindSearchTermsKey(url.ref());
 766   if (query_key.empty() == ref_key.empty())
 767     return;  // No key or multiple keys found.  We only handle having one key.
 768   search_term_key_ = query_key.empty() ? ref_key : query_key;
 769   search_term_key_location_ =
 770       query_key.empty() ? url::Parsed::REF : url::Parsed::QUERY;
 771   host_ = url.host();
 772   path_ = url.path();
 773 }
 774
 775 void TemplateURLRef::HandleReplacement(const std::string& name,
 776                                        const std::string& value,
 777                                        const Replacement& replacement,
 778                                        std::string* url) const {
 779   size_t pos = replacement.index;
 780   if (replacement.is_post_param) {
 781     DCHECK_LT(pos, post_params_.size());
 782     DCHECK(!post_params_[pos].first.empty());
 783     post_params_[pos].second = value;
 784   } else {
 785     url->insert(pos, name.empty() ? value : (name + "=" + value + "&"));
 786   }
 787 }
 788
 789 std::string TemplateURLRef::HandleReplacements(
 790     const SearchTermsArgs& search_terms_args,
 791     const SearchTermsData& search_terms_data,
 792     PostContent* post_content) const {
 793   if (replacements_.empty()) {
 794     if (!post_params_.empty())
 795       EncodeFormData(post_params_, post_content);
 796     return parsed_url_;
 797   }
 798
 799   // Determine if the search terms are in the query or before. We're escaping
 800   // space as '+' in the former case and as '%20' in the latter case.
 801   bool is_in_query = true;
 802   for (Replacements::iterator i = replacements_.begin();
 803        i != replacements_.end(); ++i) {
 804     if (i->type == SEARCH_TERMS) {
 805       base::string16::size_type query_start = parsed_url_.find('?');
 806       is_in_query = query_start != base::string16::npos &&
 807           (static_cast<base::string16::size_type>(i->index) > query_start);
 808       break;
 809     }
 810   }
 811
 812   std::string input_encoding;
 813   base::string16 encoded_terms;
 814   base::string16 encoded_original_query;
 815   owner_->EncodeSearchTerms(search_terms_args, is_in_query, &input_encoding,
 816                             &encoded_terms, &encoded_original_query);
 817
 818   std::string url = parsed_url_;
 819
 820   // replacements_ is ordered in ascending order, as such we need to iterate
 821   // from the back.
 822   for (Replacements::reverse_iterator i = replacements_.rbegin();
 823        i != replacements_.rend(); ++i) {
 824     switch (i->type) {
 825       case ENCODING:
 826         HandleReplacement(std::string(), input_encoding, *i, &url);
 827         break;
 828
 829       case GOOGLE_ASSISTED_QUERY_STATS:
 830         DCHECK(!i->is_post_param);
 831         if (!search_terms_args.assisted_query_stats.empty()) {
 832           // Get the base URL without substituting AQS to avoid infinite
 833           // recursion.  We need the URL to find out if it meets all
 834           // AQS requirements (e.g. HTTPS protocol check).
 835           // See TemplateURLRef::SearchTermsArgs for more details.
 836           SearchTermsArgs search_terms_args_without_aqs(search_terms_args);
 837           search_terms_args_without_aqs.assisted_query_stats.clear();
 838           GURL base_url(ReplaceSearchTerms(
 839               search_terms_args_without_aqs, search_terms_data, NULL));
 840           if (base_url.SchemeIs(url::kHttpsScheme)) {
 841             HandleReplacement(
 842                 "aqs", search_terms_args.assisted_query_stats, *i, &url);
 843           }
 844         }
 845         break;
 846
 847       case GOOGLE_BASE_URL:
 848         DCHECK(!i->is_post_param);
 849         HandleReplacement(
 850             std::string(), search_terms_data.GoogleBaseURLValue(), *i, &url);
 851         break;
 852
 853       case GOOGLE_BASE_SUGGEST_URL:
 854         DCHECK(!i->is_post_param);
 855         HandleReplacement(
 856             std::string(), search_terms_data.GoogleBaseSuggestURLValue(), *i,
 857             &url);
 858         break;
 859
 860       case GOOGLE_BOOKMARK_BAR_PINNED:
 861         if (search_terms_data.IsShowingSearchTermsOnSearchResultsPages()) {
 862           // Log whether the bookmark bar is pinned when the user is seeing
 863           // InstantExtended on the SRP.
 864           DCHECK(!i->is_post_param);
 865           HandleReplacement(
 866               "bmbp", search_terms_args.bookmark_bar_pinned ? "1" : "0", *i,
 867               &url);
 868         }
 869         break;
 870
 871       case GOOGLE_CURRENT_PAGE_URL:
 872         DCHECK(!i->is_post_param);
 873         if (!search_terms_args.current_page_url.empty()) {
 874           const std::string& escaped_current_page_url =
 875               net::EscapeQueryParamValue(search_terms_args.current_page_url,
 876                                          true);
 877           HandleReplacement("url", escaped_current_page_url, *i, &url);
 878         }
 879         break;
 880
 881       case GOOGLE_CURSOR_POSITION:
 882         DCHECK(!i->is_post_param);
 883         if (search_terms_args.cursor_position != base::string16::npos)
 884           HandleReplacement(
 885               "cp",
 886               base::StringPrintf("%" PRIuS, search_terms_args.cursor_position),
 887               *i,
 888               &url);
 889         break;
 890
 891       case GOOGLE_FORCE_INSTANT_RESULTS:
 892         DCHECK(!i->is_post_param);
 893         HandleReplacement(std::string(),
 894                           search_terms_data.ForceInstantResultsParam(
 895                               search_terms_args.force_instant_results),
 896                           *i,
 897                           &url);
 898         break;
 899
 900       case GOOGLE_INPUT_TYPE:
 901         DCHECK(!i->is_post_param);
 902         HandleReplacement(
 903             "oit", base::IntToString(search_terms_args.input_type), *i, &url);
 904         break;
 905
 906       case GOOGLE_INSTANT_EXTENDED_ENABLED:
 907         DCHECK(!i->is_post_param);
 908         HandleReplacement(std::string(),
 909                           search_terms_data.InstantExtendedEnabledParam(
 910                               type_ == SEARCH),
 911                           *i,
 912                           &url);
 913         break;
 914
 915       case GOOGLE_NTP_IS_THEMED:
 916         DCHECK(!i->is_post_param);
 917         HandleReplacement(
 918             std::string(), search_terms_data.NTPIsThemedParam(), *i, &url);
 919         break;
 920
 921       case GOOGLE_OMNIBOX_START_MARGIN:
 922         DCHECK(!i->is_post_param);
 923         if (search_terms_args.omnibox_start_margin >= 0) {
 924           HandleReplacement(
 925               "es_sm",
 926               base::IntToString(search_terms_args.omnibox_start_margin),
 927               *i,
 928               &url);
 929         }
 930         break;
 931
 932       case GOOGLE_CONTEXTUAL_SEARCH_VERSION:
 933         if (search_terms_args.contextual_search_params.version >= 0) {
 934           HandleReplacement(
 935               "ctxs",
 936               base::IntToString(
 937                   search_terms_args.contextual_search_params.version),
 938               *i,
 939               &url);
 940         }
 941         break;
 942
 943       case GOOGLE_CONTEXTUAL_SEARCH_CONTEXT_DATA: {
 944         DCHECK(!i->is_post_param);
 945         std::string context_data;
 946
 947         const SearchTermsArgs::ContextualSearchParams& params =
 948             search_terms_args.contextual_search_params;
 949
 950         if (params.start != std::string::npos) {
 951           context_data.append("ctxs_start=" + base::IntToString(
 952               params.start) + "&");
 953         }
 954
 955         if (params.end != std::string::npos) {
 956           context_data.append("ctxs_end=" + base::IntToString(
 957               params.end) + "&");
 958         }
 959
 960         if (!params.selection.empty())
 961           context_data.append("q=" + params.selection + "&");
 962
 963         if (!params.content.empty())
 964           context_data.append("ctxs_content=" + params.content + "&");
 965
 966         if (!params.base_page_url.empty())
 967           context_data.append("ctxs_url=" + params.base_page_url + "&");
 968
 969         if (!params.encoding.empty()) {
 970           context_data.append("ctxs_encoding=" + params.encoding + "&");
 971         }
 972
 973         HandleReplacement(std::string(), context_data, *i, &url);
 974         break;
 975       }
 976
 977       case GOOGLE_ORIGINAL_QUERY_FOR_SUGGESTION:
 978         DCHECK(!i->is_post_param);
 979         if (search_terms_args.accepted_suggestion >= 0 ||
 980             !search_terms_args.assisted_query_stats.empty()) {
 981           HandleReplacement(
 982               "oq", base::UTF16ToUTF8(encoded_original_query), *i, &url);
 983         }
 984         break;
 985
 986       case GOOGLE_PAGE_CLASSIFICATION:
 987         if (search_terms_args.page_classification !=
 988             metrics::OmniboxEventProto::INVALID_SPEC) {
 989           HandleReplacement(
 990               "pgcl", base::IntToString(search_terms_args.page_classification),
 991               *i, &url);
 992         }
 993         break;
 994
 995       case GOOGLE_PREFETCH_QUERY: {
 996         const std::string& query = search_terms_args.prefetch_query;
 997         const std::string& type = search_terms_args.prefetch_query_type;
 998         if (!query.empty() && !type.empty()) {
 999           HandleReplacement(
1000               std::string(), "pfq=" + query + "&qha=" + type + "&", *i, &url);
1001         }
1002         break;
1003       }
1004
1005       case GOOGLE_RLZ: {
1006         DCHECK(!i->is_post_param);
1007         // On platforms that don't have RLZ, we still want this branch
1008         // to happen so that we replace the RLZ template with the
1009         // empty string.  (If we don't handle this case, we hit a
1010         // NOTREACHED below.)
1011         base::string16 rlz_string = search_terms_data.GetRlzParameterValue(
1012             search_terms_args.from_app_list);
1013         if (!rlz_string.empty()) {
1014           HandleReplacement("rlz", base::UTF16ToUTF8(rlz_string), *i, &url);
1015         }
1016         break;
1017       }
1018
1019       case GOOGLE_SEARCH_CLIENT: {
1020         DCHECK(!i->is_post_param);
1021         std::string client = search_terms_data.GetSearchClient();
1022         if (!client.empty())
1023           HandleReplacement("client", client, *i, &url);
1024         break;
1025       }
1026
1027       case GOOGLE_SEARCH_FIELDTRIAL_GROUP:
1028         // We are not currently running any fieldtrials that modulate the search
1029         // url.  If we do, then we'd have some conditional insert such as:
1030         // url.insert(i->index, used_www ? "gcx=w&" : "gcx=c&");
1031         break;
1032
1033       case GOOGLE_SEARCH_VERSION:
1034         if (search_terms_data.EnableAnswersInSuggest())
1035           HandleReplacement("gs_rn", "42", *i, &url);
1036         break;
1037
1038       case GOOGLE_SESSION_TOKEN: {
1039         std::string token = search_terms_args.session_token;
1040         if (!token.empty())
1041           HandleReplacement("psi", token, *i, &url);
1042         break;
1043       }
1044
1045       case GOOGLE_SUGGEST_CLIENT:
1046         HandleReplacement(
1047             std::string(), search_terms_data.GetSuggestClient(), *i, &url);
1048         break;
1049
1050       case GOOGLE_SUGGEST_REQUEST_ID:
1051         HandleReplacement(
1052             std::string(), search_terms_data.GetSuggestRequestIdentifier(), *i,
1053             &url);
1054         break;
1055
1056       case GOOGLE_UNESCAPED_SEARCH_TERMS: {
1057         std::string unescaped_terms;
1058         base::UTF16ToCodepage(search_terms_args.search_terms,
1059                               input_encoding.c_str(),
1060                               base::OnStringConversionError::SKIP,
1061                               &unescaped_terms);
1062         HandleReplacement(std::string(), unescaped_terms, *i, &url);
1063         break;
1064       }
1065
1066       case LANGUAGE:
1067         HandleReplacement(
1068             std::string(), search_terms_data.GetApplicationLocale(), *i, &url);
1069         break;
1070
1071       case SEARCH_TERMS:
1072         HandleReplacement(
1073             std::string(), base::UTF16ToUTF8(encoded_terms), *i, &url);
1074         break;
1075
1076       case GOOGLE_IMAGE_THUMBNAIL:
1077         HandleReplacement(
1078             std::string(), search_terms_args.image_thumbnail_content, *i, &url);
1079         break;
1080
1081       case GOOGLE_IMAGE_URL:
1082         if (search_terms_args.image_url.is_valid()) {
1083           HandleReplacement(
1084               std::string(), search_terms_args.image_url.spec(), *i, &url);
1085         }
1086         break;
1087
1088       case GOOGLE_IMAGE_ORIGINAL_WIDTH:
1089         if (!search_terms_args.image_original_size.IsEmpty()) {
1090           HandleReplacement(
1091               std::string(),
1092               base::IntToString(search_terms_args.image_original_size.width()),
1093               *i, &url);
1094         }
1095         break;
1096
1097       case GOOGLE_IMAGE_ORIGINAL_HEIGHT:
1098         if (!search_terms_args.image_original_size.IsEmpty()) {
1099           HandleReplacement(
1100               std::string(),
1101               base::IntToString(search_terms_args.image_original_size.height()),
1102               *i, &url);
1103         }
1104         break;
1105
1106       case GOOGLE_IMAGE_SEARCH_SOURCE:
1107         HandleReplacement(
1108             std::string(), search_terms_data.GoogleImageSearchSource(), *i,
1109             &url);
1110         break;
1111
1112       default:
1113         NOTREACHED();
1114         break;
1115     }
1116   }
1117
1118   if (!post_params_.empty())
1119     EncodeFormData(post_params_, post_content);
1120
1121   return url;
1122 }
1123
1124
1125 // TemplateURL ----------------------------------------------------------------
1126
1127 TemplateURL::AssociatedExtensionInfo::AssociatedExtensionInfo(
1128     Type type,
1129     const std::string& extension_id)
1130     : type(type),
1131       extension_id(extension_id),
1132       wants_to_be_default_engine(false) {
1133   DCHECK_NE(NORMAL, type);
1134 }
1135
1136 TemplateURL::AssociatedExtensionInfo::~AssociatedExtensionInfo() {
1137 }
1138
1139 TemplateURL::TemplateURL(const TemplateURLData& data)
1140     : data_(data),
1141       url_ref_(this, TemplateURLRef::SEARCH),
1142       suggestions_url_ref_(this,
1143                            TemplateURLRef::SUGGEST),
1144       instant_url_ref_(this,
1145                        TemplateURLRef::INSTANT),
1146       image_url_ref_(this, TemplateURLRef::IMAGE),
1147       new_tab_url_ref_(this, TemplateURLRef::NEW_TAB),
1148       contextual_search_url_ref_(this, TemplateURLRef::CONTEXTUAL_SEARCH) {
1149   SetPrepopulateId(data_.prepopulate_id);
1150
1151   if (data_.search_terms_replacement_key ==
1152       kGoogleInstantExtendedEnabledKeyFull) {
1153     data_.search_terms_replacement_key = google_util::kInstantExtendedAPIParam;
1154   }
1155 }
1156
1157 TemplateURL::~TemplateURL() {
1158 }
1159
1160 // static
1161 base::string16 TemplateURL::GenerateKeyword(const GURL& url) {
1162   DCHECK(url.is_valid());
1163   // Strip "www." off the front of the keyword; otherwise the keyword won't work
1164   // properly.  See http://code.google.com/p/chromium/issues/detail?id=6984 .
1165   // Special case: if the host was exactly "www." (not sure this can happen but
1166   // perhaps with some weird intranet and custom DNS server?), ensure we at
1167   // least don't return the empty string.
1168   base::string16 keyword(net::StripWWWFromHost(url));
1169   return keyword.empty() ? base::ASCIIToUTF16("www") : keyword;
1170 }
1171
1172 // static
1173 GURL TemplateURL::GenerateFaviconURL(const GURL& url) {
1174   DCHECK(url.is_valid());
1175   GURL::Replacements rep;
1176
1177   const char favicon_path[] = "/favicon.ico";
1178   int favicon_path_len = arraysize(favicon_path) - 1;
1179
1180   rep.SetPath(favicon_path, url::Component(0, favicon_path_len));
1181   rep.ClearUsername();
1182   rep.ClearPassword();
1183   rep.ClearQuery();
1184   rep.ClearRef();
1185   return url.ReplaceComponents(rep);
1186 }
1187
1188 // static
1189 bool TemplateURL::MatchesData(const TemplateURL* t_url,
1190                               const TemplateURLData* data,
1191                               const SearchTermsData& search_terms_data) {
1192   if (!t_url || !data)
1193     return !t_url && !data;
1194
1195   return (t_url->short_name() == data->short_name) &&
1196       t_url->HasSameKeywordAs(*data, search_terms_data) &&
1197       (t_url->url() == data->url()) &&
1198       (t_url->suggestions_url() == data->suggestions_url) &&
1199       (t_url->instant_url() == data->instant_url) &&
1200       (t_url->image_url() == data->image_url) &&
1201       (t_url->new_tab_url() == data->new_tab_url) &&
1202       (t_url->search_url_post_params() == data->search_url_post_params) &&
1203       (t_url->suggestions_url_post_params() ==
1204           data->suggestions_url_post_params) &&
1205       (t_url->instant_url_post_params() == data->instant_url_post_params) &&
1206       (t_url->image_url_post_params() == data->image_url_post_params) &&
1207       (t_url->favicon_url() == data->favicon_url) &&
1208       (t_url->safe_for_autoreplace() == data->safe_for_autoreplace) &&
1209       (t_url->show_in_default_list() == data->show_in_default_list) &&
1210       (t_url->input_encodings() == data->input_encodings) &&
1211       (t_url->alternate_urls() == data->alternate_urls) &&
1212       (t_url->search_terms_replacement_key() ==
1213           data->search_terms_replacement_key);
1214 }
1215
1216 base::string16 TemplateURL::AdjustedShortNameForLocaleDirection() const {
1217   base::string16 bidi_safe_short_name = data_.short_name;
1218   base::i18n::AdjustStringForLocaleDirection(&bidi_safe_short_name);
1219   return bidi_safe_short_name;
1220 }
1221
1222 bool TemplateURL::ShowInDefaultList(
1223     const SearchTermsData& search_terms_data) const {
1224   return data_.show_in_default_list &&
1225       url_ref_.SupportsReplacement(search_terms_data);
1226 }
1227
1228 bool TemplateURL::SupportsReplacement(
1229     const SearchTermsData& search_terms_data) const {
1230   return url_ref_.SupportsReplacement(search_terms_data);
1231 }
1232
1233 bool TemplateURL::HasGoogleBaseURLs(
1234     const SearchTermsData& search_terms_data) const {
1235   return url_ref_.HasGoogleBaseURLs(search_terms_data) ||
1236       suggestions_url_ref_.HasGoogleBaseURLs(search_terms_data) ||
1237       instant_url_ref_.HasGoogleBaseURLs(search_terms_data) ||
1238       image_url_ref_.HasGoogleBaseURLs(search_terms_data) ||
1239       new_tab_url_ref_.HasGoogleBaseURLs(search_terms_data);
1240 }
1241
1242 bool TemplateURL::IsGoogleSearchURLWithReplaceableKeyword(
1243     const SearchTermsData& search_terms_data) const {
1244   return (GetType() == NORMAL) &&
1245       url_ref_.HasGoogleBaseURLs(search_terms_data) &&
1246       google_util::IsGoogleHostname(base::UTF16ToUTF8(data_.keyword()),
1247                                     google_util::DISALLOW_SUBDOMAIN);
1248 }
1249
1250 bool TemplateURL::HasSameKeywordAs(
1251     const TemplateURLData& other,
1252     const SearchTermsData& search_terms_data) const {
1253   return (data_.keyword() == other.keyword()) ||
1254       (IsGoogleSearchURLWithReplaceableKeyword(search_terms_data) &&
1255        TemplateURL(other).IsGoogleSearchURLWithReplaceableKeyword(
1256            search_terms_data));
1257 }
1258
1259 TemplateURL::Type TemplateURL::GetType() const {
1260   return extension_info_ ? extension_info_->type : NORMAL;
1261 }
1262
1263 std::string TemplateURL::GetExtensionId() const {
1264   DCHECK(extension_info_);
1265   return extension_info_->extension_id;
1266 }
1267
1268 size_t TemplateURL::URLCount() const {
1269   // Add 1 for the regular search URL.
1270   return data_.alternate_urls.size() + 1;
1271 }
1272
1273 const std::string& TemplateURL::GetURL(size_t index) const {
1274   DCHECK_LT(index, URLCount());
1275
1276   return (index < data_.alternate_urls.size()) ?
1277       data_.alternate_urls[index] : url();
1278 }
1279
1280 bool TemplateURL::ExtractSearchTermsFromURL(
1281     const GURL& url,
1282     const SearchTermsData& search_terms_data,
1283     base::string16* search_terms) {
1284   return FindSearchTermsInURL(url, search_terms_data, search_terms, NULL, NULL);
1285 }
1286
1287 bool TemplateURL::IsSearchURL(
1288     const GURL& url,
1289     const SearchTermsData& search_terms_data) {
1290   base::string16 search_terms;
1291   return ExtractSearchTermsFromURL(url, search_terms_data, &search_terms) &&
1292       !search_terms.empty();
1293 }
1294
1295 bool TemplateURL::HasSearchTermsReplacementKey(const GURL& url) const {
1296   // Look for the key both in the query and the ref.
1297   std::string params[] = {url.query(), url.ref()};
1298
1299   for (int i = 0; i < 2; ++i) {
1300     url::Component query, key, value;
1301     query.len = static_cast<int>(params[i].size());
1302     while (url::ExtractQueryKeyValue(params[i].c_str(), &query, &key, &value)) {
1303       if (key.is_nonempty() &&
1304           params[i].substr(key.begin, key.len) ==
1305               search_terms_replacement_key()) {
1306         return true;
1307       }
1308     }
1309   }
1310   return false;
1311 }
1312
1313 bool TemplateURL::ReplaceSearchTermsInURL(
1314     const GURL& url,
1315     const TemplateURLRef::SearchTermsArgs& search_terms_args,
1316     const SearchTermsData& search_terms_data,
1317     GURL* result) {
1318   // TODO(beaudoin): Use AQS from |search_terms_args| too.
1319   url::Parsed::ComponentType search_term_component;
1320   url::Component search_terms_position;
1321   base::string16 search_terms;
1322   if (!FindSearchTermsInURL(url, search_terms_data, &search_terms,
1323                             &search_term_component, &search_terms_position)) {
1324     return false;
1325   }
1326   DCHECK(search_terms_position.is_nonempty());
1327
1328   // FindSearchTermsInURL only returns true for search terms in the query or
1329   // ref, so we can call EncodeSearchTerm with |is_in_query| = true, since query
1330   // and ref are encoded in the same way.
1331   std::string input_encoding;
1332   base::string16 encoded_terms;
1333   base::string16 encoded_original_query;
1334   EncodeSearchTerms(search_terms_args, true, &input_encoding,
1335                     &encoded_terms, &encoded_original_query);
1336
1337   std::string old_params(
1338       (search_term_component == url::Parsed::REF) ? url.ref() : url.query());
1339   std::string new_params(old_params, 0, search_terms_position.begin);
1340   new_params += base::UTF16ToUTF8(search_terms_args.search_terms);
1341   new_params += old_params.substr(search_terms_position.end());
1342   url::StdStringReplacements<std::string> replacements;
1343   if (search_term_component == url::Parsed::REF)
1344     replacements.SetRefStr(new_params);
1345   else
1346     replacements.SetQueryStr(new_params);
1347   *result = url.ReplaceComponents(replacements);
1348   return true;
1349 }
1350
1351 void TemplateURL::EncodeSearchTerms(
1352     const TemplateURLRef::SearchTermsArgs& search_terms_args,
1353     bool is_in_query,
1354     std::string* input_encoding,
1355     base::string16* encoded_terms,
1356     base::string16* encoded_original_query) const {
1357
1358   std::vector<std::string> encodings(input_encodings());
1359   if (std::find(encodings.begin(), encodings.end(), "UTF-8") == encodings.end())
1360     encodings.push_back("UTF-8");
1361   for (std::vector<std::string>::const_iterator i(encodings.begin());
1362        i != encodings.end(); ++i) {
1363     if (TryEncoding(search_terms_args.search_terms,
1364                     search_terms_args.original_query, i->c_str(),
1365                     is_in_query, encoded_terms, encoded_original_query)) {
1366       *input_encoding = *i;
1367       return;
1368     }
1369   }
1370   NOTREACHED();
1371 }
1372
1373 GURL TemplateURL::GenerateSearchURL(
1374     const SearchTermsData& search_terms_data) const {
1375   if (!url_ref_.IsValid(search_terms_data))
1376     return GURL();
1377
1378   if (!url_ref_.SupportsReplacement(search_terms_data))
1379     return GURL(url());
1380
1381   // Use something obscure for the search terms argument so that in the rare
1382   // case the term replaces the URL it's unlikely another keyword would have the
1383   // same url.
1384   // TODO(jnd): Add additional parameters to get post data when the search URL
1385   // has post parameters.
1386   return GURL(url_ref_.ReplaceSearchTerms(
1387       TemplateURLRef::SearchTermsArgs(
1388           base::ASCIIToUTF16("blah.blah.blah.blah.blah")),
1389       search_terms_data, NULL));
1390 }
1391
1392 void TemplateURL::CopyFrom(const TemplateURL& other) {
1393   if (this == &other)
1394     return;
1395
1396   data_ = other.data_;
1397   url_ref_.InvalidateCachedValues();
1398   suggestions_url_ref_.InvalidateCachedValues();
1399   instant_url_ref_.InvalidateCachedValues();
1400   SetPrepopulateId(other.data_.prepopulate_id);
1401 }
1402
1403 void TemplateURL::SetURL(const std::string& url) {
1404   data_.SetURL(url);
1405   url_ref_.InvalidateCachedValues();
1406 }
1407
1408 void TemplateURL::SetPrepopulateId(int id) {
1409   data_.prepopulate_id = id;
1410   const bool prepopulated = id > 0;
1411   url_ref_.prepopulated_ = prepopulated;
1412   suggestions_url_ref_.prepopulated_ = prepopulated;
1413   instant_url_ref_.prepopulated_ = prepopulated;
1414 }
1415
1416 void TemplateURL::ResetKeywordIfNecessary(
1417     const SearchTermsData& search_terms_data,
1418     bool force) {
1419   if (IsGoogleSearchURLWithReplaceableKeyword(search_terms_data) || force) {
1420     DCHECK(GetType() != OMNIBOX_API_EXTENSION);
1421     GURL url(GenerateSearchURL(search_terms_data));
1422     if (url.is_valid())
1423       data_.SetKeyword(GenerateKeyword(url));
1424   }
1425 }
1426
1427 bool TemplateURL::FindSearchTermsInURL(
1428     const GURL& url,
1429     const SearchTermsData& search_terms_data,
1430     base::string16* search_terms,
1431     url::Parsed::ComponentType* search_term_component,
1432     url::Component* search_terms_position) {
1433   DCHECK(search_terms);
1434   search_terms->clear();
1435
1436   // Try to match with every pattern.
1437   for (size_t i = 0; i < URLCount(); ++i) {
1438     TemplateURLRef ref(this, i);
1439     if (ref.ExtractSearchTermsFromURL(url, search_terms, search_terms_data,
1440         search_term_component, search_terms_position)) {
1441       // If ExtractSearchTermsFromURL() returns true and |search_terms| is empty
1442       // it means the pattern matched but no search terms were present. In this
1443       // case we fail immediately without looking for matches in subsequent
1444       // patterns. This means that given patterns
1445       //    [ "http://foo/#q={searchTerms}", "http://foo/?q={searchTerms}" ],
1446       // calling ExtractSearchTermsFromURL() on "http://foo/?q=bar#q=' would
1447       // return false. This is important for at least Google, where such URLs
1448       // are invalid.
1449       return !search_terms->empty();
1450     }
1451   }
1452   return false;
1453 }