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 #include "url/gurl.h"
  32
  33 namespace {
  34
  35 // The TemplateURLRef has any number of terms that need to be replaced. Each of
  36 // the terms is enclosed in braces. If the character preceeding the final
  37 // brace is a ?, it indicates the term is optional and can be replaced with
  38 // an empty string.
  39 const char kStartParameter = '{';
  40 const char kEndParameter = '}';
  41 const char kOptional = '?';
  42
  43 // Known parameters found in the URL.
  44 const char kSearchTermsParameter[] = "searchTerms";
  45 const char kSearchTermsParameterFull[] = "{searchTerms}";
  46 const char kCountParameter[] = "count";
  47 const char kStartIndexParameter[] = "startIndex";
  48 const char kStartPageParameter[] = "startPage";
  49 const char kLanguageParameter[] = "language";
  50 const char kInputEncodingParameter[] = "inputEncoding";
  51 const char kOutputEncodingParameter[] = "outputEncoding";
  52
  53 const char kGoogleAssistedQueryStatsParameter[] = "google:assistedQueryStats";
  54
  55 // Host/Domain Google searches are relative to.
  56 const char kGoogleBaseURLParameter[] = "google:baseURL";
  57 const char kGoogleBaseURLParameterFull[] = "{google:baseURL}";
  58
  59 // Like google:baseURL, but for the Search Suggest capability.
  60 const char kGoogleBaseSuggestURLParameter[] = "google:baseSuggestURL";
  61 const char kGoogleBaseSuggestURLParameterFull[] = "{google:baseSuggestURL}";
  62 const char kGoogleBookmarkBarPinnedParameter[] = "google:bookmarkBarPinned";
  63 const char kGoogleContextualSearchContextData[] =
  64     "google:contextualSearchContextData";
  65 const char kGoogleContextualSearchVersion[] = "google:contextualSearchVersion";
  66 const char kGoogleCurrentPageUrlParameter[] = "google:currentPageUrl";
  67 const char kGoogleCursorPositionParameter[] = "google:cursorPosition";
  68 const char kGoogleForceInstantResultsParameter[] = "google:forceInstantResults";
  69 const char kGoogleImageSearchSource[] = "google:imageSearchSource";
  70 const char kGoogleImageThumbnailParameter[] = "google:imageThumbnail";
  71 const char kGoogleImageOriginalWidth[] = "google:imageOriginalWidth";
  72 const char kGoogleImageOriginalHeight[] = "google:imageOriginalHeight";
  73 const char kGoogleImageURLParameter[] = "google:imageURL";
  74 const char kGoogleInputTypeParameter[] = "google:inputType";
  75 const char kGoogleInstantExtendedEnabledParameter[] =
  76     "google:instantExtendedEnabledParameter";
  77 const char kGoogleInstantExtendedEnabledKey[] =
  78     "google:instantExtendedEnabledKey";
  79 const char kGoogleInstantExtendedEnabledKeyFull[] =
  80     "{google:instantExtendedEnabledKey}";
  81 const char kGoogleNTPIsThemedParameter[] = "google:ntpIsThemedParameter";
  82 const char kGoogleOmniboxStartMarginParameter[] =
  83     "google:omniboxStartMarginParameter";
  84 const char kGoogleOriginalQueryForSuggestionParameter[] =
  85     "google:originalQueryForSuggestion";
  86 const char kGooglePageClassificationParameter[] = "google:pageClassification";
  87 const char kGooglePrefetchQuery[] = "google:prefetchQuery";
  88 const char kGoogleRLZParameter[] = "google:RLZ";
  89 const char kGoogleSearchClient[] = "google:searchClient";
  90 const char kGoogleSearchFieldtrialParameter[] =
  91     "google:searchFieldtrialParameter";
  92 const char kGoogleSearchVersion[] = "google:searchVersion";
  93 const char kGoogleSessionToken[] = "google:sessionToken";
  94 const char kGoogleSourceIdParameter[] = "google:sourceId";
  95 const char kGoogleSuggestAPIKeyParameter[] = "google:suggestAPIKeyParameter";
  96 const char kGoogleSuggestClient[] = "google:suggestClient";
  97 const char kGoogleSuggestRequestId[] = "google:suggestRid";
  98
  99 // Same as kSearchTermsParameter, with no escaping.
 100 const char kGoogleUnescapedSearchTermsParameter[] =
 101     "google:unescapedSearchTerms";
 102 const char kGoogleUnescapedSearchTermsParameterFull[] =
 103     "{google:unescapedSearchTerms}";
 104
 105 // Display value for kSearchTermsParameter.
 106 const char kDisplaySearchTerms[] = "%s";
 107
 108 // Display value for kGoogleUnescapedSearchTermsParameter.
 109 const char kDisplayUnescapedSearchTerms[] = "%S";
 110
 111 // Used if the count parameter is not optional. Indicates we want 10 search
 112 // results.
 113 const char kDefaultCount[] = "10";
 114
 115 // Used if the parameter kOutputEncodingParameter is required.
 116 const char kOutputEncodingType[] = "UTF-8";
 117
 118 // Attempts to encode |terms| and |original_query| in |encoding| and escape
 119 // them.  |terms| may be escaped as path or query depending on |is_in_query|;
 120 // |original_query| is always escaped as query.  Returns whether the encoding
 121 // process succeeded.
 122 bool TryEncoding(const base::string16& terms,
 123                  const base::string16& original_query,
 124                  const char* encoding,
 125                  bool is_in_query,
 126                  base::string16* escaped_terms,
 127                  base::string16* escaped_original_query) {
 128   DCHECK(escaped_terms);
 129   DCHECK(escaped_original_query);
 130   std::string encoded_terms;
 131   if (!base::UTF16ToCodepage(terms, encoding,
 132       base::OnStringConversionError::SKIP, &encoded_terms))
 133     return false;
 134   *escaped_terms = base::UTF8ToUTF16(is_in_query ?
 135       net::EscapeQueryParamValue(encoded_terms, true) :
 136       net::EscapePath(encoded_terms));
 137   if (original_query.empty())
 138     return true;
 139   std::string encoded_original_query;
 140   if (!base::UTF16ToCodepage(original_query, encoding,
 141       base::OnStringConversionError::SKIP, &encoded_original_query))
 142     return false;
 143   *escaped_original_query = base::UTF8ToUTF16(
 144       net::EscapeQueryParamValue(encoded_original_query, true));
 145   return true;
 146 }
 147
 148 // Extract query key and host given a list of parameters coming from the URL
 149 // query or ref.
 150 std::string FindSearchTermsKey(const std::string& params) {
 151   if (params.empty())
 152     return std::string();
 153   url::Component query, key, value;
 154   query.len = static_cast<int>(params.size());
 155   while (url::ExtractQueryKeyValue(params.c_str(), &query, &key, &value)) {
 156     if (key.is_nonempty() && value.is_nonempty()) {
 157       std::string value_string = params.substr(value.begin, value.len);
 158       if (value_string.find(kSearchTermsParameterFull, 0) !=
 159           std::string::npos ||
 160           value_string.find(kGoogleUnescapedSearchTermsParameterFull, 0) !=
 161           std::string::npos) {
 162         return params.substr(key.begin, key.len);
 163       }
 164     }
 165   }
 166   return std::string();
 167 }
 168
 169 bool IsTemplateParameterString(const std::string& param) {
 170   return (param.length() > 2) && (*(param.begin()) == kStartParameter) &&
 171       (*(param.rbegin()) == kEndParameter);
 172 }
 173
 174 }  // namespace
 175
 176
 177 // TemplateURLRef::SearchTermsArgs --------------------------------------------
 178
 179 TemplateURLRef::SearchTermsArgs::SearchTermsArgs(
 180     const base::string16& search_terms)
 181     : search_terms(search_terms),
 182       input_type(metrics::OmniboxInputType::INVALID),
 183       accepted_suggestion(NO_SUGGESTIONS_AVAILABLE),
 184       cursor_position(base::string16::npos),
 185       enable_omnibox_start_margin(false),
 186       page_classification(metrics::OmniboxEventProto::INVALID_SPEC),
 187       bookmark_bar_pinned(false),
 188       append_extra_query_params(false),
 189       force_instant_results(false),
 190       from_app_list(false),
 191       contextual_search_params(ContextualSearchParams()) {
 192 }
 193
 194 TemplateURLRef::SearchTermsArgs::~SearchTermsArgs() {
 195 }
 196
 197 TemplateURLRef::SearchTermsArgs::ContextualSearchParams::
 198     ContextualSearchParams()
 199     : version(-1),
 200       start(base::string16::npos),
 201       end(base::string16::npos),
 202       resolve(true) {
 203 }
 204
 205 TemplateURLRef::SearchTermsArgs::ContextualSearchParams::
 206     ContextualSearchParams(
 207         const int version,
 208         const std::string& selection,
 209         const std::string& base_page_url,
 210         const bool resolve)
 211     : version(version),
 212       start(base::string16::npos),
 213       end(base::string16::npos),
 214       selection(selection),
 215       base_page_url(base_page_url),
 216       resolve(resolve) {
 217 }
 218
 219 TemplateURLRef::SearchTermsArgs::ContextualSearchParams::
 220     ContextualSearchParams(
 221         const int version,
 222         const size_t start,
 223         const size_t end,
 224         const std::string& selection,
 225         const std::string& content,
 226         const std::string& base_page_url,
 227         const std::string& encoding,
 228         const bool resolve)
 229     : version(version),
 230       start(start),
 231       end(end),
 232       selection(selection),
 233       content(content),
 234       base_page_url(base_page_url),
 235       encoding(encoding),
 236       resolve(resolve) {
 237 }
 238
 239 TemplateURLRef::SearchTermsArgs::ContextualSearchParams::
 240     ~ContextualSearchParams() {
 241 }
 242
 243 // TemplateURLRef -------------------------------------------------------------
 244
 245 TemplateURLRef::TemplateURLRef(TemplateURL* owner, Type type)
 246     : owner_(owner),
 247       type_(type),
 248       index_in_owner_(0),
 249       parsed_(false),
 250       valid_(false),
 251       supports_replacements_(false),
 252       search_term_key_location_(url::Parsed::QUERY),
 253       prepopulated_(false) {
 254   DCHECK(owner_);
 255   DCHECK_NE(INDEXED, type_);
 256 }
 257
 258 TemplateURLRef::TemplateURLRef(TemplateURL* owner, size_t index_in_owner)
 259     : owner_(owner),
 260       type_(INDEXED),
 261       index_in_owner_(index_in_owner),
 262       parsed_(false),
 263       valid_(false),
 264       supports_replacements_(false),
 265       search_term_key_location_(url::Parsed::QUERY),
 266       prepopulated_(false) {
 267   DCHECK(owner_);
 268   DCHECK_LT(index_in_owner_, owner_->URLCount());
 269 }
 270
 271 TemplateURLRef::~TemplateURLRef() {
 272 }
 273
 274 std::string TemplateURLRef::GetURL() const {
 275   switch (type_) {
 276     case SEARCH:            return owner_->url();
 277     case SUGGEST:           return owner_->suggestions_url();
 278     case INSTANT:           return owner_->instant_url();
 279     case IMAGE:             return owner_->image_url();
 280     case NEW_TAB:           return owner_->new_tab_url();
 281     case CONTEXTUAL_SEARCH: return owner_->contextual_search_url();
 282     case INDEXED:           return owner_->GetURL(index_in_owner_);
 283     default:       NOTREACHED(); return std::string();  // NOLINT
 284   }
 285 }
 286
 287 std::string TemplateURLRef::GetPostParamsString() const {
 288   switch (type_) {
 289     case INDEXED:
 290     case SEARCH:            return owner_->search_url_post_params();
 291     case SUGGEST:           return owner_->suggestions_url_post_params();
 292     case INSTANT:           return owner_->instant_url_post_params();
 293     case NEW_TAB:           return std::string();
 294     case CONTEXTUAL_SEARCH: return std::string();
 295     case IMAGE:             return owner_->image_url_post_params();
 296     default:      NOTREACHED(); return std::string();  // NOLINT
 297   }
 298 }
 299
 300 bool TemplateURLRef::UsesPOSTMethod(
 301     const SearchTermsData& search_terms_data) const {
 302   ParseIfNecessary(search_terms_data);
 303   return !post_params_.empty();
 304 }
 305
 306 bool TemplateURLRef::EncodeFormData(const PostParams& post_params,
 307                                     PostContent* post_content) const {
 308   if (post_params.empty())
 309     return true;
 310   if (!post_content)
 311     return false;
 312
 313   const char kUploadDataMIMEType[] = "multipart/form-data; boundary=";
 314   const char kMultipartBoundary[] = "----+*+----%016" PRIx64 "----+*+----";
 315   // Each name/value pair is stored in a body part which is preceded by a
 316   // boundary delimiter line. Uses random number generator here to create
 317   // a unique boundary delimiter for form data encoding.
 318   std::string boundary = base::StringPrintf(kMultipartBoundary,
 319                                             base::RandUint64());
 320   // Sets the content MIME type.
 321   post_content->first = kUploadDataMIMEType;
 322   post_content->first += boundary;
 323   // Encodes the post parameters.
 324   std::string* post_data = &post_content->second;
 325   post_data->clear();
 326   for (const auto& param : post_params) {
 327     DCHECK(!param.name.empty());
 328     net::AddMultipartValueForUpload(param.name, param.value, boundary,
 329                                     param.content_type, post_data);
 330   }
 331   net::AddMultipartFinalDelimiterForUpload(boundary, post_data);
 332   return true;
 333 }
 334
 335 bool TemplateURLRef::SupportsReplacement(
 336     const SearchTermsData& search_terms_data) const {
 337   ParseIfNecessary(search_terms_data);
 338   return valid_ && supports_replacements_;
 339 }
 340
 341 std::string TemplateURLRef::ReplaceSearchTerms(
 342     const SearchTermsArgs& search_terms_args,
 343     const SearchTermsData& search_terms_data,
 344     PostContent* post_content) const {
 345   ParseIfNecessary(search_terms_data);
 346   if (!valid_)
 347     return std::string();
 348
 349   std::string url(HandleReplacements(search_terms_args, search_terms_data,
 350                                      post_content));
 351
 352   GURL gurl(url);
 353   if (!gurl.is_valid())
 354     return url;
 355
 356   std::vector<std::string> query_params;
 357   if (search_terms_args.append_extra_query_params) {
 358     std::string extra_params(
 359         base::CommandLine::ForCurrentProcess()->GetSwitchValueASCII(
 360             switches::kExtraSearchQueryParams));
 361     if (!extra_params.empty())
 362       query_params.push_back(extra_params);
 363   }
 364   if (!search_terms_args.suggest_query_params.empty())
 365     query_params.push_back(search_terms_args.suggest_query_params);
 366   if (!gurl.query().empty())
 367     query_params.push_back(gurl.query());
 368
 369   if (query_params.empty())
 370     return url;
 371
 372   GURL::Replacements replacements;
 373   std::string query_str = JoinString(query_params, "&");
 374   replacements.SetQueryStr(query_str);
 375   return gurl.ReplaceComponents(replacements).possibly_invalid_spec();
 376 }
 377
 378 bool TemplateURLRef::IsValid(const SearchTermsData& search_terms_data) const {
 379   ParseIfNecessary(search_terms_data);
 380   return valid_;
 381 }
 382
 383 base::string16 TemplateURLRef::DisplayURL(
 384     const SearchTermsData& search_terms_data) const {
 385   ParseIfNecessary(search_terms_data);
 386   base::string16 result(base::UTF8ToUTF16(GetURL()));
 387   if (valid_ && !replacements_.empty()) {
 388     ReplaceSubstringsAfterOffset(&result, 0,
 389                                  base::ASCIIToUTF16(kSearchTermsParameterFull),
 390                                  base::ASCIIToUTF16(kDisplaySearchTerms));
 391     ReplaceSubstringsAfterOffset(&result, 0,
 392         base::ASCIIToUTF16(kGoogleUnescapedSearchTermsParameterFull),
 393         base::ASCIIToUTF16(kDisplayUnescapedSearchTerms));
 394   }
 395   return result;
 396 }
 397
 398 // static
 399 std::string TemplateURLRef::DisplayURLToURLRef(
 400     const base::string16& display_url) {
 401   base::string16 result = display_url;
 402   ReplaceSubstringsAfterOffset(&result, 0,
 403                                base::ASCIIToUTF16(kDisplaySearchTerms),
 404                                base::ASCIIToUTF16(kSearchTermsParameterFull));
 405   ReplaceSubstringsAfterOffset(
 406       &result, 0,
 407       base::ASCIIToUTF16(kDisplayUnescapedSearchTerms),
 408       base::ASCIIToUTF16(kGoogleUnescapedSearchTermsParameterFull));
 409   return base::UTF16ToUTF8(result);
 410 }
 411
 412 const std::string& TemplateURLRef::GetHost(
 413     const SearchTermsData& search_terms_data) const {
 414   ParseIfNecessary(search_terms_data);
 415   return host_;
 416 }
 417
 418 const std::string& TemplateURLRef::GetPath(
 419     const SearchTermsData& search_terms_data) const {
 420   ParseIfNecessary(search_terms_data);
 421   return path_;
 422 }
 423
 424 const std::string& TemplateURLRef::GetSearchTermKey(
 425     const SearchTermsData& search_terms_data) const {
 426   ParseIfNecessary(search_terms_data);
 427   return search_term_key_;
 428 }
 429
 430 base::string16 TemplateURLRef::SearchTermToString16(
 431     const std::string& term) const {
 432   const std::vector<std::string>& encodings = owner_->input_encodings();
 433   base::string16 result;
 434
 435   std::string unescaped = net::UnescapeURLComponent(
 436       term,
 437       net::UnescapeRule::REPLACE_PLUS_WITH_SPACE |
 438       net::UnescapeRule::URL_SPECIAL_CHARS);
 439   for (size_t i = 0; i < encodings.size(); ++i) {
 440     if (base::CodepageToUTF16(unescaped, encodings[i].c_str(),
 441                               base::OnStringConversionError::FAIL, &result))
 442       return result;
 443   }
 444
 445   // Always fall back on UTF-8 if it works.
 446   if (base::CodepageToUTF16(unescaped, base::kCodepageUTF8,
 447                             base::OnStringConversionError::FAIL, &result))
 448     return result;
 449
 450   // When nothing worked, just use the escaped text. We have no idea what the
 451   // encoding is. We need to substitute spaces for pluses ourselves since we're
 452   // not sending it through an unescaper.
 453   result = base::UTF8ToUTF16(term);
 454   std::replace(result.begin(), result.end(), '+', ' ');
 455   return result;
 456 }
 457
 458 bool TemplateURLRef::HasGoogleBaseURLs(
 459     const SearchTermsData& search_terms_data) const {
 460   ParseIfNecessary(search_terms_data);
 461   for (size_t i = 0; i < replacements_.size(); ++i) {
 462     if ((replacements_[i].type == GOOGLE_BASE_URL) ||
 463         (replacements_[i].type == GOOGLE_BASE_SUGGEST_URL))
 464       return true;
 465   }
 466   return false;
 467 }
 468
 469 bool TemplateURLRef::ExtractSearchTermsFromURL(
 470     const GURL& url,
 471     base::string16* search_terms,
 472     const SearchTermsData& search_terms_data,
 473     url::Parsed::ComponentType* search_terms_component,
 474     url::Component* search_terms_position) const {
 475   DCHECK(search_terms);
 476   search_terms->clear();
 477
 478   ParseIfNecessary(search_terms_data);
 479
 480   // We need a search term in the template URL to extract something.
 481   if (search_term_key_.empty())
 482     return false;
 483
 484   // TODO(beaudoin): Support patterns of the form http://foo/{searchTerms}/
 485   // See crbug.com/153798
 486
 487   // Fill-in the replacements. We don't care about search terms in the pattern,
 488   // so we use the empty string.
 489   // Currently we assume the search term only shows in URL, not in post params.
 490   GURL pattern(ReplaceSearchTerms(SearchTermsArgs(base::string16()),
 491                                   search_terms_data, NULL));
 492   // Host, path and port must match.
 493   if (url.port() != pattern.port() ||
 494       url.host() != host_ ||
 495       url.path() != path_) {
 496     return false;
 497   }
 498
 499   // Parameter must be present either in the query or the ref.
 500   const std::string& params(
 501       (search_term_key_location_ == url::Parsed::QUERY) ?
 502           url.query() : url.ref());
 503
 504   url::Component query, key, value;
 505   query.len = static_cast<int>(params.size());
 506   bool key_found = false;
 507   while (url::ExtractQueryKeyValue(params.c_str(), &query, &key, &value)) {
 508     if (key.is_nonempty()) {
 509       if (params.substr(key.begin, key.len) == search_term_key_) {
 510         // Fail if search term key is found twice.
 511         if (key_found) {
 512           search_terms->clear();
 513           return false;
 514         }
 515         key_found = true;
 516         // Extract the search term.
 517         *search_terms = net::UnescapeAndDecodeUTF8URLComponent(
 518             params.substr(value.begin, value.len),
 519             net::UnescapeRule::SPACES |
 520                 net::UnescapeRule::URL_SPECIAL_CHARS |
 521                 net::UnescapeRule::REPLACE_PLUS_WITH_SPACE);
 522         if (search_terms_component)
 523           *search_terms_component = search_term_key_location_;
 524         if (search_terms_position)
 525           *search_terms_position = value;
 526       }
 527     }
 528   }
 529   return key_found;
 530 }
 531
 532 void TemplateURLRef::InvalidateCachedValues() const {
 533   supports_replacements_ = valid_ = parsed_ = false;
 534   host_.clear();
 535   path_.clear();
 536   search_term_key_.clear();
 537   replacements_.clear();
 538   post_params_.clear();
 539 }
 540
 541 bool TemplateURLRef::ParseParameter(size_t start,
 542                                     size_t end,
 543                                     std::string* url,
 544                                     Replacements* replacements) const {
 545   DCHECK(start != std::string::npos &&
 546          end != std::string::npos && end > start);
 547   size_t length = end - start - 1;
 548   bool optional = false;
 549   if ((*url)[end - 1] == kOptional) {
 550     optional = true;
 551     length--;
 552   }
 553   std::string parameter(url->substr(start + 1, length));
 554   std::string full_parameter(url->substr(start, end - start + 1));
 555   // Remove the parameter from the string.  For parameters who replacement is
 556   // constant and already known, just replace them directly.  For other cases,
 557   // like parameters whose values may change over time, use |replacements|.
 558   url->erase(start, end - start + 1);
 559   if (parameter == kSearchTermsParameter) {
 560     replacements->push_back(Replacement(SEARCH_TERMS, start));
 561   } else if (parameter == kCountParameter) {
 562     if (!optional)
 563       url->insert(start, kDefaultCount);
 564   } else if (parameter == kGoogleAssistedQueryStatsParameter) {
 565     replacements->push_back(Replacement(GOOGLE_ASSISTED_QUERY_STATS, start));
 566   } else if (parameter == kGoogleBaseURLParameter) {
 567     replacements->push_back(Replacement(GOOGLE_BASE_URL, start));
 568   } else if (parameter == kGoogleBaseSuggestURLParameter) {
 569     replacements->push_back(Replacement(GOOGLE_BASE_SUGGEST_URL, start));
 570   } else if (parameter == kGoogleBookmarkBarPinnedParameter) {
 571     replacements->push_back(Replacement(GOOGLE_BOOKMARK_BAR_PINNED, start));
 572   } else if (parameter == kGoogleCurrentPageUrlParameter) {
 573     replacements->push_back(Replacement(GOOGLE_CURRENT_PAGE_URL, start));
 574   } else if (parameter == kGoogleCursorPositionParameter) {
 575     replacements->push_back(Replacement(GOOGLE_CURSOR_POSITION, start));
 576   } else if (parameter == kGoogleForceInstantResultsParameter) {
 577     replacements->push_back(Replacement(GOOGLE_FORCE_INSTANT_RESULTS, start));
 578   } else if (parameter == kGoogleImageOriginalHeight) {
 579     replacements->push_back(
 580         Replacement(TemplateURLRef::GOOGLE_IMAGE_ORIGINAL_HEIGHT, start));
 581   } else if (parameter == kGoogleImageOriginalWidth) {
 582     replacements->push_back(
 583         Replacement(TemplateURLRef::GOOGLE_IMAGE_ORIGINAL_WIDTH, start));
 584   } else if (parameter == kGoogleImageSearchSource) {
 585     replacements->push_back(
 586         Replacement(TemplateURLRef::GOOGLE_IMAGE_SEARCH_SOURCE, start));
 587   } else if (parameter == kGoogleImageThumbnailParameter) {
 588     replacements->push_back(
 589         Replacement(TemplateURLRef::GOOGLE_IMAGE_THUMBNAIL, start));
 590   } else if (parameter == kGoogleImageURLParameter) {
 591     replacements->push_back(Replacement(TemplateURLRef::GOOGLE_IMAGE_URL,
 592                                         start));
 593   } else if (parameter == kGoogleInputTypeParameter) {
 594     replacements->push_back(Replacement(TemplateURLRef::GOOGLE_INPUT_TYPE,
 595                                         start));
 596   } else if (parameter == kGoogleInstantExtendedEnabledParameter) {
 597     replacements->push_back(Replacement(GOOGLE_INSTANT_EXTENDED_ENABLED,
 598                                         start));
 599   } else if (parameter == kGoogleInstantExtendedEnabledKey) {
 600     url->insert(start, google_util::kInstantExtendedAPIParam);
 601   } else if (parameter == kGoogleNTPIsThemedParameter) {
 602     replacements->push_back(Replacement(GOOGLE_NTP_IS_THEMED, start));
 603   } else if (parameter == kGoogleOmniboxStartMarginParameter) {
 604     replacements->push_back(Replacement(GOOGLE_OMNIBOX_START_MARGIN, start));
 605   } else if (parameter == kGoogleContextualSearchVersion) {
 606     replacements->push_back(
 607         Replacement(GOOGLE_CONTEXTUAL_SEARCH_VERSION, start));
 608   } else if (parameter == kGoogleContextualSearchContextData) {
 609     replacements->push_back(
 610         Replacement(GOOGLE_CONTEXTUAL_SEARCH_CONTEXT_DATA, start));
 611   } else if (parameter == kGoogleOriginalQueryForSuggestionParameter) {
 612     replacements->push_back(Replacement(GOOGLE_ORIGINAL_QUERY_FOR_SUGGESTION,
 613                                         start));
 614   } else if (parameter == kGooglePageClassificationParameter) {
 615     replacements->push_back(Replacement(GOOGLE_PAGE_CLASSIFICATION, start));
 616   } else if (parameter == kGooglePrefetchQuery) {
 617     replacements->push_back(Replacement(GOOGLE_PREFETCH_QUERY, start));
 618   } else if (parameter == kGoogleRLZParameter) {
 619     replacements->push_back(Replacement(GOOGLE_RLZ, start));
 620   } else if (parameter == kGoogleSearchClient) {
 621     replacements->push_back(Replacement(GOOGLE_SEARCH_CLIENT, start));
 622   } else if (parameter == kGoogleSearchFieldtrialParameter) {
 623     replacements->push_back(Replacement(GOOGLE_SEARCH_FIELDTRIAL_GROUP, start));
 624   } else if (parameter == kGoogleSearchVersion) {
 625     replacements->push_back(Replacement(GOOGLE_SEARCH_VERSION, start));
 626   } else if (parameter == kGoogleSessionToken) {
 627     replacements->push_back(Replacement(GOOGLE_SESSION_TOKEN, start));
 628   } else if (parameter == kGoogleSourceIdParameter) {
 629 #if defined(OS_ANDROID)
 630     url->insert(start, "sourceid=chrome-mobile&");
 631 #else
 632     url->insert(start, "sourceid=chrome&");
 633 #endif
 634   } else if (parameter == kGoogleSuggestAPIKeyParameter) {
 635     url->insert(start,
 636                 net::EscapeQueryParamValue(google_apis::GetAPIKey(), false));
 637   } else if (parameter == kGoogleSuggestClient) {
 638     replacements->push_back(Replacement(GOOGLE_SUGGEST_CLIENT, start));
 639   } else if (parameter == kGoogleSuggestRequestId) {
 640     replacements->push_back(Replacement(GOOGLE_SUGGEST_REQUEST_ID, start));
 641   } else if (parameter == kGoogleUnescapedSearchTermsParameter) {
 642     replacements->push_back(Replacement(GOOGLE_UNESCAPED_SEARCH_TERMS, start));
 643   } else if (parameter == kInputEncodingParameter) {
 644     replacements->push_back(Replacement(ENCODING, start));
 645   } else if (parameter == kLanguageParameter) {
 646     replacements->push_back(Replacement(LANGUAGE, start));
 647   } else if (parameter == kOutputEncodingParameter) {
 648     if (!optional)
 649       url->insert(start, kOutputEncodingType);
 650   } else if ((parameter == kStartIndexParameter) ||
 651              (parameter == kStartPageParameter)) {
 652     // We don't support these.
 653     if (!optional)
 654       url->insert(start, "1");
 655   } else if (!prepopulated_) {
 656     // If it's a prepopulated URL, we know that it's safe to remove unknown
 657     // parameters, so just ignore this and return true below. Otherwise it could
 658     // be some garbage but can also be a javascript block. Put it back.
 659     url->insert(start, full_parameter);
 660     return false;
 661   }
 662   return true;
 663 }
 664
 665 std::string TemplateURLRef::ParseURL(const std::string& url,
 666                                      Replacements* replacements,
 667                                      PostParams* post_params,
 668                                      bool* valid) const {
 669   *valid = false;
 670   std::string parsed_url = url;
 671   for (size_t last = 0; last != std::string::npos; ) {
 672     last = parsed_url.find(kStartParameter, last);
 673     if (last != std::string::npos) {
 674       size_t template_end = parsed_url.find(kEndParameter, last);
 675       if (template_end != std::string::npos) {
 676         // Since we allow Javascript in the URL, {} pairs could be nested. Match
 677         // only leaf pairs with supported parameters.
 678         size_t next_template_start = parsed_url.find(kStartParameter, last + 1);
 679         if (next_template_start == std::string::npos ||
 680             next_template_start > template_end) {
 681           // If successful, ParseParameter erases from the string as such no
 682           // need to update |last|. If failed, move |last| to the end of pair.
 683           if (!ParseParameter(last, template_end, &parsed_url, replacements)) {
 684             // |template_end| + 1 may be beyond the end of the string.
 685             last = template_end;
 686           }
 687         } else {
 688           last = next_template_start;
 689         }
 690       } else {
 691         // Open brace without a closing brace, return.
 692         return std::string();
 693       }
 694     }
 695   }
 696
 697   // Handles the post parameters.
 698   const std::string& post_params_string = GetPostParamsString();
 699   if (!post_params_string.empty()) {
 700     typedef std::vector<std::string> Strings;
 701     Strings param_list;
 702     base::SplitString(post_params_string, ',', &param_list);
 703
 704     for (Strings::const_iterator iterator = param_list.begin();
 705          iterator != param_list.end(); ++iterator) {
 706       Strings parts;
 707       // The '=' delimiter is required and the name must be not empty.
 708       base::SplitString(*iterator, '=', &parts);
 709       if ((parts.size() != 2U) || parts[0].empty())
 710         return std::string();
 711
 712       std::string& value = parts[1];
 713       size_t replacements_size = replacements->size();
 714       if (IsTemplateParameterString(value))
 715         ParseParameter(0, value.length() - 1, &value, replacements);
 716       PostParam param = { parts[0], value };
 717       post_params->push_back(param);
 718       // If there was a replacement added, points its index to last added
 719       // PostParam.
 720       if (replacements->size() > replacements_size) {
 721         DCHECK_EQ(replacements_size + 1, replacements->size());
 722         Replacement* r = &replacements->back();
 723         r->is_post_param = true;
 724         r->index = post_params->size() - 1;
 725       }
 726     }
 727     DCHECK(!post_params->empty());
 728   }
 729
 730   *valid = true;
 731   return parsed_url;
 732 }
 733
 734 void TemplateURLRef::ParseIfNecessary(
 735     const SearchTermsData& search_terms_data) const {
 736   if (!parsed_) {
 737     InvalidateCachedValues();
 738     parsed_ = true;
 739     parsed_url_ = ParseURL(GetURL(), &replacements_, &post_params_, &valid_);
 740     supports_replacements_ = false;
 741     if (valid_) {
 742       bool has_only_one_search_term = false;
 743       for (Replacements::const_iterator i = replacements_.begin();
 744            i != replacements_.end(); ++i) {
 745         if ((i->type == SEARCH_TERMS) ||
 746             (i->type == GOOGLE_UNESCAPED_SEARCH_TERMS)) {
 747           if (has_only_one_search_term) {
 748             has_only_one_search_term = false;
 749             break;
 750           }
 751           has_only_one_search_term = true;
 752           supports_replacements_ = true;
 753         }
 754       }
 755       // Only parse the host/key if there is one search term. Technically there
 756       // could be more than one term, but it's uncommon; so we punt.
 757       if (has_only_one_search_term)
 758         ParseHostAndSearchTermKey(search_terms_data);
 759     }
 760   }
 761 }
 762
 763 void TemplateURLRef::ParseHostAndSearchTermKey(
 764     const SearchTermsData& search_terms_data) const {
 765   std::string url_string(GetURL());
 766   ReplaceSubstringsAfterOffset(&url_string, 0,
 767                                kGoogleBaseURLParameterFull,
 768                                search_terms_data.GoogleBaseURLValue());
 769   ReplaceSubstringsAfterOffset(&url_string, 0,
 770                                kGoogleBaseSuggestURLParameterFull,
 771                                search_terms_data.GoogleBaseSuggestURLValue());
 772
 773   search_term_key_.clear();
 774   host_.clear();
 775   path_.clear();
 776   search_term_key_location_ = url::Parsed::REF;
 777
 778   GURL url(url_string);
 779   if (!url.is_valid())
 780     return;
 781
 782   std::string query_key = FindSearchTermsKey(url.query());
 783   std::string ref_key = FindSearchTermsKey(url.ref());
 784   if (query_key.empty() == ref_key.empty())
 785     return;  // No key or multiple keys found.  We only handle having one key.
 786   search_term_key_ = query_key.empty() ? ref_key : query_key;
 787   search_term_key_location_ =
 788       query_key.empty() ? url::Parsed::REF : url::Parsed::QUERY;
 789   host_ = url.host();
 790   path_ = url.path();
 791 }
 792
 793 void TemplateURLRef::HandleReplacement(const std::string& name,
 794                                        const std::string& value,
 795                                        const Replacement& replacement,
 796                                        std::string* url) const {
 797   size_t pos = replacement.index;
 798   if (replacement.is_post_param) {
 799     DCHECK_LT(pos, post_params_.size());
 800     DCHECK(!post_params_[pos].name.empty());
 801     post_params_[pos].value = value;
 802   } else {
 803     url->insert(pos, name.empty() ? value : (name + "=" + value + "&"));
 804   }
 805 }
 806
 807 std::string TemplateURLRef::HandleReplacements(
 808     const SearchTermsArgs& search_terms_args,
 809     const SearchTermsData& search_terms_data,
 810     PostContent* post_content) const {
 811   if (replacements_.empty()) {
 812     if (!post_params_.empty())
 813       EncodeFormData(post_params_, post_content);
 814     return parsed_url_;
 815   }
 816
 817   // Determine if the search terms are in the query or before. We're escaping
 818   // space as '+' in the former case and as '%20' in the latter case.
 819   bool is_in_query = true;
 820   for (Replacements::iterator i = replacements_.begin();
 821        i != replacements_.end(); ++i) {
 822     if (i->type == SEARCH_TERMS) {
 823       base::string16::size_type query_start = parsed_url_.find('?');
 824       is_in_query = query_start != base::string16::npos &&
 825           (static_cast<base::string16::size_type>(i->index) > query_start);
 826       break;
 827     }
 828   }
 829
 830   std::string input_encoding;
 831   base::string16 encoded_terms;
 832   base::string16 encoded_original_query;
 833   owner_->EncodeSearchTerms(search_terms_args, is_in_query, &input_encoding,
 834                             &encoded_terms, &encoded_original_query);
 835
 836   std::string url = parsed_url_;
 837
 838   // replacements_ is ordered in ascending order, as such we need to iterate
 839   // from the back.
 840   for (Replacements::reverse_iterator i = replacements_.rbegin();
 841        i != replacements_.rend(); ++i) {
 842     switch (i->type) {
 843       case ENCODING:
 844         HandleReplacement(std::string(), input_encoding, *i, &url);
 845         break;
 846
 847       case GOOGLE_ASSISTED_QUERY_STATS:
 848         DCHECK(!i->is_post_param);
 849         if (!search_terms_args.assisted_query_stats.empty()) {
 850           // Get the base URL without substituting AQS to avoid infinite
 851           // recursion.  We need the URL to find out if it meets all
 852           // AQS requirements (e.g. HTTPS protocol check).
 853           // See TemplateURLRef::SearchTermsArgs for more details.
 854           SearchTermsArgs search_terms_args_without_aqs(search_terms_args);
 855           search_terms_args_without_aqs.assisted_query_stats.clear();
 856           GURL base_url(ReplaceSearchTerms(
 857               search_terms_args_without_aqs, search_terms_data, NULL));
 858           if (base_url.SchemeIs(url::kHttpsScheme)) {
 859             HandleReplacement(
 860                 "aqs", search_terms_args.assisted_query_stats, *i, &url);
 861           }
 862         }
 863         break;
 864
 865       case GOOGLE_BASE_URL:
 866         DCHECK(!i->is_post_param);
 867         HandleReplacement(
 868             std::string(), search_terms_data.GoogleBaseURLValue(), *i, &url);
 869         break;
 870
 871       case GOOGLE_BASE_SUGGEST_URL:
 872         DCHECK(!i->is_post_param);
 873         HandleReplacement(
 874             std::string(), search_terms_data.GoogleBaseSuggestURLValue(), *i,
 875             &url);
 876         break;
 877
 878       case GOOGLE_BOOKMARK_BAR_PINNED:
 879         if (search_terms_data.IsShowingSearchTermsOnSearchResultsPages()) {
 880           // Log whether the bookmark bar is pinned when the user is seeing
 881           // InstantExtended on the SRP.
 882           DCHECK(!i->is_post_param);
 883           HandleReplacement(
 884               "bmbp", search_terms_args.bookmark_bar_pinned ? "1" : "0", *i,
 885               &url);
 886         }
 887         break;
 888
 889       case GOOGLE_CURRENT_PAGE_URL:
 890         DCHECK(!i->is_post_param);
 891         if (!search_terms_args.current_page_url.empty()) {
 892           const std::string& escaped_current_page_url =
 893               net::EscapeQueryParamValue(search_terms_args.current_page_url,
 894                                          true);
 895           HandleReplacement("url", escaped_current_page_url, *i, &url);
 896         }
 897         break;
 898
 899       case GOOGLE_CURSOR_POSITION:
 900         DCHECK(!i->is_post_param);
 901         if (search_terms_args.cursor_position != base::string16::npos)
 902           HandleReplacement(
 903               "cp",
 904               base::StringPrintf("%" PRIuS, search_terms_args.cursor_position),
 905               *i,
 906               &url);
 907         break;
 908
 909       case GOOGLE_FORCE_INSTANT_RESULTS:
 910         DCHECK(!i->is_post_param);
 911         HandleReplacement(std::string(),
 912                           search_terms_data.ForceInstantResultsParam(
 913                               search_terms_args.force_instant_results),
 914                           *i,
 915                           &url);
 916         break;
 917
 918       case GOOGLE_INPUT_TYPE:
 919         DCHECK(!i->is_post_param);
 920         HandleReplacement(
 921             "oit", base::IntToString(search_terms_args.input_type), *i, &url);
 922         break;
 923
 924       case GOOGLE_INSTANT_EXTENDED_ENABLED:
 925         DCHECK(!i->is_post_param);
 926         HandleReplacement(std::string(),
 927                           search_terms_data.InstantExtendedEnabledParam(
 928                               type_ == SEARCH),
 929                           *i,
 930                           &url);
 931         break;
 932
 933       case GOOGLE_NTP_IS_THEMED:
 934         DCHECK(!i->is_post_param);
 935         HandleReplacement(
 936             std::string(), search_terms_data.NTPIsThemedParam(), *i, &url);
 937         break;
 938
 939       case GOOGLE_OMNIBOX_START_MARGIN:
 940         DCHECK(!i->is_post_param);
 941         if (search_terms_args.enable_omnibox_start_margin) {
 942           int omnibox_start_margin = search_terms_data.OmniboxStartMargin();
 943           if (omnibox_start_margin >= 0) {
 944             HandleReplacement("es_sm", base::IntToString(omnibox_start_margin),
 945                               *i, &url);
 946           }
 947         }
 948         break;
 949
 950       case GOOGLE_CONTEXTUAL_SEARCH_VERSION:
 951         if (search_terms_args.contextual_search_params.version >= 0) {
 952           HandleReplacement(
 953               "ctxs",
 954               base::IntToString(
 955                   search_terms_args.contextual_search_params.version),
 956               *i,
 957               &url);
 958         }
 959         break;
 960
 961       case GOOGLE_CONTEXTUAL_SEARCH_CONTEXT_DATA: {
 962         DCHECK(!i->is_post_param);
 963         std::string context_data;
 964
 965         const SearchTermsArgs::ContextualSearchParams& params =
 966             search_terms_args.contextual_search_params;
 967
 968         if (params.start != std::string::npos) {
 969           context_data.append("ctxs_start=" + base::IntToString(
 970               params.start) + "&");
 971         }
 972
 973         if (params.end != std::string::npos) {
 974           context_data.append("ctxs_end=" + base::IntToString(
 975               params.end) + "&");
 976         }
 977
 978         if (!params.selection.empty())
 979           context_data.append("q=" + params.selection + "&");
 980
 981         if (!params.content.empty())
 982           context_data.append("ctxs_content=" + params.content + "&");
 983
 984         if (!params.base_page_url.empty())
 985           context_data.append("ctxsl_url=" + params.base_page_url + "&");
 986
 987         if (!params.encoding.empty()) {
 988           context_data.append("ctxs_encoding=" + params.encoding + "&");
 989         }
 990
 991         context_data.append(
 992             params.resolve ? "ctxsl_resolve=1" : "ctxsl_resolve=0");
 993
 994         HandleReplacement(std::string(), context_data, *i, &url);
 995         break;
 996       }
 997
 998       case GOOGLE_ORIGINAL_QUERY_FOR_SUGGESTION:
 999         DCHECK(!i->is_post_param);
1000         if (search_terms_args.accepted_suggestion >= 0 ||
1001             !search_terms_args.assisted_query_stats.empty()) {
1002           HandleReplacement(
1003               "oq", base::UTF16ToUTF8(encoded_original_query), *i, &url);
1004         }
1005         break;
1006
1007       case GOOGLE_PAGE_CLASSIFICATION:
1008         if (search_terms_args.page_classification !=
1009             metrics::OmniboxEventProto::INVALID_SPEC) {
1010           HandleReplacement(
1011               "pgcl", base::IntToString(search_terms_args.page_classification),
1012               *i, &url);
1013         }
1014         break;
1015
1016       case GOOGLE_PREFETCH_QUERY: {
1017         const std::string& query = search_terms_args.prefetch_query;
1018         const std::string& type = search_terms_args.prefetch_query_type;
1019         if (!query.empty() && !type.empty()) {
1020           HandleReplacement(
1021               std::string(), "pfq=" + query + "&qha=" + type + "&", *i, &url);
1022         }
1023         break;
1024       }
1025
1026       case GOOGLE_RLZ: {
1027         DCHECK(!i->is_post_param);
1028         // On platforms that don't have RLZ, we still want this branch
1029         // to happen so that we replace the RLZ template with the
1030         // empty string.  (If we don't handle this case, we hit a
1031         // NOTREACHED below.)
1032         base::string16 rlz_string = search_terms_data.GetRlzParameterValue(
1033             search_terms_args.from_app_list);
1034         if (!rlz_string.empty()) {
1035           HandleReplacement("rlz", base::UTF16ToUTF8(rlz_string), *i, &url);
1036         }
1037         break;
1038       }
1039
1040       case GOOGLE_SEARCH_CLIENT: {
1041         DCHECK(!i->is_post_param);
1042         std::string client = search_terms_data.GetSearchClient();
1043         if (!client.empty())
1044           HandleReplacement("client", client, *i, &url);
1045         break;
1046       }
1047
1048       case GOOGLE_SEARCH_FIELDTRIAL_GROUP:
1049         // We are not currently running any fieldtrials that modulate the search
1050         // url.  If we do, then we'd have some conditional insert such as:
1051         // url.insert(i->index, used_www ? "gcx=w&" : "gcx=c&");
1052         break;
1053
1054       case GOOGLE_SEARCH_VERSION:
1055         if (search_terms_data.EnableAnswersInSuggest())
1056           HandleReplacement("gs_rn", "42", *i, &url);
1057         break;
1058
1059       case GOOGLE_SESSION_TOKEN: {
1060         std::string token = search_terms_args.session_token;
1061         if (!token.empty())
1062           HandleReplacement("psi", token, *i, &url);
1063         break;
1064       }
1065
1066       case GOOGLE_SUGGEST_CLIENT:
1067         HandleReplacement(
1068             std::string(), search_terms_data.GetSuggestClient(), *i, &url);
1069         break;
1070
1071       case GOOGLE_SUGGEST_REQUEST_ID:
1072         HandleReplacement(
1073             std::string(), search_terms_data.GetSuggestRequestIdentifier(), *i,
1074             &url);
1075         break;
1076
1077       case GOOGLE_UNESCAPED_SEARCH_TERMS: {
1078         std::string unescaped_terms;
1079         base::UTF16ToCodepage(search_terms_args.search_terms,
1080                               input_encoding.c_str(),
1081                               base::OnStringConversionError::SKIP,
1082                               &unescaped_terms);
1083         HandleReplacement(std::string(), unescaped_terms, *i, &url);
1084         break;
1085       }
1086
1087       case LANGUAGE:
1088         HandleReplacement(
1089             std::string(), search_terms_data.GetApplicationLocale(), *i, &url);
1090         break;
1091
1092       case SEARCH_TERMS:
1093         HandleReplacement(
1094             std::string(), base::UTF16ToUTF8(encoded_terms), *i, &url);
1095         break;
1096
1097       case GOOGLE_IMAGE_THUMBNAIL:
1098         HandleReplacement(
1099             std::string(), search_terms_args.image_thumbnail_content, *i, &url);
1100         post_params_[i->index].content_type = "image/jpeg";
1101         break;
1102
1103       case GOOGLE_IMAGE_URL:
1104         if (search_terms_args.image_url.is_valid()) {
1105           HandleReplacement(
1106               std::string(), search_terms_args.image_url.spec(), *i, &url);
1107         }
1108         break;
1109
1110       case GOOGLE_IMAGE_ORIGINAL_WIDTH:
1111         if (!search_terms_args.image_original_size.IsEmpty()) {
1112           HandleReplacement(
1113               std::string(),
1114               base::IntToString(search_terms_args.image_original_size.width()),
1115               *i, &url);
1116         }
1117         break;
1118
1119       case GOOGLE_IMAGE_ORIGINAL_HEIGHT:
1120         if (!search_terms_args.image_original_size.IsEmpty()) {
1121           HandleReplacement(
1122               std::string(),
1123               base::IntToString(search_terms_args.image_original_size.height()),
1124               *i, &url);
1125         }
1126         break;
1127
1128       case GOOGLE_IMAGE_SEARCH_SOURCE:
1129         HandleReplacement(
1130             std::string(), search_terms_data.GoogleImageSearchSource(), *i,
1131             &url);
1132         break;
1133
1134       default:
1135         NOTREACHED();
1136         break;
1137     }
1138   }
1139
1140   if (!post_params_.empty())
1141     EncodeFormData(post_params_, post_content);
1142
1143   return url;
1144 }
1145
1146
1147 // TemplateURL ----------------------------------------------------------------
1148
1149 TemplateURL::AssociatedExtensionInfo::AssociatedExtensionInfo(
1150     Type type,
1151     const std::string& extension_id)
1152     : type(type),
1153       extension_id(extension_id),
1154       wants_to_be_default_engine(false) {
1155   DCHECK_NE(NORMAL, type);
1156 }
1157
1158 TemplateURL::AssociatedExtensionInfo::~AssociatedExtensionInfo() {
1159 }
1160
1161 TemplateURL::TemplateURL(const TemplateURLData& data)
1162     : data_(data),
1163       url_ref_(this, TemplateURLRef::SEARCH),
1164       suggestions_url_ref_(this,
1165                            TemplateURLRef::SUGGEST),
1166       instant_url_ref_(this,
1167                        TemplateURLRef::INSTANT),
1168       image_url_ref_(this, TemplateURLRef::IMAGE),
1169       new_tab_url_ref_(this, TemplateURLRef::NEW_TAB),
1170       contextual_search_url_ref_(this, TemplateURLRef::CONTEXTUAL_SEARCH) {
1171   SetPrepopulateId(data_.prepopulate_id);
1172
1173   if (data_.search_terms_replacement_key ==
1174       kGoogleInstantExtendedEnabledKeyFull) {
1175     data_.search_terms_replacement_key = google_util::kInstantExtendedAPIParam;
1176   }
1177 }
1178
1179 TemplateURL::~TemplateURL() {
1180 }
1181
1182 // static
1183 base::string16 TemplateURL::GenerateKeyword(const GURL& url) {
1184   DCHECK(url.is_valid());
1185   // Strip "www." off the front of the keyword; otherwise the keyword won't work
1186   // properly.  See http://code.google.com/p/chromium/issues/detail?id=6984 .
1187   // Special case: if the host was exactly "www." (not sure this can happen but
1188   // perhaps with some weird intranet and custom DNS server?), ensure we at
1189   // least don't return the empty string.
1190   base::string16 keyword(net::StripWWWFromHost(url));
1191   return keyword.empty() ? base::ASCIIToUTF16("www") : keyword;
1192 }
1193
1194 // static
1195 GURL TemplateURL::GenerateFaviconURL(const GURL& url) {
1196   DCHECK(url.is_valid());
1197   GURL::Replacements rep;
1198
1199   const char favicon_path[] = "/favicon.ico";
1200   int favicon_path_len = arraysize(favicon_path) - 1;
1201
1202   rep.SetPath(favicon_path, url::Component(0, favicon_path_len));
1203   rep.ClearUsername();
1204   rep.ClearPassword();
1205   rep.ClearQuery();
1206   rep.ClearRef();
1207   return url.ReplaceComponents(rep);
1208 }
1209
1210 // static
1211 bool TemplateURL::MatchesData(const TemplateURL* t_url,
1212                               const TemplateURLData* data,
1213                               const SearchTermsData& search_terms_data) {
1214   if (!t_url || !data)
1215     return !t_url && !data;
1216
1217   return (t_url->short_name() == data->short_name) &&
1218       t_url->HasSameKeywordAs(*data, search_terms_data) &&
1219       (t_url->url() == data->url()) &&
1220       (t_url->suggestions_url() == data->suggestions_url) &&
1221       (t_url->instant_url() == data->instant_url) &&
1222       (t_url->image_url() == data->image_url) &&
1223       (t_url->new_tab_url() == data->new_tab_url) &&
1224       (t_url->search_url_post_params() == data->search_url_post_params) &&
1225       (t_url->suggestions_url_post_params() ==
1226           data->suggestions_url_post_params) &&
1227       (t_url->instant_url_post_params() == data->instant_url_post_params) &&
1228       (t_url->image_url_post_params() == data->image_url_post_params) &&
1229       (t_url->favicon_url() == data->favicon_url) &&
1230       (t_url->safe_for_autoreplace() == data->safe_for_autoreplace) &&
1231       (t_url->show_in_default_list() == data->show_in_default_list) &&
1232       (t_url->input_encodings() == data->input_encodings) &&
1233       (t_url->alternate_urls() == data->alternate_urls) &&
1234       (t_url->search_terms_replacement_key() ==
1235           data->search_terms_replacement_key);
1236 }
1237
1238 base::string16 TemplateURL::AdjustedShortNameForLocaleDirection() const {
1239   base::string16 bidi_safe_short_name = data_.short_name;
1240   base::i18n::AdjustStringForLocaleDirection(&bidi_safe_short_name);
1241   return bidi_safe_short_name;
1242 }
1243
1244 bool TemplateURL::ShowInDefaultList(
1245     const SearchTermsData& search_terms_data) const {
1246   return data_.show_in_default_list &&
1247       url_ref_.SupportsReplacement(search_terms_data);
1248 }
1249
1250 bool TemplateURL::SupportsReplacement(
1251     const SearchTermsData& search_terms_data) const {
1252   return url_ref_.SupportsReplacement(search_terms_data);
1253 }
1254
1255 bool TemplateURL::HasGoogleBaseURLs(
1256     const SearchTermsData& search_terms_data) const {
1257   return url_ref_.HasGoogleBaseURLs(search_terms_data) ||
1258       suggestions_url_ref_.HasGoogleBaseURLs(search_terms_data) ||
1259       instant_url_ref_.HasGoogleBaseURLs(search_terms_data) ||
1260       image_url_ref_.HasGoogleBaseURLs(search_terms_data) ||
1261       new_tab_url_ref_.HasGoogleBaseURLs(search_terms_data);
1262 }
1263
1264 bool TemplateURL::IsGoogleSearchURLWithReplaceableKeyword(
1265     const SearchTermsData& search_terms_data) const {
1266   return (GetType() == NORMAL) &&
1267       url_ref_.HasGoogleBaseURLs(search_terms_data) &&
1268       google_util::IsGoogleHostname(base::UTF16ToUTF8(data_.keyword()),
1269                                     google_util::DISALLOW_SUBDOMAIN);
1270 }
1271
1272 bool TemplateURL::HasSameKeywordAs(
1273     const TemplateURLData& other,
1274     const SearchTermsData& search_terms_data) const {
1275   return (data_.keyword() == other.keyword()) ||
1276       (IsGoogleSearchURLWithReplaceableKeyword(search_terms_data) &&
1277        TemplateURL(other).IsGoogleSearchURLWithReplaceableKeyword(
1278            search_terms_data));
1279 }
1280
1281 TemplateURL::Type TemplateURL::GetType() const {
1282   return extension_info_ ? extension_info_->type : NORMAL;
1283 }
1284
1285 std::string TemplateURL::GetExtensionId() const {
1286   DCHECK(extension_info_);
1287   return extension_info_->extension_id;
1288 }
1289
1290 size_t TemplateURL::URLCount() const {
1291   // Add 1 for the regular search URL.
1292   return data_.alternate_urls.size() + 1;
1293 }
1294
1295 const std::string& TemplateURL::GetURL(size_t index) const {
1296   DCHECK_LT(index, URLCount());
1297
1298   return (index < data_.alternate_urls.size()) ?
1299       data_.alternate_urls[index] : url();
1300 }
1301
1302 bool TemplateURL::ExtractSearchTermsFromURL(
1303     const GURL& url,
1304     const SearchTermsData& search_terms_data,
1305     base::string16* search_terms) {
1306   return FindSearchTermsInURL(url, search_terms_data, search_terms, NULL, NULL);
1307 }
1308
1309 bool TemplateURL::IsSearchURL(
1310     const GURL& url,
1311     const SearchTermsData& search_terms_data) {
1312   base::string16 search_terms;
1313   return ExtractSearchTermsFromURL(url, search_terms_data, &search_terms) &&
1314       !search_terms.empty();
1315 }
1316
1317 bool TemplateURL::HasSearchTermsReplacementKey(const GURL& url) const {
1318   // Look for the key both in the query and the ref.
1319   std::string params[] = {url.query(), url.ref()};
1320
1321   for (int i = 0; i < 2; ++i) {
1322     url::Component query, key, value;
1323     query.len = static_cast<int>(params[i].size());
1324     while (url::ExtractQueryKeyValue(params[i].c_str(), &query, &key, &value)) {
1325       if (key.is_nonempty() &&
1326           params[i].substr(key.begin, key.len) ==
1327               search_terms_replacement_key()) {
1328         return true;
1329       }
1330     }
1331   }
1332   return false;
1333 }
1334
1335 bool TemplateURL::ReplaceSearchTermsInURL(
1336     const GURL& url,
1337     const TemplateURLRef::SearchTermsArgs& search_terms_args,
1338     const SearchTermsData& search_terms_data,
1339     GURL* result) {
1340   // TODO(beaudoin): Use AQS from |search_terms_args| too.
1341   url::Parsed::ComponentType search_term_component;
1342   url::Component search_terms_position;
1343   base::string16 search_terms;
1344   if (!FindSearchTermsInURL(url, search_terms_data, &search_terms,
1345                             &search_term_component, &search_terms_position)) {
1346     return false;
1347   }
1348   DCHECK(search_terms_position.is_nonempty());
1349
1350   // FindSearchTermsInURL only returns true for search terms in the query or
1351   // ref, so we can call EncodeSearchTerm with |is_in_query| = true, since query
1352   // and ref are encoded in the same way.
1353   std::string input_encoding;
1354   base::string16 encoded_terms;
1355   base::string16 encoded_original_query;
1356   EncodeSearchTerms(search_terms_args, true, &input_encoding,
1357                     &encoded_terms, &encoded_original_query);
1358
1359   std::string old_params(
1360       (search_term_component == url::Parsed::REF) ? url.ref() : url.query());
1361   std::string new_params(old_params, 0, search_terms_position.begin);
1362   new_params += base::UTF16ToUTF8(search_terms_args.search_terms);
1363   new_params += old_params.substr(search_terms_position.end());
1364   GURL::Replacements replacements;
1365   if (search_term_component == url::Parsed::REF)
1366     replacements.SetRefStr(new_params);
1367   else
1368     replacements.SetQueryStr(new_params);
1369   *result = url.ReplaceComponents(replacements);
1370   return true;
1371 }
1372
1373 void TemplateURL::EncodeSearchTerms(
1374     const TemplateURLRef::SearchTermsArgs& search_terms_args,
1375     bool is_in_query,
1376     std::string* input_encoding,
1377     base::string16* encoded_terms,
1378     base::string16* encoded_original_query) const {
1379
1380   std::vector<std::string> encodings(input_encodings());
1381   if (std::find(encodings.begin(), encodings.end(), "UTF-8") == encodings.end())
1382     encodings.push_back("UTF-8");
1383   for (std::vector<std::string>::const_iterator i(encodings.begin());
1384        i != encodings.end(); ++i) {
1385     if (TryEncoding(search_terms_args.search_terms,
1386                     search_terms_args.original_query, i->c_str(),
1387                     is_in_query, encoded_terms, encoded_original_query)) {
1388       *input_encoding = *i;
1389       return;
1390     }
1391   }
1392   NOTREACHED();
1393 }
1394
1395 GURL TemplateURL::GenerateSearchURL(
1396     const SearchTermsData& search_terms_data) const {
1397   if (!url_ref_.IsValid(search_terms_data))
1398     return GURL();
1399
1400   if (!url_ref_.SupportsReplacement(search_terms_data))
1401     return GURL(url());
1402
1403   // Use something obscure for the search terms argument so that in the rare
1404   // case the term replaces the URL it's unlikely another keyword would have the
1405   // same url.
1406   // TODO(jnd): Add additional parameters to get post data when the search URL
1407   // has post parameters.
1408   return GURL(url_ref_.ReplaceSearchTerms(
1409       TemplateURLRef::SearchTermsArgs(
1410           base::ASCIIToUTF16("blah.blah.blah.blah.blah")),
1411       search_terms_data, NULL));
1412 }
1413
1414 void TemplateURL::CopyFrom(const TemplateURL& other) {
1415   if (this == &other)
1416     return;
1417
1418   data_ = other.data_;
1419   url_ref_.InvalidateCachedValues();
1420   suggestions_url_ref_.InvalidateCachedValues();
1421   instant_url_ref_.InvalidateCachedValues();
1422   SetPrepopulateId(other.data_.prepopulate_id);
1423 }
1424
1425 void TemplateURL::SetURL(const std::string& url) {
1426   data_.SetURL(url);
1427   url_ref_.InvalidateCachedValues();
1428 }
1429
1430 void TemplateURL::SetPrepopulateId(int id) {
1431   data_.prepopulate_id = id;
1432   const bool prepopulated = id > 0;
1433   url_ref_.prepopulated_ = prepopulated;
1434   suggestions_url_ref_.prepopulated_ = prepopulated;
1435   instant_url_ref_.prepopulated_ = prepopulated;
1436 }
1437
1438 void TemplateURL::ResetKeywordIfNecessary(
1439     const SearchTermsData& search_terms_data,
1440     bool force) {
1441   if (IsGoogleSearchURLWithReplaceableKeyword(search_terms_data) || force) {
1442     DCHECK(GetType() != OMNIBOX_API_EXTENSION);
1443     GURL url(GenerateSearchURL(search_terms_data));
1444     if (url.is_valid())
1445       data_.SetKeyword(GenerateKeyword(url));
1446   }
1447 }
1448
1449 bool TemplateURL::FindSearchTermsInURL(
1450     const GURL& url,
1451     const SearchTermsData& search_terms_data,
1452     base::string16* search_terms,
1453     url::Parsed::ComponentType* search_term_component,
1454     url::Component* search_terms_position) {
1455   DCHECK(search_terms);
1456   search_terms->clear();
1457
1458   // Try to match with every pattern.
1459   for (size_t i = 0; i < URLCount(); ++i) {
1460     TemplateURLRef ref(this, i);
1461     if (ref.ExtractSearchTermsFromURL(url, search_terms, search_terms_data,
1462         search_term_component, search_terms_position)) {
1463       // If ExtractSearchTermsFromURL() returns true and |search_terms| is empty
1464       // it means the pattern matched but no search terms were present. In this
1465       // case we fail immediately without looking for matches in subsequent
1466       // patterns. This means that given patterns
1467       //    [ "http://foo/#q={searchTerms}", "http://foo/?q={searchTerms}" ],
1468       // calling ExtractSearchTermsFromURL() on "http://foo/?q=bar#q=' would
1469       // return false. This is important for at least Google, where such URLs
1470       // are invalid.
1471       return !search_terms->empty();
1472     }
1473   }
1474   return false;
1475 }