Cleanup: Remove some unneeded UTF8 to UTF16 conversions.
[chromium-blink-merge.git] / components / search_engines / template_url.cc
blobf95acb7c1f6fce7431c4a264073d8d03e369ad58
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.
5 #include "components/search_engines/template_url.h"
7 #include <string>
8 #include <vector>
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"
33 namespace {
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 = '?';
43 // Known parameters found in the URL.
44 const char kSearchTermsParameter[] = "searchTerms";
45 const char kSearchTermsParameterFull[] = "{searchTerms}";
46 const char kSearchTermsParameterFullEscaped[] = "%7BsearchTerms%7D";
47 const char kCountParameter[] = "count";
48 const char kStartIndexParameter[] = "startIndex";
49 const char kStartPageParameter[] = "startPage";
50 const char kLanguageParameter[] = "language";
51 const char kInputEncodingParameter[] = "inputEncoding";
52 const char kOutputEncodingParameter[] = "outputEncoding";
54 const char kGoogleAssistedQueryStatsParameter[] = "google:assistedQueryStats";
56 // Host/Domain Google searches are relative to.
57 const char kGoogleBaseURLParameter[] = "google:baseURL";
58 const char kGoogleBaseURLParameterFull[] = "{google:baseURL}";
60 // Like google:baseURL, but for the Search Suggest capability.
61 const char kGoogleBaseSuggestURLParameter[] = "google:baseSuggestURL";
62 const char kGoogleBaseSuggestURLParameterFull[] = "{google:baseSuggestURL}";
63 const char kGoogleBookmarkBarPinnedParameter[] = "google:bookmarkBarPinned";
64 const char kGoogleContextualSearchContextData[] =
65 "google:contextualSearchContextData";
66 const char kGoogleContextualSearchVersion[] = "google:contextualSearchVersion";
67 const char kGoogleCurrentPageUrlParameter[] = "google:currentPageUrl";
68 const char kGoogleCursorPositionParameter[] = "google:cursorPosition";
69 const char kGoogleForceInstantResultsParameter[] = "google:forceInstantResults";
70 const char kGoogleImageSearchSource[] = "google:imageSearchSource";
71 const char kGoogleImageThumbnailParameter[] = "google:imageThumbnail";
72 const char kGoogleImageOriginalWidth[] = "google:imageOriginalWidth";
73 const char kGoogleImageOriginalHeight[] = "google:imageOriginalHeight";
74 const char kGoogleImageURLParameter[] = "google:imageURL";
75 const char kGoogleInputTypeParameter[] = "google:inputType";
76 const char kGoogleInstantExtendedEnabledParameter[] =
77 "google:instantExtendedEnabledParameter";
78 const char kGoogleInstantExtendedEnabledKey[] =
79 "google:instantExtendedEnabledKey";
80 const char kGoogleInstantExtendedEnabledKeyFull[] =
81 "{google:instantExtendedEnabledKey}";
82 const char kGoogleNTPIsThemedParameter[] = "google:ntpIsThemedParameter";
83 const char kGoogleOmniboxStartMarginParameter[] =
84 "google:omniboxStartMarginParameter";
85 const char kGoogleOriginalQueryForSuggestionParameter[] =
86 "google:originalQueryForSuggestion";
87 const char kGooglePageClassificationParameter[] = "google:pageClassification";
88 const char kGooglePrefetchQuery[] = "google:prefetchQuery";
89 const char kGoogleRLZParameter[] = "google:RLZ";
90 const char kGoogleSearchClient[] = "google:searchClient";
91 const char kGoogleSearchFieldtrialParameter[] =
92 "google:searchFieldtrialParameter";
93 const char kGoogleSearchVersion[] = "google:searchVersion";
94 const char kGoogleSessionToken[] = "google:sessionToken";
95 const char kGoogleSourceIdParameter[] = "google:sourceId";
96 const char kGoogleSuggestAPIKeyParameter[] = "google:suggestAPIKeyParameter";
97 const char kGoogleSuggestClient[] = "google:suggestClient";
98 const char kGoogleSuggestRequestId[] = "google:suggestRid";
100 // Same as kSearchTermsParameter, with no escaping.
101 const char kGoogleUnescapedSearchTermsParameter[] =
102 "google:unescapedSearchTerms";
103 const char kGoogleUnescapedSearchTermsParameterFull[] =
104 "{google:unescapedSearchTerms}";
106 // Display value for kSearchTermsParameter.
107 const char kDisplaySearchTerms[] = "%s";
109 // Display value for kGoogleUnescapedSearchTermsParameter.
110 const char kDisplayUnescapedSearchTerms[] = "%S";
112 // Used if the count parameter is not optional. Indicates we want 10 search
113 // results.
114 const char kDefaultCount[] = "10";
116 // Used if the parameter kOutputEncodingParameter is required.
117 const char kOutputEncodingType[] = "UTF-8";
119 // Attempts to encode |terms| and |original_query| in |encoding| and escape
120 // them. |terms| may be escaped as path or query depending on |is_in_query|;
121 // |original_query| is always escaped as query. Returns whether the encoding
122 // process succeeded.
123 bool TryEncoding(const base::string16& terms,
124 const base::string16& original_query,
125 const char* encoding,
126 bool is_in_query,
127 base::string16* escaped_terms,
128 base::string16* escaped_original_query) {
129 DCHECK(escaped_terms);
130 DCHECK(escaped_original_query);
131 std::string encoded_terms;
132 if (!base::UTF16ToCodepage(terms, encoding,
133 base::OnStringConversionError::SKIP, &encoded_terms))
134 return false;
135 *escaped_terms = base::UTF8ToUTF16(is_in_query ?
136 net::EscapeQueryParamValue(encoded_terms, true) :
137 net::EscapePath(encoded_terms));
138 if (original_query.empty())
139 return true;
140 std::string encoded_original_query;
141 if (!base::UTF16ToCodepage(original_query, encoding,
142 base::OnStringConversionError::SKIP, &encoded_original_query))
143 return false;
144 *escaped_original_query = base::UTF8ToUTF16(
145 net::EscapeQueryParamValue(encoded_original_query, true));
146 return true;
149 // Extract query key and host given a list of parameters coming from the URL
150 // query or ref.
151 std::string FindSearchTermsKey(const std::string& params) {
152 if (params.empty())
153 return std::string();
154 url::Component query, key, value;
155 query.len = static_cast<int>(params.size());
156 while (url::ExtractQueryKeyValue(params.c_str(), &query, &key, &value)) {
157 if (key.is_nonempty() && value.is_nonempty()) {
158 std::string value_string = params.substr(value.begin, value.len);
159 if (value_string.find(kSearchTermsParameterFull, 0) !=
160 std::string::npos ||
161 value_string.find(kGoogleUnescapedSearchTermsParameterFull, 0) !=
162 std::string::npos) {
163 return params.substr(key.begin, key.len);
167 return std::string();
170 // Extract the position of the search terms' parameter in the URL path.
171 bool FindSearchTermsInPath(const std::string& path,
172 url::Component* parameter_position) {
173 DCHECK(parameter_position);
174 parameter_position->reset();
175 const size_t begin = path.find(kSearchTermsParameterFullEscaped);
176 if (begin == std::string::npos)
177 return false;
178 parameter_position->begin = begin;
179 parameter_position->len = arraysize(kSearchTermsParameterFullEscaped) - 1;
180 return true;
183 bool IsTemplateParameterString(const std::string& param) {
184 return (param.length() > 2) && (*(param.begin()) == kStartParameter) &&
185 (*(param.rbegin()) == kEndParameter);
188 } // namespace
191 // TemplateURLRef::SearchTermsArgs --------------------------------------------
193 TemplateURLRef::SearchTermsArgs::SearchTermsArgs(
194 const base::string16& search_terms)
195 : search_terms(search_terms),
196 input_type(metrics::OmniboxInputType::INVALID),
197 accepted_suggestion(NO_SUGGESTIONS_AVAILABLE),
198 cursor_position(base::string16::npos),
199 enable_omnibox_start_margin(false),
200 page_classification(metrics::OmniboxEventProto::INVALID_SPEC),
201 bookmark_bar_pinned(false),
202 append_extra_query_params(false),
203 force_instant_results(false),
204 from_app_list(false),
205 contextual_search_params(ContextualSearchParams()) {
208 TemplateURLRef::SearchTermsArgs::~SearchTermsArgs() {
211 TemplateURLRef::SearchTermsArgs::ContextualSearchParams::
212 ContextualSearchParams()
213 : version(-1),
214 start(base::string16::npos),
215 end(base::string16::npos),
216 resolve(true) {
219 TemplateURLRef::SearchTermsArgs::ContextualSearchParams::
220 ContextualSearchParams(
221 const int version,
222 const std::string& selection,
223 const std::string& base_page_url,
224 const bool resolve)
225 : version(version),
226 start(base::string16::npos),
227 end(base::string16::npos),
228 selection(selection),
229 base_page_url(base_page_url),
230 resolve(resolve) {
233 TemplateURLRef::SearchTermsArgs::ContextualSearchParams::
234 ContextualSearchParams(
235 const int version,
236 const size_t start,
237 const size_t end,
238 const std::string& selection,
239 const std::string& content,
240 const std::string& base_page_url,
241 const std::string& encoding,
242 const bool resolve)
243 : version(version),
244 start(start),
245 end(end),
246 selection(selection),
247 content(content),
248 base_page_url(base_page_url),
249 encoding(encoding),
250 resolve(resolve) {
253 TemplateURLRef::SearchTermsArgs::ContextualSearchParams::
254 ~ContextualSearchParams() {
257 // TemplateURLRef -------------------------------------------------------------
259 TemplateURLRef::TemplateURLRef(TemplateURL* owner, Type type)
260 : owner_(owner),
261 type_(type),
262 index_in_owner_(0),
263 parsed_(false),
264 valid_(false),
265 supports_replacements_(false),
266 search_term_position_in_path_(std::string::npos),
267 search_term_key_location_(url::Parsed::QUERY),
268 prepopulated_(false) {
269 DCHECK(owner_);
270 DCHECK_NE(INDEXED, type_);
273 TemplateURLRef::TemplateURLRef(TemplateURL* owner, size_t index_in_owner)
274 : owner_(owner),
275 type_(INDEXED),
276 index_in_owner_(index_in_owner),
277 parsed_(false),
278 valid_(false),
279 supports_replacements_(false),
280 search_term_position_in_path_(std::string::npos),
281 search_term_key_location_(url::Parsed::QUERY),
282 prepopulated_(false) {
283 DCHECK(owner_);
284 DCHECK_LT(index_in_owner_, owner_->URLCount());
287 TemplateURLRef::~TemplateURLRef() {
290 std::string TemplateURLRef::GetURL() const {
291 switch (type_) {
292 case SEARCH: return owner_->url();
293 case SUGGEST: return owner_->suggestions_url();
294 case INSTANT: return owner_->instant_url();
295 case IMAGE: return owner_->image_url();
296 case NEW_TAB: return owner_->new_tab_url();
297 case CONTEXTUAL_SEARCH: return owner_->contextual_search_url();
298 case INDEXED: return owner_->GetURL(index_in_owner_);
299 default: NOTREACHED(); return std::string(); // NOLINT
303 std::string TemplateURLRef::GetPostParamsString() const {
304 switch (type_) {
305 case INDEXED:
306 case SEARCH: return owner_->search_url_post_params();
307 case SUGGEST: return owner_->suggestions_url_post_params();
308 case INSTANT: return owner_->instant_url_post_params();
309 case NEW_TAB: return std::string();
310 case CONTEXTUAL_SEARCH: return std::string();
311 case IMAGE: return owner_->image_url_post_params();
312 default: NOTREACHED(); return std::string(); // NOLINT
316 bool TemplateURLRef::UsesPOSTMethod(
317 const SearchTermsData& search_terms_data) const {
318 ParseIfNecessary(search_terms_data);
319 return !post_params_.empty();
322 bool TemplateURLRef::EncodeFormData(const PostParams& post_params,
323 PostContent* post_content) const {
324 if (post_params.empty())
325 return true;
326 if (!post_content)
327 return false;
329 const char kUploadDataMIMEType[] = "multipart/form-data; boundary=";
330 const char kMultipartBoundary[] = "----+*+----%016" PRIx64 "----+*+----";
331 // Each name/value pair is stored in a body part which is preceded by a
332 // boundary delimiter line. Uses random number generator here to create
333 // a unique boundary delimiter for form data encoding.
334 std::string boundary = base::StringPrintf(kMultipartBoundary,
335 base::RandUint64());
336 // Sets the content MIME type.
337 post_content->first = kUploadDataMIMEType;
338 post_content->first += boundary;
339 // Encodes the post parameters.
340 std::string* post_data = &post_content->second;
341 post_data->clear();
342 for (const auto& param : post_params) {
343 DCHECK(!param.name.empty());
344 net::AddMultipartValueForUpload(param.name, param.value, boundary,
345 param.content_type, post_data);
347 net::AddMultipartFinalDelimiterForUpload(boundary, post_data);
348 return true;
351 bool TemplateURLRef::SupportsReplacement(
352 const SearchTermsData& search_terms_data) const {
353 ParseIfNecessary(search_terms_data);
354 return valid_ && supports_replacements_;
357 std::string TemplateURLRef::ReplaceSearchTerms(
358 const SearchTermsArgs& search_terms_args,
359 const SearchTermsData& search_terms_data,
360 PostContent* post_content) const {
361 ParseIfNecessary(search_terms_data);
362 if (!valid_)
363 return std::string();
365 std::string url(HandleReplacements(search_terms_args, search_terms_data,
366 post_content));
368 GURL gurl(url);
369 if (!gurl.is_valid())
370 return url;
372 std::vector<std::string> query_params;
373 if (search_terms_args.append_extra_query_params) {
374 std::string extra_params(
375 base::CommandLine::ForCurrentProcess()->GetSwitchValueASCII(
376 switches::kExtraSearchQueryParams));
377 if (!extra_params.empty())
378 query_params.push_back(extra_params);
380 if (!search_terms_args.suggest_query_params.empty())
381 query_params.push_back(search_terms_args.suggest_query_params);
382 if (!gurl.query().empty())
383 query_params.push_back(gurl.query());
385 if (query_params.empty())
386 return url;
388 GURL::Replacements replacements;
389 std::string query_str = JoinString(query_params, "&");
390 replacements.SetQueryStr(query_str);
391 return gurl.ReplaceComponents(replacements).possibly_invalid_spec();
394 bool TemplateURLRef::IsValid(const SearchTermsData& search_terms_data) const {
395 ParseIfNecessary(search_terms_data);
396 return valid_;
399 base::string16 TemplateURLRef::DisplayURL(
400 const SearchTermsData& search_terms_data) const {
401 ParseIfNecessary(search_terms_data);
402 std::string result(GetURL());
403 if (valid_ && !replacements_.empty()) {
404 ReplaceSubstringsAfterOffset(&result, 0,
405 kSearchTermsParameterFull,
406 kDisplaySearchTerms);
407 ReplaceSubstringsAfterOffset(&result, 0,
408 kGoogleUnescapedSearchTermsParameterFull,
409 kDisplayUnescapedSearchTerms);
411 return base::UTF8ToUTF16(result);
414 // static
415 std::string TemplateURLRef::DisplayURLToURLRef(
416 const base::string16& display_url) {
417 std::string result = base::UTF16ToUTF8(display_url);
418 ReplaceSubstringsAfterOffset(&result, 0,
419 kDisplaySearchTerms,
420 kSearchTermsParameterFull);
421 ReplaceSubstringsAfterOffset(&result, 0,
422 kDisplayUnescapedSearchTerms,
423 kGoogleUnescapedSearchTermsParameterFull);
424 return result;
427 const std::string& TemplateURLRef::GetHost(
428 const SearchTermsData& search_terms_data) const {
429 ParseIfNecessary(search_terms_data);
430 return host_;
433 const std::string& TemplateURLRef::GetPath(
434 const SearchTermsData& search_terms_data) const {
435 ParseIfNecessary(search_terms_data);
436 return path_;
439 const std::string& TemplateURLRef::GetSearchTermKey(
440 const SearchTermsData& search_terms_data) const {
441 ParseIfNecessary(search_terms_data);
442 return search_term_key_;
445 size_t TemplateURLRef::GetSearchTermPositionInPath(
446 const SearchTermsData& search_terms_data) const {
447 ParseIfNecessary(search_terms_data);
448 return search_term_position_in_path_;
451 url::Parsed::ComponentType TemplateURLRef::GetSearchTermKeyLocation(
452 const SearchTermsData& search_terms_data) const {
453 ParseIfNecessary(search_terms_data);
454 return search_term_key_location_;
457 base::string16 TemplateURLRef::SearchTermToString16(
458 const std::string& term) const {
459 const std::vector<std::string>& encodings = owner_->input_encodings();
460 base::string16 result;
462 std::string unescaped = net::UnescapeURLComponent(
463 term,
464 net::UnescapeRule::REPLACE_PLUS_WITH_SPACE |
465 net::UnescapeRule::URL_SPECIAL_CHARS);
466 for (size_t i = 0; i < encodings.size(); ++i) {
467 if (base::CodepageToUTF16(unescaped, encodings[i].c_str(),
468 base::OnStringConversionError::FAIL, &result))
469 return result;
472 // Always fall back on UTF-8 if it works.
473 if (base::CodepageToUTF16(unescaped, base::kCodepageUTF8,
474 base::OnStringConversionError::FAIL, &result))
475 return result;
477 // When nothing worked, just use the escaped text. We have no idea what the
478 // encoding is. We need to substitute spaces for pluses ourselves since we're
479 // not sending it through an unescaper.
480 result = base::UTF8ToUTF16(term);
481 std::replace(result.begin(), result.end(), '+', ' ');
482 return result;
485 bool TemplateURLRef::HasGoogleBaseURLs(
486 const SearchTermsData& search_terms_data) const {
487 ParseIfNecessary(search_terms_data);
488 for (size_t i = 0; i < replacements_.size(); ++i) {
489 if ((replacements_[i].type == GOOGLE_BASE_URL) ||
490 (replacements_[i].type == GOOGLE_BASE_SUGGEST_URL))
491 return true;
493 return false;
496 bool TemplateURLRef::ExtractSearchTermsFromURL(
497 const GURL& url,
498 base::string16* search_terms,
499 const SearchTermsData& search_terms_data,
500 url::Parsed::ComponentType* search_terms_component,
501 url::Component* search_terms_position) const {
502 DCHECK(search_terms);
503 search_terms->clear();
505 ParseIfNecessary(search_terms_data);
507 // We need a search term in the template URL to extract something.
508 if (search_term_key_.empty() &&
509 (search_term_key_location_ != url::Parsed::PATH))
510 return false;
512 // Fill-in the replacements. We don't care about search terms in the pattern,
513 // so we use the empty string.
514 // Currently we assume the search term only shows in URL, not in post params.
515 GURL pattern(ReplaceSearchTerms(SearchTermsArgs(base::string16()),
516 search_terms_data, NULL));
517 // Host, path and port must match.
518 if ((url.port() != pattern.port()) ||
519 (url.host() != host_) ||
520 ((url.path() != path_) &&
521 (search_term_key_location_ != url::Parsed::PATH))) {
522 return false;
525 std::string source;
526 url::Component position;
527 net::UnescapeRule::Type unescape_rules =
528 net::UnescapeRule::SPACES | net::UnescapeRule::URL_SPECIAL_CHARS;
530 if (search_term_key_location_ == url::Parsed::PATH) {
531 source = url.path();
533 // Characters in the path before and after search terms must match.
534 if (source.length() < path_.length())
535 return false;
536 position.begin = search_term_position_in_path_;
537 position.len = source.length() - path_.length();
538 if (source.substr(0, position.begin) + source.substr(position.end()) !=
539 path_)
540 return false;
541 } else {
542 DCHECK(search_term_key_location_ == url::Parsed::QUERY ||
543 search_term_key_location_ == url::Parsed::REF);
544 source = (search_term_key_location_ == url::Parsed::QUERY) ?
545 url.query() : url.ref();
547 url::Component query, key, value;
548 query.len = static_cast<int>(source.size());
549 bool key_found = false;
550 while (url::ExtractQueryKeyValue(source.c_str(), &query, &key, &value)) {
551 if (key.is_nonempty()) {
552 if (source.substr(key.begin, key.len) == search_term_key_) {
553 // Fail if search term key is found twice.
554 if (key_found)
555 return false;
556 key_found = true;
557 position = value;
561 if (!key_found)
562 return false;
563 unescape_rules |= net::UnescapeRule::REPLACE_PLUS_WITH_SPACE;
566 // Extract the search term.
567 *search_terms = net::UnescapeAndDecodeUTF8URLComponent(
568 source.substr(position.begin, position.len), unescape_rules);
569 if (search_terms_component)
570 *search_terms_component = search_term_key_location_;
571 if (search_terms_position)
572 *search_terms_position = position;
573 return true;
576 void TemplateURLRef::InvalidateCachedValues() const {
577 supports_replacements_ = valid_ = parsed_ = false;
578 host_.clear();
579 path_.clear();
580 search_term_key_.clear();
581 search_term_position_in_path_ = std::string::npos;
582 search_term_key_location_ = url::Parsed::QUERY;
583 replacements_.clear();
584 post_params_.clear();
587 bool TemplateURLRef::ParseParameter(size_t start,
588 size_t end,
589 std::string* url,
590 Replacements* replacements) const {
591 DCHECK(start != std::string::npos &&
592 end != std::string::npos && end > start);
593 size_t length = end - start - 1;
594 bool optional = false;
595 if ((*url)[end - 1] == kOptional) {
596 optional = true;
597 length--;
599 std::string parameter(url->substr(start + 1, length));
600 std::string full_parameter(url->substr(start, end - start + 1));
601 // Remove the parameter from the string. For parameters who replacement is
602 // constant and already known, just replace them directly. For other cases,
603 // like parameters whose values may change over time, use |replacements|.
604 url->erase(start, end - start + 1);
605 if (parameter == kSearchTermsParameter) {
606 replacements->push_back(Replacement(SEARCH_TERMS, start));
607 } else if (parameter == kCountParameter) {
608 if (!optional)
609 url->insert(start, kDefaultCount);
610 } else if (parameter == kGoogleAssistedQueryStatsParameter) {
611 replacements->push_back(Replacement(GOOGLE_ASSISTED_QUERY_STATS, start));
612 } else if (parameter == kGoogleBaseURLParameter) {
613 replacements->push_back(Replacement(GOOGLE_BASE_URL, start));
614 } else if (parameter == kGoogleBaseSuggestURLParameter) {
615 replacements->push_back(Replacement(GOOGLE_BASE_SUGGEST_URL, start));
616 } else if (parameter == kGoogleBookmarkBarPinnedParameter) {
617 replacements->push_back(Replacement(GOOGLE_BOOKMARK_BAR_PINNED, start));
618 } else if (parameter == kGoogleCurrentPageUrlParameter) {
619 replacements->push_back(Replacement(GOOGLE_CURRENT_PAGE_URL, start));
620 } else if (parameter == kGoogleCursorPositionParameter) {
621 replacements->push_back(Replacement(GOOGLE_CURSOR_POSITION, start));
622 } else if (parameter == kGoogleForceInstantResultsParameter) {
623 replacements->push_back(Replacement(GOOGLE_FORCE_INSTANT_RESULTS, start));
624 } else if (parameter == kGoogleImageOriginalHeight) {
625 replacements->push_back(
626 Replacement(TemplateURLRef::GOOGLE_IMAGE_ORIGINAL_HEIGHT, start));
627 } else if (parameter == kGoogleImageOriginalWidth) {
628 replacements->push_back(
629 Replacement(TemplateURLRef::GOOGLE_IMAGE_ORIGINAL_WIDTH, start));
630 } else if (parameter == kGoogleImageSearchSource) {
631 replacements->push_back(
632 Replacement(TemplateURLRef::GOOGLE_IMAGE_SEARCH_SOURCE, start));
633 } else if (parameter == kGoogleImageThumbnailParameter) {
634 replacements->push_back(
635 Replacement(TemplateURLRef::GOOGLE_IMAGE_THUMBNAIL, start));
636 } else if (parameter == kGoogleImageURLParameter) {
637 replacements->push_back(Replacement(TemplateURLRef::GOOGLE_IMAGE_URL,
638 start));
639 } else if (parameter == kGoogleInputTypeParameter) {
640 replacements->push_back(Replacement(TemplateURLRef::GOOGLE_INPUT_TYPE,
641 start));
642 } else if (parameter == kGoogleInstantExtendedEnabledParameter) {
643 replacements->push_back(Replacement(GOOGLE_INSTANT_EXTENDED_ENABLED,
644 start));
645 } else if (parameter == kGoogleInstantExtendedEnabledKey) {
646 url->insert(start, google_util::kInstantExtendedAPIParam);
647 } else if (parameter == kGoogleNTPIsThemedParameter) {
648 replacements->push_back(Replacement(GOOGLE_NTP_IS_THEMED, start));
649 } else if (parameter == kGoogleOmniboxStartMarginParameter) {
650 replacements->push_back(Replacement(GOOGLE_OMNIBOX_START_MARGIN, start));
651 } else if (parameter == kGoogleContextualSearchVersion) {
652 replacements->push_back(
653 Replacement(GOOGLE_CONTEXTUAL_SEARCH_VERSION, start));
654 } else if (parameter == kGoogleContextualSearchContextData) {
655 replacements->push_back(
656 Replacement(GOOGLE_CONTEXTUAL_SEARCH_CONTEXT_DATA, start));
657 } else if (parameter == kGoogleOriginalQueryForSuggestionParameter) {
658 replacements->push_back(Replacement(GOOGLE_ORIGINAL_QUERY_FOR_SUGGESTION,
659 start));
660 } else if (parameter == kGooglePageClassificationParameter) {
661 replacements->push_back(Replacement(GOOGLE_PAGE_CLASSIFICATION, start));
662 } else if (parameter == kGooglePrefetchQuery) {
663 replacements->push_back(Replacement(GOOGLE_PREFETCH_QUERY, start));
664 } else if (parameter == kGoogleRLZParameter) {
665 replacements->push_back(Replacement(GOOGLE_RLZ, start));
666 } else if (parameter == kGoogleSearchClient) {
667 replacements->push_back(Replacement(GOOGLE_SEARCH_CLIENT, start));
668 } else if (parameter == kGoogleSearchFieldtrialParameter) {
669 replacements->push_back(Replacement(GOOGLE_SEARCH_FIELDTRIAL_GROUP, start));
670 } else if (parameter == kGoogleSearchVersion) {
671 replacements->push_back(Replacement(GOOGLE_SEARCH_VERSION, start));
672 } else if (parameter == kGoogleSessionToken) {
673 replacements->push_back(Replacement(GOOGLE_SESSION_TOKEN, start));
674 } else if (parameter == kGoogleSourceIdParameter) {
675 #if defined(OS_ANDROID)
676 url->insert(start, "sourceid=chrome-mobile&");
677 #else
678 url->insert(start, "sourceid=chrome&");
679 #endif
680 } else if (parameter == kGoogleSuggestAPIKeyParameter) {
681 url->insert(start,
682 net::EscapeQueryParamValue(google_apis::GetAPIKey(), false));
683 } else if (parameter == kGoogleSuggestClient) {
684 replacements->push_back(Replacement(GOOGLE_SUGGEST_CLIENT, start));
685 } else if (parameter == kGoogleSuggestRequestId) {
686 replacements->push_back(Replacement(GOOGLE_SUGGEST_REQUEST_ID, start));
687 } else if (parameter == kGoogleUnescapedSearchTermsParameter) {
688 replacements->push_back(Replacement(GOOGLE_UNESCAPED_SEARCH_TERMS, start));
689 } else if (parameter == kInputEncodingParameter) {
690 replacements->push_back(Replacement(ENCODING, start));
691 } else if (parameter == kLanguageParameter) {
692 replacements->push_back(Replacement(LANGUAGE, start));
693 } else if (parameter == kOutputEncodingParameter) {
694 if (!optional)
695 url->insert(start, kOutputEncodingType);
696 } else if ((parameter == kStartIndexParameter) ||
697 (parameter == kStartPageParameter)) {
698 // We don't support these.
699 if (!optional)
700 url->insert(start, "1");
701 } else if (!prepopulated_) {
702 // If it's a prepopulated URL, we know that it's safe to remove unknown
703 // parameters, so just ignore this and return true below. Otherwise it could
704 // be some garbage but can also be a javascript block. Put it back.
705 url->insert(start, full_parameter);
706 return false;
708 return true;
711 std::string TemplateURLRef::ParseURL(const std::string& url,
712 Replacements* replacements,
713 PostParams* post_params,
714 bool* valid) const {
715 *valid = false;
716 std::string parsed_url = url;
717 for (size_t last = 0; last != std::string::npos; ) {
718 last = parsed_url.find(kStartParameter, last);
719 if (last != std::string::npos) {
720 size_t template_end = parsed_url.find(kEndParameter, last);
721 if (template_end != std::string::npos) {
722 // Since we allow Javascript in the URL, {} pairs could be nested. Match
723 // only leaf pairs with supported parameters.
724 size_t next_template_start = parsed_url.find(kStartParameter, last + 1);
725 if (next_template_start == std::string::npos ||
726 next_template_start > template_end) {
727 // If successful, ParseParameter erases from the string as such no
728 // need to update |last|. If failed, move |last| to the end of pair.
729 if (!ParseParameter(last, template_end, &parsed_url, replacements)) {
730 // |template_end| + 1 may be beyond the end of the string.
731 last = template_end;
733 } else {
734 last = next_template_start;
736 } else {
737 // Open brace without a closing brace, return.
738 return std::string();
743 // Handles the post parameters.
744 const std::string& post_params_string = GetPostParamsString();
745 if (!post_params_string.empty()) {
746 typedef std::vector<std::string> Strings;
747 Strings param_list;
748 base::SplitString(post_params_string, ',', &param_list);
750 for (Strings::const_iterator iterator = param_list.begin();
751 iterator != param_list.end(); ++iterator) {
752 Strings parts;
753 // The '=' delimiter is required and the name must be not empty.
754 base::SplitString(*iterator, '=', &parts);
755 if ((parts.size() != 2U) || parts[0].empty())
756 return std::string();
758 std::string& value = parts[1];
759 size_t replacements_size = replacements->size();
760 if (IsTemplateParameterString(value))
761 ParseParameter(0, value.length() - 1, &value, replacements);
762 PostParam param = { parts[0], value };
763 post_params->push_back(param);
764 // If there was a replacement added, points its index to last added
765 // PostParam.
766 if (replacements->size() > replacements_size) {
767 DCHECK_EQ(replacements_size + 1, replacements->size());
768 Replacement* r = &replacements->back();
769 r->is_post_param = true;
770 r->index = post_params->size() - 1;
773 DCHECK(!post_params->empty());
776 *valid = true;
777 return parsed_url;
780 void TemplateURLRef::ParseIfNecessary(
781 const SearchTermsData& search_terms_data) const {
782 if (!parsed_) {
783 InvalidateCachedValues();
784 parsed_ = true;
785 parsed_url_ = ParseURL(GetURL(), &replacements_, &post_params_, &valid_);
786 supports_replacements_ = false;
787 if (valid_) {
788 bool has_only_one_search_term = false;
789 for (Replacements::const_iterator i = replacements_.begin();
790 i != replacements_.end(); ++i) {
791 if ((i->type == SEARCH_TERMS) ||
792 (i->type == GOOGLE_UNESCAPED_SEARCH_TERMS)) {
793 if (has_only_one_search_term) {
794 has_only_one_search_term = false;
795 break;
797 has_only_one_search_term = true;
798 supports_replacements_ = true;
801 // Only parse the host/key if there is one search term. Technically there
802 // could be more than one term, but it's uncommon; so we punt.
803 if (has_only_one_search_term)
804 ParseHostAndSearchTermKey(search_terms_data);
809 void TemplateURLRef::ParseHostAndSearchTermKey(
810 const SearchTermsData& search_terms_data) const {
811 std::string url_string(GetURL());
812 ReplaceSubstringsAfterOffset(&url_string, 0,
813 kGoogleBaseURLParameterFull,
814 search_terms_data.GoogleBaseURLValue());
815 ReplaceSubstringsAfterOffset(&url_string, 0,
816 kGoogleBaseSuggestURLParameterFull,
817 search_terms_data.GoogleBaseSuggestURLValue());
819 search_term_key_.clear();
820 search_term_position_in_path_ = std::string::npos;
821 host_.clear();
822 path_.clear();
823 search_term_key_location_ = url::Parsed::QUERY;
825 GURL url(url_string);
826 if (!url.is_valid())
827 return;
829 std::string query_key = FindSearchTermsKey(url.query());
830 std::string ref_key = FindSearchTermsKey(url.ref());
831 url::Component parameter_position;
832 const bool in_query = !query_key.empty();
833 const bool in_ref = !ref_key.empty();
834 const bool in_path = FindSearchTermsInPath(url.path(), &parameter_position);
835 if (in_query ? (in_ref || in_path) : (in_ref == in_path))
836 return; // No key or multiple keys found. We only handle having one key.
838 host_ = url.host();
839 path_ = url.path();
840 if (in_query) {
841 search_term_key_ = query_key;
842 search_term_key_location_ = url::Parsed::QUERY;
843 } else if (in_ref) {
844 search_term_key_ = ref_key;
845 search_term_key_location_ = url::Parsed::REF;
846 } else {
847 DCHECK(in_path);
848 DCHECK_GE(parameter_position.begin, 1); // Path must start with '/'.
849 search_term_key_location_ = url::Parsed::PATH;
850 search_term_position_in_path_ = parameter_position.begin;
851 // Remove the "{searchTerms}" itself from |path_|.
852 path_.erase(parameter_position.begin, parameter_position.len);
856 void TemplateURLRef::HandleReplacement(const std::string& name,
857 const std::string& value,
858 const Replacement& replacement,
859 std::string* url) const {
860 size_t pos = replacement.index;
861 if (replacement.is_post_param) {
862 DCHECK_LT(pos, post_params_.size());
863 DCHECK(!post_params_[pos].name.empty());
864 post_params_[pos].value = value;
865 } else {
866 url->insert(pos, name.empty() ? value : (name + "=" + value + "&"));
870 std::string TemplateURLRef::HandleReplacements(
871 const SearchTermsArgs& search_terms_args,
872 const SearchTermsData& search_terms_data,
873 PostContent* post_content) const {
874 if (replacements_.empty()) {
875 if (!post_params_.empty())
876 EncodeFormData(post_params_, post_content);
877 return parsed_url_;
880 // Determine if the search terms are in the query or before. We're escaping
881 // space as '+' in the former case and as '%20' in the latter case.
882 bool is_in_query = true;
883 for (Replacements::iterator i = replacements_.begin();
884 i != replacements_.end(); ++i) {
885 if (i->type == SEARCH_TERMS) {
886 base::string16::size_type query_start = parsed_url_.find('?');
887 is_in_query = query_start != base::string16::npos &&
888 (static_cast<base::string16::size_type>(i->index) > query_start);
889 break;
893 std::string input_encoding;
894 base::string16 encoded_terms;
895 base::string16 encoded_original_query;
896 owner_->EncodeSearchTerms(search_terms_args, is_in_query, &input_encoding,
897 &encoded_terms, &encoded_original_query);
899 std::string url = parsed_url_;
901 // replacements_ is ordered in ascending order, as such we need to iterate
902 // from the back.
903 for (Replacements::reverse_iterator i = replacements_.rbegin();
904 i != replacements_.rend(); ++i) {
905 switch (i->type) {
906 case ENCODING:
907 HandleReplacement(std::string(), input_encoding, *i, &url);
908 break;
910 case GOOGLE_ASSISTED_QUERY_STATS:
911 DCHECK(!i->is_post_param);
912 if (!search_terms_args.assisted_query_stats.empty()) {
913 // Get the base URL without substituting AQS to avoid infinite
914 // recursion. We need the URL to find out if it meets all
915 // AQS requirements (e.g. HTTPS protocol check).
916 // See TemplateURLRef::SearchTermsArgs for more details.
917 SearchTermsArgs search_terms_args_without_aqs(search_terms_args);
918 search_terms_args_without_aqs.assisted_query_stats.clear();
919 GURL base_url(ReplaceSearchTerms(
920 search_terms_args_without_aqs, search_terms_data, NULL));
921 if (base_url.SchemeIs(url::kHttpsScheme)) {
922 HandleReplacement(
923 "aqs", search_terms_args.assisted_query_stats, *i, &url);
926 break;
928 case GOOGLE_BASE_URL:
929 DCHECK(!i->is_post_param);
930 HandleReplacement(
931 std::string(), search_terms_data.GoogleBaseURLValue(), *i, &url);
932 break;
934 case GOOGLE_BASE_SUGGEST_URL:
935 DCHECK(!i->is_post_param);
936 HandleReplacement(
937 std::string(), search_terms_data.GoogleBaseSuggestURLValue(), *i,
938 &url);
939 break;
941 case GOOGLE_BOOKMARK_BAR_PINNED:
942 if (search_terms_data.IsShowingSearchTermsOnSearchResultsPages()) {
943 // Log whether the bookmark bar is pinned when the user is seeing
944 // InstantExtended on the SRP.
945 DCHECK(!i->is_post_param);
946 HandleReplacement(
947 "bmbp", search_terms_args.bookmark_bar_pinned ? "1" : "0", *i,
948 &url);
950 break;
952 case GOOGLE_CURRENT_PAGE_URL:
953 DCHECK(!i->is_post_param);
954 if (!search_terms_args.current_page_url.empty()) {
955 const std::string& escaped_current_page_url =
956 net::EscapeQueryParamValue(search_terms_args.current_page_url,
957 true);
958 HandleReplacement("url", escaped_current_page_url, *i, &url);
960 break;
962 case GOOGLE_CURSOR_POSITION:
963 DCHECK(!i->is_post_param);
964 if (search_terms_args.cursor_position != base::string16::npos)
965 HandleReplacement(
966 "cp",
967 base::StringPrintf("%" PRIuS, search_terms_args.cursor_position),
969 &url);
970 break;
972 case GOOGLE_FORCE_INSTANT_RESULTS:
973 DCHECK(!i->is_post_param);
974 HandleReplacement(std::string(),
975 search_terms_data.ForceInstantResultsParam(
976 search_terms_args.force_instant_results),
978 &url);
979 break;
981 case GOOGLE_INPUT_TYPE:
982 DCHECK(!i->is_post_param);
983 HandleReplacement(
984 "oit", base::IntToString(search_terms_args.input_type), *i, &url);
985 break;
987 case GOOGLE_INSTANT_EXTENDED_ENABLED:
988 DCHECK(!i->is_post_param);
989 HandleReplacement(std::string(),
990 search_terms_data.InstantExtendedEnabledParam(
991 type_ == SEARCH),
993 &url);
994 break;
996 case GOOGLE_NTP_IS_THEMED:
997 DCHECK(!i->is_post_param);
998 HandleReplacement(
999 std::string(), search_terms_data.NTPIsThemedParam(), *i, &url);
1000 break;
1002 case GOOGLE_OMNIBOX_START_MARGIN:
1003 DCHECK(!i->is_post_param);
1004 if (search_terms_args.enable_omnibox_start_margin) {
1005 int omnibox_start_margin = search_terms_data.OmniboxStartMargin();
1006 if (omnibox_start_margin >= 0) {
1007 HandleReplacement("es_sm", base::IntToString(omnibox_start_margin),
1008 *i, &url);
1011 break;
1013 case GOOGLE_CONTEXTUAL_SEARCH_VERSION:
1014 if (search_terms_args.contextual_search_params.version >= 0) {
1015 HandleReplacement(
1016 "ctxs",
1017 base::IntToString(
1018 search_terms_args.contextual_search_params.version),
1020 &url);
1022 break;
1024 case GOOGLE_CONTEXTUAL_SEARCH_CONTEXT_DATA: {
1025 DCHECK(!i->is_post_param);
1026 std::string context_data;
1028 const SearchTermsArgs::ContextualSearchParams& params =
1029 search_terms_args.contextual_search_params;
1031 if (params.start != std::string::npos) {
1032 context_data.append("ctxs_start=" + base::IntToString(
1033 params.start) + "&");
1036 if (params.end != std::string::npos) {
1037 context_data.append("ctxs_end=" + base::IntToString(
1038 params.end) + "&");
1041 if (!params.selection.empty())
1042 context_data.append("q=" + params.selection + "&");
1044 if (!params.content.empty())
1045 context_data.append("ctxs_content=" + params.content + "&");
1047 if (!params.base_page_url.empty())
1048 context_data.append("ctxsl_url=" + params.base_page_url + "&");
1050 if (!params.encoding.empty()) {
1051 context_data.append("ctxs_encoding=" + params.encoding + "&");
1054 context_data.append(
1055 params.resolve ? "ctxsl_resolve=1" : "ctxsl_resolve=0");
1057 HandleReplacement(std::string(), context_data, *i, &url);
1058 break;
1061 case GOOGLE_ORIGINAL_QUERY_FOR_SUGGESTION:
1062 DCHECK(!i->is_post_param);
1063 if (search_terms_args.accepted_suggestion >= 0 ||
1064 !search_terms_args.assisted_query_stats.empty()) {
1065 HandleReplacement(
1066 "oq", base::UTF16ToUTF8(encoded_original_query), *i, &url);
1068 break;
1070 case GOOGLE_PAGE_CLASSIFICATION:
1071 if (search_terms_args.page_classification !=
1072 metrics::OmniboxEventProto::INVALID_SPEC) {
1073 HandleReplacement(
1074 "pgcl", base::IntToString(search_terms_args.page_classification),
1075 *i, &url);
1077 break;
1079 case GOOGLE_PREFETCH_QUERY: {
1080 const std::string& query = search_terms_args.prefetch_query;
1081 const std::string& type = search_terms_args.prefetch_query_type;
1082 if (!query.empty() && !type.empty()) {
1083 HandleReplacement(
1084 std::string(), "pfq=" + query + "&qha=" + type + "&", *i, &url);
1086 break;
1089 case GOOGLE_RLZ: {
1090 DCHECK(!i->is_post_param);
1091 // On platforms that don't have RLZ, we still want this branch
1092 // to happen so that we replace the RLZ template with the
1093 // empty string. (If we don't handle this case, we hit a
1094 // NOTREACHED below.)
1095 base::string16 rlz_string = search_terms_data.GetRlzParameterValue(
1096 search_terms_args.from_app_list);
1097 if (!rlz_string.empty()) {
1098 HandleReplacement("rlz", base::UTF16ToUTF8(rlz_string), *i, &url);
1100 break;
1103 case GOOGLE_SEARCH_CLIENT: {
1104 DCHECK(!i->is_post_param);
1105 std::string client = search_terms_data.GetSearchClient();
1106 if (!client.empty())
1107 HandleReplacement("client", client, *i, &url);
1108 break;
1111 case GOOGLE_SEARCH_FIELDTRIAL_GROUP:
1112 // We are not currently running any fieldtrials that modulate the search
1113 // url. If we do, then we'd have some conditional insert such as:
1114 // url.insert(i->index, used_www ? "gcx=w&" : "gcx=c&");
1115 break;
1117 case GOOGLE_SEARCH_VERSION:
1118 if (search_terms_data.EnableAnswersInSuggest())
1119 HandleReplacement("gs_rn", "42", *i, &url);
1120 break;
1122 case GOOGLE_SESSION_TOKEN: {
1123 std::string token = search_terms_args.session_token;
1124 if (!token.empty())
1125 HandleReplacement("psi", token, *i, &url);
1126 break;
1129 case GOOGLE_SUGGEST_CLIENT:
1130 HandleReplacement(
1131 std::string(), search_terms_data.GetSuggestClient(), *i, &url);
1132 break;
1134 case GOOGLE_SUGGEST_REQUEST_ID:
1135 HandleReplacement(
1136 std::string(), search_terms_data.GetSuggestRequestIdentifier(), *i,
1137 &url);
1138 break;
1140 case GOOGLE_UNESCAPED_SEARCH_TERMS: {
1141 std::string unescaped_terms;
1142 base::UTF16ToCodepage(search_terms_args.search_terms,
1143 input_encoding.c_str(),
1144 base::OnStringConversionError::SKIP,
1145 &unescaped_terms);
1146 HandleReplacement(std::string(), unescaped_terms, *i, &url);
1147 break;
1150 case LANGUAGE:
1151 HandleReplacement(
1152 std::string(), search_terms_data.GetApplicationLocale(), *i, &url);
1153 break;
1155 case SEARCH_TERMS:
1156 HandleReplacement(
1157 std::string(), base::UTF16ToUTF8(encoded_terms), *i, &url);
1158 break;
1160 case GOOGLE_IMAGE_THUMBNAIL:
1161 HandleReplacement(
1162 std::string(), search_terms_args.image_thumbnail_content, *i, &url);
1163 post_params_[i->index].content_type = "image/jpeg";
1164 break;
1166 case GOOGLE_IMAGE_URL:
1167 if (search_terms_args.image_url.is_valid()) {
1168 HandleReplacement(
1169 std::string(), search_terms_args.image_url.spec(), *i, &url);
1171 break;
1173 case GOOGLE_IMAGE_ORIGINAL_WIDTH:
1174 if (!search_terms_args.image_original_size.IsEmpty()) {
1175 HandleReplacement(
1176 std::string(),
1177 base::IntToString(search_terms_args.image_original_size.width()),
1178 *i, &url);
1180 break;
1182 case GOOGLE_IMAGE_ORIGINAL_HEIGHT:
1183 if (!search_terms_args.image_original_size.IsEmpty()) {
1184 HandleReplacement(
1185 std::string(),
1186 base::IntToString(search_terms_args.image_original_size.height()),
1187 *i, &url);
1189 break;
1191 case GOOGLE_IMAGE_SEARCH_SOURCE:
1192 HandleReplacement(
1193 std::string(), search_terms_data.GoogleImageSearchSource(), *i,
1194 &url);
1195 break;
1197 default:
1198 NOTREACHED();
1199 break;
1203 if (!post_params_.empty())
1204 EncodeFormData(post_params_, post_content);
1206 return url;
1210 // TemplateURL ----------------------------------------------------------------
1212 TemplateURL::AssociatedExtensionInfo::AssociatedExtensionInfo(
1213 Type type,
1214 const std::string& extension_id)
1215 : type(type),
1216 extension_id(extension_id),
1217 wants_to_be_default_engine(false) {
1218 DCHECK_NE(NORMAL, type);
1221 TemplateURL::AssociatedExtensionInfo::~AssociatedExtensionInfo() {
1224 TemplateURL::TemplateURL(const TemplateURLData& data)
1225 : data_(data),
1226 url_ref_(this, TemplateURLRef::SEARCH),
1227 suggestions_url_ref_(this,
1228 TemplateURLRef::SUGGEST),
1229 instant_url_ref_(this,
1230 TemplateURLRef::INSTANT),
1231 image_url_ref_(this, TemplateURLRef::IMAGE),
1232 new_tab_url_ref_(this, TemplateURLRef::NEW_TAB),
1233 contextual_search_url_ref_(this, TemplateURLRef::CONTEXTUAL_SEARCH) {
1234 SetPrepopulateId(data_.prepopulate_id);
1236 if (data_.search_terms_replacement_key ==
1237 kGoogleInstantExtendedEnabledKeyFull) {
1238 data_.search_terms_replacement_key = google_util::kInstantExtendedAPIParam;
1242 TemplateURL::~TemplateURL() {
1245 // static
1246 base::string16 TemplateURL::GenerateKeyword(const GURL& url) {
1247 DCHECK(url.is_valid());
1248 // Strip "www." off the front of the keyword; otherwise the keyword won't work
1249 // properly. See http://code.google.com/p/chromium/issues/detail?id=6984 .
1250 // Special case: if the host was exactly "www." (not sure this can happen but
1251 // perhaps with some weird intranet and custom DNS server?), ensure we at
1252 // least don't return the empty string.
1253 base::string16 keyword(net::StripWWWFromHost(url));
1254 return keyword.empty() ? base::ASCIIToUTF16("www") : keyword;
1257 // static
1258 GURL TemplateURL::GenerateFaviconURL(const GURL& url) {
1259 DCHECK(url.is_valid());
1260 GURL::Replacements rep;
1262 const char favicon_path[] = "/favicon.ico";
1263 int favicon_path_len = arraysize(favicon_path) - 1;
1265 rep.SetPath(favicon_path, url::Component(0, favicon_path_len));
1266 rep.ClearUsername();
1267 rep.ClearPassword();
1268 rep.ClearQuery();
1269 rep.ClearRef();
1270 return url.ReplaceComponents(rep);
1273 // static
1274 bool TemplateURL::MatchesData(const TemplateURL* t_url,
1275 const TemplateURLData* data,
1276 const SearchTermsData& search_terms_data) {
1277 if (!t_url || !data)
1278 return !t_url && !data;
1280 return (t_url->short_name() == data->short_name) &&
1281 t_url->HasSameKeywordAs(*data, search_terms_data) &&
1282 (t_url->url() == data->url()) &&
1283 (t_url->suggestions_url() == data->suggestions_url) &&
1284 (t_url->instant_url() == data->instant_url) &&
1285 (t_url->image_url() == data->image_url) &&
1286 (t_url->new_tab_url() == data->new_tab_url) &&
1287 (t_url->search_url_post_params() == data->search_url_post_params) &&
1288 (t_url->suggestions_url_post_params() ==
1289 data->suggestions_url_post_params) &&
1290 (t_url->instant_url_post_params() == data->instant_url_post_params) &&
1291 (t_url->image_url_post_params() == data->image_url_post_params) &&
1292 (t_url->favicon_url() == data->favicon_url) &&
1293 (t_url->safe_for_autoreplace() == data->safe_for_autoreplace) &&
1294 (t_url->show_in_default_list() == data->show_in_default_list) &&
1295 (t_url->input_encodings() == data->input_encodings) &&
1296 (t_url->alternate_urls() == data->alternate_urls) &&
1297 (t_url->search_terms_replacement_key() ==
1298 data->search_terms_replacement_key);
1301 base::string16 TemplateURL::AdjustedShortNameForLocaleDirection() const {
1302 base::string16 bidi_safe_short_name = data_.short_name;
1303 base::i18n::AdjustStringForLocaleDirection(&bidi_safe_short_name);
1304 return bidi_safe_short_name;
1307 bool TemplateURL::ShowInDefaultList(
1308 const SearchTermsData& search_terms_data) const {
1309 return data_.show_in_default_list &&
1310 url_ref_.SupportsReplacement(search_terms_data);
1313 bool TemplateURL::SupportsReplacement(
1314 const SearchTermsData& search_terms_data) const {
1315 return url_ref_.SupportsReplacement(search_terms_data);
1318 bool TemplateURL::HasGoogleBaseURLs(
1319 const SearchTermsData& search_terms_data) const {
1320 return url_ref_.HasGoogleBaseURLs(search_terms_data) ||
1321 suggestions_url_ref_.HasGoogleBaseURLs(search_terms_data) ||
1322 instant_url_ref_.HasGoogleBaseURLs(search_terms_data) ||
1323 image_url_ref_.HasGoogleBaseURLs(search_terms_data) ||
1324 new_tab_url_ref_.HasGoogleBaseURLs(search_terms_data);
1327 bool TemplateURL::IsGoogleSearchURLWithReplaceableKeyword(
1328 const SearchTermsData& search_terms_data) const {
1329 return (GetType() == NORMAL) &&
1330 url_ref_.HasGoogleBaseURLs(search_terms_data) &&
1331 google_util::IsGoogleHostname(base::UTF16ToUTF8(data_.keyword()),
1332 google_util::DISALLOW_SUBDOMAIN);
1335 bool TemplateURL::HasSameKeywordAs(
1336 const TemplateURLData& other,
1337 const SearchTermsData& search_terms_data) const {
1338 return (data_.keyword() == other.keyword()) ||
1339 (IsGoogleSearchURLWithReplaceableKeyword(search_terms_data) &&
1340 TemplateURL(other).IsGoogleSearchURLWithReplaceableKeyword(
1341 search_terms_data));
1344 TemplateURL::Type TemplateURL::GetType() const {
1345 return extension_info_ ? extension_info_->type : NORMAL;
1348 std::string TemplateURL::GetExtensionId() const {
1349 DCHECK(extension_info_);
1350 return extension_info_->extension_id;
1353 size_t TemplateURL::URLCount() const {
1354 // Add 1 for the regular search URL.
1355 return data_.alternate_urls.size() + 1;
1358 const std::string& TemplateURL::GetURL(size_t index) const {
1359 DCHECK_LT(index, URLCount());
1361 return (index < data_.alternate_urls.size()) ?
1362 data_.alternate_urls[index] : url();
1365 bool TemplateURL::ExtractSearchTermsFromURL(
1366 const GURL& url,
1367 const SearchTermsData& search_terms_data,
1368 base::string16* search_terms) {
1369 return FindSearchTermsInURL(url, search_terms_data, search_terms, NULL, NULL);
1372 bool TemplateURL::IsSearchURL(
1373 const GURL& url,
1374 const SearchTermsData& search_terms_data) {
1375 base::string16 search_terms;
1376 return ExtractSearchTermsFromURL(url, search_terms_data, &search_terms) &&
1377 !search_terms.empty();
1380 bool TemplateURL::HasSearchTermsReplacementKey(const GURL& url) const {
1381 // Look for the key both in the query and the ref.
1382 std::string params[] = {url.query(), url.ref()};
1384 for (int i = 0; i < 2; ++i) {
1385 url::Component query, key, value;
1386 query.len = static_cast<int>(params[i].size());
1387 while (url::ExtractQueryKeyValue(params[i].c_str(), &query, &key, &value)) {
1388 if (key.is_nonempty() &&
1389 params[i].substr(key.begin, key.len) ==
1390 search_terms_replacement_key()) {
1391 return true;
1395 return false;
1398 bool TemplateURL::ReplaceSearchTermsInURL(
1399 const GURL& url,
1400 const TemplateURLRef::SearchTermsArgs& search_terms_args,
1401 const SearchTermsData& search_terms_data,
1402 GURL* result) {
1403 // TODO(beaudoin): Use AQS from |search_terms_args| too.
1404 url::Parsed::ComponentType search_term_component;
1405 url::Component search_terms_position;
1406 base::string16 search_terms;
1407 if (!FindSearchTermsInURL(url, search_terms_data, &search_terms,
1408 &search_term_component, &search_terms_position)) {
1409 return false;
1411 DCHECK(search_terms_position.is_nonempty());
1413 // Query and ref are encoded in the same way.
1414 const bool is_in_query = (search_term_component != url::Parsed::PATH);
1416 std::string input_encoding;
1417 base::string16 encoded_terms;
1418 base::string16 encoded_original_query;
1419 EncodeSearchTerms(search_terms_args, is_in_query, &input_encoding,
1420 &encoded_terms, &encoded_original_query);
1422 std::string old_params;
1423 if (search_term_component == url::Parsed::QUERY) {
1424 old_params = url.query();
1425 } else if (search_term_component == url::Parsed::REF) {
1426 old_params = url.ref();
1427 } else {
1428 DCHECK_EQ(search_term_component, url::Parsed::PATH);
1429 old_params = url.path();
1432 std::string new_params(old_params, 0, search_terms_position.begin);
1433 new_params += base::UTF16ToUTF8(search_terms_args.search_terms);
1434 new_params += old_params.substr(search_terms_position.end());
1435 GURL::Replacements replacements;
1437 if (search_term_component == url::Parsed::QUERY) {
1438 replacements.SetQueryStr(new_params);
1439 } else if (search_term_component == url::Parsed::REF) {
1440 replacements.SetRefStr(new_params);
1441 } else {
1442 DCHECK_EQ(search_term_component, url::Parsed::PATH);
1443 replacements.SetPathStr(new_params);
1446 *result = url.ReplaceComponents(replacements);
1447 return true;
1450 void TemplateURL::EncodeSearchTerms(
1451 const TemplateURLRef::SearchTermsArgs& search_terms_args,
1452 bool is_in_query,
1453 std::string* input_encoding,
1454 base::string16* encoded_terms,
1455 base::string16* encoded_original_query) const {
1457 std::vector<std::string> encodings(input_encodings());
1458 if (std::find(encodings.begin(), encodings.end(), "UTF-8") == encodings.end())
1459 encodings.push_back("UTF-8");
1460 for (std::vector<std::string>::const_iterator i(encodings.begin());
1461 i != encodings.end(); ++i) {
1462 if (TryEncoding(search_terms_args.search_terms,
1463 search_terms_args.original_query, i->c_str(),
1464 is_in_query, encoded_terms, encoded_original_query)) {
1465 *input_encoding = *i;
1466 return;
1469 NOTREACHED();
1472 GURL TemplateURL::GenerateSearchURL(
1473 const SearchTermsData& search_terms_data) const {
1474 if (!url_ref_.IsValid(search_terms_data))
1475 return GURL();
1477 if (!url_ref_.SupportsReplacement(search_terms_data))
1478 return GURL(url());
1480 // Use something obscure for the search terms argument so that in the rare
1481 // case the term replaces the URL it's unlikely another keyword would have the
1482 // same url.
1483 // TODO(jnd): Add additional parameters to get post data when the search URL
1484 // has post parameters.
1485 return GURL(url_ref_.ReplaceSearchTerms(
1486 TemplateURLRef::SearchTermsArgs(
1487 base::ASCIIToUTF16("blah.blah.blah.blah.blah")),
1488 search_terms_data, NULL));
1491 void TemplateURL::CopyFrom(const TemplateURL& other) {
1492 if (this == &other)
1493 return;
1495 data_ = other.data_;
1496 url_ref_.InvalidateCachedValues();
1497 suggestions_url_ref_.InvalidateCachedValues();
1498 instant_url_ref_.InvalidateCachedValues();
1499 SetPrepopulateId(other.data_.prepopulate_id);
1502 void TemplateURL::SetURL(const std::string& url) {
1503 data_.SetURL(url);
1504 url_ref_.InvalidateCachedValues();
1507 void TemplateURL::SetPrepopulateId(int id) {
1508 data_.prepopulate_id = id;
1509 const bool prepopulated = id > 0;
1510 url_ref_.prepopulated_ = prepopulated;
1511 suggestions_url_ref_.prepopulated_ = prepopulated;
1512 instant_url_ref_.prepopulated_ = prepopulated;
1515 void TemplateURL::ResetKeywordIfNecessary(
1516 const SearchTermsData& search_terms_data,
1517 bool force) {
1518 if (IsGoogleSearchURLWithReplaceableKeyword(search_terms_data) || force) {
1519 DCHECK(GetType() != OMNIBOX_API_EXTENSION);
1520 GURL url(GenerateSearchURL(search_terms_data));
1521 if (url.is_valid())
1522 data_.SetKeyword(GenerateKeyword(url));
1526 bool TemplateURL::FindSearchTermsInURL(
1527 const GURL& url,
1528 const SearchTermsData& search_terms_data,
1529 base::string16* search_terms,
1530 url::Parsed::ComponentType* search_term_component,
1531 url::Component* search_terms_position) {
1532 DCHECK(search_terms);
1533 search_terms->clear();
1535 // Try to match with every pattern.
1536 for (size_t i = 0; i < URLCount(); ++i) {
1537 TemplateURLRef ref(this, i);
1538 if (ref.ExtractSearchTermsFromURL(url, search_terms, search_terms_data,
1539 search_term_component, search_terms_position)) {
1540 // If ExtractSearchTermsFromURL() returns true and |search_terms| is empty
1541 // it means the pattern matched but no search terms were present. In this
1542 // case we fail immediately without looking for matches in subsequent
1543 // patterns. This means that given patterns
1544 // [ "http://foo/#q={searchTerms}", "http://foo/?q={searchTerms}" ],
1545 // calling ExtractSearchTermsFromURL() on "http://foo/?q=bar#q=' would
1546 // return false. This is important for at least Google, where such URLs
1547 // are invalid.
1548 return !search_terms->empty();
1551 return false;