Home | History | Annotate | Download | only in search_engines
      1 // Copyright (c) 2012 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 "chrome/browser/search_engines/util.h"
      6 
      7 #include <map>
      8 #include <set>
      9 #include <string>
     10 #include <vector>
     11 
     12 #include "base/logging.h"
     13 #include "base/memory/scoped_vector.h"
     14 #include "base/prefs/pref_service.h"
     15 #include "base/time/time.h"
     16 #include "chrome/browser/profiles/profile.h"
     17 #include "chrome/browser/search_engines/template_url.h"
     18 #include "chrome/browser/search_engines/template_url_prepopulate_data.h"
     19 #include "chrome/browser/search_engines/template_url_service.h"
     20 #include "chrome/browser/search_engines/template_url_service_factory.h"
     21 #include "content/public/browser/browser_thread.h"
     22 
     23 using content::BrowserThread;
     24 
     25 base::string16 GetDefaultSearchEngineName(Profile* profile) {
     26   if (!profile) {
     27     NOTREACHED();
     28     return base::string16();
     29   }
     30   const TemplateURL* const default_provider =
     31       TemplateURLServiceFactory::GetForProfile(profile)->
     32       GetDefaultSearchProvider();
     33   if (!default_provider) {
     34     // TODO(cpu): bug 1187517. It is possible to have no default provider.
     35     // returning an empty string is a stopgap measure for the crash
     36     // http://code.google.com/p/chromium/issues/detail?id=2573
     37     return base::string16();
     38   }
     39   return default_provider->short_name();
     40 }
     41 
     42 GURL GetDefaultSearchURLForSearchTerms(Profile* profile,
     43                                        const base::string16& terms) {
     44   DCHECK(profile);
     45   const TemplateURL* default_provider =
     46       TemplateURLServiceFactory::GetForProfile(profile)->
     47       GetDefaultSearchProvider();
     48   if (!default_provider)
     49     return GURL();
     50   const TemplateURLRef& search_url = default_provider->url_ref();
     51   DCHECK(search_url.SupportsReplacement());
     52   TemplateURLRef::SearchTermsArgs search_terms_args(terms);
     53   search_terms_args.append_extra_query_params = true;
     54   return GURL(search_url.ReplaceSearchTerms(search_terms_args));
     55 }
     56 
     57 void RemoveDuplicatePrepopulateIDs(
     58     WebDataService* service,
     59     const ScopedVector<TemplateURL>& prepopulated_urls,
     60     TemplateURL* default_search_provider,
     61     TemplateURLService::TemplateURLVector* template_urls,
     62     std::set<std::string>* removed_keyword_guids) {
     63   DCHECK(service == NULL || BrowserThread::CurrentlyOn(BrowserThread::UI));
     64   DCHECK(template_urls);
     65 
     66   // For convenience construct an ID->TemplateURL* map from |prepopulated_urls|.
     67   typedef std::map<int, TemplateURL*> PrepopulatedURLMap;
     68   PrepopulatedURLMap prepopulated_url_map;
     69   for (std::vector<TemplateURL*>::const_iterator i(prepopulated_urls.begin());
     70        i != prepopulated_urls.end(); ++i)
     71     prepopulated_url_map[(*i)->prepopulate_id()] = *i;
     72 
     73   // Separate |template_urls| into prepopulated and non-prepopulated groups.
     74   typedef std::multimap<int, TemplateURL*> UncheckedURLMap;
     75   UncheckedURLMap unchecked_urls;
     76   TemplateURLService::TemplateURLVector checked_urls;
     77   for (TemplateURLService::TemplateURLVector::iterator i(
     78        template_urls->begin()); i != template_urls->end(); ++i) {
     79     TemplateURL* turl = *i;
     80     int prepopulate_id = turl->prepopulate_id();
     81     if (prepopulate_id)
     82       unchecked_urls.insert(std::make_pair(prepopulate_id, turl));
     83     else
     84       checked_urls.push_back(turl);
     85   }
     86 
     87   // For each group of prepopulated URLs with one ID, find the best URL to use
     88   // and add it to the (initially all non-prepopulated) URLs we've already OKed.
     89   // Delete the others from the service and from memory.
     90   while (!unchecked_urls.empty()) {
     91     // Find the best URL.
     92     int prepopulate_id = unchecked_urls.begin()->first;
     93     PrepopulatedURLMap::const_iterator prepopulated_url =
     94         prepopulated_url_map.find(prepopulate_id);
     95     UncheckedURLMap::iterator end = unchecked_urls.upper_bound(prepopulate_id);
     96     UncheckedURLMap::iterator best = unchecked_urls.begin();
     97     bool matched_keyword = false;
     98     for (UncheckedURLMap::iterator i = unchecked_urls.begin(); i != end; ++i) {
     99       // A URL is automatically the best if it's the default search engine.
    100       if (i->second == default_search_provider) {
    101         best = i;
    102         break;
    103       }
    104 
    105       // Otherwise, a URL is best if it matches the prepopulated data's keyword;
    106       // if none match, just fall back to using the one with the lowest ID.
    107       if (matched_keyword)
    108         continue;
    109       if ((prepopulated_url != prepopulated_url_map.end()) &&
    110            i->second->HasSameKeywordAs(*prepopulated_url->second)) {
    111         best = i;
    112         matched_keyword = true;
    113       } else if (i->second->id() < best->second->id()) {
    114         best = i;
    115       }
    116     }
    117 
    118     // Add the best URL to the checked group and delete the rest.
    119     checked_urls.push_back(best->second);
    120     for (UncheckedURLMap::iterator i = unchecked_urls.begin(); i != end; ++i) {
    121       if (i == best)
    122         continue;
    123       if (service) {
    124         service->RemoveKeyword(i->second->id());
    125         if (removed_keyword_guids)
    126           removed_keyword_guids->insert(i->second->sync_guid());
    127       }
    128       delete i->second;
    129     }
    130 
    131     // Done with this group.
    132     unchecked_urls.erase(unchecked_urls.begin(), end);
    133   }
    134 
    135   // Return the checked URLs.
    136   template_urls->swap(checked_urls);
    137 }
    138 
    139 // Returns the TemplateURL with id specified from the list of TemplateURLs.
    140 // If not found, returns NULL.
    141 TemplateURL* GetTemplateURLByID(
    142     const TemplateURLService::TemplateURLVector& template_urls,
    143     int64 id) {
    144   for (TemplateURLService::TemplateURLVector::const_iterator i(
    145        template_urls.begin()); i != template_urls.end(); ++i) {
    146     if ((*i)->id() == id) {
    147       return *i;
    148     }
    149   }
    150   return NULL;
    151 }
    152 
    153 TemplateURL* FindURLByPrepopulateID(
    154     const TemplateURLService::TemplateURLVector& template_urls,
    155     int prepopulate_id) {
    156   for (std::vector<TemplateURL*>::const_iterator i = template_urls.begin();
    157        i < template_urls.end(); ++i) {
    158     if ((*i)->prepopulate_id() == prepopulate_id)
    159       return *i;
    160   }
    161   return NULL;
    162 }
    163 
    164 void MergeIntoPrepopulatedEngineData(TemplateURLData* prepopulated_url,
    165                                      const TemplateURL* original_turl) {
    166   DCHECK_EQ(original_turl->prepopulate_id(), prepopulated_url->prepopulate_id);
    167   if (!original_turl->safe_for_autoreplace()) {
    168     prepopulated_url->safe_for_autoreplace = false;
    169     prepopulated_url->SetKeyword(original_turl->keyword());
    170     prepopulated_url->short_name = original_turl->short_name();
    171   }
    172   prepopulated_url->id = original_turl->id();
    173   prepopulated_url->sync_guid = original_turl->sync_guid();
    174   prepopulated_url->date_created = original_turl->date_created();
    175   prepopulated_url->last_modified = original_turl->last_modified();
    176 }
    177 
    178 ActionsFromPrepopulateData::ActionsFromPrepopulateData() {}
    179 
    180 ActionsFromPrepopulateData::~ActionsFromPrepopulateData() {}
    181 
    182 // This is invoked when the version of the prepopulate data changes.
    183 // If |removed_keyword_guids| is not NULL, the Sync GUID of each item removed
    184 // from the DB will be added to it.  Note that this function will take
    185 // ownership of |prepopulated_urls| and will clear the vector.
    186 void MergeEnginesFromPrepopulateData(
    187     Profile* profile,
    188     WebDataService* service,
    189     ScopedVector<TemplateURL>* prepopulated_urls,
    190     size_t default_search_index,
    191     TemplateURLService::TemplateURLVector* template_urls,
    192     TemplateURL** default_search_provider,
    193     std::set<std::string>* removed_keyword_guids) {
    194   DCHECK(service == NULL || BrowserThread::CurrentlyOn(BrowserThread::UI));
    195   DCHECK(prepopulated_urls);
    196   DCHECK(template_urls);
    197   DCHECK(default_search_provider);
    198 
    199   int default_prepopulated_id =
    200       (*prepopulated_urls)[default_search_index]->prepopulate_id();
    201   ActionsFromPrepopulateData actions(CreateActionsFromCurrentPrepopulateData(
    202       prepopulated_urls, *template_urls, *default_search_provider));
    203 
    204   // Remove items.
    205   for (std::vector<TemplateURL*>::iterator i = actions.removed_engines.begin();
    206        i < actions.removed_engines.end(); ++i) {
    207     scoped_ptr<TemplateURL> template_url(*i);
    208     TemplateURLService::TemplateURLVector::iterator j =
    209         std::find(template_urls->begin(), template_urls->end(), template_url);
    210     DCHECK(j != template_urls->end());
    211     DCHECK(*j != *default_search_provider);
    212     template_urls->erase(j);
    213     if (service) {
    214       service->RemoveKeyword(template_url->id());
    215       if (removed_keyword_guids)
    216         removed_keyword_guids->insert(template_url->sync_guid());
    217     }
    218   }
    219 
    220   // Edit items.
    221   for (EditedEngines::iterator i(actions.edited_engines.begin());
    222        i < actions.edited_engines.end(); ++i) {
    223     TemplateURLData& data = i->second;
    224     scoped_ptr<TemplateURL> existing_url(i->first);
    225     if (service)
    226       service->UpdateKeyword(data);
    227 
    228     // Replace the entry in |template_urls| with the updated one.
    229     TemplateURLService::TemplateURLVector::iterator j = std::find(
    230         template_urls->begin(), template_urls->end(), existing_url.get());
    231     *j = new TemplateURL(profile, data);
    232     if (*default_search_provider == existing_url.get())
    233       *default_search_provider = *j;
    234   }
    235 
    236   // Add items.
    237   template_urls->insert(template_urls->end(), actions.added_engines.begin(),
    238                         actions.added_engines.end());
    239 
    240   if (!*default_search_provider) {
    241     // The user had no existing default search provider, so set the
    242     // default to the default prepopulated engine.
    243     *default_search_provider = FindURLByPrepopulateID(*template_urls,
    244                                                       default_prepopulated_id);
    245   }
    246 }
    247 
    248 ActionsFromPrepopulateData CreateActionsFromCurrentPrepopulateData(
    249     ScopedVector<TemplateURL>* prepopulated_urls,
    250     const TemplateURLService::TemplateURLVector& existing_urls,
    251     const TemplateURL* default_search_provider) {
    252   // Create a map to hold all provided |template_urls| that originally came from
    253   // prepopulate data (i.e. have a non-zero prepopulate_id()).
    254   typedef std::map<int, TemplateURL*> IDMap;
    255   IDMap id_to_turl;
    256   for (TemplateURLService::TemplateURLVector::const_iterator i(
    257        existing_urls.begin()); i != existing_urls.end(); ++i) {
    258     int prepopulate_id = (*i)->prepopulate_id();
    259     if (prepopulate_id > 0)
    260       id_to_turl[prepopulate_id] = *i;
    261   }
    262 
    263   // For each current prepopulated URL, check whether |template_urls| contained
    264   // a matching prepopulated URL.  If so, update the passed-in URL to match the
    265   // current data.  (If the passed-in URL was user-edited, we persist the user's
    266   // name and keyword.)  If not, add the prepopulated URL.
    267   ActionsFromPrepopulateData actions;
    268   for (size_t i = 0; i < prepopulated_urls->size(); ++i) {
    269     // We take ownership of |prepopulated_urls[i]|.
    270     scoped_ptr<TemplateURL> prepopulated_url((*prepopulated_urls)[i]);
    271     const int prepopulated_id = prepopulated_url->prepopulate_id();
    272     DCHECK_NE(0, prepopulated_id);
    273 
    274     IDMap::iterator existing_url_iter(id_to_turl.find(prepopulated_id));
    275     if (existing_url_iter != id_to_turl.end()) {
    276       // Update the data store with the new prepopulated data. Preserve user
    277       // edits to the name and keyword.
    278       TemplateURLData data(prepopulated_url->data());
    279       TemplateURL* existing_url(existing_url_iter->second);
    280       id_to_turl.erase(existing_url_iter);
    281       MergeIntoPrepopulatedEngineData(&data, existing_url);
    282       // Update last_modified to ensure that if this entry is later merged with
    283       // entries from Sync, the conflict resolution logic knows that this was
    284       // updated and propagates the new values to the server.
    285       data.last_modified = base::Time::Now();
    286       actions.edited_engines.push_back(std::make_pair(existing_url, data));
    287     } else {
    288       actions.added_engines.push_back(prepopulated_url.release());
    289     }
    290   }
    291   // The above loop takes ownership of all the contents of prepopulated_urls.
    292   // Clear the pointers.
    293   prepopulated_urls->weak_erase(prepopulated_urls->begin(),
    294                                 prepopulated_urls->end());
    295 
    296   // The block above removed all the URLs from the |id_to_turl| map that were
    297   // found in the prepopulate data.  Any remaining URLs that haven't been
    298   // user-edited or made default can be removed from the data store.
    299   for (IDMap::iterator i(id_to_turl.begin()); i != id_to_turl.end(); ++i) {
    300     TemplateURL* template_url = i->second;
    301     if ((template_url->safe_for_autoreplace()) &&
    302         (template_url != default_search_provider))
    303       actions.removed_engines.push_back(template_url);
    304   }
    305 
    306   return actions;
    307 }
    308 
    309 void GetSearchProvidersUsingKeywordResult(
    310     const WDTypedResult& result,
    311     WebDataService* service,
    312     Profile* profile,
    313     TemplateURLService::TemplateURLVector* template_urls,
    314     TemplateURL** default_search_provider,
    315     int* new_resource_keyword_version,
    316     std::set<std::string>* removed_keyword_guids) {
    317   DCHECK(service == NULL || BrowserThread::CurrentlyOn(BrowserThread::UI));
    318   DCHECK(template_urls);
    319   DCHECK(template_urls->empty());
    320   DCHECK(default_search_provider);
    321   DCHECK(*default_search_provider == NULL);
    322   DCHECK_EQ(KEYWORDS_RESULT, result.GetType());
    323   DCHECK(new_resource_keyword_version);
    324 
    325   WDKeywordsResult keyword_result = reinterpret_cast<
    326       const WDResult<WDKeywordsResult>*>(&result)->GetValue();
    327 
    328   for (KeywordTable::Keywords::iterator i(keyword_result.keywords.begin());
    329        i != keyword_result.keywords.end(); ++i) {
    330     // Fix any duplicate encodings in the local database.  Note that we don't
    331     // adjust the last_modified time of this keyword; this way, we won't later
    332     // overwrite any changes on the sync server that happened to this keyword
    333     // since the last time we synced.  Instead, we also run a de-duping pass on
    334     // the server-provided data in
    335     // TemplateURLService::CreateTemplateURLFromTemplateURLAndSyncData() and
    336     // update the server with the merged, de-duped results at that time.  We
    337     // still fix here, though, to correct problems in clients that have disabled
    338     // search engine sync, since in that case that code will never be reached.
    339     if (DeDupeEncodings(&i->input_encodings) && service)
    340       service->UpdateKeyword(*i);
    341     template_urls->push_back(new TemplateURL(profile, *i));
    342   }
    343 
    344   int64 default_search_provider_id = keyword_result.default_search_provider_id;
    345   if (default_search_provider_id) {
    346     *default_search_provider =
    347         GetTemplateURLByID(*template_urls, default_search_provider_id);
    348   }
    349 
    350   *new_resource_keyword_version = keyword_result.builtin_keyword_version;
    351   GetSearchProvidersUsingLoadedEngines(service, profile, template_urls,
    352                                        default_search_provider,
    353                                        new_resource_keyword_version,
    354                                        removed_keyword_guids);
    355 }
    356 
    357 void GetSearchProvidersUsingLoadedEngines(
    358     WebDataService* service,
    359     Profile* profile,
    360     TemplateURLService::TemplateURLVector* template_urls,
    361     TemplateURL** default_search_provider,
    362     int* resource_keyword_version,
    363     std::set<std::string>* removed_keyword_guids) {
    364   DCHECK(service == NULL || BrowserThread::CurrentlyOn(BrowserThread::UI));
    365   DCHECK(template_urls);
    366   DCHECK(default_search_provider);
    367   DCHECK(resource_keyword_version);
    368 
    369   size_t default_search_index;
    370   ScopedVector<TemplateURL> prepopulated_urls =
    371       TemplateURLPrepopulateData::GetPrepopulatedEngines(profile,
    372                                                          &default_search_index);
    373   RemoveDuplicatePrepopulateIDs(service, prepopulated_urls,
    374                                 *default_search_provider, template_urls,
    375                                 removed_keyword_guids);
    376 
    377   const int prepopulate_resource_keyword_version =
    378       TemplateURLPrepopulateData::GetDataVersion(
    379           profile ? profile->GetPrefs() : NULL);
    380   if (*resource_keyword_version < prepopulate_resource_keyword_version) {
    381     MergeEnginesFromPrepopulateData(profile, service, &prepopulated_urls,
    382         default_search_index, template_urls, default_search_provider,
    383         removed_keyword_guids);
    384     *resource_keyword_version = prepopulate_resource_keyword_version;
    385   } else {
    386     *resource_keyword_version = 0;
    387   }
    388 }
    389 
    390 bool DeDupeEncodings(std::vector<std::string>* encodings) {
    391   std::vector<std::string> deduped_encodings;
    392   std::set<std::string> encoding_set;
    393   for (std::vector<std::string>::const_iterator i(encodings->begin());
    394        i != encodings->end(); ++i) {
    395     if (encoding_set.insert(*i).second)
    396       deduped_encodings.push_back(*i);
    397   }
    398   encodings->swap(deduped_encodings);
    399   return encodings->size() != deduped_encodings.size();
    400 }
    401