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   TemplateURLService* service =
     46       TemplateURLServiceFactory::GetForProfile(profile);
     47   const TemplateURL* default_provider = service->GetDefaultSearchProvider();
     48   if (!default_provider)
     49     return GURL();
     50   const TemplateURLRef& search_url = default_provider->url_ref();
     51   DCHECK(search_url.SupportsReplacement(service->search_terms_data()));
     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                                             service->search_terms_data()));
     56 }
     57 
     58 void RemoveDuplicatePrepopulateIDs(
     59     WebDataService* service,
     60     const ScopedVector<TemplateURLData>& prepopulated_urls,
     61     TemplateURL* default_search_provider,
     62     TemplateURLService::TemplateURLVector* template_urls,
     63     const SearchTermsData& search_terms_data,
     64     std::set<std::string>* removed_keyword_guids) {
     65   DCHECK(service == NULL || BrowserThread::CurrentlyOn(BrowserThread::UI));
     66   DCHECK(template_urls);
     67 
     68   // For convenience construct an ID->TemplateURL* map from |prepopulated_urls|.
     69   typedef std::map<int, TemplateURLData*> PrepopulatedURLMap;
     70   PrepopulatedURLMap prepopulated_url_map;
     71   for (std::vector<TemplateURLData*>::const_iterator i(
     72            prepopulated_urls.begin());
     73        i != prepopulated_urls.end();
     74        ++i)
     75     prepopulated_url_map[(*i)->prepopulate_id] = *i;
     76 
     77   // Separate |template_urls| into prepopulated and non-prepopulated groups.
     78   typedef std::multimap<int, TemplateURL*> UncheckedURLMap;
     79   UncheckedURLMap unchecked_urls;
     80   TemplateURLService::TemplateURLVector checked_urls;
     81   for (TemplateURLService::TemplateURLVector::iterator i(
     82        template_urls->begin()); i != template_urls->end(); ++i) {
     83     TemplateURL* turl = *i;
     84     int prepopulate_id = turl->prepopulate_id();
     85     if (prepopulate_id)
     86       unchecked_urls.insert(std::make_pair(prepopulate_id, turl));
     87     else
     88       checked_urls.push_back(turl);
     89   }
     90 
     91   // For each group of prepopulated URLs with one ID, find the best URL to use
     92   // and add it to the (initially all non-prepopulated) URLs we've already OKed.
     93   // Delete the others from the service and from memory.
     94   while (!unchecked_urls.empty()) {
     95     // Find the best URL.
     96     int prepopulate_id = unchecked_urls.begin()->first;
     97     PrepopulatedURLMap::const_iterator prepopulated_url =
     98         prepopulated_url_map.find(prepopulate_id);
     99     UncheckedURLMap::iterator end = unchecked_urls.upper_bound(prepopulate_id);
    100     UncheckedURLMap::iterator best = unchecked_urls.begin();
    101     bool matched_keyword = false;
    102     for (UncheckedURLMap::iterator i = unchecked_urls.begin(); i != end; ++i) {
    103       // If the user-selected DSE is a prepopulated engine its properties will
    104       // either come from the prepopulation origin or from the user preferences
    105       // file (see DefaultSearchManager). Those properties will end up
    106       // overwriting whatever we load now anyway. If we are eliminating
    107       // duplicates, then, we err on the side of keeping the thing that looks
    108       // more like the value we will end up with in the end.
    109       if (default_search_provider &&
    110           (default_search_provider->prepopulate_id() ==
    111               i->second->prepopulate_id()) &&
    112           default_search_provider->HasSameKeywordAs(i->second->data(),
    113                                                     search_terms_data)) {
    114         best = i;
    115         break;
    116       }
    117 
    118       // Otherwise, a URL is best if it matches the prepopulated data's keyword;
    119       // if none match, just fall back to using the one with the lowest ID.
    120       if (matched_keyword)
    121         continue;
    122       if ((prepopulated_url != prepopulated_url_map.end()) &&
    123           i->second->HasSameKeywordAs(*prepopulated_url->second,
    124                                       search_terms_data)) {
    125         best = i;
    126         matched_keyword = true;
    127       } else if (i->second->id() < best->second->id()) {
    128         best = i;
    129       }
    130     }
    131 
    132     // Add the best URL to the checked group and delete the rest.
    133     checked_urls.push_back(best->second);
    134     for (UncheckedURLMap::iterator i = unchecked_urls.begin(); i != end; ++i) {
    135       if (i == best)
    136         continue;
    137       if (service) {
    138         service->RemoveKeyword(i->second->id());
    139         if (removed_keyword_guids)
    140           removed_keyword_guids->insert(i->second->sync_guid());
    141       }
    142       delete i->second;
    143     }
    144 
    145     // Done with this group.
    146     unchecked_urls.erase(unchecked_urls.begin(), end);
    147   }
    148 
    149   // Return the checked URLs.
    150   template_urls->swap(checked_urls);
    151 }
    152 
    153 // Returns the TemplateURL with id specified from the list of TemplateURLs.
    154 // If not found, returns NULL.
    155 TemplateURL* GetTemplateURLByID(
    156     const TemplateURLService::TemplateURLVector& template_urls,
    157     int64 id) {
    158   for (TemplateURLService::TemplateURLVector::const_iterator i(
    159        template_urls.begin()); i != template_urls.end(); ++i) {
    160     if ((*i)->id() == id) {
    161       return *i;
    162     }
    163   }
    164   return NULL;
    165 }
    166 
    167 TemplateURL* FindURLByPrepopulateID(
    168     const TemplateURLService::TemplateURLVector& template_urls,
    169     int prepopulate_id) {
    170   for (std::vector<TemplateURL*>::const_iterator i = template_urls.begin();
    171        i < template_urls.end(); ++i) {
    172     if ((*i)->prepopulate_id() == prepopulate_id)
    173       return *i;
    174   }
    175   return NULL;
    176 }
    177 
    178 void MergeIntoPrepopulatedEngineData(const TemplateURL* original_turl,
    179                                      TemplateURLData* prepopulated_url) {
    180   DCHECK_EQ(original_turl->prepopulate_id(), prepopulated_url->prepopulate_id);
    181   if (!original_turl->safe_for_autoreplace()) {
    182     prepopulated_url->safe_for_autoreplace = false;
    183     prepopulated_url->SetKeyword(original_turl->keyword());
    184     prepopulated_url->short_name = original_turl->short_name();
    185   }
    186   prepopulated_url->id = original_turl->id();
    187   prepopulated_url->sync_guid = original_turl->sync_guid();
    188   prepopulated_url->date_created = original_turl->date_created();
    189   prepopulated_url->last_modified = original_turl->last_modified();
    190 }
    191 
    192 ActionsFromPrepopulateData::ActionsFromPrepopulateData() {}
    193 
    194 ActionsFromPrepopulateData::~ActionsFromPrepopulateData() {}
    195 
    196 // This is invoked when the version of the prepopulate data changes.
    197 // If |removed_keyword_guids| is not NULL, the Sync GUID of each item removed
    198 // from the DB will be added to it.  Note that this function will take
    199 // ownership of |prepopulated_urls| and will clear the vector.
    200 void MergeEnginesFromPrepopulateData(
    201     Profile* profile,
    202     WebDataService* service,
    203     ScopedVector<TemplateURLData>* prepopulated_urls,
    204     size_t default_search_index,
    205     TemplateURLService::TemplateURLVector* template_urls,
    206     TemplateURL* default_search_provider,
    207     std::set<std::string>* removed_keyword_guids) {
    208   DCHECK(service == NULL || BrowserThread::CurrentlyOn(BrowserThread::UI));
    209   DCHECK(prepopulated_urls);
    210   DCHECK(template_urls);
    211 
    212   ActionsFromPrepopulateData actions(CreateActionsFromCurrentPrepopulateData(
    213       prepopulated_urls, *template_urls, default_search_provider));
    214 
    215   // Remove items.
    216   for (std::vector<TemplateURL*>::iterator i = actions.removed_engines.begin();
    217        i < actions.removed_engines.end(); ++i) {
    218     scoped_ptr<TemplateURL> template_url(*i);
    219     TemplateURLService::TemplateURLVector::iterator j =
    220         std::find(template_urls->begin(), template_urls->end(), template_url);
    221     DCHECK(j != template_urls->end());
    222     DCHECK(!default_search_provider ||
    223            (*j)->prepopulate_id() != default_search_provider->prepopulate_id());
    224     template_urls->erase(j);
    225     if (service) {
    226       service->RemoveKeyword(template_url->id());
    227       if (removed_keyword_guids)
    228         removed_keyword_guids->insert(template_url->sync_guid());
    229     }
    230   }
    231 
    232   // Edit items.
    233   for (EditedEngines::iterator i(actions.edited_engines.begin());
    234        i < actions.edited_engines.end(); ++i) {
    235     TemplateURLData& data = i->second;
    236     scoped_ptr<TemplateURL> existing_url(i->first);
    237     if (service)
    238       service->UpdateKeyword(data);
    239 
    240     // Replace the entry in |template_urls| with the updated one.
    241     TemplateURLService::TemplateURLVector::iterator j = std::find(
    242         template_urls->begin(), template_urls->end(), existing_url.get());
    243     *j = new TemplateURL(data);
    244   }
    245 
    246   // Add items.
    247   for (std::vector<TemplateURLData>::const_iterator it =
    248            actions.added_engines.begin();
    249        it != actions.added_engines.end();
    250        ++it) {
    251     template_urls->push_back(new TemplateURL(*it));
    252   }
    253 }
    254 
    255 ActionsFromPrepopulateData CreateActionsFromCurrentPrepopulateData(
    256     ScopedVector<TemplateURLData>* prepopulated_urls,
    257     const TemplateURLService::TemplateURLVector& existing_urls,
    258     const TemplateURL* default_search_provider) {
    259   // Create a map to hold all provided |template_urls| that originally came from
    260   // prepopulate data (i.e. have a non-zero prepopulate_id()).
    261   typedef std::map<int, TemplateURL*> IDMap;
    262   IDMap id_to_turl;
    263   for (TemplateURLService::TemplateURLVector::const_iterator i(
    264        existing_urls.begin()); i != existing_urls.end(); ++i) {
    265     int prepopulate_id = (*i)->prepopulate_id();
    266     if (prepopulate_id > 0)
    267       id_to_turl[prepopulate_id] = *i;
    268   }
    269 
    270   // For each current prepopulated URL, check whether |template_urls| contained
    271   // a matching prepopulated URL.  If so, update the passed-in URL to match the
    272   // current data.  (If the passed-in URL was user-edited, we persist the user's
    273   // name and keyword.)  If not, add the prepopulated URL.
    274   ActionsFromPrepopulateData actions;
    275   for (size_t i = 0; i < prepopulated_urls->size(); ++i) {
    276     // We take ownership of |prepopulated_urls[i]|.
    277     scoped_ptr<TemplateURLData> prepopulated_url((*prepopulated_urls)[i]);
    278     const int prepopulated_id = prepopulated_url->prepopulate_id;
    279     DCHECK_NE(0, prepopulated_id);
    280 
    281     IDMap::iterator existing_url_iter(id_to_turl.find(prepopulated_id));
    282     if (existing_url_iter != id_to_turl.end()) {
    283       // Update the data store with the new prepopulated data. Preserve user
    284       // edits to the name and keyword.
    285       TemplateURL* existing_url(existing_url_iter->second);
    286       id_to_turl.erase(existing_url_iter);
    287       MergeIntoPrepopulatedEngineData(existing_url, prepopulated_url.get());
    288       // Update last_modified to ensure that if this entry is later merged with
    289       // entries from Sync, the conflict resolution logic knows that this was
    290       // updated and propagates the new values to the server.
    291       prepopulated_url->last_modified = base::Time::Now();
    292       actions.edited_engines.push_back(
    293           std::make_pair(existing_url, *prepopulated_url));
    294     } else {
    295       actions.added_engines.push_back(*prepopulated_url);
    296     }
    297   }
    298   // The above loop takes ownership of all the contents of prepopulated_urls.
    299   // Clear the pointers.
    300   prepopulated_urls->weak_erase(prepopulated_urls->begin(),
    301                                 prepopulated_urls->end());
    302 
    303   // The block above removed all the URLs from the |id_to_turl| map that were
    304   // found in the prepopulate data.  Any remaining URLs that haven't been
    305   // user-edited or made default can be removed from the data store.
    306   // We assume that this entry is equivalent to the DSE if its prepopulate ID
    307   // and keyword both match. If the prepopulate ID _does_ match all properties
    308   // will be replaced with those from |default_search_provider| anyway.
    309   for (IDMap::iterator i(id_to_turl.begin()); i != id_to_turl.end(); ++i) {
    310     TemplateURL* template_url = i->second;
    311     if ((template_url->safe_for_autoreplace()) &&
    312         (!default_search_provider ||
    313          (template_url->prepopulate_id() !=
    314              default_search_provider->prepopulate_id()) ||
    315          (template_url->keyword() != default_search_provider->keyword())))
    316       actions.removed_engines.push_back(template_url);
    317   }
    318 
    319   return actions;
    320 }
    321 
    322 void GetSearchProvidersUsingKeywordResult(
    323     const WDTypedResult& result,
    324     WebDataService* service,
    325     Profile* profile,
    326     TemplateURLService::TemplateURLVector* template_urls,
    327     TemplateURL* default_search_provider,
    328     const SearchTermsData& search_terms_data,
    329     int* new_resource_keyword_version,
    330     std::set<std::string>* removed_keyword_guids) {
    331   DCHECK(service == NULL || BrowserThread::CurrentlyOn(BrowserThread::UI));
    332   DCHECK(template_urls);
    333   DCHECK(template_urls->empty());
    334   DCHECK_EQ(KEYWORDS_RESULT, result.GetType());
    335   DCHECK(new_resource_keyword_version);
    336 
    337   WDKeywordsResult keyword_result = reinterpret_cast<
    338       const WDResult<WDKeywordsResult>*>(&result)->GetValue();
    339 
    340   for (KeywordTable::Keywords::iterator i(keyword_result.keywords.begin());
    341        i != keyword_result.keywords.end(); ++i) {
    342     // Fix any duplicate encodings in the local database.  Note that we don't
    343     // adjust the last_modified time of this keyword; this way, we won't later
    344     // overwrite any changes on the sync server that happened to this keyword
    345     // since the last time we synced.  Instead, we also run a de-duping pass on
    346     // the server-provided data in
    347     // TemplateURLService::CreateTemplateURLFromTemplateURLAndSyncData() and
    348     // update the server with the merged, de-duped results at that time.  We
    349     // still fix here, though, to correct problems in clients that have disabled
    350     // search engine sync, since in that case that code will never be reached.
    351     if (DeDupeEncodings(&i->input_encodings) && service)
    352       service->UpdateKeyword(*i);
    353     template_urls->push_back(new TemplateURL(*i));
    354   }
    355 
    356   *new_resource_keyword_version = keyword_result.builtin_keyword_version;
    357   GetSearchProvidersUsingLoadedEngines(service, profile, template_urls,
    358                                        default_search_provider,
    359                                        search_terms_data,
    360                                        new_resource_keyword_version,
    361                                        removed_keyword_guids);
    362 }
    363 
    364 void GetSearchProvidersUsingLoadedEngines(
    365     WebDataService* service,
    366     Profile* profile,
    367     TemplateURLService::TemplateURLVector* template_urls,
    368     TemplateURL* default_search_provider,
    369     const SearchTermsData& search_terms_data,
    370     int* resource_keyword_version,
    371     std::set<std::string>* removed_keyword_guids) {
    372   DCHECK(service == NULL || BrowserThread::CurrentlyOn(BrowserThread::UI));
    373   DCHECK(template_urls);
    374   DCHECK(resource_keyword_version);
    375   PrefService* prefs = profile ? profile->GetPrefs() : NULL;
    376   size_t default_search_index;
    377   ScopedVector<TemplateURLData> prepopulated_urls =
    378       TemplateURLPrepopulateData::GetPrepopulatedEngines(prefs,
    379                                                          &default_search_index);
    380   RemoveDuplicatePrepopulateIDs(service, prepopulated_urls,
    381                                 default_search_provider, template_urls,
    382                                 search_terms_data, removed_keyword_guids);
    383 
    384   const int prepopulate_resource_keyword_version =
    385       TemplateURLPrepopulateData::GetDataVersion(prefs);
    386   if (*resource_keyword_version < prepopulate_resource_keyword_version) {
    387     MergeEnginesFromPrepopulateData(profile, service, &prepopulated_urls,
    388         default_search_index, template_urls, default_search_provider,
    389         removed_keyword_guids);
    390     *resource_keyword_version = prepopulate_resource_keyword_version;
    391   } else {
    392     *resource_keyword_version = 0;
    393   }
    394 }
    395 
    396 bool DeDupeEncodings(std::vector<std::string>* encodings) {
    397   std::vector<std::string> deduped_encodings;
    398   std::set<std::string> encoding_set;
    399   for (std::vector<std::string>::const_iterator i(encodings->begin());
    400        i != encodings->end(); ++i) {
    401     if (encoding_set.insert(*i).second)
    402       deduped_encodings.push_back(*i);
    403   }
    404   encodings->swap(deduped_encodings);
    405   return encodings->size() != deduped_encodings.size();
    406 }
    407