Home | History | Annotate | Download | only in search_engines
      1 // Copyright 2014 The Chromium Authors. All rights reserved.
      2 // Use of this source code is governed by a BSD-style license that can be
      3 // found in the LICENSE file.
      4 
      5 #include "components/search_engines/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 "components/search_engines/template_url.h"
     17 #include "components/search_engines/template_url_prepopulate_data.h"
     18 #include "components/search_engines/template_url_service.h"
     19 
     20 base::string16 GetDefaultSearchEngineName(TemplateURLService* service) {
     21   DCHECK(service);
     22   const TemplateURL* const default_provider =
     23       service->GetDefaultSearchProvider();
     24   if (!default_provider) {
     25     // TODO(cpu): bug 1187517. It is possible to have no default provider.
     26     // returning an empty string is a stopgap measure for the crash
     27     // http://code.google.com/p/chromium/issues/detail?id=2573
     28     return base::string16();
     29   }
     30   return default_provider->short_name();
     31 }
     32 
     33 GURL GetDefaultSearchURLForSearchTerms(TemplateURLService* service,
     34                                        const base::string16& terms) {
     35   DCHECK(service);
     36   const TemplateURL* default_provider = service->GetDefaultSearchProvider();
     37   if (!default_provider)
     38     return GURL();
     39   const TemplateURLRef& search_url = default_provider->url_ref();
     40   DCHECK(search_url.SupportsReplacement(service->search_terms_data()));
     41   TemplateURLRef::SearchTermsArgs search_terms_args(terms);
     42   search_terms_args.append_extra_query_params = true;
     43   return GURL(search_url.ReplaceSearchTerms(search_terms_args,
     44                                             service->search_terms_data()));
     45 }
     46 
     47 void RemoveDuplicatePrepopulateIDs(
     48     KeywordWebDataService* service,
     49     const ScopedVector<TemplateURLData>& prepopulated_urls,
     50     TemplateURL* default_search_provider,
     51     TemplateURLService::TemplateURLVector* template_urls,
     52     const SearchTermsData& search_terms_data,
     53     std::set<std::string>* removed_keyword_guids) {
     54   DCHECK(template_urls);
     55 
     56   // For convenience construct an ID->TemplateURL* map from |prepopulated_urls|.
     57   typedef std::map<int, TemplateURLData*> PrepopulatedURLMap;
     58   PrepopulatedURLMap prepopulated_url_map;
     59   for (std::vector<TemplateURLData*>::const_iterator i(
     60            prepopulated_urls.begin());
     61        i != prepopulated_urls.end();
     62        ++i)
     63     prepopulated_url_map[(*i)->prepopulate_id] = *i;
     64 
     65   // Separate |template_urls| into prepopulated and non-prepopulated groups.
     66   typedef std::multimap<int, TemplateURL*> UncheckedURLMap;
     67   UncheckedURLMap unchecked_urls;
     68   TemplateURLService::TemplateURLVector checked_urls;
     69   for (TemplateURLService::TemplateURLVector::iterator i(
     70        template_urls->begin()); i != template_urls->end(); ++i) {
     71     TemplateURL* turl = *i;
     72     int prepopulate_id = turl->prepopulate_id();
     73     if (prepopulate_id)
     74       unchecked_urls.insert(std::make_pair(prepopulate_id, turl));
     75     else
     76       checked_urls.push_back(turl);
     77   }
     78 
     79   // For each group of prepopulated URLs with one ID, find the best URL to use
     80   // and add it to the (initially all non-prepopulated) URLs we've already OKed.
     81   // Delete the others from the service and from memory.
     82   while (!unchecked_urls.empty()) {
     83     // Find the best URL.
     84     int prepopulate_id = unchecked_urls.begin()->first;
     85     PrepopulatedURLMap::const_iterator prepopulated_url =
     86         prepopulated_url_map.find(prepopulate_id);
     87     UncheckedURLMap::iterator end = unchecked_urls.upper_bound(prepopulate_id);
     88     UncheckedURLMap::iterator best = unchecked_urls.begin();
     89     bool matched_keyword = false;
     90     for (UncheckedURLMap::iterator i = unchecked_urls.begin(); i != end; ++i) {
     91       // If the user-selected DSE is a prepopulated engine its properties will
     92       // either come from the prepopulation origin or from the user preferences
     93       // file (see DefaultSearchManager). Those properties will end up
     94       // overwriting whatever we load now anyway. If we are eliminating
     95       // duplicates, then, we err on the side of keeping the thing that looks
     96       // more like the value we will end up with in the end.
     97       if (default_search_provider &&
     98           (default_search_provider->prepopulate_id() ==
     99               i->second->prepopulate_id()) &&
    100           default_search_provider->HasSameKeywordAs(i->second->data(),
    101                                                     search_terms_data)) {
    102         best = i;
    103         break;
    104       }
    105 
    106       // Otherwise, a URL is best if it matches the prepopulated data's keyword;
    107       // if none match, just fall back to using the one with the lowest ID.
    108       if (matched_keyword)
    109         continue;
    110       if ((prepopulated_url != prepopulated_url_map.end()) &&
    111           i->second->HasSameKeywordAs(*prepopulated_url->second,
    112                                       search_terms_data)) {
    113         best = i;
    114         matched_keyword = true;
    115       } else if (i->second->id() < best->second->id()) {
    116         best = i;
    117       }
    118     }
    119 
    120     // Add the best URL to the checked group and delete the rest.
    121     checked_urls.push_back(best->second);
    122     for (UncheckedURLMap::iterator i = unchecked_urls.begin(); i != end; ++i) {
    123       if (i == best)
    124         continue;
    125       if (service) {
    126         service->RemoveKeyword(i->second->id());
    127         if (removed_keyword_guids)
    128           removed_keyword_guids->insert(i->second->sync_guid());
    129       }
    130       delete i->second;
    131     }
    132 
    133     // Done with this group.
    134     unchecked_urls.erase(unchecked_urls.begin(), end);
    135   }
    136 
    137   // Return the checked URLs.
    138   template_urls->swap(checked_urls);
    139 }
    140 
    141 // Returns the TemplateURL with id specified from the list of TemplateURLs.
    142 // If not found, returns NULL.
    143 TemplateURL* GetTemplateURLByID(
    144     const TemplateURLService::TemplateURLVector& template_urls,
    145     int64 id) {
    146   for (TemplateURLService::TemplateURLVector::const_iterator i(
    147        template_urls.begin()); i != template_urls.end(); ++i) {
    148     if ((*i)->id() == id) {
    149       return *i;
    150     }
    151   }
    152   return NULL;
    153 }
    154 
    155 TemplateURL* FindURLByPrepopulateID(
    156     const TemplateURLService::TemplateURLVector& template_urls,
    157     int prepopulate_id) {
    158   for (std::vector<TemplateURL*>::const_iterator i = template_urls.begin();
    159        i < template_urls.end(); ++i) {
    160     if ((*i)->prepopulate_id() == prepopulate_id)
    161       return *i;
    162   }
    163   return NULL;
    164 }
    165 
    166 void MergeIntoPrepopulatedEngineData(const TemplateURL* original_turl,
    167                                      TemplateURLData* prepopulated_url) {
    168   DCHECK_EQ(original_turl->prepopulate_id(), prepopulated_url->prepopulate_id);
    169   if (!original_turl->safe_for_autoreplace()) {
    170     prepopulated_url->safe_for_autoreplace = false;
    171     prepopulated_url->SetKeyword(original_turl->keyword());
    172     prepopulated_url->short_name = original_turl->short_name();
    173   }
    174   prepopulated_url->id = original_turl->id();
    175   prepopulated_url->sync_guid = original_turl->sync_guid();
    176   prepopulated_url->date_created = original_turl->date_created();
    177   prepopulated_url->last_modified = original_turl->last_modified();
    178 }
    179 
    180 ActionsFromPrepopulateData::ActionsFromPrepopulateData() {}
    181 
    182 ActionsFromPrepopulateData::~ActionsFromPrepopulateData() {}
    183 
    184 // This is invoked when the version of the prepopulate data changes.
    185 // If |removed_keyword_guids| is not NULL, the Sync GUID of each item removed
    186 // from the DB will be added to it.  Note that this function will take
    187 // ownership of |prepopulated_urls| and will clear the vector.
    188 void MergeEnginesFromPrepopulateData(
    189     KeywordWebDataService* service,
    190     ScopedVector<TemplateURLData>* prepopulated_urls,
    191     size_t default_search_index,
    192     TemplateURLService::TemplateURLVector* template_urls,
    193     TemplateURL* default_search_provider,
    194     std::set<std::string>* removed_keyword_guids) {
    195   DCHECK(prepopulated_urls);
    196   DCHECK(template_urls);
    197 
    198   ActionsFromPrepopulateData actions(CreateActionsFromCurrentPrepopulateData(
    199       prepopulated_urls, *template_urls, default_search_provider));
    200 
    201   // Remove items.
    202   for (std::vector<TemplateURL*>::iterator i = actions.removed_engines.begin();
    203        i < actions.removed_engines.end(); ++i) {
    204     scoped_ptr<TemplateURL> template_url(*i);
    205     TemplateURLService::TemplateURLVector::iterator j =
    206         std::find(template_urls->begin(), template_urls->end(), template_url);
    207     DCHECK(j != template_urls->end());
    208     DCHECK(!default_search_provider ||
    209            (*j)->prepopulate_id() != default_search_provider->prepopulate_id());
    210     template_urls->erase(j);
    211     if (service) {
    212       service->RemoveKeyword(template_url->id());
    213       if (removed_keyword_guids)
    214         removed_keyword_guids->insert(template_url->sync_guid());
    215     }
    216   }
    217 
    218   // Edit items.
    219   for (EditedEngines::iterator i(actions.edited_engines.begin());
    220        i < actions.edited_engines.end(); ++i) {
    221     TemplateURLData& data = i->second;
    222     scoped_ptr<TemplateURL> existing_url(i->first);
    223     if (service)
    224       service->UpdateKeyword(data);
    225 
    226     // Replace the entry in |template_urls| with the updated one.
    227     TemplateURLService::TemplateURLVector::iterator j = std::find(
    228         template_urls->begin(), template_urls->end(), existing_url.get());
    229     *j = new TemplateURL(data);
    230   }
    231 
    232   // Add items.
    233   for (std::vector<TemplateURLData>::const_iterator it =
    234            actions.added_engines.begin();
    235        it != actions.added_engines.end();
    236        ++it) {
    237     template_urls->push_back(new TemplateURL(*it));
    238   }
    239 }
    240 
    241 ActionsFromPrepopulateData CreateActionsFromCurrentPrepopulateData(
    242     ScopedVector<TemplateURLData>* prepopulated_urls,
    243     const TemplateURLService::TemplateURLVector& existing_urls,
    244     const TemplateURL* default_search_provider) {
    245   // Create a map to hold all provided |template_urls| that originally came from
    246   // prepopulate data (i.e. have a non-zero prepopulate_id()).
    247   typedef std::map<int, TemplateURL*> IDMap;
    248   IDMap id_to_turl;
    249   for (TemplateURLService::TemplateURLVector::const_iterator i(
    250        existing_urls.begin()); i != existing_urls.end(); ++i) {
    251     int prepopulate_id = (*i)->prepopulate_id();
    252     if (prepopulate_id > 0)
    253       id_to_turl[prepopulate_id] = *i;
    254   }
    255 
    256   // For each current prepopulated URL, check whether |template_urls| contained
    257   // a matching prepopulated URL.  If so, update the passed-in URL to match the
    258   // current data.  (If the passed-in URL was user-edited, we persist the user's
    259   // name and keyword.)  If not, add the prepopulated URL.
    260   ActionsFromPrepopulateData actions;
    261   for (size_t i = 0; i < prepopulated_urls->size(); ++i) {
    262     // We take ownership of |prepopulated_urls[i]|.
    263     scoped_ptr<TemplateURLData> prepopulated_url((*prepopulated_urls)[i]);
    264     const int prepopulated_id = prepopulated_url->prepopulate_id;
    265     DCHECK_NE(0, prepopulated_id);
    266 
    267     IDMap::iterator existing_url_iter(id_to_turl.find(prepopulated_id));
    268     if (existing_url_iter != id_to_turl.end()) {
    269       // Update the data store with the new prepopulated data. Preserve user
    270       // edits to the name and keyword.
    271       TemplateURL* existing_url(existing_url_iter->second);
    272       id_to_turl.erase(existing_url_iter);
    273       MergeIntoPrepopulatedEngineData(existing_url, prepopulated_url.get());
    274       // Update last_modified to ensure that if this entry is later merged with
    275       // entries from Sync, the conflict resolution logic knows that this was
    276       // updated and propagates the new values to the server.
    277       prepopulated_url->last_modified = base::Time::Now();
    278       actions.edited_engines.push_back(
    279           std::make_pair(existing_url, *prepopulated_url));
    280     } else {
    281       actions.added_engines.push_back(*prepopulated_url);
    282     }
    283   }
    284   // The above loop takes ownership of all the contents of prepopulated_urls.
    285   // Clear the pointers.
    286   prepopulated_urls->weak_erase(prepopulated_urls->begin(),
    287                                 prepopulated_urls->end());
    288 
    289   // The block above removed all the URLs from the |id_to_turl| map that were
    290   // found in the prepopulate data.  Any remaining URLs that haven't been
    291   // user-edited or made default can be removed from the data store.
    292   // We assume that this entry is equivalent to the DSE if its prepopulate ID
    293   // and keyword both match. If the prepopulate ID _does_ match all properties
    294   // will be replaced with those from |default_search_provider| anyway.
    295   for (IDMap::iterator i(id_to_turl.begin()); i != id_to_turl.end(); ++i) {
    296     TemplateURL* template_url = i->second;
    297     if ((template_url->safe_for_autoreplace()) &&
    298         (!default_search_provider ||
    299          (template_url->prepopulate_id() !=
    300              default_search_provider->prepopulate_id()) ||
    301          (template_url->keyword() != default_search_provider->keyword())))
    302       actions.removed_engines.push_back(template_url);
    303   }
    304 
    305   return actions;
    306 }
    307 
    308 void GetSearchProvidersUsingKeywordResult(
    309     const WDTypedResult& result,
    310     KeywordWebDataService* service,
    311     PrefService* prefs,
    312     TemplateURLService::TemplateURLVector* template_urls,
    313     TemplateURL* default_search_provider,
    314     const SearchTermsData& search_terms_data,
    315     int* new_resource_keyword_version,
    316     std::set<std::string>* removed_keyword_guids) {
    317   DCHECK(template_urls);
    318   DCHECK(template_urls->empty());
    319   DCHECK_EQ(KEYWORDS_RESULT, result.GetType());
    320   DCHECK(new_resource_keyword_version);
    321 
    322   WDKeywordsResult keyword_result = reinterpret_cast<
    323       const WDResult<WDKeywordsResult>*>(&result)->GetValue();
    324 
    325   for (KeywordTable::Keywords::iterator i(keyword_result.keywords.begin());
    326        i != keyword_result.keywords.end(); ++i) {
    327     // Fix any duplicate encodings in the local database.  Note that we don't
    328     // adjust the last_modified time of this keyword; this way, we won't later
    329     // overwrite any changes on the sync server that happened to this keyword
    330     // since the last time we synced.  Instead, we also run a de-duping pass on
    331     // the server-provided data in
    332     // TemplateURLService::CreateTemplateURLFromTemplateURLAndSyncData() and
    333     // update the server with the merged, de-duped results at that time.  We
    334     // still fix here, though, to correct problems in clients that have disabled
    335     // search engine sync, since in that case that code will never be reached.
    336     if (DeDupeEncodings(&i->input_encodings) && service)
    337       service->UpdateKeyword(*i);
    338     template_urls->push_back(new TemplateURL(*i));
    339   }
    340 
    341   *new_resource_keyword_version = keyword_result.builtin_keyword_version;
    342   GetSearchProvidersUsingLoadedEngines(service, prefs, template_urls,
    343                                        default_search_provider,
    344                                        search_terms_data,
    345                                        new_resource_keyword_version,
    346                                        removed_keyword_guids);
    347 }
    348 
    349 void GetSearchProvidersUsingLoadedEngines(
    350     KeywordWebDataService* service,
    351     PrefService* prefs,
    352     TemplateURLService::TemplateURLVector* template_urls,
    353     TemplateURL* default_search_provider,
    354     const SearchTermsData& search_terms_data,
    355     int* resource_keyword_version,
    356     std::set<std::string>* removed_keyword_guids) {
    357   DCHECK(template_urls);
    358   DCHECK(resource_keyword_version);
    359   size_t default_search_index;
    360   ScopedVector<TemplateURLData> prepopulated_urls =
    361       TemplateURLPrepopulateData::GetPrepopulatedEngines(prefs,
    362                                                          &default_search_index);
    363   RemoveDuplicatePrepopulateIDs(service, prepopulated_urls,
    364                                 default_search_provider, template_urls,
    365                                 search_terms_data, removed_keyword_guids);
    366 
    367   const int prepopulate_resource_keyword_version =
    368       TemplateURLPrepopulateData::GetDataVersion(prefs);
    369   if (*resource_keyword_version < prepopulate_resource_keyword_version) {
    370     MergeEnginesFromPrepopulateData(
    371         service, &prepopulated_urls, default_search_index, template_urls,
    372         default_search_provider, removed_keyword_guids);
    373     *resource_keyword_version = prepopulate_resource_keyword_version;
    374   } else {
    375     *resource_keyword_version = 0;
    376   }
    377 }
    378 
    379 bool DeDupeEncodings(std::vector<std::string>* encodings) {
    380   std::vector<std::string> deduped_encodings;
    381   std::set<std::string> encoding_set;
    382   for (std::vector<std::string>::const_iterator i(encodings->begin());
    383        i != encodings->end(); ++i) {
    384     if (encoding_set.insert(*i).second)
    385       deduped_encodings.push_back(*i);
    386   }
    387   encodings->swap(deduped_encodings);
    388   return encodings->size() != deduped_encodings.size();
    389 }
    390