[booking] Add translated name and address

generator/booking_dataset.cpp

@@ -50,6 +50,13 @@ BookingDataset::Hotel::Hotel(string const & src)
   descUrl = rec[Index(Fields::DescUrl)];
 
   strings::to_uint(rec[Index(Fields::Type)], type);
+
+  langCode = rec[Index(Fields::Language)];
+  if (!langCode.empty())
+  {
+     nameLoc = rec[Index(Fields::NameLoc)];
+     addressLoc = rec[Index(Fields::AddressLoc)];
+  }
 }
 
 ostream & operator<<(ostream & s, BookingDataset::Hotel const & h)
@@ -171,6 +178,12 @@ void BookingDataset::BuildFeatures(function<void(OsmElement *)> const & fn) cons
     e.AddTag("price_rate", strings::to_string(hotel.priceCategory));
     e.AddTag("addr:full", hotel.address);
 
+    if (!hotel.langCode.empty())
+    {
+      e.AddTag("name:" + hotel.langCode, hotel.nameLoc);
+      e.AddTag("addr:full:" + hotel.langCode, hotel.addressLoc);
+    }
+
     switch (hotel.type)
     {
     case 19:

generator/booking_dataset.hpp

@@ -36,6 +36,9 @@ public:
       RatingUsers = 8,
       DescUrl = 9,
       Type = 10,
+      Language = 11,
+      NameLoc = 12,
+      AddressLoc = 13,
 
       Counter
     };
@@ -51,6 +54,9 @@ public:
     double ratingUser = 0.0;
     string descUrl;
     uint32_t type = 0;
+    string langCode;
+    string nameLoc;
+    string addressLoc;
 
     static constexpr size_t Index(Fields field) { return static_cast<size_t>(field); }
     static constexpr size_t FieldsCount() { return static_cast<size_t>(Fields::Counter); }

tools/python/booking_hotels.py

@@ -81,26 +81,45 @@ def download(user, password, path):
         countrycode = country['countrycode']
         logging.info(u'Download[{0}]: {1}'.format(countrycode, country['name']))
 
-        allhotels = []
+        allhotels = {}
         while True:
             hotels = api.call('getHotels',
                               dict(new_hotel_type=1, offset=len(allhotels), rows=maxrows, countrycodes=countrycode))
 
             # Check for error.
-            if not hotels:
+            if hotels is None:
                 exit(1)
 
-            allhotels.extend(hotels)
+            for h in hotels:
+                allhotels[h['hotel_id']] = h
 
             # If hotels in answer less then maxrows, we reach end of data.
             if len(hotels) < maxrows:
                 break
 
-        logging.info('Num of hotels: {0}'.format(len(allhotels)))
+        # Now the same for hotel translations
+        offset = 0
+        while True:
+            hotels = api.call('getHotelTranslations', dict(offset=offset, rows=maxrows, countrycodes=countrycode))
+            if hotels is None:
+                exit(1)
+
+            # Add translations for each hotel
+            for h in hotels:
+                if h['hotel_id'] in allhotels:
+                    if 'translations' not in allhotels[h['hotel_id']]:
+                        allhotels[h['hotel_id']]['translations'] = {}
+                    allhotels[h['hotel_id']]['translations'][h['languagecode']] = {'name': h['name'], 'address': h['address']}
+
+            offset += len(hotels)
+            if len(hotels) < maxrows:
+                break
+
+        logging.info('Num of hotels: {0}, translations: {1}'.format(len(allhotels), offset))
         filename = os.path.join(path,
                                 '{0} - {1}.pkl'.format(country['area'].encode('utf8'), country['name'].encode('utf8')))
         with open(filename, 'wb') as fd:
-            pickle.dump(allhotels, fd, pickle.HIGHEST_PROTOCOL)
+            pickle.dump(allhotels.values(), fd, pickle.HIGHEST_PROTOCOL)
 
 
 def translate(source, output):
@@ -110,7 +129,7 @@ def translate(source, output):
     files = [filename for filename in os.listdir(source) if filename.endswith('.pkl')]
 
     data = []
-    for filename in files:
+    for filename in sorted(files):
         logging.info('Processing {0}'.format(filename))
         with open(filename, 'rb') as fd:
             data += pickle.load(fd)
@@ -155,8 +174,17 @@ def translate(source, output):
                 # Find a range that contains the price
                 while rate <= len(rates) and price > avg * rates[rate - 1]:
                     rate += 1
-            l = [unicode(get_hotel_field(hotel, e, rate)).encode('utf8').replace('\t', ' ') for e in HOTEL_FIELDS]
-            print('\t'.join(l), file=fd)
+            l = [get_hotel_field(hotel, e, rate) for e in HOTEL_FIELDS]
+            # Add translations for hotel name and address if present.
+            if 'translations' in hotel:
+                tr_lang = hotel['languagecode']
+                if tr_lang not in hotel['translations']:
+                    tr_lang = hotel['translations'].keys()[0]
+                l.append(tr_lang)
+                l.extend([hotel['translations'][tr_lang][e] for e in ('name', 'address')])
+            else:
+                l.extend([''] * 3)
+            print('\t'.join([unicode(f).encode('utf8').replace('\t', ' ') for f in l]), file=fd)
 
 
 def process_options():