diff --git a/scripts/geodata/osm/fetch_osm_address_data.sh b/scripts/geodata/osm/fetch_osm_address_data.sh index 13fbe9d0..ac97dd37 100755 --- a/scripts/geodata/osm/fetch_osm_address_data.sh +++ b/scripts/geodata/osm/fetch_osm_address_data.sh @@ -45,6 +45,8 @@ echo "Converting to o5m: `date`" PLANET_PBF="planet-latest.osm.pbf" PLANET_O5M="planet-latest.o5m" +TOURISM_KEYS="tourism=hotel or tourism=attraction or tourism=guest_house or tourism=museum or tourism=chalet or tourism=motel or tourism=hostel or tourism=alpine_hut or tourism=theme_park or tourism=zoo or tourism=apartment or tourism=wilderness_hut or tourism=gallery or tourism=bed_and_breakfast or tourism=hanami or tourism=wine_cellar or tourism=resort or tourism=aquarium or tourism=apartments or tourism=cabin or tourism=winery or tourism=hut" + # Needs to be in O5M for some of the subsequent steps to work whereas PBF is smaller for download osmconvert $PLANET_PBF -o=$PLANET_O5M rm $PLANET_PBF @@ -52,12 +54,12 @@ rm $PLANET_PBF # Address data set for use in parser, language detection echo "Filtering for records with address tags: `date`" PLANET_ADDRESSES_O5M="planet-addresses.o5m" -osmfilter $PLANET_O5M --keep="( ( name= and ( amenity= or leisure= or tourism= ) ) or ( addr:street= and ( addr:housename= or addr:housenumber= ) ) )" --drop-author --drop-version -o=$PLANET_ADDRESSES_O5M +osmfilter $PLANET_O5M --keep="( ( name= and ( amenity= or leisure= or $TOURISM_KEYS ) ) or ( addr:street= and ( addr:housename= or addr:housenumber= ) ) )" --drop-author --drop-version -o=$PLANET_ADDRESSES_O5M PLANET_ADDRESSES_LATLONS="planet-addresses-latlons.o5m" osmconvert $PLANET_ADDRESSES_O5M --max-objects=1000000000 --all-to-nodes -o=$PLANET_ADDRESSES_LATLONS rm $PLANET_ADDRESSES_O5M PLANET_ADDRESSES="planet-addresses.osm" -osmfilter $PLANET_ADDRESSES_LATLONS --keep="( ( name= and ( amenity= or leisure= or tourism= ) ) or ( addr:street= and ( addr:housename= or addr:housenumber= ) ) )" -o=$PLANET_ADDRESSES +osmfilter $PLANET_ADDRESSES_LATLONS --keep="( ( name= and ( amenity= or leisure= or $TOURISM_KEYS ) ) or ( addr:street= and ( addr:housename= or addr:housenumber= ) ) )" -o=$PLANET_ADDRESSES rm $PLANET_ADDRESSES_LATLONS # Border data set for use in R-tree index/reverse geocoding, parsing, language detection @@ -80,12 +82,12 @@ osmfilter $PLANET_O5M --keep="name= and ( place=neighbourhood or place=suburb or # Venue data set for use in venue classification echo "Filtering for venue records: `date`" PLANET_VENUES_O5M="planet-venues.o5m" -osmfilter $PLANET_O5M --keep="name= and ( amenity= or building= or leisure= or tourism= )" --drop-author --drop-version -o=$PLANET_VENUES_O5M +osmfilter $PLANET_O5M --keep="name= and ( amenity= or building= or leisure= or $TOURISM_KEYS )" --drop-author --drop-version -o=$PLANET_VENUES_O5M PLANET_VENUES_LATLONS="planet-venues-latlons.o5m" osmconvert $PLANET_VENUES_O5M --max-objects=1000000000 --all-to-nodes -o=$PLANET_VENUES_LATLONS rm $PLANET_VENUES_O5M PLANET_VENUES="planet-venues.osm" -osmfilter $PLANET_VENUES_LATLONS --keep="name= and ( amenity= or building= or leisure= or tourism= )" -o=$PLANET_VENUES +osmfilter $PLANET_VENUES_LATLONS --keep="name= and ( amenity= or building= or leisure= or $TOURISM_KEYS )" -o=$PLANET_VENUES rm $PLANET_VENUES_LATLONS # Streets data set for use in language classification