pelias-openstreetmap 5.13.0

package/config/category_map.js

@@ -0,0 +1,360 @@
+
+/**
+ default category mapping, openstreetmap features on the left
+ correspond to the Pelias taxonomy on the right.
+
+ you can modify this file to suit your specific use-case, or
+ alternatively you can inject your own custom taxonomy mapping
+ at runtime.
+
+ A special key '*' is used to match any tag value for that key.
+ eg. 'aerialway:*' would match 'aerialway:foo' and 'aerialway:bar'.
+
+ Categories are cumulative, so if a document matches on many different
+ mappings then it will inherit *all* of those categories.
+
+ @see: http://wiki.openstreetmap.org/wiki/Map_Features
+ @see: https://github.com/pelias/pelias/wiki/Taxonomy-v1
+**/
+
+const mapping = {
+
+  'aerialway': {
+    '*':                        ['transport'],
+  },
+
+  'aeroway': {
+    '*':                        ['transport','transport:air'],
+    'aerodrome':                ['transport','transport:air','transport:air:aerodrome'],
+    'international':            ['transport','transport:air','transport:air:airport']
+  },
+
+  'amenity': {
+
+    'bbq':                      ['recreation'],
+    'dojo':                     ['recreation'],
+    'gym':                      ['recreation'],
+
+    'place_of_worship':         ['religion'],
+
+    'arts_centre':              ['education','entertainment'],
+    'community_centre':         ['education','entertainment'],
+    'social_centre':            ['education','entertainment'],
+    'library':                  ['education','entertainment'],
+    'planetarium':              ['education','entertainment'],
+    'theatre':                  ['education','entertainment'],
+    'college':                  ['education'],
+    'kindergarten':             ['education'],
+    'school':                   ['education'],
+    'university':               ['education'],
+
+    'bar':                      ['nightlife'],
+    'biergarten':               ['nightlife','food'],
+    'cinema':                   ['entertainment','nightlife'],
+    'casino':                   ['nightlife'],
+    'gambling':                 ['nightlife'],
+    'nightclub':                ['nightlife'],
+    'pub':                      ['nightlife'],
+
+    'courthouse':               ['government'],
+    'embassy':                  ['government'],
+    'fire_station':             ['government'],
+    'police':                   ['government'],
+    'post_office':              ['government'],
+    'ranger_station':           ['government','recreation'],
+    'register_office':          ['government'],
+    'townhall':                 ['government'],
+
+    'coworking_space':          ['professional'],
+
+    'atm':                      ['finance'],
+    'bank':                     ['finance','professional'],
+    'bureau_de_change':         ['finance','professional'],
+
+    'clinic':                   ['health'],
+    'dentist':                  ['health'],
+    'doctors':                  ['health'],
+    'hospital':                 ['health'],
+    'nursing_home':             ['health'],
+    'pharmacy':                 ['health'],
+    'social_facility':          ['health'],
+    'veterinary':               ['professional'],
+
+    'cafe':                     ['food','retail'],
+    'fast_food':                ['food','retail'],
+    'food_court':               ['food','retail'],
+    'ice_cream':                ['food','retail'],
+    'marketplace':              ['food','retail'],
+    'restaurant':               ['food','retail','nightlife'],
+
+    'bus_station':              ['transport','transport:public','transport:bus'],
+    'taxi':                     ['transport','transport:taxi'],
+
+    'car_rental':               ['transport','professional'],
+    'car_wash':                 ['professional'],
+    'charging_station':         ['transport','professional'],
+    'fuel':                     ['transport','professional'],
+
+    'ferry_terminal':           ['transport','transport:sea']
+  },
+
+  'building': {
+    'hotel':                    ['accommodation'],
+    'commercial':               ['professional'],
+    'retail':                   ['retail'],
+
+    'chapel':                   ['religion'],
+    'church':                   ['religion'],
+    'mosque':                   ['religion'],
+    'temple':                   ['religion'],
+    'synagogue':                ['religion'],
+    'shrine':                   ['religion'],
+
+    'civic':                    ['government'],
+    'hospital':                 ['health'],
+    'school':                   ['education'],
+    'stadium':                  ['entertainment'],
+    'university':               ['education'],
+    'public':                   ['government'],
+
+    'farm':                     ['industry','industry:agriculture'],
+
+    'train_station':            ['transport','transport:station'],
+    'transportation':           ['transport','transport:station']
+  },
+
+  // experimental, import cuisines for food-related venues
+  'cuisine': {
+    '*':                        ['food'],
+
+    // menu focus
+    'bagel':                    ['food:bagel'],
+    'barbecue':                 ['food:barbecue'],
+    'bougatsa':                 ['food:bougatsa'],
+    'burger':                   ['food:burger'],
+    'burrito':                  ['food:burrito'],
+    'cake':                     ['food:cake'],
+    'casserole':                ['food:casserole'],
+    'chicken':                  ['food:chicken'],
+    'coffee_shop':              ['food:coffee_shop'],
+    'crepe':                    ['food:crepe'],
+    'couscous':                 ['food:couscous'],
+    'curry':                    ['food:curry'],
+    'dessert':                  ['food:dessert'],
+    'donut':                    ['food:donut'],
+    'doughnut':                 ['food:donut'],
+    'empanada':                 ['food:empanada'],
+    'fish':                     ['food:fish'],
+    'fish_and_chips':           ['food:fish_and_chips'],
+    'fried_food':               ['food:fried_food'],
+    'friture':                  ['food:friture'],
+    'gyro':                     ['food:gyro'],
+    'ice_cream':                ['food:ice_cream'],
+    'kebab':                    ['food:kebab'],
+    'mediterranean':            ['food:mediterranean'],
+    'noodle':                   ['food:noodle'],
+    'pancake':                  ['food:pancake'],
+    'pasta':                    ['food:pasta'],
+    'pie':                      ['food:pie'],
+    'pizza':                    ['food:pizza'],
+    'regional':                 ['food:regional'],
+    'sandwich':                 ['food:sandwich'],
+    'sausage':                  ['food:sausage'],
+    'savory_pancakes':          ['food:savory_pancakes'],
+    'seafood':                  ['food:seafood'],
+    'steak_house':              ['food:steak'],
+    'sub':                      ['food:sub'],
+    'sushi':                    ['food:sushi'],
+    'tapas':                    ['food:tapas'],
+    'vegan':                    ['food:vegan'],
+    'vegetarian':               ['food:vegetarian'],
+    'wings':                    ['food:wings'],
+
+    // regional/cultural focus
+    'african':                  ['food:cuisine:african'],
+    'american':                 ['food:cuisine:american'],
+    'arab':                     ['food:cuisine:arab'],
+    'argentinian':              ['food:cuisine:argentinian'],
+    'asian':                    ['food:cuisine:asian'],
+    'australian':               ['food:cuisine:australian'],
+    'baiana':                   ['food:cuisine:baiana'],
+    'balkan':                   ['food:cuisine:balkan'],
+    'basque':                   ['food:cuisine:basque'],
+    'bavarian':                 ['food:cuisine:bavarian'],
+    'belarusian':               ['food:cuisine:belarusian'],
+    'brazilian':                ['food:cuisine:brazilian'],
+    'cantonese':                ['food:cuisine:cantonese'],
+    'capixaba':                 ['food:cuisine:capixaba'],
+    'caribbean':                ['food:cuisine:caribbean'],
+    'chinese':                  ['food:cuisine:chinese'],
+    'croatian':                 ['food:cuisine:croatian'],
+    'czech':                    ['food:cuisine:czech'],
+    'danish':                   ['food:cuisine:danish'],
+    'french':                   ['food:cuisine:french'],
+    'gaucho':                   ['food:cuisine:gaucho'],
+    'german':                   ['food:cuisine:german'],
+    'greek':                    ['food:cuisine:greek'],
+    'hunan':                    ['food:cuisine:hunan'],
+    'hungarian':                ['food:cuisine:hungarian'],
+    'indian':                   ['food:cuisine:indian'],
+    'international':            ['food:cuisine:international'],
+    'iranian':                  ['food:cuisine:iranian'],
+    'italian':                  ['food:cuisine:italian'],
+    'japanese':                 ['food:cuisine:japanese'],
+    'korean':                   ['food:cuisine:korean'],
+    'kyo_ryouri':               ['food:cuisine:kyo_ryouri'],
+    'latin_american':           ['food:cuisine:latin_american'],
+    'lebanese':                 ['food:cuisine:lebanese'],
+    'malagasy':                 ['food:cuisine:malagasy'],
+    'mexican':                  ['food:cuisine:mexican'],
+    'mineira':                  ['food:cuisine:mineira'],
+    'okinawa_ryori':            ['food:cuisine:okinawa_ryori'],
+    'pakistani':                ['food:cuisine:pakistani'],
+    'peruvian':                 ['food:cuisine:peruvian'],
+    'polish':                   ['food:cuisine:polish'],
+    'portuguese':               ['food:cuisine:portuguese'],
+    'rhenish':                  ['food:cuisine:rhenish'],
+    'russian':                  ['food:cuisine:russian'],
+    'shandong':                 ['food:cuisine:shandong'],
+    'sichuan':                  ['food:cuisine:sichuan'],
+    'spanish':                  ['food:cuisine:spanish'],
+    'thai':                     ['food:cuisine:thai'],
+    'turkish':                  ['food:cuisine:turkish'],
+    'vietnamese':               ['food:cuisine:vietnamese'],
+    'westphalian':              ['food:cuisine:westphalian']
+  },
+
+  'craft': {
+    '*':                        ['professional']
+  },
+
+  'emergency': {
+    'ambulance_station':        ['health','government']
+  },
+
+  'historic': {
+    'archaeological_site':      ['education'],
+    'monument':                 ['education']
+  },
+
+  'leisure': {
+    'adult_gaming_centre':      ['nightlife'],
+    'amusement_arcade':         ['entertainment'],
+    'beach_resort':             ['entertainment','recreation'],
+    'bandstand':                ['entertainment'],
+    'dance':                    ['nightlife'],
+
+    'dog_park':                 ['recreation'],
+    'fishing':                  ['recreation'],
+    'garden':                   ['recreation'],
+    'golf_course':              ['recreation','entertainment'],
+    'ice_rink':                 ['entertainment'],
+    'miniature_golf':           ['entertainment'],
+    'nature_reserve':           ['recreation'],
+    'park':                     ['recreation'],
+    'pitch':                    ['recreation','entertainment'],
+    'playground':               ['recreation'],
+    'sports_centre':            ['recreation','education','entertainment'],
+    'stadium':                  ['entertainment'],
+    'summer_camp':              ['recreation','education'],
+    'swimming_pool':            ['recreation'],
+    'track':                    ['recreation'],
+    'water_park':               ['entertainment'],
+
+    'hackerspace':              ['education','entertainment'],
+  },
+
+  'military': {
+    '*':                        ['government:military','government'],
+  },
+
+  'natural': {
+    'wood':                     ['natural','recreation'],
+    'water':                    ['natural','natural:water','recreation'],
+    'glacier':                  ['natural','recreation'],
+    'beach':                    ['natural','recreation'],
+  },
+
+  'office': {
+    '*':                        ['professional']
+  },
+
+  'public_transport': {
+    '*':                        ['transport','transport:public'],
+    'station':                  ['transport','transport:station']
+  },
+
+  'railway': {
+    'light_rail':               ['transport','transport:rail'],
+    'subway':                   ['transport','transport:rail'],
+    'tram':                     ['transport','transport:rail'],
+    'station':                  ['transport','transport:rail','transport:station']
+  },
+
+  'shop': {
+    '*':                        ['retail'],
+    'bakery':                   ['food'],
+    'butcher':                  ['food'],
+    'cheese':                   ['food'],
+    'chocolate':                ['food'],
+    'coffee':                   ['food'],
+    'deli':                     ['food'],
+    'greengrocer':              ['food'],
+    'seafood':                  ['food'],
+    'supermarket':              ['food'],
+
+    'tailor':                   ['professional'],
+    'copyshop':                 ['professional'],
+    'dry_cleaning':             ['professional'],
+
+    'chemist':                  ['health'],
+    'medical_supply':           ['health'],
+    'optician':                 ['health']
+  },
+
+  'sport': {
+    '*':                        ['recreation'],
+    'american_football':        ['entertainment'],
+    'australian_football':      ['entertainment'],
+    'badminton':                ['entertainment'],
+    'baseball':                 ['entertainment'],
+    'basketball':               ['entertainment'],
+    'beachvolleyball':          ['entertainment'],
+    'billiards':                ['entertainment'],
+    'canadian_football':        ['entertainment'],
+    'chess':                    ['entertainment'],
+    'cricket':                  ['entertainment'],
+    'dog_racing':               ['entertainment'],
+    'field_hockey':             ['entertainment'],
+    'gaelic_games':             ['entertainment'],
+    'horse_racing':             ['entertainment'],
+    'ice_hockey':               ['entertainment'],
+    'karting':                  ['entertainment'],
+    'rc_car':                   ['entertainment'],
+    'rugby_league':             ['entertainment'],
+    'rugby_union':              ['entertainment'],
+    'safety_training':          ['education']
+  },
+
+  'tourism': {
+    'hotel':                    ['accommodation'],
+    'motel':                    ['accommodation'],
+    'alpine_hut':               ['accommodation'],
+    'apartment':                ['accommodation'],
+    'guest_house':              ['accommodation'],
+    'chalet':                   ['accommodation'],
+    'caravan_site':             ['accommodation'],
+    'camp_site':                ['accommodation'],
+    'wilderness_hut':           ['accommodation'],
+    'information':              ['government'],
+    'attraction':               ['entertainment'],
+    'theme_park':               ['entertainment'],
+    'viewpoint':                ['recreation'],
+    'museum':                   ['education','entertainment'],
+    'gallery':                  ['education','entertainment'],
+    'zoo':                      ['education','entertainment']
+  }
+
+};
+
+module.exports = mapping;

package/config/features.js

@@ -0,0 +1,52 @@
+
+/**
+ default list of tags to extract from the pbf file when running
+ imports. @see: https://github.com/pelias/pbf2json for more info.
+**/
+
+// default tags imported
+const tags = [
+  'addr:housenumber+addr:street'
+];
+
+// tags corresponding to venues
+const venue_tags = [
+  'amenity+name',
+  'building+name',
+  'shop+name',
+  'office+name',
+  'public_transport+name',
+  'cuisine+name',
+  'railway~station+name',
+  'railway~tram_stop+name',
+  'railway~halt+name',
+  'railway~subway_entrance+name',
+  'railway~train_station_entrance+name',
+  'highway~pedestrian+area~yes+name',
+  'place~square+name',
+  'sport+name',
+  'natural+name',
+  'tourism+name',
+  'leisure+name',
+  'historic+name',
+  'man_made+name',
+  'landuse+name',
+  'waterway+name',
+  'aerialway+name',
+  'craft+name',
+  'military+name',
+  'aeroway~terminal+name',
+  'aeroway~aerodrome+name',
+  'aeroway~helipad+name',
+  'aeroway~airstrip+name',
+  'aeroway~heliport+name',
+  'aeroway~areodrome+name',
+  'aeroway~spaceport+name',
+  'aeroway~landing_strip+name',
+  'aeroway~airfield+name',
+  'aeroway~airport+name',
+  'brand+name',
+  'healthcare+name'
+];
+
+module.exports = {tags,venue_tags};

package/config/localized_name_keys.js

@@ -0,0 +1,44 @@
+
+/**
+ list of iso-639-1 language codes to extract from osm tags
+
+ this file is nessesary due to the free-tagging nature of
+ osm. @see ./test/fixtures for an example list of tags
+ contained in the planet file (sourced from taginfo)
+
+ some common keys we need to defend against:
+ 'name:ဗမာ', 'name:<阪神電鉄>', 'name:*'
+
+ Here is an example of the usage.
+ All these tags might appear on the same element:
+
+  name=Irgendwas        (the default name, used locally)
+  name:en=Something     (the name in English)
+  name:el=Κάτι          (the name in Greek)
+  name:de=Irgendwas     (the name in German)
+  name:pl=Coś           (the name in Polish)
+  name:fr=Quelque chose (the name in French)
+  name:es=Algo          (the name in Spanish)
+  name:it=Qualcosa      (the name in Italian)
+  name:ja=何か          (the name in Japanese)
+  name:ko=뭔가           (the name in Korean)
+  name:ko_rm=Mweonga    (the name in Romanised Korean)
+
+  @note: 'ko_rm' is present in the osm database but not supported
+  by pelias due to it not being listed in the ISO specification.
+
+  @see: http://wiki.openstreetmap.org/wiki/Talk:Names
+**/
+
+const _ = require('lodash');
+const iso6393 = require('iso-639-3');
+const keys = [];
+
+_.each(iso6393, (lang) => {
+  let code = _.get(lang, 'iso6391');
+  if( code ){
+    keys.push( code );
+  }
+});
+
+module.exports = keys;

package/config/popularity_map.js

@@ -0,0 +1,163 @@
+/**
+  The popularity mapper is responsible for generating a 'popularity'
+  value by inspecting OSM tags.
+
+  Disused and abandoned places are given a strong negative score.
+  If the popularity score is less than zero then the document is discarded.
+
+  Feel free to make changes to this mapping file!
+**/
+
+const mapping = {
+  // https://taginfo.openstreetmap.org/keys/importance
+  importance: {
+    international: { _score: 50000 },
+    national: { _score: 10000 },
+    regional: { _score: 5000 },
+    urban: { _score: 1000 },
+    suburban: { _score: 500 },
+    local: { _score: 100 },
+  },
+
+  // concordances
+  wikipedia: { _score: 3000 },
+  wikidata: { _score: 3000 },
+
+  // properties found on major landmarks &
+  // public buildings.
+  architect: { _score: 5000 },
+  heritage: { _score: 5000 },
+  'heritage:operator': { _score: 2000 },
+  historic: {
+    building: { _score: 5000 },
+    church: { _score: 5000 },
+    heritage: { _score: 7000 },
+    _score: 1000,
+  },
+  building: {
+    colour: { _score: 2000 },
+    material: { _score: 2000 },
+    supermarket: { _score: 2000 },
+    civic: { _score: 2000 },
+    government: { _score: 2000 },
+    hospital: { _score: 2000 },
+    train_station: { _score: 5000 },
+    transportation: { _score: 2000 },
+    university: { _score: 2000 },
+    public: { _score: 2000 },
+    sports_hall: { _score: 2000 },
+    historic: { _score: 5000 },
+  },
+  height: { _score: 2000 },
+  start_date: { _score: 2000 },
+  opening_date: { _score: 2000 },
+
+  // properties found on tourist attractions
+  tourism: {
+    visitors: { _score: 2000 },
+    aquarium: { _score: 2000 },
+    attraction: { _score: 1000 },
+    museum: { _score: 2000 },
+    theme_park: { _score: 2000 },
+    zoo: { _score: 2000 },
+    _score: 1000
+  },
+  museum: { _score: 2000 },
+  museum_type: { _score: 2000 },
+  opening_hours: { _score: 1000 },
+  operator: { _score: 1000 },
+  fee: { _score: 2000 },
+
+  // closed and abandoned places
+  // https://wiki.openstreetmap.org/wiki/Key:disused:
+  disused: { _score: -100000 },
+  'disused:shop': { _score: -100000 },
+  'disused:amenity': { _score: -100000 },
+  'disused:railway': { _score: -100000 },
+  'disused:leisure': { _score: -100000 },
+  // https://wiki.openstreetmap.org/wiki/Key:abandoned:
+  abandoned: { _score: -100000 },
+  'abandoned:shop': { _score: -100000 },
+  'abandoned:amenity': {
+    // note: some places such as 'Angkor Wat' are tagged 'abandoned:amenity=place_of_worship'.
+    // we use a positive integer to negate the effects of the base _score below (zeroing it out).
+    place_of_worship: { _score: +100000 },
+    _score: -100000
+  },
+  'abandoned:railway': { _score: -100000 },
+  'abandoned:leisure': { _score: -100000 },
+
+  // types of public amenities
+  amenity: {
+    disused: { _score: -100000 },
+    college: { _score: 2000 },
+    library: { _score: 2000 },
+    school: { _score: 1000 },
+    university: { _score: 2000 },
+    bank: { _score: 1000 },
+    clinic: { _score: 1000 },
+    dentist: { _score: 1000 },
+    doctors: { _score: 1000 },
+    hospital: { _score: 2000 },
+    nursing_home: { _score: 1000 },
+    pharmacy: { _score: 1000 },
+    social_facility: { _score: 1000 },
+    veterinary: { _score: 1000 },
+    community_centre: { _score: 1000 },
+    music_venue: { _score: 1000 },
+    nightclub: { _score: 1000 },
+    planetarium: { _score: 1000 },
+    social_centre: { _score: 1000 },
+    theatre: { _score: 1000 },
+    courthouse: { _score: 1000 },
+    coworking_space: { _score: 1000 },
+    dojo: { _score: 1000 },
+    embassy: { _score: 1000 },
+    ferry_terminal: { _score: 2000 },
+    fire_station: { _score: 1000 },
+    marketplace: { _score: 1000 },
+    place_of_worship: { _score: 1000 },
+    police: { _score: 1000 },
+    post_office: { _score: 1000 },
+    prison: { _score: 1000 },
+    public_bath: { _score: 1000 },
+    townhall: { _score: 1000 }
+  },
+
+  // transportation
+  aerodrome: {
+    international: { _score: 10000 },
+    regional: { _score: 5000 }
+  },
+  iata: {
+    _score: 5000,
+    none: { _score: -5000 }
+  },
+  railway: {
+    station: { _score: 2000 },
+    subway_entrance: { _score: 2000 },
+  },
+  public_transport: {
+    station: { _score: 2000 }
+    // no scoring boost for `public_transport:*`, as many unimportant records have other tags
+  },
+
+  // contact information
+  website: { _score: 200 },
+  phone: { _score: 200 },
+  'contact:website': { _score: 200 },
+  'contact:email': { _score: 200 },
+  'contact:phone': { _score: 200 },
+  'contact:fax': { _score: 200 },
+
+  // social media
+  'contact:foursquare': { _score: 200 },
+  'contact:facebook': { _score: 200 },
+  'contact:linkedin': { _score: 200 },
+  'contact:instagram': { _score: 200 },
+  'contact:skype': { _score: 200 },
+  'contact:flickr': { _score: 200 },
+  'contact:youtube': { _score: 200 },
+};
+
+module.exports = mapping;

package/index.js

@@ -0,0 +1,12 @@
+const peliasConfig = require('pelias-config').generate(require('./schema'));
+const _ = require('lodash');
+const logger = require('pelias-logger').get('openstreetmap');
+
+if (_.has(peliasConfig, 'imports.openstreetmap.adminLookup')) {
+  logger.info('imports.openstreetmap.adminLookup has been deprecated, ' +
+              'enable adminLookup using imports.adminLookup.enabled = true');
+}
+
+const importPipeline = require('./stream/importPipeline');
+
+importPipeline.import();