--- dao/DaoAdresseService/src/main/java/dk/daoas/daoadresseservice/AddressSearch.java 2015/07/15 06:52:05 2608 +++ dao/DaoAdresseService/src/main/java/dk/daoas/daoadresseservice/AddressSearch.java 2016/01/13 11:04:36 2804 @@ -30,11 +30,13 @@ import dk.daoas.daoadresseservice.util.DaoUtils; import dk.daoas.daoadresseservice.util.DeduplicateHelper; import dk.daoas.daoadresseservice.util.NaturalOrderComparator; +import dk.daoas.daoadresseservice.util.NearestShortTreeMap; public class AddressSearch { private Map> searchPostnrVejnavnGadeid; - private Map> searchGadeidentAdresser; + private Map>> searchGadeidentAdresser; // Gadeident -> husnrumre -> litra + private List
alleAdresser; @@ -43,26 +45,52 @@ private DataStatisticsBean stats = new DataStatisticsBean(); private DatabaseLayer db; + private ServiceConfig config; - ServiceConfig config; StreetnameHelper levenshteinHelper; + StreetnameHelper metaphoneHelper; StreetnameHelper googleHelper; StreetnameHelper osmHelper; public AddressSearch(DatabaseLayer db, ServiceConfig config) { - this.config = config; this.db = db; + this.config = config; levenshteinHelper = new LevenshteinStreetnameHelper( this ); + metaphoneHelper = new DoubleMetaphoneStreetnameHelper(config); googleHelper = new GoogleStreetnameHelper( config ); osmHelper = new OSMStreetnameHelper( config ); } + public Address getAddressSafe(int gadeid, short husnr, String litra) throws Exception { + + Map> gade = searchGadeidentAdresser.get(gadeid); + + if (gade == null) { + throw new Exception("GadeID ikke fundet"); + } + + Map litraList = gade.get(husnr); + + Address addr = litraList.get(litra); + + if (addr == null) { + throw new Exception("Adresse ikke ikke fundet"); + } + + return addr; + } + public SearchResult search(String postnrStr, String adresse) { + return search(postnrStr, adresse, false); + } + + + public SearchResult search(String postnrStr, String adresse, boolean naermesteHusnr) { postnrStr = postnrStr.trim(); adresse = adresse.trim(); @@ -154,7 +182,7 @@ result.gadeident = gadeident; - Map gade = searchGadeidentAdresser.get(gadeident); + Map> gade = searchGadeidentAdresser.get(gadeident); if (gade == null) { //Denne søgning må ikke fejle result.status = Status.ERROR_INTERNAL; return result; @@ -162,15 +190,40 @@ //Kunne evt klares med Iterables.get fra Guava/Collections - Address firstAddress = gade.values().iterator().next(); + Address firstAddress = gade.values().iterator().next().values().iterator().next(); //Første husnr -> første litra result.vej = firstAddress.vejnavn; - String husnrSearch = "" + result.splitResult.husnr + result.splitResult.litra; - Address addr = gade.get(husnrSearch); + short husnrSearch = Short.parseShort(result.splitResult.husnr ); + Map litraList = gade.get(husnrSearch); + - if (addr == null) { - result.status = Status.ERROR_UNKNOWN_ADDRESSPOINT; - return result; + if (litraList == null) { //Husnr ikke fundet + + if (naermesteHusnr) { + TreeMap> gadeTreeMap = (TreeMap>) gade; + result.anvendtHusnr = NearestShortTreeMap.getNearestKey(husnrSearch, gadeTreeMap); + + result.nearestHusnr = true; + litraList = gade.get(result.anvendtHusnr); + + + } else { + result.status = Status.ERROR_UNKNOWN_ADDRESSPOINT; + return result; + } + } + + Address addr = litraList.get( result.splitResult.litra ); + + if (addr == null) { //litra ikke fundet + if (config.nearestLitra == true || naermesteHusnr == true) { + result.nearestLitra = true; + addr = litraList.values().iterator().next(); + } else { + //husnr fundet men litra blev ikke fundet + result.status = Status.ERROR_UNKNOWN_ADDRESSPOINT; + return result; + } } result.address = addr; @@ -211,6 +264,13 @@ } } + if (gadeident == null) { + String vej = metaphoneHelper.proposeStreetName(request, result); + if (vej != null) { + gadeident = request.streetNames.get(vej); + } + } + // Brug OpenStreetMap før vi prøver google // For google har en request limit, det har OSM ikke! @@ -248,7 +308,7 @@ public void buildSearchStructures() throws SQLException{ searchPostnrVejnavnGadeid = new HashMap>(); - searchGadeidentAdresser = new HashMap>(); + searchGadeidentAdresser = new HashMap>>(); helperCache = new ConcurrentHashMap(); long start1 = System.currentTimeMillis(); @@ -260,16 +320,21 @@ /* Mapper mellem db Row ID og adresse noden */ Map idAddressMap = new HashMap( alleAdresser.size() ); DeduplicateHelper aliasHelper = new DeduplicateHelper(); - DeduplicateHelper husnrHelper = new DeduplicateHelper(); + + DeduplicateHelper shortHelper = new DeduplicateHelper(); + DeduplicateHelper intHelper = new DeduplicateHelper(); + for (Address a : alleAdresser) { idAddressMap.put(a.id, a); - Map postnrVeje = searchPostnrVejnavnGadeid.get(a.postnr); + Short postnr = shortHelper.getInstance(a.postnr); + + Map postnrVeje = searchPostnrVejnavnGadeid.get(postnr); if (postnrVeje == null) { - postnrVeje = new ConcurrentHashMap(); - searchPostnrVejnavnGadeid.put(a.postnr, postnrVeje); + postnrVeje = new HashMap(10240); //lige nu indeholder den største 9500 entries + searchPostnrVejnavnGadeid.put(postnr, postnrVeje); } @@ -278,7 +343,7 @@ if (gadeident == null) { //postnrVeje.put(vasketVejnavn, a.gadeid); - gadeident = a.gadeid; + gadeident = intHelper.getInstance( a.gadeid ); Set aliaser = findVejAliaser(a.vejnavn); for(String alias : aliaser) { @@ -286,19 +351,26 @@ vasketAlias = aliasHelper.getInstance(vasketAlias); postnrVeje.put(vasketAlias, gadeident); - } + } + } - Map gade = searchGadeidentAdresser.get(gadeident); + + Map> gade = searchGadeidentAdresser.get(gadeident); if (gade == null) { - gade = new HashMap(); + gade = new TreeMap>(); searchGadeidentAdresser.put(gadeident, gade); } - String husnrSearch = "" + a.husnr + a.husnrbogstav; - husnrSearch = husnrHelper.getInstance(husnrSearch); - gade.put(husnrSearch, a); + + Map litraList = gade.get(a.husnr); + if (litraList == null) { + litraList = new HashMap(); + gade.put(a.husnr, litraList); + } + + litraList.put(a.husnrbogstav, a); } - + //////////////////////////////////////////////////////////////////////////////////////// long start2 = System.currentTimeMillis(); System.out.println("Build, stage1 elapsed: " + (start2-start1) ); @@ -441,7 +513,7 @@ //////////////////////////////////////////////////////////////////////////////////// long stop = System.currentTimeMillis(); - System.out.println("Build, stage3 elapsed: " + (stop-start4) ); + System.out.println("Build, stage4 elapsed: " + (stop-start4) ); System.out.println("Build -- Gathering statistics"); for (Address addr : alleAdresser) { @@ -506,7 +578,7 @@ return ".." + targetAddress.rute; } else if (eb.afstand < 0.701) { return "..." + targetAddress.rute; - } else if (eb.afstand < 0.501) { + } else if (eb.afstand < 1.001) { return "...." + targetAddress.rute; } break; @@ -532,6 +604,8 @@ return "...." + targetAddress.rute; } break; + default: + System.out.println("Ukendt transport type: " + eb ); } return null; } @@ -547,6 +621,17 @@ return result; } + public List
get100PctAddresses() { + List
result = new ArrayList
(120000); + for (Address a : alleAdresser) { + if ( a.daekningsType == DaekningsType.DAEKNING_100PCT) { + result.add(a); + } + } + Collections.sort(result ); + return result; + } + /////////////////////////////////////////////////////////// private String calculateExtendedBkKoreliste(ExtendedBean eb, Address orgAddress, Address targetAddress) { String inject; @@ -585,6 +670,9 @@ aliasSet.add(vejnavn.replace("gl", "gammel") ); aliasSet.add(vejnavn.replace("gammel", "gl") ); + aliasSet.add(vejnavn.replace("gl", "gamle") ); + aliasSet.add(vejnavn.replace("gamle", "gl") ); + aliasSet.add(vejnavn.replace("lille", "ll") ); aliasSet.add(vejnavn.replace("ll ", "lille") ); aliasSet.add(vejnavn.replace("ll.", "lille") ); @@ -640,6 +728,12 @@ aliasSet.add(vejnavn.replace("christians", "chr") ); aliasSet.add(vejnavn.replace("kristian", "kr") ); aliasSet.add(vejnavn.replace("kristians", "kr") ); + + aliasSet.add(vejnavn.replace("mylius", "myl") ); + aliasSet.add(vejnavn.replace("myl", "mylius") ); + + aliasSet.add(vejnavn.replace("ludvig", "ludv") ); + aliasSet.add(vejnavn.replace("ludv", "ludvig") ); // alternative stavninger @@ -711,8 +805,7 @@ // Opbyg æøå varianter over alle fundne aliaser - @SuppressWarnings("unchecked") - HashSet variants = (HashSet) aliasSet.clone(); + HashSet variants = new HashSet( aliasSet ); for (String vVejnavn : variants) { // danske tegn 1 @@ -732,19 +825,19 @@ aliasSet.add( vVejnavn.replace("oe", "ø").replace("aa", "å") ); //danske tegn 3 - aliasSet.add( vejnavn.replace("æ", "ae").replace("ø", "oe").replace("å", "aa") ); - aliasSet.add( vejnavn.replace("ae", "æ").replace("oe", "ø").replace("aa", "å") ); + aliasSet.add( vVejnavn.replace("æ", "ae").replace("ø", "oe").replace("å", "aa") ); + aliasSet.add( vVejnavn.replace("ae", "æ").replace("oe", "ø").replace("aa", "å") ); //udenlandsketegn - aliasSet.add(vejnavn.replace("u", "ü") ); - aliasSet.add(vejnavn.replace("ü", "u") ); + aliasSet.add( vVejnavn.replace("u", "ü") ); + aliasSet.add( vVejnavn.replace("ü", "u") ); - aliasSet.add(vejnavn.replace("ä", "æ") ); - aliasSet.add(vejnavn.replace("æ", "ä") ); + aliasSet.add( vVejnavn.replace("ä", "æ") ); + aliasSet.add( vVejnavn.replace("æ", "ä") ); - aliasSet.add(vejnavn.replace("ö", "ø") ); - aliasSet.add(vejnavn.replace("ø", "ö") ); + aliasSet.add( vVejnavn.replace("ö", "ø") ); + aliasSet.add( vVejnavn.replace("ø", "ö") ); } return aliasSet; @@ -756,28 +849,33 @@ public Set> getVejnavne(short postnr) { Map map = searchPostnrVejnavnGadeid.get(postnr); - TreeMap newMap = new TreeMap(); - for (Entry entry : map.entrySet() ) { - newMap.put(entry.getKey(), entry.getValue()); - } + + TreeMap newMap = new TreeMap( map );//Lav et nyt TreeMap for at sikre sortering return newMap.entrySet(); } public Set getHusnumre(int gadeid) { - Map gade = searchGadeidentAdresser.get(gadeid); + Map> gade = searchGadeidentAdresser.get(gadeid); Set set = new TreeSet( new NaturalOrderComparator() ); - set.addAll( gade.keySet() ); + for(Map litraList : gade.values()) { + for (Address a: litraList.values()) { + set.add( a.husnr + a.husnrbogstav ); + } + } return set; } - public Address getAdresse(int gadeid, String husnr) { - - Map gade = searchGadeidentAdresser.get(gadeid); - return gade.get(husnr); + public Address getAdresse(int gadeid, String husnrStr) { + short husnr = Short.parseShort( husnrStr.replaceAll("[^\\d]","") ); + String litra = husnrStr.replaceAll("\\d", ""); + + Map> gade = searchGadeidentAdresser.get(gadeid); + Map litraList = gade.get(husnr); + return litraList.get(litra); }