| 1 |
torben |
2807 |
package dk.daoas.daoadresseservice;
|
| 2 |
|
|
|
| 3 |
torben |
2808 |
import java.util.ArrayList;
|
| 4 |
torben |
2814 |
import java.util.HashMap;
|
| 5 |
torben |
2807 |
import java.util.HashSet;
|
| 6 |
torben |
2814 |
import java.util.Map;
|
| 7 |
torben |
2807 |
import java.util.Set;
|
| 8 |
|
|
|
| 9 |
torben |
2815 |
import dk.daoas.daoadresseservice.util.DeduplicateHelper;
|
| 10 |
|
|
|
| 11 |
torben |
2807 |
public class AliasGenerator {
|
| 12 |
torben |
2808 |
|
| 13 |
|
|
static class Replacement {
|
| 14 |
|
|
public String search;
|
| 15 |
|
|
public String replace;
|
| 16 |
torben |
2807 |
|
| 17 |
torben |
2808 |
public Replacement(String search, String replace) {
|
| 18 |
|
|
this.search = search;
|
| 19 |
|
|
this.replace = replace;
|
| 20 |
|
|
}
|
| 21 |
|
|
}
|
| 22 |
|
|
|
| 23 |
torben |
2814 |
private ArrayList<Replacement> replacements = new ArrayList<Replacement>();
|
| 24 |
|
|
private Map<String, Set<String>> aliasCache = new HashMap<String, Set<String>>(8192);
|
| 25 |
torben |
2808 |
|
| 26 |
torben |
2815 |
DeduplicateHelper<String> aliasHelper = new DeduplicateHelper<String>(400000); // Indeholder pt 379664 entries
|
| 27 |
torben |
2814 |
|
| 28 |
torben |
2815 |
|
| 29 |
torben |
2808 |
public AliasGenerator() {
|
| 30 |
|
|
replacements.add( new Replacement("alle", "allé") );
|
| 31 |
|
|
replacements.add( new Replacement("allé", "alle") );
|
| 32 |
torben |
2807 |
|
| 33 |
|
|
// forkortelser
|
| 34 |
torben |
2808 |
replacements.add( new Replacement("dronningens", "dr") );
|
| 35 |
|
|
replacements.add( new Replacement("dr.", "dronningens") );
|
| 36 |
|
|
replacements.add( new Replacement("dr ", "dronningens") );
|
| 37 |
torben |
2807 |
|
| 38 |
torben |
2808 |
replacements.add( new Replacement("dronning", "dr") );
|
| 39 |
|
|
replacements.add( new Replacement("dr.", "dronning") );
|
| 40 |
|
|
replacements.add( new Replacement("dr ", "dronning") );
|
| 41 |
torben |
2807 |
|
| 42 |
torben |
2808 |
replacements.add( new Replacement("kng", "kongen") );
|
| 43 |
|
|
replacements.add( new Replacement("kongen", "kng") );
|
| 44 |
torben |
2807 |
|
| 45 |
torben |
2808 |
replacements.add( new Replacement("kvt", "kvarter") );
|
| 46 |
|
|
replacements.add( new Replacement("kvarter", "kvt") );
|
| 47 |
torben |
2807 |
|
| 48 |
torben |
2808 |
replacements.add( new Replacement("gl", "gammel") );
|
| 49 |
|
|
replacements.add( new Replacement("gammel", "gl") );
|
| 50 |
torben |
2807 |
|
| 51 |
torben |
2808 |
replacements.add( new Replacement("gl", "gamle") );
|
| 52 |
|
|
replacements.add( new Replacement("gamle", "gl") );
|
| 53 |
torben |
2807 |
|
| 54 |
torben |
2808 |
replacements.add( new Replacement("lille", "ll") );
|
| 55 |
|
|
replacements.add( new Replacement("ll ", "lille") );
|
| 56 |
|
|
replacements.add( new Replacement("ll.", "lille") );
|
| 57 |
torben |
2807 |
|
| 58 |
torben |
2808 |
replacements.add( new Replacement("store", "st") );
|
| 59 |
|
|
replacements.add( new Replacement("st ", "store") );
|
| 60 |
|
|
replacements.add( new Replacement("st.", "store") );
|
| 61 |
torben |
2807 |
|
| 62 |
torben |
2808 |
replacements.add( new Replacement("søndre", "sdr") );
|
| 63 |
|
|
replacements.add( new Replacement("sdr", "søndre") );
|
| 64 |
|
|
replacements.add( new Replacement("sønder", "sdr") );
|
| 65 |
|
|
replacements.add( new Replacement("sdr", "sønder") );
|
| 66 |
torben |
2807 |
|
| 67 |
|
|
|
| 68 |
torben |
2808 |
replacements.add( new Replacement("nørre", "nr") );
|
| 69 |
|
|
replacements.add( new Replacement("nr", "nørre") );
|
| 70 |
torben |
2807 |
|
| 71 |
torben |
2808 |
replacements.add( new Replacement("nordre", "ndr") );
|
| 72 |
|
|
replacements.add( new Replacement("ndr", "nordre") );
|
| 73 |
torben |
2807 |
|
| 74 |
|
|
|
| 75 |
torben |
2808 |
replacements.add( new Replacement("vestre", "vester") );
|
| 76 |
|
|
replacements.add( new Replacement("vester", "vestre") );
|
| 77 |
torben |
2807 |
|
| 78 |
torben |
2808 |
replacements.add( new Replacement("søndre", "sønder") );
|
| 79 |
|
|
replacements.add( new Replacement("sønder", "søndre") );
|
| 80 |
torben |
2807 |
|
| 81 |
torben |
2808 |
replacements.add( new Replacement("østre", "øster") );
|
| 82 |
|
|
replacements.add( new Replacement("øster", "østre") );
|
| 83 |
torben |
2807 |
|
| 84 |
torben |
2813 |
replacements.add( new Replacement("nordre", "nørre") );
|
| 85 |
|
|
replacements.add( new Replacement("nørre", "nordre") );
|
| 86 |
|
|
|
| 87 |
torben |
2808 |
replacements.add( new Replacement("borgmester", "borgm") );
|
| 88 |
|
|
replacements.add( new Replacement("borgm", "borgmester") );
|
| 89 |
torben |
2807 |
|
| 90 |
|
|
|
| 91 |
torben |
2808 |
replacements.add( new Replacement("sankt", "skt") );
|
| 92 |
|
|
replacements.add( new Replacement("sankt", "sct") );
|
| 93 |
|
|
replacements.add( new Replacement("skt", "sankt") );
|
| 94 |
torben |
2807 |
|
| 95 |
torben |
2808 |
replacements.add( new Replacement("skt", "sct") );
|
| 96 |
|
|
replacements.add( new Replacement("sct", "skt") );
|
| 97 |
torben |
2807 |
|
| 98 |
|
|
|
| 99 |
|
|
|
| 100 |
|
|
// fornavn forkortelser
|
| 101 |
torben |
2808 |
replacements.add( new Replacement("john", "j") );
|
| 102 |
|
|
replacements.add( new Replacement("thomas", "th") );
|
| 103 |
|
|
replacements.add( new Replacement("thorvald", "th") );
|
| 104 |
|
|
replacements.add( new Replacement("frederik", "fr") );
|
| 105 |
|
|
replacements.add( new Replacement("frederiks", "fr") );
|
| 106 |
torben |
2807 |
|
| 107 |
|
|
|
| 108 |
torben |
2808 |
replacements.add( new Replacement("christian", "chr") );
|
| 109 |
|
|
replacements.add( new Replacement("christians", "chr") );
|
| 110 |
|
|
replacements.add( new Replacement("kristian", "kr") );
|
| 111 |
|
|
replacements.add( new Replacement("kristians", "kr") );
|
| 112 |
torben |
2807 |
|
| 113 |
torben |
2808 |
replacements.add( new Replacement("mylius", "myl") );
|
| 114 |
|
|
replacements.add( new Replacement("myl", "mylius") );
|
| 115 |
torben |
2807 |
|
| 116 |
torben |
2808 |
replacements.add( new Replacement("ludvig", "ludv") );
|
| 117 |
|
|
replacements.add( new Replacement("ludv", "ludvig") );
|
| 118 |
torben |
2807 |
|
| 119 |
|
|
|
| 120 |
|
|
// alternative stavninger
|
| 121 |
torben |
2808 |
replacements.add( new Replacement("peder", "peter") );
|
| 122 |
|
|
replacements.add( new Replacement("peter", "peder") );
|
| 123 |
torben |
2807 |
|
| 124 |
|
|
|
| 125 |
|
|
|
| 126 |
|
|
// Romertal
|
| 127 |
torben |
2808 |
replacements.add( new Replacement("1", "i") );
|
| 128 |
|
|
replacements.add( new Replacement("i", "1") );
|
| 129 |
torben |
2807 |
|
| 130 |
torben |
2808 |
replacements.add( new Replacement("2", "ii") );
|
| 131 |
|
|
replacements.add( new Replacement("ii", "2") );
|
| 132 |
torben |
2807 |
|
| 133 |
torben |
2808 |
replacements.add( new Replacement("3", "iii") );
|
| 134 |
|
|
replacements.add( new Replacement("iii", "3") );
|
| 135 |
torben |
2807 |
|
| 136 |
torben |
2808 |
replacements.add( new Replacement("4", "iv") );
|
| 137 |
|
|
replacements.add( new Replacement("iv", "4") );
|
| 138 |
torben |
2807 |
|
| 139 |
torben |
2808 |
replacements.add( new Replacement("5", "v") );
|
| 140 |
|
|
replacements.add( new Replacement("v", "5") );
|
| 141 |
torben |
2807 |
|
| 142 |
torben |
2808 |
replacements.add( new Replacement("6", "vi") );
|
| 143 |
|
|
replacements.add( new Replacement("vi", "6") );
|
| 144 |
torben |
2807 |
|
| 145 |
torben |
2808 |
replacements.add( new Replacement("7", "vii") );
|
| 146 |
|
|
replacements.add( new Replacement("vii", "7") );
|
| 147 |
torben |
2807 |
|
| 148 |
torben |
2808 |
replacements.add( new Replacement("8", "viii") );
|
| 149 |
|
|
replacements.add( new Replacement("viii", "8") );
|
| 150 |
torben |
2807 |
|
| 151 |
torben |
2808 |
replacements.add( new Replacement("9", "ix") );
|
| 152 |
|
|
replacements.add( new Replacement("ix", "9") );
|
| 153 |
torben |
2807 |
|
| 154 |
torben |
2808 |
replacements.add( new Replacement("10", "x") );
|
| 155 |
|
|
replacements.add( new Replacement("x", "10") );
|
| 156 |
torben |
2807 |
|
| 157 |
|
|
|
| 158 |
|
|
//alternative måder at stave vej/gade/alle
|
| 159 |
torben |
2808 |
replacements.add( new Replacement("vej", "ve") );
|
| 160 |
|
|
replacements.add( new Replacement("vej", "vj") );
|
| 161 |
|
|
replacements.add( new Replacement("vej", "v") );
|
| 162 |
|
|
replacements.add( new Replacement("vej", "vejen") );
|
| 163 |
|
|
replacements.add( new Replacement("vejen", "vej") );
|
| 164 |
torben |
2807 |
|
| 165 |
torben |
2808 |
replacements.add( new Replacement("v", "vej") );//Vi kan have vejnavne i DB der bare slutter på v istedet for vej
|
| 166 |
torben |
2807 |
|
| 167 |
torben |
2808 |
replacements.add( new Replacement("alle", "all") );
|
| 168 |
|
|
replacements.add( new Replacement("allé", "all") );
|
| 169 |
|
|
replacements.add( new Replacement("alle", "allú") ); //Fundet i logs.hentruteinfo
|
| 170 |
|
|
replacements.add( new Replacement("allé", "allú") );
|
| 171 |
torben |
2807 |
|
| 172 |
torben |
2808 |
replacements.add( new Replacement("gade", "gaed") ); //Fundet i logs.hentruteinfo
|
| 173 |
|
|
replacements.add( new Replacement("gade", "gde") );
|
| 174 |
torben |
2807 |
|
| 175 |
torben |
2808 |
replacements.add( new Replacement("plads", "pl") );
|
| 176 |
|
|
replacements.add( new Replacement("vænget", "vænge") );
|
| 177 |
torben |
2807 |
|
| 178 |
torben |
2808 |
replacements.add( new Replacement("boulevard", "blvd") );
|
| 179 |
|
|
replacements.add( new Replacement("blvd", "boulevard") );
|
| 180 |
torben |
2807 |
|
| 181 |
torben |
2808 |
replacements.add( new Replacement("boulevard", "boule") );
|
| 182 |
|
|
replacements.add( new Replacement("boule", "boulevard") );
|
| 183 |
torben |
2807 |
|
| 184 |
torben |
2808 |
replacements.add( new Replacement("kronborg", "krbg") ); //Krbg ladegårdsvej, 3000
|
| 185 |
|
|
replacements.add( new Replacement("krbg", "kronborg") );
|
| 186 |
torben |
2810 |
|
| 187 |
|
|
System.out.println("AliasGenetator, loaded replacements: " + replacements.size() );
|
| 188 |
torben |
2808 |
|
| 189 |
|
|
}
|
| 190 |
|
|
|
| 191 |
torben |
2815 |
public void clear() {
|
| 192 |
|
|
aliasCache.clear();
|
| 193 |
|
|
aliasHelper.clear();
|
| 194 |
|
|
}
|
| 195 |
torben |
2808 |
|
| 196 |
torben |
2815 |
|
| 197 |
torben |
2808 |
public Set<String> findVejAliaser(String vejnavn) {
|
| 198 |
|
|
vejnavn = vejnavn.toLowerCase();
|
| 199 |
torben |
2814 |
|
| 200 |
|
|
Set<String> aliasSet = aliasCache.get(vejnavn);
|
| 201 |
|
|
if (aliasSet == null) {
|
| 202 |
|
|
aliasSet = genererVejAliaser(vejnavn);
|
| 203 |
torben |
2815 |
|
| 204 |
|
|
aliasSet = vaskAliaser( aliasSet );
|
| 205 |
|
|
|
| 206 |
torben |
2814 |
aliasCache.put(vejnavn, aliasSet);
|
| 207 |
|
|
}
|
| 208 |
|
|
|
| 209 |
|
|
return aliasSet;
|
| 210 |
|
|
|
| 211 |
|
|
}
|
| 212 |
|
|
|
| 213 |
torben |
2815 |
private Set<String> vaskAliaser(Set<String> input) {
|
| 214 |
|
|
HashSet<String> output = new HashSet<String>( input.size() );
|
| 215 |
|
|
|
| 216 |
|
|
for (String alias : input) {
|
| 217 |
|
|
String vasketAlias = AddressUtils.vaskVejnavn(alias);
|
| 218 |
|
|
vasketAlias = aliasHelper.getInstance(vasketAlias);
|
| 219 |
|
|
output.add(vasketAlias);
|
| 220 |
|
|
}
|
| 221 |
|
|
|
| 222 |
|
|
return output;
|
| 223 |
|
|
}
|
| 224 |
torben |
2814 |
|
| 225 |
torben |
2815 |
|
| 226 |
torben |
2814 |
private Set<String> genererVejAliaser(String vejnavn) {
|
| 227 |
|
|
|
| 228 |
torben |
2808 |
HashSet<String> aliasSet = new HashSet<String>();
|
| 229 |
|
|
aliasSet.add(vejnavn);
|
| 230 |
torben |
2807 |
|
| 231 |
torben |
2808 |
for (Replacement repl : replacements) {
|
| 232 |
torben |
2809 |
if (vejnavn.indexOf(repl.search) >= 0) {
|
| 233 |
|
|
aliasSet.add( vejnavn.replace(repl.search, repl.replace) );
|
| 234 |
|
|
}
|
| 235 |
|
|
|
| 236 |
torben |
2808 |
}
|
| 237 |
|
|
|
| 238 |
|
|
|
| 239 |
|
|
|
| 240 |
torben |
2807 |
// Opbyg æøå varianter over alle fundne aliaser
|
| 241 |
|
|
|
| 242 |
torben |
2808 |
HashSet<String> variants = new HashSet<String>( aliasSet );//brug Copy-Constructor
|
| 243 |
torben |
2807 |
|
| 244 |
|
|
for (String vVejnavn : variants) {
|
| 245 |
|
|
// danske tegn 1
|
| 246 |
|
|
aliasSet.add( vVejnavn.replace("æ", "ae") );
|
| 247 |
|
|
aliasSet.add( vVejnavn.replace("ø", "oe") );
|
| 248 |
|
|
aliasSet.add( vVejnavn.replace("å", "aa") );
|
| 249 |
|
|
aliasSet.add( vVejnavn.replace("ae", "æ") );
|
| 250 |
|
|
aliasSet.add( vVejnavn.replace("oe", "ø") );
|
| 251 |
|
|
aliasSet.add( vVejnavn.replace("aa", "å") );
|
| 252 |
|
|
|
| 253 |
|
|
//danske tegn 2
|
| 254 |
|
|
aliasSet.add( vVejnavn.replace("æ", "ae").replace("ø", "oe") );
|
| 255 |
|
|
aliasSet.add( vVejnavn.replace("æ", "ae").replace("å", "aa") );
|
| 256 |
|
|
aliasSet.add( vVejnavn.replace("ø", "oe").replace("å", "aa") );
|
| 257 |
|
|
aliasSet.add( vVejnavn.replace("ae", "æ").replace("oe","ø") );
|
| 258 |
|
|
aliasSet.add( vVejnavn.replace("ae", "æ").replace("aa","å") );
|
| 259 |
|
|
aliasSet.add( vVejnavn.replace("oe", "ø").replace("aa", "å") );
|
| 260 |
|
|
|
| 261 |
|
|
//danske tegn 3
|
| 262 |
|
|
aliasSet.add( vVejnavn.replace("æ", "ae").replace("ø", "oe").replace("å", "aa") );
|
| 263 |
|
|
aliasSet.add( vVejnavn.replace("ae", "æ").replace("oe", "ø").replace("aa", "å") );
|
| 264 |
|
|
|
| 265 |
|
|
|
| 266 |
|
|
//udenlandsketegn
|
| 267 |
|
|
aliasSet.add( vVejnavn.replace("u", "ü") );
|
| 268 |
|
|
aliasSet.add( vVejnavn.replace("ü", "u") );
|
| 269 |
|
|
|
| 270 |
|
|
aliasSet.add( vVejnavn.replace("ä", "æ") );
|
| 271 |
|
|
aliasSet.add( vVejnavn.replace("æ", "ä") );
|
| 272 |
|
|
|
| 273 |
|
|
aliasSet.add( vVejnavn.replace("ö", "ø") );
|
| 274 |
|
|
aliasSet.add( vVejnavn.replace("ø", "ö") );
|
| 275 |
|
|
}
|
| 276 |
|
|
|
| 277 |
|
|
return aliasSet;
|
| 278 |
|
|
}
|
| 279 |
|
|
}
|
Parent Directory
| 
Revision Log